diff options
| author | Guido van Rossum <guido@python.org> | 2007-05-02 19:09:54 (GMT) |
|---|---|---|
| committer | Guido van Rossum <guido@python.org> | 2007-05-02 19:09:54 (GMT) |
| commit | ef87d6ed94780fe00250a551031023aeb2898365 (patch) | |
| tree | 1f8989aaaec7ec5f8b2f26498317f2022bf85531 /Lib/encodings | |
| parent | 572dbf8f1320c0b34b9c786e5c30ba4a4b61b292 (diff) | |
| download | cpython-ef87d6ed94780fe00250a551031023aeb2898365.zip cpython-ef87d6ed94780fe00250a551031023aeb2898365.tar.gz cpython-ef87d6ed94780fe00250a551031023aeb2898365.tar.bz2 | |
Rip out all the u"..." literals and calls to unicode().
Diffstat (limited to 'Lib/encodings')
65 files changed, 15638 insertions, 15638 deletions
diff --git a/Lib/encodings/__init__.py b/Lib/encodings/__init__.py index 383f3b0..5bfee44 100644 --- a/Lib/encodings/__init__.py +++ b/Lib/encodings/__init__.py @@ -60,7 +60,7 @@ def normalize_encoding(encoding): """ # Make sure we have an 8-bit string, because .translate() works # differently for Unicode strings. - if isinstance(encoding, unicode): + if isinstance(encoding, str): # Note that .encode('latin-1') does *not* use the codec # registry, so this call doesn't recurse. (See unicodeobject.c # PyUnicode_AsEncodedString() for details) diff --git a/Lib/encodings/cp037.py b/Lib/encodings/cp037.py index c802b89..4edd708 100644 --- a/Lib/encodings/cp037.py +++ b/Lib/encodings/cp037.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x9c' # 0x04 -> CONTROL - u'\t' # 0x05 -> HORIZONTAL TABULATION - u'\x86' # 0x06 -> CONTROL - u'\x7f' # 0x07 -> DELETE - u'\x97' # 0x08 -> CONTROL - u'\x8d' # 0x09 -> CONTROL - u'\x8e' # 0x0A -> CONTROL - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x9d' # 0x14 -> CONTROL - u'\x85' # 0x15 -> CONTROL - u'\x08' # 0x16 -> BACKSPACE - u'\x87' # 0x17 -> CONTROL - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x92' # 0x1A -> CONTROL - u'\x8f' # 0x1B -> CONTROL - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u'\x80' # 0x20 -> CONTROL - u'\x81' # 0x21 -> CONTROL - u'\x82' # 0x22 -> CONTROL - u'\x83' # 0x23 -> CONTROL - u'\x84' # 0x24 -> CONTROL - u'\n' # 0x25 -> LINE FEED - u'\x17' # 0x26 -> END OF TRANSMISSION BLOCK - u'\x1b' # 0x27 -> ESCAPE - u'\x88' # 0x28 -> CONTROL - u'\x89' # 0x29 -> CONTROL - u'\x8a' # 0x2A -> CONTROL - u'\x8b' # 0x2B -> CONTROL - u'\x8c' # 0x2C -> CONTROL - u'\x05' # 0x2D -> ENQUIRY - u'\x06' # 0x2E -> ACKNOWLEDGE - u'\x07' # 0x2F -> BELL - u'\x90' # 0x30 -> CONTROL - u'\x91' # 0x31 -> CONTROL - u'\x16' # 0x32 -> SYNCHRONOUS IDLE - u'\x93' # 0x33 -> CONTROL - u'\x94' # 0x34 -> CONTROL - u'\x95' # 0x35 -> CONTROL - u'\x96' # 0x36 -> CONTROL - u'\x04' # 0x37 -> END OF TRANSMISSION - u'\x98' # 0x38 -> CONTROL - u'\x99' # 0x39 -> CONTROL - u'\x9a' # 0x3A -> CONTROL - u'\x9b' # 0x3B -> CONTROL - u'\x14' # 0x3C -> DEVICE CONTROL FOUR - u'\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE - u'\x9e' # 0x3E -> CONTROL - u'\x1a' # 0x3F -> SUBSTITUTE - u' ' # 0x40 -> SPACE - u'\xa0' # 0x41 -> NO-BREAK SPACE - u'\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x48 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE - u'\xa2' # 0x4A -> CENT SIGN - u'.' # 0x4B -> FULL STOP - u'<' # 0x4C -> LESS-THAN SIGN - u'(' # 0x4D -> LEFT PARENTHESIS - u'+' # 0x4E -> PLUS SIGN - u'|' # 0x4F -> VERTICAL LINE - u'&' # 0x50 -> AMPERSAND - u'\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE - u'\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE - u'\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) - u'!' # 0x5A -> EXCLAMATION MARK - u'$' # 0x5B -> DOLLAR SIGN - u'*' # 0x5C -> ASTERISK - u')' # 0x5D -> RIGHT PARENTHESIS - u';' # 0x5E -> SEMICOLON - u'\xac' # 0x5F -> NOT SIGN - u'-' # 0x60 -> HYPHEN-MINUS - u'/' # 0x61 -> SOLIDUS - u'\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xa6' # 0x6A -> BROKEN BAR - u',' # 0x6B -> COMMA - u'%' # 0x6C -> PERCENT SIGN - u'_' # 0x6D -> LOW LINE - u'>' # 0x6E -> GREATER-THAN SIGN - u'?' # 0x6F -> QUESTION MARK - u'\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE - u'\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE - u'`' # 0x79 -> GRAVE ACCENT - u':' # 0x7A -> COLON - u'#' # 0x7B -> NUMBER SIGN - u'@' # 0x7C -> COMMERCIAL AT - u"'" # 0x7D -> APOSTROPHE - u'=' # 0x7E -> EQUALS SIGN - u'"' # 0x7F -> QUOTATION MARK - u'\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE - u'a' # 0x81 -> LATIN SMALL LETTER A - u'b' # 0x82 -> LATIN SMALL LETTER B - u'c' # 0x83 -> LATIN SMALL LETTER C - u'd' # 0x84 -> LATIN SMALL LETTER D - u'e' # 0x85 -> LATIN SMALL LETTER E - u'f' # 0x86 -> LATIN SMALL LETTER F - u'g' # 0x87 -> LATIN SMALL LETTER G - u'h' # 0x88 -> LATIN SMALL LETTER H - u'i' # 0x89 -> LATIN SMALL LETTER I - u'\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xf0' # 0x8C -> LATIN SMALL LETTER ETH (ICELANDIC) - u'\xfd' # 0x8D -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0x8E -> LATIN SMALL LETTER THORN (ICELANDIC) - u'\xb1' # 0x8F -> PLUS-MINUS SIGN - u'\xb0' # 0x90 -> DEGREE SIGN - u'j' # 0x91 -> LATIN SMALL LETTER J - u'k' # 0x92 -> LATIN SMALL LETTER K - u'l' # 0x93 -> LATIN SMALL LETTER L - u'm' # 0x94 -> LATIN SMALL LETTER M - u'n' # 0x95 -> LATIN SMALL LETTER N - u'o' # 0x96 -> LATIN SMALL LETTER O - u'p' # 0x97 -> LATIN SMALL LETTER P - u'q' # 0x98 -> LATIN SMALL LETTER Q - u'r' # 0x99 -> LATIN SMALL LETTER R - u'\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR - u'\xe6' # 0x9C -> LATIN SMALL LIGATURE AE - u'\xb8' # 0x9D -> CEDILLA - u'\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE - u'\xa4' # 0x9F -> CURRENCY SIGN - u'\xb5' # 0xA0 -> MICRO SIGN - u'~' # 0xA1 -> TILDE - u's' # 0xA2 -> LATIN SMALL LETTER S - u't' # 0xA3 -> LATIN SMALL LETTER T - u'u' # 0xA4 -> LATIN SMALL LETTER U - u'v' # 0xA5 -> LATIN SMALL LETTER V - u'w' # 0xA6 -> LATIN SMALL LETTER W - u'x' # 0xA7 -> LATIN SMALL LETTER X - u'y' # 0xA8 -> LATIN SMALL LETTER Y - u'z' # 0xA9 -> LATIN SMALL LETTER Z - u'\xa1' # 0xAA -> INVERTED EXCLAMATION MARK - u'\xbf' # 0xAB -> INVERTED QUESTION MARK - u'\xd0' # 0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC) - u'\xdd' # 0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC) - u'\xae' # 0xAF -> REGISTERED SIGN - u'^' # 0xB0 -> CIRCUMFLEX ACCENT - u'\xa3' # 0xB1 -> POUND SIGN - u'\xa5' # 0xB2 -> YEN SIGN - u'\xb7' # 0xB3 -> MIDDLE DOT - u'\xa9' # 0xB4 -> COPYRIGHT SIGN - u'\xa7' # 0xB5 -> SECTION SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS - u'[' # 0xBA -> LEFT SQUARE BRACKET - u']' # 0xBB -> RIGHT SQUARE BRACKET - u'\xaf' # 0xBC -> MACRON - u'\xa8' # 0xBD -> DIAERESIS - u'\xb4' # 0xBE -> ACUTE ACCENT - u'\xd7' # 0xBF -> MULTIPLICATION SIGN - u'{' # 0xC0 -> LEFT CURLY BRACKET - u'A' # 0xC1 -> LATIN CAPITAL LETTER A - u'B' # 0xC2 -> LATIN CAPITAL LETTER B - u'C' # 0xC3 -> LATIN CAPITAL LETTER C - u'D' # 0xC4 -> LATIN CAPITAL LETTER D - u'E' # 0xC5 -> LATIN CAPITAL LETTER E - u'F' # 0xC6 -> LATIN CAPITAL LETTER F - u'G' # 0xC7 -> LATIN CAPITAL LETTER G - u'H' # 0xC8 -> LATIN CAPITAL LETTER H - u'I' # 0xC9 -> LATIN CAPITAL LETTER I - u'\xad' # 0xCA -> SOFT HYPHEN - u'\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0xCC -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE - u'\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE - u'}' # 0xD0 -> RIGHT CURLY BRACKET - u'J' # 0xD1 -> LATIN CAPITAL LETTER J - u'K' # 0xD2 -> LATIN CAPITAL LETTER K - u'L' # 0xD3 -> LATIN CAPITAL LETTER L - u'M' # 0xD4 -> LATIN CAPITAL LETTER M - u'N' # 0xD5 -> LATIN CAPITAL LETTER N - u'O' # 0xD6 -> LATIN CAPITAL LETTER O - u'P' # 0xD7 -> LATIN CAPITAL LETTER P - u'Q' # 0xD8 -> LATIN CAPITAL LETTER Q - u'R' # 0xD9 -> LATIN CAPITAL LETTER R - u'\xb9' # 0xDA -> SUPERSCRIPT ONE - u'\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xDC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE - u'\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\\' # 0xE0 -> REVERSE SOLIDUS - u'\xf7' # 0xE1 -> DIVISION SIGN - u'S' # 0xE2 -> LATIN CAPITAL LETTER S - u'T' # 0xE3 -> LATIN CAPITAL LETTER T - u'U' # 0xE4 -> LATIN CAPITAL LETTER U - u'V' # 0xE5 -> LATIN CAPITAL LETTER V - u'W' # 0xE6 -> LATIN CAPITAL LETTER W - u'X' # 0xE7 -> LATIN CAPITAL LETTER X - u'Y' # 0xE8 -> LATIN CAPITAL LETTER Y - u'Z' # 0xE9 -> LATIN CAPITAL LETTER Z - u'\xb2' # 0xEA -> SUPERSCRIPT TWO - u'\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd6' # 0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE - u'0' # 0xF0 -> DIGIT ZERO - u'1' # 0xF1 -> DIGIT ONE - u'2' # 0xF2 -> DIGIT TWO - u'3' # 0xF3 -> DIGIT THREE - u'4' # 0xF4 -> DIGIT FOUR - u'5' # 0xF5 -> DIGIT FIVE - u'6' # 0xF6 -> DIGIT SIX - u'7' # 0xF7 -> DIGIT SEVEN - u'8' # 0xF8 -> DIGIT EIGHT - u'9' # 0xF9 -> DIGIT NINE - u'\xb3' # 0xFA -> SUPERSCRIPT THREE - u'\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE - u'\x9f' # 0xFF -> CONTROL + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x9c' # 0x04 -> CONTROL + '\t' # 0x05 -> HORIZONTAL TABULATION + '\x86' # 0x06 -> CONTROL + '\x7f' # 0x07 -> DELETE + '\x97' # 0x08 -> CONTROL + '\x8d' # 0x09 -> CONTROL + '\x8e' # 0x0A -> CONTROL + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x9d' # 0x14 -> CONTROL + '\x85' # 0x15 -> CONTROL + '\x08' # 0x16 -> BACKSPACE + '\x87' # 0x17 -> CONTROL + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x92' # 0x1A -> CONTROL + '\x8f' # 0x1B -> CONTROL + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + '\x80' # 0x20 -> CONTROL + '\x81' # 0x21 -> CONTROL + '\x82' # 0x22 -> CONTROL + '\x83' # 0x23 -> CONTROL + '\x84' # 0x24 -> CONTROL + '\n' # 0x25 -> LINE FEED + '\x17' # 0x26 -> END OF TRANSMISSION BLOCK + '\x1b' # 0x27 -> ESCAPE + '\x88' # 0x28 -> CONTROL + '\x89' # 0x29 -> CONTROL + '\x8a' # 0x2A -> CONTROL + '\x8b' # 0x2B -> CONTROL + '\x8c' # 0x2C -> CONTROL + '\x05' # 0x2D -> ENQUIRY + '\x06' # 0x2E -> ACKNOWLEDGE + '\x07' # 0x2F -> BELL + '\x90' # 0x30 -> CONTROL + '\x91' # 0x31 -> CONTROL + '\x16' # 0x32 -> SYNCHRONOUS IDLE + '\x93' # 0x33 -> CONTROL + '\x94' # 0x34 -> CONTROL + '\x95' # 0x35 -> CONTROL + '\x96' # 0x36 -> CONTROL + '\x04' # 0x37 -> END OF TRANSMISSION + '\x98' # 0x38 -> CONTROL + '\x99' # 0x39 -> CONTROL + '\x9a' # 0x3A -> CONTROL + '\x9b' # 0x3B -> CONTROL + '\x14' # 0x3C -> DEVICE CONTROL FOUR + '\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE + '\x9e' # 0x3E -> CONTROL + '\x1a' # 0x3F -> SUBSTITUTE + ' ' # 0x40 -> SPACE + '\xa0' # 0x41 -> NO-BREAK SPACE + '\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE + '\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x48 -> LATIN SMALL LETTER C WITH CEDILLA + '\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE + '\xa2' # 0x4A -> CENT SIGN + '.' # 0x4B -> FULL STOP + '<' # 0x4C -> LESS-THAN SIGN + '(' # 0x4D -> LEFT PARENTHESIS + '+' # 0x4E -> PLUS SIGN + '|' # 0x4F -> VERTICAL LINE + '&' # 0x50 -> AMPERSAND + '\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE + '\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE + '\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) + '!' # 0x5A -> EXCLAMATION MARK + '$' # 0x5B -> DOLLAR SIGN + '*' # 0x5C -> ASTERISK + ')' # 0x5D -> RIGHT PARENTHESIS + ';' # 0x5E -> SEMICOLON + '\xac' # 0x5F -> NOT SIGN + '-' # 0x60 -> HYPHEN-MINUS + '/' # 0x61 -> SOLIDUS + '\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE + '\xa6' # 0x6A -> BROKEN BAR + ',' # 0x6B -> COMMA + '%' # 0x6C -> PERCENT SIGN + '_' # 0x6D -> LOW LINE + '>' # 0x6E -> GREATER-THAN SIGN + '?' # 0x6F -> QUESTION MARK + '\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE + '\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE + '`' # 0x79 -> GRAVE ACCENT + ':' # 0x7A -> COLON + '#' # 0x7B -> NUMBER SIGN + '@' # 0x7C -> COMMERCIAL AT + "'" # 0x7D -> APOSTROPHE + '=' # 0x7E -> EQUALS SIGN + '"' # 0x7F -> QUOTATION MARK + '\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE + 'a' # 0x81 -> LATIN SMALL LETTER A + 'b' # 0x82 -> LATIN SMALL LETTER B + 'c' # 0x83 -> LATIN SMALL LETTER C + 'd' # 0x84 -> LATIN SMALL LETTER D + 'e' # 0x85 -> LATIN SMALL LETTER E + 'f' # 0x86 -> LATIN SMALL LETTER F + 'g' # 0x87 -> LATIN SMALL LETTER G + 'h' # 0x88 -> LATIN SMALL LETTER H + 'i' # 0x89 -> LATIN SMALL LETTER I + '\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xf0' # 0x8C -> LATIN SMALL LETTER ETH (ICELANDIC) + '\xfd' # 0x8D -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0x8E -> LATIN SMALL LETTER THORN (ICELANDIC) + '\xb1' # 0x8F -> PLUS-MINUS SIGN + '\xb0' # 0x90 -> DEGREE SIGN + 'j' # 0x91 -> LATIN SMALL LETTER J + 'k' # 0x92 -> LATIN SMALL LETTER K + 'l' # 0x93 -> LATIN SMALL LETTER L + 'm' # 0x94 -> LATIN SMALL LETTER M + 'n' # 0x95 -> LATIN SMALL LETTER N + 'o' # 0x96 -> LATIN SMALL LETTER O + 'p' # 0x97 -> LATIN SMALL LETTER P + 'q' # 0x98 -> LATIN SMALL LETTER Q + 'r' # 0x99 -> LATIN SMALL LETTER R + '\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR + '\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR + '\xe6' # 0x9C -> LATIN SMALL LIGATURE AE + '\xb8' # 0x9D -> CEDILLA + '\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE + '\xa4' # 0x9F -> CURRENCY SIGN + '\xb5' # 0xA0 -> MICRO SIGN + '~' # 0xA1 -> TILDE + 's' # 0xA2 -> LATIN SMALL LETTER S + 't' # 0xA3 -> LATIN SMALL LETTER T + 'u' # 0xA4 -> LATIN SMALL LETTER U + 'v' # 0xA5 -> LATIN SMALL LETTER V + 'w' # 0xA6 -> LATIN SMALL LETTER W + 'x' # 0xA7 -> LATIN SMALL LETTER X + 'y' # 0xA8 -> LATIN SMALL LETTER Y + 'z' # 0xA9 -> LATIN SMALL LETTER Z + '\xa1' # 0xAA -> INVERTED EXCLAMATION MARK + '\xbf' # 0xAB -> INVERTED QUESTION MARK + '\xd0' # 0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC) + '\xdd' # 0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC) + '\xae' # 0xAF -> REGISTERED SIGN + '^' # 0xB0 -> CIRCUMFLEX ACCENT + '\xa3' # 0xB1 -> POUND SIGN + '\xa5' # 0xB2 -> YEN SIGN + '\xb7' # 0xB3 -> MIDDLE DOT + '\xa9' # 0xB4 -> COPYRIGHT SIGN + '\xa7' # 0xB5 -> SECTION SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF + '\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS + '[' # 0xBA -> LEFT SQUARE BRACKET + ']' # 0xBB -> RIGHT SQUARE BRACKET + '\xaf' # 0xBC -> MACRON + '\xa8' # 0xBD -> DIAERESIS + '\xb4' # 0xBE -> ACUTE ACCENT + '\xd7' # 0xBF -> MULTIPLICATION SIGN + '{' # 0xC0 -> LEFT CURLY BRACKET + 'A' # 0xC1 -> LATIN CAPITAL LETTER A + 'B' # 0xC2 -> LATIN CAPITAL LETTER B + 'C' # 0xC3 -> LATIN CAPITAL LETTER C + 'D' # 0xC4 -> LATIN CAPITAL LETTER D + 'E' # 0xC5 -> LATIN CAPITAL LETTER E + 'F' # 0xC6 -> LATIN CAPITAL LETTER F + 'G' # 0xC7 -> LATIN CAPITAL LETTER G + 'H' # 0xC8 -> LATIN CAPITAL LETTER H + 'I' # 0xC9 -> LATIN CAPITAL LETTER I + '\xad' # 0xCA -> SOFT HYPHEN + '\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0xCC -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE + '\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE + '}' # 0xD0 -> RIGHT CURLY BRACKET + 'J' # 0xD1 -> LATIN CAPITAL LETTER J + 'K' # 0xD2 -> LATIN CAPITAL LETTER K + 'L' # 0xD3 -> LATIN CAPITAL LETTER L + 'M' # 0xD4 -> LATIN CAPITAL LETTER M + 'N' # 0xD5 -> LATIN CAPITAL LETTER N + 'O' # 0xD6 -> LATIN CAPITAL LETTER O + 'P' # 0xD7 -> LATIN CAPITAL LETTER P + 'Q' # 0xD8 -> LATIN CAPITAL LETTER Q + 'R' # 0xD9 -> LATIN CAPITAL LETTER R + '\xb9' # 0xDA -> SUPERSCRIPT ONE + '\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xDC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE + '\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\\' # 0xE0 -> REVERSE SOLIDUS + '\xf7' # 0xE1 -> DIVISION SIGN + 'S' # 0xE2 -> LATIN CAPITAL LETTER S + 'T' # 0xE3 -> LATIN CAPITAL LETTER T + 'U' # 0xE4 -> LATIN CAPITAL LETTER U + 'V' # 0xE5 -> LATIN CAPITAL LETTER V + 'W' # 0xE6 -> LATIN CAPITAL LETTER W + 'X' # 0xE7 -> LATIN CAPITAL LETTER X + 'Y' # 0xE8 -> LATIN CAPITAL LETTER Y + 'Z' # 0xE9 -> LATIN CAPITAL LETTER Z + '\xb2' # 0xEA -> SUPERSCRIPT TWO + '\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd6' # 0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE + '0' # 0xF0 -> DIGIT ZERO + '1' # 0xF1 -> DIGIT ONE + '2' # 0xF2 -> DIGIT TWO + '3' # 0xF3 -> DIGIT THREE + '4' # 0xF4 -> DIGIT FOUR + '5' # 0xF5 -> DIGIT FIVE + '6' # 0xF6 -> DIGIT SIX + '7' # 0xF7 -> DIGIT SEVEN + '8' # 0xF8 -> DIGIT EIGHT + '9' # 0xF9 -> DIGIT NINE + '\xb3' # 0xFA -> SUPERSCRIPT THREE + '\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE + '\x9f' # 0xFF -> CONTROL ) ### Encoding table diff --git a/Lib/encodings/cp1006.py b/Lib/encodings/cp1006.py index e21e804..a1221c3 100644 --- a/Lib/encodings/cp1006.py +++ b/Lib/encodings/cp1006.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u06f0' # 0xA1 -> EXTENDED ARABIC-INDIC DIGIT ZERO - u'\u06f1' # 0xA2 -> EXTENDED ARABIC-INDIC DIGIT ONE - u'\u06f2' # 0xA3 -> EXTENDED ARABIC-INDIC DIGIT TWO - u'\u06f3' # 0xA4 -> EXTENDED ARABIC-INDIC DIGIT THREE - u'\u06f4' # 0xA5 -> EXTENDED ARABIC-INDIC DIGIT FOUR - u'\u06f5' # 0xA6 -> EXTENDED ARABIC-INDIC DIGIT FIVE - u'\u06f6' # 0xA7 -> EXTENDED ARABIC-INDIC DIGIT SIX - u'\u06f7' # 0xA8 -> EXTENDED ARABIC-INDIC DIGIT SEVEN - u'\u06f8' # 0xA9 -> EXTENDED ARABIC-INDIC DIGIT EIGHT - u'\u06f9' # 0xAA -> EXTENDED ARABIC-INDIC DIGIT NINE - u'\u060c' # 0xAB -> ARABIC COMMA - u'\u061b' # 0xAC -> ARABIC SEMICOLON - u'\xad' # 0xAD -> SOFT HYPHEN - u'\u061f' # 0xAE -> ARABIC QUESTION MARK - u'\ufe81' # 0xAF -> ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM - u'\ufe8d' # 0xB0 -> ARABIC LETTER ALEF ISOLATED FORM - u'\ufe8e' # 0xB1 -> ARABIC LETTER ALEF FINAL FORM - u'\ufe8e' # 0xB2 -> ARABIC LETTER ALEF FINAL FORM - u'\ufe8f' # 0xB3 -> ARABIC LETTER BEH ISOLATED FORM - u'\ufe91' # 0xB4 -> ARABIC LETTER BEH INITIAL FORM - u'\ufb56' # 0xB5 -> ARABIC LETTER PEH ISOLATED FORM - u'\ufb58' # 0xB6 -> ARABIC LETTER PEH INITIAL FORM - u'\ufe93' # 0xB7 -> ARABIC LETTER TEH MARBUTA ISOLATED FORM - u'\ufe95' # 0xB8 -> ARABIC LETTER TEH ISOLATED FORM - u'\ufe97' # 0xB9 -> ARABIC LETTER TEH INITIAL FORM - u'\ufb66' # 0xBA -> ARABIC LETTER TTEH ISOLATED FORM - u'\ufb68' # 0xBB -> ARABIC LETTER TTEH INITIAL FORM - u'\ufe99' # 0xBC -> ARABIC LETTER THEH ISOLATED FORM - u'\ufe9b' # 0xBD -> ARABIC LETTER THEH INITIAL FORM - u'\ufe9d' # 0xBE -> ARABIC LETTER JEEM ISOLATED FORM - u'\ufe9f' # 0xBF -> ARABIC LETTER JEEM INITIAL FORM - u'\ufb7a' # 0xC0 -> ARABIC LETTER TCHEH ISOLATED FORM - u'\ufb7c' # 0xC1 -> ARABIC LETTER TCHEH INITIAL FORM - u'\ufea1' # 0xC2 -> ARABIC LETTER HAH ISOLATED FORM - u'\ufea3' # 0xC3 -> ARABIC LETTER HAH INITIAL FORM - u'\ufea5' # 0xC4 -> ARABIC LETTER KHAH ISOLATED FORM - u'\ufea7' # 0xC5 -> ARABIC LETTER KHAH INITIAL FORM - u'\ufea9' # 0xC6 -> ARABIC LETTER DAL ISOLATED FORM - u'\ufb84' # 0xC7 -> ARABIC LETTER DAHAL ISOLATED FORMN - u'\ufeab' # 0xC8 -> ARABIC LETTER THAL ISOLATED FORM - u'\ufead' # 0xC9 -> ARABIC LETTER REH ISOLATED FORM - u'\ufb8c' # 0xCA -> ARABIC LETTER RREH ISOLATED FORM - u'\ufeaf' # 0xCB -> ARABIC LETTER ZAIN ISOLATED FORM - u'\ufb8a' # 0xCC -> ARABIC LETTER JEH ISOLATED FORM - u'\ufeb1' # 0xCD -> ARABIC LETTER SEEN ISOLATED FORM - u'\ufeb3' # 0xCE -> ARABIC LETTER SEEN INITIAL FORM - u'\ufeb5' # 0xCF -> ARABIC LETTER SHEEN ISOLATED FORM - u'\ufeb7' # 0xD0 -> ARABIC LETTER SHEEN INITIAL FORM - u'\ufeb9' # 0xD1 -> ARABIC LETTER SAD ISOLATED FORM - u'\ufebb' # 0xD2 -> ARABIC LETTER SAD INITIAL FORM - u'\ufebd' # 0xD3 -> ARABIC LETTER DAD ISOLATED FORM - u'\ufebf' # 0xD4 -> ARABIC LETTER DAD INITIAL FORM - u'\ufec1' # 0xD5 -> ARABIC LETTER TAH ISOLATED FORM - u'\ufec5' # 0xD6 -> ARABIC LETTER ZAH ISOLATED FORM - u'\ufec9' # 0xD7 -> ARABIC LETTER AIN ISOLATED FORM - u'\ufeca' # 0xD8 -> ARABIC LETTER AIN FINAL FORM - u'\ufecb' # 0xD9 -> ARABIC LETTER AIN INITIAL FORM - u'\ufecc' # 0xDA -> ARABIC LETTER AIN MEDIAL FORM - u'\ufecd' # 0xDB -> ARABIC LETTER GHAIN ISOLATED FORM - u'\ufece' # 0xDC -> ARABIC LETTER GHAIN FINAL FORM - u'\ufecf' # 0xDD -> ARABIC LETTER GHAIN INITIAL FORM - u'\ufed0' # 0xDE -> ARABIC LETTER GHAIN MEDIAL FORM - u'\ufed1' # 0xDF -> ARABIC LETTER FEH ISOLATED FORM - u'\ufed3' # 0xE0 -> ARABIC LETTER FEH INITIAL FORM - u'\ufed5' # 0xE1 -> ARABIC LETTER QAF ISOLATED FORM - u'\ufed7' # 0xE2 -> ARABIC LETTER QAF INITIAL FORM - u'\ufed9' # 0xE3 -> ARABIC LETTER KAF ISOLATED FORM - u'\ufedb' # 0xE4 -> ARABIC LETTER KAF INITIAL FORM - u'\ufb92' # 0xE5 -> ARABIC LETTER GAF ISOLATED FORM - u'\ufb94' # 0xE6 -> ARABIC LETTER GAF INITIAL FORM - u'\ufedd' # 0xE7 -> ARABIC LETTER LAM ISOLATED FORM - u'\ufedf' # 0xE8 -> ARABIC LETTER LAM INITIAL FORM - u'\ufee0' # 0xE9 -> ARABIC LETTER LAM MEDIAL FORM - u'\ufee1' # 0xEA -> ARABIC LETTER MEEM ISOLATED FORM - u'\ufee3' # 0xEB -> ARABIC LETTER MEEM INITIAL FORM - u'\ufb9e' # 0xEC -> ARABIC LETTER NOON GHUNNA ISOLATED FORM - u'\ufee5' # 0xED -> ARABIC LETTER NOON ISOLATED FORM - u'\ufee7' # 0xEE -> ARABIC LETTER NOON INITIAL FORM - u'\ufe85' # 0xEF -> ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM - u'\ufeed' # 0xF0 -> ARABIC LETTER WAW ISOLATED FORM - u'\ufba6' # 0xF1 -> ARABIC LETTER HEH GOAL ISOLATED FORM - u'\ufba8' # 0xF2 -> ARABIC LETTER HEH GOAL INITIAL FORM - u'\ufba9' # 0xF3 -> ARABIC LETTER HEH GOAL MEDIAL FORM - u'\ufbaa' # 0xF4 -> ARABIC LETTER HEH DOACHASHMEE ISOLATED FORM - u'\ufe80' # 0xF5 -> ARABIC LETTER HAMZA ISOLATED FORM - u'\ufe89' # 0xF6 -> ARABIC LETTER YEH WITH HAMZA ABOVE ISOLATED FORM - u'\ufe8a' # 0xF7 -> ARABIC LETTER YEH WITH HAMZA ABOVE FINAL FORM - u'\ufe8b' # 0xF8 -> ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM - u'\ufef1' # 0xF9 -> ARABIC LETTER YEH ISOLATED FORM - u'\ufef2' # 0xFA -> ARABIC LETTER YEH FINAL FORM - u'\ufef3' # 0xFB -> ARABIC LETTER YEH INITIAL FORM - u'\ufbb0' # 0xFC -> ARABIC LETTER YEH BARREE WITH HAMZA ABOVE ISOLATED FORM - u'\ufbae' # 0xFD -> ARABIC LETTER YEH BARREE ISOLATED FORM - u'\ufe7c' # 0xFE -> ARABIC SHADDA ISOLATED FORM - u'\ufe7d' # 0xFF -> ARABIC SHADDA MEDIAL FORM + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u06f0' # 0xA1 -> EXTENDED ARABIC-INDIC DIGIT ZERO + '\u06f1' # 0xA2 -> EXTENDED ARABIC-INDIC DIGIT ONE + '\u06f2' # 0xA3 -> EXTENDED ARABIC-INDIC DIGIT TWO + '\u06f3' # 0xA4 -> EXTENDED ARABIC-INDIC DIGIT THREE + '\u06f4' # 0xA5 -> EXTENDED ARABIC-INDIC DIGIT FOUR + '\u06f5' # 0xA6 -> EXTENDED ARABIC-INDIC DIGIT FIVE + '\u06f6' # 0xA7 -> EXTENDED ARABIC-INDIC DIGIT SIX + '\u06f7' # 0xA8 -> EXTENDED ARABIC-INDIC DIGIT SEVEN + '\u06f8' # 0xA9 -> EXTENDED ARABIC-INDIC DIGIT EIGHT + '\u06f9' # 0xAA -> EXTENDED ARABIC-INDIC DIGIT NINE + '\u060c' # 0xAB -> ARABIC COMMA + '\u061b' # 0xAC -> ARABIC SEMICOLON + '\xad' # 0xAD -> SOFT HYPHEN + '\u061f' # 0xAE -> ARABIC QUESTION MARK + '\ufe81' # 0xAF -> ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM + '\ufe8d' # 0xB0 -> ARABIC LETTER ALEF ISOLATED FORM + '\ufe8e' # 0xB1 -> ARABIC LETTER ALEF FINAL FORM + '\ufe8e' # 0xB2 -> ARABIC LETTER ALEF FINAL FORM + '\ufe8f' # 0xB3 -> ARABIC LETTER BEH ISOLATED FORM + '\ufe91' # 0xB4 -> ARABIC LETTER BEH INITIAL FORM + '\ufb56' # 0xB5 -> ARABIC LETTER PEH ISOLATED FORM + '\ufb58' # 0xB6 -> ARABIC LETTER PEH INITIAL FORM + '\ufe93' # 0xB7 -> ARABIC LETTER TEH MARBUTA ISOLATED FORM + '\ufe95' # 0xB8 -> ARABIC LETTER TEH ISOLATED FORM + '\ufe97' # 0xB9 -> ARABIC LETTER TEH INITIAL FORM + '\ufb66' # 0xBA -> ARABIC LETTER TTEH ISOLATED FORM + '\ufb68' # 0xBB -> ARABIC LETTER TTEH INITIAL FORM + '\ufe99' # 0xBC -> ARABIC LETTER THEH ISOLATED FORM + '\ufe9b' # 0xBD -> ARABIC LETTER THEH INITIAL FORM + '\ufe9d' # 0xBE -> ARABIC LETTER JEEM ISOLATED FORM + '\ufe9f' # 0xBF -> ARABIC LETTER JEEM INITIAL FORM + '\ufb7a' # 0xC0 -> ARABIC LETTER TCHEH ISOLATED FORM + '\ufb7c' # 0xC1 -> ARABIC LETTER TCHEH INITIAL FORM + '\ufea1' # 0xC2 -> ARABIC LETTER HAH ISOLATED FORM + '\ufea3' # 0xC3 -> ARABIC LETTER HAH INITIAL FORM + '\ufea5' # 0xC4 -> ARABIC LETTER KHAH ISOLATED FORM + '\ufea7' # 0xC5 -> ARABIC LETTER KHAH INITIAL FORM + '\ufea9' # 0xC6 -> ARABIC LETTER DAL ISOLATED FORM + '\ufb84' # 0xC7 -> ARABIC LETTER DAHAL ISOLATED FORMN + '\ufeab' # 0xC8 -> ARABIC LETTER THAL ISOLATED FORM + '\ufead' # 0xC9 -> ARABIC LETTER REH ISOLATED FORM + '\ufb8c' # 0xCA -> ARABIC LETTER RREH ISOLATED FORM + '\ufeaf' # 0xCB -> ARABIC LETTER ZAIN ISOLATED FORM + '\ufb8a' # 0xCC -> ARABIC LETTER JEH ISOLATED FORM + '\ufeb1' # 0xCD -> ARABIC LETTER SEEN ISOLATED FORM + '\ufeb3' # 0xCE -> ARABIC LETTER SEEN INITIAL FORM + '\ufeb5' # 0xCF -> ARABIC LETTER SHEEN ISOLATED FORM + '\ufeb7' # 0xD0 -> ARABIC LETTER SHEEN INITIAL FORM + '\ufeb9' # 0xD1 -> ARABIC LETTER SAD ISOLATED FORM + '\ufebb' # 0xD2 -> ARABIC LETTER SAD INITIAL FORM + '\ufebd' # 0xD3 -> ARABIC LETTER DAD ISOLATED FORM + '\ufebf' # 0xD4 -> ARABIC LETTER DAD INITIAL FORM + '\ufec1' # 0xD5 -> ARABIC LETTER TAH ISOLATED FORM + '\ufec5' # 0xD6 -> ARABIC LETTER ZAH ISOLATED FORM + '\ufec9' # 0xD7 -> ARABIC LETTER AIN ISOLATED FORM + '\ufeca' # 0xD8 -> ARABIC LETTER AIN FINAL FORM + '\ufecb' # 0xD9 -> ARABIC LETTER AIN INITIAL FORM + '\ufecc' # 0xDA -> ARABIC LETTER AIN MEDIAL FORM + '\ufecd' # 0xDB -> ARABIC LETTER GHAIN ISOLATED FORM + '\ufece' # 0xDC -> ARABIC LETTER GHAIN FINAL FORM + '\ufecf' # 0xDD -> ARABIC LETTER GHAIN INITIAL FORM + '\ufed0' # 0xDE -> ARABIC LETTER GHAIN MEDIAL FORM + '\ufed1' # 0xDF -> ARABIC LETTER FEH ISOLATED FORM + '\ufed3' # 0xE0 -> ARABIC LETTER FEH INITIAL FORM + '\ufed5' # 0xE1 -> ARABIC LETTER QAF ISOLATED FORM + '\ufed7' # 0xE2 -> ARABIC LETTER QAF INITIAL FORM + '\ufed9' # 0xE3 -> ARABIC LETTER KAF ISOLATED FORM + '\ufedb' # 0xE4 -> ARABIC LETTER KAF INITIAL FORM + '\ufb92' # 0xE5 -> ARABIC LETTER GAF ISOLATED FORM + '\ufb94' # 0xE6 -> ARABIC LETTER GAF INITIAL FORM + '\ufedd' # 0xE7 -> ARABIC LETTER LAM ISOLATED FORM + '\ufedf' # 0xE8 -> ARABIC LETTER LAM INITIAL FORM + '\ufee0' # 0xE9 -> ARABIC LETTER LAM MEDIAL FORM + '\ufee1' # 0xEA -> ARABIC LETTER MEEM ISOLATED FORM + '\ufee3' # 0xEB -> ARABIC LETTER MEEM INITIAL FORM + '\ufb9e' # 0xEC -> ARABIC LETTER NOON GHUNNA ISOLATED FORM + '\ufee5' # 0xED -> ARABIC LETTER NOON ISOLATED FORM + '\ufee7' # 0xEE -> ARABIC LETTER NOON INITIAL FORM + '\ufe85' # 0xEF -> ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM + '\ufeed' # 0xF0 -> ARABIC LETTER WAW ISOLATED FORM + '\ufba6' # 0xF1 -> ARABIC LETTER HEH GOAL ISOLATED FORM + '\ufba8' # 0xF2 -> ARABIC LETTER HEH GOAL INITIAL FORM + '\ufba9' # 0xF3 -> ARABIC LETTER HEH GOAL MEDIAL FORM + '\ufbaa' # 0xF4 -> ARABIC LETTER HEH DOACHASHMEE ISOLATED FORM + '\ufe80' # 0xF5 -> ARABIC LETTER HAMZA ISOLATED FORM + '\ufe89' # 0xF6 -> ARABIC LETTER YEH WITH HAMZA ABOVE ISOLATED FORM + '\ufe8a' # 0xF7 -> ARABIC LETTER YEH WITH HAMZA ABOVE FINAL FORM + '\ufe8b' # 0xF8 -> ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM + '\ufef1' # 0xF9 -> ARABIC LETTER YEH ISOLATED FORM + '\ufef2' # 0xFA -> ARABIC LETTER YEH FINAL FORM + '\ufef3' # 0xFB -> ARABIC LETTER YEH INITIAL FORM + '\ufbb0' # 0xFC -> ARABIC LETTER YEH BARREE WITH HAMZA ABOVE ISOLATED FORM + '\ufbae' # 0xFD -> ARABIC LETTER YEH BARREE ISOLATED FORM + '\ufe7c' # 0xFE -> ARABIC SHADDA ISOLATED FORM + '\ufe7d' # 0xFF -> ARABIC SHADDA MEDIAL FORM ) ### Encoding table diff --git a/Lib/encodings/cp1026.py b/Lib/encodings/cp1026.py index 45bbe62..46f71f7 100644 --- a/Lib/encodings/cp1026.py +++ b/Lib/encodings/cp1026.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x9c' # 0x04 -> CONTROL - u'\t' # 0x05 -> HORIZONTAL TABULATION - u'\x86' # 0x06 -> CONTROL - u'\x7f' # 0x07 -> DELETE - u'\x97' # 0x08 -> CONTROL - u'\x8d' # 0x09 -> CONTROL - u'\x8e' # 0x0A -> CONTROL - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x9d' # 0x14 -> CONTROL - u'\x85' # 0x15 -> CONTROL - u'\x08' # 0x16 -> BACKSPACE - u'\x87' # 0x17 -> CONTROL - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x92' # 0x1A -> CONTROL - u'\x8f' # 0x1B -> CONTROL - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u'\x80' # 0x20 -> CONTROL - u'\x81' # 0x21 -> CONTROL - u'\x82' # 0x22 -> CONTROL - u'\x83' # 0x23 -> CONTROL - u'\x84' # 0x24 -> CONTROL - u'\n' # 0x25 -> LINE FEED - u'\x17' # 0x26 -> END OF TRANSMISSION BLOCK - u'\x1b' # 0x27 -> ESCAPE - u'\x88' # 0x28 -> CONTROL - u'\x89' # 0x29 -> CONTROL - u'\x8a' # 0x2A -> CONTROL - u'\x8b' # 0x2B -> CONTROL - u'\x8c' # 0x2C -> CONTROL - u'\x05' # 0x2D -> ENQUIRY - u'\x06' # 0x2E -> ACKNOWLEDGE - u'\x07' # 0x2F -> BELL - u'\x90' # 0x30 -> CONTROL - u'\x91' # 0x31 -> CONTROL - u'\x16' # 0x32 -> SYNCHRONOUS IDLE - u'\x93' # 0x33 -> CONTROL - u'\x94' # 0x34 -> CONTROL - u'\x95' # 0x35 -> CONTROL - u'\x96' # 0x36 -> CONTROL - u'\x04' # 0x37 -> END OF TRANSMISSION - u'\x98' # 0x38 -> CONTROL - u'\x99' # 0x39 -> CONTROL - u'\x9a' # 0x3A -> CONTROL - u'\x9b' # 0x3B -> CONTROL - u'\x14' # 0x3C -> DEVICE CONTROL FOUR - u'\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE - u'\x9e' # 0x3E -> CONTROL - u'\x1a' # 0x3F -> SUBSTITUTE - u' ' # 0x40 -> SPACE - u'\xa0' # 0x41 -> NO-BREAK SPACE - u'\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE - u'{' # 0x48 -> LEFT CURLY BRACKET - u'\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE - u'\xc7' # 0x4A -> LATIN CAPITAL LETTER C WITH CEDILLA - u'.' # 0x4B -> FULL STOP - u'<' # 0x4C -> LESS-THAN SIGN - u'(' # 0x4D -> LEFT PARENTHESIS - u'+' # 0x4E -> PLUS SIGN - u'!' # 0x4F -> EXCLAMATION MARK - u'&' # 0x50 -> AMPERSAND - u'\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE - u'\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE - u'\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) - u'\u011e' # 0x5A -> LATIN CAPITAL LETTER G WITH BREVE - u'\u0130' # 0x5B -> LATIN CAPITAL LETTER I WITH DOT ABOVE - u'*' # 0x5C -> ASTERISK - u')' # 0x5D -> RIGHT PARENTHESIS - u';' # 0x5E -> SEMICOLON - u'^' # 0x5F -> CIRCUMFLEX ACCENT - u'-' # 0x60 -> HYPHEN-MINUS - u'/' # 0x61 -> SOLIDUS - u'\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'[' # 0x68 -> LEFT SQUARE BRACKET - u'\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE - u'\u015f' # 0x6A -> LATIN SMALL LETTER S WITH CEDILLA - u',' # 0x6B -> COMMA - u'%' # 0x6C -> PERCENT SIGN - u'_' # 0x6D -> LOW LINE - u'>' # 0x6E -> GREATER-THAN SIGN - u'?' # 0x6F -> QUESTION MARK - u'\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE - u'\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE - u'\u0131' # 0x79 -> LATIN SMALL LETTER DOTLESS I - u':' # 0x7A -> COLON - u'\xd6' # 0x7B -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\u015e' # 0x7C -> LATIN CAPITAL LETTER S WITH CEDILLA - u"'" # 0x7D -> APOSTROPHE - u'=' # 0x7E -> EQUALS SIGN - u'\xdc' # 0x7F -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE - u'a' # 0x81 -> LATIN SMALL LETTER A - u'b' # 0x82 -> LATIN SMALL LETTER B - u'c' # 0x83 -> LATIN SMALL LETTER C - u'd' # 0x84 -> LATIN SMALL LETTER D - u'e' # 0x85 -> LATIN SMALL LETTER E - u'f' # 0x86 -> LATIN SMALL LETTER F - u'g' # 0x87 -> LATIN SMALL LETTER G - u'h' # 0x88 -> LATIN SMALL LETTER H - u'i' # 0x89 -> LATIN SMALL LETTER I - u'\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'}' # 0x8C -> RIGHT CURLY BRACKET - u'`' # 0x8D -> GRAVE ACCENT - u'\xa6' # 0x8E -> BROKEN BAR - u'\xb1' # 0x8F -> PLUS-MINUS SIGN - u'\xb0' # 0x90 -> DEGREE SIGN - u'j' # 0x91 -> LATIN SMALL LETTER J - u'k' # 0x92 -> LATIN SMALL LETTER K - u'l' # 0x93 -> LATIN SMALL LETTER L - u'm' # 0x94 -> LATIN SMALL LETTER M - u'n' # 0x95 -> LATIN SMALL LETTER N - u'o' # 0x96 -> LATIN SMALL LETTER O - u'p' # 0x97 -> LATIN SMALL LETTER P - u'q' # 0x98 -> LATIN SMALL LETTER Q - u'r' # 0x99 -> LATIN SMALL LETTER R - u'\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR - u'\xe6' # 0x9C -> LATIN SMALL LIGATURE AE - u'\xb8' # 0x9D -> CEDILLA - u'\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE - u'\xa4' # 0x9F -> CURRENCY SIGN - u'\xb5' # 0xA0 -> MICRO SIGN - u'\xf6' # 0xA1 -> LATIN SMALL LETTER O WITH DIAERESIS - u's' # 0xA2 -> LATIN SMALL LETTER S - u't' # 0xA3 -> LATIN SMALL LETTER T - u'u' # 0xA4 -> LATIN SMALL LETTER U - u'v' # 0xA5 -> LATIN SMALL LETTER V - u'w' # 0xA6 -> LATIN SMALL LETTER W - u'x' # 0xA7 -> LATIN SMALL LETTER X - u'y' # 0xA8 -> LATIN SMALL LETTER Y - u'z' # 0xA9 -> LATIN SMALL LETTER Z - u'\xa1' # 0xAA -> INVERTED EXCLAMATION MARK - u'\xbf' # 0xAB -> INVERTED QUESTION MARK - u']' # 0xAC -> RIGHT SQUARE BRACKET - u'$' # 0xAD -> DOLLAR SIGN - u'@' # 0xAE -> COMMERCIAL AT - u'\xae' # 0xAF -> REGISTERED SIGN - u'\xa2' # 0xB0 -> CENT SIGN - u'\xa3' # 0xB1 -> POUND SIGN - u'\xa5' # 0xB2 -> YEN SIGN - u'\xb7' # 0xB3 -> MIDDLE DOT - u'\xa9' # 0xB4 -> COPYRIGHT SIGN - u'\xa7' # 0xB5 -> SECTION SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS - u'\xac' # 0xBA -> NOT SIGN - u'|' # 0xBB -> VERTICAL LINE - u'\xaf' # 0xBC -> MACRON - u'\xa8' # 0xBD -> DIAERESIS - u'\xb4' # 0xBE -> ACUTE ACCENT - u'\xd7' # 0xBF -> MULTIPLICATION SIGN - u'\xe7' # 0xC0 -> LATIN SMALL LETTER C WITH CEDILLA - u'A' # 0xC1 -> LATIN CAPITAL LETTER A - u'B' # 0xC2 -> LATIN CAPITAL LETTER B - u'C' # 0xC3 -> LATIN CAPITAL LETTER C - u'D' # 0xC4 -> LATIN CAPITAL LETTER D - u'E' # 0xC5 -> LATIN CAPITAL LETTER E - u'F' # 0xC6 -> LATIN CAPITAL LETTER F - u'G' # 0xC7 -> LATIN CAPITAL LETTER G - u'H' # 0xC8 -> LATIN CAPITAL LETTER H - u'I' # 0xC9 -> LATIN CAPITAL LETTER I - u'\xad' # 0xCA -> SOFT HYPHEN - u'\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'~' # 0xCC -> TILDE - u'\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE - u'\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE - u'\u011f' # 0xD0 -> LATIN SMALL LETTER G WITH BREVE - u'J' # 0xD1 -> LATIN CAPITAL LETTER J - u'K' # 0xD2 -> LATIN CAPITAL LETTER K - u'L' # 0xD3 -> LATIN CAPITAL LETTER L - u'M' # 0xD4 -> LATIN CAPITAL LETTER M - u'N' # 0xD5 -> LATIN CAPITAL LETTER N - u'O' # 0xD6 -> LATIN CAPITAL LETTER O - u'P' # 0xD7 -> LATIN CAPITAL LETTER P - u'Q' # 0xD8 -> LATIN CAPITAL LETTER Q - u'R' # 0xD9 -> LATIN CAPITAL LETTER R - u'\xb9' # 0xDA -> SUPERSCRIPT ONE - u'\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\\' # 0xDC -> REVERSE SOLIDUS - u'\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE - u'\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\xfc' # 0xE0 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xf7' # 0xE1 -> DIVISION SIGN - u'S' # 0xE2 -> LATIN CAPITAL LETTER S - u'T' # 0xE3 -> LATIN CAPITAL LETTER T - u'U' # 0xE4 -> LATIN CAPITAL LETTER U - u'V' # 0xE5 -> LATIN CAPITAL LETTER V - u'W' # 0xE6 -> LATIN CAPITAL LETTER W - u'X' # 0xE7 -> LATIN CAPITAL LETTER X - u'Y' # 0xE8 -> LATIN CAPITAL LETTER Y - u'Z' # 0xE9 -> LATIN CAPITAL LETTER Z - u'\xb2' # 0xEA -> SUPERSCRIPT TWO - u'\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'#' # 0xEC -> NUMBER SIGN - u'\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE - u'0' # 0xF0 -> DIGIT ZERO - u'1' # 0xF1 -> DIGIT ONE - u'2' # 0xF2 -> DIGIT TWO - u'3' # 0xF3 -> DIGIT THREE - u'4' # 0xF4 -> DIGIT FOUR - u'5' # 0xF5 -> DIGIT FIVE - u'6' # 0xF6 -> DIGIT SIX - u'7' # 0xF7 -> DIGIT SEVEN - u'8' # 0xF8 -> DIGIT EIGHT - u'9' # 0xF9 -> DIGIT NINE - u'\xb3' # 0xFA -> SUPERSCRIPT THREE - u'\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'"' # 0xFC -> QUOTATION MARK - u'\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE - u'\x9f' # 0xFF -> CONTROL + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x9c' # 0x04 -> CONTROL + '\t' # 0x05 -> HORIZONTAL TABULATION + '\x86' # 0x06 -> CONTROL + '\x7f' # 0x07 -> DELETE + '\x97' # 0x08 -> CONTROL + '\x8d' # 0x09 -> CONTROL + '\x8e' # 0x0A -> CONTROL + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x9d' # 0x14 -> CONTROL + '\x85' # 0x15 -> CONTROL + '\x08' # 0x16 -> BACKSPACE + '\x87' # 0x17 -> CONTROL + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x92' # 0x1A -> CONTROL + '\x8f' # 0x1B -> CONTROL + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + '\x80' # 0x20 -> CONTROL + '\x81' # 0x21 -> CONTROL + '\x82' # 0x22 -> CONTROL + '\x83' # 0x23 -> CONTROL + '\x84' # 0x24 -> CONTROL + '\n' # 0x25 -> LINE FEED + '\x17' # 0x26 -> END OF TRANSMISSION BLOCK + '\x1b' # 0x27 -> ESCAPE + '\x88' # 0x28 -> CONTROL + '\x89' # 0x29 -> CONTROL + '\x8a' # 0x2A -> CONTROL + '\x8b' # 0x2B -> CONTROL + '\x8c' # 0x2C -> CONTROL + '\x05' # 0x2D -> ENQUIRY + '\x06' # 0x2E -> ACKNOWLEDGE + '\x07' # 0x2F -> BELL + '\x90' # 0x30 -> CONTROL + '\x91' # 0x31 -> CONTROL + '\x16' # 0x32 -> SYNCHRONOUS IDLE + '\x93' # 0x33 -> CONTROL + '\x94' # 0x34 -> CONTROL + '\x95' # 0x35 -> CONTROL + '\x96' # 0x36 -> CONTROL + '\x04' # 0x37 -> END OF TRANSMISSION + '\x98' # 0x38 -> CONTROL + '\x99' # 0x39 -> CONTROL + '\x9a' # 0x3A -> CONTROL + '\x9b' # 0x3B -> CONTROL + '\x14' # 0x3C -> DEVICE CONTROL FOUR + '\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE + '\x9e' # 0x3E -> CONTROL + '\x1a' # 0x3F -> SUBSTITUTE + ' ' # 0x40 -> SPACE + '\xa0' # 0x41 -> NO-BREAK SPACE + '\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE + '\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE + '{' # 0x48 -> LEFT CURLY BRACKET + '\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE + '\xc7' # 0x4A -> LATIN CAPITAL LETTER C WITH CEDILLA + '.' # 0x4B -> FULL STOP + '<' # 0x4C -> LESS-THAN SIGN + '(' # 0x4D -> LEFT PARENTHESIS + '+' # 0x4E -> PLUS SIGN + '!' # 0x4F -> EXCLAMATION MARK + '&' # 0x50 -> AMPERSAND + '\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE + '\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE + '\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) + '\u011e' # 0x5A -> LATIN CAPITAL LETTER G WITH BREVE + '\u0130' # 0x5B -> LATIN CAPITAL LETTER I WITH DOT ABOVE + '*' # 0x5C -> ASTERISK + ')' # 0x5D -> RIGHT PARENTHESIS + ';' # 0x5E -> SEMICOLON + '^' # 0x5F -> CIRCUMFLEX ACCENT + '-' # 0x60 -> HYPHEN-MINUS + '/' # 0x61 -> SOLIDUS + '\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '[' # 0x68 -> LEFT SQUARE BRACKET + '\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE + '\u015f' # 0x6A -> LATIN SMALL LETTER S WITH CEDILLA + ',' # 0x6B -> COMMA + '%' # 0x6C -> PERCENT SIGN + '_' # 0x6D -> LOW LINE + '>' # 0x6E -> GREATER-THAN SIGN + '?' # 0x6F -> QUESTION MARK + '\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE + '\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE + '\u0131' # 0x79 -> LATIN SMALL LETTER DOTLESS I + ':' # 0x7A -> COLON + '\xd6' # 0x7B -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\u015e' # 0x7C -> LATIN CAPITAL LETTER S WITH CEDILLA + "'" # 0x7D -> APOSTROPHE + '=' # 0x7E -> EQUALS SIGN + '\xdc' # 0x7F -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE + 'a' # 0x81 -> LATIN SMALL LETTER A + 'b' # 0x82 -> LATIN SMALL LETTER B + 'c' # 0x83 -> LATIN SMALL LETTER C + 'd' # 0x84 -> LATIN SMALL LETTER D + 'e' # 0x85 -> LATIN SMALL LETTER E + 'f' # 0x86 -> LATIN SMALL LETTER F + 'g' # 0x87 -> LATIN SMALL LETTER G + 'h' # 0x88 -> LATIN SMALL LETTER H + 'i' # 0x89 -> LATIN SMALL LETTER I + '\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '}' # 0x8C -> RIGHT CURLY BRACKET + '`' # 0x8D -> GRAVE ACCENT + '\xa6' # 0x8E -> BROKEN BAR + '\xb1' # 0x8F -> PLUS-MINUS SIGN + '\xb0' # 0x90 -> DEGREE SIGN + 'j' # 0x91 -> LATIN SMALL LETTER J + 'k' # 0x92 -> LATIN SMALL LETTER K + 'l' # 0x93 -> LATIN SMALL LETTER L + 'm' # 0x94 -> LATIN SMALL LETTER M + 'n' # 0x95 -> LATIN SMALL LETTER N + 'o' # 0x96 -> LATIN SMALL LETTER O + 'p' # 0x97 -> LATIN SMALL LETTER P + 'q' # 0x98 -> LATIN SMALL LETTER Q + 'r' # 0x99 -> LATIN SMALL LETTER R + '\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR + '\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR + '\xe6' # 0x9C -> LATIN SMALL LIGATURE AE + '\xb8' # 0x9D -> CEDILLA + '\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE + '\xa4' # 0x9F -> CURRENCY SIGN + '\xb5' # 0xA0 -> MICRO SIGN + '\xf6' # 0xA1 -> LATIN SMALL LETTER O WITH DIAERESIS + 's' # 0xA2 -> LATIN SMALL LETTER S + 't' # 0xA3 -> LATIN SMALL LETTER T + 'u' # 0xA4 -> LATIN SMALL LETTER U + 'v' # 0xA5 -> LATIN SMALL LETTER V + 'w' # 0xA6 -> LATIN SMALL LETTER W + 'x' # 0xA7 -> LATIN SMALL LETTER X + 'y' # 0xA8 -> LATIN SMALL LETTER Y + 'z' # 0xA9 -> LATIN SMALL LETTER Z + '\xa1' # 0xAA -> INVERTED EXCLAMATION MARK + '\xbf' # 0xAB -> INVERTED QUESTION MARK + ']' # 0xAC -> RIGHT SQUARE BRACKET + '$' # 0xAD -> DOLLAR SIGN + '@' # 0xAE -> COMMERCIAL AT + '\xae' # 0xAF -> REGISTERED SIGN + '\xa2' # 0xB0 -> CENT SIGN + '\xa3' # 0xB1 -> POUND SIGN + '\xa5' # 0xB2 -> YEN SIGN + '\xb7' # 0xB3 -> MIDDLE DOT + '\xa9' # 0xB4 -> COPYRIGHT SIGN + '\xa7' # 0xB5 -> SECTION SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF + '\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS + '\xac' # 0xBA -> NOT SIGN + '|' # 0xBB -> VERTICAL LINE + '\xaf' # 0xBC -> MACRON + '\xa8' # 0xBD -> DIAERESIS + '\xb4' # 0xBE -> ACUTE ACCENT + '\xd7' # 0xBF -> MULTIPLICATION SIGN + '\xe7' # 0xC0 -> LATIN SMALL LETTER C WITH CEDILLA + 'A' # 0xC1 -> LATIN CAPITAL LETTER A + 'B' # 0xC2 -> LATIN CAPITAL LETTER B + 'C' # 0xC3 -> LATIN CAPITAL LETTER C + 'D' # 0xC4 -> LATIN CAPITAL LETTER D + 'E' # 0xC5 -> LATIN CAPITAL LETTER E + 'F' # 0xC6 -> LATIN CAPITAL LETTER F + 'G' # 0xC7 -> LATIN CAPITAL LETTER G + 'H' # 0xC8 -> LATIN CAPITAL LETTER H + 'I' # 0xC9 -> LATIN CAPITAL LETTER I + '\xad' # 0xCA -> SOFT HYPHEN + '\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '~' # 0xCC -> TILDE + '\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE + '\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE + '\u011f' # 0xD0 -> LATIN SMALL LETTER G WITH BREVE + 'J' # 0xD1 -> LATIN CAPITAL LETTER J + 'K' # 0xD2 -> LATIN CAPITAL LETTER K + 'L' # 0xD3 -> LATIN CAPITAL LETTER L + 'M' # 0xD4 -> LATIN CAPITAL LETTER M + 'N' # 0xD5 -> LATIN CAPITAL LETTER N + 'O' # 0xD6 -> LATIN CAPITAL LETTER O + 'P' # 0xD7 -> LATIN CAPITAL LETTER P + 'Q' # 0xD8 -> LATIN CAPITAL LETTER Q + 'R' # 0xD9 -> LATIN CAPITAL LETTER R + '\xb9' # 0xDA -> SUPERSCRIPT ONE + '\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\\' # 0xDC -> REVERSE SOLIDUS + '\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE + '\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\xfc' # 0xE0 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xf7' # 0xE1 -> DIVISION SIGN + 'S' # 0xE2 -> LATIN CAPITAL LETTER S + 'T' # 0xE3 -> LATIN CAPITAL LETTER T + 'U' # 0xE4 -> LATIN CAPITAL LETTER U + 'V' # 0xE5 -> LATIN CAPITAL LETTER V + 'W' # 0xE6 -> LATIN CAPITAL LETTER W + 'X' # 0xE7 -> LATIN CAPITAL LETTER X + 'Y' # 0xE8 -> LATIN CAPITAL LETTER Y + 'Z' # 0xE9 -> LATIN CAPITAL LETTER Z + '\xb2' # 0xEA -> SUPERSCRIPT TWO + '\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '#' # 0xEC -> NUMBER SIGN + '\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE + '0' # 0xF0 -> DIGIT ZERO + '1' # 0xF1 -> DIGIT ONE + '2' # 0xF2 -> DIGIT TWO + '3' # 0xF3 -> DIGIT THREE + '4' # 0xF4 -> DIGIT FOUR + '5' # 0xF5 -> DIGIT FIVE + '6' # 0xF6 -> DIGIT SIX + '7' # 0xF7 -> DIGIT SEVEN + '8' # 0xF8 -> DIGIT EIGHT + '9' # 0xF9 -> DIGIT NINE + '\xb3' # 0xFA -> SUPERSCRIPT THREE + '\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '"' # 0xFC -> QUOTATION MARK + '\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE + '\x9f' # 0xFF -> CONTROL ) ### Encoding table diff --git a/Lib/encodings/cp1140.py b/Lib/encodings/cp1140.py index 7e507fd..0a919d8 100644 --- a/Lib/encodings/cp1140.py +++ b/Lib/encodings/cp1140.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x9c' # 0x04 -> CONTROL - u'\t' # 0x05 -> HORIZONTAL TABULATION - u'\x86' # 0x06 -> CONTROL - u'\x7f' # 0x07 -> DELETE - u'\x97' # 0x08 -> CONTROL - u'\x8d' # 0x09 -> CONTROL - u'\x8e' # 0x0A -> CONTROL - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x9d' # 0x14 -> CONTROL - u'\x85' # 0x15 -> CONTROL - u'\x08' # 0x16 -> BACKSPACE - u'\x87' # 0x17 -> CONTROL - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x92' # 0x1A -> CONTROL - u'\x8f' # 0x1B -> CONTROL - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u'\x80' # 0x20 -> CONTROL - u'\x81' # 0x21 -> CONTROL - u'\x82' # 0x22 -> CONTROL - u'\x83' # 0x23 -> CONTROL - u'\x84' # 0x24 -> CONTROL - u'\n' # 0x25 -> LINE FEED - u'\x17' # 0x26 -> END OF TRANSMISSION BLOCK - u'\x1b' # 0x27 -> ESCAPE - u'\x88' # 0x28 -> CONTROL - u'\x89' # 0x29 -> CONTROL - u'\x8a' # 0x2A -> CONTROL - u'\x8b' # 0x2B -> CONTROL - u'\x8c' # 0x2C -> CONTROL - u'\x05' # 0x2D -> ENQUIRY - u'\x06' # 0x2E -> ACKNOWLEDGE - u'\x07' # 0x2F -> BELL - u'\x90' # 0x30 -> CONTROL - u'\x91' # 0x31 -> CONTROL - u'\x16' # 0x32 -> SYNCHRONOUS IDLE - u'\x93' # 0x33 -> CONTROL - u'\x94' # 0x34 -> CONTROL - u'\x95' # 0x35 -> CONTROL - u'\x96' # 0x36 -> CONTROL - u'\x04' # 0x37 -> END OF TRANSMISSION - u'\x98' # 0x38 -> CONTROL - u'\x99' # 0x39 -> CONTROL - u'\x9a' # 0x3A -> CONTROL - u'\x9b' # 0x3B -> CONTROL - u'\x14' # 0x3C -> DEVICE CONTROL FOUR - u'\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE - u'\x9e' # 0x3E -> CONTROL - u'\x1a' # 0x3F -> SUBSTITUTE - u' ' # 0x40 -> SPACE - u'\xa0' # 0x41 -> NO-BREAK SPACE - u'\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x48 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE - u'\xa2' # 0x4A -> CENT SIGN - u'.' # 0x4B -> FULL STOP - u'<' # 0x4C -> LESS-THAN SIGN - u'(' # 0x4D -> LEFT PARENTHESIS - u'+' # 0x4E -> PLUS SIGN - u'|' # 0x4F -> VERTICAL LINE - u'&' # 0x50 -> AMPERSAND - u'\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE - u'\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE - u'\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) - u'!' # 0x5A -> EXCLAMATION MARK - u'$' # 0x5B -> DOLLAR SIGN - u'*' # 0x5C -> ASTERISK - u')' # 0x5D -> RIGHT PARENTHESIS - u';' # 0x5E -> SEMICOLON - u'\xac' # 0x5F -> NOT SIGN - u'-' # 0x60 -> HYPHEN-MINUS - u'/' # 0x61 -> SOLIDUS - u'\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xa6' # 0x6A -> BROKEN BAR - u',' # 0x6B -> COMMA - u'%' # 0x6C -> PERCENT SIGN - u'_' # 0x6D -> LOW LINE - u'>' # 0x6E -> GREATER-THAN SIGN - u'?' # 0x6F -> QUESTION MARK - u'\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE - u'\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE - u'`' # 0x79 -> GRAVE ACCENT - u':' # 0x7A -> COLON - u'#' # 0x7B -> NUMBER SIGN - u'@' # 0x7C -> COMMERCIAL AT - u"'" # 0x7D -> APOSTROPHE - u'=' # 0x7E -> EQUALS SIGN - u'"' # 0x7F -> QUOTATION MARK - u'\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE - u'a' # 0x81 -> LATIN SMALL LETTER A - u'b' # 0x82 -> LATIN SMALL LETTER B - u'c' # 0x83 -> LATIN SMALL LETTER C - u'd' # 0x84 -> LATIN SMALL LETTER D - u'e' # 0x85 -> LATIN SMALL LETTER E - u'f' # 0x86 -> LATIN SMALL LETTER F - u'g' # 0x87 -> LATIN SMALL LETTER G - u'h' # 0x88 -> LATIN SMALL LETTER H - u'i' # 0x89 -> LATIN SMALL LETTER I - u'\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xf0' # 0x8C -> LATIN SMALL LETTER ETH (ICELANDIC) - u'\xfd' # 0x8D -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0x8E -> LATIN SMALL LETTER THORN (ICELANDIC) - u'\xb1' # 0x8F -> PLUS-MINUS SIGN - u'\xb0' # 0x90 -> DEGREE SIGN - u'j' # 0x91 -> LATIN SMALL LETTER J - u'k' # 0x92 -> LATIN SMALL LETTER K - u'l' # 0x93 -> LATIN SMALL LETTER L - u'm' # 0x94 -> LATIN SMALL LETTER M - u'n' # 0x95 -> LATIN SMALL LETTER N - u'o' # 0x96 -> LATIN SMALL LETTER O - u'p' # 0x97 -> LATIN SMALL LETTER P - u'q' # 0x98 -> LATIN SMALL LETTER Q - u'r' # 0x99 -> LATIN SMALL LETTER R - u'\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR - u'\xe6' # 0x9C -> LATIN SMALL LIGATURE AE - u'\xb8' # 0x9D -> CEDILLA - u'\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE - u'\u20ac' # 0x9F -> EURO SIGN - u'\xb5' # 0xA0 -> MICRO SIGN - u'~' # 0xA1 -> TILDE - u's' # 0xA2 -> LATIN SMALL LETTER S - u't' # 0xA3 -> LATIN SMALL LETTER T - u'u' # 0xA4 -> LATIN SMALL LETTER U - u'v' # 0xA5 -> LATIN SMALL LETTER V - u'w' # 0xA6 -> LATIN SMALL LETTER W - u'x' # 0xA7 -> LATIN SMALL LETTER X - u'y' # 0xA8 -> LATIN SMALL LETTER Y - u'z' # 0xA9 -> LATIN SMALL LETTER Z - u'\xa1' # 0xAA -> INVERTED EXCLAMATION MARK - u'\xbf' # 0xAB -> INVERTED QUESTION MARK - u'\xd0' # 0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC) - u'\xdd' # 0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC) - u'\xae' # 0xAF -> REGISTERED SIGN - u'^' # 0xB0 -> CIRCUMFLEX ACCENT - u'\xa3' # 0xB1 -> POUND SIGN - u'\xa5' # 0xB2 -> YEN SIGN - u'\xb7' # 0xB3 -> MIDDLE DOT - u'\xa9' # 0xB4 -> COPYRIGHT SIGN - u'\xa7' # 0xB5 -> SECTION SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS - u'[' # 0xBA -> LEFT SQUARE BRACKET - u']' # 0xBB -> RIGHT SQUARE BRACKET - u'\xaf' # 0xBC -> MACRON - u'\xa8' # 0xBD -> DIAERESIS - u'\xb4' # 0xBE -> ACUTE ACCENT - u'\xd7' # 0xBF -> MULTIPLICATION SIGN - u'{' # 0xC0 -> LEFT CURLY BRACKET - u'A' # 0xC1 -> LATIN CAPITAL LETTER A - u'B' # 0xC2 -> LATIN CAPITAL LETTER B - u'C' # 0xC3 -> LATIN CAPITAL LETTER C - u'D' # 0xC4 -> LATIN CAPITAL LETTER D - u'E' # 0xC5 -> LATIN CAPITAL LETTER E - u'F' # 0xC6 -> LATIN CAPITAL LETTER F - u'G' # 0xC7 -> LATIN CAPITAL LETTER G - u'H' # 0xC8 -> LATIN CAPITAL LETTER H - u'I' # 0xC9 -> LATIN CAPITAL LETTER I - u'\xad' # 0xCA -> SOFT HYPHEN - u'\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0xCC -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE - u'\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE - u'}' # 0xD0 -> RIGHT CURLY BRACKET - u'J' # 0xD1 -> LATIN CAPITAL LETTER J - u'K' # 0xD2 -> LATIN CAPITAL LETTER K - u'L' # 0xD3 -> LATIN CAPITAL LETTER L - u'M' # 0xD4 -> LATIN CAPITAL LETTER M - u'N' # 0xD5 -> LATIN CAPITAL LETTER N - u'O' # 0xD6 -> LATIN CAPITAL LETTER O - u'P' # 0xD7 -> LATIN CAPITAL LETTER P - u'Q' # 0xD8 -> LATIN CAPITAL LETTER Q - u'R' # 0xD9 -> LATIN CAPITAL LETTER R - u'\xb9' # 0xDA -> SUPERSCRIPT ONE - u'\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xDC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE - u'\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\\' # 0xE0 -> REVERSE SOLIDUS - u'\xf7' # 0xE1 -> DIVISION SIGN - u'S' # 0xE2 -> LATIN CAPITAL LETTER S - u'T' # 0xE3 -> LATIN CAPITAL LETTER T - u'U' # 0xE4 -> LATIN CAPITAL LETTER U - u'V' # 0xE5 -> LATIN CAPITAL LETTER V - u'W' # 0xE6 -> LATIN CAPITAL LETTER W - u'X' # 0xE7 -> LATIN CAPITAL LETTER X - u'Y' # 0xE8 -> LATIN CAPITAL LETTER Y - u'Z' # 0xE9 -> LATIN CAPITAL LETTER Z - u'\xb2' # 0xEA -> SUPERSCRIPT TWO - u'\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd6' # 0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE - u'0' # 0xF0 -> DIGIT ZERO - u'1' # 0xF1 -> DIGIT ONE - u'2' # 0xF2 -> DIGIT TWO - u'3' # 0xF3 -> DIGIT THREE - u'4' # 0xF4 -> DIGIT FOUR - u'5' # 0xF5 -> DIGIT FIVE - u'6' # 0xF6 -> DIGIT SIX - u'7' # 0xF7 -> DIGIT SEVEN - u'8' # 0xF8 -> DIGIT EIGHT - u'9' # 0xF9 -> DIGIT NINE - u'\xb3' # 0xFA -> SUPERSCRIPT THREE - u'\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE - u'\x9f' # 0xFF -> CONTROL + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x9c' # 0x04 -> CONTROL + '\t' # 0x05 -> HORIZONTAL TABULATION + '\x86' # 0x06 -> CONTROL + '\x7f' # 0x07 -> DELETE + '\x97' # 0x08 -> CONTROL + '\x8d' # 0x09 -> CONTROL + '\x8e' # 0x0A -> CONTROL + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x9d' # 0x14 -> CONTROL + '\x85' # 0x15 -> CONTROL + '\x08' # 0x16 -> BACKSPACE + '\x87' # 0x17 -> CONTROL + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x92' # 0x1A -> CONTROL + '\x8f' # 0x1B -> CONTROL + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + '\x80' # 0x20 -> CONTROL + '\x81' # 0x21 -> CONTROL + '\x82' # 0x22 -> CONTROL + '\x83' # 0x23 -> CONTROL + '\x84' # 0x24 -> CONTROL + '\n' # 0x25 -> LINE FEED + '\x17' # 0x26 -> END OF TRANSMISSION BLOCK + '\x1b' # 0x27 -> ESCAPE + '\x88' # 0x28 -> CONTROL + '\x89' # 0x29 -> CONTROL + '\x8a' # 0x2A -> CONTROL + '\x8b' # 0x2B -> CONTROL + '\x8c' # 0x2C -> CONTROL + '\x05' # 0x2D -> ENQUIRY + '\x06' # 0x2E -> ACKNOWLEDGE + '\x07' # 0x2F -> BELL + '\x90' # 0x30 -> CONTROL + '\x91' # 0x31 -> CONTROL + '\x16' # 0x32 -> SYNCHRONOUS IDLE + '\x93' # 0x33 -> CONTROL + '\x94' # 0x34 -> CONTROL + '\x95' # 0x35 -> CONTROL + '\x96' # 0x36 -> CONTROL + '\x04' # 0x37 -> END OF TRANSMISSION + '\x98' # 0x38 -> CONTROL + '\x99' # 0x39 -> CONTROL + '\x9a' # 0x3A -> CONTROL + '\x9b' # 0x3B -> CONTROL + '\x14' # 0x3C -> DEVICE CONTROL FOUR + '\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE + '\x9e' # 0x3E -> CONTROL + '\x1a' # 0x3F -> SUBSTITUTE + ' ' # 0x40 -> SPACE + '\xa0' # 0x41 -> NO-BREAK SPACE + '\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE + '\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x48 -> LATIN SMALL LETTER C WITH CEDILLA + '\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE + '\xa2' # 0x4A -> CENT SIGN + '.' # 0x4B -> FULL STOP + '<' # 0x4C -> LESS-THAN SIGN + '(' # 0x4D -> LEFT PARENTHESIS + '+' # 0x4E -> PLUS SIGN + '|' # 0x4F -> VERTICAL LINE + '&' # 0x50 -> AMPERSAND + '\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE + '\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE + '\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) + '!' # 0x5A -> EXCLAMATION MARK + '$' # 0x5B -> DOLLAR SIGN + '*' # 0x5C -> ASTERISK + ')' # 0x5D -> RIGHT PARENTHESIS + ';' # 0x5E -> SEMICOLON + '\xac' # 0x5F -> NOT SIGN + '-' # 0x60 -> HYPHEN-MINUS + '/' # 0x61 -> SOLIDUS + '\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE + '\xa6' # 0x6A -> BROKEN BAR + ',' # 0x6B -> COMMA + '%' # 0x6C -> PERCENT SIGN + '_' # 0x6D -> LOW LINE + '>' # 0x6E -> GREATER-THAN SIGN + '?' # 0x6F -> QUESTION MARK + '\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE + '\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE + '`' # 0x79 -> GRAVE ACCENT + ':' # 0x7A -> COLON + '#' # 0x7B -> NUMBER SIGN + '@' # 0x7C -> COMMERCIAL AT + "'" # 0x7D -> APOSTROPHE + '=' # 0x7E -> EQUALS SIGN + '"' # 0x7F -> QUOTATION MARK + '\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE + 'a' # 0x81 -> LATIN SMALL LETTER A + 'b' # 0x82 -> LATIN SMALL LETTER B + 'c' # 0x83 -> LATIN SMALL LETTER C + 'd' # 0x84 -> LATIN SMALL LETTER D + 'e' # 0x85 -> LATIN SMALL LETTER E + 'f' # 0x86 -> LATIN SMALL LETTER F + 'g' # 0x87 -> LATIN SMALL LETTER G + 'h' # 0x88 -> LATIN SMALL LETTER H + 'i' # 0x89 -> LATIN SMALL LETTER I + '\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xf0' # 0x8C -> LATIN SMALL LETTER ETH (ICELANDIC) + '\xfd' # 0x8D -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0x8E -> LATIN SMALL LETTER THORN (ICELANDIC) + '\xb1' # 0x8F -> PLUS-MINUS SIGN + '\xb0' # 0x90 -> DEGREE SIGN + 'j' # 0x91 -> LATIN SMALL LETTER J + 'k' # 0x92 -> LATIN SMALL LETTER K + 'l' # 0x93 -> LATIN SMALL LETTER L + 'm' # 0x94 -> LATIN SMALL LETTER M + 'n' # 0x95 -> LATIN SMALL LETTER N + 'o' # 0x96 -> LATIN SMALL LETTER O + 'p' # 0x97 -> LATIN SMALL LETTER P + 'q' # 0x98 -> LATIN SMALL LETTER Q + 'r' # 0x99 -> LATIN SMALL LETTER R + '\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR + '\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR + '\xe6' # 0x9C -> LATIN SMALL LIGATURE AE + '\xb8' # 0x9D -> CEDILLA + '\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE + '\u20ac' # 0x9F -> EURO SIGN + '\xb5' # 0xA0 -> MICRO SIGN + '~' # 0xA1 -> TILDE + 's' # 0xA2 -> LATIN SMALL LETTER S + 't' # 0xA3 -> LATIN SMALL LETTER T + 'u' # 0xA4 -> LATIN SMALL LETTER U + 'v' # 0xA5 -> LATIN SMALL LETTER V + 'w' # 0xA6 -> LATIN SMALL LETTER W + 'x' # 0xA7 -> LATIN SMALL LETTER X + 'y' # 0xA8 -> LATIN SMALL LETTER Y + 'z' # 0xA9 -> LATIN SMALL LETTER Z + '\xa1' # 0xAA -> INVERTED EXCLAMATION MARK + '\xbf' # 0xAB -> INVERTED QUESTION MARK + '\xd0' # 0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC) + '\xdd' # 0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC) + '\xae' # 0xAF -> REGISTERED SIGN + '^' # 0xB0 -> CIRCUMFLEX ACCENT + '\xa3' # 0xB1 -> POUND SIGN + '\xa5' # 0xB2 -> YEN SIGN + '\xb7' # 0xB3 -> MIDDLE DOT + '\xa9' # 0xB4 -> COPYRIGHT SIGN + '\xa7' # 0xB5 -> SECTION SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF + '\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS + '[' # 0xBA -> LEFT SQUARE BRACKET + ']' # 0xBB -> RIGHT SQUARE BRACKET + '\xaf' # 0xBC -> MACRON + '\xa8' # 0xBD -> DIAERESIS + '\xb4' # 0xBE -> ACUTE ACCENT + '\xd7' # 0xBF -> MULTIPLICATION SIGN + '{' # 0xC0 -> LEFT CURLY BRACKET + 'A' # 0xC1 -> LATIN CAPITAL LETTER A + 'B' # 0xC2 -> LATIN CAPITAL LETTER B + 'C' # 0xC3 -> LATIN CAPITAL LETTER C + 'D' # 0xC4 -> LATIN CAPITAL LETTER D + 'E' # 0xC5 -> LATIN CAPITAL LETTER E + 'F' # 0xC6 -> LATIN CAPITAL LETTER F + 'G' # 0xC7 -> LATIN CAPITAL LETTER G + 'H' # 0xC8 -> LATIN CAPITAL LETTER H + 'I' # 0xC9 -> LATIN CAPITAL LETTER I + '\xad' # 0xCA -> SOFT HYPHEN + '\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0xCC -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE + '\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE + '}' # 0xD0 -> RIGHT CURLY BRACKET + 'J' # 0xD1 -> LATIN CAPITAL LETTER J + 'K' # 0xD2 -> LATIN CAPITAL LETTER K + 'L' # 0xD3 -> LATIN CAPITAL LETTER L + 'M' # 0xD4 -> LATIN CAPITAL LETTER M + 'N' # 0xD5 -> LATIN CAPITAL LETTER N + 'O' # 0xD6 -> LATIN CAPITAL LETTER O + 'P' # 0xD7 -> LATIN CAPITAL LETTER P + 'Q' # 0xD8 -> LATIN CAPITAL LETTER Q + 'R' # 0xD9 -> LATIN CAPITAL LETTER R + '\xb9' # 0xDA -> SUPERSCRIPT ONE + '\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xDC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE + '\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\\' # 0xE0 -> REVERSE SOLIDUS + '\xf7' # 0xE1 -> DIVISION SIGN + 'S' # 0xE2 -> LATIN CAPITAL LETTER S + 'T' # 0xE3 -> LATIN CAPITAL LETTER T + 'U' # 0xE4 -> LATIN CAPITAL LETTER U + 'V' # 0xE5 -> LATIN CAPITAL LETTER V + 'W' # 0xE6 -> LATIN CAPITAL LETTER W + 'X' # 0xE7 -> LATIN CAPITAL LETTER X + 'Y' # 0xE8 -> LATIN CAPITAL LETTER Y + 'Z' # 0xE9 -> LATIN CAPITAL LETTER Z + '\xb2' # 0xEA -> SUPERSCRIPT TWO + '\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd6' # 0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE + '0' # 0xF0 -> DIGIT ZERO + '1' # 0xF1 -> DIGIT ONE + '2' # 0xF2 -> DIGIT TWO + '3' # 0xF3 -> DIGIT THREE + '4' # 0xF4 -> DIGIT FOUR + '5' # 0xF5 -> DIGIT FIVE + '6' # 0xF6 -> DIGIT SIX + '7' # 0xF7 -> DIGIT SEVEN + '8' # 0xF8 -> DIGIT EIGHT + '9' # 0xF9 -> DIGIT NINE + '\xb3' # 0xFA -> SUPERSCRIPT THREE + '\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE + '\x9f' # 0xFF -> CONTROL ) ### Encoding table diff --git a/Lib/encodings/cp1250.py b/Lib/encodings/cp1250.py index d620b89..c2c83aa 100644 --- a/Lib/encodings/cp1250.py +++ b/Lib/encodings/cp1250.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\ufffe' # 0x83 -> UNDEFINED - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\ufffe' # 0x88 -> UNDEFINED - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\u0160' # 0x8A -> LATIN CAPITAL LETTER S WITH CARON - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u015a' # 0x8C -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u0164' # 0x8D -> LATIN CAPITAL LETTER T WITH CARON - u'\u017d' # 0x8E -> LATIN CAPITAL LETTER Z WITH CARON - u'\u0179' # 0x8F -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\ufffe' # 0x98 -> UNDEFINED - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\u0161' # 0x9A -> LATIN SMALL LETTER S WITH CARON - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u015b' # 0x9C -> LATIN SMALL LETTER S WITH ACUTE - u'\u0165' # 0x9D -> LATIN SMALL LETTER T WITH CARON - u'\u017e' # 0x9E -> LATIN SMALL LETTER Z WITH CARON - u'\u017a' # 0x9F -> LATIN SMALL LETTER Z WITH ACUTE - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u02c7' # 0xA1 -> CARON - u'\u02d8' # 0xA2 -> BREVE - u'\u0141' # 0xA3 -> LATIN CAPITAL LETTER L WITH STROKE - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\u0104' # 0xA5 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u015e' # 0xAA -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u02db' # 0xB2 -> OGONEK - u'\u0142' # 0xB3 -> LATIN SMALL LETTER L WITH STROKE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\u0105' # 0xB9 -> LATIN SMALL LETTER A WITH OGONEK - u'\u015f' # 0xBA -> LATIN SMALL LETTER S WITH CEDILLA - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u013d' # 0xBC -> LATIN CAPITAL LETTER L WITH CARON - u'\u02dd' # 0xBD -> DOUBLE ACUTE ACCENT - u'\u013e' # 0xBE -> LATIN SMALL LETTER L WITH CARON - u'\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u0154' # 0xC0 -> LATIN CAPITAL LETTER R WITH ACUTE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\u0139' # 0xC5 -> LATIN CAPITAL LETTER L WITH ACUTE - u'\u0106' # 0xC6 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u011a' # 0xCC -> LATIN CAPITAL LETTER E WITH CARON - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\u010e' # 0xCF -> LATIN CAPITAL LETTER D WITH CARON - u'\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\u0147' # 0xD2 -> LATIN CAPITAL LETTER N WITH CARON - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u0150' # 0xD5 -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\u0158' # 0xD8 -> LATIN CAPITAL LETTER R WITH CARON - u'\u016e' # 0xD9 -> LATIN CAPITAL LETTER U WITH RING ABOVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\u0170' # 0xDB -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\u0162' # 0xDE -> LATIN CAPITAL LETTER T WITH CEDILLA - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\u0155' # 0xE0 -> LATIN SMALL LETTER R WITH ACUTE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u013a' # 0xE5 -> LATIN SMALL LETTER L WITH ACUTE - u'\u0107' # 0xE6 -> LATIN SMALL LETTER C WITH ACUTE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u011b' # 0xEC -> LATIN SMALL LETTER E WITH CARON - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\u010f' # 0xEF -> LATIN SMALL LETTER D WITH CARON - u'\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE - u'\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0148' # 0xF2 -> LATIN SMALL LETTER N WITH CARON - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\u0151' # 0xF5 -> LATIN SMALL LETTER O WITH DOUBLE ACUTE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\u0159' # 0xF8 -> LATIN SMALL LETTER R WITH CARON - u'\u016f' # 0xF9 -> LATIN SMALL LETTER U WITH RING ABOVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\u0171' # 0xFB -> LATIN SMALL LETTER U WITH DOUBLE ACUTE - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\u0163' # 0xFE -> LATIN SMALL LETTER T WITH CEDILLA - u'\u02d9' # 0xFF -> DOT ABOVE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\ufffe' # 0x83 -> UNDEFINED + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\ufffe' # 0x88 -> UNDEFINED + '\u2030' # 0x89 -> PER MILLE SIGN + '\u0160' # 0x8A -> LATIN CAPITAL LETTER S WITH CARON + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u015a' # 0x8C -> LATIN CAPITAL LETTER S WITH ACUTE + '\u0164' # 0x8D -> LATIN CAPITAL LETTER T WITH CARON + '\u017d' # 0x8E -> LATIN CAPITAL LETTER Z WITH CARON + '\u0179' # 0x8F -> LATIN CAPITAL LETTER Z WITH ACUTE + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\ufffe' # 0x98 -> UNDEFINED + '\u2122' # 0x99 -> TRADE MARK SIGN + '\u0161' # 0x9A -> LATIN SMALL LETTER S WITH CARON + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u015b' # 0x9C -> LATIN SMALL LETTER S WITH ACUTE + '\u0165' # 0x9D -> LATIN SMALL LETTER T WITH CARON + '\u017e' # 0x9E -> LATIN SMALL LETTER Z WITH CARON + '\u017a' # 0x9F -> LATIN SMALL LETTER Z WITH ACUTE + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u02c7' # 0xA1 -> CARON + '\u02d8' # 0xA2 -> BREVE + '\u0141' # 0xA3 -> LATIN CAPITAL LETTER L WITH STROKE + '\xa4' # 0xA4 -> CURRENCY SIGN + '\u0104' # 0xA5 -> LATIN CAPITAL LETTER A WITH OGONEK + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u015e' # 0xAA -> LATIN CAPITAL LETTER S WITH CEDILLA + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u02db' # 0xB2 -> OGONEK + '\u0142' # 0xB3 -> LATIN SMALL LETTER L WITH STROKE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\u0105' # 0xB9 -> LATIN SMALL LETTER A WITH OGONEK + '\u015f' # 0xBA -> LATIN SMALL LETTER S WITH CEDILLA + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u013d' # 0xBC -> LATIN CAPITAL LETTER L WITH CARON + '\u02dd' # 0xBD -> DOUBLE ACUTE ACCENT + '\u013e' # 0xBE -> LATIN SMALL LETTER L WITH CARON + '\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u0154' # 0xC0 -> LATIN CAPITAL LETTER R WITH ACUTE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\u0139' # 0xC5 -> LATIN CAPITAL LETTER L WITH ACUTE + '\u0106' # 0xC6 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u011a' # 0xCC -> LATIN CAPITAL LETTER E WITH CARON + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\u010e' # 0xCF -> LATIN CAPITAL LETTER D WITH CARON + '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE + '\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE + '\u0147' # 0xD2 -> LATIN CAPITAL LETTER N WITH CARON + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u0150' # 0xD5 -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\u0158' # 0xD8 -> LATIN CAPITAL LETTER R WITH CARON + '\u016e' # 0xD9 -> LATIN CAPITAL LETTER U WITH RING ABOVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\u0170' # 0xDB -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\u0162' # 0xDE -> LATIN CAPITAL LETTER T WITH CEDILLA + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\u0155' # 0xE0 -> LATIN SMALL LETTER R WITH ACUTE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\u013a' # 0xE5 -> LATIN SMALL LETTER L WITH ACUTE + '\u0107' # 0xE6 -> LATIN SMALL LETTER C WITH ACUTE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\u011b' # 0xEC -> LATIN SMALL LETTER E WITH CARON + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\u010f' # 0xEF -> LATIN SMALL LETTER D WITH CARON + '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE + '\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE + '\u0148' # 0xF2 -> LATIN SMALL LETTER N WITH CARON + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\u0151' # 0xF5 -> LATIN SMALL LETTER O WITH DOUBLE ACUTE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\u0159' # 0xF8 -> LATIN SMALL LETTER R WITH CARON + '\u016f' # 0xF9 -> LATIN SMALL LETTER U WITH RING ABOVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\u0171' # 0xFB -> LATIN SMALL LETTER U WITH DOUBLE ACUTE + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\u0163' # 0xFE -> LATIN SMALL LETTER T WITH CEDILLA + '\u02d9' # 0xFF -> DOT ABOVE ) ### Encoding table diff --git a/Lib/encodings/cp1251.py b/Lib/encodings/cp1251.py index 216771f..22bc660 100644 --- a/Lib/encodings/cp1251.py +++ b/Lib/encodings/cp1251.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u0402' # 0x80 -> CYRILLIC CAPITAL LETTER DJE - u'\u0403' # 0x81 -> CYRILLIC CAPITAL LETTER GJE - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0453' # 0x83 -> CYRILLIC SMALL LETTER GJE - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\u20ac' # 0x88 -> EURO SIGN - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\u0409' # 0x8A -> CYRILLIC CAPITAL LETTER LJE - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u040a' # 0x8C -> CYRILLIC CAPITAL LETTER NJE - u'\u040c' # 0x8D -> CYRILLIC CAPITAL LETTER KJE - u'\u040b' # 0x8E -> CYRILLIC CAPITAL LETTER TSHE - u'\u040f' # 0x8F -> CYRILLIC CAPITAL LETTER DZHE - u'\u0452' # 0x90 -> CYRILLIC SMALL LETTER DJE - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\ufffe' # 0x98 -> UNDEFINED - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\u0459' # 0x9A -> CYRILLIC SMALL LETTER LJE - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u045a' # 0x9C -> CYRILLIC SMALL LETTER NJE - u'\u045c' # 0x9D -> CYRILLIC SMALL LETTER KJE - u'\u045b' # 0x9E -> CYRILLIC SMALL LETTER TSHE - u'\u045f' # 0x9F -> CYRILLIC SMALL LETTER DZHE - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u040e' # 0xA1 -> CYRILLIC CAPITAL LETTER SHORT U - u'\u045e' # 0xA2 -> CYRILLIC SMALL LETTER SHORT U - u'\u0408' # 0xA3 -> CYRILLIC CAPITAL LETTER JE - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\u0490' # 0xA5 -> CYRILLIC CAPITAL LETTER GHE WITH UPTURN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\u0401' # 0xA8 -> CYRILLIC CAPITAL LETTER IO - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u0404' # 0xAA -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\u0407' # 0xAF -> CYRILLIC CAPITAL LETTER YI - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u0406' # 0xB2 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0456' # 0xB3 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0491' # 0xB4 -> CYRILLIC SMALL LETTER GHE WITH UPTURN - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\u0451' # 0xB8 -> CYRILLIC SMALL LETTER IO - u'\u2116' # 0xB9 -> NUMERO SIGN - u'\u0454' # 0xBA -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u0458' # 0xBC -> CYRILLIC SMALL LETTER JE - u'\u0405' # 0xBD -> CYRILLIC CAPITAL LETTER DZE - u'\u0455' # 0xBE -> CYRILLIC SMALL LETTER DZE - u'\u0457' # 0xBF -> CYRILLIC SMALL LETTER YI - u'\u0410' # 0xC0 -> CYRILLIC CAPITAL LETTER A - u'\u0411' # 0xC1 -> CYRILLIC CAPITAL LETTER BE - u'\u0412' # 0xC2 -> CYRILLIC CAPITAL LETTER VE - u'\u0413' # 0xC3 -> CYRILLIC CAPITAL LETTER GHE - u'\u0414' # 0xC4 -> CYRILLIC CAPITAL LETTER DE - u'\u0415' # 0xC5 -> CYRILLIC CAPITAL LETTER IE - u'\u0416' # 0xC6 -> CYRILLIC CAPITAL LETTER ZHE - u'\u0417' # 0xC7 -> CYRILLIC CAPITAL LETTER ZE - u'\u0418' # 0xC8 -> CYRILLIC CAPITAL LETTER I - u'\u0419' # 0xC9 -> CYRILLIC CAPITAL LETTER SHORT I - u'\u041a' # 0xCA -> CYRILLIC CAPITAL LETTER KA - u'\u041b' # 0xCB -> CYRILLIC CAPITAL LETTER EL - u'\u041c' # 0xCC -> CYRILLIC CAPITAL LETTER EM - u'\u041d' # 0xCD -> CYRILLIC CAPITAL LETTER EN - u'\u041e' # 0xCE -> CYRILLIC CAPITAL LETTER O - u'\u041f' # 0xCF -> CYRILLIC CAPITAL LETTER PE - u'\u0420' # 0xD0 -> CYRILLIC CAPITAL LETTER ER - u'\u0421' # 0xD1 -> CYRILLIC CAPITAL LETTER ES - u'\u0422' # 0xD2 -> CYRILLIC CAPITAL LETTER TE - u'\u0423' # 0xD3 -> CYRILLIC CAPITAL LETTER U - u'\u0424' # 0xD4 -> CYRILLIC CAPITAL LETTER EF - u'\u0425' # 0xD5 -> CYRILLIC CAPITAL LETTER HA - u'\u0426' # 0xD6 -> CYRILLIC CAPITAL LETTER TSE - u'\u0427' # 0xD7 -> CYRILLIC CAPITAL LETTER CHE - u'\u0428' # 0xD8 -> CYRILLIC CAPITAL LETTER SHA - u'\u0429' # 0xD9 -> CYRILLIC CAPITAL LETTER SHCHA - u'\u042a' # 0xDA -> CYRILLIC CAPITAL LETTER HARD SIGN - u'\u042b' # 0xDB -> CYRILLIC CAPITAL LETTER YERU - u'\u042c' # 0xDC -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u042d' # 0xDD -> CYRILLIC CAPITAL LETTER E - u'\u042e' # 0xDE -> CYRILLIC CAPITAL LETTER YU - u'\u042f' # 0xDF -> CYRILLIC CAPITAL LETTER YA - u'\u0430' # 0xE0 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0xE1 -> CYRILLIC SMALL LETTER BE - u'\u0432' # 0xE2 -> CYRILLIC SMALL LETTER VE - u'\u0433' # 0xE3 -> CYRILLIC SMALL LETTER GHE - u'\u0434' # 0xE4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0xE5 -> CYRILLIC SMALL LETTER IE - u'\u0436' # 0xE6 -> CYRILLIC SMALL LETTER ZHE - u'\u0437' # 0xE7 -> CYRILLIC SMALL LETTER ZE - u'\u0438' # 0xE8 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0xE9 -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0xEA -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0xEB -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0xEC -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0xED -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0xEE -> CYRILLIC SMALL LETTER O - u'\u043f' # 0xEF -> CYRILLIC SMALL LETTER PE - u'\u0440' # 0xF0 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0xF1 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0xF2 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0xF3 -> CYRILLIC SMALL LETTER U - u'\u0444' # 0xF4 -> CYRILLIC SMALL LETTER EF - u'\u0445' # 0xF5 -> CYRILLIC SMALL LETTER HA - u'\u0446' # 0xF6 -> CYRILLIC SMALL LETTER TSE - u'\u0447' # 0xF7 -> CYRILLIC SMALL LETTER CHE - u'\u0448' # 0xF8 -> CYRILLIC SMALL LETTER SHA - u'\u0449' # 0xF9 -> CYRILLIC SMALL LETTER SHCHA - u'\u044a' # 0xFA -> CYRILLIC SMALL LETTER HARD SIGN - u'\u044b' # 0xFB -> CYRILLIC SMALL LETTER YERU - u'\u044c' # 0xFC -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044d' # 0xFD -> CYRILLIC SMALL LETTER E - u'\u044e' # 0xFE -> CYRILLIC SMALL LETTER YU - u'\u044f' # 0xFF -> CYRILLIC SMALL LETTER YA + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u0402' # 0x80 -> CYRILLIC CAPITAL LETTER DJE + '\u0403' # 0x81 -> CYRILLIC CAPITAL LETTER GJE + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0453' # 0x83 -> CYRILLIC SMALL LETTER GJE + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\u20ac' # 0x88 -> EURO SIGN + '\u2030' # 0x89 -> PER MILLE SIGN + '\u0409' # 0x8A -> CYRILLIC CAPITAL LETTER LJE + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u040a' # 0x8C -> CYRILLIC CAPITAL LETTER NJE + '\u040c' # 0x8D -> CYRILLIC CAPITAL LETTER KJE + '\u040b' # 0x8E -> CYRILLIC CAPITAL LETTER TSHE + '\u040f' # 0x8F -> CYRILLIC CAPITAL LETTER DZHE + '\u0452' # 0x90 -> CYRILLIC SMALL LETTER DJE + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\ufffe' # 0x98 -> UNDEFINED + '\u2122' # 0x99 -> TRADE MARK SIGN + '\u0459' # 0x9A -> CYRILLIC SMALL LETTER LJE + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u045a' # 0x9C -> CYRILLIC SMALL LETTER NJE + '\u045c' # 0x9D -> CYRILLIC SMALL LETTER KJE + '\u045b' # 0x9E -> CYRILLIC SMALL LETTER TSHE + '\u045f' # 0x9F -> CYRILLIC SMALL LETTER DZHE + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u040e' # 0xA1 -> CYRILLIC CAPITAL LETTER SHORT U + '\u045e' # 0xA2 -> CYRILLIC SMALL LETTER SHORT U + '\u0408' # 0xA3 -> CYRILLIC CAPITAL LETTER JE + '\xa4' # 0xA4 -> CURRENCY SIGN + '\u0490' # 0xA5 -> CYRILLIC CAPITAL LETTER GHE WITH UPTURN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\u0401' # 0xA8 -> CYRILLIC CAPITAL LETTER IO + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u0404' # 0xAA -> CYRILLIC CAPITAL LETTER UKRAINIAN IE + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\u0407' # 0xAF -> CYRILLIC CAPITAL LETTER YI + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u0406' # 0xB2 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0456' # 0xB3 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0491' # 0xB4 -> CYRILLIC SMALL LETTER GHE WITH UPTURN + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\u0451' # 0xB8 -> CYRILLIC SMALL LETTER IO + '\u2116' # 0xB9 -> NUMERO SIGN + '\u0454' # 0xBA -> CYRILLIC SMALL LETTER UKRAINIAN IE + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u0458' # 0xBC -> CYRILLIC SMALL LETTER JE + '\u0405' # 0xBD -> CYRILLIC CAPITAL LETTER DZE + '\u0455' # 0xBE -> CYRILLIC SMALL LETTER DZE + '\u0457' # 0xBF -> CYRILLIC SMALL LETTER YI + '\u0410' # 0xC0 -> CYRILLIC CAPITAL LETTER A + '\u0411' # 0xC1 -> CYRILLIC CAPITAL LETTER BE + '\u0412' # 0xC2 -> CYRILLIC CAPITAL LETTER VE + '\u0413' # 0xC3 -> CYRILLIC CAPITAL LETTER GHE + '\u0414' # 0xC4 -> CYRILLIC CAPITAL LETTER DE + '\u0415' # 0xC5 -> CYRILLIC CAPITAL LETTER IE + '\u0416' # 0xC6 -> CYRILLIC CAPITAL LETTER ZHE + '\u0417' # 0xC7 -> CYRILLIC CAPITAL LETTER ZE + '\u0418' # 0xC8 -> CYRILLIC CAPITAL LETTER I + '\u0419' # 0xC9 -> CYRILLIC CAPITAL LETTER SHORT I + '\u041a' # 0xCA -> CYRILLIC CAPITAL LETTER KA + '\u041b' # 0xCB -> CYRILLIC CAPITAL LETTER EL + '\u041c' # 0xCC -> CYRILLIC CAPITAL LETTER EM + '\u041d' # 0xCD -> CYRILLIC CAPITAL LETTER EN + '\u041e' # 0xCE -> CYRILLIC CAPITAL LETTER O + '\u041f' # 0xCF -> CYRILLIC CAPITAL LETTER PE + '\u0420' # 0xD0 -> CYRILLIC CAPITAL LETTER ER + '\u0421' # 0xD1 -> CYRILLIC CAPITAL LETTER ES + '\u0422' # 0xD2 -> CYRILLIC CAPITAL LETTER TE + '\u0423' # 0xD3 -> CYRILLIC CAPITAL LETTER U + '\u0424' # 0xD4 -> CYRILLIC CAPITAL LETTER EF + '\u0425' # 0xD5 -> CYRILLIC CAPITAL LETTER HA + '\u0426' # 0xD6 -> CYRILLIC CAPITAL LETTER TSE + '\u0427' # 0xD7 -> CYRILLIC CAPITAL LETTER CHE + '\u0428' # 0xD8 -> CYRILLIC CAPITAL LETTER SHA + '\u0429' # 0xD9 -> CYRILLIC CAPITAL LETTER SHCHA + '\u042a' # 0xDA -> CYRILLIC CAPITAL LETTER HARD SIGN + '\u042b' # 0xDB -> CYRILLIC CAPITAL LETTER YERU + '\u042c' # 0xDC -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u042d' # 0xDD -> CYRILLIC CAPITAL LETTER E + '\u042e' # 0xDE -> CYRILLIC CAPITAL LETTER YU + '\u042f' # 0xDF -> CYRILLIC CAPITAL LETTER YA + '\u0430' # 0xE0 -> CYRILLIC SMALL LETTER A + '\u0431' # 0xE1 -> CYRILLIC SMALL LETTER BE + '\u0432' # 0xE2 -> CYRILLIC SMALL LETTER VE + '\u0433' # 0xE3 -> CYRILLIC SMALL LETTER GHE + '\u0434' # 0xE4 -> CYRILLIC SMALL LETTER DE + '\u0435' # 0xE5 -> CYRILLIC SMALL LETTER IE + '\u0436' # 0xE6 -> CYRILLIC SMALL LETTER ZHE + '\u0437' # 0xE7 -> CYRILLIC SMALL LETTER ZE + '\u0438' # 0xE8 -> CYRILLIC SMALL LETTER I + '\u0439' # 0xE9 -> CYRILLIC SMALL LETTER SHORT I + '\u043a' # 0xEA -> CYRILLIC SMALL LETTER KA + '\u043b' # 0xEB -> CYRILLIC SMALL LETTER EL + '\u043c' # 0xEC -> CYRILLIC SMALL LETTER EM + '\u043d' # 0xED -> CYRILLIC SMALL LETTER EN + '\u043e' # 0xEE -> CYRILLIC SMALL LETTER O + '\u043f' # 0xEF -> CYRILLIC SMALL LETTER PE + '\u0440' # 0xF0 -> CYRILLIC SMALL LETTER ER + '\u0441' # 0xF1 -> CYRILLIC SMALL LETTER ES + '\u0442' # 0xF2 -> CYRILLIC SMALL LETTER TE + '\u0443' # 0xF3 -> CYRILLIC SMALL LETTER U + '\u0444' # 0xF4 -> CYRILLIC SMALL LETTER EF + '\u0445' # 0xF5 -> CYRILLIC SMALL LETTER HA + '\u0446' # 0xF6 -> CYRILLIC SMALL LETTER TSE + '\u0447' # 0xF7 -> CYRILLIC SMALL LETTER CHE + '\u0448' # 0xF8 -> CYRILLIC SMALL LETTER SHA + '\u0449' # 0xF9 -> CYRILLIC SMALL LETTER SHCHA + '\u044a' # 0xFA -> CYRILLIC SMALL LETTER HARD SIGN + '\u044b' # 0xFB -> CYRILLIC SMALL LETTER YERU + '\u044c' # 0xFC -> CYRILLIC SMALL LETTER SOFT SIGN + '\u044d' # 0xFD -> CYRILLIC SMALL LETTER E + '\u044e' # 0xFE -> CYRILLIC SMALL LETTER YU + '\u044f' # 0xFF -> CYRILLIC SMALL LETTER YA ) ### Encoding table diff --git a/Lib/encodings/cp1252.py b/Lib/encodings/cp1252.py index e60a328..c0e8088 100644 --- a/Lib/encodings/cp1252.py +++ b/Lib/encodings/cp1252.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\u0160' # 0x8A -> LATIN CAPITAL LETTER S WITH CARON - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE - u'\ufffe' # 0x8D -> UNDEFINED - u'\u017d' # 0x8E -> LATIN CAPITAL LETTER Z WITH CARON - u'\ufffe' # 0x8F -> UNDEFINED - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\u02dc' # 0x98 -> SMALL TILDE - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\u0161' # 0x9A -> LATIN SMALL LETTER S WITH CARON - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u0153' # 0x9C -> LATIN SMALL LIGATURE OE - u'\ufffe' # 0x9D -> UNDEFINED - u'\u017e' # 0x9E -> LATIN SMALL LETTER Z WITH CARON - u'\u0178' # 0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xDE -> LATIN CAPITAL LETTER THORN - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf0' # 0xF0 -> LATIN SMALL LETTER ETH - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0xFE -> LATIN SMALL LETTER THORN - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u2030' # 0x89 -> PER MILLE SIGN + '\u0160' # 0x8A -> LATIN CAPITAL LETTER S WITH CARON + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE + '\ufffe' # 0x8D -> UNDEFINED + '\u017d' # 0x8E -> LATIN CAPITAL LETTER Z WITH CARON + '\ufffe' # 0x8F -> UNDEFINED + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\u02dc' # 0x98 -> SMALL TILDE + '\u2122' # 0x99 -> TRADE MARK SIGN + '\u0161' # 0x9A -> LATIN SMALL LETTER S WITH CARON + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u0153' # 0x9C -> LATIN SMALL LIGATURE OE + '\ufffe' # 0x9D -> UNDEFINED + '\u017e' # 0x9E -> LATIN SMALL LETTER Z WITH CARON + '\u0178' # 0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xDE -> LATIN CAPITAL LETTER THORN + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf0' # 0xF0 -> LATIN SMALL LETTER ETH + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0xFE -> LATIN SMALL LETTER THORN + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/cp1253.py b/Lib/encodings/cp1253.py index 49f6ccc..ec9c097 100644 --- a/Lib/encodings/cp1253.py +++ b/Lib/encodings/cp1253.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\ufffe' # 0x88 -> UNDEFINED - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\ufffe' # 0x8A -> UNDEFINED - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\ufffe' # 0x8C -> UNDEFINED - u'\ufffe' # 0x8D -> UNDEFINED - u'\ufffe' # 0x8E -> UNDEFINED - u'\ufffe' # 0x8F -> UNDEFINED - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\ufffe' # 0x98 -> UNDEFINED - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\ufffe' # 0x9A -> UNDEFINED - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\ufffe' # 0x9C -> UNDEFINED - u'\ufffe' # 0x9D -> UNDEFINED - u'\ufffe' # 0x9E -> UNDEFINED - u'\ufffe' # 0x9F -> UNDEFINED - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0385' # 0xA1 -> GREEK DIALYTIKA TONOS - u'\u0386' # 0xA2 -> GREEK CAPITAL LETTER ALPHA WITH TONOS - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\ufffe' # 0xAA -> UNDEFINED - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\u2015' # 0xAF -> HORIZONTAL BAR - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\u0384' # 0xB4 -> GREEK TONOS - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\u0388' # 0xB8 -> GREEK CAPITAL LETTER EPSILON WITH TONOS - u'\u0389' # 0xB9 -> GREEK CAPITAL LETTER ETA WITH TONOS - u'\u038a' # 0xBA -> GREEK CAPITAL LETTER IOTA WITH TONOS - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u038c' # 0xBC -> GREEK CAPITAL LETTER OMICRON WITH TONOS - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\u038e' # 0xBE -> GREEK CAPITAL LETTER UPSILON WITH TONOS - u'\u038f' # 0xBF -> GREEK CAPITAL LETTER OMEGA WITH TONOS - u'\u0390' # 0xC0 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - u'\u0391' # 0xC1 -> GREEK CAPITAL LETTER ALPHA - u'\u0392' # 0xC2 -> GREEK CAPITAL LETTER BETA - u'\u0393' # 0xC3 -> GREEK CAPITAL LETTER GAMMA - u'\u0394' # 0xC4 -> GREEK CAPITAL LETTER DELTA - u'\u0395' # 0xC5 -> GREEK CAPITAL LETTER EPSILON - u'\u0396' # 0xC6 -> GREEK CAPITAL LETTER ZETA - u'\u0397' # 0xC7 -> GREEK CAPITAL LETTER ETA - u'\u0398' # 0xC8 -> GREEK CAPITAL LETTER THETA - u'\u0399' # 0xC9 -> GREEK CAPITAL LETTER IOTA - u'\u039a' # 0xCA -> GREEK CAPITAL LETTER KAPPA - u'\u039b' # 0xCB -> GREEK CAPITAL LETTER LAMDA - u'\u039c' # 0xCC -> GREEK CAPITAL LETTER MU - u'\u039d' # 0xCD -> GREEK CAPITAL LETTER NU - u'\u039e' # 0xCE -> GREEK CAPITAL LETTER XI - u'\u039f' # 0xCF -> GREEK CAPITAL LETTER OMICRON - u'\u03a0' # 0xD0 -> GREEK CAPITAL LETTER PI - u'\u03a1' # 0xD1 -> GREEK CAPITAL LETTER RHO - u'\ufffe' # 0xD2 -> UNDEFINED - u'\u03a3' # 0xD3 -> GREEK CAPITAL LETTER SIGMA - u'\u03a4' # 0xD4 -> GREEK CAPITAL LETTER TAU - u'\u03a5' # 0xD5 -> GREEK CAPITAL LETTER UPSILON - u'\u03a6' # 0xD6 -> GREEK CAPITAL LETTER PHI - u'\u03a7' # 0xD7 -> GREEK CAPITAL LETTER CHI - u'\u03a8' # 0xD8 -> GREEK CAPITAL LETTER PSI - u'\u03a9' # 0xD9 -> GREEK CAPITAL LETTER OMEGA - u'\u03aa' # 0xDA -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - u'\u03ab' # 0xDB -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - u'\u03ac' # 0xDC -> GREEK SMALL LETTER ALPHA WITH TONOS - u'\u03ad' # 0xDD -> GREEK SMALL LETTER EPSILON WITH TONOS - u'\u03ae' # 0xDE -> GREEK SMALL LETTER ETA WITH TONOS - u'\u03af' # 0xDF -> GREEK SMALL LETTER IOTA WITH TONOS - u'\u03b0' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - u'\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA - u'\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA - u'\u03b3' # 0xE3 -> GREEK SMALL LETTER GAMMA - u'\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA - u'\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON - u'\u03b6' # 0xE6 -> GREEK SMALL LETTER ZETA - u'\u03b7' # 0xE7 -> GREEK SMALL LETTER ETA - u'\u03b8' # 0xE8 -> GREEK SMALL LETTER THETA - u'\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA - u'\u03ba' # 0xEA -> GREEK SMALL LETTER KAPPA - u'\u03bb' # 0xEB -> GREEK SMALL LETTER LAMDA - u'\u03bc' # 0xEC -> GREEK SMALL LETTER MU - u'\u03bd' # 0xED -> GREEK SMALL LETTER NU - u'\u03be' # 0xEE -> GREEK SMALL LETTER XI - u'\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON - u'\u03c0' # 0xF0 -> GREEK SMALL LETTER PI - u'\u03c1' # 0xF1 -> GREEK SMALL LETTER RHO - u'\u03c2' # 0xF2 -> GREEK SMALL LETTER FINAL SIGMA - u'\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA - u'\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU - u'\u03c5' # 0xF5 -> GREEK SMALL LETTER UPSILON - u'\u03c6' # 0xF6 -> GREEK SMALL LETTER PHI - u'\u03c7' # 0xF7 -> GREEK SMALL LETTER CHI - u'\u03c8' # 0xF8 -> GREEK SMALL LETTER PSI - u'\u03c9' # 0xF9 -> GREEK SMALL LETTER OMEGA - u'\u03ca' # 0xFA -> GREEK SMALL LETTER IOTA WITH DIALYTIKA - u'\u03cb' # 0xFB -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA - u'\u03cc' # 0xFC -> GREEK SMALL LETTER OMICRON WITH TONOS - u'\u03cd' # 0xFD -> GREEK SMALL LETTER UPSILON WITH TONOS - u'\u03ce' # 0xFE -> GREEK SMALL LETTER OMEGA WITH TONOS - u'\ufffe' # 0xFF -> UNDEFINED + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\ufffe' # 0x88 -> UNDEFINED + '\u2030' # 0x89 -> PER MILLE SIGN + '\ufffe' # 0x8A -> UNDEFINED + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\ufffe' # 0x8C -> UNDEFINED + '\ufffe' # 0x8D -> UNDEFINED + '\ufffe' # 0x8E -> UNDEFINED + '\ufffe' # 0x8F -> UNDEFINED + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\ufffe' # 0x98 -> UNDEFINED + '\u2122' # 0x99 -> TRADE MARK SIGN + '\ufffe' # 0x9A -> UNDEFINED + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\ufffe' # 0x9C -> UNDEFINED + '\ufffe' # 0x9D -> UNDEFINED + '\ufffe' # 0x9E -> UNDEFINED + '\ufffe' # 0x9F -> UNDEFINED + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0385' # 0xA1 -> GREEK DIALYTIKA TONOS + '\u0386' # 0xA2 -> GREEK CAPITAL LETTER ALPHA WITH TONOS + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\ufffe' # 0xAA -> UNDEFINED + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\u2015' # 0xAF -> HORIZONTAL BAR + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\u0384' # 0xB4 -> GREEK TONOS + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\u0388' # 0xB8 -> GREEK CAPITAL LETTER EPSILON WITH TONOS + '\u0389' # 0xB9 -> GREEK CAPITAL LETTER ETA WITH TONOS + '\u038a' # 0xBA -> GREEK CAPITAL LETTER IOTA WITH TONOS + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u038c' # 0xBC -> GREEK CAPITAL LETTER OMICRON WITH TONOS + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\u038e' # 0xBE -> GREEK CAPITAL LETTER UPSILON WITH TONOS + '\u038f' # 0xBF -> GREEK CAPITAL LETTER OMEGA WITH TONOS + '\u0390' # 0xC0 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS + '\u0391' # 0xC1 -> GREEK CAPITAL LETTER ALPHA + '\u0392' # 0xC2 -> GREEK CAPITAL LETTER BETA + '\u0393' # 0xC3 -> GREEK CAPITAL LETTER GAMMA + '\u0394' # 0xC4 -> GREEK CAPITAL LETTER DELTA + '\u0395' # 0xC5 -> GREEK CAPITAL LETTER EPSILON + '\u0396' # 0xC6 -> GREEK CAPITAL LETTER ZETA + '\u0397' # 0xC7 -> GREEK CAPITAL LETTER ETA + '\u0398' # 0xC8 -> GREEK CAPITAL LETTER THETA + '\u0399' # 0xC9 -> GREEK CAPITAL LETTER IOTA + '\u039a' # 0xCA -> GREEK CAPITAL LETTER KAPPA + '\u039b' # 0xCB -> GREEK CAPITAL LETTER LAMDA + '\u039c' # 0xCC -> GREEK CAPITAL LETTER MU + '\u039d' # 0xCD -> GREEK CAPITAL LETTER NU + '\u039e' # 0xCE -> GREEK CAPITAL LETTER XI + '\u039f' # 0xCF -> GREEK CAPITAL LETTER OMICRON + '\u03a0' # 0xD0 -> GREEK CAPITAL LETTER PI + '\u03a1' # 0xD1 -> GREEK CAPITAL LETTER RHO + '\ufffe' # 0xD2 -> UNDEFINED + '\u03a3' # 0xD3 -> GREEK CAPITAL LETTER SIGMA + '\u03a4' # 0xD4 -> GREEK CAPITAL LETTER TAU + '\u03a5' # 0xD5 -> GREEK CAPITAL LETTER UPSILON + '\u03a6' # 0xD6 -> GREEK CAPITAL LETTER PHI + '\u03a7' # 0xD7 -> GREEK CAPITAL LETTER CHI + '\u03a8' # 0xD8 -> GREEK CAPITAL LETTER PSI + '\u03a9' # 0xD9 -> GREEK CAPITAL LETTER OMEGA + '\u03aa' # 0xDA -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA + '\u03ab' # 0xDB -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA + '\u03ac' # 0xDC -> GREEK SMALL LETTER ALPHA WITH TONOS + '\u03ad' # 0xDD -> GREEK SMALL LETTER EPSILON WITH TONOS + '\u03ae' # 0xDE -> GREEK SMALL LETTER ETA WITH TONOS + '\u03af' # 0xDF -> GREEK SMALL LETTER IOTA WITH TONOS + '\u03b0' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS + '\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA + '\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA + '\u03b3' # 0xE3 -> GREEK SMALL LETTER GAMMA + '\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA + '\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON + '\u03b6' # 0xE6 -> GREEK SMALL LETTER ZETA + '\u03b7' # 0xE7 -> GREEK SMALL LETTER ETA + '\u03b8' # 0xE8 -> GREEK SMALL LETTER THETA + '\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA + '\u03ba' # 0xEA -> GREEK SMALL LETTER KAPPA + '\u03bb' # 0xEB -> GREEK SMALL LETTER LAMDA + '\u03bc' # 0xEC -> GREEK SMALL LETTER MU + '\u03bd' # 0xED -> GREEK SMALL LETTER NU + '\u03be' # 0xEE -> GREEK SMALL LETTER XI + '\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON + '\u03c0' # 0xF0 -> GREEK SMALL LETTER PI + '\u03c1' # 0xF1 -> GREEK SMALL LETTER RHO + '\u03c2' # 0xF2 -> GREEK SMALL LETTER FINAL SIGMA + '\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA + '\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU + '\u03c5' # 0xF5 -> GREEK SMALL LETTER UPSILON + '\u03c6' # 0xF6 -> GREEK SMALL LETTER PHI + '\u03c7' # 0xF7 -> GREEK SMALL LETTER CHI + '\u03c8' # 0xF8 -> GREEK SMALL LETTER PSI + '\u03c9' # 0xF9 -> GREEK SMALL LETTER OMEGA + '\u03ca' # 0xFA -> GREEK SMALL LETTER IOTA WITH DIALYTIKA + '\u03cb' # 0xFB -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA + '\u03cc' # 0xFC -> GREEK SMALL LETTER OMICRON WITH TONOS + '\u03cd' # 0xFD -> GREEK SMALL LETTER UPSILON WITH TONOS + '\u03ce' # 0xFE -> GREEK SMALL LETTER OMEGA WITH TONOS + '\ufffe' # 0xFF -> UNDEFINED ) ### Encoding table diff --git a/Lib/encodings/cp1254.py b/Lib/encodings/cp1254.py index 65530ab..4912327 100644 --- a/Lib/encodings/cp1254.py +++ b/Lib/encodings/cp1254.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\u0160' # 0x8A -> LATIN CAPITAL LETTER S WITH CARON - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE - u'\ufffe' # 0x8D -> UNDEFINED - u'\ufffe' # 0x8E -> UNDEFINED - u'\ufffe' # 0x8F -> UNDEFINED - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\u02dc' # 0x98 -> SMALL TILDE - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\u0161' # 0x9A -> LATIN SMALL LETTER S WITH CARON - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u0153' # 0x9C -> LATIN SMALL LIGATURE OE - u'\ufffe' # 0x9D -> UNDEFINED - u'\ufffe' # 0x9E -> UNDEFINED - u'\u0178' # 0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u011e' # 0xD0 -> LATIN CAPITAL LETTER G WITH BREVE - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u0130' # 0xDD -> LATIN CAPITAL LETTER I WITH DOT ABOVE - u'\u015e' # 0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u011f' # 0xF0 -> LATIN SMALL LETTER G WITH BREVE - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u0131' # 0xFD -> LATIN SMALL LETTER DOTLESS I - u'\u015f' # 0xFE -> LATIN SMALL LETTER S WITH CEDILLA - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u2030' # 0x89 -> PER MILLE SIGN + '\u0160' # 0x8A -> LATIN CAPITAL LETTER S WITH CARON + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE + '\ufffe' # 0x8D -> UNDEFINED + '\ufffe' # 0x8E -> UNDEFINED + '\ufffe' # 0x8F -> UNDEFINED + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\u02dc' # 0x98 -> SMALL TILDE + '\u2122' # 0x99 -> TRADE MARK SIGN + '\u0161' # 0x9A -> LATIN SMALL LETTER S WITH CARON + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u0153' # 0x9C -> LATIN SMALL LIGATURE OE + '\ufffe' # 0x9D -> UNDEFINED + '\ufffe' # 0x9E -> UNDEFINED + '\u0178' # 0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u011e' # 0xD0 -> LATIN CAPITAL LETTER G WITH BREVE + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u0130' # 0xDD -> LATIN CAPITAL LETTER I WITH DOT ABOVE + '\u015e' # 0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\u011f' # 0xF0 -> LATIN SMALL LETTER G WITH BREVE + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u0131' # 0xFD -> LATIN SMALL LETTER DOTLESS I + '\u015f' # 0xFE -> LATIN SMALL LETTER S WITH CEDILLA + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/cp1255.py b/Lib/encodings/cp1255.py index fd1456fa..91ce26b 100644 --- a/Lib/encodings/cp1255.py +++ b/Lib/encodings/cp1255.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\ufffe' # 0x8A -> UNDEFINED - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\ufffe' # 0x8C -> UNDEFINED - u'\ufffe' # 0x8D -> UNDEFINED - u'\ufffe' # 0x8E -> UNDEFINED - u'\ufffe' # 0x8F -> UNDEFINED - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\u02dc' # 0x98 -> SMALL TILDE - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\ufffe' # 0x9A -> UNDEFINED - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\ufffe' # 0x9C -> UNDEFINED - u'\ufffe' # 0x9D -> UNDEFINED - u'\ufffe' # 0x9E -> UNDEFINED - u'\ufffe' # 0x9F -> UNDEFINED - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\u20aa' # 0xA4 -> NEW SHEQEL SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xd7' # 0xAA -> MULTIPLICATION SIGN - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xf7' # 0xBA -> DIVISION SIGN - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\u05b0' # 0xC0 -> HEBREW POINT SHEVA - u'\u05b1' # 0xC1 -> HEBREW POINT HATAF SEGOL - u'\u05b2' # 0xC2 -> HEBREW POINT HATAF PATAH - u'\u05b3' # 0xC3 -> HEBREW POINT HATAF QAMATS - u'\u05b4' # 0xC4 -> HEBREW POINT HIRIQ - u'\u05b5' # 0xC5 -> HEBREW POINT TSERE - u'\u05b6' # 0xC6 -> HEBREW POINT SEGOL - u'\u05b7' # 0xC7 -> HEBREW POINT PATAH - u'\u05b8' # 0xC8 -> HEBREW POINT QAMATS - u'\u05b9' # 0xC9 -> HEBREW POINT HOLAM - u'\ufffe' # 0xCA -> UNDEFINED - u'\u05bb' # 0xCB -> HEBREW POINT QUBUTS - u'\u05bc' # 0xCC -> HEBREW POINT DAGESH OR MAPIQ - u'\u05bd' # 0xCD -> HEBREW POINT METEG - u'\u05be' # 0xCE -> HEBREW PUNCTUATION MAQAF - u'\u05bf' # 0xCF -> HEBREW POINT RAFE - u'\u05c0' # 0xD0 -> HEBREW PUNCTUATION PASEQ - u'\u05c1' # 0xD1 -> HEBREW POINT SHIN DOT - u'\u05c2' # 0xD2 -> HEBREW POINT SIN DOT - u'\u05c3' # 0xD3 -> HEBREW PUNCTUATION SOF PASUQ - u'\u05f0' # 0xD4 -> HEBREW LIGATURE YIDDISH DOUBLE VAV - u'\u05f1' # 0xD5 -> HEBREW LIGATURE YIDDISH VAV YOD - u'\u05f2' # 0xD6 -> HEBREW LIGATURE YIDDISH DOUBLE YOD - u'\u05f3' # 0xD7 -> HEBREW PUNCTUATION GERESH - u'\u05f4' # 0xD8 -> HEBREW PUNCTUATION GERSHAYIM - u'\ufffe' # 0xD9 -> UNDEFINED - u'\ufffe' # 0xDA -> UNDEFINED - u'\ufffe' # 0xDB -> UNDEFINED - u'\ufffe' # 0xDC -> UNDEFINED - u'\ufffe' # 0xDD -> UNDEFINED - u'\ufffe' # 0xDE -> UNDEFINED - u'\ufffe' # 0xDF -> UNDEFINED - u'\u05d0' # 0xE0 -> HEBREW LETTER ALEF - u'\u05d1' # 0xE1 -> HEBREW LETTER BET - u'\u05d2' # 0xE2 -> HEBREW LETTER GIMEL - u'\u05d3' # 0xE3 -> HEBREW LETTER DALET - u'\u05d4' # 0xE4 -> HEBREW LETTER HE - u'\u05d5' # 0xE5 -> HEBREW LETTER VAV - u'\u05d6' # 0xE6 -> HEBREW LETTER ZAYIN - u'\u05d7' # 0xE7 -> HEBREW LETTER HET - u'\u05d8' # 0xE8 -> HEBREW LETTER TET - u'\u05d9' # 0xE9 -> HEBREW LETTER YOD - u'\u05da' # 0xEA -> HEBREW LETTER FINAL KAF - u'\u05db' # 0xEB -> HEBREW LETTER KAF - u'\u05dc' # 0xEC -> HEBREW LETTER LAMED - u'\u05dd' # 0xED -> HEBREW LETTER FINAL MEM - u'\u05de' # 0xEE -> HEBREW LETTER MEM - u'\u05df' # 0xEF -> HEBREW LETTER FINAL NUN - u'\u05e0' # 0xF0 -> HEBREW LETTER NUN - u'\u05e1' # 0xF1 -> HEBREW LETTER SAMEKH - u'\u05e2' # 0xF2 -> HEBREW LETTER AYIN - u'\u05e3' # 0xF3 -> HEBREW LETTER FINAL PE - u'\u05e4' # 0xF4 -> HEBREW LETTER PE - u'\u05e5' # 0xF5 -> HEBREW LETTER FINAL TSADI - u'\u05e6' # 0xF6 -> HEBREW LETTER TSADI - u'\u05e7' # 0xF7 -> HEBREW LETTER QOF - u'\u05e8' # 0xF8 -> HEBREW LETTER RESH - u'\u05e9' # 0xF9 -> HEBREW LETTER SHIN - u'\u05ea' # 0xFA -> HEBREW LETTER TAV - u'\ufffe' # 0xFB -> UNDEFINED - u'\ufffe' # 0xFC -> UNDEFINED - u'\u200e' # 0xFD -> LEFT-TO-RIGHT MARK - u'\u200f' # 0xFE -> RIGHT-TO-LEFT MARK - u'\ufffe' # 0xFF -> UNDEFINED + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u2030' # 0x89 -> PER MILLE SIGN + '\ufffe' # 0x8A -> UNDEFINED + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\ufffe' # 0x8C -> UNDEFINED + '\ufffe' # 0x8D -> UNDEFINED + '\ufffe' # 0x8E -> UNDEFINED + '\ufffe' # 0x8F -> UNDEFINED + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\u02dc' # 0x98 -> SMALL TILDE + '\u2122' # 0x99 -> TRADE MARK SIGN + '\ufffe' # 0x9A -> UNDEFINED + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\ufffe' # 0x9C -> UNDEFINED + '\ufffe' # 0x9D -> UNDEFINED + '\ufffe' # 0x9E -> UNDEFINED + '\ufffe' # 0x9F -> UNDEFINED + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\u20aa' # 0xA4 -> NEW SHEQEL SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xd7' # 0xAA -> MULTIPLICATION SIGN + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xf7' # 0xBA -> DIVISION SIGN + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\u05b0' # 0xC0 -> HEBREW POINT SHEVA + '\u05b1' # 0xC1 -> HEBREW POINT HATAF SEGOL + '\u05b2' # 0xC2 -> HEBREW POINT HATAF PATAH + '\u05b3' # 0xC3 -> HEBREW POINT HATAF QAMATS + '\u05b4' # 0xC4 -> HEBREW POINT HIRIQ + '\u05b5' # 0xC5 -> HEBREW POINT TSERE + '\u05b6' # 0xC6 -> HEBREW POINT SEGOL + '\u05b7' # 0xC7 -> HEBREW POINT PATAH + '\u05b8' # 0xC8 -> HEBREW POINT QAMATS + '\u05b9' # 0xC9 -> HEBREW POINT HOLAM + '\ufffe' # 0xCA -> UNDEFINED + '\u05bb' # 0xCB -> HEBREW POINT QUBUTS + '\u05bc' # 0xCC -> HEBREW POINT DAGESH OR MAPIQ + '\u05bd' # 0xCD -> HEBREW POINT METEG + '\u05be' # 0xCE -> HEBREW PUNCTUATION MAQAF + '\u05bf' # 0xCF -> HEBREW POINT RAFE + '\u05c0' # 0xD0 -> HEBREW PUNCTUATION PASEQ + '\u05c1' # 0xD1 -> HEBREW POINT SHIN DOT + '\u05c2' # 0xD2 -> HEBREW POINT SIN DOT + '\u05c3' # 0xD3 -> HEBREW PUNCTUATION SOF PASUQ + '\u05f0' # 0xD4 -> HEBREW LIGATURE YIDDISH DOUBLE VAV + '\u05f1' # 0xD5 -> HEBREW LIGATURE YIDDISH VAV YOD + '\u05f2' # 0xD6 -> HEBREW LIGATURE YIDDISH DOUBLE YOD + '\u05f3' # 0xD7 -> HEBREW PUNCTUATION GERESH + '\u05f4' # 0xD8 -> HEBREW PUNCTUATION GERSHAYIM + '\ufffe' # 0xD9 -> UNDEFINED + '\ufffe' # 0xDA -> UNDEFINED + '\ufffe' # 0xDB -> UNDEFINED + '\ufffe' # 0xDC -> UNDEFINED + '\ufffe' # 0xDD -> UNDEFINED + '\ufffe' # 0xDE -> UNDEFINED + '\ufffe' # 0xDF -> UNDEFINED + '\u05d0' # 0xE0 -> HEBREW LETTER ALEF + '\u05d1' # 0xE1 -> HEBREW LETTER BET + '\u05d2' # 0xE2 -> HEBREW LETTER GIMEL + '\u05d3' # 0xE3 -> HEBREW LETTER DALET + '\u05d4' # 0xE4 -> HEBREW LETTER HE + '\u05d5' # 0xE5 -> HEBREW LETTER VAV + '\u05d6' # 0xE6 -> HEBREW LETTER ZAYIN + '\u05d7' # 0xE7 -> HEBREW LETTER HET + '\u05d8' # 0xE8 -> HEBREW LETTER TET + '\u05d9' # 0xE9 -> HEBREW LETTER YOD + '\u05da' # 0xEA -> HEBREW LETTER FINAL KAF + '\u05db' # 0xEB -> HEBREW LETTER KAF + '\u05dc' # 0xEC -> HEBREW LETTER LAMED + '\u05dd' # 0xED -> HEBREW LETTER FINAL MEM + '\u05de' # 0xEE -> HEBREW LETTER MEM + '\u05df' # 0xEF -> HEBREW LETTER FINAL NUN + '\u05e0' # 0xF0 -> HEBREW LETTER NUN + '\u05e1' # 0xF1 -> HEBREW LETTER SAMEKH + '\u05e2' # 0xF2 -> HEBREW LETTER AYIN + '\u05e3' # 0xF3 -> HEBREW LETTER FINAL PE + '\u05e4' # 0xF4 -> HEBREW LETTER PE + '\u05e5' # 0xF5 -> HEBREW LETTER FINAL TSADI + '\u05e6' # 0xF6 -> HEBREW LETTER TSADI + '\u05e7' # 0xF7 -> HEBREW LETTER QOF + '\u05e8' # 0xF8 -> HEBREW LETTER RESH + '\u05e9' # 0xF9 -> HEBREW LETTER SHIN + '\u05ea' # 0xFA -> HEBREW LETTER TAV + '\ufffe' # 0xFB -> UNDEFINED + '\ufffe' # 0xFC -> UNDEFINED + '\u200e' # 0xFD -> LEFT-TO-RIGHT MARK + '\u200f' # 0xFE -> RIGHT-TO-LEFT MARK + '\ufffe' # 0xFF -> UNDEFINED ) ### Encoding table diff --git a/Lib/encodings/cp1256.py b/Lib/encodings/cp1256.py index 302b5fa..fd6afab 100644 --- a/Lib/encodings/cp1256.py +++ b/Lib/encodings/cp1256.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\u067e' # 0x81 -> ARABIC LETTER PEH - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\u0679' # 0x8A -> ARABIC LETTER TTEH - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE - u'\u0686' # 0x8D -> ARABIC LETTER TCHEH - u'\u0698' # 0x8E -> ARABIC LETTER JEH - u'\u0688' # 0x8F -> ARABIC LETTER DDAL - u'\u06af' # 0x90 -> ARABIC LETTER GAF - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\u06a9' # 0x98 -> ARABIC LETTER KEHEH - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\u0691' # 0x9A -> ARABIC LETTER RREH - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u0153' # 0x9C -> LATIN SMALL LIGATURE OE - u'\u200c' # 0x9D -> ZERO WIDTH NON-JOINER - u'\u200d' # 0x9E -> ZERO WIDTH JOINER - u'\u06ba' # 0x9F -> ARABIC LETTER NOON GHUNNA - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u060c' # 0xA1 -> ARABIC COMMA - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u06be' # 0xAA -> ARABIC LETTER HEH DOACHASHMEE - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\u061b' # 0xBA -> ARABIC SEMICOLON - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\u061f' # 0xBF -> ARABIC QUESTION MARK - u'\u06c1' # 0xC0 -> ARABIC LETTER HEH GOAL - u'\u0621' # 0xC1 -> ARABIC LETTER HAMZA - u'\u0622' # 0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE - u'\u0623' # 0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE - u'\u0624' # 0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE - u'\u0625' # 0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW - u'\u0626' # 0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE - u'\u0627' # 0xC7 -> ARABIC LETTER ALEF - u'\u0628' # 0xC8 -> ARABIC LETTER BEH - u'\u0629' # 0xC9 -> ARABIC LETTER TEH MARBUTA - u'\u062a' # 0xCA -> ARABIC LETTER TEH - u'\u062b' # 0xCB -> ARABIC LETTER THEH - u'\u062c' # 0xCC -> ARABIC LETTER JEEM - u'\u062d' # 0xCD -> ARABIC LETTER HAH - u'\u062e' # 0xCE -> ARABIC LETTER KHAH - u'\u062f' # 0xCF -> ARABIC LETTER DAL - u'\u0630' # 0xD0 -> ARABIC LETTER THAL - u'\u0631' # 0xD1 -> ARABIC LETTER REH - u'\u0632' # 0xD2 -> ARABIC LETTER ZAIN - u'\u0633' # 0xD3 -> ARABIC LETTER SEEN - u'\u0634' # 0xD4 -> ARABIC LETTER SHEEN - u'\u0635' # 0xD5 -> ARABIC LETTER SAD - u'\u0636' # 0xD6 -> ARABIC LETTER DAD - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\u0637' # 0xD8 -> ARABIC LETTER TAH - u'\u0638' # 0xD9 -> ARABIC LETTER ZAH - u'\u0639' # 0xDA -> ARABIC LETTER AIN - u'\u063a' # 0xDB -> ARABIC LETTER GHAIN - u'\u0640' # 0xDC -> ARABIC TATWEEL - u'\u0641' # 0xDD -> ARABIC LETTER FEH - u'\u0642' # 0xDE -> ARABIC LETTER QAF - u'\u0643' # 0xDF -> ARABIC LETTER KAF - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\u0644' # 0xE1 -> ARABIC LETTER LAM - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\u0645' # 0xE3 -> ARABIC LETTER MEEM - u'\u0646' # 0xE4 -> ARABIC LETTER NOON - u'\u0647' # 0xE5 -> ARABIC LETTER HEH - u'\u0648' # 0xE6 -> ARABIC LETTER WAW - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u0649' # 0xEC -> ARABIC LETTER ALEF MAKSURA - u'\u064a' # 0xED -> ARABIC LETTER YEH - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u064b' # 0xF0 -> ARABIC FATHATAN - u'\u064c' # 0xF1 -> ARABIC DAMMATAN - u'\u064d' # 0xF2 -> ARABIC KASRATAN - u'\u064e' # 0xF3 -> ARABIC FATHA - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\u064f' # 0xF5 -> ARABIC DAMMA - u'\u0650' # 0xF6 -> ARABIC KASRA - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\u0651' # 0xF8 -> ARABIC SHADDA - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\u0652' # 0xFA -> ARABIC SUKUN - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u200e' # 0xFD -> LEFT-TO-RIGHT MARK - u'\u200f' # 0xFE -> RIGHT-TO-LEFT MARK - u'\u06d2' # 0xFF -> ARABIC LETTER YEH BARREE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\u067e' # 0x81 -> ARABIC LETTER PEH + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u2030' # 0x89 -> PER MILLE SIGN + '\u0679' # 0x8A -> ARABIC LETTER TTEH + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE + '\u0686' # 0x8D -> ARABIC LETTER TCHEH + '\u0698' # 0x8E -> ARABIC LETTER JEH + '\u0688' # 0x8F -> ARABIC LETTER DDAL + '\u06af' # 0x90 -> ARABIC LETTER GAF + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\u06a9' # 0x98 -> ARABIC LETTER KEHEH + '\u2122' # 0x99 -> TRADE MARK SIGN + '\u0691' # 0x9A -> ARABIC LETTER RREH + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u0153' # 0x9C -> LATIN SMALL LIGATURE OE + '\u200c' # 0x9D -> ZERO WIDTH NON-JOINER + '\u200d' # 0x9E -> ZERO WIDTH JOINER + '\u06ba' # 0x9F -> ARABIC LETTER NOON GHUNNA + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u060c' # 0xA1 -> ARABIC COMMA + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u06be' # 0xAA -> ARABIC LETTER HEH DOACHASHMEE + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\u061b' # 0xBA -> ARABIC SEMICOLON + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\u061f' # 0xBF -> ARABIC QUESTION MARK + '\u06c1' # 0xC0 -> ARABIC LETTER HEH GOAL + '\u0621' # 0xC1 -> ARABIC LETTER HAMZA + '\u0622' # 0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE + '\u0623' # 0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE + '\u0624' # 0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE + '\u0625' # 0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW + '\u0626' # 0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE + '\u0627' # 0xC7 -> ARABIC LETTER ALEF + '\u0628' # 0xC8 -> ARABIC LETTER BEH + '\u0629' # 0xC9 -> ARABIC LETTER TEH MARBUTA + '\u062a' # 0xCA -> ARABIC LETTER TEH + '\u062b' # 0xCB -> ARABIC LETTER THEH + '\u062c' # 0xCC -> ARABIC LETTER JEEM + '\u062d' # 0xCD -> ARABIC LETTER HAH + '\u062e' # 0xCE -> ARABIC LETTER KHAH + '\u062f' # 0xCF -> ARABIC LETTER DAL + '\u0630' # 0xD0 -> ARABIC LETTER THAL + '\u0631' # 0xD1 -> ARABIC LETTER REH + '\u0632' # 0xD2 -> ARABIC LETTER ZAIN + '\u0633' # 0xD3 -> ARABIC LETTER SEEN + '\u0634' # 0xD4 -> ARABIC LETTER SHEEN + '\u0635' # 0xD5 -> ARABIC LETTER SAD + '\u0636' # 0xD6 -> ARABIC LETTER DAD + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\u0637' # 0xD8 -> ARABIC LETTER TAH + '\u0638' # 0xD9 -> ARABIC LETTER ZAH + '\u0639' # 0xDA -> ARABIC LETTER AIN + '\u063a' # 0xDB -> ARABIC LETTER GHAIN + '\u0640' # 0xDC -> ARABIC TATWEEL + '\u0641' # 0xDD -> ARABIC LETTER FEH + '\u0642' # 0xDE -> ARABIC LETTER QAF + '\u0643' # 0xDF -> ARABIC LETTER KAF + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\u0644' # 0xE1 -> ARABIC LETTER LAM + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\u0645' # 0xE3 -> ARABIC LETTER MEEM + '\u0646' # 0xE4 -> ARABIC LETTER NOON + '\u0647' # 0xE5 -> ARABIC LETTER HEH + '\u0648' # 0xE6 -> ARABIC LETTER WAW + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\u0649' # 0xEC -> ARABIC LETTER ALEF MAKSURA + '\u064a' # 0xED -> ARABIC LETTER YEH + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\u064b' # 0xF0 -> ARABIC FATHATAN + '\u064c' # 0xF1 -> ARABIC DAMMATAN + '\u064d' # 0xF2 -> ARABIC KASRATAN + '\u064e' # 0xF3 -> ARABIC FATHA + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\u064f' # 0xF5 -> ARABIC DAMMA + '\u0650' # 0xF6 -> ARABIC KASRA + '\xf7' # 0xF7 -> DIVISION SIGN + '\u0651' # 0xF8 -> ARABIC SHADDA + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\u0652' # 0xFA -> ARABIC SUKUN + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u200e' # 0xFD -> LEFT-TO-RIGHT MARK + '\u200f' # 0xFE -> RIGHT-TO-LEFT MARK + '\u06d2' # 0xFF -> ARABIC LETTER YEH BARREE ) ### Encoding table diff --git a/Lib/encodings/cp1257.py b/Lib/encodings/cp1257.py index 53a6b29..9ebc90d 100644 --- a/Lib/encodings/cp1257.py +++ b/Lib/encodings/cp1257.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\ufffe' # 0x83 -> UNDEFINED - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\ufffe' # 0x88 -> UNDEFINED - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\ufffe' # 0x8A -> UNDEFINED - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\ufffe' # 0x8C -> UNDEFINED - u'\xa8' # 0x8D -> DIAERESIS - u'\u02c7' # 0x8E -> CARON - u'\xb8' # 0x8F -> CEDILLA - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\ufffe' # 0x98 -> UNDEFINED - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\ufffe' # 0x9A -> UNDEFINED - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\ufffe' # 0x9C -> UNDEFINED - u'\xaf' # 0x9D -> MACRON - u'\u02db' # 0x9E -> OGONEK - u'\ufffe' # 0x9F -> UNDEFINED - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\ufffe' # 0xA1 -> UNDEFINED - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\ufffe' # 0xA5 -> UNDEFINED - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xd8' # 0xA8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u0156' # 0xAA -> LATIN CAPITAL LETTER R WITH CEDILLA - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xc6' # 0xAF -> LATIN CAPITAL LETTER AE - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xf8' # 0xB8 -> LATIN SMALL LETTER O WITH STROKE - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\u0157' # 0xBA -> LATIN SMALL LETTER R WITH CEDILLA - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xe6' # 0xBF -> LATIN SMALL LETTER AE - u'\u0104' # 0xC0 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u012e' # 0xC1 -> LATIN CAPITAL LETTER I WITH OGONEK - u'\u0100' # 0xC2 -> LATIN CAPITAL LETTER A WITH MACRON - u'\u0106' # 0xC3 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\u0118' # 0xC6 -> LATIN CAPITAL LETTER E WITH OGONEK - u'\u0112' # 0xC7 -> LATIN CAPITAL LETTER E WITH MACRON - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0179' # 0xCA -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\u0116' # 0xCB -> LATIN CAPITAL LETTER E WITH DOT ABOVE - u'\u0122' # 0xCC -> LATIN CAPITAL LETTER G WITH CEDILLA - u'\u0136' # 0xCD -> LATIN CAPITAL LETTER K WITH CEDILLA - u'\u012a' # 0xCE -> LATIN CAPITAL LETTER I WITH MACRON - u'\u013b' # 0xCF -> LATIN CAPITAL LETTER L WITH CEDILLA - u'\u0160' # 0xD0 -> LATIN CAPITAL LETTER S WITH CARON - u'\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\u0145' # 0xD2 -> LATIN CAPITAL LETTER N WITH CEDILLA - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\u014c' # 0xD4 -> LATIN CAPITAL LETTER O WITH MACRON - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\u0172' # 0xD8 -> LATIN CAPITAL LETTER U WITH OGONEK - u'\u0141' # 0xD9 -> LATIN CAPITAL LETTER L WITH STROKE - u'\u015a' # 0xDA -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u016a' # 0xDB -> LATIN CAPITAL LETTER U WITH MACRON - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u017b' # 0xDD -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\u017d' # 0xDE -> LATIN CAPITAL LETTER Z WITH CARON - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\u0105' # 0xE0 -> LATIN SMALL LETTER A WITH OGONEK - u'\u012f' # 0xE1 -> LATIN SMALL LETTER I WITH OGONEK - u'\u0101' # 0xE2 -> LATIN SMALL LETTER A WITH MACRON - u'\u0107' # 0xE3 -> LATIN SMALL LETTER C WITH ACUTE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\u0119' # 0xE6 -> LATIN SMALL LETTER E WITH OGONEK - u'\u0113' # 0xE7 -> LATIN SMALL LETTER E WITH MACRON - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\u017a' # 0xEA -> LATIN SMALL LETTER Z WITH ACUTE - u'\u0117' # 0xEB -> LATIN SMALL LETTER E WITH DOT ABOVE - u'\u0123' # 0xEC -> LATIN SMALL LETTER G WITH CEDILLA - u'\u0137' # 0xED -> LATIN SMALL LETTER K WITH CEDILLA - u'\u012b' # 0xEE -> LATIN SMALL LETTER I WITH MACRON - u'\u013c' # 0xEF -> LATIN SMALL LETTER L WITH CEDILLA - u'\u0161' # 0xF0 -> LATIN SMALL LETTER S WITH CARON - u'\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0146' # 0xF2 -> LATIN SMALL LETTER N WITH CEDILLA - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\u014d' # 0xF4 -> LATIN SMALL LETTER O WITH MACRON - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\u0173' # 0xF8 -> LATIN SMALL LETTER U WITH OGONEK - u'\u0142' # 0xF9 -> LATIN SMALL LETTER L WITH STROKE - u'\u015b' # 0xFA -> LATIN SMALL LETTER S WITH ACUTE - u'\u016b' # 0xFB -> LATIN SMALL LETTER U WITH MACRON - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u017c' # 0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u017e' # 0xFE -> LATIN SMALL LETTER Z WITH CARON - u'\u02d9' # 0xFF -> DOT ABOVE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\ufffe' # 0x83 -> UNDEFINED + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\ufffe' # 0x88 -> UNDEFINED + '\u2030' # 0x89 -> PER MILLE SIGN + '\ufffe' # 0x8A -> UNDEFINED + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\ufffe' # 0x8C -> UNDEFINED + '\xa8' # 0x8D -> DIAERESIS + '\u02c7' # 0x8E -> CARON + '\xb8' # 0x8F -> CEDILLA + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\ufffe' # 0x98 -> UNDEFINED + '\u2122' # 0x99 -> TRADE MARK SIGN + '\ufffe' # 0x9A -> UNDEFINED + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\ufffe' # 0x9C -> UNDEFINED + '\xaf' # 0x9D -> MACRON + '\u02db' # 0x9E -> OGONEK + '\ufffe' # 0x9F -> UNDEFINED + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\ufffe' # 0xA1 -> UNDEFINED + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\ufffe' # 0xA5 -> UNDEFINED + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xd8' # 0xA8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u0156' # 0xAA -> LATIN CAPITAL LETTER R WITH CEDILLA + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xc6' # 0xAF -> LATIN CAPITAL LETTER AE + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xf8' # 0xB8 -> LATIN SMALL LETTER O WITH STROKE + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\u0157' # 0xBA -> LATIN SMALL LETTER R WITH CEDILLA + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xe6' # 0xBF -> LATIN SMALL LETTER AE + '\u0104' # 0xC0 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u012e' # 0xC1 -> LATIN CAPITAL LETTER I WITH OGONEK + '\u0100' # 0xC2 -> LATIN CAPITAL LETTER A WITH MACRON + '\u0106' # 0xC3 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\u0118' # 0xC6 -> LATIN CAPITAL LETTER E WITH OGONEK + '\u0112' # 0xC7 -> LATIN CAPITAL LETTER E WITH MACRON + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0179' # 0xCA -> LATIN CAPITAL LETTER Z WITH ACUTE + '\u0116' # 0xCB -> LATIN CAPITAL LETTER E WITH DOT ABOVE + '\u0122' # 0xCC -> LATIN CAPITAL LETTER G WITH CEDILLA + '\u0136' # 0xCD -> LATIN CAPITAL LETTER K WITH CEDILLA + '\u012a' # 0xCE -> LATIN CAPITAL LETTER I WITH MACRON + '\u013b' # 0xCF -> LATIN CAPITAL LETTER L WITH CEDILLA + '\u0160' # 0xD0 -> LATIN CAPITAL LETTER S WITH CARON + '\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE + '\u0145' # 0xD2 -> LATIN CAPITAL LETTER N WITH CEDILLA + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\u014c' # 0xD4 -> LATIN CAPITAL LETTER O WITH MACRON + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\u0172' # 0xD8 -> LATIN CAPITAL LETTER U WITH OGONEK + '\u0141' # 0xD9 -> LATIN CAPITAL LETTER L WITH STROKE + '\u015a' # 0xDA -> LATIN CAPITAL LETTER S WITH ACUTE + '\u016a' # 0xDB -> LATIN CAPITAL LETTER U WITH MACRON + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u017b' # 0xDD -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\u017d' # 0xDE -> LATIN CAPITAL LETTER Z WITH CARON + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\u0105' # 0xE0 -> LATIN SMALL LETTER A WITH OGONEK + '\u012f' # 0xE1 -> LATIN SMALL LETTER I WITH OGONEK + '\u0101' # 0xE2 -> LATIN SMALL LETTER A WITH MACRON + '\u0107' # 0xE3 -> LATIN SMALL LETTER C WITH ACUTE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\u0119' # 0xE6 -> LATIN SMALL LETTER E WITH OGONEK + '\u0113' # 0xE7 -> LATIN SMALL LETTER E WITH MACRON + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\u017a' # 0xEA -> LATIN SMALL LETTER Z WITH ACUTE + '\u0117' # 0xEB -> LATIN SMALL LETTER E WITH DOT ABOVE + '\u0123' # 0xEC -> LATIN SMALL LETTER G WITH CEDILLA + '\u0137' # 0xED -> LATIN SMALL LETTER K WITH CEDILLA + '\u012b' # 0xEE -> LATIN SMALL LETTER I WITH MACRON + '\u013c' # 0xEF -> LATIN SMALL LETTER L WITH CEDILLA + '\u0161' # 0xF0 -> LATIN SMALL LETTER S WITH CARON + '\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE + '\u0146' # 0xF2 -> LATIN SMALL LETTER N WITH CEDILLA + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\u014d' # 0xF4 -> LATIN SMALL LETTER O WITH MACRON + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\u0173' # 0xF8 -> LATIN SMALL LETTER U WITH OGONEK + '\u0142' # 0xF9 -> LATIN SMALL LETTER L WITH STROKE + '\u015b' # 0xFA -> LATIN SMALL LETTER S WITH ACUTE + '\u016b' # 0xFB -> LATIN SMALL LETTER U WITH MACRON + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u017c' # 0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u017e' # 0xFE -> LATIN SMALL LETTER Z WITH CARON + '\u02d9' # 0xFF -> DOT ABOVE ) ### Encoding table diff --git a/Lib/encodings/cp1258.py b/Lib/encodings/cp1258.py index 4b25d8e..784378a 100644 --- a/Lib/encodings/cp1258.py +++ b/Lib/encodings/cp1258.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK - u'\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK - u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\u2020' # 0x86 -> DAGGER - u'\u2021' # 0x87 -> DOUBLE DAGGER - u'\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u2030' # 0x89 -> PER MILLE SIGN - u'\ufffe' # 0x8A -> UNDEFINED - u'\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE - u'\ufffe' # 0x8D -> UNDEFINED - u'\ufffe' # 0x8E -> UNDEFINED - u'\ufffe' # 0x8F -> UNDEFINED - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\u02dc' # 0x98 -> SMALL TILDE - u'\u2122' # 0x99 -> TRADE MARK SIGN - u'\ufffe' # 0x9A -> UNDEFINED - u'\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u0153' # 0x9C -> LATIN SMALL LIGATURE OE - u'\ufffe' # 0x9D -> UNDEFINED - u'\ufffe' # 0x9E -> UNDEFINED - u'\u0178' # 0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u0300' # 0xCC -> COMBINING GRAVE ACCENT - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\u0309' # 0xD2 -> COMBINING HOOK ABOVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u01a0' # 0xD5 -> LATIN CAPITAL LETTER O WITH HORN - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u01af' # 0xDD -> LATIN CAPITAL LETTER U WITH HORN - u'\u0303' # 0xDE -> COMBINING TILDE - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u0301' # 0xEC -> COMBINING ACUTE ACCENT - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\u0323' # 0xF2 -> COMBINING DOT BELOW - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\u01a1' # 0xF5 -> LATIN SMALL LETTER O WITH HORN - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u01b0' # 0xFD -> LATIN SMALL LETTER U WITH HORN - u'\u20ab' # 0xFE -> DONG SIGN - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK + '\u0192' # 0x83 -> LATIN SMALL LETTER F WITH HOOK + '\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\u2020' # 0x86 -> DAGGER + '\u2021' # 0x87 -> DOUBLE DAGGER + '\u02c6' # 0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u2030' # 0x89 -> PER MILLE SIGN + '\ufffe' # 0x8A -> UNDEFINED + '\u2039' # 0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u0152' # 0x8C -> LATIN CAPITAL LIGATURE OE + '\ufffe' # 0x8D -> UNDEFINED + '\ufffe' # 0x8E -> UNDEFINED + '\ufffe' # 0x8F -> UNDEFINED + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\u02dc' # 0x98 -> SMALL TILDE + '\u2122' # 0x99 -> TRADE MARK SIGN + '\ufffe' # 0x9A -> UNDEFINED + '\u203a' # 0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u0153' # 0x9C -> LATIN SMALL LIGATURE OE + '\ufffe' # 0x9D -> UNDEFINED + '\ufffe' # 0x9E -> UNDEFINED + '\u0178' # 0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u0300' # 0xCC -> COMBINING GRAVE ACCENT + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\u0309' # 0xD2 -> COMBINING HOOK ABOVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u01a0' # 0xD5 -> LATIN CAPITAL LETTER O WITH HORN + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u01af' # 0xDD -> LATIN CAPITAL LETTER U WITH HORN + '\u0303' # 0xDE -> COMBINING TILDE + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\u0301' # 0xEC -> COMBINING ACUTE ACCENT + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\u0323' # 0xF2 -> COMBINING DOT BELOW + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\u01a1' # 0xF5 -> LATIN SMALL LETTER O WITH HORN + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u01b0' # 0xFD -> LATIN SMALL LETTER U WITH HORN + '\u20ab' # 0xFE -> DONG SIGN + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/cp424.py b/Lib/encodings/cp424.py index d3ade22..6753daf 100644 --- a/Lib/encodings/cp424.py +++ b/Lib/encodings/cp424.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x9c' # 0x04 -> SELECT - u'\t' # 0x05 -> HORIZONTAL TABULATION - u'\x86' # 0x06 -> REQUIRED NEW LINE - u'\x7f' # 0x07 -> DELETE - u'\x97' # 0x08 -> GRAPHIC ESCAPE - u'\x8d' # 0x09 -> SUPERSCRIPT - u'\x8e' # 0x0A -> REPEAT - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x9d' # 0x14 -> RESTORE/ENABLE PRESENTATION - u'\x85' # 0x15 -> NEW LINE - u'\x08' # 0x16 -> BACKSPACE - u'\x87' # 0x17 -> PROGRAM OPERATOR COMMUNICATION - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x92' # 0x1A -> UNIT BACK SPACE - u'\x8f' # 0x1B -> CUSTOMER USE ONE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u'\x80' # 0x20 -> DIGIT SELECT - u'\x81' # 0x21 -> START OF SIGNIFICANCE - u'\x82' # 0x22 -> FIELD SEPARATOR - u'\x83' # 0x23 -> WORD UNDERSCORE - u'\x84' # 0x24 -> BYPASS OR INHIBIT PRESENTATION - u'\n' # 0x25 -> LINE FEED - u'\x17' # 0x26 -> END OF TRANSMISSION BLOCK - u'\x1b' # 0x27 -> ESCAPE - u'\x88' # 0x28 -> SET ATTRIBUTE - u'\x89' # 0x29 -> START FIELD EXTENDED - u'\x8a' # 0x2A -> SET MODE OR SWITCH - u'\x8b' # 0x2B -> CONTROL SEQUENCE PREFIX - u'\x8c' # 0x2C -> MODIFY FIELD ATTRIBUTE - u'\x05' # 0x2D -> ENQUIRY - u'\x06' # 0x2E -> ACKNOWLEDGE - u'\x07' # 0x2F -> BELL - u'\x90' # 0x30 -> <reserved> - u'\x91' # 0x31 -> <reserved> - u'\x16' # 0x32 -> SYNCHRONOUS IDLE - u'\x93' # 0x33 -> INDEX RETURN - u'\x94' # 0x34 -> PRESENTATION POSITION - u'\x95' # 0x35 -> TRANSPARENT - u'\x96' # 0x36 -> NUMERIC BACKSPACE - u'\x04' # 0x37 -> END OF TRANSMISSION - u'\x98' # 0x38 -> SUBSCRIPT - u'\x99' # 0x39 -> INDENT TABULATION - u'\x9a' # 0x3A -> REVERSE FORM FEED - u'\x9b' # 0x3B -> CUSTOMER USE THREE - u'\x14' # 0x3C -> DEVICE CONTROL FOUR - u'\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE - u'\x9e' # 0x3E -> <reserved> - u'\x1a' # 0x3F -> SUBSTITUTE - u' ' # 0x40 -> SPACE - u'\u05d0' # 0x41 -> HEBREW LETTER ALEF - u'\u05d1' # 0x42 -> HEBREW LETTER BET - u'\u05d2' # 0x43 -> HEBREW LETTER GIMEL - u'\u05d3' # 0x44 -> HEBREW LETTER DALET - u'\u05d4' # 0x45 -> HEBREW LETTER HE - u'\u05d5' # 0x46 -> HEBREW LETTER VAV - u'\u05d6' # 0x47 -> HEBREW LETTER ZAYIN - u'\u05d7' # 0x48 -> HEBREW LETTER HET - u'\u05d8' # 0x49 -> HEBREW LETTER TET - u'\xa2' # 0x4A -> CENT SIGN - u'.' # 0x4B -> FULL STOP - u'<' # 0x4C -> LESS-THAN SIGN - u'(' # 0x4D -> LEFT PARENTHESIS - u'+' # 0x4E -> PLUS SIGN - u'|' # 0x4F -> VERTICAL LINE - u'&' # 0x50 -> AMPERSAND - u'\u05d9' # 0x51 -> HEBREW LETTER YOD - u'\u05da' # 0x52 -> HEBREW LETTER FINAL KAF - u'\u05db' # 0x53 -> HEBREW LETTER KAF - u'\u05dc' # 0x54 -> HEBREW LETTER LAMED - u'\u05dd' # 0x55 -> HEBREW LETTER FINAL MEM - u'\u05de' # 0x56 -> HEBREW LETTER MEM - u'\u05df' # 0x57 -> HEBREW LETTER FINAL NUN - u'\u05e0' # 0x58 -> HEBREW LETTER NUN - u'\u05e1' # 0x59 -> HEBREW LETTER SAMEKH - u'!' # 0x5A -> EXCLAMATION MARK - u'$' # 0x5B -> DOLLAR SIGN - u'*' # 0x5C -> ASTERISK - u')' # 0x5D -> RIGHT PARENTHESIS - u';' # 0x5E -> SEMICOLON - u'\xac' # 0x5F -> NOT SIGN - u'-' # 0x60 -> HYPHEN-MINUS - u'/' # 0x61 -> SOLIDUS - u'\u05e2' # 0x62 -> HEBREW LETTER AYIN - u'\u05e3' # 0x63 -> HEBREW LETTER FINAL PE - u'\u05e4' # 0x64 -> HEBREW LETTER PE - u'\u05e5' # 0x65 -> HEBREW LETTER FINAL TSADI - u'\u05e6' # 0x66 -> HEBREW LETTER TSADI - u'\u05e7' # 0x67 -> HEBREW LETTER QOF - u'\u05e8' # 0x68 -> HEBREW LETTER RESH - u'\u05e9' # 0x69 -> HEBREW LETTER SHIN - u'\xa6' # 0x6A -> BROKEN BAR - u',' # 0x6B -> COMMA - u'%' # 0x6C -> PERCENT SIGN - u'_' # 0x6D -> LOW LINE - u'>' # 0x6E -> GREATER-THAN SIGN - u'?' # 0x6F -> QUESTION MARK - u'\ufffe' # 0x70 -> UNDEFINED - u'\u05ea' # 0x71 -> HEBREW LETTER TAV - u'\ufffe' # 0x72 -> UNDEFINED - u'\ufffe' # 0x73 -> UNDEFINED - u'\xa0' # 0x74 -> NO-BREAK SPACE - u'\ufffe' # 0x75 -> UNDEFINED - u'\ufffe' # 0x76 -> UNDEFINED - u'\ufffe' # 0x77 -> UNDEFINED - u'\u2017' # 0x78 -> DOUBLE LOW LINE - u'`' # 0x79 -> GRAVE ACCENT - u':' # 0x7A -> COLON - u'#' # 0x7B -> NUMBER SIGN - u'@' # 0x7C -> COMMERCIAL AT - u"'" # 0x7D -> APOSTROPHE - u'=' # 0x7E -> EQUALS SIGN - u'"' # 0x7F -> QUOTATION MARK - u'\ufffe' # 0x80 -> UNDEFINED - u'a' # 0x81 -> LATIN SMALL LETTER A - u'b' # 0x82 -> LATIN SMALL LETTER B - u'c' # 0x83 -> LATIN SMALL LETTER C - u'd' # 0x84 -> LATIN SMALL LETTER D - u'e' # 0x85 -> LATIN SMALL LETTER E - u'f' # 0x86 -> LATIN SMALL LETTER F - u'g' # 0x87 -> LATIN SMALL LETTER G - u'h' # 0x88 -> LATIN SMALL LETTER H - u'i' # 0x89 -> LATIN SMALL LETTER I - u'\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\ufffe' # 0x8C -> UNDEFINED - u'\ufffe' # 0x8D -> UNDEFINED - u'\ufffe' # 0x8E -> UNDEFINED - u'\xb1' # 0x8F -> PLUS-MINUS SIGN - u'\xb0' # 0x90 -> DEGREE SIGN - u'j' # 0x91 -> LATIN SMALL LETTER J - u'k' # 0x92 -> LATIN SMALL LETTER K - u'l' # 0x93 -> LATIN SMALL LETTER L - u'm' # 0x94 -> LATIN SMALL LETTER M - u'n' # 0x95 -> LATIN SMALL LETTER N - u'o' # 0x96 -> LATIN SMALL LETTER O - u'p' # 0x97 -> LATIN SMALL LETTER P - u'q' # 0x98 -> LATIN SMALL LETTER Q - u'r' # 0x99 -> LATIN SMALL LETTER R - u'\ufffe' # 0x9A -> UNDEFINED - u'\ufffe' # 0x9B -> UNDEFINED - u'\ufffe' # 0x9C -> UNDEFINED - u'\xb8' # 0x9D -> CEDILLA - u'\ufffe' # 0x9E -> UNDEFINED - u'\xa4' # 0x9F -> CURRENCY SIGN - u'\xb5' # 0xA0 -> MICRO SIGN - u'~' # 0xA1 -> TILDE - u's' # 0xA2 -> LATIN SMALL LETTER S - u't' # 0xA3 -> LATIN SMALL LETTER T - u'u' # 0xA4 -> LATIN SMALL LETTER U - u'v' # 0xA5 -> LATIN SMALL LETTER V - u'w' # 0xA6 -> LATIN SMALL LETTER W - u'x' # 0xA7 -> LATIN SMALL LETTER X - u'y' # 0xA8 -> LATIN SMALL LETTER Y - u'z' # 0xA9 -> LATIN SMALL LETTER Z - u'\ufffe' # 0xAA -> UNDEFINED - u'\ufffe' # 0xAB -> UNDEFINED - u'\ufffe' # 0xAC -> UNDEFINED - u'\ufffe' # 0xAD -> UNDEFINED - u'\ufffe' # 0xAE -> UNDEFINED - u'\xae' # 0xAF -> REGISTERED SIGN - u'^' # 0xB0 -> CIRCUMFLEX ACCENT - u'\xa3' # 0xB1 -> POUND SIGN - u'\xa5' # 0xB2 -> YEN SIGN - u'\xb7' # 0xB3 -> MIDDLE DOT - u'\xa9' # 0xB4 -> COPYRIGHT SIGN - u'\xa7' # 0xB5 -> SECTION SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS - u'[' # 0xBA -> LEFT SQUARE BRACKET - u']' # 0xBB -> RIGHT SQUARE BRACKET - u'\xaf' # 0xBC -> MACRON - u'\xa8' # 0xBD -> DIAERESIS - u'\xb4' # 0xBE -> ACUTE ACCENT - u'\xd7' # 0xBF -> MULTIPLICATION SIGN - u'{' # 0xC0 -> LEFT CURLY BRACKET - u'A' # 0xC1 -> LATIN CAPITAL LETTER A - u'B' # 0xC2 -> LATIN CAPITAL LETTER B - u'C' # 0xC3 -> LATIN CAPITAL LETTER C - u'D' # 0xC4 -> LATIN CAPITAL LETTER D - u'E' # 0xC5 -> LATIN CAPITAL LETTER E - u'F' # 0xC6 -> LATIN CAPITAL LETTER F - u'G' # 0xC7 -> LATIN CAPITAL LETTER G - u'H' # 0xC8 -> LATIN CAPITAL LETTER H - u'I' # 0xC9 -> LATIN CAPITAL LETTER I - u'\xad' # 0xCA -> SOFT HYPHEN - u'\ufffe' # 0xCB -> UNDEFINED - u'\ufffe' # 0xCC -> UNDEFINED - u'\ufffe' # 0xCD -> UNDEFINED - u'\ufffe' # 0xCE -> UNDEFINED - u'\ufffe' # 0xCF -> UNDEFINED - u'}' # 0xD0 -> RIGHT CURLY BRACKET - u'J' # 0xD1 -> LATIN CAPITAL LETTER J - u'K' # 0xD2 -> LATIN CAPITAL LETTER K - u'L' # 0xD3 -> LATIN CAPITAL LETTER L - u'M' # 0xD4 -> LATIN CAPITAL LETTER M - u'N' # 0xD5 -> LATIN CAPITAL LETTER N - u'O' # 0xD6 -> LATIN CAPITAL LETTER O - u'P' # 0xD7 -> LATIN CAPITAL LETTER P - u'Q' # 0xD8 -> LATIN CAPITAL LETTER Q - u'R' # 0xD9 -> LATIN CAPITAL LETTER R - u'\xb9' # 0xDA -> SUPERSCRIPT ONE - u'\ufffe' # 0xDB -> UNDEFINED - u'\ufffe' # 0xDC -> UNDEFINED - u'\ufffe' # 0xDD -> UNDEFINED - u'\ufffe' # 0xDE -> UNDEFINED - u'\ufffe' # 0xDF -> UNDEFINED - u'\\' # 0xE0 -> REVERSE SOLIDUS - u'\xf7' # 0xE1 -> DIVISION SIGN - u'S' # 0xE2 -> LATIN CAPITAL LETTER S - u'T' # 0xE3 -> LATIN CAPITAL LETTER T - u'U' # 0xE4 -> LATIN CAPITAL LETTER U - u'V' # 0xE5 -> LATIN CAPITAL LETTER V - u'W' # 0xE6 -> LATIN CAPITAL LETTER W - u'X' # 0xE7 -> LATIN CAPITAL LETTER X - u'Y' # 0xE8 -> LATIN CAPITAL LETTER Y - u'Z' # 0xE9 -> LATIN CAPITAL LETTER Z - u'\xb2' # 0xEA -> SUPERSCRIPT TWO - u'\ufffe' # 0xEB -> UNDEFINED - u'\ufffe' # 0xEC -> UNDEFINED - u'\ufffe' # 0xED -> UNDEFINED - u'\ufffe' # 0xEE -> UNDEFINED - u'\ufffe' # 0xEF -> UNDEFINED - u'0' # 0xF0 -> DIGIT ZERO - u'1' # 0xF1 -> DIGIT ONE - u'2' # 0xF2 -> DIGIT TWO - u'3' # 0xF3 -> DIGIT THREE - u'4' # 0xF4 -> DIGIT FOUR - u'5' # 0xF5 -> DIGIT FIVE - u'6' # 0xF6 -> DIGIT SIX - u'7' # 0xF7 -> DIGIT SEVEN - u'8' # 0xF8 -> DIGIT EIGHT - u'9' # 0xF9 -> DIGIT NINE - u'\xb3' # 0xFA -> SUPERSCRIPT THREE - u'\ufffe' # 0xFB -> UNDEFINED - u'\ufffe' # 0xFC -> UNDEFINED - u'\ufffe' # 0xFD -> UNDEFINED - u'\ufffe' # 0xFE -> UNDEFINED - u'\x9f' # 0xFF -> EIGHT ONES + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x9c' # 0x04 -> SELECT + '\t' # 0x05 -> HORIZONTAL TABULATION + '\x86' # 0x06 -> REQUIRED NEW LINE + '\x7f' # 0x07 -> DELETE + '\x97' # 0x08 -> GRAPHIC ESCAPE + '\x8d' # 0x09 -> SUPERSCRIPT + '\x8e' # 0x0A -> REPEAT + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x9d' # 0x14 -> RESTORE/ENABLE PRESENTATION + '\x85' # 0x15 -> NEW LINE + '\x08' # 0x16 -> BACKSPACE + '\x87' # 0x17 -> PROGRAM OPERATOR COMMUNICATION + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x92' # 0x1A -> UNIT BACK SPACE + '\x8f' # 0x1B -> CUSTOMER USE ONE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + '\x80' # 0x20 -> DIGIT SELECT + '\x81' # 0x21 -> START OF SIGNIFICANCE + '\x82' # 0x22 -> FIELD SEPARATOR + '\x83' # 0x23 -> WORD UNDERSCORE + '\x84' # 0x24 -> BYPASS OR INHIBIT PRESENTATION + '\n' # 0x25 -> LINE FEED + '\x17' # 0x26 -> END OF TRANSMISSION BLOCK + '\x1b' # 0x27 -> ESCAPE + '\x88' # 0x28 -> SET ATTRIBUTE + '\x89' # 0x29 -> START FIELD EXTENDED + '\x8a' # 0x2A -> SET MODE OR SWITCH + '\x8b' # 0x2B -> CONTROL SEQUENCE PREFIX + '\x8c' # 0x2C -> MODIFY FIELD ATTRIBUTE + '\x05' # 0x2D -> ENQUIRY + '\x06' # 0x2E -> ACKNOWLEDGE + '\x07' # 0x2F -> BELL + '\x90' # 0x30 -> <reserved> + '\x91' # 0x31 -> <reserved> + '\x16' # 0x32 -> SYNCHRONOUS IDLE + '\x93' # 0x33 -> INDEX RETURN + '\x94' # 0x34 -> PRESENTATION POSITION + '\x95' # 0x35 -> TRANSPARENT + '\x96' # 0x36 -> NUMERIC BACKSPACE + '\x04' # 0x37 -> END OF TRANSMISSION + '\x98' # 0x38 -> SUBSCRIPT + '\x99' # 0x39 -> INDENT TABULATION + '\x9a' # 0x3A -> REVERSE FORM FEED + '\x9b' # 0x3B -> CUSTOMER USE THREE + '\x14' # 0x3C -> DEVICE CONTROL FOUR + '\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE + '\x9e' # 0x3E -> <reserved> + '\x1a' # 0x3F -> SUBSTITUTE + ' ' # 0x40 -> SPACE + '\u05d0' # 0x41 -> HEBREW LETTER ALEF + '\u05d1' # 0x42 -> HEBREW LETTER BET + '\u05d2' # 0x43 -> HEBREW LETTER GIMEL + '\u05d3' # 0x44 -> HEBREW LETTER DALET + '\u05d4' # 0x45 -> HEBREW LETTER HE + '\u05d5' # 0x46 -> HEBREW LETTER VAV + '\u05d6' # 0x47 -> HEBREW LETTER ZAYIN + '\u05d7' # 0x48 -> HEBREW LETTER HET + '\u05d8' # 0x49 -> HEBREW LETTER TET + '\xa2' # 0x4A -> CENT SIGN + '.' # 0x4B -> FULL STOP + '<' # 0x4C -> LESS-THAN SIGN + '(' # 0x4D -> LEFT PARENTHESIS + '+' # 0x4E -> PLUS SIGN + '|' # 0x4F -> VERTICAL LINE + '&' # 0x50 -> AMPERSAND + '\u05d9' # 0x51 -> HEBREW LETTER YOD + '\u05da' # 0x52 -> HEBREW LETTER FINAL KAF + '\u05db' # 0x53 -> HEBREW LETTER KAF + '\u05dc' # 0x54 -> HEBREW LETTER LAMED + '\u05dd' # 0x55 -> HEBREW LETTER FINAL MEM + '\u05de' # 0x56 -> HEBREW LETTER MEM + '\u05df' # 0x57 -> HEBREW LETTER FINAL NUN + '\u05e0' # 0x58 -> HEBREW LETTER NUN + '\u05e1' # 0x59 -> HEBREW LETTER SAMEKH + '!' # 0x5A -> EXCLAMATION MARK + '$' # 0x5B -> DOLLAR SIGN + '*' # 0x5C -> ASTERISK + ')' # 0x5D -> RIGHT PARENTHESIS + ';' # 0x5E -> SEMICOLON + '\xac' # 0x5F -> NOT SIGN + '-' # 0x60 -> HYPHEN-MINUS + '/' # 0x61 -> SOLIDUS + '\u05e2' # 0x62 -> HEBREW LETTER AYIN + '\u05e3' # 0x63 -> HEBREW LETTER FINAL PE + '\u05e4' # 0x64 -> HEBREW LETTER PE + '\u05e5' # 0x65 -> HEBREW LETTER FINAL TSADI + '\u05e6' # 0x66 -> HEBREW LETTER TSADI + '\u05e7' # 0x67 -> HEBREW LETTER QOF + '\u05e8' # 0x68 -> HEBREW LETTER RESH + '\u05e9' # 0x69 -> HEBREW LETTER SHIN + '\xa6' # 0x6A -> BROKEN BAR + ',' # 0x6B -> COMMA + '%' # 0x6C -> PERCENT SIGN + '_' # 0x6D -> LOW LINE + '>' # 0x6E -> GREATER-THAN SIGN + '?' # 0x6F -> QUESTION MARK + '\ufffe' # 0x70 -> UNDEFINED + '\u05ea' # 0x71 -> HEBREW LETTER TAV + '\ufffe' # 0x72 -> UNDEFINED + '\ufffe' # 0x73 -> UNDEFINED + '\xa0' # 0x74 -> NO-BREAK SPACE + '\ufffe' # 0x75 -> UNDEFINED + '\ufffe' # 0x76 -> UNDEFINED + '\ufffe' # 0x77 -> UNDEFINED + '\u2017' # 0x78 -> DOUBLE LOW LINE + '`' # 0x79 -> GRAVE ACCENT + ':' # 0x7A -> COLON + '#' # 0x7B -> NUMBER SIGN + '@' # 0x7C -> COMMERCIAL AT + "'" # 0x7D -> APOSTROPHE + '=' # 0x7E -> EQUALS SIGN + '"' # 0x7F -> QUOTATION MARK + '\ufffe' # 0x80 -> UNDEFINED + 'a' # 0x81 -> LATIN SMALL LETTER A + 'b' # 0x82 -> LATIN SMALL LETTER B + 'c' # 0x83 -> LATIN SMALL LETTER C + 'd' # 0x84 -> LATIN SMALL LETTER D + 'e' # 0x85 -> LATIN SMALL LETTER E + 'f' # 0x86 -> LATIN SMALL LETTER F + 'g' # 0x87 -> LATIN SMALL LETTER G + 'h' # 0x88 -> LATIN SMALL LETTER H + 'i' # 0x89 -> LATIN SMALL LETTER I + '\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\ufffe' # 0x8C -> UNDEFINED + '\ufffe' # 0x8D -> UNDEFINED + '\ufffe' # 0x8E -> UNDEFINED + '\xb1' # 0x8F -> PLUS-MINUS SIGN + '\xb0' # 0x90 -> DEGREE SIGN + 'j' # 0x91 -> LATIN SMALL LETTER J + 'k' # 0x92 -> LATIN SMALL LETTER K + 'l' # 0x93 -> LATIN SMALL LETTER L + 'm' # 0x94 -> LATIN SMALL LETTER M + 'n' # 0x95 -> LATIN SMALL LETTER N + 'o' # 0x96 -> LATIN SMALL LETTER O + 'p' # 0x97 -> LATIN SMALL LETTER P + 'q' # 0x98 -> LATIN SMALL LETTER Q + 'r' # 0x99 -> LATIN SMALL LETTER R + '\ufffe' # 0x9A -> UNDEFINED + '\ufffe' # 0x9B -> UNDEFINED + '\ufffe' # 0x9C -> UNDEFINED + '\xb8' # 0x9D -> CEDILLA + '\ufffe' # 0x9E -> UNDEFINED + '\xa4' # 0x9F -> CURRENCY SIGN + '\xb5' # 0xA0 -> MICRO SIGN + '~' # 0xA1 -> TILDE + 's' # 0xA2 -> LATIN SMALL LETTER S + 't' # 0xA3 -> LATIN SMALL LETTER T + 'u' # 0xA4 -> LATIN SMALL LETTER U + 'v' # 0xA5 -> LATIN SMALL LETTER V + 'w' # 0xA6 -> LATIN SMALL LETTER W + 'x' # 0xA7 -> LATIN SMALL LETTER X + 'y' # 0xA8 -> LATIN SMALL LETTER Y + 'z' # 0xA9 -> LATIN SMALL LETTER Z + '\ufffe' # 0xAA -> UNDEFINED + '\ufffe' # 0xAB -> UNDEFINED + '\ufffe' # 0xAC -> UNDEFINED + '\ufffe' # 0xAD -> UNDEFINED + '\ufffe' # 0xAE -> UNDEFINED + '\xae' # 0xAF -> REGISTERED SIGN + '^' # 0xB0 -> CIRCUMFLEX ACCENT + '\xa3' # 0xB1 -> POUND SIGN + '\xa5' # 0xB2 -> YEN SIGN + '\xb7' # 0xB3 -> MIDDLE DOT + '\xa9' # 0xB4 -> COPYRIGHT SIGN + '\xa7' # 0xB5 -> SECTION SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF + '\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS + '[' # 0xBA -> LEFT SQUARE BRACKET + ']' # 0xBB -> RIGHT SQUARE BRACKET + '\xaf' # 0xBC -> MACRON + '\xa8' # 0xBD -> DIAERESIS + '\xb4' # 0xBE -> ACUTE ACCENT + '\xd7' # 0xBF -> MULTIPLICATION SIGN + '{' # 0xC0 -> LEFT CURLY BRACKET + 'A' # 0xC1 -> LATIN CAPITAL LETTER A + 'B' # 0xC2 -> LATIN CAPITAL LETTER B + 'C' # 0xC3 -> LATIN CAPITAL LETTER C + 'D' # 0xC4 -> LATIN CAPITAL LETTER D + 'E' # 0xC5 -> LATIN CAPITAL LETTER E + 'F' # 0xC6 -> LATIN CAPITAL LETTER F + 'G' # 0xC7 -> LATIN CAPITAL LETTER G + 'H' # 0xC8 -> LATIN CAPITAL LETTER H + 'I' # 0xC9 -> LATIN CAPITAL LETTER I + '\xad' # 0xCA -> SOFT HYPHEN + '\ufffe' # 0xCB -> UNDEFINED + '\ufffe' # 0xCC -> UNDEFINED + '\ufffe' # 0xCD -> UNDEFINED + '\ufffe' # 0xCE -> UNDEFINED + '\ufffe' # 0xCF -> UNDEFINED + '}' # 0xD0 -> RIGHT CURLY BRACKET + 'J' # 0xD1 -> LATIN CAPITAL LETTER J + 'K' # 0xD2 -> LATIN CAPITAL LETTER K + 'L' # 0xD3 -> LATIN CAPITAL LETTER L + 'M' # 0xD4 -> LATIN CAPITAL LETTER M + 'N' # 0xD5 -> LATIN CAPITAL LETTER N + 'O' # 0xD6 -> LATIN CAPITAL LETTER O + 'P' # 0xD7 -> LATIN CAPITAL LETTER P + 'Q' # 0xD8 -> LATIN CAPITAL LETTER Q + 'R' # 0xD9 -> LATIN CAPITAL LETTER R + '\xb9' # 0xDA -> SUPERSCRIPT ONE + '\ufffe' # 0xDB -> UNDEFINED + '\ufffe' # 0xDC -> UNDEFINED + '\ufffe' # 0xDD -> UNDEFINED + '\ufffe' # 0xDE -> UNDEFINED + '\ufffe' # 0xDF -> UNDEFINED + '\\' # 0xE0 -> REVERSE SOLIDUS + '\xf7' # 0xE1 -> DIVISION SIGN + 'S' # 0xE2 -> LATIN CAPITAL LETTER S + 'T' # 0xE3 -> LATIN CAPITAL LETTER T + 'U' # 0xE4 -> LATIN CAPITAL LETTER U + 'V' # 0xE5 -> LATIN CAPITAL LETTER V + 'W' # 0xE6 -> LATIN CAPITAL LETTER W + 'X' # 0xE7 -> LATIN CAPITAL LETTER X + 'Y' # 0xE8 -> LATIN CAPITAL LETTER Y + 'Z' # 0xE9 -> LATIN CAPITAL LETTER Z + '\xb2' # 0xEA -> SUPERSCRIPT TWO + '\ufffe' # 0xEB -> UNDEFINED + '\ufffe' # 0xEC -> UNDEFINED + '\ufffe' # 0xED -> UNDEFINED + '\ufffe' # 0xEE -> UNDEFINED + '\ufffe' # 0xEF -> UNDEFINED + '0' # 0xF0 -> DIGIT ZERO + '1' # 0xF1 -> DIGIT ONE + '2' # 0xF2 -> DIGIT TWO + '3' # 0xF3 -> DIGIT THREE + '4' # 0xF4 -> DIGIT FOUR + '5' # 0xF5 -> DIGIT FIVE + '6' # 0xF6 -> DIGIT SIX + '7' # 0xF7 -> DIGIT SEVEN + '8' # 0xF8 -> DIGIT EIGHT + '9' # 0xF9 -> DIGIT NINE + '\xb3' # 0xFA -> SUPERSCRIPT THREE + '\ufffe' # 0xFB -> UNDEFINED + '\ufffe' # 0xFC -> UNDEFINED + '\ufffe' # 0xFD -> UNDEFINED + '\ufffe' # 0xFE -> UNDEFINED + '\x9f' # 0xFF -> EIGHT ONES ) ### Encoding table diff --git a/Lib/encodings/cp437.py b/Lib/encodings/cp437.py index 52cd882..b6c75e2 100644 --- a/Lib/encodings/cp437.py +++ b/Lib/encodings/cp437.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE - u'\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE - u'\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE - u'\xff' # 0x0098 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xa2' # 0x009b -> CENT SIGN - u'\xa3' # 0x009c -> POUND SIGN - u'\xa5' # 0x009d -> YEN SIGN - u'\u20a7' # 0x009e -> PESETA SIGN - u'\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE - u'\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\u2310' # 0x00a9 -> REVERSED NOT SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA - u'\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI - u'\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA - u'\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU - u'\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI - u'\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA - u'\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA - u'\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA - u'\u221e' # 0x00ec -> INFINITY - u'\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI - u'\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON - u'\u2229' # 0x00ef -> INTERSECTION - u'\u2261' # 0x00f0 -> IDENTICAL TO - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u2320' # 0x00f4 -> TOP HALF INTEGRAL - u'\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS + '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE + '\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE + '\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE + '\xff' # 0x0098 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xa2' # 0x009b -> CENT SIGN + '\xa3' # 0x009c -> POUND SIGN + '\xa5' # 0x009d -> YEN SIGN + '\u20a7' # 0x009e -> PESETA SIGN + '\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE + '\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE + '\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR + '\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\u2310' # 0x00a9 -> REVERSED NOT SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA + '\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI + '\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA + '\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA + '\xb5' # 0x00e6 -> MICRO SIGN + '\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU + '\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI + '\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA + '\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA + '\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA + '\u221e' # 0x00ec -> INFINITY + '\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI + '\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON + '\u2229' # 0x00ef -> INTERSECTION + '\u2261' # 0x00f0 -> IDENTICAL TO + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u2320' # 0x00f4 -> TOP HALF INTEGRAL + '\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp500.py b/Lib/encodings/cp500.py index 60766c0..5f61535 100644 --- a/Lib/encodings/cp500.py +++ b/Lib/encodings/cp500.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x9c' # 0x04 -> CONTROL - u'\t' # 0x05 -> HORIZONTAL TABULATION - u'\x86' # 0x06 -> CONTROL - u'\x7f' # 0x07 -> DELETE - u'\x97' # 0x08 -> CONTROL - u'\x8d' # 0x09 -> CONTROL - u'\x8e' # 0x0A -> CONTROL - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x9d' # 0x14 -> CONTROL - u'\x85' # 0x15 -> CONTROL - u'\x08' # 0x16 -> BACKSPACE - u'\x87' # 0x17 -> CONTROL - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x92' # 0x1A -> CONTROL - u'\x8f' # 0x1B -> CONTROL - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u'\x80' # 0x20 -> CONTROL - u'\x81' # 0x21 -> CONTROL - u'\x82' # 0x22 -> CONTROL - u'\x83' # 0x23 -> CONTROL - u'\x84' # 0x24 -> CONTROL - u'\n' # 0x25 -> LINE FEED - u'\x17' # 0x26 -> END OF TRANSMISSION BLOCK - u'\x1b' # 0x27 -> ESCAPE - u'\x88' # 0x28 -> CONTROL - u'\x89' # 0x29 -> CONTROL - u'\x8a' # 0x2A -> CONTROL - u'\x8b' # 0x2B -> CONTROL - u'\x8c' # 0x2C -> CONTROL - u'\x05' # 0x2D -> ENQUIRY - u'\x06' # 0x2E -> ACKNOWLEDGE - u'\x07' # 0x2F -> BELL - u'\x90' # 0x30 -> CONTROL - u'\x91' # 0x31 -> CONTROL - u'\x16' # 0x32 -> SYNCHRONOUS IDLE - u'\x93' # 0x33 -> CONTROL - u'\x94' # 0x34 -> CONTROL - u'\x95' # 0x35 -> CONTROL - u'\x96' # 0x36 -> CONTROL - u'\x04' # 0x37 -> END OF TRANSMISSION - u'\x98' # 0x38 -> CONTROL - u'\x99' # 0x39 -> CONTROL - u'\x9a' # 0x3A -> CONTROL - u'\x9b' # 0x3B -> CONTROL - u'\x14' # 0x3C -> DEVICE CONTROL FOUR - u'\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE - u'\x9e' # 0x3E -> CONTROL - u'\x1a' # 0x3F -> SUBSTITUTE - u' ' # 0x40 -> SPACE - u'\xa0' # 0x41 -> NO-BREAK SPACE - u'\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x48 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE - u'[' # 0x4A -> LEFT SQUARE BRACKET - u'.' # 0x4B -> FULL STOP - u'<' # 0x4C -> LESS-THAN SIGN - u'(' # 0x4D -> LEFT PARENTHESIS - u'+' # 0x4E -> PLUS SIGN - u'!' # 0x4F -> EXCLAMATION MARK - u'&' # 0x50 -> AMPERSAND - u'\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE - u'\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE - u'\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) - u']' # 0x5A -> RIGHT SQUARE BRACKET - u'$' # 0x5B -> DOLLAR SIGN - u'*' # 0x5C -> ASTERISK - u')' # 0x5D -> RIGHT PARENTHESIS - u';' # 0x5E -> SEMICOLON - u'^' # 0x5F -> CIRCUMFLEX ACCENT - u'-' # 0x60 -> HYPHEN-MINUS - u'/' # 0x61 -> SOLIDUS - u'\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xa6' # 0x6A -> BROKEN BAR - u',' # 0x6B -> COMMA - u'%' # 0x6C -> PERCENT SIGN - u'_' # 0x6D -> LOW LINE - u'>' # 0x6E -> GREATER-THAN SIGN - u'?' # 0x6F -> QUESTION MARK - u'\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE - u'\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE - u'`' # 0x79 -> GRAVE ACCENT - u':' # 0x7A -> COLON - u'#' # 0x7B -> NUMBER SIGN - u'@' # 0x7C -> COMMERCIAL AT - u"'" # 0x7D -> APOSTROPHE - u'=' # 0x7E -> EQUALS SIGN - u'"' # 0x7F -> QUOTATION MARK - u'\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE - u'a' # 0x81 -> LATIN SMALL LETTER A - u'b' # 0x82 -> LATIN SMALL LETTER B - u'c' # 0x83 -> LATIN SMALL LETTER C - u'd' # 0x84 -> LATIN SMALL LETTER D - u'e' # 0x85 -> LATIN SMALL LETTER E - u'f' # 0x86 -> LATIN SMALL LETTER F - u'g' # 0x87 -> LATIN SMALL LETTER G - u'h' # 0x88 -> LATIN SMALL LETTER H - u'i' # 0x89 -> LATIN SMALL LETTER I - u'\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xf0' # 0x8C -> LATIN SMALL LETTER ETH (ICELANDIC) - u'\xfd' # 0x8D -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0x8E -> LATIN SMALL LETTER THORN (ICELANDIC) - u'\xb1' # 0x8F -> PLUS-MINUS SIGN - u'\xb0' # 0x90 -> DEGREE SIGN - u'j' # 0x91 -> LATIN SMALL LETTER J - u'k' # 0x92 -> LATIN SMALL LETTER K - u'l' # 0x93 -> LATIN SMALL LETTER L - u'm' # 0x94 -> LATIN SMALL LETTER M - u'n' # 0x95 -> LATIN SMALL LETTER N - u'o' # 0x96 -> LATIN SMALL LETTER O - u'p' # 0x97 -> LATIN SMALL LETTER P - u'q' # 0x98 -> LATIN SMALL LETTER Q - u'r' # 0x99 -> LATIN SMALL LETTER R - u'\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR - u'\xe6' # 0x9C -> LATIN SMALL LIGATURE AE - u'\xb8' # 0x9D -> CEDILLA - u'\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE - u'\xa4' # 0x9F -> CURRENCY SIGN - u'\xb5' # 0xA0 -> MICRO SIGN - u'~' # 0xA1 -> TILDE - u's' # 0xA2 -> LATIN SMALL LETTER S - u't' # 0xA3 -> LATIN SMALL LETTER T - u'u' # 0xA4 -> LATIN SMALL LETTER U - u'v' # 0xA5 -> LATIN SMALL LETTER V - u'w' # 0xA6 -> LATIN SMALL LETTER W - u'x' # 0xA7 -> LATIN SMALL LETTER X - u'y' # 0xA8 -> LATIN SMALL LETTER Y - u'z' # 0xA9 -> LATIN SMALL LETTER Z - u'\xa1' # 0xAA -> INVERTED EXCLAMATION MARK - u'\xbf' # 0xAB -> INVERTED QUESTION MARK - u'\xd0' # 0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC) - u'\xdd' # 0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC) - u'\xae' # 0xAF -> REGISTERED SIGN - u'\xa2' # 0xB0 -> CENT SIGN - u'\xa3' # 0xB1 -> POUND SIGN - u'\xa5' # 0xB2 -> YEN SIGN - u'\xb7' # 0xB3 -> MIDDLE DOT - u'\xa9' # 0xB4 -> COPYRIGHT SIGN - u'\xa7' # 0xB5 -> SECTION SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS - u'\xac' # 0xBA -> NOT SIGN - u'|' # 0xBB -> VERTICAL LINE - u'\xaf' # 0xBC -> MACRON - u'\xa8' # 0xBD -> DIAERESIS - u'\xb4' # 0xBE -> ACUTE ACCENT - u'\xd7' # 0xBF -> MULTIPLICATION SIGN - u'{' # 0xC0 -> LEFT CURLY BRACKET - u'A' # 0xC1 -> LATIN CAPITAL LETTER A - u'B' # 0xC2 -> LATIN CAPITAL LETTER B - u'C' # 0xC3 -> LATIN CAPITAL LETTER C - u'D' # 0xC4 -> LATIN CAPITAL LETTER D - u'E' # 0xC5 -> LATIN CAPITAL LETTER E - u'F' # 0xC6 -> LATIN CAPITAL LETTER F - u'G' # 0xC7 -> LATIN CAPITAL LETTER G - u'H' # 0xC8 -> LATIN CAPITAL LETTER H - u'I' # 0xC9 -> LATIN CAPITAL LETTER I - u'\xad' # 0xCA -> SOFT HYPHEN - u'\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0xCC -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE - u'\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE - u'}' # 0xD0 -> RIGHT CURLY BRACKET - u'J' # 0xD1 -> LATIN CAPITAL LETTER J - u'K' # 0xD2 -> LATIN CAPITAL LETTER K - u'L' # 0xD3 -> LATIN CAPITAL LETTER L - u'M' # 0xD4 -> LATIN CAPITAL LETTER M - u'N' # 0xD5 -> LATIN CAPITAL LETTER N - u'O' # 0xD6 -> LATIN CAPITAL LETTER O - u'P' # 0xD7 -> LATIN CAPITAL LETTER P - u'Q' # 0xD8 -> LATIN CAPITAL LETTER Q - u'R' # 0xD9 -> LATIN CAPITAL LETTER R - u'\xb9' # 0xDA -> SUPERSCRIPT ONE - u'\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xDC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE - u'\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\\' # 0xE0 -> REVERSE SOLIDUS - u'\xf7' # 0xE1 -> DIVISION SIGN - u'S' # 0xE2 -> LATIN CAPITAL LETTER S - u'T' # 0xE3 -> LATIN CAPITAL LETTER T - u'U' # 0xE4 -> LATIN CAPITAL LETTER U - u'V' # 0xE5 -> LATIN CAPITAL LETTER V - u'W' # 0xE6 -> LATIN CAPITAL LETTER W - u'X' # 0xE7 -> LATIN CAPITAL LETTER X - u'Y' # 0xE8 -> LATIN CAPITAL LETTER Y - u'Z' # 0xE9 -> LATIN CAPITAL LETTER Z - u'\xb2' # 0xEA -> SUPERSCRIPT TWO - u'\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd6' # 0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE - u'0' # 0xF0 -> DIGIT ZERO - u'1' # 0xF1 -> DIGIT ONE - u'2' # 0xF2 -> DIGIT TWO - u'3' # 0xF3 -> DIGIT THREE - u'4' # 0xF4 -> DIGIT FOUR - u'5' # 0xF5 -> DIGIT FIVE - u'6' # 0xF6 -> DIGIT SIX - u'7' # 0xF7 -> DIGIT SEVEN - u'8' # 0xF8 -> DIGIT EIGHT - u'9' # 0xF9 -> DIGIT NINE - u'\xb3' # 0xFA -> SUPERSCRIPT THREE - u'\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE - u'\x9f' # 0xFF -> CONTROL + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x9c' # 0x04 -> CONTROL + '\t' # 0x05 -> HORIZONTAL TABULATION + '\x86' # 0x06 -> CONTROL + '\x7f' # 0x07 -> DELETE + '\x97' # 0x08 -> CONTROL + '\x8d' # 0x09 -> CONTROL + '\x8e' # 0x0A -> CONTROL + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x9d' # 0x14 -> CONTROL + '\x85' # 0x15 -> CONTROL + '\x08' # 0x16 -> BACKSPACE + '\x87' # 0x17 -> CONTROL + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x92' # 0x1A -> CONTROL + '\x8f' # 0x1B -> CONTROL + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + '\x80' # 0x20 -> CONTROL + '\x81' # 0x21 -> CONTROL + '\x82' # 0x22 -> CONTROL + '\x83' # 0x23 -> CONTROL + '\x84' # 0x24 -> CONTROL + '\n' # 0x25 -> LINE FEED + '\x17' # 0x26 -> END OF TRANSMISSION BLOCK + '\x1b' # 0x27 -> ESCAPE + '\x88' # 0x28 -> CONTROL + '\x89' # 0x29 -> CONTROL + '\x8a' # 0x2A -> CONTROL + '\x8b' # 0x2B -> CONTROL + '\x8c' # 0x2C -> CONTROL + '\x05' # 0x2D -> ENQUIRY + '\x06' # 0x2E -> ACKNOWLEDGE + '\x07' # 0x2F -> BELL + '\x90' # 0x30 -> CONTROL + '\x91' # 0x31 -> CONTROL + '\x16' # 0x32 -> SYNCHRONOUS IDLE + '\x93' # 0x33 -> CONTROL + '\x94' # 0x34 -> CONTROL + '\x95' # 0x35 -> CONTROL + '\x96' # 0x36 -> CONTROL + '\x04' # 0x37 -> END OF TRANSMISSION + '\x98' # 0x38 -> CONTROL + '\x99' # 0x39 -> CONTROL + '\x9a' # 0x3A -> CONTROL + '\x9b' # 0x3B -> CONTROL + '\x14' # 0x3C -> DEVICE CONTROL FOUR + '\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE + '\x9e' # 0x3E -> CONTROL + '\x1a' # 0x3F -> SUBSTITUTE + ' ' # 0x40 -> SPACE + '\xa0' # 0x41 -> NO-BREAK SPACE + '\xe2' # 0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x43 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x44 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0x45 -> LATIN SMALL LETTER A WITH ACUTE + '\xe3' # 0x46 -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x47 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x48 -> LATIN SMALL LETTER C WITH CEDILLA + '\xf1' # 0x49 -> LATIN SMALL LETTER N WITH TILDE + '[' # 0x4A -> LEFT SQUARE BRACKET + '.' # 0x4B -> FULL STOP + '<' # 0x4C -> LESS-THAN SIGN + '(' # 0x4D -> LEFT PARENTHESIS + '+' # 0x4E -> PLUS SIGN + '!' # 0x4F -> EXCLAMATION MARK + '&' # 0x50 -> AMPERSAND + '\xe9' # 0x51 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x53 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x54 -> LATIN SMALL LETTER E WITH GRAVE + '\xed' # 0x55 -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x57 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xec' # 0x58 -> LATIN SMALL LETTER I WITH GRAVE + '\xdf' # 0x59 -> LATIN SMALL LETTER SHARP S (GERMAN) + ']' # 0x5A -> RIGHT SQUARE BRACKET + '$' # 0x5B -> DOLLAR SIGN + '*' # 0x5C -> ASTERISK + ')' # 0x5D -> RIGHT PARENTHESIS + ';' # 0x5E -> SEMICOLON + '^' # 0x5F -> CIRCUMFLEX ACCENT + '-' # 0x60 -> HYPHEN-MINUS + '/' # 0x61 -> SOLIDUS + '\xc2' # 0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc4' # 0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc0' # 0x64 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0x65 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc3' # 0x66 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc5' # 0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xd1' # 0x69 -> LATIN CAPITAL LETTER N WITH TILDE + '\xa6' # 0x6A -> BROKEN BAR + ',' # 0x6B -> COMMA + '%' # 0x6C -> PERCENT SIGN + '_' # 0x6D -> LOW LINE + '>' # 0x6E -> GREATER-THAN SIGN + '?' # 0x6F -> QUESTION MARK + '\xf8' # 0x70 -> LATIN SMALL LETTER O WITH STROKE + '\xc9' # 0x71 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0x74 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0x75 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0x78 -> LATIN CAPITAL LETTER I WITH GRAVE + '`' # 0x79 -> GRAVE ACCENT + ':' # 0x7A -> COLON + '#' # 0x7B -> NUMBER SIGN + '@' # 0x7C -> COMMERCIAL AT + "'" # 0x7D -> APOSTROPHE + '=' # 0x7E -> EQUALS SIGN + '"' # 0x7F -> QUOTATION MARK + '\xd8' # 0x80 -> LATIN CAPITAL LETTER O WITH STROKE + 'a' # 0x81 -> LATIN SMALL LETTER A + 'b' # 0x82 -> LATIN SMALL LETTER B + 'c' # 0x83 -> LATIN SMALL LETTER C + 'd' # 0x84 -> LATIN SMALL LETTER D + 'e' # 0x85 -> LATIN SMALL LETTER E + 'f' # 0x86 -> LATIN SMALL LETTER F + 'g' # 0x87 -> LATIN SMALL LETTER G + 'h' # 0x88 -> LATIN SMALL LETTER H + 'i' # 0x89 -> LATIN SMALL LETTER I + '\xab' # 0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xf0' # 0x8C -> LATIN SMALL LETTER ETH (ICELANDIC) + '\xfd' # 0x8D -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0x8E -> LATIN SMALL LETTER THORN (ICELANDIC) + '\xb1' # 0x8F -> PLUS-MINUS SIGN + '\xb0' # 0x90 -> DEGREE SIGN + 'j' # 0x91 -> LATIN SMALL LETTER J + 'k' # 0x92 -> LATIN SMALL LETTER K + 'l' # 0x93 -> LATIN SMALL LETTER L + 'm' # 0x94 -> LATIN SMALL LETTER M + 'n' # 0x95 -> LATIN SMALL LETTER N + 'o' # 0x96 -> LATIN SMALL LETTER O + 'p' # 0x97 -> LATIN SMALL LETTER P + 'q' # 0x98 -> LATIN SMALL LETTER Q + 'r' # 0x99 -> LATIN SMALL LETTER R + '\xaa' # 0x9A -> FEMININE ORDINAL INDICATOR + '\xba' # 0x9B -> MASCULINE ORDINAL INDICATOR + '\xe6' # 0x9C -> LATIN SMALL LIGATURE AE + '\xb8' # 0x9D -> CEDILLA + '\xc6' # 0x9E -> LATIN CAPITAL LIGATURE AE + '\xa4' # 0x9F -> CURRENCY SIGN + '\xb5' # 0xA0 -> MICRO SIGN + '~' # 0xA1 -> TILDE + 's' # 0xA2 -> LATIN SMALL LETTER S + 't' # 0xA3 -> LATIN SMALL LETTER T + 'u' # 0xA4 -> LATIN SMALL LETTER U + 'v' # 0xA5 -> LATIN SMALL LETTER V + 'w' # 0xA6 -> LATIN SMALL LETTER W + 'x' # 0xA7 -> LATIN SMALL LETTER X + 'y' # 0xA8 -> LATIN SMALL LETTER Y + 'z' # 0xA9 -> LATIN SMALL LETTER Z + '\xa1' # 0xAA -> INVERTED EXCLAMATION MARK + '\xbf' # 0xAB -> INVERTED QUESTION MARK + '\xd0' # 0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC) + '\xdd' # 0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC) + '\xae' # 0xAF -> REGISTERED SIGN + '\xa2' # 0xB0 -> CENT SIGN + '\xa3' # 0xB1 -> POUND SIGN + '\xa5' # 0xB2 -> YEN SIGN + '\xb7' # 0xB3 -> MIDDLE DOT + '\xa9' # 0xB4 -> COPYRIGHT SIGN + '\xa7' # 0xB5 -> SECTION SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xbc' # 0xB7 -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xB8 -> VULGAR FRACTION ONE HALF + '\xbe' # 0xB9 -> VULGAR FRACTION THREE QUARTERS + '\xac' # 0xBA -> NOT SIGN + '|' # 0xBB -> VERTICAL LINE + '\xaf' # 0xBC -> MACRON + '\xa8' # 0xBD -> DIAERESIS + '\xb4' # 0xBE -> ACUTE ACCENT + '\xd7' # 0xBF -> MULTIPLICATION SIGN + '{' # 0xC0 -> LEFT CURLY BRACKET + 'A' # 0xC1 -> LATIN CAPITAL LETTER A + 'B' # 0xC2 -> LATIN CAPITAL LETTER B + 'C' # 0xC3 -> LATIN CAPITAL LETTER C + 'D' # 0xC4 -> LATIN CAPITAL LETTER D + 'E' # 0xC5 -> LATIN CAPITAL LETTER E + 'F' # 0xC6 -> LATIN CAPITAL LETTER F + 'G' # 0xC7 -> LATIN CAPITAL LETTER G + 'H' # 0xC8 -> LATIN CAPITAL LETTER H + 'I' # 0xC9 -> LATIN CAPITAL LETTER I + '\xad' # 0xCA -> SOFT HYPHEN + '\xf4' # 0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0xCC -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0xCD -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xCE -> LATIN SMALL LETTER O WITH ACUTE + '\xf5' # 0xCF -> LATIN SMALL LETTER O WITH TILDE + '}' # 0xD0 -> RIGHT CURLY BRACKET + 'J' # 0xD1 -> LATIN CAPITAL LETTER J + 'K' # 0xD2 -> LATIN CAPITAL LETTER K + 'L' # 0xD3 -> LATIN CAPITAL LETTER L + 'M' # 0xD4 -> LATIN CAPITAL LETTER M + 'N' # 0xD5 -> LATIN CAPITAL LETTER N + 'O' # 0xD6 -> LATIN CAPITAL LETTER O + 'P' # 0xD7 -> LATIN CAPITAL LETTER P + 'Q' # 0xD8 -> LATIN CAPITAL LETTER Q + 'R' # 0xD9 -> LATIN CAPITAL LETTER R + '\xb9' # 0xDA -> SUPERSCRIPT ONE + '\xfb' # 0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xDC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xf9' # 0xDD -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xDE -> LATIN SMALL LETTER U WITH ACUTE + '\xff' # 0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\\' # 0xE0 -> REVERSE SOLIDUS + '\xf7' # 0xE1 -> DIVISION SIGN + 'S' # 0xE2 -> LATIN CAPITAL LETTER S + 'T' # 0xE3 -> LATIN CAPITAL LETTER T + 'U' # 0xE4 -> LATIN CAPITAL LETTER U + 'V' # 0xE5 -> LATIN CAPITAL LETTER V + 'W' # 0xE6 -> LATIN CAPITAL LETTER W + 'X' # 0xE7 -> LATIN CAPITAL LETTER X + 'Y' # 0xE8 -> LATIN CAPITAL LETTER Y + 'Z' # 0xE9 -> LATIN CAPITAL LETTER Z + '\xb2' # 0xEA -> SUPERSCRIPT TWO + '\xd4' # 0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd6' # 0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd2' # 0xED -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd5' # 0xEF -> LATIN CAPITAL LETTER O WITH TILDE + '0' # 0xF0 -> DIGIT ZERO + '1' # 0xF1 -> DIGIT ONE + '2' # 0xF2 -> DIGIT TWO + '3' # 0xF3 -> DIGIT THREE + '4' # 0xF4 -> DIGIT FOUR + '5' # 0xF5 -> DIGIT FIVE + '6' # 0xF6 -> DIGIT SIX + '7' # 0xF7 -> DIGIT SEVEN + '8' # 0xF8 -> DIGIT EIGHT + '9' # 0xF9 -> DIGIT NINE + '\xb3' # 0xFA -> SUPERSCRIPT THREE + '\xdb' # 0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xd9' # 0xFD -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xFE -> LATIN CAPITAL LETTER U WITH ACUTE + '\x9f' # 0xFF -> CONTROL ) ### Encoding table diff --git a/Lib/encodings/cp737.py b/Lib/encodings/cp737.py index d654448..9685bae 100644 --- a/Lib/encodings/cp737.py +++ b/Lib/encodings/cp737.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\u0391' # 0x0080 -> GREEK CAPITAL LETTER ALPHA - u'\u0392' # 0x0081 -> GREEK CAPITAL LETTER BETA - u'\u0393' # 0x0082 -> GREEK CAPITAL LETTER GAMMA - u'\u0394' # 0x0083 -> GREEK CAPITAL LETTER DELTA - u'\u0395' # 0x0084 -> GREEK CAPITAL LETTER EPSILON - u'\u0396' # 0x0085 -> GREEK CAPITAL LETTER ZETA - u'\u0397' # 0x0086 -> GREEK CAPITAL LETTER ETA - u'\u0398' # 0x0087 -> GREEK CAPITAL LETTER THETA - u'\u0399' # 0x0088 -> GREEK CAPITAL LETTER IOTA - u'\u039a' # 0x0089 -> GREEK CAPITAL LETTER KAPPA - u'\u039b' # 0x008a -> GREEK CAPITAL LETTER LAMDA - u'\u039c' # 0x008b -> GREEK CAPITAL LETTER MU - u'\u039d' # 0x008c -> GREEK CAPITAL LETTER NU - u'\u039e' # 0x008d -> GREEK CAPITAL LETTER XI - u'\u039f' # 0x008e -> GREEK CAPITAL LETTER OMICRON - u'\u03a0' # 0x008f -> GREEK CAPITAL LETTER PI - u'\u03a1' # 0x0090 -> GREEK CAPITAL LETTER RHO - u'\u03a3' # 0x0091 -> GREEK CAPITAL LETTER SIGMA - u'\u03a4' # 0x0092 -> GREEK CAPITAL LETTER TAU - u'\u03a5' # 0x0093 -> GREEK CAPITAL LETTER UPSILON - u'\u03a6' # 0x0094 -> GREEK CAPITAL LETTER PHI - u'\u03a7' # 0x0095 -> GREEK CAPITAL LETTER CHI - u'\u03a8' # 0x0096 -> GREEK CAPITAL LETTER PSI - u'\u03a9' # 0x0097 -> GREEK CAPITAL LETTER OMEGA - u'\u03b1' # 0x0098 -> GREEK SMALL LETTER ALPHA - u'\u03b2' # 0x0099 -> GREEK SMALL LETTER BETA - u'\u03b3' # 0x009a -> GREEK SMALL LETTER GAMMA - u'\u03b4' # 0x009b -> GREEK SMALL LETTER DELTA - u'\u03b5' # 0x009c -> GREEK SMALL LETTER EPSILON - u'\u03b6' # 0x009d -> GREEK SMALL LETTER ZETA - u'\u03b7' # 0x009e -> GREEK SMALL LETTER ETA - u'\u03b8' # 0x009f -> GREEK SMALL LETTER THETA - u'\u03b9' # 0x00a0 -> GREEK SMALL LETTER IOTA - u'\u03ba' # 0x00a1 -> GREEK SMALL LETTER KAPPA - u'\u03bb' # 0x00a2 -> GREEK SMALL LETTER LAMDA - u'\u03bc' # 0x00a3 -> GREEK SMALL LETTER MU - u'\u03bd' # 0x00a4 -> GREEK SMALL LETTER NU - u'\u03be' # 0x00a5 -> GREEK SMALL LETTER XI - u'\u03bf' # 0x00a6 -> GREEK SMALL LETTER OMICRON - u'\u03c0' # 0x00a7 -> GREEK SMALL LETTER PI - u'\u03c1' # 0x00a8 -> GREEK SMALL LETTER RHO - u'\u03c3' # 0x00a9 -> GREEK SMALL LETTER SIGMA - u'\u03c2' # 0x00aa -> GREEK SMALL LETTER FINAL SIGMA - u'\u03c4' # 0x00ab -> GREEK SMALL LETTER TAU - u'\u03c5' # 0x00ac -> GREEK SMALL LETTER UPSILON - u'\u03c6' # 0x00ad -> GREEK SMALL LETTER PHI - u'\u03c7' # 0x00ae -> GREEK SMALL LETTER CHI - u'\u03c8' # 0x00af -> GREEK SMALL LETTER PSI - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03c9' # 0x00e0 -> GREEK SMALL LETTER OMEGA - u'\u03ac' # 0x00e1 -> GREEK SMALL LETTER ALPHA WITH TONOS - u'\u03ad' # 0x00e2 -> GREEK SMALL LETTER EPSILON WITH TONOS - u'\u03ae' # 0x00e3 -> GREEK SMALL LETTER ETA WITH TONOS - u'\u03ca' # 0x00e4 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA - u'\u03af' # 0x00e5 -> GREEK SMALL LETTER IOTA WITH TONOS - u'\u03cc' # 0x00e6 -> GREEK SMALL LETTER OMICRON WITH TONOS - u'\u03cd' # 0x00e7 -> GREEK SMALL LETTER UPSILON WITH TONOS - u'\u03cb' # 0x00e8 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA - u'\u03ce' # 0x00e9 -> GREEK SMALL LETTER OMEGA WITH TONOS - u'\u0386' # 0x00ea -> GREEK CAPITAL LETTER ALPHA WITH TONOS - u'\u0388' # 0x00eb -> GREEK CAPITAL LETTER EPSILON WITH TONOS - u'\u0389' # 0x00ec -> GREEK CAPITAL LETTER ETA WITH TONOS - u'\u038a' # 0x00ed -> GREEK CAPITAL LETTER IOTA WITH TONOS - u'\u038c' # 0x00ee -> GREEK CAPITAL LETTER OMICRON WITH TONOS - u'\u038e' # 0x00ef -> GREEK CAPITAL LETTER UPSILON WITH TONOS - u'\u038f' # 0x00f0 -> GREEK CAPITAL LETTER OMEGA WITH TONOS - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u03aa' # 0x00f4 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - u'\u03ab' # 0x00f5 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\u0391' # 0x0080 -> GREEK CAPITAL LETTER ALPHA + '\u0392' # 0x0081 -> GREEK CAPITAL LETTER BETA + '\u0393' # 0x0082 -> GREEK CAPITAL LETTER GAMMA + '\u0394' # 0x0083 -> GREEK CAPITAL LETTER DELTA + '\u0395' # 0x0084 -> GREEK CAPITAL LETTER EPSILON + '\u0396' # 0x0085 -> GREEK CAPITAL LETTER ZETA + '\u0397' # 0x0086 -> GREEK CAPITAL LETTER ETA + '\u0398' # 0x0087 -> GREEK CAPITAL LETTER THETA + '\u0399' # 0x0088 -> GREEK CAPITAL LETTER IOTA + '\u039a' # 0x0089 -> GREEK CAPITAL LETTER KAPPA + '\u039b' # 0x008a -> GREEK CAPITAL LETTER LAMDA + '\u039c' # 0x008b -> GREEK CAPITAL LETTER MU + '\u039d' # 0x008c -> GREEK CAPITAL LETTER NU + '\u039e' # 0x008d -> GREEK CAPITAL LETTER XI + '\u039f' # 0x008e -> GREEK CAPITAL LETTER OMICRON + '\u03a0' # 0x008f -> GREEK CAPITAL LETTER PI + '\u03a1' # 0x0090 -> GREEK CAPITAL LETTER RHO + '\u03a3' # 0x0091 -> GREEK CAPITAL LETTER SIGMA + '\u03a4' # 0x0092 -> GREEK CAPITAL LETTER TAU + '\u03a5' # 0x0093 -> GREEK CAPITAL LETTER UPSILON + '\u03a6' # 0x0094 -> GREEK CAPITAL LETTER PHI + '\u03a7' # 0x0095 -> GREEK CAPITAL LETTER CHI + '\u03a8' # 0x0096 -> GREEK CAPITAL LETTER PSI + '\u03a9' # 0x0097 -> GREEK CAPITAL LETTER OMEGA + '\u03b1' # 0x0098 -> GREEK SMALL LETTER ALPHA + '\u03b2' # 0x0099 -> GREEK SMALL LETTER BETA + '\u03b3' # 0x009a -> GREEK SMALL LETTER GAMMA + '\u03b4' # 0x009b -> GREEK SMALL LETTER DELTA + '\u03b5' # 0x009c -> GREEK SMALL LETTER EPSILON + '\u03b6' # 0x009d -> GREEK SMALL LETTER ZETA + '\u03b7' # 0x009e -> GREEK SMALL LETTER ETA + '\u03b8' # 0x009f -> GREEK SMALL LETTER THETA + '\u03b9' # 0x00a0 -> GREEK SMALL LETTER IOTA + '\u03ba' # 0x00a1 -> GREEK SMALL LETTER KAPPA + '\u03bb' # 0x00a2 -> GREEK SMALL LETTER LAMDA + '\u03bc' # 0x00a3 -> GREEK SMALL LETTER MU + '\u03bd' # 0x00a4 -> GREEK SMALL LETTER NU + '\u03be' # 0x00a5 -> GREEK SMALL LETTER XI + '\u03bf' # 0x00a6 -> GREEK SMALL LETTER OMICRON + '\u03c0' # 0x00a7 -> GREEK SMALL LETTER PI + '\u03c1' # 0x00a8 -> GREEK SMALL LETTER RHO + '\u03c3' # 0x00a9 -> GREEK SMALL LETTER SIGMA + '\u03c2' # 0x00aa -> GREEK SMALL LETTER FINAL SIGMA + '\u03c4' # 0x00ab -> GREEK SMALL LETTER TAU + '\u03c5' # 0x00ac -> GREEK SMALL LETTER UPSILON + '\u03c6' # 0x00ad -> GREEK SMALL LETTER PHI + '\u03c7' # 0x00ae -> GREEK SMALL LETTER CHI + '\u03c8' # 0x00af -> GREEK SMALL LETTER PSI + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03c9' # 0x00e0 -> GREEK SMALL LETTER OMEGA + '\u03ac' # 0x00e1 -> GREEK SMALL LETTER ALPHA WITH TONOS + '\u03ad' # 0x00e2 -> GREEK SMALL LETTER EPSILON WITH TONOS + '\u03ae' # 0x00e3 -> GREEK SMALL LETTER ETA WITH TONOS + '\u03ca' # 0x00e4 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA + '\u03af' # 0x00e5 -> GREEK SMALL LETTER IOTA WITH TONOS + '\u03cc' # 0x00e6 -> GREEK SMALL LETTER OMICRON WITH TONOS + '\u03cd' # 0x00e7 -> GREEK SMALL LETTER UPSILON WITH TONOS + '\u03cb' # 0x00e8 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA + '\u03ce' # 0x00e9 -> GREEK SMALL LETTER OMEGA WITH TONOS + '\u0386' # 0x00ea -> GREEK CAPITAL LETTER ALPHA WITH TONOS + '\u0388' # 0x00eb -> GREEK CAPITAL LETTER EPSILON WITH TONOS + '\u0389' # 0x00ec -> GREEK CAPITAL LETTER ETA WITH TONOS + '\u038a' # 0x00ed -> GREEK CAPITAL LETTER IOTA WITH TONOS + '\u038c' # 0x00ee -> GREEK CAPITAL LETTER OMICRON WITH TONOS + '\u038e' # 0x00ef -> GREEK CAPITAL LETTER UPSILON WITH TONOS + '\u038f' # 0x00f0 -> GREEK CAPITAL LETTER OMEGA WITH TONOS + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u03aa' # 0x00f4 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA + '\u03ab' # 0x00f5 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp775.py b/Lib/encodings/cp775.py index 6a456a5..fe06e7b 100644 --- a/Lib/encodings/cp775.py +++ b/Lib/encodings/cp775.py @@ -177,262 +177,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\u0106' # 0x0080 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\u0101' # 0x0083 -> LATIN SMALL LETTER A WITH MACRON - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u0123' # 0x0085 -> LATIN SMALL LETTER G WITH CEDILLA - u'\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\u0107' # 0x0087 -> LATIN SMALL LETTER C WITH ACUTE - u'\u0142' # 0x0088 -> LATIN SMALL LETTER L WITH STROKE - u'\u0113' # 0x0089 -> LATIN SMALL LETTER E WITH MACRON - u'\u0156' # 0x008a -> LATIN CAPITAL LETTER R WITH CEDILLA - u'\u0157' # 0x008b -> LATIN SMALL LETTER R WITH CEDILLA - u'\u012b' # 0x008c -> LATIN SMALL LETTER I WITH MACRON - u'\u0179' # 0x008d -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE - u'\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE - u'\u014d' # 0x0093 -> LATIN SMALL LETTER O WITH MACRON - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\u0122' # 0x0095 -> LATIN CAPITAL LETTER G WITH CEDILLA - u'\xa2' # 0x0096 -> CENT SIGN - u'\u015a' # 0x0097 -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u015b' # 0x0098 -> LATIN SMALL LETTER S WITH ACUTE - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE - u'\xa3' # 0x009c -> POUND SIGN - u'\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd7' # 0x009e -> MULTIPLICATION SIGN - u'\xa4' # 0x009f -> CURRENCY SIGN - u'\u0100' # 0x00a0 -> LATIN CAPITAL LETTER A WITH MACRON - u'\u012a' # 0x00a1 -> LATIN CAPITAL LETTER I WITH MACRON - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\u017b' # 0x00a3 -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\u017c' # 0x00a4 -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u017a' # 0x00a5 -> LATIN SMALL LETTER Z WITH ACUTE - u'\u201d' # 0x00a6 -> RIGHT DOUBLE QUOTATION MARK - u'\xa6' # 0x00a7 -> BROKEN BAR - u'\xa9' # 0x00a8 -> COPYRIGHT SIGN - u'\xae' # 0x00a9 -> REGISTERED SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\u0141' # 0x00ad -> LATIN CAPITAL LETTER L WITH STROKE - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u0104' # 0x00b5 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u010c' # 0x00b6 -> LATIN CAPITAL LETTER C WITH CARON - u'\u0118' # 0x00b7 -> LATIN CAPITAL LETTER E WITH OGONEK - u'\u0116' # 0x00b8 -> LATIN CAPITAL LETTER E WITH DOT ABOVE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u012e' # 0x00bd -> LATIN CAPITAL LETTER I WITH OGONEK - u'\u0160' # 0x00be -> LATIN CAPITAL LETTER S WITH CARON - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u0172' # 0x00c6 -> LATIN CAPITAL LETTER U WITH OGONEK - u'\u016a' # 0x00c7 -> LATIN CAPITAL LETTER U WITH MACRON - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u017d' # 0x00cf -> LATIN CAPITAL LETTER Z WITH CARON - u'\u0105' # 0x00d0 -> LATIN SMALL LETTER A WITH OGONEK - u'\u010d' # 0x00d1 -> LATIN SMALL LETTER C WITH CARON - u'\u0119' # 0x00d2 -> LATIN SMALL LETTER E WITH OGONEK - u'\u0117' # 0x00d3 -> LATIN SMALL LETTER E WITH DOT ABOVE - u'\u012f' # 0x00d4 -> LATIN SMALL LETTER I WITH OGONEK - u'\u0161' # 0x00d5 -> LATIN SMALL LETTER S WITH CARON - u'\u0173' # 0x00d6 -> LATIN SMALL LETTER U WITH OGONEK - u'\u016b' # 0x00d7 -> LATIN SMALL LETTER U WITH MACRON - u'\u017e' # 0x00d8 -> LATIN SMALL LETTER Z WITH CARON - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S (GERMAN) - u'\u014c' # 0x00e2 -> LATIN CAPITAL LETTER O WITH MACRON - u'\u0143' # 0x00e3 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\xf5' # 0x00e4 -> LATIN SMALL LETTER O WITH TILDE - u'\xd5' # 0x00e5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u0144' # 0x00e7 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0136' # 0x00e8 -> LATIN CAPITAL LETTER K WITH CEDILLA - u'\u0137' # 0x00e9 -> LATIN SMALL LETTER K WITH CEDILLA - u'\u013b' # 0x00ea -> LATIN CAPITAL LETTER L WITH CEDILLA - u'\u013c' # 0x00eb -> LATIN SMALL LETTER L WITH CEDILLA - u'\u0146' # 0x00ec -> LATIN SMALL LETTER N WITH CEDILLA - u'\u0112' # 0x00ed -> LATIN CAPITAL LETTER E WITH MACRON - u'\u0145' # 0x00ee -> LATIN CAPITAL LETTER N WITH CEDILLA - u'\u2019' # 0x00ef -> RIGHT SINGLE QUOTATION MARK - u'\xad' # 0x00f0 -> SOFT HYPHEN - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u201c' # 0x00f2 -> LEFT DOUBLE QUOTATION MARK - u'\xbe' # 0x00f3 -> VULGAR FRACTION THREE QUARTERS - u'\xb6' # 0x00f4 -> PILCROW SIGN - u'\xa7' # 0x00f5 -> SECTION SIGN - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u201e' # 0x00f7 -> DOUBLE LOW-9 QUOTATION MARK - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\xb9' # 0x00fb -> SUPERSCRIPT ONE - u'\xb3' # 0x00fc -> SUPERSCRIPT THREE - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\u0106' # 0x0080 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\u0101' # 0x0083 -> LATIN SMALL LETTER A WITH MACRON + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\u0123' # 0x0085 -> LATIN SMALL LETTER G WITH CEDILLA + '\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE + '\u0107' # 0x0087 -> LATIN SMALL LETTER C WITH ACUTE + '\u0142' # 0x0088 -> LATIN SMALL LETTER L WITH STROKE + '\u0113' # 0x0089 -> LATIN SMALL LETTER E WITH MACRON + '\u0156' # 0x008a -> LATIN CAPITAL LETTER R WITH CEDILLA + '\u0157' # 0x008b -> LATIN SMALL LETTER R WITH CEDILLA + '\u012b' # 0x008c -> LATIN SMALL LETTER I WITH MACRON + '\u0179' # 0x008d -> LATIN CAPITAL LETTER Z WITH ACUTE + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE + '\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE + '\u014d' # 0x0093 -> LATIN SMALL LETTER O WITH MACRON + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\u0122' # 0x0095 -> LATIN CAPITAL LETTER G WITH CEDILLA + '\xa2' # 0x0096 -> CENT SIGN + '\u015a' # 0x0097 -> LATIN CAPITAL LETTER S WITH ACUTE + '\u015b' # 0x0098 -> LATIN SMALL LETTER S WITH ACUTE + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE + '\xa3' # 0x009c -> POUND SIGN + '\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE + '\xd7' # 0x009e -> MULTIPLICATION SIGN + '\xa4' # 0x009f -> CURRENCY SIGN + '\u0100' # 0x00a0 -> LATIN CAPITAL LETTER A WITH MACRON + '\u012a' # 0x00a1 -> LATIN CAPITAL LETTER I WITH MACRON + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\u017b' # 0x00a3 -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\u017c' # 0x00a4 -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u017a' # 0x00a5 -> LATIN SMALL LETTER Z WITH ACUTE + '\u201d' # 0x00a6 -> RIGHT DOUBLE QUOTATION MARK + '\xa6' # 0x00a7 -> BROKEN BAR + '\xa9' # 0x00a8 -> COPYRIGHT SIGN + '\xae' # 0x00a9 -> REGISTERED SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\u0141' # 0x00ad -> LATIN CAPITAL LETTER L WITH STROKE + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u0104' # 0x00b5 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u010c' # 0x00b6 -> LATIN CAPITAL LETTER C WITH CARON + '\u0118' # 0x00b7 -> LATIN CAPITAL LETTER E WITH OGONEK + '\u0116' # 0x00b8 -> LATIN CAPITAL LETTER E WITH DOT ABOVE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u012e' # 0x00bd -> LATIN CAPITAL LETTER I WITH OGONEK + '\u0160' # 0x00be -> LATIN CAPITAL LETTER S WITH CARON + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u0172' # 0x00c6 -> LATIN CAPITAL LETTER U WITH OGONEK + '\u016a' # 0x00c7 -> LATIN CAPITAL LETTER U WITH MACRON + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u017d' # 0x00cf -> LATIN CAPITAL LETTER Z WITH CARON + '\u0105' # 0x00d0 -> LATIN SMALL LETTER A WITH OGONEK + '\u010d' # 0x00d1 -> LATIN SMALL LETTER C WITH CARON + '\u0119' # 0x00d2 -> LATIN SMALL LETTER E WITH OGONEK + '\u0117' # 0x00d3 -> LATIN SMALL LETTER E WITH DOT ABOVE + '\u012f' # 0x00d4 -> LATIN SMALL LETTER I WITH OGONEK + '\u0161' # 0x00d5 -> LATIN SMALL LETTER S WITH CARON + '\u0173' # 0x00d6 -> LATIN SMALL LETTER U WITH OGONEK + '\u016b' # 0x00d7 -> LATIN SMALL LETTER U WITH MACRON + '\u017e' # 0x00d8 -> LATIN SMALL LETTER Z WITH CARON + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S (GERMAN) + '\u014c' # 0x00e2 -> LATIN CAPITAL LETTER O WITH MACRON + '\u0143' # 0x00e3 -> LATIN CAPITAL LETTER N WITH ACUTE + '\xf5' # 0x00e4 -> LATIN SMALL LETTER O WITH TILDE + '\xd5' # 0x00e5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xb5' # 0x00e6 -> MICRO SIGN + '\u0144' # 0x00e7 -> LATIN SMALL LETTER N WITH ACUTE + '\u0136' # 0x00e8 -> LATIN CAPITAL LETTER K WITH CEDILLA + '\u0137' # 0x00e9 -> LATIN SMALL LETTER K WITH CEDILLA + '\u013b' # 0x00ea -> LATIN CAPITAL LETTER L WITH CEDILLA + '\u013c' # 0x00eb -> LATIN SMALL LETTER L WITH CEDILLA + '\u0146' # 0x00ec -> LATIN SMALL LETTER N WITH CEDILLA + '\u0112' # 0x00ed -> LATIN CAPITAL LETTER E WITH MACRON + '\u0145' # 0x00ee -> LATIN CAPITAL LETTER N WITH CEDILLA + '\u2019' # 0x00ef -> RIGHT SINGLE QUOTATION MARK + '\xad' # 0x00f0 -> SOFT HYPHEN + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u201c' # 0x00f2 -> LEFT DOUBLE QUOTATION MARK + '\xbe' # 0x00f3 -> VULGAR FRACTION THREE QUARTERS + '\xb6' # 0x00f4 -> PILCROW SIGN + '\xa7' # 0x00f5 -> SECTION SIGN + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u201e' # 0x00f7 -> DOUBLE LOW-9 QUOTATION MARK + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\xb9' # 0x00fb -> SUPERSCRIPT ONE + '\xb3' # 0x00fc -> SUPERSCRIPT THREE + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp850.py b/Lib/encodings/cp850.py index 0c8478c..f98aef9 100644 --- a/Lib/encodings/cp850.py +++ b/Lib/encodings/cp850.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE - u'\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE - u'\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE - u'\xff' # 0x0098 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE - u'\xa3' # 0x009c -> POUND SIGN - u'\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd7' # 0x009e -> MULTIPLICATION SIGN - u'\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE - u'\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\xae' # 0x00a9 -> REGISTERED SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc0' # 0x00b7 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xa9' # 0x00b8 -> COPYRIGHT SIGN - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\xa2' # 0x00bd -> CENT SIGN - u'\xa5' # 0x00be -> YEN SIGN - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\xe3' # 0x00c6 -> LATIN SMALL LETTER A WITH TILDE - u'\xc3' # 0x00c7 -> LATIN CAPITAL LETTER A WITH TILDE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa4' # 0x00cf -> CURRENCY SIGN - u'\xf0' # 0x00d0 -> LATIN SMALL LETTER ETH - u'\xd0' # 0x00d1 -> LATIN CAPITAL LETTER ETH - u'\xca' # 0x00d2 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0x00d4 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\u0131' # 0x00d5 -> LATIN SMALL LETTER DOTLESS I - u'\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0x00d8 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\xa6' # 0x00dd -> BROKEN BAR - u'\xcc' # 0x00de -> LATIN CAPITAL LETTER I WITH GRAVE - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd2' # 0x00e3 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xf5' # 0x00e4 -> LATIN SMALL LETTER O WITH TILDE - u'\xd5' # 0x00e5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\xfe' # 0x00e7 -> LATIN SMALL LETTER THORN - u'\xde' # 0x00e8 -> LATIN CAPITAL LETTER THORN - u'\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0x00ea -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0x00eb -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xfd' # 0x00ec -> LATIN SMALL LETTER Y WITH ACUTE - u'\xdd' # 0x00ed -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xaf' # 0x00ee -> MACRON - u'\xb4' # 0x00ef -> ACUTE ACCENT - u'\xad' # 0x00f0 -> SOFT HYPHEN - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2017' # 0x00f2 -> DOUBLE LOW LINE - u'\xbe' # 0x00f3 -> VULGAR FRACTION THREE QUARTERS - u'\xb6' # 0x00f4 -> PILCROW SIGN - u'\xa7' # 0x00f5 -> SECTION SIGN - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\xb8' # 0x00f7 -> CEDILLA - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\xa8' # 0x00f9 -> DIAERESIS - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\xb9' # 0x00fb -> SUPERSCRIPT ONE - u'\xb3' # 0x00fc -> SUPERSCRIPT THREE - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS + '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE + '\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE + '\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE + '\xff' # 0x0098 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE + '\xa3' # 0x009c -> POUND SIGN + '\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE + '\xd7' # 0x009e -> MULTIPLICATION SIGN + '\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE + '\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE + '\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR + '\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\xae' # 0x00a9 -> REGISTERED SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc0' # 0x00b7 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xa9' # 0x00b8 -> COPYRIGHT SIGN + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\xa2' # 0x00bd -> CENT SIGN + '\xa5' # 0x00be -> YEN SIGN + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\xe3' # 0x00c6 -> LATIN SMALL LETTER A WITH TILDE + '\xc3' # 0x00c7 -> LATIN CAPITAL LETTER A WITH TILDE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa4' # 0x00cf -> CURRENCY SIGN + '\xf0' # 0x00d0 -> LATIN SMALL LETTER ETH + '\xd0' # 0x00d1 -> LATIN CAPITAL LETTER ETH + '\xca' # 0x00d2 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0x00d4 -> LATIN CAPITAL LETTER E WITH GRAVE + '\u0131' # 0x00d5 -> LATIN SMALL LETTER DOTLESS I + '\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0x00d8 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\xa6' # 0x00dd -> BROKEN BAR + '\xcc' # 0x00de -> LATIN CAPITAL LETTER I WITH GRAVE + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd2' # 0x00e3 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xf5' # 0x00e4 -> LATIN SMALL LETTER O WITH TILDE + '\xd5' # 0x00e5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xb5' # 0x00e6 -> MICRO SIGN + '\xfe' # 0x00e7 -> LATIN SMALL LETTER THORN + '\xde' # 0x00e8 -> LATIN CAPITAL LETTER THORN + '\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0x00ea -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0x00eb -> LATIN CAPITAL LETTER U WITH GRAVE + '\xfd' # 0x00ec -> LATIN SMALL LETTER Y WITH ACUTE + '\xdd' # 0x00ed -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xaf' # 0x00ee -> MACRON + '\xb4' # 0x00ef -> ACUTE ACCENT + '\xad' # 0x00f0 -> SOFT HYPHEN + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2017' # 0x00f2 -> DOUBLE LOW LINE + '\xbe' # 0x00f3 -> VULGAR FRACTION THREE QUARTERS + '\xb6' # 0x00f4 -> PILCROW SIGN + '\xa7' # 0x00f5 -> SECTION SIGN + '\xf7' # 0x00f6 -> DIVISION SIGN + '\xb8' # 0x00f7 -> CEDILLA + '\xb0' # 0x00f8 -> DEGREE SIGN + '\xa8' # 0x00f9 -> DIAERESIS + '\xb7' # 0x00fa -> MIDDLE DOT + '\xb9' # 0x00fb -> SUPERSCRIPT ONE + '\xb3' # 0x00fc -> SUPERSCRIPT THREE + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp852.py b/Lib/encodings/cp852.py index 069d547..34d8a0e 100644 --- a/Lib/encodings/cp852.py +++ b/Lib/encodings/cp852.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u016f' # 0x0085 -> LATIN SMALL LETTER U WITH RING ABOVE - u'\u0107' # 0x0086 -> LATIN SMALL LETTER C WITH ACUTE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\u0142' # 0x0088 -> LATIN SMALL LETTER L WITH STROKE - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u0150' # 0x008a -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - u'\u0151' # 0x008b -> LATIN SMALL LETTER O WITH DOUBLE ACUTE - u'\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\u0179' # 0x008d -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\u0106' # 0x008f -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0139' # 0x0091 -> LATIN CAPITAL LETTER L WITH ACUTE - u'\u013a' # 0x0092 -> LATIN SMALL LETTER L WITH ACUTE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\u013d' # 0x0095 -> LATIN CAPITAL LETTER L WITH CARON - u'\u013e' # 0x0096 -> LATIN SMALL LETTER L WITH CARON - u'\u015a' # 0x0097 -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u015b' # 0x0098 -> LATIN SMALL LETTER S WITH ACUTE - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u0164' # 0x009b -> LATIN CAPITAL LETTER T WITH CARON - u'\u0165' # 0x009c -> LATIN SMALL LETTER T WITH CARON - u'\u0141' # 0x009d -> LATIN CAPITAL LETTER L WITH STROKE - u'\xd7' # 0x009e -> MULTIPLICATION SIGN - u'\u010d' # 0x009f -> LATIN SMALL LETTER C WITH CARON - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\u0104' # 0x00a4 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u0105' # 0x00a5 -> LATIN SMALL LETTER A WITH OGONEK - u'\u017d' # 0x00a6 -> LATIN CAPITAL LETTER Z WITH CARON - u'\u017e' # 0x00a7 -> LATIN SMALL LETTER Z WITH CARON - u'\u0118' # 0x00a8 -> LATIN CAPITAL LETTER E WITH OGONEK - u'\u0119' # 0x00a9 -> LATIN SMALL LETTER E WITH OGONEK - u'\xac' # 0x00aa -> NOT SIGN - u'\u017a' # 0x00ab -> LATIN SMALL LETTER Z WITH ACUTE - u'\u010c' # 0x00ac -> LATIN CAPITAL LETTER C WITH CARON - u'\u015f' # 0x00ad -> LATIN SMALL LETTER S WITH CEDILLA - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\u011a' # 0x00b7 -> LATIN CAPITAL LETTER E WITH CARON - u'\u015e' # 0x00b8 -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u017b' # 0x00bd -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\u017c' # 0x00be -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u0102' # 0x00c6 -> LATIN CAPITAL LETTER A WITH BREVE - u'\u0103' # 0x00c7 -> LATIN SMALL LETTER A WITH BREVE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa4' # 0x00cf -> CURRENCY SIGN - u'\u0111' # 0x00d0 -> LATIN SMALL LETTER D WITH STROKE - u'\u0110' # 0x00d1 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u010e' # 0x00d2 -> LATIN CAPITAL LETTER D WITH CARON - u'\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u010f' # 0x00d4 -> LATIN SMALL LETTER D WITH CARON - u'\u0147' # 0x00d5 -> LATIN CAPITAL LETTER N WITH CARON - u'\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\u011b' # 0x00d8 -> LATIN SMALL LETTER E WITH CARON - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u0162' # 0x00dd -> LATIN CAPITAL LETTER T WITH CEDILLA - u'\u016e' # 0x00de -> LATIN CAPITAL LETTER U WITH RING ABOVE - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u0143' # 0x00e3 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\u0144' # 0x00e4 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0148' # 0x00e5 -> LATIN SMALL LETTER N WITH CARON - u'\u0160' # 0x00e6 -> LATIN CAPITAL LETTER S WITH CARON - u'\u0161' # 0x00e7 -> LATIN SMALL LETTER S WITH CARON - u'\u0154' # 0x00e8 -> LATIN CAPITAL LETTER R WITH ACUTE - u'\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\u0155' # 0x00ea -> LATIN SMALL LETTER R WITH ACUTE - u'\u0170' # 0x00eb -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - u'\xfd' # 0x00ec -> LATIN SMALL LETTER Y WITH ACUTE - u'\xdd' # 0x00ed -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\u0163' # 0x00ee -> LATIN SMALL LETTER T WITH CEDILLA - u'\xb4' # 0x00ef -> ACUTE ACCENT - u'\xad' # 0x00f0 -> SOFT HYPHEN - u'\u02dd' # 0x00f1 -> DOUBLE ACUTE ACCENT - u'\u02db' # 0x00f2 -> OGONEK - u'\u02c7' # 0x00f3 -> CARON - u'\u02d8' # 0x00f4 -> BREVE - u'\xa7' # 0x00f5 -> SECTION SIGN - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\xb8' # 0x00f7 -> CEDILLA - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\xa8' # 0x00f9 -> DIAERESIS - u'\u02d9' # 0x00fa -> DOT ABOVE - u'\u0171' # 0x00fb -> LATIN SMALL LETTER U WITH DOUBLE ACUTE - u'\u0158' # 0x00fc -> LATIN CAPITAL LETTER R WITH CARON - u'\u0159' # 0x00fd -> LATIN SMALL LETTER R WITH CARON - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\u016f' # 0x0085 -> LATIN SMALL LETTER U WITH RING ABOVE + '\u0107' # 0x0086 -> LATIN SMALL LETTER C WITH ACUTE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\u0142' # 0x0088 -> LATIN SMALL LETTER L WITH STROKE + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\u0150' # 0x008a -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE + '\u0151' # 0x008b -> LATIN SMALL LETTER O WITH DOUBLE ACUTE + '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\u0179' # 0x008d -> LATIN CAPITAL LETTER Z WITH ACUTE + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\u0106' # 0x008f -> LATIN CAPITAL LETTER C WITH ACUTE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0139' # 0x0091 -> LATIN CAPITAL LETTER L WITH ACUTE + '\u013a' # 0x0092 -> LATIN SMALL LETTER L WITH ACUTE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\u013d' # 0x0095 -> LATIN CAPITAL LETTER L WITH CARON + '\u013e' # 0x0096 -> LATIN SMALL LETTER L WITH CARON + '\u015a' # 0x0097 -> LATIN CAPITAL LETTER S WITH ACUTE + '\u015b' # 0x0098 -> LATIN SMALL LETTER S WITH ACUTE + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u0164' # 0x009b -> LATIN CAPITAL LETTER T WITH CARON + '\u0165' # 0x009c -> LATIN SMALL LETTER T WITH CARON + '\u0141' # 0x009d -> LATIN CAPITAL LETTER L WITH STROKE + '\xd7' # 0x009e -> MULTIPLICATION SIGN + '\u010d' # 0x009f -> LATIN SMALL LETTER C WITH CARON + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\u0104' # 0x00a4 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u0105' # 0x00a5 -> LATIN SMALL LETTER A WITH OGONEK + '\u017d' # 0x00a6 -> LATIN CAPITAL LETTER Z WITH CARON + '\u017e' # 0x00a7 -> LATIN SMALL LETTER Z WITH CARON + '\u0118' # 0x00a8 -> LATIN CAPITAL LETTER E WITH OGONEK + '\u0119' # 0x00a9 -> LATIN SMALL LETTER E WITH OGONEK + '\xac' # 0x00aa -> NOT SIGN + '\u017a' # 0x00ab -> LATIN SMALL LETTER Z WITH ACUTE + '\u010c' # 0x00ac -> LATIN CAPITAL LETTER C WITH CARON + '\u015f' # 0x00ad -> LATIN SMALL LETTER S WITH CEDILLA + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\u011a' # 0x00b7 -> LATIN CAPITAL LETTER E WITH CARON + '\u015e' # 0x00b8 -> LATIN CAPITAL LETTER S WITH CEDILLA + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u017b' # 0x00bd -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\u017c' # 0x00be -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u0102' # 0x00c6 -> LATIN CAPITAL LETTER A WITH BREVE + '\u0103' # 0x00c7 -> LATIN SMALL LETTER A WITH BREVE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa4' # 0x00cf -> CURRENCY SIGN + '\u0111' # 0x00d0 -> LATIN SMALL LETTER D WITH STROKE + '\u0110' # 0x00d1 -> LATIN CAPITAL LETTER D WITH STROKE + '\u010e' # 0x00d2 -> LATIN CAPITAL LETTER D WITH CARON + '\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u010f' # 0x00d4 -> LATIN SMALL LETTER D WITH CARON + '\u0147' # 0x00d5 -> LATIN CAPITAL LETTER N WITH CARON + '\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\u011b' # 0x00d8 -> LATIN SMALL LETTER E WITH CARON + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u0162' # 0x00dd -> LATIN CAPITAL LETTER T WITH CEDILLA + '\u016e' # 0x00de -> LATIN CAPITAL LETTER U WITH RING ABOVE + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u0143' # 0x00e3 -> LATIN CAPITAL LETTER N WITH ACUTE + '\u0144' # 0x00e4 -> LATIN SMALL LETTER N WITH ACUTE + '\u0148' # 0x00e5 -> LATIN SMALL LETTER N WITH CARON + '\u0160' # 0x00e6 -> LATIN CAPITAL LETTER S WITH CARON + '\u0161' # 0x00e7 -> LATIN SMALL LETTER S WITH CARON + '\u0154' # 0x00e8 -> LATIN CAPITAL LETTER R WITH ACUTE + '\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE + '\u0155' # 0x00ea -> LATIN SMALL LETTER R WITH ACUTE + '\u0170' # 0x00eb -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE + '\xfd' # 0x00ec -> LATIN SMALL LETTER Y WITH ACUTE + '\xdd' # 0x00ed -> LATIN CAPITAL LETTER Y WITH ACUTE + '\u0163' # 0x00ee -> LATIN SMALL LETTER T WITH CEDILLA + '\xb4' # 0x00ef -> ACUTE ACCENT + '\xad' # 0x00f0 -> SOFT HYPHEN + '\u02dd' # 0x00f1 -> DOUBLE ACUTE ACCENT + '\u02db' # 0x00f2 -> OGONEK + '\u02c7' # 0x00f3 -> CARON + '\u02d8' # 0x00f4 -> BREVE + '\xa7' # 0x00f5 -> SECTION SIGN + '\xf7' # 0x00f6 -> DIVISION SIGN + '\xb8' # 0x00f7 -> CEDILLA + '\xb0' # 0x00f8 -> DEGREE SIGN + '\xa8' # 0x00f9 -> DIAERESIS + '\u02d9' # 0x00fa -> DOT ABOVE + '\u0171' # 0x00fb -> LATIN SMALL LETTER U WITH DOUBLE ACUTE + '\u0158' # 0x00fc -> LATIN CAPITAL LETTER R WITH CARON + '\u0159' # 0x00fd -> LATIN SMALL LETTER R WITH CARON + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp855.py b/Lib/encodings/cp855.py index 241ef9d..4fe9210 100644 --- a/Lib/encodings/cp855.py +++ b/Lib/encodings/cp855.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\u0452' # 0x0080 -> CYRILLIC SMALL LETTER DJE - u'\u0402' # 0x0081 -> CYRILLIC CAPITAL LETTER DJE - u'\u0453' # 0x0082 -> CYRILLIC SMALL LETTER GJE - u'\u0403' # 0x0083 -> CYRILLIC CAPITAL LETTER GJE - u'\u0451' # 0x0084 -> CYRILLIC SMALL LETTER IO - u'\u0401' # 0x0085 -> CYRILLIC CAPITAL LETTER IO - u'\u0454' # 0x0086 -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\u0404' # 0x0087 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\u0455' # 0x0088 -> CYRILLIC SMALL LETTER DZE - u'\u0405' # 0x0089 -> CYRILLIC CAPITAL LETTER DZE - u'\u0456' # 0x008a -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0406' # 0x008b -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0457' # 0x008c -> CYRILLIC SMALL LETTER YI - u'\u0407' # 0x008d -> CYRILLIC CAPITAL LETTER YI - u'\u0458' # 0x008e -> CYRILLIC SMALL LETTER JE - u'\u0408' # 0x008f -> CYRILLIC CAPITAL LETTER JE - u'\u0459' # 0x0090 -> CYRILLIC SMALL LETTER LJE - u'\u0409' # 0x0091 -> CYRILLIC CAPITAL LETTER LJE - u'\u045a' # 0x0092 -> CYRILLIC SMALL LETTER NJE - u'\u040a' # 0x0093 -> CYRILLIC CAPITAL LETTER NJE - u'\u045b' # 0x0094 -> CYRILLIC SMALL LETTER TSHE - u'\u040b' # 0x0095 -> CYRILLIC CAPITAL LETTER TSHE - u'\u045c' # 0x0096 -> CYRILLIC SMALL LETTER KJE - u'\u040c' # 0x0097 -> CYRILLIC CAPITAL LETTER KJE - u'\u045e' # 0x0098 -> CYRILLIC SMALL LETTER SHORT U - u'\u040e' # 0x0099 -> CYRILLIC CAPITAL LETTER SHORT U - u'\u045f' # 0x009a -> CYRILLIC SMALL LETTER DZHE - u'\u040f' # 0x009b -> CYRILLIC CAPITAL LETTER DZHE - u'\u044e' # 0x009c -> CYRILLIC SMALL LETTER YU - u'\u042e' # 0x009d -> CYRILLIC CAPITAL LETTER YU - u'\u044a' # 0x009e -> CYRILLIC SMALL LETTER HARD SIGN - u'\u042a' # 0x009f -> CYRILLIC CAPITAL LETTER HARD SIGN - u'\u0430' # 0x00a0 -> CYRILLIC SMALL LETTER A - u'\u0410' # 0x00a1 -> CYRILLIC CAPITAL LETTER A - u'\u0431' # 0x00a2 -> CYRILLIC SMALL LETTER BE - u'\u0411' # 0x00a3 -> CYRILLIC CAPITAL LETTER BE - u'\u0446' # 0x00a4 -> CYRILLIC SMALL LETTER TSE - u'\u0426' # 0x00a5 -> CYRILLIC CAPITAL LETTER TSE - u'\u0434' # 0x00a6 -> CYRILLIC SMALL LETTER DE - u'\u0414' # 0x00a7 -> CYRILLIC CAPITAL LETTER DE - u'\u0435' # 0x00a8 -> CYRILLIC SMALL LETTER IE - u'\u0415' # 0x00a9 -> CYRILLIC CAPITAL LETTER IE - u'\u0444' # 0x00aa -> CYRILLIC SMALL LETTER EF - u'\u0424' # 0x00ab -> CYRILLIC CAPITAL LETTER EF - u'\u0433' # 0x00ac -> CYRILLIC SMALL LETTER GHE - u'\u0413' # 0x00ad -> CYRILLIC CAPITAL LETTER GHE - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u0445' # 0x00b5 -> CYRILLIC SMALL LETTER HA - u'\u0425' # 0x00b6 -> CYRILLIC CAPITAL LETTER HA - u'\u0438' # 0x00b7 -> CYRILLIC SMALL LETTER I - u'\u0418' # 0x00b8 -> CYRILLIC CAPITAL LETTER I - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u0439' # 0x00bd -> CYRILLIC SMALL LETTER SHORT I - u'\u0419' # 0x00be -> CYRILLIC CAPITAL LETTER SHORT I - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u043a' # 0x00c6 -> CYRILLIC SMALL LETTER KA - u'\u041a' # 0x00c7 -> CYRILLIC CAPITAL LETTER KA - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa4' # 0x00cf -> CURRENCY SIGN - u'\u043b' # 0x00d0 -> CYRILLIC SMALL LETTER EL - u'\u041b' # 0x00d1 -> CYRILLIC CAPITAL LETTER EL - u'\u043c' # 0x00d2 -> CYRILLIC SMALL LETTER EM - u'\u041c' # 0x00d3 -> CYRILLIC CAPITAL LETTER EM - u'\u043d' # 0x00d4 -> CYRILLIC SMALL LETTER EN - u'\u041d' # 0x00d5 -> CYRILLIC CAPITAL LETTER EN - u'\u043e' # 0x00d6 -> CYRILLIC SMALL LETTER O - u'\u041e' # 0x00d7 -> CYRILLIC CAPITAL LETTER O - u'\u043f' # 0x00d8 -> CYRILLIC SMALL LETTER PE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u041f' # 0x00dd -> CYRILLIC CAPITAL LETTER PE - u'\u044f' # 0x00de -> CYRILLIC SMALL LETTER YA - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u042f' # 0x00e0 -> CYRILLIC CAPITAL LETTER YA - u'\u0440' # 0x00e1 -> CYRILLIC SMALL LETTER ER - u'\u0420' # 0x00e2 -> CYRILLIC CAPITAL LETTER ER - u'\u0441' # 0x00e3 -> CYRILLIC SMALL LETTER ES - u'\u0421' # 0x00e4 -> CYRILLIC CAPITAL LETTER ES - u'\u0442' # 0x00e5 -> CYRILLIC SMALL LETTER TE - u'\u0422' # 0x00e6 -> CYRILLIC CAPITAL LETTER TE - u'\u0443' # 0x00e7 -> CYRILLIC SMALL LETTER U - u'\u0423' # 0x00e8 -> CYRILLIC CAPITAL LETTER U - u'\u0436' # 0x00e9 -> CYRILLIC SMALL LETTER ZHE - u'\u0416' # 0x00ea -> CYRILLIC CAPITAL LETTER ZHE - u'\u0432' # 0x00eb -> CYRILLIC SMALL LETTER VE - u'\u0412' # 0x00ec -> CYRILLIC CAPITAL LETTER VE - u'\u044c' # 0x00ed -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u042c' # 0x00ee -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u2116' # 0x00ef -> NUMERO SIGN - u'\xad' # 0x00f0 -> SOFT HYPHEN - u'\u044b' # 0x00f1 -> CYRILLIC SMALL LETTER YERU - u'\u042b' # 0x00f2 -> CYRILLIC CAPITAL LETTER YERU - u'\u0437' # 0x00f3 -> CYRILLIC SMALL LETTER ZE - u'\u0417' # 0x00f4 -> CYRILLIC CAPITAL LETTER ZE - u'\u0448' # 0x00f5 -> CYRILLIC SMALL LETTER SHA - u'\u0428' # 0x00f6 -> CYRILLIC CAPITAL LETTER SHA - u'\u044d' # 0x00f7 -> CYRILLIC SMALL LETTER E - u'\u042d' # 0x00f8 -> CYRILLIC CAPITAL LETTER E - u'\u0449' # 0x00f9 -> CYRILLIC SMALL LETTER SHCHA - u'\u0429' # 0x00fa -> CYRILLIC CAPITAL LETTER SHCHA - u'\u0447' # 0x00fb -> CYRILLIC SMALL LETTER CHE - u'\u0427' # 0x00fc -> CYRILLIC CAPITAL LETTER CHE - u'\xa7' # 0x00fd -> SECTION SIGN - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\u0452' # 0x0080 -> CYRILLIC SMALL LETTER DJE + '\u0402' # 0x0081 -> CYRILLIC CAPITAL LETTER DJE + '\u0453' # 0x0082 -> CYRILLIC SMALL LETTER GJE + '\u0403' # 0x0083 -> CYRILLIC CAPITAL LETTER GJE + '\u0451' # 0x0084 -> CYRILLIC SMALL LETTER IO + '\u0401' # 0x0085 -> CYRILLIC CAPITAL LETTER IO + '\u0454' # 0x0086 -> CYRILLIC SMALL LETTER UKRAINIAN IE + '\u0404' # 0x0087 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE + '\u0455' # 0x0088 -> CYRILLIC SMALL LETTER DZE + '\u0405' # 0x0089 -> CYRILLIC CAPITAL LETTER DZE + '\u0456' # 0x008a -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0406' # 0x008b -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0457' # 0x008c -> CYRILLIC SMALL LETTER YI + '\u0407' # 0x008d -> CYRILLIC CAPITAL LETTER YI + '\u0458' # 0x008e -> CYRILLIC SMALL LETTER JE + '\u0408' # 0x008f -> CYRILLIC CAPITAL LETTER JE + '\u0459' # 0x0090 -> CYRILLIC SMALL LETTER LJE + '\u0409' # 0x0091 -> CYRILLIC CAPITAL LETTER LJE + '\u045a' # 0x0092 -> CYRILLIC SMALL LETTER NJE + '\u040a' # 0x0093 -> CYRILLIC CAPITAL LETTER NJE + '\u045b' # 0x0094 -> CYRILLIC SMALL LETTER TSHE + '\u040b' # 0x0095 -> CYRILLIC CAPITAL LETTER TSHE + '\u045c' # 0x0096 -> CYRILLIC SMALL LETTER KJE + '\u040c' # 0x0097 -> CYRILLIC CAPITAL LETTER KJE + '\u045e' # 0x0098 -> CYRILLIC SMALL LETTER SHORT U + '\u040e' # 0x0099 -> CYRILLIC CAPITAL LETTER SHORT U + '\u045f' # 0x009a -> CYRILLIC SMALL LETTER DZHE + '\u040f' # 0x009b -> CYRILLIC CAPITAL LETTER DZHE + '\u044e' # 0x009c -> CYRILLIC SMALL LETTER YU + '\u042e' # 0x009d -> CYRILLIC CAPITAL LETTER YU + '\u044a' # 0x009e -> CYRILLIC SMALL LETTER HARD SIGN + '\u042a' # 0x009f -> CYRILLIC CAPITAL LETTER HARD SIGN + '\u0430' # 0x00a0 -> CYRILLIC SMALL LETTER A + '\u0410' # 0x00a1 -> CYRILLIC CAPITAL LETTER A + '\u0431' # 0x00a2 -> CYRILLIC SMALL LETTER BE + '\u0411' # 0x00a3 -> CYRILLIC CAPITAL LETTER BE + '\u0446' # 0x00a4 -> CYRILLIC SMALL LETTER TSE + '\u0426' # 0x00a5 -> CYRILLIC CAPITAL LETTER TSE + '\u0434' # 0x00a6 -> CYRILLIC SMALL LETTER DE + '\u0414' # 0x00a7 -> CYRILLIC CAPITAL LETTER DE + '\u0435' # 0x00a8 -> CYRILLIC SMALL LETTER IE + '\u0415' # 0x00a9 -> CYRILLIC CAPITAL LETTER IE + '\u0444' # 0x00aa -> CYRILLIC SMALL LETTER EF + '\u0424' # 0x00ab -> CYRILLIC CAPITAL LETTER EF + '\u0433' # 0x00ac -> CYRILLIC SMALL LETTER GHE + '\u0413' # 0x00ad -> CYRILLIC CAPITAL LETTER GHE + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u0445' # 0x00b5 -> CYRILLIC SMALL LETTER HA + '\u0425' # 0x00b6 -> CYRILLIC CAPITAL LETTER HA + '\u0438' # 0x00b7 -> CYRILLIC SMALL LETTER I + '\u0418' # 0x00b8 -> CYRILLIC CAPITAL LETTER I + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u0439' # 0x00bd -> CYRILLIC SMALL LETTER SHORT I + '\u0419' # 0x00be -> CYRILLIC CAPITAL LETTER SHORT I + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u043a' # 0x00c6 -> CYRILLIC SMALL LETTER KA + '\u041a' # 0x00c7 -> CYRILLIC CAPITAL LETTER KA + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa4' # 0x00cf -> CURRENCY SIGN + '\u043b' # 0x00d0 -> CYRILLIC SMALL LETTER EL + '\u041b' # 0x00d1 -> CYRILLIC CAPITAL LETTER EL + '\u043c' # 0x00d2 -> CYRILLIC SMALL LETTER EM + '\u041c' # 0x00d3 -> CYRILLIC CAPITAL LETTER EM + '\u043d' # 0x00d4 -> CYRILLIC SMALL LETTER EN + '\u041d' # 0x00d5 -> CYRILLIC CAPITAL LETTER EN + '\u043e' # 0x00d6 -> CYRILLIC SMALL LETTER O + '\u041e' # 0x00d7 -> CYRILLIC CAPITAL LETTER O + '\u043f' # 0x00d8 -> CYRILLIC SMALL LETTER PE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u041f' # 0x00dd -> CYRILLIC CAPITAL LETTER PE + '\u044f' # 0x00de -> CYRILLIC SMALL LETTER YA + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u042f' # 0x00e0 -> CYRILLIC CAPITAL LETTER YA + '\u0440' # 0x00e1 -> CYRILLIC SMALL LETTER ER + '\u0420' # 0x00e2 -> CYRILLIC CAPITAL LETTER ER + '\u0441' # 0x00e3 -> CYRILLIC SMALL LETTER ES + '\u0421' # 0x00e4 -> CYRILLIC CAPITAL LETTER ES + '\u0442' # 0x00e5 -> CYRILLIC SMALL LETTER TE + '\u0422' # 0x00e6 -> CYRILLIC CAPITAL LETTER TE + '\u0443' # 0x00e7 -> CYRILLIC SMALL LETTER U + '\u0423' # 0x00e8 -> CYRILLIC CAPITAL LETTER U + '\u0436' # 0x00e9 -> CYRILLIC SMALL LETTER ZHE + '\u0416' # 0x00ea -> CYRILLIC CAPITAL LETTER ZHE + '\u0432' # 0x00eb -> CYRILLIC SMALL LETTER VE + '\u0412' # 0x00ec -> CYRILLIC CAPITAL LETTER VE + '\u044c' # 0x00ed -> CYRILLIC SMALL LETTER SOFT SIGN + '\u042c' # 0x00ee -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u2116' # 0x00ef -> NUMERO SIGN + '\xad' # 0x00f0 -> SOFT HYPHEN + '\u044b' # 0x00f1 -> CYRILLIC SMALL LETTER YERU + '\u042b' # 0x00f2 -> CYRILLIC CAPITAL LETTER YERU + '\u0437' # 0x00f3 -> CYRILLIC SMALL LETTER ZE + '\u0417' # 0x00f4 -> CYRILLIC CAPITAL LETTER ZE + '\u0448' # 0x00f5 -> CYRILLIC SMALL LETTER SHA + '\u0428' # 0x00f6 -> CYRILLIC CAPITAL LETTER SHA + '\u044d' # 0x00f7 -> CYRILLIC SMALL LETTER E + '\u042d' # 0x00f8 -> CYRILLIC CAPITAL LETTER E + '\u0449' # 0x00f9 -> CYRILLIC SMALL LETTER SHCHA + '\u0429' # 0x00fa -> CYRILLIC CAPITAL LETTER SHCHA + '\u0447' # 0x00fb -> CYRILLIC SMALL LETTER CHE + '\u0427' # 0x00fc -> CYRILLIC CAPITAL LETTER CHE + '\xa7' # 0x00fd -> SECTION SIGN + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp856.py b/Lib/encodings/cp856.py index 203c2c4..cacbfb2 100644 --- a/Lib/encodings/cp856.py +++ b/Lib/encodings/cp856.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u05d0' # 0x80 -> HEBREW LETTER ALEF - u'\u05d1' # 0x81 -> HEBREW LETTER BET - u'\u05d2' # 0x82 -> HEBREW LETTER GIMEL - u'\u05d3' # 0x83 -> HEBREW LETTER DALET - u'\u05d4' # 0x84 -> HEBREW LETTER HE - u'\u05d5' # 0x85 -> HEBREW LETTER VAV - u'\u05d6' # 0x86 -> HEBREW LETTER ZAYIN - u'\u05d7' # 0x87 -> HEBREW LETTER HET - u'\u05d8' # 0x88 -> HEBREW LETTER TET - u'\u05d9' # 0x89 -> HEBREW LETTER YOD - u'\u05da' # 0x8A -> HEBREW LETTER FINAL KAF - u'\u05db' # 0x8B -> HEBREW LETTER KAF - u'\u05dc' # 0x8C -> HEBREW LETTER LAMED - u'\u05dd' # 0x8D -> HEBREW LETTER FINAL MEM - u'\u05de' # 0x8E -> HEBREW LETTER MEM - u'\u05df' # 0x8F -> HEBREW LETTER FINAL NUN - u'\u05e0' # 0x90 -> HEBREW LETTER NUN - u'\u05e1' # 0x91 -> HEBREW LETTER SAMEKH - u'\u05e2' # 0x92 -> HEBREW LETTER AYIN - u'\u05e3' # 0x93 -> HEBREW LETTER FINAL PE - u'\u05e4' # 0x94 -> HEBREW LETTER PE - u'\u05e5' # 0x95 -> HEBREW LETTER FINAL TSADI - u'\u05e6' # 0x96 -> HEBREW LETTER TSADI - u'\u05e7' # 0x97 -> HEBREW LETTER QOF - u'\u05e8' # 0x98 -> HEBREW LETTER RESH - u'\u05e9' # 0x99 -> HEBREW LETTER SHIN - u'\u05ea' # 0x9A -> HEBREW LETTER TAV - u'\ufffe' # 0x9B -> UNDEFINED - u'\xa3' # 0x9C -> POUND SIGN - u'\ufffe' # 0x9D -> UNDEFINED - u'\xd7' # 0x9E -> MULTIPLICATION SIGN - u'\ufffe' # 0x9F -> UNDEFINED - u'\ufffe' # 0xA0 -> UNDEFINED - u'\ufffe' # 0xA1 -> UNDEFINED - u'\ufffe' # 0xA2 -> UNDEFINED - u'\ufffe' # 0xA3 -> UNDEFINED - u'\ufffe' # 0xA4 -> UNDEFINED - u'\ufffe' # 0xA5 -> UNDEFINED - u'\ufffe' # 0xA6 -> UNDEFINED - u'\ufffe' # 0xA7 -> UNDEFINED - u'\ufffe' # 0xA8 -> UNDEFINED - u'\xae' # 0xA9 -> REGISTERED SIGN - u'\xac' # 0xAA -> NOT SIGN - u'\xbd' # 0xAB -> VULGAR FRACTION ONE HALF - u'\xbc' # 0xAC -> VULGAR FRACTION ONE QUARTER - u'\ufffe' # 0xAD -> UNDEFINED - u'\xab' # 0xAE -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xAF -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0xB0 -> LIGHT SHADE - u'\u2592' # 0xB1 -> MEDIUM SHADE - u'\u2593' # 0xB2 -> DARK SHADE - u'\u2502' # 0xB3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0xB4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\ufffe' # 0xB5 -> UNDEFINED - u'\ufffe' # 0xB6 -> UNDEFINED - u'\ufffe' # 0xB7 -> UNDEFINED - u'\xa9' # 0xB8 -> COPYRIGHT SIGN - u'\u2563' # 0xB9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0xBA -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0xBB -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0xBC -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\xa2' # 0xBD -> CENT SIGN - u'\xa5' # 0xBE -> YEN SIGN - u'\u2510' # 0xBF -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0xC0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0xC1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0xC2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0xC3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0xC4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0xC5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\ufffe' # 0xC6 -> UNDEFINED - u'\ufffe' # 0xC7 -> UNDEFINED - u'\u255a' # 0xC8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0xC9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0xCA -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0xCB -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0xCC -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0xCD -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0xCE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa4' # 0xCF -> CURRENCY SIGN - u'\ufffe' # 0xD0 -> UNDEFINED - u'\ufffe' # 0xD1 -> UNDEFINED - u'\ufffe' # 0xD2 -> UNDEFINED - u'\ufffe' # 0xD3 -> UNDEFINEDS - u'\ufffe' # 0xD4 -> UNDEFINED - u'\ufffe' # 0xD5 -> UNDEFINED - u'\ufffe' # 0xD6 -> UNDEFINEDE - u'\ufffe' # 0xD7 -> UNDEFINED - u'\ufffe' # 0xD8 -> UNDEFINED - u'\u2518' # 0xD9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0xDA -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0xDB -> FULL BLOCK - u'\u2584' # 0xDC -> LOWER HALF BLOCK - u'\xa6' # 0xDD -> BROKEN BAR - u'\ufffe' # 0xDE -> UNDEFINED - u'\u2580' # 0xDF -> UPPER HALF BLOCK - u'\ufffe' # 0xE0 -> UNDEFINED - u'\ufffe' # 0xE1 -> UNDEFINED - u'\ufffe' # 0xE2 -> UNDEFINED - u'\ufffe' # 0xE3 -> UNDEFINED - u'\ufffe' # 0xE4 -> UNDEFINED - u'\ufffe' # 0xE5 -> UNDEFINED - u'\xb5' # 0xE6 -> MICRO SIGN - u'\ufffe' # 0xE7 -> UNDEFINED - u'\ufffe' # 0xE8 -> UNDEFINED - u'\ufffe' # 0xE9 -> UNDEFINED - u'\ufffe' # 0xEA -> UNDEFINED - u'\ufffe' # 0xEB -> UNDEFINED - u'\ufffe' # 0xEC -> UNDEFINED - u'\ufffe' # 0xED -> UNDEFINED - u'\xaf' # 0xEE -> MACRON - u'\xb4' # 0xEF -> ACUTE ACCENT - u'\xad' # 0xF0 -> SOFT HYPHEN - u'\xb1' # 0xF1 -> PLUS-MINUS SIGN - u'\u2017' # 0xF2 -> DOUBLE LOW LINE - u'\xbe' # 0xF3 -> VULGAR FRACTION THREE QUARTERS - u'\xb6' # 0xF4 -> PILCROW SIGN - u'\xa7' # 0xF5 -> SECTION SIGN - u'\xf7' # 0xF6 -> DIVISION SIGN - u'\xb8' # 0xF7 -> CEDILLA - u'\xb0' # 0xF8 -> DEGREE SIGN - u'\xa8' # 0xF9 -> DIAERESIS - u'\xb7' # 0xFA -> MIDDLE DOT - u'\xb9' # 0xFB -> SUPERSCRIPT ONE - u'\xb3' # 0xFC -> SUPERSCRIPT THREE - u'\xb2' # 0xFD -> SUPERSCRIPT TWO - u'\u25a0' # 0xFE -> BLACK SQUARE - u'\xa0' # 0xFF -> NO-BREAK SPACE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u05d0' # 0x80 -> HEBREW LETTER ALEF + '\u05d1' # 0x81 -> HEBREW LETTER BET + '\u05d2' # 0x82 -> HEBREW LETTER GIMEL + '\u05d3' # 0x83 -> HEBREW LETTER DALET + '\u05d4' # 0x84 -> HEBREW LETTER HE + '\u05d5' # 0x85 -> HEBREW LETTER VAV + '\u05d6' # 0x86 -> HEBREW LETTER ZAYIN + '\u05d7' # 0x87 -> HEBREW LETTER HET + '\u05d8' # 0x88 -> HEBREW LETTER TET + '\u05d9' # 0x89 -> HEBREW LETTER YOD + '\u05da' # 0x8A -> HEBREW LETTER FINAL KAF + '\u05db' # 0x8B -> HEBREW LETTER KAF + '\u05dc' # 0x8C -> HEBREW LETTER LAMED + '\u05dd' # 0x8D -> HEBREW LETTER FINAL MEM + '\u05de' # 0x8E -> HEBREW LETTER MEM + '\u05df' # 0x8F -> HEBREW LETTER FINAL NUN + '\u05e0' # 0x90 -> HEBREW LETTER NUN + '\u05e1' # 0x91 -> HEBREW LETTER SAMEKH + '\u05e2' # 0x92 -> HEBREW LETTER AYIN + '\u05e3' # 0x93 -> HEBREW LETTER FINAL PE + '\u05e4' # 0x94 -> HEBREW LETTER PE + '\u05e5' # 0x95 -> HEBREW LETTER FINAL TSADI + '\u05e6' # 0x96 -> HEBREW LETTER TSADI + '\u05e7' # 0x97 -> HEBREW LETTER QOF + '\u05e8' # 0x98 -> HEBREW LETTER RESH + '\u05e9' # 0x99 -> HEBREW LETTER SHIN + '\u05ea' # 0x9A -> HEBREW LETTER TAV + '\ufffe' # 0x9B -> UNDEFINED + '\xa3' # 0x9C -> POUND SIGN + '\ufffe' # 0x9D -> UNDEFINED + '\xd7' # 0x9E -> MULTIPLICATION SIGN + '\ufffe' # 0x9F -> UNDEFINED + '\ufffe' # 0xA0 -> UNDEFINED + '\ufffe' # 0xA1 -> UNDEFINED + '\ufffe' # 0xA2 -> UNDEFINED + '\ufffe' # 0xA3 -> UNDEFINED + '\ufffe' # 0xA4 -> UNDEFINED + '\ufffe' # 0xA5 -> UNDEFINED + '\ufffe' # 0xA6 -> UNDEFINED + '\ufffe' # 0xA7 -> UNDEFINED + '\ufffe' # 0xA8 -> UNDEFINED + '\xae' # 0xA9 -> REGISTERED SIGN + '\xac' # 0xAA -> NOT SIGN + '\xbd' # 0xAB -> VULGAR FRACTION ONE HALF + '\xbc' # 0xAC -> VULGAR FRACTION ONE QUARTER + '\ufffe' # 0xAD -> UNDEFINED + '\xab' # 0xAE -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xAF -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0xB0 -> LIGHT SHADE + '\u2592' # 0xB1 -> MEDIUM SHADE + '\u2593' # 0xB2 -> DARK SHADE + '\u2502' # 0xB3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0xB4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\ufffe' # 0xB5 -> UNDEFINED + '\ufffe' # 0xB6 -> UNDEFINED + '\ufffe' # 0xB7 -> UNDEFINED + '\xa9' # 0xB8 -> COPYRIGHT SIGN + '\u2563' # 0xB9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0xBA -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0xBB -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0xBC -> BOX DRAWINGS DOUBLE UP AND LEFT + '\xa2' # 0xBD -> CENT SIGN + '\xa5' # 0xBE -> YEN SIGN + '\u2510' # 0xBF -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0xC0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0xC1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0xC2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0xC3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0xC4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0xC5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\ufffe' # 0xC6 -> UNDEFINED + '\ufffe' # 0xC7 -> UNDEFINED + '\u255a' # 0xC8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0xC9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0xCA -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0xCB -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0xCC -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0xCD -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0xCE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa4' # 0xCF -> CURRENCY SIGN + '\ufffe' # 0xD0 -> UNDEFINED + '\ufffe' # 0xD1 -> UNDEFINED + '\ufffe' # 0xD2 -> UNDEFINED + '\ufffe' # 0xD3 -> UNDEFINEDS + '\ufffe' # 0xD4 -> UNDEFINED + '\ufffe' # 0xD5 -> UNDEFINED + '\ufffe' # 0xD6 -> UNDEFINEDE + '\ufffe' # 0xD7 -> UNDEFINED + '\ufffe' # 0xD8 -> UNDEFINED + '\u2518' # 0xD9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0xDA -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0xDB -> FULL BLOCK + '\u2584' # 0xDC -> LOWER HALF BLOCK + '\xa6' # 0xDD -> BROKEN BAR + '\ufffe' # 0xDE -> UNDEFINED + '\u2580' # 0xDF -> UPPER HALF BLOCK + '\ufffe' # 0xE0 -> UNDEFINED + '\ufffe' # 0xE1 -> UNDEFINED + '\ufffe' # 0xE2 -> UNDEFINED + '\ufffe' # 0xE3 -> UNDEFINED + '\ufffe' # 0xE4 -> UNDEFINED + '\ufffe' # 0xE5 -> UNDEFINED + '\xb5' # 0xE6 -> MICRO SIGN + '\ufffe' # 0xE7 -> UNDEFINED + '\ufffe' # 0xE8 -> UNDEFINED + '\ufffe' # 0xE9 -> UNDEFINED + '\ufffe' # 0xEA -> UNDEFINED + '\ufffe' # 0xEB -> UNDEFINED + '\ufffe' # 0xEC -> UNDEFINED + '\ufffe' # 0xED -> UNDEFINED + '\xaf' # 0xEE -> MACRON + '\xb4' # 0xEF -> ACUTE ACCENT + '\xad' # 0xF0 -> SOFT HYPHEN + '\xb1' # 0xF1 -> PLUS-MINUS SIGN + '\u2017' # 0xF2 -> DOUBLE LOW LINE + '\xbe' # 0xF3 -> VULGAR FRACTION THREE QUARTERS + '\xb6' # 0xF4 -> PILCROW SIGN + '\xa7' # 0xF5 -> SECTION SIGN + '\xf7' # 0xF6 -> DIVISION SIGN + '\xb8' # 0xF7 -> CEDILLA + '\xb0' # 0xF8 -> DEGREE SIGN + '\xa8' # 0xF9 -> DIAERESIS + '\xb7' # 0xFA -> MIDDLE DOT + '\xb9' # 0xFB -> SUPERSCRIPT ONE + '\xb3' # 0xFC -> SUPERSCRIPT THREE + '\xb2' # 0xFD -> SUPERSCRIPT TWO + '\u25a0' # 0xFE -> BLACK SQUARE + '\xa0' # 0xFF -> NO-BREAK SPACE ) ### Encoding table diff --git a/Lib/encodings/cp857.py b/Lib/encodings/cp857.py index c24191b..741b059 100644 --- a/Lib/encodings/cp857.py +++ b/Lib/encodings/cp857.py @@ -177,262 +177,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\u0131' # 0x008d -> LATIN SMALL LETTER DOTLESS I - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE - u'\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE - u'\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE - u'\u0130' # 0x0098 -> LATIN CAPITAL LETTER I WITH DOT ABOVE - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE - u'\xa3' # 0x009c -> POUND SIGN - u'\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE - u'\u015e' # 0x009e -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\u015f' # 0x009f -> LATIN SMALL LETTER S WITH CEDILLA - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE - u'\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE - u'\u011e' # 0x00a6 -> LATIN CAPITAL LETTER G WITH BREVE - u'\u011f' # 0x00a7 -> LATIN SMALL LETTER G WITH BREVE - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\xae' # 0x00a9 -> REGISTERED SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc0' # 0x00b7 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xa9' # 0x00b8 -> COPYRIGHT SIGN - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\xa2' # 0x00bd -> CENT SIGN - u'\xa5' # 0x00be -> YEN SIGN - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\xe3' # 0x00c6 -> LATIN SMALL LETTER A WITH TILDE - u'\xc3' # 0x00c7 -> LATIN CAPITAL LETTER A WITH TILDE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa4' # 0x00cf -> CURRENCY SIGN - u'\xba' # 0x00d0 -> MASCULINE ORDINAL INDICATOR - u'\xaa' # 0x00d1 -> FEMININE ORDINAL INDICATOR - u'\xca' # 0x00d2 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0x00d4 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\ufffe' # 0x00d5 -> UNDEFINED - u'\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0x00d8 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\xa6' # 0x00dd -> BROKEN BAR - u'\xcc' # 0x00de -> LATIN CAPITAL LETTER I WITH GRAVE - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd2' # 0x00e3 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xf5' # 0x00e4 -> LATIN SMALL LETTER O WITH TILDE - u'\xd5' # 0x00e5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\ufffe' # 0x00e7 -> UNDEFINED - u'\xd7' # 0x00e8 -> MULTIPLICATION SIGN - u'\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0x00ea -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0x00eb -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xec' # 0x00ec -> LATIN SMALL LETTER I WITH GRAVE - u'\xff' # 0x00ed -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\xaf' # 0x00ee -> MACRON - u'\xb4' # 0x00ef -> ACUTE ACCENT - u'\xad' # 0x00f0 -> SOFT HYPHEN - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\ufffe' # 0x00f2 -> UNDEFINED - u'\xbe' # 0x00f3 -> VULGAR FRACTION THREE QUARTERS - u'\xb6' # 0x00f4 -> PILCROW SIGN - u'\xa7' # 0x00f5 -> SECTION SIGN - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\xb8' # 0x00f7 -> CEDILLA - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\xa8' # 0x00f9 -> DIAERESIS - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\xb9' # 0x00fb -> SUPERSCRIPT ONE - u'\xb3' # 0x00fc -> SUPERSCRIPT THREE - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS + '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\u0131' # 0x008d -> LATIN SMALL LETTER DOTLESS I + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE + '\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE + '\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE + '\u0130' # 0x0098 -> LATIN CAPITAL LETTER I WITH DOT ABOVE + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE + '\xa3' # 0x009c -> POUND SIGN + '\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE + '\u015e' # 0x009e -> LATIN CAPITAL LETTER S WITH CEDILLA + '\u015f' # 0x009f -> LATIN SMALL LETTER S WITH CEDILLA + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE + '\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE + '\u011e' # 0x00a6 -> LATIN CAPITAL LETTER G WITH BREVE + '\u011f' # 0x00a7 -> LATIN SMALL LETTER G WITH BREVE + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\xae' # 0x00a9 -> REGISTERED SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc0' # 0x00b7 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xa9' # 0x00b8 -> COPYRIGHT SIGN + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\xa2' # 0x00bd -> CENT SIGN + '\xa5' # 0x00be -> YEN SIGN + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\xe3' # 0x00c6 -> LATIN SMALL LETTER A WITH TILDE + '\xc3' # 0x00c7 -> LATIN CAPITAL LETTER A WITH TILDE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa4' # 0x00cf -> CURRENCY SIGN + '\xba' # 0x00d0 -> MASCULINE ORDINAL INDICATOR + '\xaa' # 0x00d1 -> FEMININE ORDINAL INDICATOR + '\xca' # 0x00d2 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0x00d4 -> LATIN CAPITAL LETTER E WITH GRAVE + '\ufffe' # 0x00d5 -> UNDEFINED + '\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0x00d8 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\xa6' # 0x00dd -> BROKEN BAR + '\xcc' # 0x00de -> LATIN CAPITAL LETTER I WITH GRAVE + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd2' # 0x00e3 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xf5' # 0x00e4 -> LATIN SMALL LETTER O WITH TILDE + '\xd5' # 0x00e5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xb5' # 0x00e6 -> MICRO SIGN + '\ufffe' # 0x00e7 -> UNDEFINED + '\xd7' # 0x00e8 -> MULTIPLICATION SIGN + '\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0x00ea -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0x00eb -> LATIN CAPITAL LETTER U WITH GRAVE + '\xec' # 0x00ec -> LATIN SMALL LETTER I WITH GRAVE + '\xff' # 0x00ed -> LATIN SMALL LETTER Y WITH DIAERESIS + '\xaf' # 0x00ee -> MACRON + '\xb4' # 0x00ef -> ACUTE ACCENT + '\xad' # 0x00f0 -> SOFT HYPHEN + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\ufffe' # 0x00f2 -> UNDEFINED + '\xbe' # 0x00f3 -> VULGAR FRACTION THREE QUARTERS + '\xb6' # 0x00f4 -> PILCROW SIGN + '\xa7' # 0x00f5 -> SECTION SIGN + '\xf7' # 0x00f6 -> DIVISION SIGN + '\xb8' # 0x00f7 -> CEDILLA + '\xb0' # 0x00f8 -> DEGREE SIGN + '\xa8' # 0x00f9 -> DIAERESIS + '\xb7' # 0x00fa -> MIDDLE DOT + '\xb9' # 0x00fb -> SUPERSCRIPT ONE + '\xb3' # 0x00fc -> SUPERSCRIPT THREE + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp860.py b/Lib/encodings/cp860.py index 4acb0cf..65903e7 100644 --- a/Lib/encodings/cp860.py +++ b/Lib/encodings/cp860.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0x0084 -> LATIN SMALL LETTER A WITH TILDE - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xc1' # 0x0086 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xca' # 0x0089 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xcd' # 0x008b -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xd4' # 0x008c -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE - u'\xc3' # 0x008e -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc2' # 0x008f -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xc0' # 0x0091 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc8' # 0x0092 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0x0094 -> LATIN SMALL LETTER O WITH TILDE - u'\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE - u'\xda' # 0x0096 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE - u'\xcc' # 0x0098 -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xd5' # 0x0099 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xa2' # 0x009b -> CENT SIGN - u'\xa3' # 0x009c -> POUND SIGN - u'\xd9' # 0x009d -> LATIN CAPITAL LETTER U WITH GRAVE - u'\u20a7' # 0x009e -> PESETA SIGN - u'\xd3' # 0x009f -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE - u'\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\xd2' # 0x00a9 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA - u'\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI - u'\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA - u'\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU - u'\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI - u'\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA - u'\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA - u'\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA - u'\u221e' # 0x00ec -> INFINITY - u'\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI - u'\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON - u'\u2229' # 0x00ef -> INTERSECTION - u'\u2261' # 0x00f0 -> IDENTICAL TO - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u2320' # 0x00f4 -> TOP HALF INTEGRAL - u'\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0x0084 -> LATIN SMALL LETTER A WITH TILDE + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xc1' # 0x0086 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xca' # 0x0089 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xcd' # 0x008b -> LATIN CAPITAL LETTER I WITH ACUTE + '\xd4' # 0x008c -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE + '\xc3' # 0x008e -> LATIN CAPITAL LETTER A WITH TILDE + '\xc2' # 0x008f -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xc0' # 0x0091 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc8' # 0x0092 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0x0094 -> LATIN SMALL LETTER O WITH TILDE + '\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE + '\xda' # 0x0096 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE + '\xcc' # 0x0098 -> LATIN CAPITAL LETTER I WITH GRAVE + '\xd5' # 0x0099 -> LATIN CAPITAL LETTER O WITH TILDE + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xa2' # 0x009b -> CENT SIGN + '\xa3' # 0x009c -> POUND SIGN + '\xd9' # 0x009d -> LATIN CAPITAL LETTER U WITH GRAVE + '\u20a7' # 0x009e -> PESETA SIGN + '\xd3' # 0x009f -> LATIN CAPITAL LETTER O WITH ACUTE + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE + '\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE + '\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR + '\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\xd2' # 0x00a9 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA + '\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI + '\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA + '\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA + '\xb5' # 0x00e6 -> MICRO SIGN + '\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU + '\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI + '\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA + '\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA + '\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA + '\u221e' # 0x00ec -> INFINITY + '\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI + '\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON + '\u2229' # 0x00ef -> INTERSECTION + '\u2261' # 0x00f0 -> IDENTICAL TO + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u2320' # 0x00f4 -> TOP HALF INTEGRAL + '\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp861.py b/Lib/encodings/cp861.py index 0939b5b..860a05f 100644 --- a/Lib/encodings/cp861.py +++ b/Lib/encodings/cp861.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xd0' # 0x008b -> LATIN CAPITAL LETTER ETH - u'\xf0' # 0x008c -> LATIN SMALL LETTER ETH - u'\xde' # 0x008d -> LATIN CAPITAL LETTER THORN - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE - u'\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xfe' # 0x0095 -> LATIN SMALL LETTER THORN - u'\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xdd' # 0x0097 -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xfd' # 0x0098 -> LATIN SMALL LETTER Y WITH ACUTE - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE - u'\xa3' # 0x009c -> POUND SIGN - u'\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE - u'\u20a7' # 0x009e -> PESETA SIGN - u'\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xc1' # 0x00a4 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xcd' # 0x00a5 -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xd3' # 0x00a6 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xda' # 0x00a7 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\u2310' # 0x00a9 -> REVERSED NOT SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA - u'\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI - u'\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA - u'\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU - u'\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI - u'\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA - u'\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA - u'\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA - u'\u221e' # 0x00ec -> INFINITY - u'\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI - u'\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON - u'\u2229' # 0x00ef -> INTERSECTION - u'\u2261' # 0x00f0 -> IDENTICAL TO - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u2320' # 0x00f4 -> TOP HALF INTEGRAL - u'\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xd0' # 0x008b -> LATIN CAPITAL LETTER ETH + '\xf0' # 0x008c -> LATIN SMALL LETTER ETH + '\xde' # 0x008d -> LATIN CAPITAL LETTER THORN + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE + '\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xfe' # 0x0095 -> LATIN SMALL LETTER THORN + '\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xdd' # 0x0097 -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xfd' # 0x0098 -> LATIN SMALL LETTER Y WITH ACUTE + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE + '\xa3' # 0x009c -> POUND SIGN + '\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE + '\u20a7' # 0x009e -> PESETA SIGN + '\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xc1' # 0x00a4 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xcd' # 0x00a5 -> LATIN CAPITAL LETTER I WITH ACUTE + '\xd3' # 0x00a6 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xda' # 0x00a7 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\u2310' # 0x00a9 -> REVERSED NOT SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA + '\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI + '\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA + '\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA + '\xb5' # 0x00e6 -> MICRO SIGN + '\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU + '\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI + '\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA + '\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA + '\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA + '\u221e' # 0x00ec -> INFINITY + '\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI + '\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON + '\u2229' # 0x00ef -> INTERSECTION + '\u2261' # 0x00f0 -> IDENTICAL TO + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u2320' # 0x00f4 -> TOP HALF INTEGRAL + '\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp862.py b/Lib/encodings/cp862.py index ea0405c..3df22f9 100644 --- a/Lib/encodings/cp862.py +++ b/Lib/encodings/cp862.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\u05d0' # 0x0080 -> HEBREW LETTER ALEF - u'\u05d1' # 0x0081 -> HEBREW LETTER BET - u'\u05d2' # 0x0082 -> HEBREW LETTER GIMEL - u'\u05d3' # 0x0083 -> HEBREW LETTER DALET - u'\u05d4' # 0x0084 -> HEBREW LETTER HE - u'\u05d5' # 0x0085 -> HEBREW LETTER VAV - u'\u05d6' # 0x0086 -> HEBREW LETTER ZAYIN - u'\u05d7' # 0x0087 -> HEBREW LETTER HET - u'\u05d8' # 0x0088 -> HEBREW LETTER TET - u'\u05d9' # 0x0089 -> HEBREW LETTER YOD - u'\u05da' # 0x008a -> HEBREW LETTER FINAL KAF - u'\u05db' # 0x008b -> HEBREW LETTER KAF - u'\u05dc' # 0x008c -> HEBREW LETTER LAMED - u'\u05dd' # 0x008d -> HEBREW LETTER FINAL MEM - u'\u05de' # 0x008e -> HEBREW LETTER MEM - u'\u05df' # 0x008f -> HEBREW LETTER FINAL NUN - u'\u05e0' # 0x0090 -> HEBREW LETTER NUN - u'\u05e1' # 0x0091 -> HEBREW LETTER SAMEKH - u'\u05e2' # 0x0092 -> HEBREW LETTER AYIN - u'\u05e3' # 0x0093 -> HEBREW LETTER FINAL PE - u'\u05e4' # 0x0094 -> HEBREW LETTER PE - u'\u05e5' # 0x0095 -> HEBREW LETTER FINAL TSADI - u'\u05e6' # 0x0096 -> HEBREW LETTER TSADI - u'\u05e7' # 0x0097 -> HEBREW LETTER QOF - u'\u05e8' # 0x0098 -> HEBREW LETTER RESH - u'\u05e9' # 0x0099 -> HEBREW LETTER SHIN - u'\u05ea' # 0x009a -> HEBREW LETTER TAV - u'\xa2' # 0x009b -> CENT SIGN - u'\xa3' # 0x009c -> POUND SIGN - u'\xa5' # 0x009d -> YEN SIGN - u'\u20a7' # 0x009e -> PESETA SIGN - u'\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE - u'\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\u2310' # 0x00a9 -> REVERSED NOT SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S (GERMAN) - u'\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA - u'\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI - u'\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA - u'\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU - u'\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI - u'\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA - u'\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA - u'\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA - u'\u221e' # 0x00ec -> INFINITY - u'\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI - u'\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON - u'\u2229' # 0x00ef -> INTERSECTION - u'\u2261' # 0x00f0 -> IDENTICAL TO - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u2320' # 0x00f4 -> TOP HALF INTEGRAL - u'\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\u05d0' # 0x0080 -> HEBREW LETTER ALEF + '\u05d1' # 0x0081 -> HEBREW LETTER BET + '\u05d2' # 0x0082 -> HEBREW LETTER GIMEL + '\u05d3' # 0x0083 -> HEBREW LETTER DALET + '\u05d4' # 0x0084 -> HEBREW LETTER HE + '\u05d5' # 0x0085 -> HEBREW LETTER VAV + '\u05d6' # 0x0086 -> HEBREW LETTER ZAYIN + '\u05d7' # 0x0087 -> HEBREW LETTER HET + '\u05d8' # 0x0088 -> HEBREW LETTER TET + '\u05d9' # 0x0089 -> HEBREW LETTER YOD + '\u05da' # 0x008a -> HEBREW LETTER FINAL KAF + '\u05db' # 0x008b -> HEBREW LETTER KAF + '\u05dc' # 0x008c -> HEBREW LETTER LAMED + '\u05dd' # 0x008d -> HEBREW LETTER FINAL MEM + '\u05de' # 0x008e -> HEBREW LETTER MEM + '\u05df' # 0x008f -> HEBREW LETTER FINAL NUN + '\u05e0' # 0x0090 -> HEBREW LETTER NUN + '\u05e1' # 0x0091 -> HEBREW LETTER SAMEKH + '\u05e2' # 0x0092 -> HEBREW LETTER AYIN + '\u05e3' # 0x0093 -> HEBREW LETTER FINAL PE + '\u05e4' # 0x0094 -> HEBREW LETTER PE + '\u05e5' # 0x0095 -> HEBREW LETTER FINAL TSADI + '\u05e6' # 0x0096 -> HEBREW LETTER TSADI + '\u05e7' # 0x0097 -> HEBREW LETTER QOF + '\u05e8' # 0x0098 -> HEBREW LETTER RESH + '\u05e9' # 0x0099 -> HEBREW LETTER SHIN + '\u05ea' # 0x009a -> HEBREW LETTER TAV + '\xa2' # 0x009b -> CENT SIGN + '\xa3' # 0x009c -> POUND SIGN + '\xa5' # 0x009d -> YEN SIGN + '\u20a7' # 0x009e -> PESETA SIGN + '\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE + '\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE + '\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR + '\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\u2310' # 0x00a9 -> REVERSED NOT SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S (GERMAN) + '\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA + '\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI + '\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA + '\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA + '\xb5' # 0x00e6 -> MICRO SIGN + '\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU + '\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI + '\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA + '\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA + '\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA + '\u221e' # 0x00ec -> INFINITY + '\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI + '\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON + '\u2229' # 0x00ef -> INTERSECTION + '\u2261' # 0x00f0 -> IDENTICAL TO + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u2320' # 0x00f4 -> TOP HALF INTEGRAL + '\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp863.py b/Lib/encodings/cp863.py index 62dfabf..764180b 100644 --- a/Lib/encodings/cp863.py +++ b/Lib/encodings/cp863.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xc2' # 0x0084 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xb6' # 0x0086 -> PILCROW SIGN - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\u2017' # 0x008d -> DOUBLE LOW LINE - u'\xc0' # 0x008e -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xa7' # 0x008f -> SECTION SIGN - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xc8' # 0x0091 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xca' # 0x0092 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xcb' # 0x0094 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcf' # 0x0095 -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE - u'\xa4' # 0x0098 -> CURRENCY SIGN - u'\xd4' # 0x0099 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xa2' # 0x009b -> CENT SIGN - u'\xa3' # 0x009c -> POUND SIGN - u'\xd9' # 0x009d -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xdb' # 0x009e -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK - u'\xa6' # 0x00a0 -> BROKEN BAR - u'\xb4' # 0x00a1 -> ACUTE ACCENT - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xa8' # 0x00a4 -> DIAERESIS - u'\xb8' # 0x00a5 -> CEDILLA - u'\xb3' # 0x00a6 -> SUPERSCRIPT THREE - u'\xaf' # 0x00a7 -> MACRON - u'\xce' # 0x00a8 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\u2310' # 0x00a9 -> REVERSED NOT SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xbe' # 0x00ad -> VULGAR FRACTION THREE QUARTERS - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA - u'\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI - u'\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA - u'\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU - u'\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI - u'\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA - u'\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA - u'\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA - u'\u221e' # 0x00ec -> INFINITY - u'\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI - u'\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON - u'\u2229' # 0x00ef -> INTERSECTION - u'\u2261' # 0x00f0 -> IDENTICAL TO - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u2320' # 0x00f4 -> TOP HALF INTEGRAL - u'\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xc2' # 0x0084 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xb6' # 0x0086 -> PILCROW SIGN + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS + '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\u2017' # 0x008d -> DOUBLE LOW LINE + '\xc0' # 0x008e -> LATIN CAPITAL LETTER A WITH GRAVE + '\xa7' # 0x008f -> SECTION SIGN + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xc8' # 0x0091 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xca' # 0x0092 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xcb' # 0x0094 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcf' # 0x0095 -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE + '\xa4' # 0x0098 -> CURRENCY SIGN + '\xd4' # 0x0099 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xa2' # 0x009b -> CENT SIGN + '\xa3' # 0x009c -> POUND SIGN + '\xd9' # 0x009d -> LATIN CAPITAL LETTER U WITH GRAVE + '\xdb' # 0x009e -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK + '\xa6' # 0x00a0 -> BROKEN BAR + '\xb4' # 0x00a1 -> ACUTE ACCENT + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xa8' # 0x00a4 -> DIAERESIS + '\xb8' # 0x00a5 -> CEDILLA + '\xb3' # 0x00a6 -> SUPERSCRIPT THREE + '\xaf' # 0x00a7 -> MACRON + '\xce' # 0x00a8 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\u2310' # 0x00a9 -> REVERSED NOT SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xbe' # 0x00ad -> VULGAR FRACTION THREE QUARTERS + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA + '\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI + '\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA + '\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA + '\xb5' # 0x00e6 -> MICRO SIGN + '\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU + '\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI + '\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA + '\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA + '\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA + '\u221e' # 0x00ec -> INFINITY + '\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI + '\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON + '\u2229' # 0x00ef -> INTERSECTION + '\u2261' # 0x00f0 -> IDENTICAL TO + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u2320' # 0x00f4 -> TOP HALF INTEGRAL + '\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp864.py b/Lib/encodings/cp864.py index 02a0e73..53df482 100644 --- a/Lib/encodings/cp864.py +++ b/Lib/encodings/cp864.py @@ -176,262 +176,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'\u066a' # 0x0025 -> ARABIC PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xb0' # 0x0080 -> DEGREE SIGN - u'\xb7' # 0x0081 -> MIDDLE DOT - u'\u2219' # 0x0082 -> BULLET OPERATOR - u'\u221a' # 0x0083 -> SQUARE ROOT - u'\u2592' # 0x0084 -> MEDIUM SHADE - u'\u2500' # 0x0085 -> FORMS LIGHT HORIZONTAL - u'\u2502' # 0x0086 -> FORMS LIGHT VERTICAL - u'\u253c' # 0x0087 -> FORMS LIGHT VERTICAL AND HORIZONTAL - u'\u2524' # 0x0088 -> FORMS LIGHT VERTICAL AND LEFT - u'\u252c' # 0x0089 -> FORMS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x008a -> FORMS LIGHT VERTICAL AND RIGHT - u'\u2534' # 0x008b -> FORMS LIGHT UP AND HORIZONTAL - u'\u2510' # 0x008c -> FORMS LIGHT DOWN AND LEFT - u'\u250c' # 0x008d -> FORMS LIGHT DOWN AND RIGHT - u'\u2514' # 0x008e -> FORMS LIGHT UP AND RIGHT - u'\u2518' # 0x008f -> FORMS LIGHT UP AND LEFT - u'\u03b2' # 0x0090 -> GREEK SMALL BETA - u'\u221e' # 0x0091 -> INFINITY - u'\u03c6' # 0x0092 -> GREEK SMALL PHI - u'\xb1' # 0x0093 -> PLUS-OR-MINUS SIGN - u'\xbd' # 0x0094 -> FRACTION 1/2 - u'\xbc' # 0x0095 -> FRACTION 1/4 - u'\u2248' # 0x0096 -> ALMOST EQUAL TO - u'\xab' # 0x0097 -> LEFT POINTING GUILLEMET - u'\xbb' # 0x0098 -> RIGHT POINTING GUILLEMET - u'\ufef7' # 0x0099 -> ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE ISOLATED FORM - u'\ufef8' # 0x009a -> ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE FINAL FORM - u'\ufffe' # 0x009b -> UNDEFINED - u'\ufffe' # 0x009c -> UNDEFINED - u'\ufefb' # 0x009d -> ARABIC LIGATURE LAM WITH ALEF ISOLATED FORM - u'\ufefc' # 0x009e -> ARABIC LIGATURE LAM WITH ALEF FINAL FORM - u'\ufffe' # 0x009f -> UNDEFINED - u'\xa0' # 0x00a0 -> NON-BREAKING SPACE - u'\xad' # 0x00a1 -> SOFT HYPHEN - u'\ufe82' # 0x00a2 -> ARABIC LETTER ALEF WITH MADDA ABOVE FINAL FORM - u'\xa3' # 0x00a3 -> POUND SIGN - u'\xa4' # 0x00a4 -> CURRENCY SIGN - u'\ufe84' # 0x00a5 -> ARABIC LETTER ALEF WITH HAMZA ABOVE FINAL FORM - u'\ufffe' # 0x00a6 -> UNDEFINED - u'\ufffe' # 0x00a7 -> UNDEFINED - u'\ufe8e' # 0x00a8 -> ARABIC LETTER ALEF FINAL FORM - u'\ufe8f' # 0x00a9 -> ARABIC LETTER BEH ISOLATED FORM - u'\ufe95' # 0x00aa -> ARABIC LETTER TEH ISOLATED FORM - u'\ufe99' # 0x00ab -> ARABIC LETTER THEH ISOLATED FORM - u'\u060c' # 0x00ac -> ARABIC COMMA - u'\ufe9d' # 0x00ad -> ARABIC LETTER JEEM ISOLATED FORM - u'\ufea1' # 0x00ae -> ARABIC LETTER HAH ISOLATED FORM - u'\ufea5' # 0x00af -> ARABIC LETTER KHAH ISOLATED FORM - u'\u0660' # 0x00b0 -> ARABIC-INDIC DIGIT ZERO - u'\u0661' # 0x00b1 -> ARABIC-INDIC DIGIT ONE - u'\u0662' # 0x00b2 -> ARABIC-INDIC DIGIT TWO - u'\u0663' # 0x00b3 -> ARABIC-INDIC DIGIT THREE - u'\u0664' # 0x00b4 -> ARABIC-INDIC DIGIT FOUR - u'\u0665' # 0x00b5 -> ARABIC-INDIC DIGIT FIVE - u'\u0666' # 0x00b6 -> ARABIC-INDIC DIGIT SIX - u'\u0667' # 0x00b7 -> ARABIC-INDIC DIGIT SEVEN - u'\u0668' # 0x00b8 -> ARABIC-INDIC DIGIT EIGHT - u'\u0669' # 0x00b9 -> ARABIC-INDIC DIGIT NINE - u'\ufed1' # 0x00ba -> ARABIC LETTER FEH ISOLATED FORM - u'\u061b' # 0x00bb -> ARABIC SEMICOLON - u'\ufeb1' # 0x00bc -> ARABIC LETTER SEEN ISOLATED FORM - u'\ufeb5' # 0x00bd -> ARABIC LETTER SHEEN ISOLATED FORM - u'\ufeb9' # 0x00be -> ARABIC LETTER SAD ISOLATED FORM - u'\u061f' # 0x00bf -> ARABIC QUESTION MARK - u'\xa2' # 0x00c0 -> CENT SIGN - u'\ufe80' # 0x00c1 -> ARABIC LETTER HAMZA ISOLATED FORM - u'\ufe81' # 0x00c2 -> ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM - u'\ufe83' # 0x00c3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE ISOLATED FORM - u'\ufe85' # 0x00c4 -> ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM - u'\ufeca' # 0x00c5 -> ARABIC LETTER AIN FINAL FORM - u'\ufe8b' # 0x00c6 -> ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM - u'\ufe8d' # 0x00c7 -> ARABIC LETTER ALEF ISOLATED FORM - u'\ufe91' # 0x00c8 -> ARABIC LETTER BEH INITIAL FORM - u'\ufe93' # 0x00c9 -> ARABIC LETTER TEH MARBUTA ISOLATED FORM - u'\ufe97' # 0x00ca -> ARABIC LETTER TEH INITIAL FORM - u'\ufe9b' # 0x00cb -> ARABIC LETTER THEH INITIAL FORM - u'\ufe9f' # 0x00cc -> ARABIC LETTER JEEM INITIAL FORM - u'\ufea3' # 0x00cd -> ARABIC LETTER HAH INITIAL FORM - u'\ufea7' # 0x00ce -> ARABIC LETTER KHAH INITIAL FORM - u'\ufea9' # 0x00cf -> ARABIC LETTER DAL ISOLATED FORM - u'\ufeab' # 0x00d0 -> ARABIC LETTER THAL ISOLATED FORM - u'\ufead' # 0x00d1 -> ARABIC LETTER REH ISOLATED FORM - u'\ufeaf' # 0x00d2 -> ARABIC LETTER ZAIN ISOLATED FORM - u'\ufeb3' # 0x00d3 -> ARABIC LETTER SEEN INITIAL FORM - u'\ufeb7' # 0x00d4 -> ARABIC LETTER SHEEN INITIAL FORM - u'\ufebb' # 0x00d5 -> ARABIC LETTER SAD INITIAL FORM - u'\ufebf' # 0x00d6 -> ARABIC LETTER DAD INITIAL FORM - u'\ufec1' # 0x00d7 -> ARABIC LETTER TAH ISOLATED FORM - u'\ufec5' # 0x00d8 -> ARABIC LETTER ZAH ISOLATED FORM - u'\ufecb' # 0x00d9 -> ARABIC LETTER AIN INITIAL FORM - u'\ufecf' # 0x00da -> ARABIC LETTER GHAIN INITIAL FORM - u'\xa6' # 0x00db -> BROKEN VERTICAL BAR - u'\xac' # 0x00dc -> NOT SIGN - u'\xf7' # 0x00dd -> DIVISION SIGN - u'\xd7' # 0x00de -> MULTIPLICATION SIGN - u'\ufec9' # 0x00df -> ARABIC LETTER AIN ISOLATED FORM - u'\u0640' # 0x00e0 -> ARABIC TATWEEL - u'\ufed3' # 0x00e1 -> ARABIC LETTER FEH INITIAL FORM - u'\ufed7' # 0x00e2 -> ARABIC LETTER QAF INITIAL FORM - u'\ufedb' # 0x00e3 -> ARABIC LETTER KAF INITIAL FORM - u'\ufedf' # 0x00e4 -> ARABIC LETTER LAM INITIAL FORM - u'\ufee3' # 0x00e5 -> ARABIC LETTER MEEM INITIAL FORM - u'\ufee7' # 0x00e6 -> ARABIC LETTER NOON INITIAL FORM - u'\ufeeb' # 0x00e7 -> ARABIC LETTER HEH INITIAL FORM - u'\ufeed' # 0x00e8 -> ARABIC LETTER WAW ISOLATED FORM - u'\ufeef' # 0x00e9 -> ARABIC LETTER ALEF MAKSURA ISOLATED FORM - u'\ufef3' # 0x00ea -> ARABIC LETTER YEH INITIAL FORM - u'\ufebd' # 0x00eb -> ARABIC LETTER DAD ISOLATED FORM - u'\ufecc' # 0x00ec -> ARABIC LETTER AIN MEDIAL FORM - u'\ufece' # 0x00ed -> ARABIC LETTER GHAIN FINAL FORM - u'\ufecd' # 0x00ee -> ARABIC LETTER GHAIN ISOLATED FORM - u'\ufee1' # 0x00ef -> ARABIC LETTER MEEM ISOLATED FORM - u'\ufe7d' # 0x00f0 -> ARABIC SHADDA MEDIAL FORM - u'\u0651' # 0x00f1 -> ARABIC SHADDAH - u'\ufee5' # 0x00f2 -> ARABIC LETTER NOON ISOLATED FORM - u'\ufee9' # 0x00f3 -> ARABIC LETTER HEH ISOLATED FORM - u'\ufeec' # 0x00f4 -> ARABIC LETTER HEH MEDIAL FORM - u'\ufef0' # 0x00f5 -> ARABIC LETTER ALEF MAKSURA FINAL FORM - u'\ufef2' # 0x00f6 -> ARABIC LETTER YEH FINAL FORM - u'\ufed0' # 0x00f7 -> ARABIC LETTER GHAIN MEDIAL FORM - u'\ufed5' # 0x00f8 -> ARABIC LETTER QAF ISOLATED FORM - u'\ufef5' # 0x00f9 -> ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE ISOLATED FORM - u'\ufef6' # 0x00fa -> ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE FINAL FORM - u'\ufedd' # 0x00fb -> ARABIC LETTER LAM ISOLATED FORM - u'\ufed9' # 0x00fc -> ARABIC LETTER KAF ISOLATED FORM - u'\ufef1' # 0x00fd -> ARABIC LETTER YEH ISOLATED FORM - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\ufffe' # 0x00ff -> UNDEFINED + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '\u066a' # 0x0025 -> ARABIC PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xb0' # 0x0080 -> DEGREE SIGN + '\xb7' # 0x0081 -> MIDDLE DOT + '\u2219' # 0x0082 -> BULLET OPERATOR + '\u221a' # 0x0083 -> SQUARE ROOT + '\u2592' # 0x0084 -> MEDIUM SHADE + '\u2500' # 0x0085 -> FORMS LIGHT HORIZONTAL + '\u2502' # 0x0086 -> FORMS LIGHT VERTICAL + '\u253c' # 0x0087 -> FORMS LIGHT VERTICAL AND HORIZONTAL + '\u2524' # 0x0088 -> FORMS LIGHT VERTICAL AND LEFT + '\u252c' # 0x0089 -> FORMS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x008a -> FORMS LIGHT VERTICAL AND RIGHT + '\u2534' # 0x008b -> FORMS LIGHT UP AND HORIZONTAL + '\u2510' # 0x008c -> FORMS LIGHT DOWN AND LEFT + '\u250c' # 0x008d -> FORMS LIGHT DOWN AND RIGHT + '\u2514' # 0x008e -> FORMS LIGHT UP AND RIGHT + '\u2518' # 0x008f -> FORMS LIGHT UP AND LEFT + '\u03b2' # 0x0090 -> GREEK SMALL BETA + '\u221e' # 0x0091 -> INFINITY + '\u03c6' # 0x0092 -> GREEK SMALL PHI + '\xb1' # 0x0093 -> PLUS-OR-MINUS SIGN + '\xbd' # 0x0094 -> FRACTION 1/2 + '\xbc' # 0x0095 -> FRACTION 1/4 + '\u2248' # 0x0096 -> ALMOST EQUAL TO + '\xab' # 0x0097 -> LEFT POINTING GUILLEMET + '\xbb' # 0x0098 -> RIGHT POINTING GUILLEMET + '\ufef7' # 0x0099 -> ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE ISOLATED FORM + '\ufef8' # 0x009a -> ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE FINAL FORM + '\ufffe' # 0x009b -> UNDEFINED + '\ufffe' # 0x009c -> UNDEFINED + '\ufefb' # 0x009d -> ARABIC LIGATURE LAM WITH ALEF ISOLATED FORM + '\ufefc' # 0x009e -> ARABIC LIGATURE LAM WITH ALEF FINAL FORM + '\ufffe' # 0x009f -> UNDEFINED + '\xa0' # 0x00a0 -> NON-BREAKING SPACE + '\xad' # 0x00a1 -> SOFT HYPHEN + '\ufe82' # 0x00a2 -> ARABIC LETTER ALEF WITH MADDA ABOVE FINAL FORM + '\xa3' # 0x00a3 -> POUND SIGN + '\xa4' # 0x00a4 -> CURRENCY SIGN + '\ufe84' # 0x00a5 -> ARABIC LETTER ALEF WITH HAMZA ABOVE FINAL FORM + '\ufffe' # 0x00a6 -> UNDEFINED + '\ufffe' # 0x00a7 -> UNDEFINED + '\ufe8e' # 0x00a8 -> ARABIC LETTER ALEF FINAL FORM + '\ufe8f' # 0x00a9 -> ARABIC LETTER BEH ISOLATED FORM + '\ufe95' # 0x00aa -> ARABIC LETTER TEH ISOLATED FORM + '\ufe99' # 0x00ab -> ARABIC LETTER THEH ISOLATED FORM + '\u060c' # 0x00ac -> ARABIC COMMA + '\ufe9d' # 0x00ad -> ARABIC LETTER JEEM ISOLATED FORM + '\ufea1' # 0x00ae -> ARABIC LETTER HAH ISOLATED FORM + '\ufea5' # 0x00af -> ARABIC LETTER KHAH ISOLATED FORM + '\u0660' # 0x00b0 -> ARABIC-INDIC DIGIT ZERO + '\u0661' # 0x00b1 -> ARABIC-INDIC DIGIT ONE + '\u0662' # 0x00b2 -> ARABIC-INDIC DIGIT TWO + '\u0663' # 0x00b3 -> ARABIC-INDIC DIGIT THREE + '\u0664' # 0x00b4 -> ARABIC-INDIC DIGIT FOUR + '\u0665' # 0x00b5 -> ARABIC-INDIC DIGIT FIVE + '\u0666' # 0x00b6 -> ARABIC-INDIC DIGIT SIX + '\u0667' # 0x00b7 -> ARABIC-INDIC DIGIT SEVEN + '\u0668' # 0x00b8 -> ARABIC-INDIC DIGIT EIGHT + '\u0669' # 0x00b9 -> ARABIC-INDIC DIGIT NINE + '\ufed1' # 0x00ba -> ARABIC LETTER FEH ISOLATED FORM + '\u061b' # 0x00bb -> ARABIC SEMICOLON + '\ufeb1' # 0x00bc -> ARABIC LETTER SEEN ISOLATED FORM + '\ufeb5' # 0x00bd -> ARABIC LETTER SHEEN ISOLATED FORM + '\ufeb9' # 0x00be -> ARABIC LETTER SAD ISOLATED FORM + '\u061f' # 0x00bf -> ARABIC QUESTION MARK + '\xa2' # 0x00c0 -> CENT SIGN + '\ufe80' # 0x00c1 -> ARABIC LETTER HAMZA ISOLATED FORM + '\ufe81' # 0x00c2 -> ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM + '\ufe83' # 0x00c3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE ISOLATED FORM + '\ufe85' # 0x00c4 -> ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM + '\ufeca' # 0x00c5 -> ARABIC LETTER AIN FINAL FORM + '\ufe8b' # 0x00c6 -> ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM + '\ufe8d' # 0x00c7 -> ARABIC LETTER ALEF ISOLATED FORM + '\ufe91' # 0x00c8 -> ARABIC LETTER BEH INITIAL FORM + '\ufe93' # 0x00c9 -> ARABIC LETTER TEH MARBUTA ISOLATED FORM + '\ufe97' # 0x00ca -> ARABIC LETTER TEH INITIAL FORM + '\ufe9b' # 0x00cb -> ARABIC LETTER THEH INITIAL FORM + '\ufe9f' # 0x00cc -> ARABIC LETTER JEEM INITIAL FORM + '\ufea3' # 0x00cd -> ARABIC LETTER HAH INITIAL FORM + '\ufea7' # 0x00ce -> ARABIC LETTER KHAH INITIAL FORM + '\ufea9' # 0x00cf -> ARABIC LETTER DAL ISOLATED FORM + '\ufeab' # 0x00d0 -> ARABIC LETTER THAL ISOLATED FORM + '\ufead' # 0x00d1 -> ARABIC LETTER REH ISOLATED FORM + '\ufeaf' # 0x00d2 -> ARABIC LETTER ZAIN ISOLATED FORM + '\ufeb3' # 0x00d3 -> ARABIC LETTER SEEN INITIAL FORM + '\ufeb7' # 0x00d4 -> ARABIC LETTER SHEEN INITIAL FORM + '\ufebb' # 0x00d5 -> ARABIC LETTER SAD INITIAL FORM + '\ufebf' # 0x00d6 -> ARABIC LETTER DAD INITIAL FORM + '\ufec1' # 0x00d7 -> ARABIC LETTER TAH ISOLATED FORM + '\ufec5' # 0x00d8 -> ARABIC LETTER ZAH ISOLATED FORM + '\ufecb' # 0x00d9 -> ARABIC LETTER AIN INITIAL FORM + '\ufecf' # 0x00da -> ARABIC LETTER GHAIN INITIAL FORM + '\xa6' # 0x00db -> BROKEN VERTICAL BAR + '\xac' # 0x00dc -> NOT SIGN + '\xf7' # 0x00dd -> DIVISION SIGN + '\xd7' # 0x00de -> MULTIPLICATION SIGN + '\ufec9' # 0x00df -> ARABIC LETTER AIN ISOLATED FORM + '\u0640' # 0x00e0 -> ARABIC TATWEEL + '\ufed3' # 0x00e1 -> ARABIC LETTER FEH INITIAL FORM + '\ufed7' # 0x00e2 -> ARABIC LETTER QAF INITIAL FORM + '\ufedb' # 0x00e3 -> ARABIC LETTER KAF INITIAL FORM + '\ufedf' # 0x00e4 -> ARABIC LETTER LAM INITIAL FORM + '\ufee3' # 0x00e5 -> ARABIC LETTER MEEM INITIAL FORM + '\ufee7' # 0x00e6 -> ARABIC LETTER NOON INITIAL FORM + '\ufeeb' # 0x00e7 -> ARABIC LETTER HEH INITIAL FORM + '\ufeed' # 0x00e8 -> ARABIC LETTER WAW ISOLATED FORM + '\ufeef' # 0x00e9 -> ARABIC LETTER ALEF MAKSURA ISOLATED FORM + '\ufef3' # 0x00ea -> ARABIC LETTER YEH INITIAL FORM + '\ufebd' # 0x00eb -> ARABIC LETTER DAD ISOLATED FORM + '\ufecc' # 0x00ec -> ARABIC LETTER AIN MEDIAL FORM + '\ufece' # 0x00ed -> ARABIC LETTER GHAIN FINAL FORM + '\ufecd' # 0x00ee -> ARABIC LETTER GHAIN ISOLATED FORM + '\ufee1' # 0x00ef -> ARABIC LETTER MEEM ISOLATED FORM + '\ufe7d' # 0x00f0 -> ARABIC SHADDA MEDIAL FORM + '\u0651' # 0x00f1 -> ARABIC SHADDAH + '\ufee5' # 0x00f2 -> ARABIC LETTER NOON ISOLATED FORM + '\ufee9' # 0x00f3 -> ARABIC LETTER HEH ISOLATED FORM + '\ufeec' # 0x00f4 -> ARABIC LETTER HEH MEDIAL FORM + '\ufef0' # 0x00f5 -> ARABIC LETTER ALEF MAKSURA FINAL FORM + '\ufef2' # 0x00f6 -> ARABIC LETTER YEH FINAL FORM + '\ufed0' # 0x00f7 -> ARABIC LETTER GHAIN MEDIAL FORM + '\ufed5' # 0x00f8 -> ARABIC LETTER QAF ISOLATED FORM + '\ufef5' # 0x00f9 -> ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE ISOLATED FORM + '\ufef6' # 0x00fa -> ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE FINAL FORM + '\ufedd' # 0x00fb -> ARABIC LETTER LAM ISOLATED FORM + '\ufed9' # 0x00fc -> ARABIC LETTER KAF ISOLATED FORM + '\ufef1' # 0x00fd -> ARABIC LETTER YEH ISOLATED FORM + '\u25a0' # 0x00fe -> BLACK SQUARE + '\ufffe' # 0x00ff -> UNDEFINED ) ### Encoding Map diff --git a/Lib/encodings/cp865.py b/Lib/encodings/cp865.py index e9f45f1..6726cf3 100644 --- a/Lib/encodings/cp865.py +++ b/Lib/encodings/cp865.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE - u'\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE - u'\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE - u'\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE - u'\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE - u'\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE - u'\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE - u'\xff' # 0x0098 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE - u'\xa3' # 0x009c -> POUND SIGN - u'\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE - u'\u20a7' # 0x009e -> PESETA SIGN - u'\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK - u'\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE - u'\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE - u'\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE - u'\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE - u'\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE - u'\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR - u'\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR - u'\xbf' # 0x00a8 -> INVERTED QUESTION MARK - u'\u2310' # 0x00a9 -> REVERSED NOT SIGN - u'\xac' # 0x00aa -> NOT SIGN - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER - u'\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xa4' # 0x00af -> CURRENCY SIGN - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA - u'\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S - u'\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA - u'\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI - u'\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA - u'\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA - u'\xb5' # 0x00e6 -> MICRO SIGN - u'\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU - u'\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI - u'\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA - u'\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA - u'\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA - u'\u221e' # 0x00ec -> INFINITY - u'\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI - u'\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON - u'\u2229' # 0x00ef -> INTERSECTION - u'\u2261' # 0x00f0 -> IDENTICAL TO - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO - u'\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO - u'\u2320' # 0x00f4 -> TOP HALF INTEGRAL - u'\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL - u'\xf7' # 0x00f6 -> DIVISION SIGN - u'\u2248' # 0x00f7 -> ALMOST EQUAL TO - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N - u'\xb2' # 0x00fd -> SUPERSCRIPT TWO - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS + '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE + '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe0' # 0x0085 -> LATIN SMALL LETTER A WITH GRAVE + '\xe5' # 0x0086 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA + '\xea' # 0x0088 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xe8' # 0x008a -> LATIN SMALL LETTER E WITH GRAVE + '\xef' # 0x008b -> LATIN SMALL LETTER I WITH DIAERESIS + '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xec' # 0x008d -> LATIN SMALL LETTER I WITH GRAVE + '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x008f -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xe6' # 0x0091 -> LATIN SMALL LIGATURE AE + '\xc6' # 0x0092 -> LATIN CAPITAL LIGATURE AE + '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf2' # 0x0095 -> LATIN SMALL LETTER O WITH GRAVE + '\xfb' # 0x0096 -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xf9' # 0x0097 -> LATIN SMALL LETTER U WITH GRAVE + '\xff' # 0x0098 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xf8' # 0x009b -> LATIN SMALL LETTER O WITH STROKE + '\xa3' # 0x009c -> POUND SIGN + '\xd8' # 0x009d -> LATIN CAPITAL LETTER O WITH STROKE + '\u20a7' # 0x009e -> PESETA SIGN + '\u0192' # 0x009f -> LATIN SMALL LETTER F WITH HOOK + '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE + '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE + '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE + '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE + '\xf1' # 0x00a4 -> LATIN SMALL LETTER N WITH TILDE + '\xd1' # 0x00a5 -> LATIN CAPITAL LETTER N WITH TILDE + '\xaa' # 0x00a6 -> FEMININE ORDINAL INDICATOR + '\xba' # 0x00a7 -> MASCULINE ORDINAL INDICATOR + '\xbf' # 0x00a8 -> INVERTED QUESTION MARK + '\u2310' # 0x00a9 -> REVERSED NOT SIGN + '\xac' # 0x00aa -> NOT SIGN + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\xbc' # 0x00ac -> VULGAR FRACTION ONE QUARTER + '\xa1' # 0x00ad -> INVERTED EXCLAMATION MARK + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xa4' # 0x00af -> CURRENCY SIGN + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b1' # 0x00e0 -> GREEK SMALL LETTER ALPHA + '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S + '\u0393' # 0x00e2 -> GREEK CAPITAL LETTER GAMMA + '\u03c0' # 0x00e3 -> GREEK SMALL LETTER PI + '\u03a3' # 0x00e4 -> GREEK CAPITAL LETTER SIGMA + '\u03c3' # 0x00e5 -> GREEK SMALL LETTER SIGMA + '\xb5' # 0x00e6 -> MICRO SIGN + '\u03c4' # 0x00e7 -> GREEK SMALL LETTER TAU + '\u03a6' # 0x00e8 -> GREEK CAPITAL LETTER PHI + '\u0398' # 0x00e9 -> GREEK CAPITAL LETTER THETA + '\u03a9' # 0x00ea -> GREEK CAPITAL LETTER OMEGA + '\u03b4' # 0x00eb -> GREEK SMALL LETTER DELTA + '\u221e' # 0x00ec -> INFINITY + '\u03c6' # 0x00ed -> GREEK SMALL LETTER PHI + '\u03b5' # 0x00ee -> GREEK SMALL LETTER EPSILON + '\u2229' # 0x00ef -> INTERSECTION + '\u2261' # 0x00f0 -> IDENTICAL TO + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u2265' # 0x00f2 -> GREATER-THAN OR EQUAL TO + '\u2264' # 0x00f3 -> LESS-THAN OR EQUAL TO + '\u2320' # 0x00f4 -> TOP HALF INTEGRAL + '\u2321' # 0x00f5 -> BOTTOM HALF INTEGRAL + '\xf7' # 0x00f6 -> DIVISION SIGN + '\u2248' # 0x00f7 -> ALMOST EQUAL TO + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u207f' # 0x00fc -> SUPERSCRIPT LATIN SMALL LETTER N + '\xb2' # 0x00fd -> SUPERSCRIPT TWO + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp866.py b/Lib/encodings/cp866.py index 29cd85a..bec7ae3 100644 --- a/Lib/encodings/cp866.py +++ b/Lib/encodings/cp866.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\u0410' # 0x0080 -> CYRILLIC CAPITAL LETTER A - u'\u0411' # 0x0081 -> CYRILLIC CAPITAL LETTER BE - u'\u0412' # 0x0082 -> CYRILLIC CAPITAL LETTER VE - u'\u0413' # 0x0083 -> CYRILLIC CAPITAL LETTER GHE - u'\u0414' # 0x0084 -> CYRILLIC CAPITAL LETTER DE - u'\u0415' # 0x0085 -> CYRILLIC CAPITAL LETTER IE - u'\u0416' # 0x0086 -> CYRILLIC CAPITAL LETTER ZHE - u'\u0417' # 0x0087 -> CYRILLIC CAPITAL LETTER ZE - u'\u0418' # 0x0088 -> CYRILLIC CAPITAL LETTER I - u'\u0419' # 0x0089 -> CYRILLIC CAPITAL LETTER SHORT I - u'\u041a' # 0x008a -> CYRILLIC CAPITAL LETTER KA - u'\u041b' # 0x008b -> CYRILLIC CAPITAL LETTER EL - u'\u041c' # 0x008c -> CYRILLIC CAPITAL LETTER EM - u'\u041d' # 0x008d -> CYRILLIC CAPITAL LETTER EN - u'\u041e' # 0x008e -> CYRILLIC CAPITAL LETTER O - u'\u041f' # 0x008f -> CYRILLIC CAPITAL LETTER PE - u'\u0420' # 0x0090 -> CYRILLIC CAPITAL LETTER ER - u'\u0421' # 0x0091 -> CYRILLIC CAPITAL LETTER ES - u'\u0422' # 0x0092 -> CYRILLIC CAPITAL LETTER TE - u'\u0423' # 0x0093 -> CYRILLIC CAPITAL LETTER U - u'\u0424' # 0x0094 -> CYRILLIC CAPITAL LETTER EF - u'\u0425' # 0x0095 -> CYRILLIC CAPITAL LETTER HA - u'\u0426' # 0x0096 -> CYRILLIC CAPITAL LETTER TSE - u'\u0427' # 0x0097 -> CYRILLIC CAPITAL LETTER CHE - u'\u0428' # 0x0098 -> CYRILLIC CAPITAL LETTER SHA - u'\u0429' # 0x0099 -> CYRILLIC CAPITAL LETTER SHCHA - u'\u042a' # 0x009a -> CYRILLIC CAPITAL LETTER HARD SIGN - u'\u042b' # 0x009b -> CYRILLIC CAPITAL LETTER YERU - u'\u042c' # 0x009c -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u042d' # 0x009d -> CYRILLIC CAPITAL LETTER E - u'\u042e' # 0x009e -> CYRILLIC CAPITAL LETTER YU - u'\u042f' # 0x009f -> CYRILLIC CAPITAL LETTER YA - u'\u0430' # 0x00a0 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0x00a1 -> CYRILLIC SMALL LETTER BE - u'\u0432' # 0x00a2 -> CYRILLIC SMALL LETTER VE - u'\u0433' # 0x00a3 -> CYRILLIC SMALL LETTER GHE - u'\u0434' # 0x00a4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0x00a5 -> CYRILLIC SMALL LETTER IE - u'\u0436' # 0x00a6 -> CYRILLIC SMALL LETTER ZHE - u'\u0437' # 0x00a7 -> CYRILLIC SMALL LETTER ZE - u'\u0438' # 0x00a8 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0x00a9 -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0x00aa -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0x00ab -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0x00ac -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0x00ad -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0x00ae -> CYRILLIC SMALL LETTER O - u'\u043f' # 0x00af -> CYRILLIC SMALL LETTER PE - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u258c' # 0x00dd -> LEFT HALF BLOCK - u'\u2590' # 0x00de -> RIGHT HALF BLOCK - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u0440' # 0x00e0 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0x00e1 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0x00e2 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0x00e3 -> CYRILLIC SMALL LETTER U - u'\u0444' # 0x00e4 -> CYRILLIC SMALL LETTER EF - u'\u0445' # 0x00e5 -> CYRILLIC SMALL LETTER HA - u'\u0446' # 0x00e6 -> CYRILLIC SMALL LETTER TSE - u'\u0447' # 0x00e7 -> CYRILLIC SMALL LETTER CHE - u'\u0448' # 0x00e8 -> CYRILLIC SMALL LETTER SHA - u'\u0449' # 0x00e9 -> CYRILLIC SMALL LETTER SHCHA - u'\u044a' # 0x00ea -> CYRILLIC SMALL LETTER HARD SIGN - u'\u044b' # 0x00eb -> CYRILLIC SMALL LETTER YERU - u'\u044c' # 0x00ec -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044d' # 0x00ed -> CYRILLIC SMALL LETTER E - u'\u044e' # 0x00ee -> CYRILLIC SMALL LETTER YU - u'\u044f' # 0x00ef -> CYRILLIC SMALL LETTER YA - u'\u0401' # 0x00f0 -> CYRILLIC CAPITAL LETTER IO - u'\u0451' # 0x00f1 -> CYRILLIC SMALL LETTER IO - u'\u0404' # 0x00f2 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\u0454' # 0x00f3 -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\u0407' # 0x00f4 -> CYRILLIC CAPITAL LETTER YI - u'\u0457' # 0x00f5 -> CYRILLIC SMALL LETTER YI - u'\u040e' # 0x00f6 -> CYRILLIC CAPITAL LETTER SHORT U - u'\u045e' # 0x00f7 -> CYRILLIC SMALL LETTER SHORT U - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\u2219' # 0x00f9 -> BULLET OPERATOR - u'\xb7' # 0x00fa -> MIDDLE DOT - u'\u221a' # 0x00fb -> SQUARE ROOT - u'\u2116' # 0x00fc -> NUMERO SIGN - u'\xa4' # 0x00fd -> CURRENCY SIGN - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\u0410' # 0x0080 -> CYRILLIC CAPITAL LETTER A + '\u0411' # 0x0081 -> CYRILLIC CAPITAL LETTER BE + '\u0412' # 0x0082 -> CYRILLIC CAPITAL LETTER VE + '\u0413' # 0x0083 -> CYRILLIC CAPITAL LETTER GHE + '\u0414' # 0x0084 -> CYRILLIC CAPITAL LETTER DE + '\u0415' # 0x0085 -> CYRILLIC CAPITAL LETTER IE + '\u0416' # 0x0086 -> CYRILLIC CAPITAL LETTER ZHE + '\u0417' # 0x0087 -> CYRILLIC CAPITAL LETTER ZE + '\u0418' # 0x0088 -> CYRILLIC CAPITAL LETTER I + '\u0419' # 0x0089 -> CYRILLIC CAPITAL LETTER SHORT I + '\u041a' # 0x008a -> CYRILLIC CAPITAL LETTER KA + '\u041b' # 0x008b -> CYRILLIC CAPITAL LETTER EL + '\u041c' # 0x008c -> CYRILLIC CAPITAL LETTER EM + '\u041d' # 0x008d -> CYRILLIC CAPITAL LETTER EN + '\u041e' # 0x008e -> CYRILLIC CAPITAL LETTER O + '\u041f' # 0x008f -> CYRILLIC CAPITAL LETTER PE + '\u0420' # 0x0090 -> CYRILLIC CAPITAL LETTER ER + '\u0421' # 0x0091 -> CYRILLIC CAPITAL LETTER ES + '\u0422' # 0x0092 -> CYRILLIC CAPITAL LETTER TE + '\u0423' # 0x0093 -> CYRILLIC CAPITAL LETTER U + '\u0424' # 0x0094 -> CYRILLIC CAPITAL LETTER EF + '\u0425' # 0x0095 -> CYRILLIC CAPITAL LETTER HA + '\u0426' # 0x0096 -> CYRILLIC CAPITAL LETTER TSE + '\u0427' # 0x0097 -> CYRILLIC CAPITAL LETTER CHE + '\u0428' # 0x0098 -> CYRILLIC CAPITAL LETTER SHA + '\u0429' # 0x0099 -> CYRILLIC CAPITAL LETTER SHCHA + '\u042a' # 0x009a -> CYRILLIC CAPITAL LETTER HARD SIGN + '\u042b' # 0x009b -> CYRILLIC CAPITAL LETTER YERU + '\u042c' # 0x009c -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u042d' # 0x009d -> CYRILLIC CAPITAL LETTER E + '\u042e' # 0x009e -> CYRILLIC CAPITAL LETTER YU + '\u042f' # 0x009f -> CYRILLIC CAPITAL LETTER YA + '\u0430' # 0x00a0 -> CYRILLIC SMALL LETTER A + '\u0431' # 0x00a1 -> CYRILLIC SMALL LETTER BE + '\u0432' # 0x00a2 -> CYRILLIC SMALL LETTER VE + '\u0433' # 0x00a3 -> CYRILLIC SMALL LETTER GHE + '\u0434' # 0x00a4 -> CYRILLIC SMALL LETTER DE + '\u0435' # 0x00a5 -> CYRILLIC SMALL LETTER IE + '\u0436' # 0x00a6 -> CYRILLIC SMALL LETTER ZHE + '\u0437' # 0x00a7 -> CYRILLIC SMALL LETTER ZE + '\u0438' # 0x00a8 -> CYRILLIC SMALL LETTER I + '\u0439' # 0x00a9 -> CYRILLIC SMALL LETTER SHORT I + '\u043a' # 0x00aa -> CYRILLIC SMALL LETTER KA + '\u043b' # 0x00ab -> CYRILLIC SMALL LETTER EL + '\u043c' # 0x00ac -> CYRILLIC SMALL LETTER EM + '\u043d' # 0x00ad -> CYRILLIC SMALL LETTER EN + '\u043e' # 0x00ae -> CYRILLIC SMALL LETTER O + '\u043f' # 0x00af -> CYRILLIC SMALL LETTER PE + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u2561' # 0x00b5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u2562' # 0x00b6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2556' # 0x00b7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2555' # 0x00b8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255c' # 0x00bd -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255b' # 0x00be -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u255e' # 0x00c6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0x00c7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u2567' # 0x00cf -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0x00d0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2564' # 0x00d1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0x00d2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2559' # 0x00d3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u2558' # 0x00d4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2552' # 0x00d5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u2553' # 0x00d6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u256b' # 0x00d7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256a' # 0x00d8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u258c' # 0x00dd -> LEFT HALF BLOCK + '\u2590' # 0x00de -> RIGHT HALF BLOCK + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u0440' # 0x00e0 -> CYRILLIC SMALL LETTER ER + '\u0441' # 0x00e1 -> CYRILLIC SMALL LETTER ES + '\u0442' # 0x00e2 -> CYRILLIC SMALL LETTER TE + '\u0443' # 0x00e3 -> CYRILLIC SMALL LETTER U + '\u0444' # 0x00e4 -> CYRILLIC SMALL LETTER EF + '\u0445' # 0x00e5 -> CYRILLIC SMALL LETTER HA + '\u0446' # 0x00e6 -> CYRILLIC SMALL LETTER TSE + '\u0447' # 0x00e7 -> CYRILLIC SMALL LETTER CHE + '\u0448' # 0x00e8 -> CYRILLIC SMALL LETTER SHA + '\u0449' # 0x00e9 -> CYRILLIC SMALL LETTER SHCHA + '\u044a' # 0x00ea -> CYRILLIC SMALL LETTER HARD SIGN + '\u044b' # 0x00eb -> CYRILLIC SMALL LETTER YERU + '\u044c' # 0x00ec -> CYRILLIC SMALL LETTER SOFT SIGN + '\u044d' # 0x00ed -> CYRILLIC SMALL LETTER E + '\u044e' # 0x00ee -> CYRILLIC SMALL LETTER YU + '\u044f' # 0x00ef -> CYRILLIC SMALL LETTER YA + '\u0401' # 0x00f0 -> CYRILLIC CAPITAL LETTER IO + '\u0451' # 0x00f1 -> CYRILLIC SMALL LETTER IO + '\u0404' # 0x00f2 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE + '\u0454' # 0x00f3 -> CYRILLIC SMALL LETTER UKRAINIAN IE + '\u0407' # 0x00f4 -> CYRILLIC CAPITAL LETTER YI + '\u0457' # 0x00f5 -> CYRILLIC SMALL LETTER YI + '\u040e' # 0x00f6 -> CYRILLIC CAPITAL LETTER SHORT U + '\u045e' # 0x00f7 -> CYRILLIC SMALL LETTER SHORT U + '\xb0' # 0x00f8 -> DEGREE SIGN + '\u2219' # 0x00f9 -> BULLET OPERATOR + '\xb7' # 0x00fa -> MIDDLE DOT + '\u221a' # 0x00fb -> SQUARE ROOT + '\u2116' # 0x00fc -> NUMERO SIGN + '\xa4' # 0x00fd -> CURRENCY SIGN + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp869.py b/Lib/encodings/cp869.py index b4dc99b..8d8a29b 100644 --- a/Lib/encodings/cp869.py +++ b/Lib/encodings/cp869.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> NULL - u'\x01' # 0x0001 -> START OF HEADING - u'\x02' # 0x0002 -> START OF TEXT - u'\x03' # 0x0003 -> END OF TEXT - u'\x04' # 0x0004 -> END OF TRANSMISSION - u'\x05' # 0x0005 -> ENQUIRY - u'\x06' # 0x0006 -> ACKNOWLEDGE - u'\x07' # 0x0007 -> BELL - u'\x08' # 0x0008 -> BACKSPACE - u'\t' # 0x0009 -> HORIZONTAL TABULATION - u'\n' # 0x000a -> LINE FEED - u'\x0b' # 0x000b -> VERTICAL TABULATION - u'\x0c' # 0x000c -> FORM FEED - u'\r' # 0x000d -> CARRIAGE RETURN - u'\x0e' # 0x000e -> SHIFT OUT - u'\x0f' # 0x000f -> SHIFT IN - u'\x10' # 0x0010 -> DATA LINK ESCAPE - u'\x11' # 0x0011 -> DEVICE CONTROL ONE - u'\x12' # 0x0012 -> DEVICE CONTROL TWO - u'\x13' # 0x0013 -> DEVICE CONTROL THREE - u'\x14' # 0x0014 -> DEVICE CONTROL FOUR - u'\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x0016 -> SYNCHRONOUS IDLE - u'\x17' # 0x0017 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x0018 -> CANCEL - u'\x19' # 0x0019 -> END OF MEDIUM - u'\x1a' # 0x001a -> SUBSTITUTE - u'\x1b' # 0x001b -> ESCAPE - u'\x1c' # 0x001c -> FILE SEPARATOR - u'\x1d' # 0x001d -> GROUP SEPARATOR - u'\x1e' # 0x001e -> RECORD SEPARATOR - u'\x1f' # 0x001f -> UNIT SEPARATOR - u' ' # 0x0020 -> SPACE - u'!' # 0x0021 -> EXCLAMATION MARK - u'"' # 0x0022 -> QUOTATION MARK - u'#' # 0x0023 -> NUMBER SIGN - u'$' # 0x0024 -> DOLLAR SIGN - u'%' # 0x0025 -> PERCENT SIGN - u'&' # 0x0026 -> AMPERSAND - u"'" # 0x0027 -> APOSTROPHE - u'(' # 0x0028 -> LEFT PARENTHESIS - u')' # 0x0029 -> RIGHT PARENTHESIS - u'*' # 0x002a -> ASTERISK - u'+' # 0x002b -> PLUS SIGN - u',' # 0x002c -> COMMA - u'-' # 0x002d -> HYPHEN-MINUS - u'.' # 0x002e -> FULL STOP - u'/' # 0x002f -> SOLIDUS - u'0' # 0x0030 -> DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE - u':' # 0x003a -> COLON - u';' # 0x003b -> SEMICOLON - u'<' # 0x003c -> LESS-THAN SIGN - u'=' # 0x003d -> EQUALS SIGN - u'>' # 0x003e -> GREATER-THAN SIGN - u'?' # 0x003f -> QUESTION MARK - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET - u'\\' # 0x005c -> REVERSE SOLIDUS - u']' # 0x005d -> RIGHT SQUARE BRACKET - u'^' # 0x005e -> CIRCUMFLEX ACCENT - u'_' # 0x005f -> LOW LINE - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET - u'|' # 0x007c -> VERTICAL LINE - u'}' # 0x007d -> RIGHT CURLY BRACKET - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> DELETE - u'\ufffe' # 0x0080 -> UNDEFINED - u'\ufffe' # 0x0081 -> UNDEFINED - u'\ufffe' # 0x0082 -> UNDEFINED - u'\ufffe' # 0x0083 -> UNDEFINED - u'\ufffe' # 0x0084 -> UNDEFINED - u'\ufffe' # 0x0085 -> UNDEFINED - u'\u0386' # 0x0086 -> GREEK CAPITAL LETTER ALPHA WITH TONOS - u'\ufffe' # 0x0087 -> UNDEFINED - u'\xb7' # 0x0088 -> MIDDLE DOT - u'\xac' # 0x0089 -> NOT SIGN - u'\xa6' # 0x008a -> BROKEN BAR - u'\u2018' # 0x008b -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x008c -> RIGHT SINGLE QUOTATION MARK - u'\u0388' # 0x008d -> GREEK CAPITAL LETTER EPSILON WITH TONOS - u'\u2015' # 0x008e -> HORIZONTAL BAR - u'\u0389' # 0x008f -> GREEK CAPITAL LETTER ETA WITH TONOS - u'\u038a' # 0x0090 -> GREEK CAPITAL LETTER IOTA WITH TONOS - u'\u03aa' # 0x0091 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - u'\u038c' # 0x0092 -> GREEK CAPITAL LETTER OMICRON WITH TONOS - u'\ufffe' # 0x0093 -> UNDEFINED - u'\ufffe' # 0x0094 -> UNDEFINED - u'\u038e' # 0x0095 -> GREEK CAPITAL LETTER UPSILON WITH TONOS - u'\u03ab' # 0x0096 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - u'\xa9' # 0x0097 -> COPYRIGHT SIGN - u'\u038f' # 0x0098 -> GREEK CAPITAL LETTER OMEGA WITH TONOS - u'\xb2' # 0x0099 -> SUPERSCRIPT TWO - u'\xb3' # 0x009a -> SUPERSCRIPT THREE - u'\u03ac' # 0x009b -> GREEK SMALL LETTER ALPHA WITH TONOS - u'\xa3' # 0x009c -> POUND SIGN - u'\u03ad' # 0x009d -> GREEK SMALL LETTER EPSILON WITH TONOS - u'\u03ae' # 0x009e -> GREEK SMALL LETTER ETA WITH TONOS - u'\u03af' # 0x009f -> GREEK SMALL LETTER IOTA WITH TONOS - u'\u03ca' # 0x00a0 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA - u'\u0390' # 0x00a1 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - u'\u03cc' # 0x00a2 -> GREEK SMALL LETTER OMICRON WITH TONOS - u'\u03cd' # 0x00a3 -> GREEK SMALL LETTER UPSILON WITH TONOS - u'\u0391' # 0x00a4 -> GREEK CAPITAL LETTER ALPHA - u'\u0392' # 0x00a5 -> GREEK CAPITAL LETTER BETA - u'\u0393' # 0x00a6 -> GREEK CAPITAL LETTER GAMMA - u'\u0394' # 0x00a7 -> GREEK CAPITAL LETTER DELTA - u'\u0395' # 0x00a8 -> GREEK CAPITAL LETTER EPSILON - u'\u0396' # 0x00a9 -> GREEK CAPITAL LETTER ZETA - u'\u0397' # 0x00aa -> GREEK CAPITAL LETTER ETA - u'\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF - u'\u0398' # 0x00ac -> GREEK CAPITAL LETTER THETA - u'\u0399' # 0x00ad -> GREEK CAPITAL LETTER IOTA - u'\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2591' # 0x00b0 -> LIGHT SHADE - u'\u2592' # 0x00b1 -> MEDIUM SHADE - u'\u2593' # 0x00b2 -> DARK SHADE - u'\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL - u'\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u039a' # 0x00b5 -> GREEK CAPITAL LETTER KAPPA - u'\u039b' # 0x00b6 -> GREEK CAPITAL LETTER LAMDA - u'\u039c' # 0x00b7 -> GREEK CAPITAL LETTER MU - u'\u039d' # 0x00b8 -> GREEK CAPITAL LETTER NU - u'\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u039e' # 0x00bd -> GREEK CAPITAL LETTER XI - u'\u039f' # 0x00be -> GREEK CAPITAL LETTER OMICRON - u'\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u03a0' # 0x00c6 -> GREEK CAPITAL LETTER PI - u'\u03a1' # 0x00c7 -> GREEK CAPITAL LETTER RHO - u'\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\u03a3' # 0x00cf -> GREEK CAPITAL LETTER SIGMA - u'\u03a4' # 0x00d0 -> GREEK CAPITAL LETTER TAU - u'\u03a5' # 0x00d1 -> GREEK CAPITAL LETTER UPSILON - u'\u03a6' # 0x00d2 -> GREEK CAPITAL LETTER PHI - u'\u03a7' # 0x00d3 -> GREEK CAPITAL LETTER CHI - u'\u03a8' # 0x00d4 -> GREEK CAPITAL LETTER PSI - u'\u03a9' # 0x00d5 -> GREEK CAPITAL LETTER OMEGA - u'\u03b1' # 0x00d6 -> GREEK SMALL LETTER ALPHA - u'\u03b2' # 0x00d7 -> GREEK SMALL LETTER BETA - u'\u03b3' # 0x00d8 -> GREEK SMALL LETTER GAMMA - u'\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2588' # 0x00db -> FULL BLOCK - u'\u2584' # 0x00dc -> LOWER HALF BLOCK - u'\u03b4' # 0x00dd -> GREEK SMALL LETTER DELTA - u'\u03b5' # 0x00de -> GREEK SMALL LETTER EPSILON - u'\u2580' # 0x00df -> UPPER HALF BLOCK - u'\u03b6' # 0x00e0 -> GREEK SMALL LETTER ZETA - u'\u03b7' # 0x00e1 -> GREEK SMALL LETTER ETA - u'\u03b8' # 0x00e2 -> GREEK SMALL LETTER THETA - u'\u03b9' # 0x00e3 -> GREEK SMALL LETTER IOTA - u'\u03ba' # 0x00e4 -> GREEK SMALL LETTER KAPPA - u'\u03bb' # 0x00e5 -> GREEK SMALL LETTER LAMDA - u'\u03bc' # 0x00e6 -> GREEK SMALL LETTER MU - u'\u03bd' # 0x00e7 -> GREEK SMALL LETTER NU - u'\u03be' # 0x00e8 -> GREEK SMALL LETTER XI - u'\u03bf' # 0x00e9 -> GREEK SMALL LETTER OMICRON - u'\u03c0' # 0x00ea -> GREEK SMALL LETTER PI - u'\u03c1' # 0x00eb -> GREEK SMALL LETTER RHO - u'\u03c3' # 0x00ec -> GREEK SMALL LETTER SIGMA - u'\u03c2' # 0x00ed -> GREEK SMALL LETTER FINAL SIGMA - u'\u03c4' # 0x00ee -> GREEK SMALL LETTER TAU - u'\u0384' # 0x00ef -> GREEK TONOS - u'\xad' # 0x00f0 -> SOFT HYPHEN - u'\xb1' # 0x00f1 -> PLUS-MINUS SIGN - u'\u03c5' # 0x00f2 -> GREEK SMALL LETTER UPSILON - u'\u03c6' # 0x00f3 -> GREEK SMALL LETTER PHI - u'\u03c7' # 0x00f4 -> GREEK SMALL LETTER CHI - u'\xa7' # 0x00f5 -> SECTION SIGN - u'\u03c8' # 0x00f6 -> GREEK SMALL LETTER PSI - u'\u0385' # 0x00f7 -> GREEK DIALYTIKA TONOS - u'\xb0' # 0x00f8 -> DEGREE SIGN - u'\xa8' # 0x00f9 -> DIAERESIS - u'\u03c9' # 0x00fa -> GREEK SMALL LETTER OMEGA - u'\u03cb' # 0x00fb -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA - u'\u03b0' # 0x00fc -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - u'\u03ce' # 0x00fd -> GREEK SMALL LETTER OMEGA WITH TONOS - u'\u25a0' # 0x00fe -> BLACK SQUARE - u'\xa0' # 0x00ff -> NO-BREAK SPACE + '\x00' # 0x0000 -> NULL + '\x01' # 0x0001 -> START OF HEADING + '\x02' # 0x0002 -> START OF TEXT + '\x03' # 0x0003 -> END OF TEXT + '\x04' # 0x0004 -> END OF TRANSMISSION + '\x05' # 0x0005 -> ENQUIRY + '\x06' # 0x0006 -> ACKNOWLEDGE + '\x07' # 0x0007 -> BELL + '\x08' # 0x0008 -> BACKSPACE + '\t' # 0x0009 -> HORIZONTAL TABULATION + '\n' # 0x000a -> LINE FEED + '\x0b' # 0x000b -> VERTICAL TABULATION + '\x0c' # 0x000c -> FORM FEED + '\r' # 0x000d -> CARRIAGE RETURN + '\x0e' # 0x000e -> SHIFT OUT + '\x0f' # 0x000f -> SHIFT IN + '\x10' # 0x0010 -> DATA LINK ESCAPE + '\x11' # 0x0011 -> DEVICE CONTROL ONE + '\x12' # 0x0012 -> DEVICE CONTROL TWO + '\x13' # 0x0013 -> DEVICE CONTROL THREE + '\x14' # 0x0014 -> DEVICE CONTROL FOUR + '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x0016 -> SYNCHRONOUS IDLE + '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK + '\x18' # 0x0018 -> CANCEL + '\x19' # 0x0019 -> END OF MEDIUM + '\x1a' # 0x001a -> SUBSTITUTE + '\x1b' # 0x001b -> ESCAPE + '\x1c' # 0x001c -> FILE SEPARATOR + '\x1d' # 0x001d -> GROUP SEPARATOR + '\x1e' # 0x001e -> RECORD SEPARATOR + '\x1f' # 0x001f -> UNIT SEPARATOR + ' ' # 0x0020 -> SPACE + '!' # 0x0021 -> EXCLAMATION MARK + '"' # 0x0022 -> QUOTATION MARK + '#' # 0x0023 -> NUMBER SIGN + '$' # 0x0024 -> DOLLAR SIGN + '%' # 0x0025 -> PERCENT SIGN + '&' # 0x0026 -> AMPERSAND + "'" # 0x0027 -> APOSTROPHE + '(' # 0x0028 -> LEFT PARENTHESIS + ')' # 0x0029 -> RIGHT PARENTHESIS + '*' # 0x002a -> ASTERISK + '+' # 0x002b -> PLUS SIGN + ',' # 0x002c -> COMMA + '-' # 0x002d -> HYPHEN-MINUS + '.' # 0x002e -> FULL STOP + '/' # 0x002f -> SOLIDUS + '0' # 0x0030 -> DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE + '2' # 0x0032 -> DIGIT TWO + '3' # 0x0033 -> DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE + ':' # 0x003a -> COLON + ';' # 0x003b -> SEMICOLON + '<' # 0x003c -> LESS-THAN SIGN + '=' # 0x003d -> EQUALS SIGN + '>' # 0x003e -> GREATER-THAN SIGN + '?' # 0x003f -> QUESTION MARK + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET + '\\' # 0x005c -> REVERSE SOLIDUS + ']' # 0x005d -> RIGHT SQUARE BRACKET + '^' # 0x005e -> CIRCUMFLEX ACCENT + '_' # 0x005f -> LOW LINE + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET + '|' # 0x007c -> VERTICAL LINE + '}' # 0x007d -> RIGHT CURLY BRACKET + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> DELETE + '\ufffe' # 0x0080 -> UNDEFINED + '\ufffe' # 0x0081 -> UNDEFINED + '\ufffe' # 0x0082 -> UNDEFINED + '\ufffe' # 0x0083 -> UNDEFINED + '\ufffe' # 0x0084 -> UNDEFINED + '\ufffe' # 0x0085 -> UNDEFINED + '\u0386' # 0x0086 -> GREEK CAPITAL LETTER ALPHA WITH TONOS + '\ufffe' # 0x0087 -> UNDEFINED + '\xb7' # 0x0088 -> MIDDLE DOT + '\xac' # 0x0089 -> NOT SIGN + '\xa6' # 0x008a -> BROKEN BAR + '\u2018' # 0x008b -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x008c -> RIGHT SINGLE QUOTATION MARK + '\u0388' # 0x008d -> GREEK CAPITAL LETTER EPSILON WITH TONOS + '\u2015' # 0x008e -> HORIZONTAL BAR + '\u0389' # 0x008f -> GREEK CAPITAL LETTER ETA WITH TONOS + '\u038a' # 0x0090 -> GREEK CAPITAL LETTER IOTA WITH TONOS + '\u03aa' # 0x0091 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA + '\u038c' # 0x0092 -> GREEK CAPITAL LETTER OMICRON WITH TONOS + '\ufffe' # 0x0093 -> UNDEFINED + '\ufffe' # 0x0094 -> UNDEFINED + '\u038e' # 0x0095 -> GREEK CAPITAL LETTER UPSILON WITH TONOS + '\u03ab' # 0x0096 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA + '\xa9' # 0x0097 -> COPYRIGHT SIGN + '\u038f' # 0x0098 -> GREEK CAPITAL LETTER OMEGA WITH TONOS + '\xb2' # 0x0099 -> SUPERSCRIPT TWO + '\xb3' # 0x009a -> SUPERSCRIPT THREE + '\u03ac' # 0x009b -> GREEK SMALL LETTER ALPHA WITH TONOS + '\xa3' # 0x009c -> POUND SIGN + '\u03ad' # 0x009d -> GREEK SMALL LETTER EPSILON WITH TONOS + '\u03ae' # 0x009e -> GREEK SMALL LETTER ETA WITH TONOS + '\u03af' # 0x009f -> GREEK SMALL LETTER IOTA WITH TONOS + '\u03ca' # 0x00a0 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA + '\u0390' # 0x00a1 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS + '\u03cc' # 0x00a2 -> GREEK SMALL LETTER OMICRON WITH TONOS + '\u03cd' # 0x00a3 -> GREEK SMALL LETTER UPSILON WITH TONOS + '\u0391' # 0x00a4 -> GREEK CAPITAL LETTER ALPHA + '\u0392' # 0x00a5 -> GREEK CAPITAL LETTER BETA + '\u0393' # 0x00a6 -> GREEK CAPITAL LETTER GAMMA + '\u0394' # 0x00a7 -> GREEK CAPITAL LETTER DELTA + '\u0395' # 0x00a8 -> GREEK CAPITAL LETTER EPSILON + '\u0396' # 0x00a9 -> GREEK CAPITAL LETTER ZETA + '\u0397' # 0x00aa -> GREEK CAPITAL LETTER ETA + '\xbd' # 0x00ab -> VULGAR FRACTION ONE HALF + '\u0398' # 0x00ac -> GREEK CAPITAL LETTER THETA + '\u0399' # 0x00ad -> GREEK CAPITAL LETTER IOTA + '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2591' # 0x00b0 -> LIGHT SHADE + '\u2592' # 0x00b1 -> MEDIUM SHADE + '\u2593' # 0x00b2 -> DARK SHADE + '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL + '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u039a' # 0x00b5 -> GREEK CAPITAL LETTER KAPPA + '\u039b' # 0x00b6 -> GREEK CAPITAL LETTER LAMDA + '\u039c' # 0x00b7 -> GREEK CAPITAL LETTER MU + '\u039d' # 0x00b8 -> GREEK CAPITAL LETTER NU + '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL + '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u039e' # 0x00bd -> GREEK CAPITAL LETTER XI + '\u039f' # 0x00be -> GREEK CAPITAL LETTER OMICRON + '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u03a0' # 0x00c6 -> GREEK CAPITAL LETTER PI + '\u03a1' # 0x00c7 -> GREEK CAPITAL LETTER RHO + '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\u03a3' # 0x00cf -> GREEK CAPITAL LETTER SIGMA + '\u03a4' # 0x00d0 -> GREEK CAPITAL LETTER TAU + '\u03a5' # 0x00d1 -> GREEK CAPITAL LETTER UPSILON + '\u03a6' # 0x00d2 -> GREEK CAPITAL LETTER PHI + '\u03a7' # 0x00d3 -> GREEK CAPITAL LETTER CHI + '\u03a8' # 0x00d4 -> GREEK CAPITAL LETTER PSI + '\u03a9' # 0x00d5 -> GREEK CAPITAL LETTER OMEGA + '\u03b1' # 0x00d6 -> GREEK SMALL LETTER ALPHA + '\u03b2' # 0x00d7 -> GREEK SMALL LETTER BETA + '\u03b3' # 0x00d8 -> GREEK SMALL LETTER GAMMA + '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2588' # 0x00db -> FULL BLOCK + '\u2584' # 0x00dc -> LOWER HALF BLOCK + '\u03b4' # 0x00dd -> GREEK SMALL LETTER DELTA + '\u03b5' # 0x00de -> GREEK SMALL LETTER EPSILON + '\u2580' # 0x00df -> UPPER HALF BLOCK + '\u03b6' # 0x00e0 -> GREEK SMALL LETTER ZETA + '\u03b7' # 0x00e1 -> GREEK SMALL LETTER ETA + '\u03b8' # 0x00e2 -> GREEK SMALL LETTER THETA + '\u03b9' # 0x00e3 -> GREEK SMALL LETTER IOTA + '\u03ba' # 0x00e4 -> GREEK SMALL LETTER KAPPA + '\u03bb' # 0x00e5 -> GREEK SMALL LETTER LAMDA + '\u03bc' # 0x00e6 -> GREEK SMALL LETTER MU + '\u03bd' # 0x00e7 -> GREEK SMALL LETTER NU + '\u03be' # 0x00e8 -> GREEK SMALL LETTER XI + '\u03bf' # 0x00e9 -> GREEK SMALL LETTER OMICRON + '\u03c0' # 0x00ea -> GREEK SMALL LETTER PI + '\u03c1' # 0x00eb -> GREEK SMALL LETTER RHO + '\u03c3' # 0x00ec -> GREEK SMALL LETTER SIGMA + '\u03c2' # 0x00ed -> GREEK SMALL LETTER FINAL SIGMA + '\u03c4' # 0x00ee -> GREEK SMALL LETTER TAU + '\u0384' # 0x00ef -> GREEK TONOS + '\xad' # 0x00f0 -> SOFT HYPHEN + '\xb1' # 0x00f1 -> PLUS-MINUS SIGN + '\u03c5' # 0x00f2 -> GREEK SMALL LETTER UPSILON + '\u03c6' # 0x00f3 -> GREEK SMALL LETTER PHI + '\u03c7' # 0x00f4 -> GREEK SMALL LETTER CHI + '\xa7' # 0x00f5 -> SECTION SIGN + '\u03c8' # 0x00f6 -> GREEK SMALL LETTER PSI + '\u0385' # 0x00f7 -> GREEK DIALYTIKA TONOS + '\xb0' # 0x00f8 -> DEGREE SIGN + '\xa8' # 0x00f9 -> DIAERESIS + '\u03c9' # 0x00fa -> GREEK SMALL LETTER OMEGA + '\u03cb' # 0x00fb -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA + '\u03b0' # 0x00fc -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS + '\u03ce' # 0x00fd -> GREEK SMALL LETTER OMEGA WITH TONOS + '\u25a0' # 0x00fe -> BLACK SQUARE + '\xa0' # 0x00ff -> NO-BREAK SPACE ) ### Encoding Map diff --git a/Lib/encodings/cp874.py b/Lib/encodings/cp874.py index 6110f46..59bfcbc 100644 --- a/Lib/encodings/cp874.py +++ b/Lib/encodings/cp874.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u20ac' # 0x80 -> EURO SIGN - u'\ufffe' # 0x81 -> UNDEFINED - u'\ufffe' # 0x82 -> UNDEFINED - u'\ufffe' # 0x83 -> UNDEFINED - u'\ufffe' # 0x84 -> UNDEFINED - u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS - u'\ufffe' # 0x86 -> UNDEFINED - u'\ufffe' # 0x87 -> UNDEFINED - u'\ufffe' # 0x88 -> UNDEFINED - u'\ufffe' # 0x89 -> UNDEFINED - u'\ufffe' # 0x8A -> UNDEFINED - u'\ufffe' # 0x8B -> UNDEFINED - u'\ufffe' # 0x8C -> UNDEFINED - u'\ufffe' # 0x8D -> UNDEFINED - u'\ufffe' # 0x8E -> UNDEFINED - u'\ufffe' # 0x8F -> UNDEFINED - u'\ufffe' # 0x90 -> UNDEFINED - u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK - u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK - u'\u2022' # 0x95 -> BULLET - u'\u2013' # 0x96 -> EN DASH - u'\u2014' # 0x97 -> EM DASH - u'\ufffe' # 0x98 -> UNDEFINED - u'\ufffe' # 0x99 -> UNDEFINED - u'\ufffe' # 0x9A -> UNDEFINED - u'\ufffe' # 0x9B -> UNDEFINED - u'\ufffe' # 0x9C -> UNDEFINED - u'\ufffe' # 0x9D -> UNDEFINED - u'\ufffe' # 0x9E -> UNDEFINED - u'\ufffe' # 0x9F -> UNDEFINED - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0e01' # 0xA1 -> THAI CHARACTER KO KAI - u'\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI - u'\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT - u'\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI - u'\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON - u'\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG - u'\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU - u'\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN - u'\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING - u'\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG - u'\u0e0b' # 0xAB -> THAI CHARACTER SO SO - u'\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE - u'\u0e0d' # 0xAD -> THAI CHARACTER YO YING - u'\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA - u'\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK - u'\u0e10' # 0xB0 -> THAI CHARACTER THO THAN - u'\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO - u'\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO - u'\u0e13' # 0xB3 -> THAI CHARACTER NO NEN - u'\u0e14' # 0xB4 -> THAI CHARACTER DO DEK - u'\u0e15' # 0xB5 -> THAI CHARACTER TO TAO - u'\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG - u'\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN - u'\u0e18' # 0xB8 -> THAI CHARACTER THO THONG - u'\u0e19' # 0xB9 -> THAI CHARACTER NO NU - u'\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI - u'\u0e1b' # 0xBB -> THAI CHARACTER PO PLA - u'\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG - u'\u0e1d' # 0xBD -> THAI CHARACTER FO FA - u'\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN - u'\u0e1f' # 0xBF -> THAI CHARACTER FO FAN - u'\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO - u'\u0e21' # 0xC1 -> THAI CHARACTER MO MA - u'\u0e22' # 0xC2 -> THAI CHARACTER YO YAK - u'\u0e23' # 0xC3 -> THAI CHARACTER RO RUA - u'\u0e24' # 0xC4 -> THAI CHARACTER RU - u'\u0e25' # 0xC5 -> THAI CHARACTER LO LING - u'\u0e26' # 0xC6 -> THAI CHARACTER LU - u'\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN - u'\u0e28' # 0xC8 -> THAI CHARACTER SO SALA - u'\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI - u'\u0e2a' # 0xCA -> THAI CHARACTER SO SUA - u'\u0e2b' # 0xCB -> THAI CHARACTER HO HIP - u'\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA - u'\u0e2d' # 0xCD -> THAI CHARACTER O ANG - u'\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK - u'\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI - u'\u0e30' # 0xD0 -> THAI CHARACTER SARA A - u'\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT - u'\u0e32' # 0xD2 -> THAI CHARACTER SARA AA - u'\u0e33' # 0xD3 -> THAI CHARACTER SARA AM - u'\u0e34' # 0xD4 -> THAI CHARACTER SARA I - u'\u0e35' # 0xD5 -> THAI CHARACTER SARA II - u'\u0e36' # 0xD6 -> THAI CHARACTER SARA UE - u'\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE - u'\u0e38' # 0xD8 -> THAI CHARACTER SARA U - u'\u0e39' # 0xD9 -> THAI CHARACTER SARA UU - u'\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU - u'\ufffe' # 0xDB -> UNDEFINED - u'\ufffe' # 0xDC -> UNDEFINED - u'\ufffe' # 0xDD -> UNDEFINED - u'\ufffe' # 0xDE -> UNDEFINED - u'\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT - u'\u0e40' # 0xE0 -> THAI CHARACTER SARA E - u'\u0e41' # 0xE1 -> THAI CHARACTER SARA AE - u'\u0e42' # 0xE2 -> THAI CHARACTER SARA O - u'\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN - u'\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI - u'\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO - u'\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK - u'\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU - u'\u0e48' # 0xE8 -> THAI CHARACTER MAI EK - u'\u0e49' # 0xE9 -> THAI CHARACTER MAI THO - u'\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI - u'\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA - u'\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT - u'\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT - u'\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN - u'\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN - u'\u0e50' # 0xF0 -> THAI DIGIT ZERO - u'\u0e51' # 0xF1 -> THAI DIGIT ONE - u'\u0e52' # 0xF2 -> THAI DIGIT TWO - u'\u0e53' # 0xF3 -> THAI DIGIT THREE - u'\u0e54' # 0xF4 -> THAI DIGIT FOUR - u'\u0e55' # 0xF5 -> THAI DIGIT FIVE - u'\u0e56' # 0xF6 -> THAI DIGIT SIX - u'\u0e57' # 0xF7 -> THAI DIGIT SEVEN - u'\u0e58' # 0xF8 -> THAI DIGIT EIGHT - u'\u0e59' # 0xF9 -> THAI DIGIT NINE - u'\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU - u'\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT - u'\ufffe' # 0xFC -> UNDEFINED - u'\ufffe' # 0xFD -> UNDEFINED - u'\ufffe' # 0xFE -> UNDEFINED - u'\ufffe' # 0xFF -> UNDEFINED + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u20ac' # 0x80 -> EURO SIGN + '\ufffe' # 0x81 -> UNDEFINED + '\ufffe' # 0x82 -> UNDEFINED + '\ufffe' # 0x83 -> UNDEFINED + '\ufffe' # 0x84 -> UNDEFINED + '\u2026' # 0x85 -> HORIZONTAL ELLIPSIS + '\ufffe' # 0x86 -> UNDEFINED + '\ufffe' # 0x87 -> UNDEFINED + '\ufffe' # 0x88 -> UNDEFINED + '\ufffe' # 0x89 -> UNDEFINED + '\ufffe' # 0x8A -> UNDEFINED + '\ufffe' # 0x8B -> UNDEFINED + '\ufffe' # 0x8C -> UNDEFINED + '\ufffe' # 0x8D -> UNDEFINED + '\ufffe' # 0x8E -> UNDEFINED + '\ufffe' # 0x8F -> UNDEFINED + '\ufffe' # 0x90 -> UNDEFINED + '\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK + '\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK + '\u2022' # 0x95 -> BULLET + '\u2013' # 0x96 -> EN DASH + '\u2014' # 0x97 -> EM DASH + '\ufffe' # 0x98 -> UNDEFINED + '\ufffe' # 0x99 -> UNDEFINED + '\ufffe' # 0x9A -> UNDEFINED + '\ufffe' # 0x9B -> UNDEFINED + '\ufffe' # 0x9C -> UNDEFINED + '\ufffe' # 0x9D -> UNDEFINED + '\ufffe' # 0x9E -> UNDEFINED + '\ufffe' # 0x9F -> UNDEFINED + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0e01' # 0xA1 -> THAI CHARACTER KO KAI + '\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI + '\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT + '\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI + '\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON + '\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG + '\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU + '\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN + '\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING + '\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG + '\u0e0b' # 0xAB -> THAI CHARACTER SO SO + '\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE + '\u0e0d' # 0xAD -> THAI CHARACTER YO YING + '\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA + '\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK + '\u0e10' # 0xB0 -> THAI CHARACTER THO THAN + '\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO + '\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO + '\u0e13' # 0xB3 -> THAI CHARACTER NO NEN + '\u0e14' # 0xB4 -> THAI CHARACTER DO DEK + '\u0e15' # 0xB5 -> THAI CHARACTER TO TAO + '\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG + '\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN + '\u0e18' # 0xB8 -> THAI CHARACTER THO THONG + '\u0e19' # 0xB9 -> THAI CHARACTER NO NU + '\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI + '\u0e1b' # 0xBB -> THAI CHARACTER PO PLA + '\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG + '\u0e1d' # 0xBD -> THAI CHARACTER FO FA + '\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN + '\u0e1f' # 0xBF -> THAI CHARACTER FO FAN + '\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO + '\u0e21' # 0xC1 -> THAI CHARACTER MO MA + '\u0e22' # 0xC2 -> THAI CHARACTER YO YAK + '\u0e23' # 0xC3 -> THAI CHARACTER RO RUA + '\u0e24' # 0xC4 -> THAI CHARACTER RU + '\u0e25' # 0xC5 -> THAI CHARACTER LO LING + '\u0e26' # 0xC6 -> THAI CHARACTER LU + '\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN + '\u0e28' # 0xC8 -> THAI CHARACTER SO SALA + '\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI + '\u0e2a' # 0xCA -> THAI CHARACTER SO SUA + '\u0e2b' # 0xCB -> THAI CHARACTER HO HIP + '\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA + '\u0e2d' # 0xCD -> THAI CHARACTER O ANG + '\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK + '\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI + '\u0e30' # 0xD0 -> THAI CHARACTER SARA A + '\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT + '\u0e32' # 0xD2 -> THAI CHARACTER SARA AA + '\u0e33' # 0xD3 -> THAI CHARACTER SARA AM + '\u0e34' # 0xD4 -> THAI CHARACTER SARA I + '\u0e35' # 0xD5 -> THAI CHARACTER SARA II + '\u0e36' # 0xD6 -> THAI CHARACTER SARA UE + '\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE + '\u0e38' # 0xD8 -> THAI CHARACTER SARA U + '\u0e39' # 0xD9 -> THAI CHARACTER SARA UU + '\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU + '\ufffe' # 0xDB -> UNDEFINED + '\ufffe' # 0xDC -> UNDEFINED + '\ufffe' # 0xDD -> UNDEFINED + '\ufffe' # 0xDE -> UNDEFINED + '\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT + '\u0e40' # 0xE0 -> THAI CHARACTER SARA E + '\u0e41' # 0xE1 -> THAI CHARACTER SARA AE + '\u0e42' # 0xE2 -> THAI CHARACTER SARA O + '\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN + '\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI + '\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO + '\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK + '\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU + '\u0e48' # 0xE8 -> THAI CHARACTER MAI EK + '\u0e49' # 0xE9 -> THAI CHARACTER MAI THO + '\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI + '\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA + '\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT + '\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT + '\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN + '\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN + '\u0e50' # 0xF0 -> THAI DIGIT ZERO + '\u0e51' # 0xF1 -> THAI DIGIT ONE + '\u0e52' # 0xF2 -> THAI DIGIT TWO + '\u0e53' # 0xF3 -> THAI DIGIT THREE + '\u0e54' # 0xF4 -> THAI DIGIT FOUR + '\u0e55' # 0xF5 -> THAI DIGIT FIVE + '\u0e56' # 0xF6 -> THAI DIGIT SIX + '\u0e57' # 0xF7 -> THAI DIGIT SEVEN + '\u0e58' # 0xF8 -> THAI DIGIT EIGHT + '\u0e59' # 0xF9 -> THAI DIGIT NINE + '\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU + '\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT + '\ufffe' # 0xFC -> UNDEFINED + '\ufffe' # 0xFD -> UNDEFINED + '\ufffe' # 0xFE -> UNDEFINED + '\ufffe' # 0xFF -> UNDEFINED ) ### Encoding table diff --git a/Lib/encodings/cp875.py b/Lib/encodings/cp875.py index 72b160b..c25a5a4 100644 --- a/Lib/encodings/cp875.py +++ b/Lib/encodings/cp875.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x9c' # 0x04 -> CONTROL - u'\t' # 0x05 -> HORIZONTAL TABULATION - u'\x86' # 0x06 -> CONTROL - u'\x7f' # 0x07 -> DELETE - u'\x97' # 0x08 -> CONTROL - u'\x8d' # 0x09 -> CONTROL - u'\x8e' # 0x0A -> CONTROL - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x9d' # 0x14 -> CONTROL - u'\x85' # 0x15 -> CONTROL - u'\x08' # 0x16 -> BACKSPACE - u'\x87' # 0x17 -> CONTROL - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x92' # 0x1A -> CONTROL - u'\x8f' # 0x1B -> CONTROL - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u'\x80' # 0x20 -> CONTROL - u'\x81' # 0x21 -> CONTROL - u'\x82' # 0x22 -> CONTROL - u'\x83' # 0x23 -> CONTROL - u'\x84' # 0x24 -> CONTROL - u'\n' # 0x25 -> LINE FEED - u'\x17' # 0x26 -> END OF TRANSMISSION BLOCK - u'\x1b' # 0x27 -> ESCAPE - u'\x88' # 0x28 -> CONTROL - u'\x89' # 0x29 -> CONTROL - u'\x8a' # 0x2A -> CONTROL - u'\x8b' # 0x2B -> CONTROL - u'\x8c' # 0x2C -> CONTROL - u'\x05' # 0x2D -> ENQUIRY - u'\x06' # 0x2E -> ACKNOWLEDGE - u'\x07' # 0x2F -> BELL - u'\x90' # 0x30 -> CONTROL - u'\x91' # 0x31 -> CONTROL - u'\x16' # 0x32 -> SYNCHRONOUS IDLE - u'\x93' # 0x33 -> CONTROL - u'\x94' # 0x34 -> CONTROL - u'\x95' # 0x35 -> CONTROL - u'\x96' # 0x36 -> CONTROL - u'\x04' # 0x37 -> END OF TRANSMISSION - u'\x98' # 0x38 -> CONTROL - u'\x99' # 0x39 -> CONTROL - u'\x9a' # 0x3A -> CONTROL - u'\x9b' # 0x3B -> CONTROL - u'\x14' # 0x3C -> DEVICE CONTROL FOUR - u'\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE - u'\x9e' # 0x3E -> CONTROL - u'\x1a' # 0x3F -> SUBSTITUTE - u' ' # 0x40 -> SPACE - u'\u0391' # 0x41 -> GREEK CAPITAL LETTER ALPHA - u'\u0392' # 0x42 -> GREEK CAPITAL LETTER BETA - u'\u0393' # 0x43 -> GREEK CAPITAL LETTER GAMMA - u'\u0394' # 0x44 -> GREEK CAPITAL LETTER DELTA - u'\u0395' # 0x45 -> GREEK CAPITAL LETTER EPSILON - u'\u0396' # 0x46 -> GREEK CAPITAL LETTER ZETA - u'\u0397' # 0x47 -> GREEK CAPITAL LETTER ETA - u'\u0398' # 0x48 -> GREEK CAPITAL LETTER THETA - u'\u0399' # 0x49 -> GREEK CAPITAL LETTER IOTA - u'[' # 0x4A -> LEFT SQUARE BRACKET - u'.' # 0x4B -> FULL STOP - u'<' # 0x4C -> LESS-THAN SIGN - u'(' # 0x4D -> LEFT PARENTHESIS - u'+' # 0x4E -> PLUS SIGN - u'!' # 0x4F -> EXCLAMATION MARK - u'&' # 0x50 -> AMPERSAND - u'\u039a' # 0x51 -> GREEK CAPITAL LETTER KAPPA - u'\u039b' # 0x52 -> GREEK CAPITAL LETTER LAMDA - u'\u039c' # 0x53 -> GREEK CAPITAL LETTER MU - u'\u039d' # 0x54 -> GREEK CAPITAL LETTER NU - u'\u039e' # 0x55 -> GREEK CAPITAL LETTER XI - u'\u039f' # 0x56 -> GREEK CAPITAL LETTER OMICRON - u'\u03a0' # 0x57 -> GREEK CAPITAL LETTER PI - u'\u03a1' # 0x58 -> GREEK CAPITAL LETTER RHO - u'\u03a3' # 0x59 -> GREEK CAPITAL LETTER SIGMA - u']' # 0x5A -> RIGHT SQUARE BRACKET - u'$' # 0x5B -> DOLLAR SIGN - u'*' # 0x5C -> ASTERISK - u')' # 0x5D -> RIGHT PARENTHESIS - u';' # 0x5E -> SEMICOLON - u'^' # 0x5F -> CIRCUMFLEX ACCENT - u'-' # 0x60 -> HYPHEN-MINUS - u'/' # 0x61 -> SOLIDUS - u'\u03a4' # 0x62 -> GREEK CAPITAL LETTER TAU - u'\u03a5' # 0x63 -> GREEK CAPITAL LETTER UPSILON - u'\u03a6' # 0x64 -> GREEK CAPITAL LETTER PHI - u'\u03a7' # 0x65 -> GREEK CAPITAL LETTER CHI - u'\u03a8' # 0x66 -> GREEK CAPITAL LETTER PSI - u'\u03a9' # 0x67 -> GREEK CAPITAL LETTER OMEGA - u'\u03aa' # 0x68 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - u'\u03ab' # 0x69 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - u'|' # 0x6A -> VERTICAL LINE - u',' # 0x6B -> COMMA - u'%' # 0x6C -> PERCENT SIGN - u'_' # 0x6D -> LOW LINE - u'>' # 0x6E -> GREATER-THAN SIGN - u'?' # 0x6F -> QUESTION MARK - u'\xa8' # 0x70 -> DIAERESIS - u'\u0386' # 0x71 -> GREEK CAPITAL LETTER ALPHA WITH TONOS - u'\u0388' # 0x72 -> GREEK CAPITAL LETTER EPSILON WITH TONOS - u'\u0389' # 0x73 -> GREEK CAPITAL LETTER ETA WITH TONOS - u'\xa0' # 0x74 -> NO-BREAK SPACE - u'\u038a' # 0x75 -> GREEK CAPITAL LETTER IOTA WITH TONOS - u'\u038c' # 0x76 -> GREEK CAPITAL LETTER OMICRON WITH TONOS - u'\u038e' # 0x77 -> GREEK CAPITAL LETTER UPSILON WITH TONOS - u'\u038f' # 0x78 -> GREEK CAPITAL LETTER OMEGA WITH TONOS - u'`' # 0x79 -> GRAVE ACCENT - u':' # 0x7A -> COLON - u'#' # 0x7B -> NUMBER SIGN - u'@' # 0x7C -> COMMERCIAL AT - u"'" # 0x7D -> APOSTROPHE - u'=' # 0x7E -> EQUALS SIGN - u'"' # 0x7F -> QUOTATION MARK - u'\u0385' # 0x80 -> GREEK DIALYTIKA TONOS - u'a' # 0x81 -> LATIN SMALL LETTER A - u'b' # 0x82 -> LATIN SMALL LETTER B - u'c' # 0x83 -> LATIN SMALL LETTER C - u'd' # 0x84 -> LATIN SMALL LETTER D - u'e' # 0x85 -> LATIN SMALL LETTER E - u'f' # 0x86 -> LATIN SMALL LETTER F - u'g' # 0x87 -> LATIN SMALL LETTER G - u'h' # 0x88 -> LATIN SMALL LETTER H - u'i' # 0x89 -> LATIN SMALL LETTER I - u'\u03b1' # 0x8A -> GREEK SMALL LETTER ALPHA - u'\u03b2' # 0x8B -> GREEK SMALL LETTER BETA - u'\u03b3' # 0x8C -> GREEK SMALL LETTER GAMMA - u'\u03b4' # 0x8D -> GREEK SMALL LETTER DELTA - u'\u03b5' # 0x8E -> GREEK SMALL LETTER EPSILON - u'\u03b6' # 0x8F -> GREEK SMALL LETTER ZETA - u'\xb0' # 0x90 -> DEGREE SIGN - u'j' # 0x91 -> LATIN SMALL LETTER J - u'k' # 0x92 -> LATIN SMALL LETTER K - u'l' # 0x93 -> LATIN SMALL LETTER L - u'm' # 0x94 -> LATIN SMALL LETTER M - u'n' # 0x95 -> LATIN SMALL LETTER N - u'o' # 0x96 -> LATIN SMALL LETTER O - u'p' # 0x97 -> LATIN SMALL LETTER P - u'q' # 0x98 -> LATIN SMALL LETTER Q - u'r' # 0x99 -> LATIN SMALL LETTER R - u'\u03b7' # 0x9A -> GREEK SMALL LETTER ETA - u'\u03b8' # 0x9B -> GREEK SMALL LETTER THETA - u'\u03b9' # 0x9C -> GREEK SMALL LETTER IOTA - u'\u03ba' # 0x9D -> GREEK SMALL LETTER KAPPA - u'\u03bb' # 0x9E -> GREEK SMALL LETTER LAMDA - u'\u03bc' # 0x9F -> GREEK SMALL LETTER MU - u'\xb4' # 0xA0 -> ACUTE ACCENT - u'~' # 0xA1 -> TILDE - u's' # 0xA2 -> LATIN SMALL LETTER S - u't' # 0xA3 -> LATIN SMALL LETTER T - u'u' # 0xA4 -> LATIN SMALL LETTER U - u'v' # 0xA5 -> LATIN SMALL LETTER V - u'w' # 0xA6 -> LATIN SMALL LETTER W - u'x' # 0xA7 -> LATIN SMALL LETTER X - u'y' # 0xA8 -> LATIN SMALL LETTER Y - u'z' # 0xA9 -> LATIN SMALL LETTER Z - u'\u03bd' # 0xAA -> GREEK SMALL LETTER NU - u'\u03be' # 0xAB -> GREEK SMALL LETTER XI - u'\u03bf' # 0xAC -> GREEK SMALL LETTER OMICRON - u'\u03c0' # 0xAD -> GREEK SMALL LETTER PI - u'\u03c1' # 0xAE -> GREEK SMALL LETTER RHO - u'\u03c3' # 0xAF -> GREEK SMALL LETTER SIGMA - u'\xa3' # 0xB0 -> POUND SIGN - u'\u03ac' # 0xB1 -> GREEK SMALL LETTER ALPHA WITH TONOS - u'\u03ad' # 0xB2 -> GREEK SMALL LETTER EPSILON WITH TONOS - u'\u03ae' # 0xB3 -> GREEK SMALL LETTER ETA WITH TONOS - u'\u03ca' # 0xB4 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA - u'\u03af' # 0xB5 -> GREEK SMALL LETTER IOTA WITH TONOS - u'\u03cc' # 0xB6 -> GREEK SMALL LETTER OMICRON WITH TONOS - u'\u03cd' # 0xB7 -> GREEK SMALL LETTER UPSILON WITH TONOS - u'\u03cb' # 0xB8 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA - u'\u03ce' # 0xB9 -> GREEK SMALL LETTER OMEGA WITH TONOS - u'\u03c2' # 0xBA -> GREEK SMALL LETTER FINAL SIGMA - u'\u03c4' # 0xBB -> GREEK SMALL LETTER TAU - u'\u03c5' # 0xBC -> GREEK SMALL LETTER UPSILON - u'\u03c6' # 0xBD -> GREEK SMALL LETTER PHI - u'\u03c7' # 0xBE -> GREEK SMALL LETTER CHI - u'\u03c8' # 0xBF -> GREEK SMALL LETTER PSI - u'{' # 0xC0 -> LEFT CURLY BRACKET - u'A' # 0xC1 -> LATIN CAPITAL LETTER A - u'B' # 0xC2 -> LATIN CAPITAL LETTER B - u'C' # 0xC3 -> LATIN CAPITAL LETTER C - u'D' # 0xC4 -> LATIN CAPITAL LETTER D - u'E' # 0xC5 -> LATIN CAPITAL LETTER E - u'F' # 0xC6 -> LATIN CAPITAL LETTER F - u'G' # 0xC7 -> LATIN CAPITAL LETTER G - u'H' # 0xC8 -> LATIN CAPITAL LETTER H - u'I' # 0xC9 -> LATIN CAPITAL LETTER I - u'\xad' # 0xCA -> SOFT HYPHEN - u'\u03c9' # 0xCB -> GREEK SMALL LETTER OMEGA - u'\u0390' # 0xCC -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - u'\u03b0' # 0xCD -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - u'\u2018' # 0xCE -> LEFT SINGLE QUOTATION MARK - u'\u2015' # 0xCF -> HORIZONTAL BAR - u'}' # 0xD0 -> RIGHT CURLY BRACKET - u'J' # 0xD1 -> LATIN CAPITAL LETTER J - u'K' # 0xD2 -> LATIN CAPITAL LETTER K - u'L' # 0xD3 -> LATIN CAPITAL LETTER L - u'M' # 0xD4 -> LATIN CAPITAL LETTER M - u'N' # 0xD5 -> LATIN CAPITAL LETTER N - u'O' # 0xD6 -> LATIN CAPITAL LETTER O - u'P' # 0xD7 -> LATIN CAPITAL LETTER P - u'Q' # 0xD8 -> LATIN CAPITAL LETTER Q - u'R' # 0xD9 -> LATIN CAPITAL LETTER R - u'\xb1' # 0xDA -> PLUS-MINUS SIGN - u'\xbd' # 0xDB -> VULGAR FRACTION ONE HALF - u'\x1a' # 0xDC -> SUBSTITUTE - u'\u0387' # 0xDD -> GREEK ANO TELEIA - u'\u2019' # 0xDE -> RIGHT SINGLE QUOTATION MARK - u'\xa6' # 0xDF -> BROKEN BAR - u'\\' # 0xE0 -> REVERSE SOLIDUS - u'\x1a' # 0xE1 -> SUBSTITUTE - u'S' # 0xE2 -> LATIN CAPITAL LETTER S - u'T' # 0xE3 -> LATIN CAPITAL LETTER T - u'U' # 0xE4 -> LATIN CAPITAL LETTER U - u'V' # 0xE5 -> LATIN CAPITAL LETTER V - u'W' # 0xE6 -> LATIN CAPITAL LETTER W - u'X' # 0xE7 -> LATIN CAPITAL LETTER X - u'Y' # 0xE8 -> LATIN CAPITAL LETTER Y - u'Z' # 0xE9 -> LATIN CAPITAL LETTER Z - u'\xb2' # 0xEA -> SUPERSCRIPT TWO - u'\xa7' # 0xEB -> SECTION SIGN - u'\x1a' # 0xEC -> SUBSTITUTE - u'\x1a' # 0xED -> SUBSTITUTE - u'\xab' # 0xEE -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xEF -> NOT SIGN - u'0' # 0xF0 -> DIGIT ZERO - u'1' # 0xF1 -> DIGIT ONE - u'2' # 0xF2 -> DIGIT TWO - u'3' # 0xF3 -> DIGIT THREE - u'4' # 0xF4 -> DIGIT FOUR - u'5' # 0xF5 -> DIGIT FIVE - u'6' # 0xF6 -> DIGIT SIX - u'7' # 0xF7 -> DIGIT SEVEN - u'8' # 0xF8 -> DIGIT EIGHT - u'9' # 0xF9 -> DIGIT NINE - u'\xb3' # 0xFA -> SUPERSCRIPT THREE - u'\xa9' # 0xFB -> COPYRIGHT SIGN - u'\x1a' # 0xFC -> SUBSTITUTE - u'\x1a' # 0xFD -> SUBSTITUTE - u'\xbb' # 0xFE -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\x9f' # 0xFF -> CONTROL + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x9c' # 0x04 -> CONTROL + '\t' # 0x05 -> HORIZONTAL TABULATION + '\x86' # 0x06 -> CONTROL + '\x7f' # 0x07 -> DELETE + '\x97' # 0x08 -> CONTROL + '\x8d' # 0x09 -> CONTROL + '\x8e' # 0x0A -> CONTROL + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x9d' # 0x14 -> CONTROL + '\x85' # 0x15 -> CONTROL + '\x08' # 0x16 -> BACKSPACE + '\x87' # 0x17 -> CONTROL + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x92' # 0x1A -> CONTROL + '\x8f' # 0x1B -> CONTROL + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + '\x80' # 0x20 -> CONTROL + '\x81' # 0x21 -> CONTROL + '\x82' # 0x22 -> CONTROL + '\x83' # 0x23 -> CONTROL + '\x84' # 0x24 -> CONTROL + '\n' # 0x25 -> LINE FEED + '\x17' # 0x26 -> END OF TRANSMISSION BLOCK + '\x1b' # 0x27 -> ESCAPE + '\x88' # 0x28 -> CONTROL + '\x89' # 0x29 -> CONTROL + '\x8a' # 0x2A -> CONTROL + '\x8b' # 0x2B -> CONTROL + '\x8c' # 0x2C -> CONTROL + '\x05' # 0x2D -> ENQUIRY + '\x06' # 0x2E -> ACKNOWLEDGE + '\x07' # 0x2F -> BELL + '\x90' # 0x30 -> CONTROL + '\x91' # 0x31 -> CONTROL + '\x16' # 0x32 -> SYNCHRONOUS IDLE + '\x93' # 0x33 -> CONTROL + '\x94' # 0x34 -> CONTROL + '\x95' # 0x35 -> CONTROL + '\x96' # 0x36 -> CONTROL + '\x04' # 0x37 -> END OF TRANSMISSION + '\x98' # 0x38 -> CONTROL + '\x99' # 0x39 -> CONTROL + '\x9a' # 0x3A -> CONTROL + '\x9b' # 0x3B -> CONTROL + '\x14' # 0x3C -> DEVICE CONTROL FOUR + '\x15' # 0x3D -> NEGATIVE ACKNOWLEDGE + '\x9e' # 0x3E -> CONTROL + '\x1a' # 0x3F -> SUBSTITUTE + ' ' # 0x40 -> SPACE + '\u0391' # 0x41 -> GREEK CAPITAL LETTER ALPHA + '\u0392' # 0x42 -> GREEK CAPITAL LETTER BETA + '\u0393' # 0x43 -> GREEK CAPITAL LETTER GAMMA + '\u0394' # 0x44 -> GREEK CAPITAL LETTER DELTA + '\u0395' # 0x45 -> GREEK CAPITAL LETTER EPSILON + '\u0396' # 0x46 -> GREEK CAPITAL LETTER ZETA + '\u0397' # 0x47 -> GREEK CAPITAL LETTER ETA + '\u0398' # 0x48 -> GREEK CAPITAL LETTER THETA + '\u0399' # 0x49 -> GREEK CAPITAL LETTER IOTA + '[' # 0x4A -> LEFT SQUARE BRACKET + '.' # 0x4B -> FULL STOP + '<' # 0x4C -> LESS-THAN SIGN + '(' # 0x4D -> LEFT PARENTHESIS + '+' # 0x4E -> PLUS SIGN + '!' # 0x4F -> EXCLAMATION MARK + '&' # 0x50 -> AMPERSAND + '\u039a' # 0x51 -> GREEK CAPITAL LETTER KAPPA + '\u039b' # 0x52 -> GREEK CAPITAL LETTER LAMDA + '\u039c' # 0x53 -> GREEK CAPITAL LETTER MU + '\u039d' # 0x54 -> GREEK CAPITAL LETTER NU + '\u039e' # 0x55 -> GREEK CAPITAL LETTER XI + '\u039f' # 0x56 -> GREEK CAPITAL LETTER OMICRON + '\u03a0' # 0x57 -> GREEK CAPITAL LETTER PI + '\u03a1' # 0x58 -> GREEK CAPITAL LETTER RHO + '\u03a3' # 0x59 -> GREEK CAPITAL LETTER SIGMA + ']' # 0x5A -> RIGHT SQUARE BRACKET + '$' # 0x5B -> DOLLAR SIGN + '*' # 0x5C -> ASTERISK + ')' # 0x5D -> RIGHT PARENTHESIS + ';' # 0x5E -> SEMICOLON + '^' # 0x5F -> CIRCUMFLEX ACCENT + '-' # 0x60 -> HYPHEN-MINUS + '/' # 0x61 -> SOLIDUS + '\u03a4' # 0x62 -> GREEK CAPITAL LETTER TAU + '\u03a5' # 0x63 -> GREEK CAPITAL LETTER UPSILON + '\u03a6' # 0x64 -> GREEK CAPITAL LETTER PHI + '\u03a7' # 0x65 -> GREEK CAPITAL LETTER CHI + '\u03a8' # 0x66 -> GREEK CAPITAL LETTER PSI + '\u03a9' # 0x67 -> GREEK CAPITAL LETTER OMEGA + '\u03aa' # 0x68 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA + '\u03ab' # 0x69 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA + '|' # 0x6A -> VERTICAL LINE + ',' # 0x6B -> COMMA + '%' # 0x6C -> PERCENT SIGN + '_' # 0x6D -> LOW LINE + '>' # 0x6E -> GREATER-THAN SIGN + '?' # 0x6F -> QUESTION MARK + '\xa8' # 0x70 -> DIAERESIS + '\u0386' # 0x71 -> GREEK CAPITAL LETTER ALPHA WITH TONOS + '\u0388' # 0x72 -> GREEK CAPITAL LETTER EPSILON WITH TONOS + '\u0389' # 0x73 -> GREEK CAPITAL LETTER ETA WITH TONOS + '\xa0' # 0x74 -> NO-BREAK SPACE + '\u038a' # 0x75 -> GREEK CAPITAL LETTER IOTA WITH TONOS + '\u038c' # 0x76 -> GREEK CAPITAL LETTER OMICRON WITH TONOS + '\u038e' # 0x77 -> GREEK CAPITAL LETTER UPSILON WITH TONOS + '\u038f' # 0x78 -> GREEK CAPITAL LETTER OMEGA WITH TONOS + '`' # 0x79 -> GRAVE ACCENT + ':' # 0x7A -> COLON + '#' # 0x7B -> NUMBER SIGN + '@' # 0x7C -> COMMERCIAL AT + "'" # 0x7D -> APOSTROPHE + '=' # 0x7E -> EQUALS SIGN + '"' # 0x7F -> QUOTATION MARK + '\u0385' # 0x80 -> GREEK DIALYTIKA TONOS + 'a' # 0x81 -> LATIN SMALL LETTER A + 'b' # 0x82 -> LATIN SMALL LETTER B + 'c' # 0x83 -> LATIN SMALL LETTER C + 'd' # 0x84 -> LATIN SMALL LETTER D + 'e' # 0x85 -> LATIN SMALL LETTER E + 'f' # 0x86 -> LATIN SMALL LETTER F + 'g' # 0x87 -> LATIN SMALL LETTER G + 'h' # 0x88 -> LATIN SMALL LETTER H + 'i' # 0x89 -> LATIN SMALL LETTER I + '\u03b1' # 0x8A -> GREEK SMALL LETTER ALPHA + '\u03b2' # 0x8B -> GREEK SMALL LETTER BETA + '\u03b3' # 0x8C -> GREEK SMALL LETTER GAMMA + '\u03b4' # 0x8D -> GREEK SMALL LETTER DELTA + '\u03b5' # 0x8E -> GREEK SMALL LETTER EPSILON + '\u03b6' # 0x8F -> GREEK SMALL LETTER ZETA + '\xb0' # 0x90 -> DEGREE SIGN + 'j' # 0x91 -> LATIN SMALL LETTER J + 'k' # 0x92 -> LATIN SMALL LETTER K + 'l' # 0x93 -> LATIN SMALL LETTER L + 'm' # 0x94 -> LATIN SMALL LETTER M + 'n' # 0x95 -> LATIN SMALL LETTER N + 'o' # 0x96 -> LATIN SMALL LETTER O + 'p' # 0x97 -> LATIN SMALL LETTER P + 'q' # 0x98 -> LATIN SMALL LETTER Q + 'r' # 0x99 -> LATIN SMALL LETTER R + '\u03b7' # 0x9A -> GREEK SMALL LETTER ETA + '\u03b8' # 0x9B -> GREEK SMALL LETTER THETA + '\u03b9' # 0x9C -> GREEK SMALL LETTER IOTA + '\u03ba' # 0x9D -> GREEK SMALL LETTER KAPPA + '\u03bb' # 0x9E -> GREEK SMALL LETTER LAMDA + '\u03bc' # 0x9F -> GREEK SMALL LETTER MU + '\xb4' # 0xA0 -> ACUTE ACCENT + '~' # 0xA1 -> TILDE + 's' # 0xA2 -> LATIN SMALL LETTER S + 't' # 0xA3 -> LATIN SMALL LETTER T + 'u' # 0xA4 -> LATIN SMALL LETTER U + 'v' # 0xA5 -> LATIN SMALL LETTER V + 'w' # 0xA6 -> LATIN SMALL LETTER W + 'x' # 0xA7 -> LATIN SMALL LETTER X + 'y' # 0xA8 -> LATIN SMALL LETTER Y + 'z' # 0xA9 -> LATIN SMALL LETTER Z + '\u03bd' # 0xAA -> GREEK SMALL LETTER NU + '\u03be' # 0xAB -> GREEK SMALL LETTER XI + '\u03bf' # 0xAC -> GREEK SMALL LETTER OMICRON + '\u03c0' # 0xAD -> GREEK SMALL LETTER PI + '\u03c1' # 0xAE -> GREEK SMALL LETTER RHO + '\u03c3' # 0xAF -> GREEK SMALL LETTER SIGMA + '\xa3' # 0xB0 -> POUND SIGN + '\u03ac' # 0xB1 -> GREEK SMALL LETTER ALPHA WITH TONOS + '\u03ad' # 0xB2 -> GREEK SMALL LETTER EPSILON WITH TONOS + '\u03ae' # 0xB3 -> GREEK SMALL LETTER ETA WITH TONOS + '\u03ca' # 0xB4 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA + '\u03af' # 0xB5 -> GREEK SMALL LETTER IOTA WITH TONOS + '\u03cc' # 0xB6 -> GREEK SMALL LETTER OMICRON WITH TONOS + '\u03cd' # 0xB7 -> GREEK SMALL LETTER UPSILON WITH TONOS + '\u03cb' # 0xB8 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA + '\u03ce' # 0xB9 -> GREEK SMALL LETTER OMEGA WITH TONOS + '\u03c2' # 0xBA -> GREEK SMALL LETTER FINAL SIGMA + '\u03c4' # 0xBB -> GREEK SMALL LETTER TAU + '\u03c5' # 0xBC -> GREEK SMALL LETTER UPSILON + '\u03c6' # 0xBD -> GREEK SMALL LETTER PHI + '\u03c7' # 0xBE -> GREEK SMALL LETTER CHI + '\u03c8' # 0xBF -> GREEK SMALL LETTER PSI + '{' # 0xC0 -> LEFT CURLY BRACKET + 'A' # 0xC1 -> LATIN CAPITAL LETTER A + 'B' # 0xC2 -> LATIN CAPITAL LETTER B + 'C' # 0xC3 -> LATIN CAPITAL LETTER C + 'D' # 0xC4 -> LATIN CAPITAL LETTER D + 'E' # 0xC5 -> LATIN CAPITAL LETTER E + 'F' # 0xC6 -> LATIN CAPITAL LETTER F + 'G' # 0xC7 -> LATIN CAPITAL LETTER G + 'H' # 0xC8 -> LATIN CAPITAL LETTER H + 'I' # 0xC9 -> LATIN CAPITAL LETTER I + '\xad' # 0xCA -> SOFT HYPHEN + '\u03c9' # 0xCB -> GREEK SMALL LETTER OMEGA + '\u0390' # 0xCC -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS + '\u03b0' # 0xCD -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS + '\u2018' # 0xCE -> LEFT SINGLE QUOTATION MARK + '\u2015' # 0xCF -> HORIZONTAL BAR + '}' # 0xD0 -> RIGHT CURLY BRACKET + 'J' # 0xD1 -> LATIN CAPITAL LETTER J + 'K' # 0xD2 -> LATIN CAPITAL LETTER K + 'L' # 0xD3 -> LATIN CAPITAL LETTER L + 'M' # 0xD4 -> LATIN CAPITAL LETTER M + 'N' # 0xD5 -> LATIN CAPITAL LETTER N + 'O' # 0xD6 -> LATIN CAPITAL LETTER O + 'P' # 0xD7 -> LATIN CAPITAL LETTER P + 'Q' # 0xD8 -> LATIN CAPITAL LETTER Q + 'R' # 0xD9 -> LATIN CAPITAL LETTER R + '\xb1' # 0xDA -> PLUS-MINUS SIGN + '\xbd' # 0xDB -> VULGAR FRACTION ONE HALF + '\x1a' # 0xDC -> SUBSTITUTE + '\u0387' # 0xDD -> GREEK ANO TELEIA + '\u2019' # 0xDE -> RIGHT SINGLE QUOTATION MARK + '\xa6' # 0xDF -> BROKEN BAR + '\\' # 0xE0 -> REVERSE SOLIDUS + '\x1a' # 0xE1 -> SUBSTITUTE + 'S' # 0xE2 -> LATIN CAPITAL LETTER S + 'T' # 0xE3 -> LATIN CAPITAL LETTER T + 'U' # 0xE4 -> LATIN CAPITAL LETTER U + 'V' # 0xE5 -> LATIN CAPITAL LETTER V + 'W' # 0xE6 -> LATIN CAPITAL LETTER W + 'X' # 0xE7 -> LATIN CAPITAL LETTER X + 'Y' # 0xE8 -> LATIN CAPITAL LETTER Y + 'Z' # 0xE9 -> LATIN CAPITAL LETTER Z + '\xb2' # 0xEA -> SUPERSCRIPT TWO + '\xa7' # 0xEB -> SECTION SIGN + '\x1a' # 0xEC -> SUBSTITUTE + '\x1a' # 0xED -> SUBSTITUTE + '\xab' # 0xEE -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xEF -> NOT SIGN + '0' # 0xF0 -> DIGIT ZERO + '1' # 0xF1 -> DIGIT ONE + '2' # 0xF2 -> DIGIT TWO + '3' # 0xF3 -> DIGIT THREE + '4' # 0xF4 -> DIGIT FOUR + '5' # 0xF5 -> DIGIT FIVE + '6' # 0xF6 -> DIGIT SIX + '7' # 0xF7 -> DIGIT SEVEN + '8' # 0xF8 -> DIGIT EIGHT + '9' # 0xF9 -> DIGIT NINE + '\xb3' # 0xFA -> SUPERSCRIPT THREE + '\xa9' # 0xFB -> COPYRIGHT SIGN + '\x1a' # 0xFC -> SUBSTITUTE + '\x1a' # 0xFD -> SUBSTITUTE + '\xbb' # 0xFE -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\x9f' # 0xFF -> CONTROL ) ### Encoding table diff --git a/Lib/encodings/idna.py b/Lib/encodings/idna.py index ea90d67..d9322a1 100644 --- a/Lib/encodings/idna.py +++ b/Lib/encodings/idna.py @@ -4,11 +4,11 @@ import stringprep, re, codecs from unicodedata import ucd_3_2_0 as unicodedata # IDNA section 3.1 -dots = re.compile(u"[\u002E\u3002\uFF0E\uFF61]") +dots = re.compile("[\u002E\u3002\uFF0E\uFF61]") # IDNA section 5 ace_prefix = "xn--" -uace_prefix = unicode(ace_prefix, "ascii") +uace_prefix = str(ace_prefix, "ascii") # This assumes query strings, so AllowUnassigned is true def nameprep(label): @@ -19,7 +19,7 @@ def nameprep(label): # Map to nothing continue newlabel.append(stringprep.map_table_b2(c)) - label = u"".join(newlabel) + label = "".join(newlabel) # Normalize label = unicodedata.normalize("NFKC", label) @@ -122,7 +122,7 @@ def ToUnicode(label): raise UnicodeError("Invalid character in IDN label") # Step 3: Check for ACE prefix if not label.startswith(ace_prefix): - return unicode(label, "ascii") + return str(label, "ascii") # Step 4: Remove ACE prefix label1 = label[len(ace_prefix):] @@ -171,28 +171,28 @@ class Codec(codecs.Codec): raise UnicodeError("Unsupported error handling "+errors) if not input: - return u"", 0 + return "", 0 # IDNA allows decoding to operate on Unicode strings, too. - if isinstance(input, unicode): + if isinstance(input, str): labels = dots.split(input) else: # Must be ASCII string input = str(input) - unicode(input, "ascii") + str(input, "ascii") labels = input.split(".") if labels and len(labels[-1]) == 0: - trailing_dot = u'.' + trailing_dot = '.' del labels[-1] else: - trailing_dot = u'' + trailing_dot = '' result = [] for label in labels: result.append(ToUnicode(label)) - return u".".join(result)+trailing_dot, len(input) + return ".".join(result)+trailing_dot, len(input) class IncrementalEncoder(codecs.BufferedIncrementalEncoder): def _buffer_encode(self, input, errors, final): @@ -204,7 +204,7 @@ class IncrementalEncoder(codecs.BufferedIncrementalEncoder): return ("", 0) labels = dots.split(input) - trailing_dot = u'' + trailing_dot = '' if labels: if not labels[-1]: trailing_dot = '.' @@ -234,27 +234,27 @@ class IncrementalDecoder(codecs.BufferedIncrementalDecoder): raise UnicodeError("Unsupported error handling "+errors) if not input: - return (u"", 0) + return ("", 0) # IDNA allows decoding to operate on Unicode strings, too. - if isinstance(input, unicode): + if isinstance(input, str): labels = dots.split(input) else: # Must be ASCII string input = str(input) - unicode(input, "ascii") + str(input, "ascii") labels = input.split(".") - trailing_dot = u'' + trailing_dot = '' if labels: if not labels[-1]: - trailing_dot = u'.' + trailing_dot = '.' del labels[-1] elif not final: # Keep potentially unfinished label until the next call del labels[-1] if labels: - trailing_dot = u'.' + trailing_dot = '.' result = [] size = 0 @@ -264,7 +264,7 @@ class IncrementalDecoder(codecs.BufferedIncrementalDecoder): size += 1 size += len(label) - result = u".".join(result) + trailing_dot + result = ".".join(result) + trailing_dot size += len(trailing_dot) return (result, size) diff --git a/Lib/encodings/iso8859_1.py b/Lib/encodings/iso8859_1.py index 71bc13f..8cfc01f 100644 --- a/Lib/encodings/iso8859_1.py +++ b/Lib/encodings/iso8859_1.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH (Icelandic) - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xDE -> LATIN CAPITAL LETTER THORN (Icelandic) - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S (German) - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf0' # 0xF0 -> LATIN SMALL LETTER ETH (Icelandic) - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0xFE -> LATIN SMALL LETTER THORN (Icelandic) - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH (Icelandic) + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xDE -> LATIN CAPITAL LETTER THORN (Icelandic) + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S (German) + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf0' # 0xF0 -> LATIN SMALL LETTER ETH (Icelandic) + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0xFE -> LATIN SMALL LETTER THORN (Icelandic) + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/iso8859_10.py b/Lib/encodings/iso8859_10.py index 757e5c5..b4fb041 100644 --- a/Lib/encodings/iso8859_10.py +++ b/Lib/encodings/iso8859_10.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u0112' # 0xA2 -> LATIN CAPITAL LETTER E WITH MACRON - u'\u0122' # 0xA3 -> LATIN CAPITAL LETTER G WITH CEDILLA - u'\u012a' # 0xA4 -> LATIN CAPITAL LETTER I WITH MACRON - u'\u0128' # 0xA5 -> LATIN CAPITAL LETTER I WITH TILDE - u'\u0136' # 0xA6 -> LATIN CAPITAL LETTER K WITH CEDILLA - u'\xa7' # 0xA7 -> SECTION SIGN - u'\u013b' # 0xA8 -> LATIN CAPITAL LETTER L WITH CEDILLA - u'\u0110' # 0xA9 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u0160' # 0xAA -> LATIN CAPITAL LETTER S WITH CARON - u'\u0166' # 0xAB -> LATIN CAPITAL LETTER T WITH STROKE - u'\u017d' # 0xAC -> LATIN CAPITAL LETTER Z WITH CARON - u'\xad' # 0xAD -> SOFT HYPHEN - u'\u016a' # 0xAE -> LATIN CAPITAL LETTER U WITH MACRON - u'\u014a' # 0xAF -> LATIN CAPITAL LETTER ENG - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK - u'\u0113' # 0xB2 -> LATIN SMALL LETTER E WITH MACRON - u'\u0123' # 0xB3 -> LATIN SMALL LETTER G WITH CEDILLA - u'\u012b' # 0xB4 -> LATIN SMALL LETTER I WITH MACRON - u'\u0129' # 0xB5 -> LATIN SMALL LETTER I WITH TILDE - u'\u0137' # 0xB6 -> LATIN SMALL LETTER K WITH CEDILLA - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\u013c' # 0xB8 -> LATIN SMALL LETTER L WITH CEDILLA - u'\u0111' # 0xB9 -> LATIN SMALL LETTER D WITH STROKE - u'\u0161' # 0xBA -> LATIN SMALL LETTER S WITH CARON - u'\u0167' # 0xBB -> LATIN SMALL LETTER T WITH STROKE - u'\u017e' # 0xBC -> LATIN SMALL LETTER Z WITH CARON - u'\u2015' # 0xBD -> HORIZONTAL BAR - u'\u016b' # 0xBE -> LATIN SMALL LETTER U WITH MACRON - u'\u014b' # 0xBF -> LATIN SMALL LETTER ENG - u'\u0100' # 0xC0 -> LATIN CAPITAL LETTER A WITH MACRON - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\u012e' # 0xC7 -> LATIN CAPITAL LETTER I WITH OGONEK - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u0116' # 0xCC -> LATIN CAPITAL LETTER E WITH DOT ABOVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH (Icelandic) - u'\u0145' # 0xD1 -> LATIN CAPITAL LETTER N WITH CEDILLA - u'\u014c' # 0xD2 -> LATIN CAPITAL LETTER O WITH MACRON - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\u0168' # 0xD7 -> LATIN CAPITAL LETTER U WITH TILDE - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\u0172' # 0xD9 -> LATIN CAPITAL LETTER U WITH OGONEK - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xDE -> LATIN CAPITAL LETTER THORN (Icelandic) - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S (German) - u'\u0101' # 0xE0 -> LATIN SMALL LETTER A WITH MACRON - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\u012f' # 0xE7 -> LATIN SMALL LETTER I WITH OGONEK - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u0117' # 0xEC -> LATIN SMALL LETTER E WITH DOT ABOVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf0' # 0xF0 -> LATIN SMALL LETTER ETH (Icelandic) - u'\u0146' # 0xF1 -> LATIN SMALL LETTER N WITH CEDILLA - u'\u014d' # 0xF2 -> LATIN SMALL LETTER O WITH MACRON - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\u0169' # 0xF7 -> LATIN SMALL LETTER U WITH TILDE - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\u0173' # 0xF9 -> LATIN SMALL LETTER U WITH OGONEK - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0xFE -> LATIN SMALL LETTER THORN (Icelandic) - u'\u0138' # 0xFF -> LATIN SMALL LETTER KRA + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u0112' # 0xA2 -> LATIN CAPITAL LETTER E WITH MACRON + '\u0122' # 0xA3 -> LATIN CAPITAL LETTER G WITH CEDILLA + '\u012a' # 0xA4 -> LATIN CAPITAL LETTER I WITH MACRON + '\u0128' # 0xA5 -> LATIN CAPITAL LETTER I WITH TILDE + '\u0136' # 0xA6 -> LATIN CAPITAL LETTER K WITH CEDILLA + '\xa7' # 0xA7 -> SECTION SIGN + '\u013b' # 0xA8 -> LATIN CAPITAL LETTER L WITH CEDILLA + '\u0110' # 0xA9 -> LATIN CAPITAL LETTER D WITH STROKE + '\u0160' # 0xAA -> LATIN CAPITAL LETTER S WITH CARON + '\u0166' # 0xAB -> LATIN CAPITAL LETTER T WITH STROKE + '\u017d' # 0xAC -> LATIN CAPITAL LETTER Z WITH CARON + '\xad' # 0xAD -> SOFT HYPHEN + '\u016a' # 0xAE -> LATIN CAPITAL LETTER U WITH MACRON + '\u014a' # 0xAF -> LATIN CAPITAL LETTER ENG + '\xb0' # 0xB0 -> DEGREE SIGN + '\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK + '\u0113' # 0xB2 -> LATIN SMALL LETTER E WITH MACRON + '\u0123' # 0xB3 -> LATIN SMALL LETTER G WITH CEDILLA + '\u012b' # 0xB4 -> LATIN SMALL LETTER I WITH MACRON + '\u0129' # 0xB5 -> LATIN SMALL LETTER I WITH TILDE + '\u0137' # 0xB6 -> LATIN SMALL LETTER K WITH CEDILLA + '\xb7' # 0xB7 -> MIDDLE DOT + '\u013c' # 0xB8 -> LATIN SMALL LETTER L WITH CEDILLA + '\u0111' # 0xB9 -> LATIN SMALL LETTER D WITH STROKE + '\u0161' # 0xBA -> LATIN SMALL LETTER S WITH CARON + '\u0167' # 0xBB -> LATIN SMALL LETTER T WITH STROKE + '\u017e' # 0xBC -> LATIN SMALL LETTER Z WITH CARON + '\u2015' # 0xBD -> HORIZONTAL BAR + '\u016b' # 0xBE -> LATIN SMALL LETTER U WITH MACRON + '\u014b' # 0xBF -> LATIN SMALL LETTER ENG + '\u0100' # 0xC0 -> LATIN CAPITAL LETTER A WITH MACRON + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\u012e' # 0xC7 -> LATIN CAPITAL LETTER I WITH OGONEK + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u0116' # 0xCC -> LATIN CAPITAL LETTER E WITH DOT ABOVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH (Icelandic) + '\u0145' # 0xD1 -> LATIN CAPITAL LETTER N WITH CEDILLA + '\u014c' # 0xD2 -> LATIN CAPITAL LETTER O WITH MACRON + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\u0168' # 0xD7 -> LATIN CAPITAL LETTER U WITH TILDE + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\u0172' # 0xD9 -> LATIN CAPITAL LETTER U WITH OGONEK + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xDE -> LATIN CAPITAL LETTER THORN (Icelandic) + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S (German) + '\u0101' # 0xE0 -> LATIN SMALL LETTER A WITH MACRON + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\u012f' # 0xE7 -> LATIN SMALL LETTER I WITH OGONEK + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\u0117' # 0xEC -> LATIN SMALL LETTER E WITH DOT ABOVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf0' # 0xF0 -> LATIN SMALL LETTER ETH (Icelandic) + '\u0146' # 0xF1 -> LATIN SMALL LETTER N WITH CEDILLA + '\u014d' # 0xF2 -> LATIN SMALL LETTER O WITH MACRON + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\u0169' # 0xF7 -> LATIN SMALL LETTER U WITH TILDE + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\u0173' # 0xF9 -> LATIN SMALL LETTER U WITH OGONEK + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0xFE -> LATIN SMALL LETTER THORN (Icelandic) + '\u0138' # 0xFF -> LATIN SMALL LETTER KRA ) ### Encoding table diff --git a/Lib/encodings/iso8859_11.py b/Lib/encodings/iso8859_11.py index 27ece8d..c7258ec 100644 --- a/Lib/encodings/iso8859_11.py +++ b/Lib/encodings/iso8859_11.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0e01' # 0xA1 -> THAI CHARACTER KO KAI - u'\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI - u'\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT - u'\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI - u'\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON - u'\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG - u'\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU - u'\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN - u'\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING - u'\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG - u'\u0e0b' # 0xAB -> THAI CHARACTER SO SO - u'\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE - u'\u0e0d' # 0xAD -> THAI CHARACTER YO YING - u'\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA - u'\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK - u'\u0e10' # 0xB0 -> THAI CHARACTER THO THAN - u'\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO - u'\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO - u'\u0e13' # 0xB3 -> THAI CHARACTER NO NEN - u'\u0e14' # 0xB4 -> THAI CHARACTER DO DEK - u'\u0e15' # 0xB5 -> THAI CHARACTER TO TAO - u'\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG - u'\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN - u'\u0e18' # 0xB8 -> THAI CHARACTER THO THONG - u'\u0e19' # 0xB9 -> THAI CHARACTER NO NU - u'\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI - u'\u0e1b' # 0xBB -> THAI CHARACTER PO PLA - u'\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG - u'\u0e1d' # 0xBD -> THAI CHARACTER FO FA - u'\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN - u'\u0e1f' # 0xBF -> THAI CHARACTER FO FAN - u'\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO - u'\u0e21' # 0xC1 -> THAI CHARACTER MO MA - u'\u0e22' # 0xC2 -> THAI CHARACTER YO YAK - u'\u0e23' # 0xC3 -> THAI CHARACTER RO RUA - u'\u0e24' # 0xC4 -> THAI CHARACTER RU - u'\u0e25' # 0xC5 -> THAI CHARACTER LO LING - u'\u0e26' # 0xC6 -> THAI CHARACTER LU - u'\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN - u'\u0e28' # 0xC8 -> THAI CHARACTER SO SALA - u'\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI - u'\u0e2a' # 0xCA -> THAI CHARACTER SO SUA - u'\u0e2b' # 0xCB -> THAI CHARACTER HO HIP - u'\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA - u'\u0e2d' # 0xCD -> THAI CHARACTER O ANG - u'\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK - u'\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI - u'\u0e30' # 0xD0 -> THAI CHARACTER SARA A - u'\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT - u'\u0e32' # 0xD2 -> THAI CHARACTER SARA AA - u'\u0e33' # 0xD3 -> THAI CHARACTER SARA AM - u'\u0e34' # 0xD4 -> THAI CHARACTER SARA I - u'\u0e35' # 0xD5 -> THAI CHARACTER SARA II - u'\u0e36' # 0xD6 -> THAI CHARACTER SARA UE - u'\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE - u'\u0e38' # 0xD8 -> THAI CHARACTER SARA U - u'\u0e39' # 0xD9 -> THAI CHARACTER SARA UU - u'\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT - u'\u0e40' # 0xE0 -> THAI CHARACTER SARA E - u'\u0e41' # 0xE1 -> THAI CHARACTER SARA AE - u'\u0e42' # 0xE2 -> THAI CHARACTER SARA O - u'\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN - u'\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI - u'\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO - u'\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK - u'\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU - u'\u0e48' # 0xE8 -> THAI CHARACTER MAI EK - u'\u0e49' # 0xE9 -> THAI CHARACTER MAI THO - u'\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI - u'\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA - u'\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT - u'\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT - u'\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN - u'\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN - u'\u0e50' # 0xF0 -> THAI DIGIT ZERO - u'\u0e51' # 0xF1 -> THAI DIGIT ONE - u'\u0e52' # 0xF2 -> THAI DIGIT TWO - u'\u0e53' # 0xF3 -> THAI DIGIT THREE - u'\u0e54' # 0xF4 -> THAI DIGIT FOUR - u'\u0e55' # 0xF5 -> THAI DIGIT FIVE - u'\u0e56' # 0xF6 -> THAI DIGIT SIX - u'\u0e57' # 0xF7 -> THAI DIGIT SEVEN - u'\u0e58' # 0xF8 -> THAI DIGIT EIGHT - u'\u0e59' # 0xF9 -> THAI DIGIT NINE - u'\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU - u'\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0e01' # 0xA1 -> THAI CHARACTER KO KAI + '\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI + '\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT + '\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI + '\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON + '\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG + '\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU + '\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN + '\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING + '\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG + '\u0e0b' # 0xAB -> THAI CHARACTER SO SO + '\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE + '\u0e0d' # 0xAD -> THAI CHARACTER YO YING + '\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA + '\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK + '\u0e10' # 0xB0 -> THAI CHARACTER THO THAN + '\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO + '\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO + '\u0e13' # 0xB3 -> THAI CHARACTER NO NEN + '\u0e14' # 0xB4 -> THAI CHARACTER DO DEK + '\u0e15' # 0xB5 -> THAI CHARACTER TO TAO + '\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG + '\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN + '\u0e18' # 0xB8 -> THAI CHARACTER THO THONG + '\u0e19' # 0xB9 -> THAI CHARACTER NO NU + '\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI + '\u0e1b' # 0xBB -> THAI CHARACTER PO PLA + '\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG + '\u0e1d' # 0xBD -> THAI CHARACTER FO FA + '\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN + '\u0e1f' # 0xBF -> THAI CHARACTER FO FAN + '\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO + '\u0e21' # 0xC1 -> THAI CHARACTER MO MA + '\u0e22' # 0xC2 -> THAI CHARACTER YO YAK + '\u0e23' # 0xC3 -> THAI CHARACTER RO RUA + '\u0e24' # 0xC4 -> THAI CHARACTER RU + '\u0e25' # 0xC5 -> THAI CHARACTER LO LING + '\u0e26' # 0xC6 -> THAI CHARACTER LU + '\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN + '\u0e28' # 0xC8 -> THAI CHARACTER SO SALA + '\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI + '\u0e2a' # 0xCA -> THAI CHARACTER SO SUA + '\u0e2b' # 0xCB -> THAI CHARACTER HO HIP + '\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA + '\u0e2d' # 0xCD -> THAI CHARACTER O ANG + '\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK + '\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI + '\u0e30' # 0xD0 -> THAI CHARACTER SARA A + '\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT + '\u0e32' # 0xD2 -> THAI CHARACTER SARA AA + '\u0e33' # 0xD3 -> THAI CHARACTER SARA AM + '\u0e34' # 0xD4 -> THAI CHARACTER SARA I + '\u0e35' # 0xD5 -> THAI CHARACTER SARA II + '\u0e36' # 0xD6 -> THAI CHARACTER SARA UE + '\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE + '\u0e38' # 0xD8 -> THAI CHARACTER SARA U + '\u0e39' # 0xD9 -> THAI CHARACTER SARA UU + '\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT + '\u0e40' # 0xE0 -> THAI CHARACTER SARA E + '\u0e41' # 0xE1 -> THAI CHARACTER SARA AE + '\u0e42' # 0xE2 -> THAI CHARACTER SARA O + '\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN + '\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI + '\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO + '\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK + '\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU + '\u0e48' # 0xE8 -> THAI CHARACTER MAI EK + '\u0e49' # 0xE9 -> THAI CHARACTER MAI THO + '\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI + '\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA + '\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT + '\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT + '\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN + '\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN + '\u0e50' # 0xF0 -> THAI DIGIT ZERO + '\u0e51' # 0xF1 -> THAI DIGIT ONE + '\u0e52' # 0xF2 -> THAI DIGIT TWO + '\u0e53' # 0xF3 -> THAI DIGIT THREE + '\u0e54' # 0xF4 -> THAI DIGIT FOUR + '\u0e55' # 0xF5 -> THAI DIGIT FIVE + '\u0e56' # 0xF6 -> THAI DIGIT SIX + '\u0e57' # 0xF7 -> THAI DIGIT SEVEN + '\u0e58' # 0xF8 -> THAI DIGIT EIGHT + '\u0e59' # 0xF9 -> THAI DIGIT NINE + '\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU + '\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' ) ### Encoding table diff --git a/Lib/encodings/iso8859_13.py b/Lib/encodings/iso8859_13.py index 71adb5c..6f8eab2 100644 --- a/Lib/encodings/iso8859_13.py +++ b/Lib/encodings/iso8859_13.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u201d' # 0xA1 -> RIGHT DOUBLE QUOTATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\u201e' # 0xA5 -> DOUBLE LOW-9 QUOTATION MARK - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xd8' # 0xA8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u0156' # 0xAA -> LATIN CAPITAL LETTER R WITH CEDILLA - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xc6' # 0xAF -> LATIN CAPITAL LETTER AE - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\u201c' # 0xB4 -> LEFT DOUBLE QUOTATION MARK - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xf8' # 0xB8 -> LATIN SMALL LETTER O WITH STROKE - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\u0157' # 0xBA -> LATIN SMALL LETTER R WITH CEDILLA - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xe6' # 0xBF -> LATIN SMALL LETTER AE - u'\u0104' # 0xC0 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u012e' # 0xC1 -> LATIN CAPITAL LETTER I WITH OGONEK - u'\u0100' # 0xC2 -> LATIN CAPITAL LETTER A WITH MACRON - u'\u0106' # 0xC3 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\u0118' # 0xC6 -> LATIN CAPITAL LETTER E WITH OGONEK - u'\u0112' # 0xC7 -> LATIN CAPITAL LETTER E WITH MACRON - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0179' # 0xCA -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\u0116' # 0xCB -> LATIN CAPITAL LETTER E WITH DOT ABOVE - u'\u0122' # 0xCC -> LATIN CAPITAL LETTER G WITH CEDILLA - u'\u0136' # 0xCD -> LATIN CAPITAL LETTER K WITH CEDILLA - u'\u012a' # 0xCE -> LATIN CAPITAL LETTER I WITH MACRON - u'\u013b' # 0xCF -> LATIN CAPITAL LETTER L WITH CEDILLA - u'\u0160' # 0xD0 -> LATIN CAPITAL LETTER S WITH CARON - u'\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\u0145' # 0xD2 -> LATIN CAPITAL LETTER N WITH CEDILLA - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\u014c' # 0xD4 -> LATIN CAPITAL LETTER O WITH MACRON - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\u0172' # 0xD8 -> LATIN CAPITAL LETTER U WITH OGONEK - u'\u0141' # 0xD9 -> LATIN CAPITAL LETTER L WITH STROKE - u'\u015a' # 0xDA -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u016a' # 0xDB -> LATIN CAPITAL LETTER U WITH MACRON - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u017b' # 0xDD -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\u017d' # 0xDE -> LATIN CAPITAL LETTER Z WITH CARON - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S (German) - u'\u0105' # 0xE0 -> LATIN SMALL LETTER A WITH OGONEK - u'\u012f' # 0xE1 -> LATIN SMALL LETTER I WITH OGONEK - u'\u0101' # 0xE2 -> LATIN SMALL LETTER A WITH MACRON - u'\u0107' # 0xE3 -> LATIN SMALL LETTER C WITH ACUTE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\u0119' # 0xE6 -> LATIN SMALL LETTER E WITH OGONEK - u'\u0113' # 0xE7 -> LATIN SMALL LETTER E WITH MACRON - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\u017a' # 0xEA -> LATIN SMALL LETTER Z WITH ACUTE - u'\u0117' # 0xEB -> LATIN SMALL LETTER E WITH DOT ABOVE - u'\u0123' # 0xEC -> LATIN SMALL LETTER G WITH CEDILLA - u'\u0137' # 0xED -> LATIN SMALL LETTER K WITH CEDILLA - u'\u012b' # 0xEE -> LATIN SMALL LETTER I WITH MACRON - u'\u013c' # 0xEF -> LATIN SMALL LETTER L WITH CEDILLA - u'\u0161' # 0xF0 -> LATIN SMALL LETTER S WITH CARON - u'\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0146' # 0xF2 -> LATIN SMALL LETTER N WITH CEDILLA - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\u014d' # 0xF4 -> LATIN SMALL LETTER O WITH MACRON - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\u0173' # 0xF8 -> LATIN SMALL LETTER U WITH OGONEK - u'\u0142' # 0xF9 -> LATIN SMALL LETTER L WITH STROKE - u'\u015b' # 0xFA -> LATIN SMALL LETTER S WITH ACUTE - u'\u016b' # 0xFB -> LATIN SMALL LETTER U WITH MACRON - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u017c' # 0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u017e' # 0xFE -> LATIN SMALL LETTER Z WITH CARON - u'\u2019' # 0xFF -> RIGHT SINGLE QUOTATION MARK + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u201d' # 0xA1 -> RIGHT DOUBLE QUOTATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\u201e' # 0xA5 -> DOUBLE LOW-9 QUOTATION MARK + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xd8' # 0xA8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u0156' # 0xAA -> LATIN CAPITAL LETTER R WITH CEDILLA + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xc6' # 0xAF -> LATIN CAPITAL LETTER AE + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\u201c' # 0xB4 -> LEFT DOUBLE QUOTATION MARK + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xf8' # 0xB8 -> LATIN SMALL LETTER O WITH STROKE + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\u0157' # 0xBA -> LATIN SMALL LETTER R WITH CEDILLA + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xe6' # 0xBF -> LATIN SMALL LETTER AE + '\u0104' # 0xC0 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u012e' # 0xC1 -> LATIN CAPITAL LETTER I WITH OGONEK + '\u0100' # 0xC2 -> LATIN CAPITAL LETTER A WITH MACRON + '\u0106' # 0xC3 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\u0118' # 0xC6 -> LATIN CAPITAL LETTER E WITH OGONEK + '\u0112' # 0xC7 -> LATIN CAPITAL LETTER E WITH MACRON + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0179' # 0xCA -> LATIN CAPITAL LETTER Z WITH ACUTE + '\u0116' # 0xCB -> LATIN CAPITAL LETTER E WITH DOT ABOVE + '\u0122' # 0xCC -> LATIN CAPITAL LETTER G WITH CEDILLA + '\u0136' # 0xCD -> LATIN CAPITAL LETTER K WITH CEDILLA + '\u012a' # 0xCE -> LATIN CAPITAL LETTER I WITH MACRON + '\u013b' # 0xCF -> LATIN CAPITAL LETTER L WITH CEDILLA + '\u0160' # 0xD0 -> LATIN CAPITAL LETTER S WITH CARON + '\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE + '\u0145' # 0xD2 -> LATIN CAPITAL LETTER N WITH CEDILLA + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\u014c' # 0xD4 -> LATIN CAPITAL LETTER O WITH MACRON + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\u0172' # 0xD8 -> LATIN CAPITAL LETTER U WITH OGONEK + '\u0141' # 0xD9 -> LATIN CAPITAL LETTER L WITH STROKE + '\u015a' # 0xDA -> LATIN CAPITAL LETTER S WITH ACUTE + '\u016a' # 0xDB -> LATIN CAPITAL LETTER U WITH MACRON + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u017b' # 0xDD -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\u017d' # 0xDE -> LATIN CAPITAL LETTER Z WITH CARON + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S (German) + '\u0105' # 0xE0 -> LATIN SMALL LETTER A WITH OGONEK + '\u012f' # 0xE1 -> LATIN SMALL LETTER I WITH OGONEK + '\u0101' # 0xE2 -> LATIN SMALL LETTER A WITH MACRON + '\u0107' # 0xE3 -> LATIN SMALL LETTER C WITH ACUTE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\u0119' # 0xE6 -> LATIN SMALL LETTER E WITH OGONEK + '\u0113' # 0xE7 -> LATIN SMALL LETTER E WITH MACRON + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\u017a' # 0xEA -> LATIN SMALL LETTER Z WITH ACUTE + '\u0117' # 0xEB -> LATIN SMALL LETTER E WITH DOT ABOVE + '\u0123' # 0xEC -> LATIN SMALL LETTER G WITH CEDILLA + '\u0137' # 0xED -> LATIN SMALL LETTER K WITH CEDILLA + '\u012b' # 0xEE -> LATIN SMALL LETTER I WITH MACRON + '\u013c' # 0xEF -> LATIN SMALL LETTER L WITH CEDILLA + '\u0161' # 0xF0 -> LATIN SMALL LETTER S WITH CARON + '\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE + '\u0146' # 0xF2 -> LATIN SMALL LETTER N WITH CEDILLA + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\u014d' # 0xF4 -> LATIN SMALL LETTER O WITH MACRON + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\u0173' # 0xF8 -> LATIN SMALL LETTER U WITH OGONEK + '\u0142' # 0xF9 -> LATIN SMALL LETTER L WITH STROKE + '\u015b' # 0xFA -> LATIN SMALL LETTER S WITH ACUTE + '\u016b' # 0xFB -> LATIN SMALL LETTER U WITH MACRON + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u017c' # 0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u017e' # 0xFE -> LATIN SMALL LETTER Z WITH CARON + '\u2019' # 0xFF -> RIGHT SINGLE QUOTATION MARK ) ### Encoding table diff --git a/Lib/encodings/iso8859_14.py b/Lib/encodings/iso8859_14.py index 56843d5..7568d4e 100644 --- a/Lib/encodings/iso8859_14.py +++ b/Lib/encodings/iso8859_14.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u1e02' # 0xA1 -> LATIN CAPITAL LETTER B WITH DOT ABOVE - u'\u1e03' # 0xA2 -> LATIN SMALL LETTER B WITH DOT ABOVE - u'\xa3' # 0xA3 -> POUND SIGN - u'\u010a' # 0xA4 -> LATIN CAPITAL LETTER C WITH DOT ABOVE - u'\u010b' # 0xA5 -> LATIN SMALL LETTER C WITH DOT ABOVE - u'\u1e0a' # 0xA6 -> LATIN CAPITAL LETTER D WITH DOT ABOVE - u'\xa7' # 0xA7 -> SECTION SIGN - u'\u1e80' # 0xA8 -> LATIN CAPITAL LETTER W WITH GRAVE - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u1e82' # 0xAA -> LATIN CAPITAL LETTER W WITH ACUTE - u'\u1e0b' # 0xAB -> LATIN SMALL LETTER D WITH DOT ABOVE - u'\u1ef2' # 0xAC -> LATIN CAPITAL LETTER Y WITH GRAVE - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\u0178' # 0xAF -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\u1e1e' # 0xB0 -> LATIN CAPITAL LETTER F WITH DOT ABOVE - u'\u1e1f' # 0xB1 -> LATIN SMALL LETTER F WITH DOT ABOVE - u'\u0120' # 0xB2 -> LATIN CAPITAL LETTER G WITH DOT ABOVE - u'\u0121' # 0xB3 -> LATIN SMALL LETTER G WITH DOT ABOVE - u'\u1e40' # 0xB4 -> LATIN CAPITAL LETTER M WITH DOT ABOVE - u'\u1e41' # 0xB5 -> LATIN SMALL LETTER M WITH DOT ABOVE - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\u1e56' # 0xB7 -> LATIN CAPITAL LETTER P WITH DOT ABOVE - u'\u1e81' # 0xB8 -> LATIN SMALL LETTER W WITH GRAVE - u'\u1e57' # 0xB9 -> LATIN SMALL LETTER P WITH DOT ABOVE - u'\u1e83' # 0xBA -> LATIN SMALL LETTER W WITH ACUTE - u'\u1e60' # 0xBB -> LATIN CAPITAL LETTER S WITH DOT ABOVE - u'\u1ef3' # 0xBC -> LATIN SMALL LETTER Y WITH GRAVE - u'\u1e84' # 0xBD -> LATIN CAPITAL LETTER W WITH DIAERESIS - u'\u1e85' # 0xBE -> LATIN SMALL LETTER W WITH DIAERESIS - u'\u1e61' # 0xBF -> LATIN SMALL LETTER S WITH DOT ABOVE - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u0174' # 0xD0 -> LATIN CAPITAL LETTER W WITH CIRCUMFLEX - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\u1e6a' # 0xD7 -> LATIN CAPITAL LETTER T WITH DOT ABOVE - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\u0176' # 0xDE -> LATIN CAPITAL LETTER Y WITH CIRCUMFLEX - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u0175' # 0xF0 -> LATIN SMALL LETTER W WITH CIRCUMFLEX - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\u1e6b' # 0xF7 -> LATIN SMALL LETTER T WITH DOT ABOVE - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\u0177' # 0xFE -> LATIN SMALL LETTER Y WITH CIRCUMFLEX - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u1e02' # 0xA1 -> LATIN CAPITAL LETTER B WITH DOT ABOVE + '\u1e03' # 0xA2 -> LATIN SMALL LETTER B WITH DOT ABOVE + '\xa3' # 0xA3 -> POUND SIGN + '\u010a' # 0xA4 -> LATIN CAPITAL LETTER C WITH DOT ABOVE + '\u010b' # 0xA5 -> LATIN SMALL LETTER C WITH DOT ABOVE + '\u1e0a' # 0xA6 -> LATIN CAPITAL LETTER D WITH DOT ABOVE + '\xa7' # 0xA7 -> SECTION SIGN + '\u1e80' # 0xA8 -> LATIN CAPITAL LETTER W WITH GRAVE + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u1e82' # 0xAA -> LATIN CAPITAL LETTER W WITH ACUTE + '\u1e0b' # 0xAB -> LATIN SMALL LETTER D WITH DOT ABOVE + '\u1ef2' # 0xAC -> LATIN CAPITAL LETTER Y WITH GRAVE + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\u0178' # 0xAF -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\u1e1e' # 0xB0 -> LATIN CAPITAL LETTER F WITH DOT ABOVE + '\u1e1f' # 0xB1 -> LATIN SMALL LETTER F WITH DOT ABOVE + '\u0120' # 0xB2 -> LATIN CAPITAL LETTER G WITH DOT ABOVE + '\u0121' # 0xB3 -> LATIN SMALL LETTER G WITH DOT ABOVE + '\u1e40' # 0xB4 -> LATIN CAPITAL LETTER M WITH DOT ABOVE + '\u1e41' # 0xB5 -> LATIN SMALL LETTER M WITH DOT ABOVE + '\xb6' # 0xB6 -> PILCROW SIGN + '\u1e56' # 0xB7 -> LATIN CAPITAL LETTER P WITH DOT ABOVE + '\u1e81' # 0xB8 -> LATIN SMALL LETTER W WITH GRAVE + '\u1e57' # 0xB9 -> LATIN SMALL LETTER P WITH DOT ABOVE + '\u1e83' # 0xBA -> LATIN SMALL LETTER W WITH ACUTE + '\u1e60' # 0xBB -> LATIN CAPITAL LETTER S WITH DOT ABOVE + '\u1ef3' # 0xBC -> LATIN SMALL LETTER Y WITH GRAVE + '\u1e84' # 0xBD -> LATIN CAPITAL LETTER W WITH DIAERESIS + '\u1e85' # 0xBE -> LATIN SMALL LETTER W WITH DIAERESIS + '\u1e61' # 0xBF -> LATIN SMALL LETTER S WITH DOT ABOVE + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u0174' # 0xD0 -> LATIN CAPITAL LETTER W WITH CIRCUMFLEX + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\u1e6a' # 0xD7 -> LATIN CAPITAL LETTER T WITH DOT ABOVE + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\u0176' # 0xDE -> LATIN CAPITAL LETTER Y WITH CIRCUMFLEX + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\u0175' # 0xF0 -> LATIN SMALL LETTER W WITH CIRCUMFLEX + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\u1e6b' # 0xF7 -> LATIN SMALL LETTER T WITH DOT ABOVE + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\u0177' # 0xFE -> LATIN SMALL LETTER Y WITH CIRCUMFLEX + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/iso8859_15.py b/Lib/encodings/iso8859_15.py index 13b140c..43bdecd 100644 --- a/Lib/encodings/iso8859_15.py +++ b/Lib/encodings/iso8859_15.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\u20ac' # 0xA4 -> EURO SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\u0160' # 0xA6 -> LATIN CAPITAL LETTER S WITH CARON - u'\xa7' # 0xA7 -> SECTION SIGN - u'\u0161' # 0xA8 -> LATIN SMALL LETTER S WITH CARON - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\u017d' # 0xB4 -> LATIN CAPITAL LETTER Z WITH CARON - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\u017e' # 0xB8 -> LATIN SMALL LETTER Z WITH CARON - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u0152' # 0xBC -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xBD -> LATIN SMALL LIGATURE OE - u'\u0178' # 0xBE -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xde' # 0xDE -> LATIN CAPITAL LETTER THORN - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf0' # 0xF0 -> LATIN SMALL LETTER ETH - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\xfe' # 0xFE -> LATIN SMALL LETTER THORN - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\u20ac' # 0xA4 -> EURO SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\u0160' # 0xA6 -> LATIN CAPITAL LETTER S WITH CARON + '\xa7' # 0xA7 -> SECTION SIGN + '\u0161' # 0xA8 -> LATIN SMALL LETTER S WITH CARON + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\u017d' # 0xB4 -> LATIN CAPITAL LETTER Z WITH CARON + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\u017e' # 0xB8 -> LATIN SMALL LETTER Z WITH CARON + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u0152' # 0xBC -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xBD -> LATIN SMALL LIGATURE OE + '\u0178' # 0xBE -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xd0' # 0xD0 -> LATIN CAPITAL LETTER ETH + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xde' # 0xDE -> LATIN CAPITAL LETTER THORN + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf0' # 0xF0 -> LATIN SMALL LETTER ETH + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\xfe' # 0xFE -> LATIN SMALL LETTER THORN + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/iso8859_16.py b/Lib/encodings/iso8859_16.py index 00b9ac8..e70c96e 100644 --- a/Lib/encodings/iso8859_16.py +++ b/Lib/encodings/iso8859_16.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u0105' # 0xA2 -> LATIN SMALL LETTER A WITH OGONEK - u'\u0141' # 0xA3 -> LATIN CAPITAL LETTER L WITH STROKE - u'\u20ac' # 0xA4 -> EURO SIGN - u'\u201e' # 0xA5 -> DOUBLE LOW-9 QUOTATION MARK - u'\u0160' # 0xA6 -> LATIN CAPITAL LETTER S WITH CARON - u'\xa7' # 0xA7 -> SECTION SIGN - u'\u0161' # 0xA8 -> LATIN SMALL LETTER S WITH CARON - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u0218' # 0xAA -> LATIN CAPITAL LETTER S WITH COMMA BELOW - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u0179' # 0xAC -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\xad' # 0xAD -> SOFT HYPHEN - u'\u017a' # 0xAE -> LATIN SMALL LETTER Z WITH ACUTE - u'\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u010c' # 0xB2 -> LATIN CAPITAL LETTER C WITH CARON - u'\u0142' # 0xB3 -> LATIN SMALL LETTER L WITH STROKE - u'\u017d' # 0xB4 -> LATIN CAPITAL LETTER Z WITH CARON - u'\u201d' # 0xB5 -> RIGHT DOUBLE QUOTATION MARK - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\u017e' # 0xB8 -> LATIN SMALL LETTER Z WITH CARON - u'\u010d' # 0xB9 -> LATIN SMALL LETTER C WITH CARON - u'\u0219' # 0xBA -> LATIN SMALL LETTER S WITH COMMA BELOW - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u0152' # 0xBC -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xBD -> LATIN SMALL LIGATURE OE - u'\u0178' # 0xBE -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\u0106' # 0xC5 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u0150' # 0xD5 -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\u015a' # 0xD7 -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u0170' # 0xD8 -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u0118' # 0xDD -> LATIN CAPITAL LETTER E WITH OGONEK - u'\u021a' # 0xDE -> LATIN CAPITAL LETTER T WITH COMMA BELOW - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u0107' # 0xE5 -> LATIN SMALL LETTER C WITH ACUTE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE - u'\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\u0151' # 0xF5 -> LATIN SMALL LETTER O WITH DOUBLE ACUTE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\u015b' # 0xF7 -> LATIN SMALL LETTER S WITH ACUTE - u'\u0171' # 0xF8 -> LATIN SMALL LETTER U WITH DOUBLE ACUTE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u0119' # 0xFD -> LATIN SMALL LETTER E WITH OGONEK - u'\u021b' # 0xFE -> LATIN SMALL LETTER T WITH COMMA BELOW - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u0105' # 0xA2 -> LATIN SMALL LETTER A WITH OGONEK + '\u0141' # 0xA3 -> LATIN CAPITAL LETTER L WITH STROKE + '\u20ac' # 0xA4 -> EURO SIGN + '\u201e' # 0xA5 -> DOUBLE LOW-9 QUOTATION MARK + '\u0160' # 0xA6 -> LATIN CAPITAL LETTER S WITH CARON + '\xa7' # 0xA7 -> SECTION SIGN + '\u0161' # 0xA8 -> LATIN SMALL LETTER S WITH CARON + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u0218' # 0xAA -> LATIN CAPITAL LETTER S WITH COMMA BELOW + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u0179' # 0xAC -> LATIN CAPITAL LETTER Z WITH ACUTE + '\xad' # 0xAD -> SOFT HYPHEN + '\u017a' # 0xAE -> LATIN SMALL LETTER Z WITH ACUTE + '\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u010c' # 0xB2 -> LATIN CAPITAL LETTER C WITH CARON + '\u0142' # 0xB3 -> LATIN SMALL LETTER L WITH STROKE + '\u017d' # 0xB4 -> LATIN CAPITAL LETTER Z WITH CARON + '\u201d' # 0xB5 -> RIGHT DOUBLE QUOTATION MARK + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\u017e' # 0xB8 -> LATIN SMALL LETTER Z WITH CARON + '\u010d' # 0xB9 -> LATIN SMALL LETTER C WITH CARON + '\u0219' # 0xBA -> LATIN SMALL LETTER S WITH COMMA BELOW + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u0152' # 0xBC -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xBD -> LATIN SMALL LIGATURE OE + '\u0178' # 0xBE -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\u0106' # 0xC5 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE + '\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u0150' # 0xD5 -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\u015a' # 0xD7 -> LATIN CAPITAL LETTER S WITH ACUTE + '\u0170' # 0xD8 -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u0118' # 0xDD -> LATIN CAPITAL LETTER E WITH OGONEK + '\u021a' # 0xDE -> LATIN CAPITAL LETTER T WITH COMMA BELOW + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\u0107' # 0xE5 -> LATIN SMALL LETTER C WITH ACUTE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE + '\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\u0151' # 0xF5 -> LATIN SMALL LETTER O WITH DOUBLE ACUTE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\u015b' # 0xF7 -> LATIN SMALL LETTER S WITH ACUTE + '\u0171' # 0xF8 -> LATIN SMALL LETTER U WITH DOUBLE ACUTE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u0119' # 0xFD -> LATIN SMALL LETTER E WITH OGONEK + '\u021b' # 0xFE -> LATIN SMALL LETTER T WITH COMMA BELOW + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/iso8859_2.py b/Lib/encodings/iso8859_2.py index 38e91d8..3698747 100644 --- a/Lib/encodings/iso8859_2.py +++ b/Lib/encodings/iso8859_2.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u02d8' # 0xA2 -> BREVE - u'\u0141' # 0xA3 -> LATIN CAPITAL LETTER L WITH STROKE - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\u013d' # 0xA5 -> LATIN CAPITAL LETTER L WITH CARON - u'\u015a' # 0xA6 -> LATIN CAPITAL LETTER S WITH ACUTE - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON - u'\u015e' # 0xAA -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\u0164' # 0xAB -> LATIN CAPITAL LETTER T WITH CARON - u'\u0179' # 0xAC -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\xad' # 0xAD -> SOFT HYPHEN - u'\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON - u'\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK - u'\u02db' # 0xB2 -> OGONEK - u'\u0142' # 0xB3 -> LATIN SMALL LETTER L WITH STROKE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\u013e' # 0xB5 -> LATIN SMALL LETTER L WITH CARON - u'\u015b' # 0xB6 -> LATIN SMALL LETTER S WITH ACUTE - u'\u02c7' # 0xB7 -> CARON - u'\xb8' # 0xB8 -> CEDILLA - u'\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON - u'\u015f' # 0xBA -> LATIN SMALL LETTER S WITH CEDILLA - u'\u0165' # 0xBB -> LATIN SMALL LETTER T WITH CARON - u'\u017a' # 0xBC -> LATIN SMALL LETTER Z WITH ACUTE - u'\u02dd' # 0xBD -> DOUBLE ACUTE ACCENT - u'\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON - u'\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u0154' # 0xC0 -> LATIN CAPITAL LETTER R WITH ACUTE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\u0139' # 0xC5 -> LATIN CAPITAL LETTER L WITH ACUTE - u'\u0106' # 0xC6 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u011a' # 0xCC -> LATIN CAPITAL LETTER E WITH CARON - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\u010e' # 0xCF -> LATIN CAPITAL LETTER D WITH CARON - u'\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\u0147' # 0xD2 -> LATIN CAPITAL LETTER N WITH CARON - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u0150' # 0xD5 -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\u0158' # 0xD8 -> LATIN CAPITAL LETTER R WITH CARON - u'\u016e' # 0xD9 -> LATIN CAPITAL LETTER U WITH RING ABOVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\u0170' # 0xDB -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\u0162' # 0xDE -> LATIN CAPITAL LETTER T WITH CEDILLA - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\u0155' # 0xE0 -> LATIN SMALL LETTER R WITH ACUTE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u013a' # 0xE5 -> LATIN SMALL LETTER L WITH ACUTE - u'\u0107' # 0xE6 -> LATIN SMALL LETTER C WITH ACUTE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u011b' # 0xEC -> LATIN SMALL LETTER E WITH CARON - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\u010f' # 0xEF -> LATIN SMALL LETTER D WITH CARON - u'\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE - u'\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0148' # 0xF2 -> LATIN SMALL LETTER N WITH CARON - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\u0151' # 0xF5 -> LATIN SMALL LETTER O WITH DOUBLE ACUTE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\u0159' # 0xF8 -> LATIN SMALL LETTER R WITH CARON - u'\u016f' # 0xF9 -> LATIN SMALL LETTER U WITH RING ABOVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\u0171' # 0xFB -> LATIN SMALL LETTER U WITH DOUBLE ACUTE - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE - u'\u0163' # 0xFE -> LATIN SMALL LETTER T WITH CEDILLA - u'\u02d9' # 0xFF -> DOT ABOVE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u02d8' # 0xA2 -> BREVE + '\u0141' # 0xA3 -> LATIN CAPITAL LETTER L WITH STROKE + '\xa4' # 0xA4 -> CURRENCY SIGN + '\u013d' # 0xA5 -> LATIN CAPITAL LETTER L WITH CARON + '\u015a' # 0xA6 -> LATIN CAPITAL LETTER S WITH ACUTE + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON + '\u015e' # 0xAA -> LATIN CAPITAL LETTER S WITH CEDILLA + '\u0164' # 0xAB -> LATIN CAPITAL LETTER T WITH CARON + '\u0179' # 0xAC -> LATIN CAPITAL LETTER Z WITH ACUTE + '\xad' # 0xAD -> SOFT HYPHEN + '\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON + '\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\xb0' # 0xB0 -> DEGREE SIGN + '\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK + '\u02db' # 0xB2 -> OGONEK + '\u0142' # 0xB3 -> LATIN SMALL LETTER L WITH STROKE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\u013e' # 0xB5 -> LATIN SMALL LETTER L WITH CARON + '\u015b' # 0xB6 -> LATIN SMALL LETTER S WITH ACUTE + '\u02c7' # 0xB7 -> CARON + '\xb8' # 0xB8 -> CEDILLA + '\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON + '\u015f' # 0xBA -> LATIN SMALL LETTER S WITH CEDILLA + '\u0165' # 0xBB -> LATIN SMALL LETTER T WITH CARON + '\u017a' # 0xBC -> LATIN SMALL LETTER Z WITH ACUTE + '\u02dd' # 0xBD -> DOUBLE ACUTE ACCENT + '\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON + '\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u0154' # 0xC0 -> LATIN CAPITAL LETTER R WITH ACUTE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\u0102' # 0xC3 -> LATIN CAPITAL LETTER A WITH BREVE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\u0139' # 0xC5 -> LATIN CAPITAL LETTER L WITH ACUTE + '\u0106' # 0xC6 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u011a' # 0xCC -> LATIN CAPITAL LETTER E WITH CARON + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\u010e' # 0xCF -> LATIN CAPITAL LETTER D WITH CARON + '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE + '\u0143' # 0xD1 -> LATIN CAPITAL LETTER N WITH ACUTE + '\u0147' # 0xD2 -> LATIN CAPITAL LETTER N WITH CARON + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u0150' # 0xD5 -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\u0158' # 0xD8 -> LATIN CAPITAL LETTER R WITH CARON + '\u016e' # 0xD9 -> LATIN CAPITAL LETTER U WITH RING ABOVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\u0170' # 0xDB -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xdd' # 0xDD -> LATIN CAPITAL LETTER Y WITH ACUTE + '\u0162' # 0xDE -> LATIN CAPITAL LETTER T WITH CEDILLA + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\u0155' # 0xE0 -> LATIN SMALL LETTER R WITH ACUTE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\u0103' # 0xE3 -> LATIN SMALL LETTER A WITH BREVE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\u013a' # 0xE5 -> LATIN SMALL LETTER L WITH ACUTE + '\u0107' # 0xE6 -> LATIN SMALL LETTER C WITH ACUTE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\u011b' # 0xEC -> LATIN SMALL LETTER E WITH CARON + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\u010f' # 0xEF -> LATIN SMALL LETTER D WITH CARON + '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE + '\u0144' # 0xF1 -> LATIN SMALL LETTER N WITH ACUTE + '\u0148' # 0xF2 -> LATIN SMALL LETTER N WITH CARON + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\u0151' # 0xF5 -> LATIN SMALL LETTER O WITH DOUBLE ACUTE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\u0159' # 0xF8 -> LATIN SMALL LETTER R WITH CARON + '\u016f' # 0xF9 -> LATIN SMALL LETTER U WITH RING ABOVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\u0171' # 0xFB -> LATIN SMALL LETTER U WITH DOUBLE ACUTE + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\xfd' # 0xFD -> LATIN SMALL LETTER Y WITH ACUTE + '\u0163' # 0xFE -> LATIN SMALL LETTER T WITH CEDILLA + '\u02d9' # 0xFF -> DOT ABOVE ) ### Encoding table diff --git a/Lib/encodings/iso8859_3.py b/Lib/encodings/iso8859_3.py index 23daafd..96d3063 100644 --- a/Lib/encodings/iso8859_3.py +++ b/Lib/encodings/iso8859_3.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0126' # 0xA1 -> LATIN CAPITAL LETTER H WITH STROKE - u'\u02d8' # 0xA2 -> BREVE - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\ufffe' - u'\u0124' # 0xA6 -> LATIN CAPITAL LETTER H WITH CIRCUMFLEX - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\u0130' # 0xA9 -> LATIN CAPITAL LETTER I WITH DOT ABOVE - u'\u015e' # 0xAA -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\u011e' # 0xAB -> LATIN CAPITAL LETTER G WITH BREVE - u'\u0134' # 0xAC -> LATIN CAPITAL LETTER J WITH CIRCUMFLEX - u'\xad' # 0xAD -> SOFT HYPHEN - u'\ufffe' - u'\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\u0127' # 0xB1 -> LATIN SMALL LETTER H WITH STROKE - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u0125' # 0xB6 -> LATIN SMALL LETTER H WITH CIRCUMFLEX - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\u0131' # 0xB9 -> LATIN SMALL LETTER DOTLESS I - u'\u015f' # 0xBA -> LATIN SMALL LETTER S WITH CEDILLA - u'\u011f' # 0xBB -> LATIN SMALL LETTER G WITH BREVE - u'\u0135' # 0xBC -> LATIN SMALL LETTER J WITH CIRCUMFLEX - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\ufffe' - u'\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\ufffe' - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\u010a' # 0xC5 -> LATIN CAPITAL LETTER C WITH DOT ABOVE - u'\u0108' # 0xC6 -> LATIN CAPITAL LETTER C WITH CIRCUMFLEX - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\ufffe' - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u0120' # 0xD5 -> LATIN CAPITAL LETTER G WITH DOT ABOVE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\u011c' # 0xD8 -> LATIN CAPITAL LETTER G WITH CIRCUMFLEX - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u016c' # 0xDD -> LATIN CAPITAL LETTER U WITH BREVE - u'\u015c' # 0xDE -> LATIN CAPITAL LETTER S WITH CIRCUMFLEX - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\ufffe' - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u010b' # 0xE5 -> LATIN SMALL LETTER C WITH DOT ABOVE - u'\u0109' # 0xE6 -> LATIN SMALL LETTER C WITH CIRCUMFLEX - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\ufffe' - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\u0121' # 0xF5 -> LATIN SMALL LETTER G WITH DOT ABOVE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\u011d' # 0xF8 -> LATIN SMALL LETTER G WITH CIRCUMFLEX - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u016d' # 0xFD -> LATIN SMALL LETTER U WITH BREVE - u'\u015d' # 0xFE -> LATIN SMALL LETTER S WITH CIRCUMFLEX - u'\u02d9' # 0xFF -> DOT ABOVE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0126' # 0xA1 -> LATIN CAPITAL LETTER H WITH STROKE + '\u02d8' # 0xA2 -> BREVE + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\ufffe' + '\u0124' # 0xA6 -> LATIN CAPITAL LETTER H WITH CIRCUMFLEX + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\u0130' # 0xA9 -> LATIN CAPITAL LETTER I WITH DOT ABOVE + '\u015e' # 0xAA -> LATIN CAPITAL LETTER S WITH CEDILLA + '\u011e' # 0xAB -> LATIN CAPITAL LETTER G WITH BREVE + '\u0134' # 0xAC -> LATIN CAPITAL LETTER J WITH CIRCUMFLEX + '\xad' # 0xAD -> SOFT HYPHEN + '\ufffe' + '\u017b' # 0xAF -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\xb0' # 0xB0 -> DEGREE SIGN + '\u0127' # 0xB1 -> LATIN SMALL LETTER H WITH STROKE + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\u0125' # 0xB6 -> LATIN SMALL LETTER H WITH CIRCUMFLEX + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\u0131' # 0xB9 -> LATIN SMALL LETTER DOTLESS I + '\u015f' # 0xBA -> LATIN SMALL LETTER S WITH CEDILLA + '\u011f' # 0xBB -> LATIN SMALL LETTER G WITH BREVE + '\u0135' # 0xBC -> LATIN SMALL LETTER J WITH CIRCUMFLEX + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\ufffe' + '\u017c' # 0xBF -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\ufffe' + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\u010a' # 0xC5 -> LATIN CAPITAL LETTER C WITH DOT ABOVE + '\u0108' # 0xC6 -> LATIN CAPITAL LETTER C WITH CIRCUMFLEX + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\ufffe' + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u0120' # 0xD5 -> LATIN CAPITAL LETTER G WITH DOT ABOVE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\u011c' # 0xD8 -> LATIN CAPITAL LETTER G WITH CIRCUMFLEX + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u016c' # 0xDD -> LATIN CAPITAL LETTER U WITH BREVE + '\u015c' # 0xDE -> LATIN CAPITAL LETTER S WITH CIRCUMFLEX + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\ufffe' + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\u010b' # 0xE5 -> LATIN SMALL LETTER C WITH DOT ABOVE + '\u0109' # 0xE6 -> LATIN SMALL LETTER C WITH CIRCUMFLEX + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\ufffe' + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\u0121' # 0xF5 -> LATIN SMALL LETTER G WITH DOT ABOVE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\u011d' # 0xF8 -> LATIN SMALL LETTER G WITH CIRCUMFLEX + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u016d' # 0xFD -> LATIN SMALL LETTER U WITH BREVE + '\u015d' # 0xFE -> LATIN SMALL LETTER S WITH CIRCUMFLEX + '\u02d9' # 0xFF -> DOT ABOVE ) ### Encoding table diff --git a/Lib/encodings/iso8859_4.py b/Lib/encodings/iso8859_4.py index c8e03b5..65c1e00 100644 --- a/Lib/encodings/iso8859_4.py +++ b/Lib/encodings/iso8859_4.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\u0138' # 0xA2 -> LATIN SMALL LETTER KRA - u'\u0156' # 0xA3 -> LATIN CAPITAL LETTER R WITH CEDILLA - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\u0128' # 0xA5 -> LATIN CAPITAL LETTER I WITH TILDE - u'\u013b' # 0xA6 -> LATIN CAPITAL LETTER L WITH CEDILLA - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON - u'\u0112' # 0xAA -> LATIN CAPITAL LETTER E WITH MACRON - u'\u0122' # 0xAB -> LATIN CAPITAL LETTER G WITH CEDILLA - u'\u0166' # 0xAC -> LATIN CAPITAL LETTER T WITH STROKE - u'\xad' # 0xAD -> SOFT HYPHEN - u'\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK - u'\u02db' # 0xB2 -> OGONEK - u'\u0157' # 0xB3 -> LATIN SMALL LETTER R WITH CEDILLA - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\u0129' # 0xB5 -> LATIN SMALL LETTER I WITH TILDE - u'\u013c' # 0xB6 -> LATIN SMALL LETTER L WITH CEDILLA - u'\u02c7' # 0xB7 -> CARON - u'\xb8' # 0xB8 -> CEDILLA - u'\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON - u'\u0113' # 0xBA -> LATIN SMALL LETTER E WITH MACRON - u'\u0123' # 0xBB -> LATIN SMALL LETTER G WITH CEDILLA - u'\u0167' # 0xBC -> LATIN SMALL LETTER T WITH STROKE - u'\u014a' # 0xBD -> LATIN CAPITAL LETTER ENG - u'\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON - u'\u014b' # 0xBF -> LATIN SMALL LETTER ENG - u'\u0100' # 0xC0 -> LATIN CAPITAL LETTER A WITH MACRON - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\u012e' # 0xC7 -> LATIN CAPITAL LETTER I WITH OGONEK - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u0116' # 0xCC -> LATIN CAPITAL LETTER E WITH DOT ABOVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\u012a' # 0xCF -> LATIN CAPITAL LETTER I WITH MACRON - u'\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u0145' # 0xD1 -> LATIN CAPITAL LETTER N WITH CEDILLA - u'\u014c' # 0xD2 -> LATIN CAPITAL LETTER O WITH MACRON - u'\u0136' # 0xD3 -> LATIN CAPITAL LETTER K WITH CEDILLA - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\u0172' # 0xD9 -> LATIN CAPITAL LETTER U WITH OGONEK - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u0168' # 0xDD -> LATIN CAPITAL LETTER U WITH TILDE - u'\u016a' # 0xDE -> LATIN CAPITAL LETTER U WITH MACRON - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\u0101' # 0xE0 -> LATIN SMALL LETTER A WITH MACRON - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\u012f' # 0xE7 -> LATIN SMALL LETTER I WITH OGONEK - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\u0117' # 0xEC -> LATIN SMALL LETTER E WITH DOT ABOVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\u012b' # 0xEF -> LATIN SMALL LETTER I WITH MACRON - u'\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE - u'\u0146' # 0xF1 -> LATIN SMALL LETTER N WITH CEDILLA - u'\u014d' # 0xF2 -> LATIN SMALL LETTER O WITH MACRON - u'\u0137' # 0xF3 -> LATIN SMALL LETTER K WITH CEDILLA - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\u0173' # 0xF9 -> LATIN SMALL LETTER U WITH OGONEK - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u0169' # 0xFD -> LATIN SMALL LETTER U WITH TILDE - u'\u016b' # 0xFE -> LATIN SMALL LETTER U WITH MACRON - u'\u02d9' # 0xFF -> DOT ABOVE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK + '\u0138' # 0xA2 -> LATIN SMALL LETTER KRA + '\u0156' # 0xA3 -> LATIN CAPITAL LETTER R WITH CEDILLA + '\xa4' # 0xA4 -> CURRENCY SIGN + '\u0128' # 0xA5 -> LATIN CAPITAL LETTER I WITH TILDE + '\u013b' # 0xA6 -> LATIN CAPITAL LETTER L WITH CEDILLA + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON + '\u0112' # 0xAA -> LATIN CAPITAL LETTER E WITH MACRON + '\u0122' # 0xAB -> LATIN CAPITAL LETTER G WITH CEDILLA + '\u0166' # 0xAC -> LATIN CAPITAL LETTER T WITH STROKE + '\xad' # 0xAD -> SOFT HYPHEN + '\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK + '\u02db' # 0xB2 -> OGONEK + '\u0157' # 0xB3 -> LATIN SMALL LETTER R WITH CEDILLA + '\xb4' # 0xB4 -> ACUTE ACCENT + '\u0129' # 0xB5 -> LATIN SMALL LETTER I WITH TILDE + '\u013c' # 0xB6 -> LATIN SMALL LETTER L WITH CEDILLA + '\u02c7' # 0xB7 -> CARON + '\xb8' # 0xB8 -> CEDILLA + '\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON + '\u0113' # 0xBA -> LATIN SMALL LETTER E WITH MACRON + '\u0123' # 0xBB -> LATIN SMALL LETTER G WITH CEDILLA + '\u0167' # 0xBC -> LATIN SMALL LETTER T WITH STROKE + '\u014a' # 0xBD -> LATIN CAPITAL LETTER ENG + '\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON + '\u014b' # 0xBF -> LATIN SMALL LETTER ENG + '\u0100' # 0xC0 -> LATIN CAPITAL LETTER A WITH MACRON + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\u012e' # 0xC7 -> LATIN CAPITAL LETTER I WITH OGONEK + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u0116' # 0xCC -> LATIN CAPITAL LETTER E WITH DOT ABOVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\u012a' # 0xCF -> LATIN CAPITAL LETTER I WITH MACRON + '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE + '\u0145' # 0xD1 -> LATIN CAPITAL LETTER N WITH CEDILLA + '\u014c' # 0xD2 -> LATIN CAPITAL LETTER O WITH MACRON + '\u0136' # 0xD3 -> LATIN CAPITAL LETTER K WITH CEDILLA + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\u0172' # 0xD9 -> LATIN CAPITAL LETTER U WITH OGONEK + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u0168' # 0xDD -> LATIN CAPITAL LETTER U WITH TILDE + '\u016a' # 0xDE -> LATIN CAPITAL LETTER U WITH MACRON + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\u0101' # 0xE0 -> LATIN SMALL LETTER A WITH MACRON + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\u012f' # 0xE7 -> LATIN SMALL LETTER I WITH OGONEK + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\u0117' # 0xEC -> LATIN SMALL LETTER E WITH DOT ABOVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\u012b' # 0xEF -> LATIN SMALL LETTER I WITH MACRON + '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE + '\u0146' # 0xF1 -> LATIN SMALL LETTER N WITH CEDILLA + '\u014d' # 0xF2 -> LATIN SMALL LETTER O WITH MACRON + '\u0137' # 0xF3 -> LATIN SMALL LETTER K WITH CEDILLA + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\u0173' # 0xF9 -> LATIN SMALL LETTER U WITH OGONEK + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u0169' # 0xFD -> LATIN SMALL LETTER U WITH TILDE + '\u016b' # 0xFE -> LATIN SMALL LETTER U WITH MACRON + '\u02d9' # 0xFF -> DOT ABOVE ) ### Encoding table diff --git a/Lib/encodings/iso8859_5.py b/Lib/encodings/iso8859_5.py index c01cd1c..a3c868a 100644 --- a/Lib/encodings/iso8859_5.py +++ b/Lib/encodings/iso8859_5.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u0401' # 0xA1 -> CYRILLIC CAPITAL LETTER IO - u'\u0402' # 0xA2 -> CYRILLIC CAPITAL LETTER DJE - u'\u0403' # 0xA3 -> CYRILLIC CAPITAL LETTER GJE - u'\u0404' # 0xA4 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\u0405' # 0xA5 -> CYRILLIC CAPITAL LETTER DZE - u'\u0406' # 0xA6 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0407' # 0xA7 -> CYRILLIC CAPITAL LETTER YI - u'\u0408' # 0xA8 -> CYRILLIC CAPITAL LETTER JE - u'\u0409' # 0xA9 -> CYRILLIC CAPITAL LETTER LJE - u'\u040a' # 0xAA -> CYRILLIC CAPITAL LETTER NJE - u'\u040b' # 0xAB -> CYRILLIC CAPITAL LETTER TSHE - u'\u040c' # 0xAC -> CYRILLIC CAPITAL LETTER KJE - u'\xad' # 0xAD -> SOFT HYPHEN - u'\u040e' # 0xAE -> CYRILLIC CAPITAL LETTER SHORT U - u'\u040f' # 0xAF -> CYRILLIC CAPITAL LETTER DZHE - u'\u0410' # 0xB0 -> CYRILLIC CAPITAL LETTER A - u'\u0411' # 0xB1 -> CYRILLIC CAPITAL LETTER BE - u'\u0412' # 0xB2 -> CYRILLIC CAPITAL LETTER VE - u'\u0413' # 0xB3 -> CYRILLIC CAPITAL LETTER GHE - u'\u0414' # 0xB4 -> CYRILLIC CAPITAL LETTER DE - u'\u0415' # 0xB5 -> CYRILLIC CAPITAL LETTER IE - u'\u0416' # 0xB6 -> CYRILLIC CAPITAL LETTER ZHE - u'\u0417' # 0xB7 -> CYRILLIC CAPITAL LETTER ZE - u'\u0418' # 0xB8 -> CYRILLIC CAPITAL LETTER I - u'\u0419' # 0xB9 -> CYRILLIC CAPITAL LETTER SHORT I - u'\u041a' # 0xBA -> CYRILLIC CAPITAL LETTER KA - u'\u041b' # 0xBB -> CYRILLIC CAPITAL LETTER EL - u'\u041c' # 0xBC -> CYRILLIC CAPITAL LETTER EM - u'\u041d' # 0xBD -> CYRILLIC CAPITAL LETTER EN - u'\u041e' # 0xBE -> CYRILLIC CAPITAL LETTER O - u'\u041f' # 0xBF -> CYRILLIC CAPITAL LETTER PE - u'\u0420' # 0xC0 -> CYRILLIC CAPITAL LETTER ER - u'\u0421' # 0xC1 -> CYRILLIC CAPITAL LETTER ES - u'\u0422' # 0xC2 -> CYRILLIC CAPITAL LETTER TE - u'\u0423' # 0xC3 -> CYRILLIC CAPITAL LETTER U - u'\u0424' # 0xC4 -> CYRILLIC CAPITAL LETTER EF - u'\u0425' # 0xC5 -> CYRILLIC CAPITAL LETTER HA - u'\u0426' # 0xC6 -> CYRILLIC CAPITAL LETTER TSE - u'\u0427' # 0xC7 -> CYRILLIC CAPITAL LETTER CHE - u'\u0428' # 0xC8 -> CYRILLIC CAPITAL LETTER SHA - u'\u0429' # 0xC9 -> CYRILLIC CAPITAL LETTER SHCHA - u'\u042a' # 0xCA -> CYRILLIC CAPITAL LETTER HARD SIGN - u'\u042b' # 0xCB -> CYRILLIC CAPITAL LETTER YERU - u'\u042c' # 0xCC -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u042d' # 0xCD -> CYRILLIC CAPITAL LETTER E - u'\u042e' # 0xCE -> CYRILLIC CAPITAL LETTER YU - u'\u042f' # 0xCF -> CYRILLIC CAPITAL LETTER YA - u'\u0430' # 0xD0 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0xD1 -> CYRILLIC SMALL LETTER BE - u'\u0432' # 0xD2 -> CYRILLIC SMALL LETTER VE - u'\u0433' # 0xD3 -> CYRILLIC SMALL LETTER GHE - u'\u0434' # 0xD4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0xD5 -> CYRILLIC SMALL LETTER IE - u'\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE - u'\u0437' # 0xD7 -> CYRILLIC SMALL LETTER ZE - u'\u0438' # 0xD8 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0xD9 -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0xDA -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0xDB -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0xDC -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0xDD -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0xDE -> CYRILLIC SMALL LETTER O - u'\u043f' # 0xDF -> CYRILLIC SMALL LETTER PE - u'\u0440' # 0xE0 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0xE1 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0xE2 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0xE3 -> CYRILLIC SMALL LETTER U - u'\u0444' # 0xE4 -> CYRILLIC SMALL LETTER EF - u'\u0445' # 0xE5 -> CYRILLIC SMALL LETTER HA - u'\u0446' # 0xE6 -> CYRILLIC SMALL LETTER TSE - u'\u0447' # 0xE7 -> CYRILLIC SMALL LETTER CHE - u'\u0448' # 0xE8 -> CYRILLIC SMALL LETTER SHA - u'\u0449' # 0xE9 -> CYRILLIC SMALL LETTER SHCHA - u'\u044a' # 0xEA -> CYRILLIC SMALL LETTER HARD SIGN - u'\u044b' # 0xEB -> CYRILLIC SMALL LETTER YERU - u'\u044c' # 0xEC -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044d' # 0xED -> CYRILLIC SMALL LETTER E - u'\u044e' # 0xEE -> CYRILLIC SMALL LETTER YU - u'\u044f' # 0xEF -> CYRILLIC SMALL LETTER YA - u'\u2116' # 0xF0 -> NUMERO SIGN - u'\u0451' # 0xF1 -> CYRILLIC SMALL LETTER IO - u'\u0452' # 0xF2 -> CYRILLIC SMALL LETTER DJE - u'\u0453' # 0xF3 -> CYRILLIC SMALL LETTER GJE - u'\u0454' # 0xF4 -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\u0455' # 0xF5 -> CYRILLIC SMALL LETTER DZE - u'\u0456' # 0xF6 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0457' # 0xF7 -> CYRILLIC SMALL LETTER YI - u'\u0458' # 0xF8 -> CYRILLIC SMALL LETTER JE - u'\u0459' # 0xF9 -> CYRILLIC SMALL LETTER LJE - u'\u045a' # 0xFA -> CYRILLIC SMALL LETTER NJE - u'\u045b' # 0xFB -> CYRILLIC SMALL LETTER TSHE - u'\u045c' # 0xFC -> CYRILLIC SMALL LETTER KJE - u'\xa7' # 0xFD -> SECTION SIGN - u'\u045e' # 0xFE -> CYRILLIC SMALL LETTER SHORT U - u'\u045f' # 0xFF -> CYRILLIC SMALL LETTER DZHE + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u0401' # 0xA1 -> CYRILLIC CAPITAL LETTER IO + '\u0402' # 0xA2 -> CYRILLIC CAPITAL LETTER DJE + '\u0403' # 0xA3 -> CYRILLIC CAPITAL LETTER GJE + '\u0404' # 0xA4 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE + '\u0405' # 0xA5 -> CYRILLIC CAPITAL LETTER DZE + '\u0406' # 0xA6 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0407' # 0xA7 -> CYRILLIC CAPITAL LETTER YI + '\u0408' # 0xA8 -> CYRILLIC CAPITAL LETTER JE + '\u0409' # 0xA9 -> CYRILLIC CAPITAL LETTER LJE + '\u040a' # 0xAA -> CYRILLIC CAPITAL LETTER NJE + '\u040b' # 0xAB -> CYRILLIC CAPITAL LETTER TSHE + '\u040c' # 0xAC -> CYRILLIC CAPITAL LETTER KJE + '\xad' # 0xAD -> SOFT HYPHEN + '\u040e' # 0xAE -> CYRILLIC CAPITAL LETTER SHORT U + '\u040f' # 0xAF -> CYRILLIC CAPITAL LETTER DZHE + '\u0410' # 0xB0 -> CYRILLIC CAPITAL LETTER A + '\u0411' # 0xB1 -> CYRILLIC CAPITAL LETTER BE + '\u0412' # 0xB2 -> CYRILLIC CAPITAL LETTER VE + '\u0413' # 0xB3 -> CYRILLIC CAPITAL LETTER GHE + '\u0414' # 0xB4 -> CYRILLIC CAPITAL LETTER DE + '\u0415' # 0xB5 -> CYRILLIC CAPITAL LETTER IE + '\u0416' # 0xB6 -> CYRILLIC CAPITAL LETTER ZHE + '\u0417' # 0xB7 -> CYRILLIC CAPITAL LETTER ZE + '\u0418' # 0xB8 -> CYRILLIC CAPITAL LETTER I + '\u0419' # 0xB9 -> CYRILLIC CAPITAL LETTER SHORT I + '\u041a' # 0xBA -> CYRILLIC CAPITAL LETTER KA + '\u041b' # 0xBB -> CYRILLIC CAPITAL LETTER EL + '\u041c' # 0xBC -> CYRILLIC CAPITAL LETTER EM + '\u041d' # 0xBD -> CYRILLIC CAPITAL LETTER EN + '\u041e' # 0xBE -> CYRILLIC CAPITAL LETTER O + '\u041f' # 0xBF -> CYRILLIC CAPITAL LETTER PE + '\u0420' # 0xC0 -> CYRILLIC CAPITAL LETTER ER + '\u0421' # 0xC1 -> CYRILLIC CAPITAL LETTER ES + '\u0422' # 0xC2 -> CYRILLIC CAPITAL LETTER TE + '\u0423' # 0xC3 -> CYRILLIC CAPITAL LETTER U + '\u0424' # 0xC4 -> CYRILLIC CAPITAL LETTER EF + '\u0425' # 0xC5 -> CYRILLIC CAPITAL LETTER HA + '\u0426' # 0xC6 -> CYRILLIC CAPITAL LETTER TSE + '\u0427' # 0xC7 -> CYRILLIC CAPITAL LETTER CHE + '\u0428' # 0xC8 -> CYRILLIC CAPITAL LETTER SHA + '\u0429' # 0xC9 -> CYRILLIC CAPITAL LETTER SHCHA + '\u042a' # 0xCA -> CYRILLIC CAPITAL LETTER HARD SIGN + '\u042b' # 0xCB -> CYRILLIC CAPITAL LETTER YERU + '\u042c' # 0xCC -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u042d' # 0xCD -> CYRILLIC CAPITAL LETTER E + '\u042e' # 0xCE -> CYRILLIC CAPITAL LETTER YU + '\u042f' # 0xCF -> CYRILLIC CAPITAL LETTER YA + '\u0430' # 0xD0 -> CYRILLIC SMALL LETTER A + '\u0431' # 0xD1 -> CYRILLIC SMALL LETTER BE + '\u0432' # 0xD2 -> CYRILLIC SMALL LETTER VE + '\u0433' # 0xD3 -> CYRILLIC SMALL LETTER GHE + '\u0434' # 0xD4 -> CYRILLIC SMALL LETTER DE + '\u0435' # 0xD5 -> CYRILLIC SMALL LETTER IE + '\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE + '\u0437' # 0xD7 -> CYRILLIC SMALL LETTER ZE + '\u0438' # 0xD8 -> CYRILLIC SMALL LETTER I + '\u0439' # 0xD9 -> CYRILLIC SMALL LETTER SHORT I + '\u043a' # 0xDA -> CYRILLIC SMALL LETTER KA + '\u043b' # 0xDB -> CYRILLIC SMALL LETTER EL + '\u043c' # 0xDC -> CYRILLIC SMALL LETTER EM + '\u043d' # 0xDD -> CYRILLIC SMALL LETTER EN + '\u043e' # 0xDE -> CYRILLIC SMALL LETTER O + '\u043f' # 0xDF -> CYRILLIC SMALL LETTER PE + '\u0440' # 0xE0 -> CYRILLIC SMALL LETTER ER + '\u0441' # 0xE1 -> CYRILLIC SMALL LETTER ES + '\u0442' # 0xE2 -> CYRILLIC SMALL LETTER TE + '\u0443' # 0xE3 -> CYRILLIC SMALL LETTER U + '\u0444' # 0xE4 -> CYRILLIC SMALL LETTER EF + '\u0445' # 0xE5 -> CYRILLIC SMALL LETTER HA + '\u0446' # 0xE6 -> CYRILLIC SMALL LETTER TSE + '\u0447' # 0xE7 -> CYRILLIC SMALL LETTER CHE + '\u0448' # 0xE8 -> CYRILLIC SMALL LETTER SHA + '\u0449' # 0xE9 -> CYRILLIC SMALL LETTER SHCHA + '\u044a' # 0xEA -> CYRILLIC SMALL LETTER HARD SIGN + '\u044b' # 0xEB -> CYRILLIC SMALL LETTER YERU + '\u044c' # 0xEC -> CYRILLIC SMALL LETTER SOFT SIGN + '\u044d' # 0xED -> CYRILLIC SMALL LETTER E + '\u044e' # 0xEE -> CYRILLIC SMALL LETTER YU + '\u044f' # 0xEF -> CYRILLIC SMALL LETTER YA + '\u2116' # 0xF0 -> NUMERO SIGN + '\u0451' # 0xF1 -> CYRILLIC SMALL LETTER IO + '\u0452' # 0xF2 -> CYRILLIC SMALL LETTER DJE + '\u0453' # 0xF3 -> CYRILLIC SMALL LETTER GJE + '\u0454' # 0xF4 -> CYRILLIC SMALL LETTER UKRAINIAN IE + '\u0455' # 0xF5 -> CYRILLIC SMALL LETTER DZE + '\u0456' # 0xF6 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0457' # 0xF7 -> CYRILLIC SMALL LETTER YI + '\u0458' # 0xF8 -> CYRILLIC SMALL LETTER JE + '\u0459' # 0xF9 -> CYRILLIC SMALL LETTER LJE + '\u045a' # 0xFA -> CYRILLIC SMALL LETTER NJE + '\u045b' # 0xFB -> CYRILLIC SMALL LETTER TSHE + '\u045c' # 0xFC -> CYRILLIC SMALL LETTER KJE + '\xa7' # 0xFD -> SECTION SIGN + '\u045e' # 0xFE -> CYRILLIC SMALL LETTER SHORT U + '\u045f' # 0xFF -> CYRILLIC SMALL LETTER DZHE ) ### Encoding table diff --git a/Lib/encodings/iso8859_6.py b/Lib/encodings/iso8859_6.py index 16c34a3..b02ade6 100644 --- a/Lib/encodings/iso8859_6.py +++ b/Lib/encodings/iso8859_6.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u060c' # 0xAC -> ARABIC COMMA - u'\xad' # 0xAD -> SOFT HYPHEN - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u061b' # 0xBB -> ARABIC SEMICOLON - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u061f' # 0xBF -> ARABIC QUESTION MARK - u'\ufffe' - u'\u0621' # 0xC1 -> ARABIC LETTER HAMZA - u'\u0622' # 0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE - u'\u0623' # 0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE - u'\u0624' # 0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE - u'\u0625' # 0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW - u'\u0626' # 0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE - u'\u0627' # 0xC7 -> ARABIC LETTER ALEF - u'\u0628' # 0xC8 -> ARABIC LETTER BEH - u'\u0629' # 0xC9 -> ARABIC LETTER TEH MARBUTA - u'\u062a' # 0xCA -> ARABIC LETTER TEH - u'\u062b' # 0xCB -> ARABIC LETTER THEH - u'\u062c' # 0xCC -> ARABIC LETTER JEEM - u'\u062d' # 0xCD -> ARABIC LETTER HAH - u'\u062e' # 0xCE -> ARABIC LETTER KHAH - u'\u062f' # 0xCF -> ARABIC LETTER DAL - u'\u0630' # 0xD0 -> ARABIC LETTER THAL - u'\u0631' # 0xD1 -> ARABIC LETTER REH - u'\u0632' # 0xD2 -> ARABIC LETTER ZAIN - u'\u0633' # 0xD3 -> ARABIC LETTER SEEN - u'\u0634' # 0xD4 -> ARABIC LETTER SHEEN - u'\u0635' # 0xD5 -> ARABIC LETTER SAD - u'\u0636' # 0xD6 -> ARABIC LETTER DAD - u'\u0637' # 0xD7 -> ARABIC LETTER TAH - u'\u0638' # 0xD8 -> ARABIC LETTER ZAH - u'\u0639' # 0xD9 -> ARABIC LETTER AIN - u'\u063a' # 0xDA -> ARABIC LETTER GHAIN - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u0640' # 0xE0 -> ARABIC TATWEEL - u'\u0641' # 0xE1 -> ARABIC LETTER FEH - u'\u0642' # 0xE2 -> ARABIC LETTER QAF - u'\u0643' # 0xE3 -> ARABIC LETTER KAF - u'\u0644' # 0xE4 -> ARABIC LETTER LAM - u'\u0645' # 0xE5 -> ARABIC LETTER MEEM - u'\u0646' # 0xE6 -> ARABIC LETTER NOON - u'\u0647' # 0xE7 -> ARABIC LETTER HEH - u'\u0648' # 0xE8 -> ARABIC LETTER WAW - u'\u0649' # 0xE9 -> ARABIC LETTER ALEF MAKSURA - u'\u064a' # 0xEA -> ARABIC LETTER YEH - u'\u064b' # 0xEB -> ARABIC FATHATAN - u'\u064c' # 0xEC -> ARABIC DAMMATAN - u'\u064d' # 0xED -> ARABIC KASRATAN - u'\u064e' # 0xEE -> ARABIC FATHA - u'\u064f' # 0xEF -> ARABIC DAMMA - u'\u0650' # 0xF0 -> ARABIC KASRA - u'\u0651' # 0xF1 -> ARABIC SHADDA - u'\u0652' # 0xF2 -> ARABIC SUKUN - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\ufffe' + '\ufffe' + '\ufffe' + '\xa4' # 0xA4 -> CURRENCY SIGN + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\u060c' # 0xAC -> ARABIC COMMA + '\xad' # 0xAD -> SOFT HYPHEN + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\u061b' # 0xBB -> ARABIC SEMICOLON + '\ufffe' + '\ufffe' + '\ufffe' + '\u061f' # 0xBF -> ARABIC QUESTION MARK + '\ufffe' + '\u0621' # 0xC1 -> ARABIC LETTER HAMZA + '\u0622' # 0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE + '\u0623' # 0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE + '\u0624' # 0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE + '\u0625' # 0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW + '\u0626' # 0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE + '\u0627' # 0xC7 -> ARABIC LETTER ALEF + '\u0628' # 0xC8 -> ARABIC LETTER BEH + '\u0629' # 0xC9 -> ARABIC LETTER TEH MARBUTA + '\u062a' # 0xCA -> ARABIC LETTER TEH + '\u062b' # 0xCB -> ARABIC LETTER THEH + '\u062c' # 0xCC -> ARABIC LETTER JEEM + '\u062d' # 0xCD -> ARABIC LETTER HAH + '\u062e' # 0xCE -> ARABIC LETTER KHAH + '\u062f' # 0xCF -> ARABIC LETTER DAL + '\u0630' # 0xD0 -> ARABIC LETTER THAL + '\u0631' # 0xD1 -> ARABIC LETTER REH + '\u0632' # 0xD2 -> ARABIC LETTER ZAIN + '\u0633' # 0xD3 -> ARABIC LETTER SEEN + '\u0634' # 0xD4 -> ARABIC LETTER SHEEN + '\u0635' # 0xD5 -> ARABIC LETTER SAD + '\u0636' # 0xD6 -> ARABIC LETTER DAD + '\u0637' # 0xD7 -> ARABIC LETTER TAH + '\u0638' # 0xD8 -> ARABIC LETTER ZAH + '\u0639' # 0xD9 -> ARABIC LETTER AIN + '\u063a' # 0xDA -> ARABIC LETTER GHAIN + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\u0640' # 0xE0 -> ARABIC TATWEEL + '\u0641' # 0xE1 -> ARABIC LETTER FEH + '\u0642' # 0xE2 -> ARABIC LETTER QAF + '\u0643' # 0xE3 -> ARABIC LETTER KAF + '\u0644' # 0xE4 -> ARABIC LETTER LAM + '\u0645' # 0xE5 -> ARABIC LETTER MEEM + '\u0646' # 0xE6 -> ARABIC LETTER NOON + '\u0647' # 0xE7 -> ARABIC LETTER HEH + '\u0648' # 0xE8 -> ARABIC LETTER WAW + '\u0649' # 0xE9 -> ARABIC LETTER ALEF MAKSURA + '\u064a' # 0xEA -> ARABIC LETTER YEH + '\u064b' # 0xEB -> ARABIC FATHATAN + '\u064c' # 0xEC -> ARABIC DAMMATAN + '\u064d' # 0xED -> ARABIC KASRATAN + '\u064e' # 0xEE -> ARABIC FATHA + '\u064f' # 0xEF -> ARABIC DAMMA + '\u0650' # 0xF0 -> ARABIC KASRA + '\u0651' # 0xF1 -> ARABIC SHADDA + '\u0652' # 0xF2 -> ARABIC SUKUN + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' ) ### Encoding table diff --git a/Lib/encodings/iso8859_7.py b/Lib/encodings/iso8859_7.py index a560023..d7b39cb 100644 --- a/Lib/encodings/iso8859_7.py +++ b/Lib/encodings/iso8859_7.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\u2018' # 0xA1 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xA2 -> RIGHT SINGLE QUOTATION MARK - u'\xa3' # 0xA3 -> POUND SIGN - u'\u20ac' # 0xA4 -> EURO SIGN - u'\u20af' # 0xA5 -> DRACHMA SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u037a' # 0xAA -> GREEK YPOGEGRAMMENI - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\ufffe' - u'\u2015' # 0xAF -> HORIZONTAL BAR - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\u0384' # 0xB4 -> GREEK TONOS - u'\u0385' # 0xB5 -> GREEK DIALYTIKA TONOS - u'\u0386' # 0xB6 -> GREEK CAPITAL LETTER ALPHA WITH TONOS - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\u0388' # 0xB8 -> GREEK CAPITAL LETTER EPSILON WITH TONOS - u'\u0389' # 0xB9 -> GREEK CAPITAL LETTER ETA WITH TONOS - u'\u038a' # 0xBA -> GREEK CAPITAL LETTER IOTA WITH TONOS - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u038c' # 0xBC -> GREEK CAPITAL LETTER OMICRON WITH TONOS - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\u038e' # 0xBE -> GREEK CAPITAL LETTER UPSILON WITH TONOS - u'\u038f' # 0xBF -> GREEK CAPITAL LETTER OMEGA WITH TONOS - u'\u0390' # 0xC0 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - u'\u0391' # 0xC1 -> GREEK CAPITAL LETTER ALPHA - u'\u0392' # 0xC2 -> GREEK CAPITAL LETTER BETA - u'\u0393' # 0xC3 -> GREEK CAPITAL LETTER GAMMA - u'\u0394' # 0xC4 -> GREEK CAPITAL LETTER DELTA - u'\u0395' # 0xC5 -> GREEK CAPITAL LETTER EPSILON - u'\u0396' # 0xC6 -> GREEK CAPITAL LETTER ZETA - u'\u0397' # 0xC7 -> GREEK CAPITAL LETTER ETA - u'\u0398' # 0xC8 -> GREEK CAPITAL LETTER THETA - u'\u0399' # 0xC9 -> GREEK CAPITAL LETTER IOTA - u'\u039a' # 0xCA -> GREEK CAPITAL LETTER KAPPA - u'\u039b' # 0xCB -> GREEK CAPITAL LETTER LAMDA - u'\u039c' # 0xCC -> GREEK CAPITAL LETTER MU - u'\u039d' # 0xCD -> GREEK CAPITAL LETTER NU - u'\u039e' # 0xCE -> GREEK CAPITAL LETTER XI - u'\u039f' # 0xCF -> GREEK CAPITAL LETTER OMICRON - u'\u03a0' # 0xD0 -> GREEK CAPITAL LETTER PI - u'\u03a1' # 0xD1 -> GREEK CAPITAL LETTER RHO - u'\ufffe' - u'\u03a3' # 0xD3 -> GREEK CAPITAL LETTER SIGMA - u'\u03a4' # 0xD4 -> GREEK CAPITAL LETTER TAU - u'\u03a5' # 0xD5 -> GREEK CAPITAL LETTER UPSILON - u'\u03a6' # 0xD6 -> GREEK CAPITAL LETTER PHI - u'\u03a7' # 0xD7 -> GREEK CAPITAL LETTER CHI - u'\u03a8' # 0xD8 -> GREEK CAPITAL LETTER PSI - u'\u03a9' # 0xD9 -> GREEK CAPITAL LETTER OMEGA - u'\u03aa' # 0xDA -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - u'\u03ab' # 0xDB -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - u'\u03ac' # 0xDC -> GREEK SMALL LETTER ALPHA WITH TONOS - u'\u03ad' # 0xDD -> GREEK SMALL LETTER EPSILON WITH TONOS - u'\u03ae' # 0xDE -> GREEK SMALL LETTER ETA WITH TONOS - u'\u03af' # 0xDF -> GREEK SMALL LETTER IOTA WITH TONOS - u'\u03b0' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - u'\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA - u'\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA - u'\u03b3' # 0xE3 -> GREEK SMALL LETTER GAMMA - u'\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA - u'\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON - u'\u03b6' # 0xE6 -> GREEK SMALL LETTER ZETA - u'\u03b7' # 0xE7 -> GREEK SMALL LETTER ETA - u'\u03b8' # 0xE8 -> GREEK SMALL LETTER THETA - u'\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA - u'\u03ba' # 0xEA -> GREEK SMALL LETTER KAPPA - u'\u03bb' # 0xEB -> GREEK SMALL LETTER LAMDA - u'\u03bc' # 0xEC -> GREEK SMALL LETTER MU - u'\u03bd' # 0xED -> GREEK SMALL LETTER NU - u'\u03be' # 0xEE -> GREEK SMALL LETTER XI - u'\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON - u'\u03c0' # 0xF0 -> GREEK SMALL LETTER PI - u'\u03c1' # 0xF1 -> GREEK SMALL LETTER RHO - u'\u03c2' # 0xF2 -> GREEK SMALL LETTER FINAL SIGMA - u'\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA - u'\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU - u'\u03c5' # 0xF5 -> GREEK SMALL LETTER UPSILON - u'\u03c6' # 0xF6 -> GREEK SMALL LETTER PHI - u'\u03c7' # 0xF7 -> GREEK SMALL LETTER CHI - u'\u03c8' # 0xF8 -> GREEK SMALL LETTER PSI - u'\u03c9' # 0xF9 -> GREEK SMALL LETTER OMEGA - u'\u03ca' # 0xFA -> GREEK SMALL LETTER IOTA WITH DIALYTIKA - u'\u03cb' # 0xFB -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA - u'\u03cc' # 0xFC -> GREEK SMALL LETTER OMICRON WITH TONOS - u'\u03cd' # 0xFD -> GREEK SMALL LETTER UPSILON WITH TONOS - u'\u03ce' # 0xFE -> GREEK SMALL LETTER OMEGA WITH TONOS - u'\ufffe' + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\u2018' # 0xA1 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xA2 -> RIGHT SINGLE QUOTATION MARK + '\xa3' # 0xA3 -> POUND SIGN + '\u20ac' # 0xA4 -> EURO SIGN + '\u20af' # 0xA5 -> DRACHMA SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u037a' # 0xAA -> GREEK YPOGEGRAMMENI + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\ufffe' + '\u2015' # 0xAF -> HORIZONTAL BAR + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\u0384' # 0xB4 -> GREEK TONOS + '\u0385' # 0xB5 -> GREEK DIALYTIKA TONOS + '\u0386' # 0xB6 -> GREEK CAPITAL LETTER ALPHA WITH TONOS + '\xb7' # 0xB7 -> MIDDLE DOT + '\u0388' # 0xB8 -> GREEK CAPITAL LETTER EPSILON WITH TONOS + '\u0389' # 0xB9 -> GREEK CAPITAL LETTER ETA WITH TONOS + '\u038a' # 0xBA -> GREEK CAPITAL LETTER IOTA WITH TONOS + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u038c' # 0xBC -> GREEK CAPITAL LETTER OMICRON WITH TONOS + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\u038e' # 0xBE -> GREEK CAPITAL LETTER UPSILON WITH TONOS + '\u038f' # 0xBF -> GREEK CAPITAL LETTER OMEGA WITH TONOS + '\u0390' # 0xC0 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS + '\u0391' # 0xC1 -> GREEK CAPITAL LETTER ALPHA + '\u0392' # 0xC2 -> GREEK CAPITAL LETTER BETA + '\u0393' # 0xC3 -> GREEK CAPITAL LETTER GAMMA + '\u0394' # 0xC4 -> GREEK CAPITAL LETTER DELTA + '\u0395' # 0xC5 -> GREEK CAPITAL LETTER EPSILON + '\u0396' # 0xC6 -> GREEK CAPITAL LETTER ZETA + '\u0397' # 0xC7 -> GREEK CAPITAL LETTER ETA + '\u0398' # 0xC8 -> GREEK CAPITAL LETTER THETA + '\u0399' # 0xC9 -> GREEK CAPITAL LETTER IOTA + '\u039a' # 0xCA -> GREEK CAPITAL LETTER KAPPA + '\u039b' # 0xCB -> GREEK CAPITAL LETTER LAMDA + '\u039c' # 0xCC -> GREEK CAPITAL LETTER MU + '\u039d' # 0xCD -> GREEK CAPITAL LETTER NU + '\u039e' # 0xCE -> GREEK CAPITAL LETTER XI + '\u039f' # 0xCF -> GREEK CAPITAL LETTER OMICRON + '\u03a0' # 0xD0 -> GREEK CAPITAL LETTER PI + '\u03a1' # 0xD1 -> GREEK CAPITAL LETTER RHO + '\ufffe' + '\u03a3' # 0xD3 -> GREEK CAPITAL LETTER SIGMA + '\u03a4' # 0xD4 -> GREEK CAPITAL LETTER TAU + '\u03a5' # 0xD5 -> GREEK CAPITAL LETTER UPSILON + '\u03a6' # 0xD6 -> GREEK CAPITAL LETTER PHI + '\u03a7' # 0xD7 -> GREEK CAPITAL LETTER CHI + '\u03a8' # 0xD8 -> GREEK CAPITAL LETTER PSI + '\u03a9' # 0xD9 -> GREEK CAPITAL LETTER OMEGA + '\u03aa' # 0xDA -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA + '\u03ab' # 0xDB -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA + '\u03ac' # 0xDC -> GREEK SMALL LETTER ALPHA WITH TONOS + '\u03ad' # 0xDD -> GREEK SMALL LETTER EPSILON WITH TONOS + '\u03ae' # 0xDE -> GREEK SMALL LETTER ETA WITH TONOS + '\u03af' # 0xDF -> GREEK SMALL LETTER IOTA WITH TONOS + '\u03b0' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS + '\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA + '\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA + '\u03b3' # 0xE3 -> GREEK SMALL LETTER GAMMA + '\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA + '\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON + '\u03b6' # 0xE6 -> GREEK SMALL LETTER ZETA + '\u03b7' # 0xE7 -> GREEK SMALL LETTER ETA + '\u03b8' # 0xE8 -> GREEK SMALL LETTER THETA + '\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA + '\u03ba' # 0xEA -> GREEK SMALL LETTER KAPPA + '\u03bb' # 0xEB -> GREEK SMALL LETTER LAMDA + '\u03bc' # 0xEC -> GREEK SMALL LETTER MU + '\u03bd' # 0xED -> GREEK SMALL LETTER NU + '\u03be' # 0xEE -> GREEK SMALL LETTER XI + '\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON + '\u03c0' # 0xF0 -> GREEK SMALL LETTER PI + '\u03c1' # 0xF1 -> GREEK SMALL LETTER RHO + '\u03c2' # 0xF2 -> GREEK SMALL LETTER FINAL SIGMA + '\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA + '\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU + '\u03c5' # 0xF5 -> GREEK SMALL LETTER UPSILON + '\u03c6' # 0xF6 -> GREEK SMALL LETTER PHI + '\u03c7' # 0xF7 -> GREEK SMALL LETTER CHI + '\u03c8' # 0xF8 -> GREEK SMALL LETTER PSI + '\u03c9' # 0xF9 -> GREEK SMALL LETTER OMEGA + '\u03ca' # 0xFA -> GREEK SMALL LETTER IOTA WITH DIALYTIKA + '\u03cb' # 0xFB -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA + '\u03cc' # 0xFC -> GREEK SMALL LETTER OMICRON WITH TONOS + '\u03cd' # 0xFD -> GREEK SMALL LETTER UPSILON WITH TONOS + '\u03ce' # 0xFE -> GREEK SMALL LETTER OMEGA WITH TONOS + '\ufffe' ) ### Encoding table diff --git a/Lib/encodings/iso8859_8.py b/Lib/encodings/iso8859_8.py index 43cf213..8184902 100644 --- a/Lib/encodings/iso8859_8.py +++ b/Lib/encodings/iso8859_8.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\ufffe' - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xd7' # 0xAA -> MULTIPLICATION SIGN - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xf7' # 0xBA -> DIVISION SIGN - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u2017' # 0xDF -> DOUBLE LOW LINE - u'\u05d0' # 0xE0 -> HEBREW LETTER ALEF - u'\u05d1' # 0xE1 -> HEBREW LETTER BET - u'\u05d2' # 0xE2 -> HEBREW LETTER GIMEL - u'\u05d3' # 0xE3 -> HEBREW LETTER DALET - u'\u05d4' # 0xE4 -> HEBREW LETTER HE - u'\u05d5' # 0xE5 -> HEBREW LETTER VAV - u'\u05d6' # 0xE6 -> HEBREW LETTER ZAYIN - u'\u05d7' # 0xE7 -> HEBREW LETTER HET - u'\u05d8' # 0xE8 -> HEBREW LETTER TET - u'\u05d9' # 0xE9 -> HEBREW LETTER YOD - u'\u05da' # 0xEA -> HEBREW LETTER FINAL KAF - u'\u05db' # 0xEB -> HEBREW LETTER KAF - u'\u05dc' # 0xEC -> HEBREW LETTER LAMED - u'\u05dd' # 0xED -> HEBREW LETTER FINAL MEM - u'\u05de' # 0xEE -> HEBREW LETTER MEM - u'\u05df' # 0xEF -> HEBREW LETTER FINAL NUN - u'\u05e0' # 0xF0 -> HEBREW LETTER NUN - u'\u05e1' # 0xF1 -> HEBREW LETTER SAMEKH - u'\u05e2' # 0xF2 -> HEBREW LETTER AYIN - u'\u05e3' # 0xF3 -> HEBREW LETTER FINAL PE - u'\u05e4' # 0xF4 -> HEBREW LETTER PE - u'\u05e5' # 0xF5 -> HEBREW LETTER FINAL TSADI - u'\u05e6' # 0xF6 -> HEBREW LETTER TSADI - u'\u05e7' # 0xF7 -> HEBREW LETTER QOF - u'\u05e8' # 0xF8 -> HEBREW LETTER RESH - u'\u05e9' # 0xF9 -> HEBREW LETTER SHIN - u'\u05ea' # 0xFA -> HEBREW LETTER TAV - u'\ufffe' - u'\ufffe' - u'\u200e' # 0xFD -> LEFT-TO-RIGHT MARK - u'\u200f' # 0xFE -> RIGHT-TO-LEFT MARK - u'\ufffe' + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\ufffe' + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xd7' # 0xAA -> MULTIPLICATION SIGN + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xf7' # 0xBA -> DIVISION SIGN + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\u2017' # 0xDF -> DOUBLE LOW LINE + '\u05d0' # 0xE0 -> HEBREW LETTER ALEF + '\u05d1' # 0xE1 -> HEBREW LETTER BET + '\u05d2' # 0xE2 -> HEBREW LETTER GIMEL + '\u05d3' # 0xE3 -> HEBREW LETTER DALET + '\u05d4' # 0xE4 -> HEBREW LETTER HE + '\u05d5' # 0xE5 -> HEBREW LETTER VAV + '\u05d6' # 0xE6 -> HEBREW LETTER ZAYIN + '\u05d7' # 0xE7 -> HEBREW LETTER HET + '\u05d8' # 0xE8 -> HEBREW LETTER TET + '\u05d9' # 0xE9 -> HEBREW LETTER YOD + '\u05da' # 0xEA -> HEBREW LETTER FINAL KAF + '\u05db' # 0xEB -> HEBREW LETTER KAF + '\u05dc' # 0xEC -> HEBREW LETTER LAMED + '\u05dd' # 0xED -> HEBREW LETTER FINAL MEM + '\u05de' # 0xEE -> HEBREW LETTER MEM + '\u05df' # 0xEF -> HEBREW LETTER FINAL NUN + '\u05e0' # 0xF0 -> HEBREW LETTER NUN + '\u05e1' # 0xF1 -> HEBREW LETTER SAMEKH + '\u05e2' # 0xF2 -> HEBREW LETTER AYIN + '\u05e3' # 0xF3 -> HEBREW LETTER FINAL PE + '\u05e4' # 0xF4 -> HEBREW LETTER PE + '\u05e5' # 0xF5 -> HEBREW LETTER FINAL TSADI + '\u05e6' # 0xF6 -> HEBREW LETTER TSADI + '\u05e7' # 0xF7 -> HEBREW LETTER QOF + '\u05e8' # 0xF8 -> HEBREW LETTER RESH + '\u05e9' # 0xF9 -> HEBREW LETTER SHIN + '\u05ea' # 0xFA -> HEBREW LETTER TAV + '\ufffe' + '\ufffe' + '\u200e' # 0xFD -> LEFT-TO-RIGHT MARK + '\u200f' # 0xFE -> RIGHT-TO-LEFT MARK + '\ufffe' ) ### Encoding table diff --git a/Lib/encodings/iso8859_9.py b/Lib/encodings/iso8859_9.py index b802938..e539fdd 100644 --- a/Lib/encodings/iso8859_9.py +++ b/Lib/encodings/iso8859_9.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\xa0' # 0xA0 -> NO-BREAK SPACE - u'\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa4' # 0xA4 -> CURRENCY SIGN - u'\xa5' # 0xA5 -> YEN SIGN - u'\xa6' # 0xA6 -> BROKEN BAR - u'\xa7' # 0xA7 -> SECTION SIGN - u'\xa8' # 0xA8 -> DIAERESIS - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR - u'\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xac' # 0xAC -> NOT SIGN - u'\xad' # 0xAD -> SOFT HYPHEN - u'\xae' # 0xAE -> REGISTERED SIGN - u'\xaf' # 0xAF -> MACRON - u'\xb0' # 0xB0 -> DEGREE SIGN - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\xb2' # 0xB2 -> SUPERSCRIPT TWO - u'\xb3' # 0xB3 -> SUPERSCRIPT THREE - u'\xb4' # 0xB4 -> ACUTE ACCENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\xb6' # 0xB6 -> PILCROW SIGN - u'\xb7' # 0xB7 -> MIDDLE DOT - u'\xb8' # 0xB8 -> CEDILLA - u'\xb9' # 0xB9 -> SUPERSCRIPT ONE - u'\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR - u'\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER - u'\xbd' # 0xBD -> VULGAR FRACTION ONE HALF - u'\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS - u'\xbf' # 0xBF -> INVERTED QUESTION MARK - u'\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE - u'\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE - u'\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\u011e' # 0xD0 -> LATIN CAPITAL LETTER G WITH BREVE - u'\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE - u'\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xd7' # 0xD7 -> MULTIPLICATION SIGN - u'\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE - u'\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u0130' # 0xDD -> LATIN CAPITAL LETTER I WITH DOT ABOVE - u'\u015e' # 0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S - u'\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE - u'\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe6' # 0xE6 -> LATIN SMALL LETTER AE - u'\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE - u'\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE - u'\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE - u'\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE - u'\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u011f' # 0xF0 -> LATIN SMALL LETTER G WITH BREVE - u'\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE - u'\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE - u'\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0xF7 -> DIVISION SIGN - u'\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE - u'\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE - u'\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE - u'\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u0131' # 0xFD -> LATIN SMALL LETTER DOTLESS I - u'\u015f' # 0xFE -> LATIN SMALL LETTER S WITH CEDILLA - u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\xa0' # 0xA0 -> NO-BREAK SPACE + '\xa1' # 0xA1 -> INVERTED EXCLAMATION MARK + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa4' # 0xA4 -> CURRENCY SIGN + '\xa5' # 0xA5 -> YEN SIGN + '\xa6' # 0xA6 -> BROKEN BAR + '\xa7' # 0xA7 -> SECTION SIGN + '\xa8' # 0xA8 -> DIAERESIS + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\xaa' # 0xAA -> FEMININE ORDINAL INDICATOR + '\xab' # 0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xac' # 0xAC -> NOT SIGN + '\xad' # 0xAD -> SOFT HYPHEN + '\xae' # 0xAE -> REGISTERED SIGN + '\xaf' # 0xAF -> MACRON + '\xb0' # 0xB0 -> DEGREE SIGN + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\xb2' # 0xB2 -> SUPERSCRIPT TWO + '\xb3' # 0xB3 -> SUPERSCRIPT THREE + '\xb4' # 0xB4 -> ACUTE ACCENT + '\xb5' # 0xB5 -> MICRO SIGN + '\xb6' # 0xB6 -> PILCROW SIGN + '\xb7' # 0xB7 -> MIDDLE DOT + '\xb8' # 0xB8 -> CEDILLA + '\xb9' # 0xB9 -> SUPERSCRIPT ONE + '\xba' # 0xBA -> MASCULINE ORDINAL INDICATOR + '\xbb' # 0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbc' # 0xBC -> VULGAR FRACTION ONE QUARTER + '\xbd' # 0xBD -> VULGAR FRACTION ONE HALF + '\xbe' # 0xBE -> VULGAR FRACTION THREE QUARTERS + '\xbf' # 0xBF -> INVERTED QUESTION MARK + '\xc0' # 0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE + '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE + '\xc7' # 0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc8' # 0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xca' # 0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xcc' # 0xCC -> LATIN CAPITAL LETTER I WITH GRAVE + '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\u011e' # 0xD0 -> LATIN CAPITAL LETTER G WITH BREVE + '\xd1' # 0xD1 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd2' # 0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xd3' # 0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE + '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xd7' # 0xD7 -> MULTIPLICATION SIGN + '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE + '\xd9' # 0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE + '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u0130' # 0xDD -> LATIN CAPITAL LETTER I WITH DOT ABOVE + '\u015e' # 0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA + '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S + '\xe0' # 0xE0 -> LATIN SMALL LETTER A WITH GRAVE + '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE + '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE + '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe6' # 0xE6 -> LATIN SMALL LETTER AE + '\xe7' # 0xE7 -> LATIN SMALL LETTER C WITH CEDILLA + '\xe8' # 0xE8 -> LATIN SMALL LETTER E WITH GRAVE + '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE + '\xea' # 0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS + '\xec' # 0xEC -> LATIN SMALL LETTER I WITH GRAVE + '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE + '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0xEF -> LATIN SMALL LETTER I WITH DIAERESIS + '\u011f' # 0xF0 -> LATIN SMALL LETTER G WITH BREVE + '\xf1' # 0xF1 -> LATIN SMALL LETTER N WITH TILDE + '\xf2' # 0xF2 -> LATIN SMALL LETTER O WITH GRAVE + '\xf3' # 0xF3 -> LATIN SMALL LETTER O WITH ACUTE + '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE + '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0xF7 -> DIVISION SIGN + '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE + '\xf9' # 0xF9 -> LATIN SMALL LETTER U WITH GRAVE + '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE + '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS + '\u0131' # 0xFD -> LATIN SMALL LETTER DOTLESS I + '\u015f' # 0xFE -> LATIN SMALL LETTER S WITH CEDILLA + '\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) ### Encoding table diff --git a/Lib/encodings/koi8_r.py b/Lib/encodings/koi8_r.py index f9eb82c..41ddde8 100644 --- a/Lib/encodings/koi8_r.py +++ b/Lib/encodings/koi8_r.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u2500' # 0x80 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u2502' # 0x81 -> BOX DRAWINGS LIGHT VERTICAL - u'\u250c' # 0x82 -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2510' # 0x83 -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x84 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2518' # 0x85 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u251c' # 0x86 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2524' # 0x87 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u252c' # 0x88 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u2534' # 0x89 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u253c' # 0x8A -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u2580' # 0x8B -> UPPER HALF BLOCK - u'\u2584' # 0x8C -> LOWER HALF BLOCK - u'\u2588' # 0x8D -> FULL BLOCK - u'\u258c' # 0x8E -> LEFT HALF BLOCK - u'\u2590' # 0x8F -> RIGHT HALF BLOCK - u'\u2591' # 0x90 -> LIGHT SHADE - u'\u2592' # 0x91 -> MEDIUM SHADE - u'\u2593' # 0x92 -> DARK SHADE - u'\u2320' # 0x93 -> TOP HALF INTEGRAL - u'\u25a0' # 0x94 -> BLACK SQUARE - u'\u2219' # 0x95 -> BULLET OPERATOR - u'\u221a' # 0x96 -> SQUARE ROOT - u'\u2248' # 0x97 -> ALMOST EQUAL TO - u'\u2264' # 0x98 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0x99 -> GREATER-THAN OR EQUAL TO - u'\xa0' # 0x9A -> NO-BREAK SPACE - u'\u2321' # 0x9B -> BOTTOM HALF INTEGRAL - u'\xb0' # 0x9C -> DEGREE SIGN - u'\xb2' # 0x9D -> SUPERSCRIPT TWO - u'\xb7' # 0x9E -> MIDDLE DOT - u'\xf7' # 0x9F -> DIVISION SIGN - u'\u2550' # 0xA0 -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u2551' # 0xA1 -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2552' # 0xA2 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u0451' # 0xA3 -> CYRILLIC SMALL LETTER IO - u'\u2553' # 0xA4 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - u'\u2554' # 0xA5 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u2555' # 0xA6 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - u'\u2556' # 0xA7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - u'\u2557' # 0xA8 -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u2558' # 0xA9 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2559' # 0xAA -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u255a' # 0xAB -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u255b' # 0xAC -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u255c' # 0xAD -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - u'\u255d' # 0xAE -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255e' # 0xAF -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0xB0 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u2560' # 0xB1 -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2561' # 0xB2 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u0401' # 0xB3 -> CYRILLIC CAPITAL LETTER IO - u'\u2562' # 0xB4 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - u'\u2563' # 0xB5 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u2564' # 0xB6 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - u'\u2565' # 0xB7 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - u'\u2566' # 0xB8 -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2567' # 0xB9 -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0xBA -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2569' # 0xBB -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u256a' # 0xBC -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u256b' # 0xBD -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - u'\u256c' # 0xBE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa9' # 0xBF -> COPYRIGHT SIGN - u'\u044e' # 0xC0 -> CYRILLIC SMALL LETTER YU - u'\u0430' # 0xC1 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0xC2 -> CYRILLIC SMALL LETTER BE - u'\u0446' # 0xC3 -> CYRILLIC SMALL LETTER TSE - u'\u0434' # 0xC4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0xC5 -> CYRILLIC SMALL LETTER IE - u'\u0444' # 0xC6 -> CYRILLIC SMALL LETTER EF - u'\u0433' # 0xC7 -> CYRILLIC SMALL LETTER GHE - u'\u0445' # 0xC8 -> CYRILLIC SMALL LETTER HA - u'\u0438' # 0xC9 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0xCA -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0xCB -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0xCC -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0xCD -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0xCE -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0xCF -> CYRILLIC SMALL LETTER O - u'\u043f' # 0xD0 -> CYRILLIC SMALL LETTER PE - u'\u044f' # 0xD1 -> CYRILLIC SMALL LETTER YA - u'\u0440' # 0xD2 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0xD3 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0xD4 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0xD5 -> CYRILLIC SMALL LETTER U - u'\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE - u'\u0432' # 0xD7 -> CYRILLIC SMALL LETTER VE - u'\u044c' # 0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044b' # 0xD9 -> CYRILLIC SMALL LETTER YERU - u'\u0437' # 0xDA -> CYRILLIC SMALL LETTER ZE - u'\u0448' # 0xDB -> CYRILLIC SMALL LETTER SHA - u'\u044d' # 0xDC -> CYRILLIC SMALL LETTER E - u'\u0449' # 0xDD -> CYRILLIC SMALL LETTER SHCHA - u'\u0447' # 0xDE -> CYRILLIC SMALL LETTER CHE - u'\u044a' # 0xDF -> CYRILLIC SMALL LETTER HARD SIGN - u'\u042e' # 0xE0 -> CYRILLIC CAPITAL LETTER YU - u'\u0410' # 0xE1 -> CYRILLIC CAPITAL LETTER A - u'\u0411' # 0xE2 -> CYRILLIC CAPITAL LETTER BE - u'\u0426' # 0xE3 -> CYRILLIC CAPITAL LETTER TSE - u'\u0414' # 0xE4 -> CYRILLIC CAPITAL LETTER DE - u'\u0415' # 0xE5 -> CYRILLIC CAPITAL LETTER IE - u'\u0424' # 0xE6 -> CYRILLIC CAPITAL LETTER EF - u'\u0413' # 0xE7 -> CYRILLIC CAPITAL LETTER GHE - u'\u0425' # 0xE8 -> CYRILLIC CAPITAL LETTER HA - u'\u0418' # 0xE9 -> CYRILLIC CAPITAL LETTER I - u'\u0419' # 0xEA -> CYRILLIC CAPITAL LETTER SHORT I - u'\u041a' # 0xEB -> CYRILLIC CAPITAL LETTER KA - u'\u041b' # 0xEC -> CYRILLIC CAPITAL LETTER EL - u'\u041c' # 0xED -> CYRILLIC CAPITAL LETTER EM - u'\u041d' # 0xEE -> CYRILLIC CAPITAL LETTER EN - u'\u041e' # 0xEF -> CYRILLIC CAPITAL LETTER O - u'\u041f' # 0xF0 -> CYRILLIC CAPITAL LETTER PE - u'\u042f' # 0xF1 -> CYRILLIC CAPITAL LETTER YA - u'\u0420' # 0xF2 -> CYRILLIC CAPITAL LETTER ER - u'\u0421' # 0xF3 -> CYRILLIC CAPITAL LETTER ES - u'\u0422' # 0xF4 -> CYRILLIC CAPITAL LETTER TE - u'\u0423' # 0xF5 -> CYRILLIC CAPITAL LETTER U - u'\u0416' # 0xF6 -> CYRILLIC CAPITAL LETTER ZHE - u'\u0412' # 0xF7 -> CYRILLIC CAPITAL LETTER VE - u'\u042c' # 0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u042b' # 0xF9 -> CYRILLIC CAPITAL LETTER YERU - u'\u0417' # 0xFA -> CYRILLIC CAPITAL LETTER ZE - u'\u0428' # 0xFB -> CYRILLIC CAPITAL LETTER SHA - u'\u042d' # 0xFC -> CYRILLIC CAPITAL LETTER E - u'\u0429' # 0xFD -> CYRILLIC CAPITAL LETTER SHCHA - u'\u0427' # 0xFE -> CYRILLIC CAPITAL LETTER CHE - u'\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u2500' # 0x80 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u2502' # 0x81 -> BOX DRAWINGS LIGHT VERTICAL + '\u250c' # 0x82 -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2510' # 0x83 -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x84 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2518' # 0x85 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u251c' # 0x86 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2524' # 0x87 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u252c' # 0x88 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u2534' # 0x89 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u253c' # 0x8A -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u2580' # 0x8B -> UPPER HALF BLOCK + '\u2584' # 0x8C -> LOWER HALF BLOCK + '\u2588' # 0x8D -> FULL BLOCK + '\u258c' # 0x8E -> LEFT HALF BLOCK + '\u2590' # 0x8F -> RIGHT HALF BLOCK + '\u2591' # 0x90 -> LIGHT SHADE + '\u2592' # 0x91 -> MEDIUM SHADE + '\u2593' # 0x92 -> DARK SHADE + '\u2320' # 0x93 -> TOP HALF INTEGRAL + '\u25a0' # 0x94 -> BLACK SQUARE + '\u2219' # 0x95 -> BULLET OPERATOR + '\u221a' # 0x96 -> SQUARE ROOT + '\u2248' # 0x97 -> ALMOST EQUAL TO + '\u2264' # 0x98 -> LESS-THAN OR EQUAL TO + '\u2265' # 0x99 -> GREATER-THAN OR EQUAL TO + '\xa0' # 0x9A -> NO-BREAK SPACE + '\u2321' # 0x9B -> BOTTOM HALF INTEGRAL + '\xb0' # 0x9C -> DEGREE SIGN + '\xb2' # 0x9D -> SUPERSCRIPT TWO + '\xb7' # 0x9E -> MIDDLE DOT + '\xf7' # 0x9F -> DIVISION SIGN + '\u2550' # 0xA0 -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u2551' # 0xA1 -> BOX DRAWINGS DOUBLE VERTICAL + '\u2552' # 0xA2 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u0451' # 0xA3 -> CYRILLIC SMALL LETTER IO + '\u2553' # 0xA4 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + '\u2554' # 0xA5 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u2555' # 0xA6 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + '\u2556' # 0xA7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + '\u2557' # 0xA8 -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u2558' # 0xA9 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2559' # 0xAA -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u255a' # 0xAB -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u255b' # 0xAC -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u255c' # 0xAD -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + '\u255d' # 0xAE -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255e' # 0xAF -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0xB0 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u2560' # 0xB1 -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2561' # 0xB2 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u0401' # 0xB3 -> CYRILLIC CAPITAL LETTER IO + '\u2562' # 0xB4 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + '\u2563' # 0xB5 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u2564' # 0xB6 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + '\u2565' # 0xB7 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + '\u2566' # 0xB8 -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2567' # 0xB9 -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0xBA -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2569' # 0xBB -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u256a' # 0xBC -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u256b' # 0xBD -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + '\u256c' # 0xBE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa9' # 0xBF -> COPYRIGHT SIGN + '\u044e' # 0xC0 -> CYRILLIC SMALL LETTER YU + '\u0430' # 0xC1 -> CYRILLIC SMALL LETTER A + '\u0431' # 0xC2 -> CYRILLIC SMALL LETTER BE + '\u0446' # 0xC3 -> CYRILLIC SMALL LETTER TSE + '\u0434' # 0xC4 -> CYRILLIC SMALL LETTER DE + '\u0435' # 0xC5 -> CYRILLIC SMALL LETTER IE + '\u0444' # 0xC6 -> CYRILLIC SMALL LETTER EF + '\u0433' # 0xC7 -> CYRILLIC SMALL LETTER GHE + '\u0445' # 0xC8 -> CYRILLIC SMALL LETTER HA + '\u0438' # 0xC9 -> CYRILLIC SMALL LETTER I + '\u0439' # 0xCA -> CYRILLIC SMALL LETTER SHORT I + '\u043a' # 0xCB -> CYRILLIC SMALL LETTER KA + '\u043b' # 0xCC -> CYRILLIC SMALL LETTER EL + '\u043c' # 0xCD -> CYRILLIC SMALL LETTER EM + '\u043d' # 0xCE -> CYRILLIC SMALL LETTER EN + '\u043e' # 0xCF -> CYRILLIC SMALL LETTER O + '\u043f' # 0xD0 -> CYRILLIC SMALL LETTER PE + '\u044f' # 0xD1 -> CYRILLIC SMALL LETTER YA + '\u0440' # 0xD2 -> CYRILLIC SMALL LETTER ER + '\u0441' # 0xD3 -> CYRILLIC SMALL LETTER ES + '\u0442' # 0xD4 -> CYRILLIC SMALL LETTER TE + '\u0443' # 0xD5 -> CYRILLIC SMALL LETTER U + '\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE + '\u0432' # 0xD7 -> CYRILLIC SMALL LETTER VE + '\u044c' # 0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN + '\u044b' # 0xD9 -> CYRILLIC SMALL LETTER YERU + '\u0437' # 0xDA -> CYRILLIC SMALL LETTER ZE + '\u0448' # 0xDB -> CYRILLIC SMALL LETTER SHA + '\u044d' # 0xDC -> CYRILLIC SMALL LETTER E + '\u0449' # 0xDD -> CYRILLIC SMALL LETTER SHCHA + '\u0447' # 0xDE -> CYRILLIC SMALL LETTER CHE + '\u044a' # 0xDF -> CYRILLIC SMALL LETTER HARD SIGN + '\u042e' # 0xE0 -> CYRILLIC CAPITAL LETTER YU + '\u0410' # 0xE1 -> CYRILLIC CAPITAL LETTER A + '\u0411' # 0xE2 -> CYRILLIC CAPITAL LETTER BE + '\u0426' # 0xE3 -> CYRILLIC CAPITAL LETTER TSE + '\u0414' # 0xE4 -> CYRILLIC CAPITAL LETTER DE + '\u0415' # 0xE5 -> CYRILLIC CAPITAL LETTER IE + '\u0424' # 0xE6 -> CYRILLIC CAPITAL LETTER EF + '\u0413' # 0xE7 -> CYRILLIC CAPITAL LETTER GHE + '\u0425' # 0xE8 -> CYRILLIC CAPITAL LETTER HA + '\u0418' # 0xE9 -> CYRILLIC CAPITAL LETTER I + '\u0419' # 0xEA -> CYRILLIC CAPITAL LETTER SHORT I + '\u041a' # 0xEB -> CYRILLIC CAPITAL LETTER KA + '\u041b' # 0xEC -> CYRILLIC CAPITAL LETTER EL + '\u041c' # 0xED -> CYRILLIC CAPITAL LETTER EM + '\u041d' # 0xEE -> CYRILLIC CAPITAL LETTER EN + '\u041e' # 0xEF -> CYRILLIC CAPITAL LETTER O + '\u041f' # 0xF0 -> CYRILLIC CAPITAL LETTER PE + '\u042f' # 0xF1 -> CYRILLIC CAPITAL LETTER YA + '\u0420' # 0xF2 -> CYRILLIC CAPITAL LETTER ER + '\u0421' # 0xF3 -> CYRILLIC CAPITAL LETTER ES + '\u0422' # 0xF4 -> CYRILLIC CAPITAL LETTER TE + '\u0423' # 0xF5 -> CYRILLIC CAPITAL LETTER U + '\u0416' # 0xF6 -> CYRILLIC CAPITAL LETTER ZHE + '\u0412' # 0xF7 -> CYRILLIC CAPITAL LETTER VE + '\u042c' # 0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u042b' # 0xF9 -> CYRILLIC CAPITAL LETTER YERU + '\u0417' # 0xFA -> CYRILLIC CAPITAL LETTER ZE + '\u0428' # 0xFB -> CYRILLIC CAPITAL LETTER SHA + '\u042d' # 0xFC -> CYRILLIC CAPITAL LETTER E + '\u0429' # 0xFD -> CYRILLIC CAPITAL LETTER SHCHA + '\u0427' # 0xFE -> CYRILLIC CAPITAL LETTER CHE + '\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN ) ### Encoding table diff --git a/Lib/encodings/koi8_u.py b/Lib/encodings/koi8_u.py index a9317b1..f9e3fae 100644 --- a/Lib/encodings/koi8_u.py +++ b/Lib/encodings/koi8_u.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\u2500' # 0x80 -> BOX DRAWINGS LIGHT HORIZONTAL - u'\u2502' # 0x81 -> BOX DRAWINGS LIGHT VERTICAL - u'\u250c' # 0x82 -> BOX DRAWINGS LIGHT DOWN AND RIGHT - u'\u2510' # 0x83 -> BOX DRAWINGS LIGHT DOWN AND LEFT - u'\u2514' # 0x84 -> BOX DRAWINGS LIGHT UP AND RIGHT - u'\u2518' # 0x85 -> BOX DRAWINGS LIGHT UP AND LEFT - u'\u251c' # 0x86 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT - u'\u2524' # 0x87 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT - u'\u252c' # 0x88 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - u'\u2534' # 0x89 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL - u'\u253c' # 0x8A -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - u'\u2580' # 0x8B -> UPPER HALF BLOCK - u'\u2584' # 0x8C -> LOWER HALF BLOCK - u'\u2588' # 0x8D -> FULL BLOCK - u'\u258c' # 0x8E -> LEFT HALF BLOCK - u'\u2590' # 0x8F -> RIGHT HALF BLOCK - u'\u2591' # 0x90 -> LIGHT SHADE - u'\u2592' # 0x91 -> MEDIUM SHADE - u'\u2593' # 0x92 -> DARK SHADE - u'\u2320' # 0x93 -> TOP HALF INTEGRAL - u'\u25a0' # 0x94 -> BLACK SQUARE - u'\u2219' # 0x95 -> BULLET OPERATOR - u'\u221a' # 0x96 -> SQUARE ROOT - u'\u2248' # 0x97 -> ALMOST EQUAL TO - u'\u2264' # 0x98 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0x99 -> GREATER-THAN OR EQUAL TO - u'\xa0' # 0x9A -> NO-BREAK SPACE - u'\u2321' # 0x9B -> BOTTOM HALF INTEGRAL - u'\xb0' # 0x9C -> DEGREE SIGN - u'\xb2' # 0x9D -> SUPERSCRIPT TWO - u'\xb7' # 0x9E -> MIDDLE DOT - u'\xf7' # 0x9F -> DIVISION SIGN - u'\u2550' # 0xA0 -> BOX DRAWINGS DOUBLE HORIZONTAL - u'\u2551' # 0xA1 -> BOX DRAWINGS DOUBLE VERTICAL - u'\u2552' # 0xA2 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - u'\u0451' # 0xA3 -> CYRILLIC SMALL LETTER IO - u'\u0454' # 0xA4 -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\u2554' # 0xA5 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT - u'\u0456' # 0xA6 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0457' # 0xA7 -> CYRILLIC SMALL LETTER YI (UKRAINIAN) - u'\u2557' # 0xA8 -> BOX DRAWINGS DOUBLE DOWN AND LEFT - u'\u2558' # 0xA9 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - u'\u2559' # 0xAA -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - u'\u255a' # 0xAB -> BOX DRAWINGS DOUBLE UP AND RIGHT - u'\u255b' # 0xAC -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - u'\u0491' # 0xAD -> CYRILLIC SMALL LETTER UKRAINIAN GHE WITH UPTURN - u'\u255d' # 0xAE -> BOX DRAWINGS DOUBLE UP AND LEFT - u'\u255e' # 0xAF -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - u'\u255f' # 0xB0 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - u'\u2560' # 0xB1 -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - u'\u2561' # 0xB2 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - u'\u0401' # 0xB3 -> CYRILLIC CAPITAL LETTER IO - u'\u0404' # 0xB4 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\u2563' # 0xB5 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT - u'\u0406' # 0xB6 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - u'\u0407' # 0xB7 -> CYRILLIC CAPITAL LETTER YI (UKRAINIAN) - u'\u2566' # 0xB8 -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - u'\u2567' # 0xB9 -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - u'\u2568' # 0xBA -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - u'\u2569' # 0xBB -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL - u'\u256a' # 0xBC -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - u'\u0490' # 0xBD -> CYRILLIC CAPITAL LETTER UKRAINIAN GHE WITH UPTURN - u'\u256c' # 0xBE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - u'\xa9' # 0xBF -> COPYRIGHT SIGN - u'\u044e' # 0xC0 -> CYRILLIC SMALL LETTER YU - u'\u0430' # 0xC1 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0xC2 -> CYRILLIC SMALL LETTER BE - u'\u0446' # 0xC3 -> CYRILLIC SMALL LETTER TSE - u'\u0434' # 0xC4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0xC5 -> CYRILLIC SMALL LETTER IE - u'\u0444' # 0xC6 -> CYRILLIC SMALL LETTER EF - u'\u0433' # 0xC7 -> CYRILLIC SMALL LETTER GHE - u'\u0445' # 0xC8 -> CYRILLIC SMALL LETTER HA - u'\u0438' # 0xC9 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0xCA -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0xCB -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0xCC -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0xCD -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0xCE -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0xCF -> CYRILLIC SMALL LETTER O - u'\u043f' # 0xD0 -> CYRILLIC SMALL LETTER PE - u'\u044f' # 0xD1 -> CYRILLIC SMALL LETTER YA - u'\u0440' # 0xD2 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0xD3 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0xD4 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0xD5 -> CYRILLIC SMALL LETTER U - u'\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE - u'\u0432' # 0xD7 -> CYRILLIC SMALL LETTER VE - u'\u044c' # 0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044b' # 0xD9 -> CYRILLIC SMALL LETTER YERU - u'\u0437' # 0xDA -> CYRILLIC SMALL LETTER ZE - u'\u0448' # 0xDB -> CYRILLIC SMALL LETTER SHA - u'\u044d' # 0xDC -> CYRILLIC SMALL LETTER E - u'\u0449' # 0xDD -> CYRILLIC SMALL LETTER SHCHA - u'\u0447' # 0xDE -> CYRILLIC SMALL LETTER CHE - u'\u044a' # 0xDF -> CYRILLIC SMALL LETTER HARD SIGN - u'\u042e' # 0xE0 -> CYRILLIC CAPITAL LETTER YU - u'\u0410' # 0xE1 -> CYRILLIC CAPITAL LETTER A - u'\u0411' # 0xE2 -> CYRILLIC CAPITAL LETTER BE - u'\u0426' # 0xE3 -> CYRILLIC CAPITAL LETTER TSE - u'\u0414' # 0xE4 -> CYRILLIC CAPITAL LETTER DE - u'\u0415' # 0xE5 -> CYRILLIC CAPITAL LETTER IE - u'\u0424' # 0xE6 -> CYRILLIC CAPITAL LETTER EF - u'\u0413' # 0xE7 -> CYRILLIC CAPITAL LETTER GHE - u'\u0425' # 0xE8 -> CYRILLIC CAPITAL LETTER HA - u'\u0418' # 0xE9 -> CYRILLIC CAPITAL LETTER I - u'\u0419' # 0xEA -> CYRILLIC CAPITAL LETTER SHORT I - u'\u041a' # 0xEB -> CYRILLIC CAPITAL LETTER KA - u'\u041b' # 0xEC -> CYRILLIC CAPITAL LETTER EL - u'\u041c' # 0xED -> CYRILLIC CAPITAL LETTER EM - u'\u041d' # 0xEE -> CYRILLIC CAPITAL LETTER EN - u'\u041e' # 0xEF -> CYRILLIC CAPITAL LETTER O - u'\u041f' # 0xF0 -> CYRILLIC CAPITAL LETTER PE - u'\u042f' # 0xF1 -> CYRILLIC CAPITAL LETTER YA - u'\u0420' # 0xF2 -> CYRILLIC CAPITAL LETTER ER - u'\u0421' # 0xF3 -> CYRILLIC CAPITAL LETTER ES - u'\u0422' # 0xF4 -> CYRILLIC CAPITAL LETTER TE - u'\u0423' # 0xF5 -> CYRILLIC CAPITAL LETTER U - u'\u0416' # 0xF6 -> CYRILLIC CAPITAL LETTER ZHE - u'\u0412' # 0xF7 -> CYRILLIC CAPITAL LETTER VE - u'\u042c' # 0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u042b' # 0xF9 -> CYRILLIC CAPITAL LETTER YERU - u'\u0417' # 0xFA -> CYRILLIC CAPITAL LETTER ZE - u'\u0428' # 0xFB -> CYRILLIC CAPITAL LETTER SHA - u'\u042d' # 0xFC -> CYRILLIC CAPITAL LETTER E - u'\u0429' # 0xFD -> CYRILLIC CAPITAL LETTER SHCHA - u'\u0427' # 0xFE -> CYRILLIC CAPITAL LETTER CHE - u'\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\u2500' # 0x80 -> BOX DRAWINGS LIGHT HORIZONTAL + '\u2502' # 0x81 -> BOX DRAWINGS LIGHT VERTICAL + '\u250c' # 0x82 -> BOX DRAWINGS LIGHT DOWN AND RIGHT + '\u2510' # 0x83 -> BOX DRAWINGS LIGHT DOWN AND LEFT + '\u2514' # 0x84 -> BOX DRAWINGS LIGHT UP AND RIGHT + '\u2518' # 0x85 -> BOX DRAWINGS LIGHT UP AND LEFT + '\u251c' # 0x86 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT + '\u2524' # 0x87 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT + '\u252c' # 0x88 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + '\u2534' # 0x89 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL + '\u253c' # 0x8A -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + '\u2580' # 0x8B -> UPPER HALF BLOCK + '\u2584' # 0x8C -> LOWER HALF BLOCK + '\u2588' # 0x8D -> FULL BLOCK + '\u258c' # 0x8E -> LEFT HALF BLOCK + '\u2590' # 0x8F -> RIGHT HALF BLOCK + '\u2591' # 0x90 -> LIGHT SHADE + '\u2592' # 0x91 -> MEDIUM SHADE + '\u2593' # 0x92 -> DARK SHADE + '\u2320' # 0x93 -> TOP HALF INTEGRAL + '\u25a0' # 0x94 -> BLACK SQUARE + '\u2219' # 0x95 -> BULLET OPERATOR + '\u221a' # 0x96 -> SQUARE ROOT + '\u2248' # 0x97 -> ALMOST EQUAL TO + '\u2264' # 0x98 -> LESS-THAN OR EQUAL TO + '\u2265' # 0x99 -> GREATER-THAN OR EQUAL TO + '\xa0' # 0x9A -> NO-BREAK SPACE + '\u2321' # 0x9B -> BOTTOM HALF INTEGRAL + '\xb0' # 0x9C -> DEGREE SIGN + '\xb2' # 0x9D -> SUPERSCRIPT TWO + '\xb7' # 0x9E -> MIDDLE DOT + '\xf7' # 0x9F -> DIVISION SIGN + '\u2550' # 0xA0 -> BOX DRAWINGS DOUBLE HORIZONTAL + '\u2551' # 0xA1 -> BOX DRAWINGS DOUBLE VERTICAL + '\u2552' # 0xA2 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + '\u0451' # 0xA3 -> CYRILLIC SMALL LETTER IO + '\u0454' # 0xA4 -> CYRILLIC SMALL LETTER UKRAINIAN IE + '\u2554' # 0xA5 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT + '\u0456' # 0xA6 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0457' # 0xA7 -> CYRILLIC SMALL LETTER YI (UKRAINIAN) + '\u2557' # 0xA8 -> BOX DRAWINGS DOUBLE DOWN AND LEFT + '\u2558' # 0xA9 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + '\u2559' # 0xAA -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + '\u255a' # 0xAB -> BOX DRAWINGS DOUBLE UP AND RIGHT + '\u255b' # 0xAC -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + '\u0491' # 0xAD -> CYRILLIC SMALL LETTER UKRAINIAN GHE WITH UPTURN + '\u255d' # 0xAE -> BOX DRAWINGS DOUBLE UP AND LEFT + '\u255e' # 0xAF -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + '\u255f' # 0xB0 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + '\u2560' # 0xB1 -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + '\u2561' # 0xB2 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + '\u0401' # 0xB3 -> CYRILLIC CAPITAL LETTER IO + '\u0404' # 0xB4 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE + '\u2563' # 0xB5 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT + '\u0406' # 0xB6 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + '\u0407' # 0xB7 -> CYRILLIC CAPITAL LETTER YI (UKRAINIAN) + '\u2566' # 0xB8 -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + '\u2567' # 0xB9 -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + '\u2568' # 0xBA -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + '\u2569' # 0xBB -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL + '\u256a' # 0xBC -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + '\u0490' # 0xBD -> CYRILLIC CAPITAL LETTER UKRAINIAN GHE WITH UPTURN + '\u256c' # 0xBE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + '\xa9' # 0xBF -> COPYRIGHT SIGN + '\u044e' # 0xC0 -> CYRILLIC SMALL LETTER YU + '\u0430' # 0xC1 -> CYRILLIC SMALL LETTER A + '\u0431' # 0xC2 -> CYRILLIC SMALL LETTER BE + '\u0446' # 0xC3 -> CYRILLIC SMALL LETTER TSE + '\u0434' # 0xC4 -> CYRILLIC SMALL LETTER DE + '\u0435' # 0xC5 -> CYRILLIC SMALL LETTER IE + '\u0444' # 0xC6 -> CYRILLIC SMALL LETTER EF + '\u0433' # 0xC7 -> CYRILLIC SMALL LETTER GHE + '\u0445' # 0xC8 -> CYRILLIC SMALL LETTER HA + '\u0438' # 0xC9 -> CYRILLIC SMALL LETTER I + '\u0439' # 0xCA -> CYRILLIC SMALL LETTER SHORT I + '\u043a' # 0xCB -> CYRILLIC SMALL LETTER KA + '\u043b' # 0xCC -> CYRILLIC SMALL LETTER EL + '\u043c' # 0xCD -> CYRILLIC SMALL LETTER EM + '\u043d' # 0xCE -> CYRILLIC SMALL LETTER EN + '\u043e' # 0xCF -> CYRILLIC SMALL LETTER O + '\u043f' # 0xD0 -> CYRILLIC SMALL LETTER PE + '\u044f' # 0xD1 -> CYRILLIC SMALL LETTER YA + '\u0440' # 0xD2 -> CYRILLIC SMALL LETTER ER + '\u0441' # 0xD3 -> CYRILLIC SMALL LETTER ES + '\u0442' # 0xD4 -> CYRILLIC SMALL LETTER TE + '\u0443' # 0xD5 -> CYRILLIC SMALL LETTER U + '\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE + '\u0432' # 0xD7 -> CYRILLIC SMALL LETTER VE + '\u044c' # 0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN + '\u044b' # 0xD9 -> CYRILLIC SMALL LETTER YERU + '\u0437' # 0xDA -> CYRILLIC SMALL LETTER ZE + '\u0448' # 0xDB -> CYRILLIC SMALL LETTER SHA + '\u044d' # 0xDC -> CYRILLIC SMALL LETTER E + '\u0449' # 0xDD -> CYRILLIC SMALL LETTER SHCHA + '\u0447' # 0xDE -> CYRILLIC SMALL LETTER CHE + '\u044a' # 0xDF -> CYRILLIC SMALL LETTER HARD SIGN + '\u042e' # 0xE0 -> CYRILLIC CAPITAL LETTER YU + '\u0410' # 0xE1 -> CYRILLIC CAPITAL LETTER A + '\u0411' # 0xE2 -> CYRILLIC CAPITAL LETTER BE + '\u0426' # 0xE3 -> CYRILLIC CAPITAL LETTER TSE + '\u0414' # 0xE4 -> CYRILLIC CAPITAL LETTER DE + '\u0415' # 0xE5 -> CYRILLIC CAPITAL LETTER IE + '\u0424' # 0xE6 -> CYRILLIC CAPITAL LETTER EF + '\u0413' # 0xE7 -> CYRILLIC CAPITAL LETTER GHE + '\u0425' # 0xE8 -> CYRILLIC CAPITAL LETTER HA + '\u0418' # 0xE9 -> CYRILLIC CAPITAL LETTER I + '\u0419' # 0xEA -> CYRILLIC CAPITAL LETTER SHORT I + '\u041a' # 0xEB -> CYRILLIC CAPITAL LETTER KA + '\u041b' # 0xEC -> CYRILLIC CAPITAL LETTER EL + '\u041c' # 0xED -> CYRILLIC CAPITAL LETTER EM + '\u041d' # 0xEE -> CYRILLIC CAPITAL LETTER EN + '\u041e' # 0xEF -> CYRILLIC CAPITAL LETTER O + '\u041f' # 0xF0 -> CYRILLIC CAPITAL LETTER PE + '\u042f' # 0xF1 -> CYRILLIC CAPITAL LETTER YA + '\u0420' # 0xF2 -> CYRILLIC CAPITAL LETTER ER + '\u0421' # 0xF3 -> CYRILLIC CAPITAL LETTER ES + '\u0422' # 0xF4 -> CYRILLIC CAPITAL LETTER TE + '\u0423' # 0xF5 -> CYRILLIC CAPITAL LETTER U + '\u0416' # 0xF6 -> CYRILLIC CAPITAL LETTER ZHE + '\u0412' # 0xF7 -> CYRILLIC CAPITAL LETTER VE + '\u042c' # 0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u042b' # 0xF9 -> CYRILLIC CAPITAL LETTER YERU + '\u0417' # 0xFA -> CYRILLIC CAPITAL LETTER ZE + '\u0428' # 0xFB -> CYRILLIC CAPITAL LETTER SHA + '\u042d' # 0xFC -> CYRILLIC CAPITAL LETTER E + '\u0429' # 0xFD -> CYRILLIC CAPITAL LETTER SHCHA + '\u0427' # 0xFE -> CYRILLIC CAPITAL LETTER CHE + '\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN ) ### Encoding table diff --git a/Lib/encodings/mac_arabic.py b/Lib/encodings/mac_arabic.py index 7a7d3c5..72847e8 100644 --- a/Lib/encodings/mac_arabic.py +++ b/Lib/encodings/mac_arabic.py @@ -178,262 +178,262 @@ decoding_map.update({ ### Decoding Table decoding_table = ( - u'\x00' # 0x0000 -> CONTROL CHARACTER - u'\x01' # 0x0001 -> CONTROL CHARACTER - u'\x02' # 0x0002 -> CONTROL CHARACTER - u'\x03' # 0x0003 -> CONTROL CHARACTER - u'\x04' # 0x0004 -> CONTROL CHARACTER - u'\x05' # 0x0005 -> CONTROL CHARACTER - u'\x06' # 0x0006 -> CONTROL CHARACTER - u'\x07' # 0x0007 -> CONTROL CHARACTER - u'\x08' # 0x0008 -> CONTROL CHARACTER - u'\t' # 0x0009 -> CONTROL CHARACTER - u'\n' # 0x000a -> CONTROL CHARACTER - u'\x0b' # 0x000b -> CONTROL CHARACTER - u'\x0c' # 0x000c -> CONTROL CHARACTER - u'\r' # 0x000d -> CONTROL CHARACTER - u'\x0e' # 0x000e -> CONTROL CHARACTER - u'\x0f' # 0x000f -> CONTROL CHARACTER - u'\x10' # 0x0010 -> CONTROL CHARACTER - u'\x11' # 0x0011 -> CONTROL CHARACTER - u'\x12' # 0x0012 -> CONTROL CHARACTER - u'\x13' # 0x0013 -> CONTROL CHARACTER - u'\x14' # 0x0014 -> CONTROL CHARACTER - u'\x15' # 0x0015 -> CONTROL CHARACTER - u'\x16' # 0x0016 -> CONTROL CHARACTER - u'\x17' # 0x0017 -> CONTROL CHARACTER - u'\x18' # 0x0018 -> CONTROL CHARACTER - u'\x19' # 0x0019 -> CONTROL CHARACTER - u'\x1a' # 0x001a -> CONTROL CHARACTER - u'\x1b' # 0x001b -> CONTROL CHARACTER - u'\x1c' # 0x001c -> CONTROL CHARACTER - u'\x1d' # 0x001d -> CONTROL CHARACTER - u'\x1e' # 0x001e -> CONTROL CHARACTER - u'\x1f' # 0x001f -> CONTROL CHARACTER - u' ' # 0x0020 -> SPACE, left-right - u'!' # 0x0021 -> EXCLAMATION MARK, left-right - u'"' # 0x0022 -> QUOTATION MARK, left-right - u'#' # 0x0023 -> NUMBER SIGN, left-right - u'$' # 0x0024 -> DOLLAR SIGN, left-right - u'%' # 0x0025 -> PERCENT SIGN, left-right - u'&' # 0x0026 -> AMPERSAND, left-right - u"'" # 0x0027 -> APOSTROPHE, left-right - u'(' # 0x0028 -> LEFT PARENTHESIS, left-right - u')' # 0x0029 -> RIGHT PARENTHESIS, left-right - u'*' # 0x002a -> ASTERISK, left-right - u'+' # 0x002b -> PLUS SIGN, left-right - u',' # 0x002c -> COMMA, left-right; in Arabic-script context, displayed as 0x066C ARABIC THOUSANDS SEPARATOR - u'-' # 0x002d -> HYPHEN-MINUS, left-right - u'.' # 0x002e -> FULL STOP, left-right; in Arabic-script context, displayed as 0x066B ARABIC DECIMAL SEPARATOR - u'/' # 0x002f -> SOLIDUS, left-right - u'0' # 0x0030 -> DIGIT ZERO; in Arabic-script context, displayed as 0x0660 ARABIC-INDIC DIGIT ZERO - u'1' # 0x0031 -> DIGIT ONE; in Arabic-script context, displayed as 0x0661 ARABIC-INDIC DIGIT ONE - u'2' # 0x0032 -> DIGIT TWO; in Arabic-script context, displayed as 0x0662 ARABIC-INDIC DIGIT TWO - u'3' # 0x0033 -> DIGIT THREE; in Arabic-script context, displayed as 0x0663 ARABIC-INDIC DIGIT THREE - u'4' # 0x0034 -> DIGIT FOUR; in Arabic-script context, displayed as 0x0664 ARABIC-INDIC DIGIT FOUR - u'5' # 0x0035 -> DIGIT FIVE; in Arabic-script context, displayed as 0x0665 ARABIC-INDIC DIGIT FIVE - u'6' # 0x0036 -> DIGIT SIX; in Arabic-script context, displayed as 0x0666 ARABIC-INDIC DIGIT SIX - u'7' # 0x0037 -> DIGIT SEVEN; in Arabic-script context, displayed as 0x0667 ARABIC-INDIC DIGIT SEVEN - u'8' # 0x0038 -> DIGIT EIGHT; in Arabic-script context, displayed as 0x0668 ARABIC-INDIC DIGIT EIGHT - u'9' # 0x0039 -> DIGIT NINE; in Arabic-script context, displayed as 0x0669 ARABIC-INDIC DIGIT NINE - u':' # 0x003a -> COLON, left-right - u';' # 0x003b -> SEMICOLON, left-right - u'<' # 0x003c -> LESS-THAN SIGN, left-right - u'=' # 0x003d -> EQUALS SIGN, left-right - u'>' # 0x003e -> GREATER-THAN SIGN, left-right - u'?' # 0x003f -> QUESTION MARK, left-right - u'@' # 0x0040 -> COMMERCIAL AT - u'A' # 0x0041 -> LATIN CAPITAL LETTER A - u'B' # 0x0042 -> LATIN CAPITAL LETTER B - u'C' # 0x0043 -> LATIN CAPITAL LETTER C - u'D' # 0x0044 -> LATIN CAPITAL LETTER D - u'E' # 0x0045 -> LATIN CAPITAL LETTER E - u'F' # 0x0046 -> LATIN CAPITAL LETTER F - u'G' # 0x0047 -> LATIN CAPITAL LETTER G - u'H' # 0x0048 -> LATIN CAPITAL LETTER H - u'I' # 0x0049 -> LATIN CAPITAL LETTER I - u'J' # 0x004a -> LATIN CAPITAL LETTER J - u'K' # 0x004b -> LATIN CAPITAL LETTER K - u'L' # 0x004c -> LATIN CAPITAL LETTER L - u'M' # 0x004d -> LATIN CAPITAL LETTER M - u'N' # 0x004e -> LATIN CAPITAL LETTER N - u'O' # 0x004f -> LATIN CAPITAL LETTER O - u'P' # 0x0050 -> LATIN CAPITAL LETTER P - u'Q' # 0x0051 -> LATIN CAPITAL LETTER Q - u'R' # 0x0052 -> LATIN CAPITAL LETTER R - u'S' # 0x0053 -> LATIN CAPITAL LETTER S - u'T' # 0x0054 -> LATIN CAPITAL LETTER T - u'U' # 0x0055 -> LATIN CAPITAL LETTER U - u'V' # 0x0056 -> LATIN CAPITAL LETTER V - u'W' # 0x0057 -> LATIN CAPITAL LETTER W - u'X' # 0x0058 -> LATIN CAPITAL LETTER X - u'Y' # 0x0059 -> LATIN CAPITAL LETTER Y - u'Z' # 0x005a -> LATIN CAPITAL LETTER Z - u'[' # 0x005b -> LEFT SQUARE BRACKET, left-right - u'\\' # 0x005c -> REVERSE SOLIDUS, left-right - u']' # 0x005d -> RIGHT SQUARE BRACKET, left-right - u'^' # 0x005e -> CIRCUMFLEX ACCENT, left-right - u'_' # 0x005f -> LOW LINE, left-right - u'`' # 0x0060 -> GRAVE ACCENT - u'a' # 0x0061 -> LATIN SMALL LETTER A - u'b' # 0x0062 -> LATIN SMALL LETTER B - u'c' # 0x0063 -> LATIN SMALL LETTER C - u'd' # 0x0064 -> LATIN SMALL LETTER D - u'e' # 0x0065 -> LATIN SMALL LETTER E - u'f' # 0x0066 -> LATIN SMALL LETTER F - u'g' # 0x0067 -> LATIN SMALL LETTER G - u'h' # 0x0068 -> LATIN SMALL LETTER H - u'i' # 0x0069 -> LATIN SMALL LETTER I - u'j' # 0x006a -> LATIN SMALL LETTER J - u'k' # 0x006b -> LATIN SMALL LETTER K - u'l' # 0x006c -> LATIN SMALL LETTER L - u'm' # 0x006d -> LATIN SMALL LETTER M - u'n' # 0x006e -> LATIN SMALL LETTER N - u'o' # 0x006f -> LATIN SMALL LETTER O - u'p' # 0x0070 -> LATIN SMALL LETTER P - u'q' # 0x0071 -> LATIN SMALL LETTER Q - u'r' # 0x0072 -> LATIN SMALL LETTER R - u's' # 0x0073 -> LATIN SMALL LETTER S - u't' # 0x0074 -> LATIN SMALL LETTER T - u'u' # 0x0075 -> LATIN SMALL LETTER U - u'v' # 0x0076 -> LATIN SMALL LETTER V - u'w' # 0x0077 -> LATIN SMALL LETTER W - u'x' # 0x0078 -> LATIN SMALL LETTER X - u'y' # 0x0079 -> LATIN SMALL LETTER Y - u'z' # 0x007a -> LATIN SMALL LETTER Z - u'{' # 0x007b -> LEFT CURLY BRACKET, left-right - u'|' # 0x007c -> VERTICAL LINE, left-right - u'}' # 0x007d -> RIGHT CURLY BRACKET, left-right - u'~' # 0x007e -> TILDE - u'\x7f' # 0x007f -> CONTROL CHARACTER - u'\xc4' # 0x0080 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xa0' # 0x0081 -> NO-BREAK SPACE, right-left - u'\xc7' # 0x0082 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x0083 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x0084 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x0085 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x0086 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x0087 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x0088 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x0089 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x008a -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u06ba' # 0x008b -> ARABIC LETTER NOON GHUNNA - u'\xab' # 0x008c -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left - u'\xe7' # 0x008d -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x008e -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x008f -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x0090 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x0091 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x0092 -> LATIN SMALL LETTER I WITH ACUTE - u'\u2026' # 0x0093 -> HORIZONTAL ELLIPSIS, right-left - u'\xee' # 0x0094 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x0095 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x0096 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x0097 -> LATIN SMALL LETTER O WITH ACUTE - u'\xbb' # 0x0098 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left - u'\xf4' # 0x0099 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x009a -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0x009b -> DIVISION SIGN, right-left - u'\xfa' # 0x009c -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x009d -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x009e -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x009f -> LATIN SMALL LETTER U WITH DIAERESIS - u' ' # 0x00a0 -> SPACE, right-left - u'!' # 0x00a1 -> EXCLAMATION MARK, right-left - u'"' # 0x00a2 -> QUOTATION MARK, right-left - u'#' # 0x00a3 -> NUMBER SIGN, right-left - u'$' # 0x00a4 -> DOLLAR SIGN, right-left - u'\u066a' # 0x00a5 -> ARABIC PERCENT SIGN - u'&' # 0x00a6 -> AMPERSAND, right-left - u"'" # 0x00a7 -> APOSTROPHE, right-left - u'(' # 0x00a8 -> LEFT PARENTHESIS, right-left - u')' # 0x00a9 -> RIGHT PARENTHESIS, right-left - u'*' # 0x00aa -> ASTERISK, right-left - u'+' # 0x00ab -> PLUS SIGN, right-left - u'\u060c' # 0x00ac -> ARABIC COMMA - u'-' # 0x00ad -> HYPHEN-MINUS, right-left - u'.' # 0x00ae -> FULL STOP, right-left - u'/' # 0x00af -> SOLIDUS, right-left - u'\u0660' # 0x00b0 -> ARABIC-INDIC DIGIT ZERO, right-left (need override) - u'\u0661' # 0x00b1 -> ARABIC-INDIC DIGIT ONE, right-left (need override) - u'\u0662' # 0x00b2 -> ARABIC-INDIC DIGIT TWO, right-left (need override) - u'\u0663' # 0x00b3 -> ARABIC-INDIC DIGIT THREE, right-left (need override) - u'\u0664' # 0x00b4 -> ARABIC-INDIC DIGIT FOUR, right-left (need override) - u'\u0665' # 0x00b5 -> ARABIC-INDIC DIGIT FIVE, right-left (need override) - u'\u0666' # 0x00b6 -> ARABIC-INDIC DIGIT SIX, right-left (need override) - u'\u0667' # 0x00b7 -> ARABIC-INDIC DIGIT SEVEN, right-left (need override) - u'\u0668' # 0x00b8 -> ARABIC-INDIC DIGIT EIGHT, right-left (need override) - u'\u0669' # 0x00b9 -> ARABIC-INDIC DIGIT NINE, right-left (need override) - u':' # 0x00ba -> COLON, right-left - u'\u061b' # 0x00bb -> ARABIC SEMICOLON - u'<' # 0x00bc -> LESS-THAN SIGN, right-left - u'=' # 0x00bd -> EQUALS SIGN, right-left - u'>' # 0x00be -> GREATER-THAN SIGN, right-left - u'\u061f' # 0x00bf -> ARABIC QUESTION MARK - u'\u274a' # 0x00c0 -> EIGHT TEARDROP-SPOKED PROPELLER ASTERISK, right-left - u'\u0621' # 0x00c1 -> ARABIC LETTER HAMZA - u'\u0622' # 0x00c2 -> ARABIC LETTER ALEF WITH MADDA ABOVE - u'\u0623' # 0x00c3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE - u'\u0624' # 0x00c4 -> ARABIC LETTER WAW WITH HAMZA ABOVE - u'\u0625' # 0x00c5 -> ARABIC LETTER ALEF WITH HAMZA BELOW - u'\u0626' # 0x00c6 -> ARABIC LETTER YEH WITH HAMZA ABOVE - u'\u0627' # 0x00c7 -> ARABIC LETTER ALEF - u'\u0628' # 0x00c8 -> ARABIC LETTER BEH - u'\u0629' # 0x00c9 -> ARABIC LETTER TEH MARBUTA - u'\u062a' # 0x00ca -> ARABIC LETTER TEH - u'\u062b' # 0x00cb -> ARABIC LETTER THEH - u'\u062c' # 0x00cc -> ARABIC LETTER JEEM - u'\u062d' # 0x00cd -> ARABIC LETTER HAH - u'\u062e' # 0x00ce -> ARABIC LETTER KHAH - u'\u062f' # 0x00cf -> ARABIC LETTER DAL - u'\u0630' # 0x00d0 -> ARABIC LETTER THAL - u'\u0631' # 0x00d1 -> ARABIC LETTER REH - u'\u0632' # 0x00d2 -> ARABIC LETTER ZAIN - u'\u0633' # 0x00d3 -> ARABIC LETTER SEEN - u'\u0634' # 0x00d4 -> ARABIC LETTER SHEEN - u'\u0635' # 0x00d5 -> ARABIC LETTER SAD - u'\u0636' # 0x00d6 -> ARABIC LETTER DAD - u'\u0637' # 0x00d7 -> ARABIC LETTER TAH - u'\u0638' # 0x00d8 -> ARABIC LETTER ZAH - u'\u0639' # 0x00d9 -> ARABIC LETTER AIN - u'\u063a' # 0x00da -> ARABIC LETTER GHAIN - u'[' # 0x00db -> LEFT SQUARE BRACKET, right-left - u'\\' # 0x00dc -> REVERSE SOLIDUS, right-left - u']' # 0x00dd -> RIGHT SQUARE BRACKET, right-left - u'^' # 0x00de -> CIRCUMFLEX ACCENT, right-left - u'_' # 0x00df -> LOW LINE, right-left - u'\u0640' # 0x00e0 -> ARABIC TATWEEL - u'\u0641' # 0x00e1 -> ARABIC LETTER FEH - u'\u0642' # 0x00e2 -> ARABIC LETTER QAF - u'\u0643' # 0x00e3 -> ARABIC LETTER KAF - u'\u0644' # 0x00e4 -> ARABIC LETTER LAM - u'\u0645' # 0x00e5 -> ARABIC LETTER MEEM - u'\u0646' # 0x00e6 -> ARABIC LETTER NOON - u'\u0647' # 0x00e7 -> ARABIC LETTER HEH - u'\u0648' # 0x00e8 -> ARABIC LETTER WAW - u'\u0649' # 0x00e9 -> ARABIC LETTER ALEF MAKSURA - u'\u064a' # 0x00ea -> ARABIC LETTER YEH - u'\u064b' # 0x00eb -> ARABIC FATHATAN - u'\u064c' # 0x00ec -> ARABIC DAMMATAN - u'\u064d' # 0x00ed -> ARABIC KASRATAN - u'\u064e' # 0x00ee -> ARABIC FATHA - u'\u064f' # 0x00ef -> ARABIC DAMMA - u'\u0650' # 0x00f0 -> ARABIC KASRA - u'\u0651' # 0x00f1 -> ARABIC SHADDA - u'\u0652' # 0x00f2 -> ARABIC SUKUN - u'\u067e' # 0x00f3 -> ARABIC LETTER PEH - u'\u0679' # 0x00f4 -> ARABIC LETTER TTEH - u'\u0686' # 0x00f5 -> ARABIC LETTER TCHEH - u'\u06d5' # 0x00f6 -> ARABIC LETTER AE - u'\u06a4' # 0x00f7 -> ARABIC LETTER VEH - u'\u06af' # 0x00f8 -> ARABIC LETTER GAF - u'\u0688' # 0x00f9 -> ARABIC LETTER DDAL - u'\u0691' # 0x00fa -> ARABIC LETTER RREH - u'{' # 0x00fb -> LEFT CURLY BRACKET, right-left - u'|' # 0x00fc -> VERTICAL LINE, right-left - u'}' # 0x00fd -> RIGHT CURLY BRACKET, right-left - u'\u0698' # 0x00fe -> ARABIC LETTER JEH - u'\u06d2' # 0x00ff -> ARABIC LETTER YEH BARREE + '\x00' # 0x0000 -> CONTROL CHARACTER + '\x01' # 0x0001 -> CONTROL CHARACTER + '\x02' # 0x0002 -> CONTROL CHARACTER + '\x03' # 0x0003 -> CONTROL CHARACTER + '\x04' # 0x0004 -> CONTROL CHARACTER + '\x05' # 0x0005 -> CONTROL CHARACTER + '\x06' # 0x0006 -> CONTROL CHARACTER + '\x07' # 0x0007 -> CONTROL CHARACTER + '\x08' # 0x0008 -> CONTROL CHARACTER + '\t' # 0x0009 -> CONTROL CHARACTER + '\n' # 0x000a -> CONTROL CHARACTER + '\x0b' # 0x000b -> CONTROL CHARACTER + '\x0c' # 0x000c -> CONTROL CHARACTER + '\r' # 0x000d -> CONTROL CHARACTER + '\x0e' # 0x000e -> CONTROL CHARACTER + '\x0f' # 0x000f -> CONTROL CHARACTER + '\x10' # 0x0010 -> CONTROL CHARACTER + '\x11' # 0x0011 -> CONTROL CHARACTER + '\x12' # 0x0012 -> CONTROL CHARACTER + '\x13' # 0x0013 -> CONTROL CHARACTER + '\x14' # 0x0014 -> CONTROL CHARACTER + '\x15' # 0x0015 -> CONTROL CHARACTER + '\x16' # 0x0016 -> CONTROL CHARACTER + '\x17' # 0x0017 -> CONTROL CHARACTER + '\x18' # 0x0018 -> CONTROL CHARACTER + '\x19' # 0x0019 -> CONTROL CHARACTER + '\x1a' # 0x001a -> CONTROL CHARACTER + '\x1b' # 0x001b -> CONTROL CHARACTER + '\x1c' # 0x001c -> CONTROL CHARACTER + '\x1d' # 0x001d -> CONTROL CHARACTER + '\x1e' # 0x001e -> CONTROL CHARACTER + '\x1f' # 0x001f -> CONTROL CHARACTER + ' ' # 0x0020 -> SPACE, left-right + '!' # 0x0021 -> EXCLAMATION MARK, left-right + '"' # 0x0022 -> QUOTATION MARK, left-right + '#' # 0x0023 -> NUMBER SIGN, left-right + '$' # 0x0024 -> DOLLAR SIGN, left-right + '%' # 0x0025 -> PERCENT SIGN, left-right + '&' # 0x0026 -> AMPERSAND, left-right + "'" # 0x0027 -> APOSTROPHE, left-right + '(' # 0x0028 -> LEFT PARENTHESIS, left-right + ')' # 0x0029 -> RIGHT PARENTHESIS, left-right + '*' # 0x002a -> ASTERISK, left-right + '+' # 0x002b -> PLUS SIGN, left-right + ',' # 0x002c -> COMMA, left-right; in Arabic-script context, displayed as 0x066C ARABIC THOUSANDS SEPARATOR + '-' # 0x002d -> HYPHEN-MINUS, left-right + '.' # 0x002e -> FULL STOP, left-right; in Arabic-script context, displayed as 0x066B ARABIC DECIMAL SEPARATOR + '/' # 0x002f -> SOLIDUS, left-right + '0' # 0x0030 -> DIGIT ZERO; in Arabic-script context, displayed as 0x0660 ARABIC-INDIC DIGIT ZERO + '1' # 0x0031 -> DIGIT ONE; in Arabic-script context, displayed as 0x0661 ARABIC-INDIC DIGIT ONE + '2' # 0x0032 -> DIGIT TWO; in Arabic-script context, displayed as 0x0662 ARABIC-INDIC DIGIT TWO + '3' # 0x0033 -> DIGIT THREE; in Arabic-script context, displayed as 0x0663 ARABIC-INDIC DIGIT THREE + '4' # 0x0034 -> DIGIT FOUR; in Arabic-script context, displayed as 0x0664 ARABIC-INDIC DIGIT FOUR + '5' # 0x0035 -> DIGIT FIVE; in Arabic-script context, displayed as 0x0665 ARABIC-INDIC DIGIT FIVE + '6' # 0x0036 -> DIGIT SIX; in Arabic-script context, displayed as 0x0666 ARABIC-INDIC DIGIT SIX + '7' # 0x0037 -> DIGIT SEVEN; in Arabic-script context, displayed as 0x0667 ARABIC-INDIC DIGIT SEVEN + '8' # 0x0038 -> DIGIT EIGHT; in Arabic-script context, displayed as 0x0668 ARABIC-INDIC DIGIT EIGHT + '9' # 0x0039 -> DIGIT NINE; in Arabic-script context, displayed as 0x0669 ARABIC-INDIC DIGIT NINE + ':' # 0x003a -> COLON, left-right + ';' # 0x003b -> SEMICOLON, left-right + '<' # 0x003c -> LESS-THAN SIGN, left-right + '=' # 0x003d -> EQUALS SIGN, left-right + '>' # 0x003e -> GREATER-THAN SIGN, left-right + '?' # 0x003f -> QUESTION MARK, left-right + '@' # 0x0040 -> COMMERCIAL AT + 'A' # 0x0041 -> LATIN CAPITAL LETTER A + 'B' # 0x0042 -> LATIN CAPITAL LETTER B + 'C' # 0x0043 -> LATIN CAPITAL LETTER C + 'D' # 0x0044 -> LATIN CAPITAL LETTER D + 'E' # 0x0045 -> LATIN CAPITAL LETTER E + 'F' # 0x0046 -> LATIN CAPITAL LETTER F + 'G' # 0x0047 -> LATIN CAPITAL LETTER G + 'H' # 0x0048 -> LATIN CAPITAL LETTER H + 'I' # 0x0049 -> LATIN CAPITAL LETTER I + 'J' # 0x004a -> LATIN CAPITAL LETTER J + 'K' # 0x004b -> LATIN CAPITAL LETTER K + 'L' # 0x004c -> LATIN CAPITAL LETTER L + 'M' # 0x004d -> LATIN CAPITAL LETTER M + 'N' # 0x004e -> LATIN CAPITAL LETTER N + 'O' # 0x004f -> LATIN CAPITAL LETTER O + 'P' # 0x0050 -> LATIN CAPITAL LETTER P + 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q + 'R' # 0x0052 -> LATIN CAPITAL LETTER R + 'S' # 0x0053 -> LATIN CAPITAL LETTER S + 'T' # 0x0054 -> LATIN CAPITAL LETTER T + 'U' # 0x0055 -> LATIN CAPITAL LETTER U + 'V' # 0x0056 -> LATIN CAPITAL LETTER V + 'W' # 0x0057 -> LATIN CAPITAL LETTER W + 'X' # 0x0058 -> LATIN CAPITAL LETTER X + 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y + 'Z' # 0x005a -> LATIN CAPITAL LETTER Z + '[' # 0x005b -> LEFT SQUARE BRACKET, left-right + '\\' # 0x005c -> REVERSE SOLIDUS, left-right + ']' # 0x005d -> RIGHT SQUARE BRACKET, left-right + '^' # 0x005e -> CIRCUMFLEX ACCENT, left-right + '_' # 0x005f -> LOW LINE, left-right + '`' # 0x0060 -> GRAVE ACCENT + 'a' # 0x0061 -> LATIN SMALL LETTER A + 'b' # 0x0062 -> LATIN SMALL LETTER B + 'c' # 0x0063 -> LATIN SMALL LETTER C + 'd' # 0x0064 -> LATIN SMALL LETTER D + 'e' # 0x0065 -> LATIN SMALL LETTER E + 'f' # 0x0066 -> LATIN SMALL LETTER F + 'g' # 0x0067 -> LATIN SMALL LETTER G + 'h' # 0x0068 -> LATIN SMALL LETTER H + 'i' # 0x0069 -> LATIN SMALL LETTER I + 'j' # 0x006a -> LATIN SMALL LETTER J + 'k' # 0x006b -> LATIN SMALL LETTER K + 'l' # 0x006c -> LATIN SMALL LETTER L + 'm' # 0x006d -> LATIN SMALL LETTER M + 'n' # 0x006e -> LATIN SMALL LETTER N + 'o' # 0x006f -> LATIN SMALL LETTER O + 'p' # 0x0070 -> LATIN SMALL LETTER P + 'q' # 0x0071 -> LATIN SMALL LETTER Q + 'r' # 0x0072 -> LATIN SMALL LETTER R + 's' # 0x0073 -> LATIN SMALL LETTER S + 't' # 0x0074 -> LATIN SMALL LETTER T + 'u' # 0x0075 -> LATIN SMALL LETTER U + 'v' # 0x0076 -> LATIN SMALL LETTER V + 'w' # 0x0077 -> LATIN SMALL LETTER W + 'x' # 0x0078 -> LATIN SMALL LETTER X + 'y' # 0x0079 -> LATIN SMALL LETTER Y + 'z' # 0x007a -> LATIN SMALL LETTER Z + '{' # 0x007b -> LEFT CURLY BRACKET, left-right + '|' # 0x007c -> VERTICAL LINE, left-right + '}' # 0x007d -> RIGHT CURLY BRACKET, left-right + '~' # 0x007e -> TILDE + '\x7f' # 0x007f -> CONTROL CHARACTER + '\xc4' # 0x0080 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xa0' # 0x0081 -> NO-BREAK SPACE, right-left + '\xc7' # 0x0082 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x0083 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x0084 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x0085 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x0086 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x0087 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x0088 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x0089 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x008a -> LATIN SMALL LETTER A WITH DIAERESIS + '\u06ba' # 0x008b -> ARABIC LETTER NOON GHUNNA + '\xab' # 0x008c -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left + '\xe7' # 0x008d -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x008e -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x008f -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x0090 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x0091 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x0092 -> LATIN SMALL LETTER I WITH ACUTE + '\u2026' # 0x0093 -> HORIZONTAL ELLIPSIS, right-left + '\xee' # 0x0094 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x0095 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x0096 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x0097 -> LATIN SMALL LETTER O WITH ACUTE + '\xbb' # 0x0098 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left + '\xf4' # 0x0099 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x009a -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0x009b -> DIVISION SIGN, right-left + '\xfa' # 0x009c -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x009d -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x009e -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x009f -> LATIN SMALL LETTER U WITH DIAERESIS + ' ' # 0x00a0 -> SPACE, right-left + '!' # 0x00a1 -> EXCLAMATION MARK, right-left + '"' # 0x00a2 -> QUOTATION MARK, right-left + '#' # 0x00a3 -> NUMBER SIGN, right-left + '$' # 0x00a4 -> DOLLAR SIGN, right-left + '\u066a' # 0x00a5 -> ARABIC PERCENT SIGN + '&' # 0x00a6 -> AMPERSAND, right-left + "'" # 0x00a7 -> APOSTROPHE, right-left + '(' # 0x00a8 -> LEFT PARENTHESIS, right-left + ')' # 0x00a9 -> RIGHT PARENTHESIS, right-left + '*' # 0x00aa -> ASTERISK, right-left + '+' # 0x00ab -> PLUS SIGN, right-left + '\u060c' # 0x00ac -> ARABIC COMMA + '-' # 0x00ad -> HYPHEN-MINUS, right-left + '.' # 0x00ae -> FULL STOP, right-left + '/' # 0x00af -> SOLIDUS, right-left + '\u0660' # 0x00b0 -> ARABIC-INDIC DIGIT ZERO, right-left (need override) + '\u0661' # 0x00b1 -> ARABIC-INDIC DIGIT ONE, right-left (need override) + '\u0662' # 0x00b2 -> ARABIC-INDIC DIGIT TWO, right-left (need override) + '\u0663' # 0x00b3 -> ARABIC-INDIC DIGIT THREE, right-left (need override) + '\u0664' # 0x00b4 -> ARABIC-INDIC DIGIT FOUR, right-left (need override) + '\u0665' # 0x00b5 -> ARABIC-INDIC DIGIT FIVE, right-left (need override) + '\u0666' # 0x00b6 -> ARABIC-INDIC DIGIT SIX, right-left (need override) + '\u0667' # 0x00b7 -> ARABIC-INDIC DIGIT SEVEN, right-left (need override) + '\u0668' # 0x00b8 -> ARABIC-INDIC DIGIT EIGHT, right-left (need override) + '\u0669' # 0x00b9 -> ARABIC-INDIC DIGIT NINE, right-left (need override) + ':' # 0x00ba -> COLON, right-left + '\u061b' # 0x00bb -> ARABIC SEMICOLON + '<' # 0x00bc -> LESS-THAN SIGN, right-left + '=' # 0x00bd -> EQUALS SIGN, right-left + '>' # 0x00be -> GREATER-THAN SIGN, right-left + '\u061f' # 0x00bf -> ARABIC QUESTION MARK + '\u274a' # 0x00c0 -> EIGHT TEARDROP-SPOKED PROPELLER ASTERISK, right-left + '\u0621' # 0x00c1 -> ARABIC LETTER HAMZA + '\u0622' # 0x00c2 -> ARABIC LETTER ALEF WITH MADDA ABOVE + '\u0623' # 0x00c3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE + '\u0624' # 0x00c4 -> ARABIC LETTER WAW WITH HAMZA ABOVE + '\u0625' # 0x00c5 -> ARABIC LETTER ALEF WITH HAMZA BELOW + '\u0626' # 0x00c6 -> ARABIC LETTER YEH WITH HAMZA ABOVE + '\u0627' # 0x00c7 -> ARABIC LETTER ALEF + '\u0628' # 0x00c8 -> ARABIC LETTER BEH + '\u0629' # 0x00c9 -> ARABIC LETTER TEH MARBUTA + '\u062a' # 0x00ca -> ARABIC LETTER TEH + '\u062b' # 0x00cb -> ARABIC LETTER THEH + '\u062c' # 0x00cc -> ARABIC LETTER JEEM + '\u062d' # 0x00cd -> ARABIC LETTER HAH + '\u062e' # 0x00ce -> ARABIC LETTER KHAH + '\u062f' # 0x00cf -> ARABIC LETTER DAL + '\u0630' # 0x00d0 -> ARABIC LETTER THAL + '\u0631' # 0x00d1 -> ARABIC LETTER REH + '\u0632' # 0x00d2 -> ARABIC LETTER ZAIN + '\u0633' # 0x00d3 -> ARABIC LETTER SEEN + '\u0634' # 0x00d4 -> ARABIC LETTER SHEEN + '\u0635' # 0x00d5 -> ARABIC LETTER SAD + '\u0636' # 0x00d6 -> ARABIC LETTER DAD + '\u0637' # 0x00d7 -> ARABIC LETTER TAH + '\u0638' # 0x00d8 -> ARABIC LETTER ZAH + '\u0639' # 0x00d9 -> ARABIC LETTER AIN + '\u063a' # 0x00da -> ARABIC LETTER GHAIN + '[' # 0x00db -> LEFT SQUARE BRACKET, right-left + '\\' # 0x00dc -> REVERSE SOLIDUS, right-left + ']' # 0x00dd -> RIGHT SQUARE BRACKET, right-left + '^' # 0x00de -> CIRCUMFLEX ACCENT, right-left + '_' # 0x00df -> LOW LINE, right-left + '\u0640' # 0x00e0 -> ARABIC TATWEEL + '\u0641' # 0x00e1 -> ARABIC LETTER FEH + '\u0642' # 0x00e2 -> ARABIC LETTER QAF + '\u0643' # 0x00e3 -> ARABIC LETTER KAF + '\u0644' # 0x00e4 -> ARABIC LETTER LAM + '\u0645' # 0x00e5 -> ARABIC LETTER MEEM + '\u0646' # 0x00e6 -> ARABIC LETTER NOON + '\u0647' # 0x00e7 -> ARABIC LETTER HEH + '\u0648' # 0x00e8 -> ARABIC LETTER WAW + '\u0649' # 0x00e9 -> ARABIC LETTER ALEF MAKSURA + '\u064a' # 0x00ea -> ARABIC LETTER YEH + '\u064b' # 0x00eb -> ARABIC FATHATAN + '\u064c' # 0x00ec -> ARABIC DAMMATAN + '\u064d' # 0x00ed -> ARABIC KASRATAN + '\u064e' # 0x00ee -> ARABIC FATHA + '\u064f' # 0x00ef -> ARABIC DAMMA + '\u0650' # 0x00f0 -> ARABIC KASRA + '\u0651' # 0x00f1 -> ARABIC SHADDA + '\u0652' # 0x00f2 -> ARABIC SUKUN + '\u067e' # 0x00f3 -> ARABIC LETTER PEH + '\u0679' # 0x00f4 -> ARABIC LETTER TTEH + '\u0686' # 0x00f5 -> ARABIC LETTER TCHEH + '\u06d5' # 0x00f6 -> ARABIC LETTER AE + '\u06a4' # 0x00f7 -> ARABIC LETTER VEH + '\u06af' # 0x00f8 -> ARABIC LETTER GAF + '\u0688' # 0x00f9 -> ARABIC LETTER DDAL + '\u0691' # 0x00fa -> ARABIC LETTER RREH + '{' # 0x00fb -> LEFT CURLY BRACKET, right-left + '|' # 0x00fc -> VERTICAL LINE, right-left + '}' # 0x00fd -> RIGHT CURLY BRACKET, right-left + '\u0698' # 0x00fe -> ARABIC LETTER JEH + '\u06d2' # 0x00ff -> ARABIC LETTER YEH BARREE ) ### Encoding Map diff --git a/Lib/encodings/mac_centeuro.py b/Lib/encodings/mac_centeuro.py index 483c821..5785a0e 100644 --- a/Lib/encodings/mac_centeuro.py +++ b/Lib/encodings/mac_centeuro.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\u0100' # 0x81 -> LATIN CAPITAL LETTER A WITH MACRON - u'\u0101' # 0x82 -> LATIN SMALL LETTER A WITH MACRON - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\u0104' # 0x84 -> LATIN CAPITAL LETTER A WITH OGONEK - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\u0105' # 0x88 -> LATIN SMALL LETTER A WITH OGONEK - u'\u010c' # 0x89 -> LATIN CAPITAL LETTER C WITH CARON - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u010d' # 0x8B -> LATIN SMALL LETTER C WITH CARON - u'\u0106' # 0x8C -> LATIN CAPITAL LETTER C WITH ACUTE - u'\u0107' # 0x8D -> LATIN SMALL LETTER C WITH ACUTE - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\u0179' # 0x8F -> LATIN CAPITAL LETTER Z WITH ACUTE - u'\u017a' # 0x90 -> LATIN SMALL LETTER Z WITH ACUTE - u'\u010e' # 0x91 -> LATIN CAPITAL LETTER D WITH CARON - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\u010f' # 0x93 -> LATIN SMALL LETTER D WITH CARON - u'\u0112' # 0x94 -> LATIN CAPITAL LETTER E WITH MACRON - u'\u0113' # 0x95 -> LATIN SMALL LETTER E WITH MACRON - u'\u0116' # 0x96 -> LATIN CAPITAL LETTER E WITH DOT ABOVE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\u0117' # 0x98 -> LATIN SMALL LETTER E WITH DOT ABOVE - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\u011a' # 0x9D -> LATIN CAPITAL LETTER E WITH CARON - u'\u011b' # 0x9E -> LATIN SMALL LETTER E WITH CARON - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u2020' # 0xA0 -> DAGGER - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\u0118' # 0xA2 -> LATIN CAPITAL LETTER E WITH OGONEK - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\u0119' # 0xAB -> LATIN SMALL LETTER E WITH OGONEK - u'\xa8' # 0xAC -> DIAERESIS - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\u0123' # 0xAE -> LATIN SMALL LETTER G WITH CEDILLA - u'\u012e' # 0xAF -> LATIN CAPITAL LETTER I WITH OGONEK - u'\u012f' # 0xB0 -> LATIN SMALL LETTER I WITH OGONEK - u'\u012a' # 0xB1 -> LATIN CAPITAL LETTER I WITH MACRON - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\u012b' # 0xB4 -> LATIN SMALL LETTER I WITH MACRON - u'\u0136' # 0xB5 -> LATIN CAPITAL LETTER K WITH CEDILLA - u'\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL - u'\u2211' # 0xB7 -> N-ARY SUMMATION - u'\u0142' # 0xB8 -> LATIN SMALL LETTER L WITH STROKE - u'\u013b' # 0xB9 -> LATIN CAPITAL LETTER L WITH CEDILLA - u'\u013c' # 0xBA -> LATIN SMALL LETTER L WITH CEDILLA - u'\u013d' # 0xBB -> LATIN CAPITAL LETTER L WITH CARON - u'\u013e' # 0xBC -> LATIN SMALL LETTER L WITH CARON - u'\u0139' # 0xBD -> LATIN CAPITAL LETTER L WITH ACUTE - u'\u013a' # 0xBE -> LATIN SMALL LETTER L WITH ACUTE - u'\u0145' # 0xBF -> LATIN CAPITAL LETTER N WITH CEDILLA - u'\u0146' # 0xC0 -> LATIN SMALL LETTER N WITH CEDILLA - u'\u0143' # 0xC1 -> LATIN CAPITAL LETTER N WITH ACUTE - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0144' # 0xC4 -> LATIN SMALL LETTER N WITH ACUTE - u'\u0147' # 0xC5 -> LATIN CAPITAL LETTER N WITH CARON - u'\u2206' # 0xC6 -> INCREMENT - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\u0148' # 0xCB -> LATIN SMALL LETTER N WITH CARON - u'\u0150' # 0xCC -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - u'\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE - u'\u0151' # 0xCE -> LATIN SMALL LETTER O WITH DOUBLE ACUTE - u'\u014c' # 0xCF -> LATIN CAPITAL LETTER O WITH MACRON - u'\u2013' # 0xD0 -> EN DASH - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u25ca' # 0xD7 -> LOZENGE - u'\u014d' # 0xD8 -> LATIN SMALL LETTER O WITH MACRON - u'\u0154' # 0xD9 -> LATIN CAPITAL LETTER R WITH ACUTE - u'\u0155' # 0xDA -> LATIN SMALL LETTER R WITH ACUTE - u'\u0158' # 0xDB -> LATIN CAPITAL LETTER R WITH CARON - u'\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u0159' # 0xDE -> LATIN SMALL LETTER R WITH CARON - u'\u0156' # 0xDF -> LATIN CAPITAL LETTER R WITH CEDILLA - u'\u0157' # 0xE0 -> LATIN SMALL LETTER R WITH CEDILLA - u'\u0160' # 0xE1 -> LATIN CAPITAL LETTER S WITH CARON - u'\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK - u'\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK - u'\u0161' # 0xE4 -> LATIN SMALL LETTER S WITH CARON - u'\u015a' # 0xE5 -> LATIN CAPITAL LETTER S WITH ACUTE - u'\u015b' # 0xE6 -> LATIN SMALL LETTER S WITH ACUTE - u'\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\u0164' # 0xE8 -> LATIN CAPITAL LETTER T WITH CARON - u'\u0165' # 0xE9 -> LATIN SMALL LETTER T WITH CARON - u'\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE - u'\u017d' # 0xEB -> LATIN CAPITAL LETTER Z WITH CARON - u'\u017e' # 0xEC -> LATIN SMALL LETTER Z WITH CARON - u'\u016a' # 0xED -> LATIN CAPITAL LETTER U WITH MACRON - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u016b' # 0xF0 -> LATIN SMALL LETTER U WITH MACRON - u'\u016e' # 0xF1 -> LATIN CAPITAL LETTER U WITH RING ABOVE - u'\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\u016f' # 0xF3 -> LATIN SMALL LETTER U WITH RING ABOVE - u'\u0170' # 0xF4 -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - u'\u0171' # 0xF5 -> LATIN SMALL LETTER U WITH DOUBLE ACUTE - u'\u0172' # 0xF6 -> LATIN CAPITAL LETTER U WITH OGONEK - u'\u0173' # 0xF7 -> LATIN SMALL LETTER U WITH OGONEK - u'\xdd' # 0xF8 -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xfd' # 0xF9 -> LATIN SMALL LETTER Y WITH ACUTE - u'\u0137' # 0xFA -> LATIN SMALL LETTER K WITH CEDILLA - u'\u017b' # 0xFB -> LATIN CAPITAL LETTER Z WITH DOT ABOVE - u'\u0141' # 0xFC -> LATIN CAPITAL LETTER L WITH STROKE - u'\u017c' # 0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE - u'\u0122' # 0xFE -> LATIN CAPITAL LETTER G WITH CEDILLA - u'\u02c7' # 0xFF -> CARON + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\u0100' # 0x81 -> LATIN CAPITAL LETTER A WITH MACRON + '\u0101' # 0x82 -> LATIN SMALL LETTER A WITH MACRON + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\u0104' # 0x84 -> LATIN CAPITAL LETTER A WITH OGONEK + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\u0105' # 0x88 -> LATIN SMALL LETTER A WITH OGONEK + '\u010c' # 0x89 -> LATIN CAPITAL LETTER C WITH CARON + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\u010d' # 0x8B -> LATIN SMALL LETTER C WITH CARON + '\u0106' # 0x8C -> LATIN CAPITAL LETTER C WITH ACUTE + '\u0107' # 0x8D -> LATIN SMALL LETTER C WITH ACUTE + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\u0179' # 0x8F -> LATIN CAPITAL LETTER Z WITH ACUTE + '\u017a' # 0x90 -> LATIN SMALL LETTER Z WITH ACUTE + '\u010e' # 0x91 -> LATIN CAPITAL LETTER D WITH CARON + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\u010f' # 0x93 -> LATIN SMALL LETTER D WITH CARON + '\u0112' # 0x94 -> LATIN CAPITAL LETTER E WITH MACRON + '\u0113' # 0x95 -> LATIN SMALL LETTER E WITH MACRON + '\u0116' # 0x96 -> LATIN CAPITAL LETTER E WITH DOT ABOVE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\u0117' # 0x98 -> LATIN SMALL LETTER E WITH DOT ABOVE + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\u011a' # 0x9D -> LATIN CAPITAL LETTER E WITH CARON + '\u011b' # 0x9E -> LATIN SMALL LETTER E WITH CARON + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\u2020' # 0xA0 -> DAGGER + '\xb0' # 0xA1 -> DEGREE SIGN + '\u0118' # 0xA2 -> LATIN CAPITAL LETTER E WITH OGONEK + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u2122' # 0xAA -> TRADE MARK SIGN + '\u0119' # 0xAB -> LATIN SMALL LETTER E WITH OGONEK + '\xa8' # 0xAC -> DIAERESIS + '\u2260' # 0xAD -> NOT EQUAL TO + '\u0123' # 0xAE -> LATIN SMALL LETTER G WITH CEDILLA + '\u012e' # 0xAF -> LATIN CAPITAL LETTER I WITH OGONEK + '\u012f' # 0xB0 -> LATIN SMALL LETTER I WITH OGONEK + '\u012a' # 0xB1 -> LATIN CAPITAL LETTER I WITH MACRON + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\u012b' # 0xB4 -> LATIN SMALL LETTER I WITH MACRON + '\u0136' # 0xB5 -> LATIN CAPITAL LETTER K WITH CEDILLA + '\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL + '\u2211' # 0xB7 -> N-ARY SUMMATION + '\u0142' # 0xB8 -> LATIN SMALL LETTER L WITH STROKE + '\u013b' # 0xB9 -> LATIN CAPITAL LETTER L WITH CEDILLA + '\u013c' # 0xBA -> LATIN SMALL LETTER L WITH CEDILLA + '\u013d' # 0xBB -> LATIN CAPITAL LETTER L WITH CARON + '\u013e' # 0xBC -> LATIN SMALL LETTER L WITH CARON + '\u0139' # 0xBD -> LATIN CAPITAL LETTER L WITH ACUTE + '\u013a' # 0xBE -> LATIN SMALL LETTER L WITH ACUTE + '\u0145' # 0xBF -> LATIN CAPITAL LETTER N WITH CEDILLA + '\u0146' # 0xC0 -> LATIN SMALL LETTER N WITH CEDILLA + '\u0143' # 0xC1 -> LATIN CAPITAL LETTER N WITH ACUTE + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0144' # 0xC4 -> LATIN SMALL LETTER N WITH ACUTE + '\u0147' # 0xC5 -> LATIN CAPITAL LETTER N WITH CARON + '\u2206' # 0xC6 -> INCREMENT + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\u0148' # 0xCB -> LATIN SMALL LETTER N WITH CARON + '\u0150' # 0xCC -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE + '\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE + '\u0151' # 0xCE -> LATIN SMALL LETTER O WITH DOUBLE ACUTE + '\u014c' # 0xCF -> LATIN CAPITAL LETTER O WITH MACRON + '\u2013' # 0xD0 -> EN DASH + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u25ca' # 0xD7 -> LOZENGE + '\u014d' # 0xD8 -> LATIN SMALL LETTER O WITH MACRON + '\u0154' # 0xD9 -> LATIN CAPITAL LETTER R WITH ACUTE + '\u0155' # 0xDA -> LATIN SMALL LETTER R WITH ACUTE + '\u0158' # 0xDB -> LATIN CAPITAL LETTER R WITH CARON + '\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u0159' # 0xDE -> LATIN SMALL LETTER R WITH CARON + '\u0156' # 0xDF -> LATIN CAPITAL LETTER R WITH CEDILLA + '\u0157' # 0xE0 -> LATIN SMALL LETTER R WITH CEDILLA + '\u0160' # 0xE1 -> LATIN CAPITAL LETTER S WITH CARON + '\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK + '\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK + '\u0161' # 0xE4 -> LATIN SMALL LETTER S WITH CARON + '\u015a' # 0xE5 -> LATIN CAPITAL LETTER S WITH ACUTE + '\u015b' # 0xE6 -> LATIN SMALL LETTER S WITH ACUTE + '\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE + '\u0164' # 0xE8 -> LATIN CAPITAL LETTER T WITH CARON + '\u0165' # 0xE9 -> LATIN SMALL LETTER T WITH CARON + '\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE + '\u017d' # 0xEB -> LATIN CAPITAL LETTER Z WITH CARON + '\u017e' # 0xEC -> LATIN SMALL LETTER Z WITH CARON + '\u016a' # 0xED -> LATIN CAPITAL LETTER U WITH MACRON + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u016b' # 0xF0 -> LATIN SMALL LETTER U WITH MACRON + '\u016e' # 0xF1 -> LATIN CAPITAL LETTER U WITH RING ABOVE + '\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE + '\u016f' # 0xF3 -> LATIN SMALL LETTER U WITH RING ABOVE + '\u0170' # 0xF4 -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE + '\u0171' # 0xF5 -> LATIN SMALL LETTER U WITH DOUBLE ACUTE + '\u0172' # 0xF6 -> LATIN CAPITAL LETTER U WITH OGONEK + '\u0173' # 0xF7 -> LATIN SMALL LETTER U WITH OGONEK + '\xdd' # 0xF8 -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xfd' # 0xF9 -> LATIN SMALL LETTER Y WITH ACUTE + '\u0137' # 0xFA -> LATIN SMALL LETTER K WITH CEDILLA + '\u017b' # 0xFB -> LATIN CAPITAL LETTER Z WITH DOT ABOVE + '\u0141' # 0xFC -> LATIN CAPITAL LETTER L WITH STROKE + '\u017c' # 0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE + '\u0122' # 0xFE -> LATIN CAPITAL LETTER G WITH CEDILLA + '\u02c7' # 0xFF -> CARON ) ### Encoding table diff --git a/Lib/encodings/mac_croatian.py b/Lib/encodings/mac_croatian.py index f57f7b4..4a92fe6 100644 --- a/Lib/encodings/mac_croatian.py +++ b/Lib/encodings/mac_croatian.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u2020' # 0xA0 -> DAGGER - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\xb4' # 0xAB -> ACUTE ACCENT - u'\xa8' # 0xAC -> DIAERESIS - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON - u'\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE - u'\u221e' # 0xB0 -> INFINITY - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\u2206' # 0xB4 -> INCREMENT - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL - u'\u2211' # 0xB7 -> N-ARY SUMMATION - u'\u220f' # 0xB8 -> N-ARY PRODUCT - u'\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON - u'\u222b' # 0xBA -> INTEGRAL - u'\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR - u'\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR - u'\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA - u'\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON - u'\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE - u'\xbf' # 0xC0 -> INVERTED QUESTION MARK - u'\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u0106' # 0xC6 -> LATIN CAPITAL LETTER C WITH ACUTE - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE - u'\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE - u'\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE - u'\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u25ca' # 0xD7 -> LOZENGE - u'\uf8ff' # 0xD8 -> Apple logo - u'\xa9' # 0xD9 -> COPYRIGHT SIGN - u'\u2044' # 0xDA -> FRACTION SLASH - u'\u20ac' # 0xDB -> EURO SIGN - u'\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\xc6' # 0xDE -> LATIN CAPITAL LETTER AE - u'\xbb' # 0xDF -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2013' # 0xE0 -> EN DASH - u'\xb7' # 0xE1 -> MIDDLE DOT - u'\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK - u'\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2030' # 0xE4 -> PER MILLE SIGN - u'\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\u0107' # 0xE6 -> LATIN SMALL LETTER C WITH ACUTE - u'\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON - u'\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE - u'\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I - u'\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u02dc' # 0xF7 -> SMALL TILDE - u'\xaf' # 0xF8 -> MACRON - u'\u03c0' # 0xF9 -> GREEK SMALL LETTER PI - u'\xcb' # 0xFA -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\u02da' # 0xFB -> RING ABOVE - u'\xb8' # 0xFC -> CEDILLA - u'\xca' # 0xFD -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xe6' # 0xFE -> LATIN SMALL LETTER AE - u'\u02c7' # 0xFF -> CARON + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\u2020' # 0xA0 -> DAGGER + '\xb0' # 0xA1 -> DEGREE SIGN + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON + '\u2122' # 0xAA -> TRADE MARK SIGN + '\xb4' # 0xAB -> ACUTE ACCENT + '\xa8' # 0xAC -> DIAERESIS + '\u2260' # 0xAD -> NOT EQUAL TO + '\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON + '\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE + '\u221e' # 0xB0 -> INFINITY + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\u2206' # 0xB4 -> INCREMENT + '\xb5' # 0xB5 -> MICRO SIGN + '\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL + '\u2211' # 0xB7 -> N-ARY SUMMATION + '\u220f' # 0xB8 -> N-ARY PRODUCT + '\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON + '\u222b' # 0xBA -> INTEGRAL + '\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR + '\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR + '\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA + '\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON + '\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE + '\xbf' # 0xC0 -> INVERTED QUESTION MARK + '\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u0106' # 0xC6 -> LATIN CAPITAL LETTER C WITH ACUTE + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE + '\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE + '\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xCF -> LATIN SMALL LIGATURE OE + '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u25ca' # 0xD7 -> LOZENGE + '\uf8ff' # 0xD8 -> Apple logo + '\xa9' # 0xD9 -> COPYRIGHT SIGN + '\u2044' # 0xDA -> FRACTION SLASH + '\u20ac' # 0xDB -> EURO SIGN + '\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\xc6' # 0xDE -> LATIN CAPITAL LETTER AE + '\xbb' # 0xDF -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2013' # 0xE0 -> EN DASH + '\xb7' # 0xE1 -> MIDDLE DOT + '\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK + '\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK + '\u2030' # 0xE4 -> PER MILLE SIGN + '\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\u0107' # 0xE6 -> LATIN SMALL LETTER C WITH ACUTE + '\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE + '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON + '\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE + '\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE + '\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I + '\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u02dc' # 0xF7 -> SMALL TILDE + '\xaf' # 0xF8 -> MACRON + '\u03c0' # 0xF9 -> GREEK SMALL LETTER PI + '\xcb' # 0xFA -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\u02da' # 0xFB -> RING ABOVE + '\xb8' # 0xFC -> CEDILLA + '\xca' # 0xFD -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xe6' # 0xFE -> LATIN SMALL LETTER AE + '\u02c7' # 0xFF -> CARON ) ### Encoding table diff --git a/Lib/encodings/mac_cyrillic.py b/Lib/encodings/mac_cyrillic.py index 63324a1..d20272a 100644 --- a/Lib/encodings/mac_cyrillic.py +++ b/Lib/encodings/mac_cyrillic.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\u0410' # 0x80 -> CYRILLIC CAPITAL LETTER A - u'\u0411' # 0x81 -> CYRILLIC CAPITAL LETTER BE - u'\u0412' # 0x82 -> CYRILLIC CAPITAL LETTER VE - u'\u0413' # 0x83 -> CYRILLIC CAPITAL LETTER GHE - u'\u0414' # 0x84 -> CYRILLIC CAPITAL LETTER DE - u'\u0415' # 0x85 -> CYRILLIC CAPITAL LETTER IE - u'\u0416' # 0x86 -> CYRILLIC CAPITAL LETTER ZHE - u'\u0417' # 0x87 -> CYRILLIC CAPITAL LETTER ZE - u'\u0418' # 0x88 -> CYRILLIC CAPITAL LETTER I - u'\u0419' # 0x89 -> CYRILLIC CAPITAL LETTER SHORT I - u'\u041a' # 0x8A -> CYRILLIC CAPITAL LETTER KA - u'\u041b' # 0x8B -> CYRILLIC CAPITAL LETTER EL - u'\u041c' # 0x8C -> CYRILLIC CAPITAL LETTER EM - u'\u041d' # 0x8D -> CYRILLIC CAPITAL LETTER EN - u'\u041e' # 0x8E -> CYRILLIC CAPITAL LETTER O - u'\u041f' # 0x8F -> CYRILLIC CAPITAL LETTER PE - u'\u0420' # 0x90 -> CYRILLIC CAPITAL LETTER ER - u'\u0421' # 0x91 -> CYRILLIC CAPITAL LETTER ES - u'\u0422' # 0x92 -> CYRILLIC CAPITAL LETTER TE - u'\u0423' # 0x93 -> CYRILLIC CAPITAL LETTER U - u'\u0424' # 0x94 -> CYRILLIC CAPITAL LETTER EF - u'\u0425' # 0x95 -> CYRILLIC CAPITAL LETTER HA - u'\u0426' # 0x96 -> CYRILLIC CAPITAL LETTER TSE - u'\u0427' # 0x97 -> CYRILLIC CAPITAL LETTER CHE - u'\u0428' # 0x98 -> CYRILLIC CAPITAL LETTER SHA - u'\u0429' # 0x99 -> CYRILLIC CAPITAL LETTER SHCHA - u'\u042a' # 0x9A -> CYRILLIC CAPITAL LETTER HARD SIGN - u'\u042b' # 0x9B -> CYRILLIC CAPITAL LETTER YERU - u'\u042c' # 0x9C -> CYRILLIC CAPITAL LETTER SOFT SIGN - u'\u042d' # 0x9D -> CYRILLIC CAPITAL LETTER E - u'\u042e' # 0x9E -> CYRILLIC CAPITAL LETTER YU - u'\u042f' # 0x9F -> CYRILLIC CAPITAL LETTER YA - u'\u2020' # 0xA0 -> DAGGER - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\u0490' # 0xA2 -> CYRILLIC CAPITAL LETTER GHE WITH UPTURN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\u0406' # 0xA7 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\u0402' # 0xAB -> CYRILLIC CAPITAL LETTER DJE - u'\u0452' # 0xAC -> CYRILLIC SMALL LETTER DJE - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\u0403' # 0xAE -> CYRILLIC CAPITAL LETTER GJE - u'\u0453' # 0xAF -> CYRILLIC SMALL LETTER GJE - u'\u221e' # 0xB0 -> INFINITY - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\u0456' # 0xB4 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u0491' # 0xB6 -> CYRILLIC SMALL LETTER GHE WITH UPTURN - u'\u0408' # 0xB7 -> CYRILLIC CAPITAL LETTER JE - u'\u0404' # 0xB8 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\u0454' # 0xB9 -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\u0407' # 0xBA -> CYRILLIC CAPITAL LETTER YI - u'\u0457' # 0xBB -> CYRILLIC SMALL LETTER YI - u'\u0409' # 0xBC -> CYRILLIC CAPITAL LETTER LJE - u'\u0459' # 0xBD -> CYRILLIC SMALL LETTER LJE - u'\u040a' # 0xBE -> CYRILLIC CAPITAL LETTER NJE - u'\u045a' # 0xBF -> CYRILLIC SMALL LETTER NJE - u'\u0458' # 0xC0 -> CYRILLIC SMALL LETTER JE - u'\u0405' # 0xC1 -> CYRILLIC CAPITAL LETTER DZE - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u2206' # 0xC6 -> INCREMENT - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\u040b' # 0xCB -> CYRILLIC CAPITAL LETTER TSHE - u'\u045b' # 0xCC -> CYRILLIC SMALL LETTER TSHE - u'\u040c' # 0xCD -> CYRILLIC CAPITAL LETTER KJE - u'\u045c' # 0xCE -> CYRILLIC SMALL LETTER KJE - u'\u0455' # 0xCF -> CYRILLIC SMALL LETTER DZE - u'\u2013' # 0xD0 -> EN DASH - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u201e' # 0xD7 -> DOUBLE LOW-9 QUOTATION MARK - u'\u040e' # 0xD8 -> CYRILLIC CAPITAL LETTER SHORT U - u'\u045e' # 0xD9 -> CYRILLIC SMALL LETTER SHORT U - u'\u040f' # 0xDA -> CYRILLIC CAPITAL LETTER DZHE - u'\u045f' # 0xDB -> CYRILLIC SMALL LETTER DZHE - u'\u2116' # 0xDC -> NUMERO SIGN - u'\u0401' # 0xDD -> CYRILLIC CAPITAL LETTER IO - u'\u0451' # 0xDE -> CYRILLIC SMALL LETTER IO - u'\u044f' # 0xDF -> CYRILLIC SMALL LETTER YA - u'\u0430' # 0xE0 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0xE1 -> CYRILLIC SMALL LETTER BE - u'\u0432' # 0xE2 -> CYRILLIC SMALL LETTER VE - u'\u0433' # 0xE3 -> CYRILLIC SMALL LETTER GHE - u'\u0434' # 0xE4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0xE5 -> CYRILLIC SMALL LETTER IE - u'\u0436' # 0xE6 -> CYRILLIC SMALL LETTER ZHE - u'\u0437' # 0xE7 -> CYRILLIC SMALL LETTER ZE - u'\u0438' # 0xE8 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0xE9 -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0xEA -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0xEB -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0xEC -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0xED -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0xEE -> CYRILLIC SMALL LETTER O - u'\u043f' # 0xEF -> CYRILLIC SMALL LETTER PE - u'\u0440' # 0xF0 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0xF1 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0xF2 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0xF3 -> CYRILLIC SMALL LETTER U - u'\u0444' # 0xF4 -> CYRILLIC SMALL LETTER EF - u'\u0445' # 0xF5 -> CYRILLIC SMALL LETTER HA - u'\u0446' # 0xF6 -> CYRILLIC SMALL LETTER TSE - u'\u0447' # 0xF7 -> CYRILLIC SMALL LETTER CHE - u'\u0448' # 0xF8 -> CYRILLIC SMALL LETTER SHA - u'\u0449' # 0xF9 -> CYRILLIC SMALL LETTER SHCHA - u'\u044a' # 0xFA -> CYRILLIC SMALL LETTER HARD SIGN - u'\u044b' # 0xFB -> CYRILLIC SMALL LETTER YERU - u'\u044c' # 0xFC -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044d' # 0xFD -> CYRILLIC SMALL LETTER E - u'\u044e' # 0xFE -> CYRILLIC SMALL LETTER YU - u'\u20ac' # 0xFF -> EURO SIGN + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\u0410' # 0x80 -> CYRILLIC CAPITAL LETTER A + '\u0411' # 0x81 -> CYRILLIC CAPITAL LETTER BE + '\u0412' # 0x82 -> CYRILLIC CAPITAL LETTER VE + '\u0413' # 0x83 -> CYRILLIC CAPITAL LETTER GHE + '\u0414' # 0x84 -> CYRILLIC CAPITAL LETTER DE + '\u0415' # 0x85 -> CYRILLIC CAPITAL LETTER IE + '\u0416' # 0x86 -> CYRILLIC CAPITAL LETTER ZHE + '\u0417' # 0x87 -> CYRILLIC CAPITAL LETTER ZE + '\u0418' # 0x88 -> CYRILLIC CAPITAL LETTER I + '\u0419' # 0x89 -> CYRILLIC CAPITAL LETTER SHORT I + '\u041a' # 0x8A -> CYRILLIC CAPITAL LETTER KA + '\u041b' # 0x8B -> CYRILLIC CAPITAL LETTER EL + '\u041c' # 0x8C -> CYRILLIC CAPITAL LETTER EM + '\u041d' # 0x8D -> CYRILLIC CAPITAL LETTER EN + '\u041e' # 0x8E -> CYRILLIC CAPITAL LETTER O + '\u041f' # 0x8F -> CYRILLIC CAPITAL LETTER PE + '\u0420' # 0x90 -> CYRILLIC CAPITAL LETTER ER + '\u0421' # 0x91 -> CYRILLIC CAPITAL LETTER ES + '\u0422' # 0x92 -> CYRILLIC CAPITAL LETTER TE + '\u0423' # 0x93 -> CYRILLIC CAPITAL LETTER U + '\u0424' # 0x94 -> CYRILLIC CAPITAL LETTER EF + '\u0425' # 0x95 -> CYRILLIC CAPITAL LETTER HA + '\u0426' # 0x96 -> CYRILLIC CAPITAL LETTER TSE + '\u0427' # 0x97 -> CYRILLIC CAPITAL LETTER CHE + '\u0428' # 0x98 -> CYRILLIC CAPITAL LETTER SHA + '\u0429' # 0x99 -> CYRILLIC CAPITAL LETTER SHCHA + '\u042a' # 0x9A -> CYRILLIC CAPITAL LETTER HARD SIGN + '\u042b' # 0x9B -> CYRILLIC CAPITAL LETTER YERU + '\u042c' # 0x9C -> CYRILLIC CAPITAL LETTER SOFT SIGN + '\u042d' # 0x9D -> CYRILLIC CAPITAL LETTER E + '\u042e' # 0x9E -> CYRILLIC CAPITAL LETTER YU + '\u042f' # 0x9F -> CYRILLIC CAPITAL LETTER YA + '\u2020' # 0xA0 -> DAGGER + '\xb0' # 0xA1 -> DEGREE SIGN + '\u0490' # 0xA2 -> CYRILLIC CAPITAL LETTER GHE WITH UPTURN + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\u0406' # 0xA7 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u2122' # 0xAA -> TRADE MARK SIGN + '\u0402' # 0xAB -> CYRILLIC CAPITAL LETTER DJE + '\u0452' # 0xAC -> CYRILLIC SMALL LETTER DJE + '\u2260' # 0xAD -> NOT EQUAL TO + '\u0403' # 0xAE -> CYRILLIC CAPITAL LETTER GJE + '\u0453' # 0xAF -> CYRILLIC SMALL LETTER GJE + '\u221e' # 0xB0 -> INFINITY + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\u0456' # 0xB4 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + '\xb5' # 0xB5 -> MICRO SIGN + '\u0491' # 0xB6 -> CYRILLIC SMALL LETTER GHE WITH UPTURN + '\u0408' # 0xB7 -> CYRILLIC CAPITAL LETTER JE + '\u0404' # 0xB8 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE + '\u0454' # 0xB9 -> CYRILLIC SMALL LETTER UKRAINIAN IE + '\u0407' # 0xBA -> CYRILLIC CAPITAL LETTER YI + '\u0457' # 0xBB -> CYRILLIC SMALL LETTER YI + '\u0409' # 0xBC -> CYRILLIC CAPITAL LETTER LJE + '\u0459' # 0xBD -> CYRILLIC SMALL LETTER LJE + '\u040a' # 0xBE -> CYRILLIC CAPITAL LETTER NJE + '\u045a' # 0xBF -> CYRILLIC SMALL LETTER NJE + '\u0458' # 0xC0 -> CYRILLIC SMALL LETTER JE + '\u0405' # 0xC1 -> CYRILLIC CAPITAL LETTER DZE + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u2206' # 0xC6 -> INCREMENT + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\u040b' # 0xCB -> CYRILLIC CAPITAL LETTER TSHE + '\u045b' # 0xCC -> CYRILLIC SMALL LETTER TSHE + '\u040c' # 0xCD -> CYRILLIC CAPITAL LETTER KJE + '\u045c' # 0xCE -> CYRILLIC SMALL LETTER KJE + '\u0455' # 0xCF -> CYRILLIC SMALL LETTER DZE + '\u2013' # 0xD0 -> EN DASH + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u201e' # 0xD7 -> DOUBLE LOW-9 QUOTATION MARK + '\u040e' # 0xD8 -> CYRILLIC CAPITAL LETTER SHORT U + '\u045e' # 0xD9 -> CYRILLIC SMALL LETTER SHORT U + '\u040f' # 0xDA -> CYRILLIC CAPITAL LETTER DZHE + '\u045f' # 0xDB -> CYRILLIC SMALL LETTER DZHE + '\u2116' # 0xDC -> NUMERO SIGN + '\u0401' # 0xDD -> CYRILLIC CAPITAL LETTER IO + '\u0451' # 0xDE -> CYRILLIC SMALL LETTER IO + '\u044f' # 0xDF -> CYRILLIC SMALL LETTER YA + '\u0430' # 0xE0 -> CYRILLIC SMALL LETTER A + '\u0431' # 0xE1 -> CYRILLIC SMALL LETTER BE + '\u0432' # 0xE2 -> CYRILLIC SMALL LETTER VE + '\u0433' # 0xE3 -> CYRILLIC SMALL LETTER GHE + '\u0434' # 0xE4 -> CYRILLIC SMALL LETTER DE + '\u0435' # 0xE5 -> CYRILLIC SMALL LETTER IE + '\u0436' # 0xE6 -> CYRILLIC SMALL LETTER ZHE + '\u0437' # 0xE7 -> CYRILLIC SMALL LETTER ZE + '\u0438' # 0xE8 -> CYRILLIC SMALL LETTER I + '\u0439' # 0xE9 -> CYRILLIC SMALL LETTER SHORT I + '\u043a' # 0xEA -> CYRILLIC SMALL LETTER KA + '\u043b' # 0xEB -> CYRILLIC SMALL LETTER EL + '\u043c' # 0xEC -> CYRILLIC SMALL LETTER EM + '\u043d' # 0xED -> CYRILLIC SMALL LETTER EN + '\u043e' # 0xEE -> CYRILLIC SMALL LETTER O + '\u043f' # 0xEF -> CYRILLIC SMALL LETTER PE + '\u0440' # 0xF0 -> CYRILLIC SMALL LETTER ER + '\u0441' # 0xF1 -> CYRILLIC SMALL LETTER ES + '\u0442' # 0xF2 -> CYRILLIC SMALL LETTER TE + '\u0443' # 0xF3 -> CYRILLIC SMALL LETTER U + '\u0444' # 0xF4 -> CYRILLIC SMALL LETTER EF + '\u0445' # 0xF5 -> CYRILLIC SMALL LETTER HA + '\u0446' # 0xF6 -> CYRILLIC SMALL LETTER TSE + '\u0447' # 0xF7 -> CYRILLIC SMALL LETTER CHE + '\u0448' # 0xF8 -> CYRILLIC SMALL LETTER SHA + '\u0449' # 0xF9 -> CYRILLIC SMALL LETTER SHCHA + '\u044a' # 0xFA -> CYRILLIC SMALL LETTER HARD SIGN + '\u044b' # 0xFB -> CYRILLIC SMALL LETTER YERU + '\u044c' # 0xFC -> CYRILLIC SMALL LETTER SOFT SIGN + '\u044d' # 0xFD -> CYRILLIC SMALL LETTER E + '\u044e' # 0xFE -> CYRILLIC SMALL LETTER YU + '\u20ac' # 0xFF -> EURO SIGN ) ### Encoding table diff --git a/Lib/encodings/mac_farsi.py b/Lib/encodings/mac_farsi.py index 9dbd76a..e357d43 100644 --- a/Lib/encodings/mac_farsi.py +++ b/Lib/encodings/mac_farsi.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE, left-right - u'!' # 0x21 -> EXCLAMATION MARK, left-right - u'"' # 0x22 -> QUOTATION MARK, left-right - u'#' # 0x23 -> NUMBER SIGN, left-right - u'$' # 0x24 -> DOLLAR SIGN, left-right - u'%' # 0x25 -> PERCENT SIGN, left-right - u'&' # 0x26 -> AMPERSAND, left-right - u"'" # 0x27 -> APOSTROPHE, left-right - u'(' # 0x28 -> LEFT PARENTHESIS, left-right - u')' # 0x29 -> RIGHT PARENTHESIS, left-right - u'*' # 0x2A -> ASTERISK, left-right - u'+' # 0x2B -> PLUS SIGN, left-right - u',' # 0x2C -> COMMA, left-right; in Arabic-script context, displayed as 0x066C ARABIC THOUSANDS SEPARATOR - u'-' # 0x2D -> HYPHEN-MINUS, left-right - u'.' # 0x2E -> FULL STOP, left-right; in Arabic-script context, displayed as 0x066B ARABIC DECIMAL SEPARATOR - u'/' # 0x2F -> SOLIDUS, left-right - u'0' # 0x30 -> DIGIT ZERO; in Arabic-script context, displayed as 0x06F0 EXTENDED ARABIC-INDIC DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE; in Arabic-script context, displayed as 0x06F1 EXTENDED ARABIC-INDIC DIGIT ONE - u'2' # 0x32 -> DIGIT TWO; in Arabic-script context, displayed as 0x06F2 EXTENDED ARABIC-INDIC DIGIT TWO - u'3' # 0x33 -> DIGIT THREE; in Arabic-script context, displayed as 0x06F3 EXTENDED ARABIC-INDIC DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR; in Arabic-script context, displayed as 0x06F4 EXTENDED ARABIC-INDIC DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE; in Arabic-script context, displayed as 0x06F5 EXTENDED ARABIC-INDIC DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX; in Arabic-script context, displayed as 0x06F6 EXTENDED ARABIC-INDIC DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN; in Arabic-script context, displayed as 0x06F7 EXTENDED ARABIC-INDIC DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT; in Arabic-script context, displayed as 0x06F8 EXTENDED ARABIC-INDIC DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE; in Arabic-script context, displayed as 0x06F9 EXTENDED ARABIC-INDIC DIGIT NINE - u':' # 0x3A -> COLON, left-right - u';' # 0x3B -> SEMICOLON, left-right - u'<' # 0x3C -> LESS-THAN SIGN, left-right - u'=' # 0x3D -> EQUALS SIGN, left-right - u'>' # 0x3E -> GREATER-THAN SIGN, left-right - u'?' # 0x3F -> QUESTION MARK, left-right - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET, left-right - u'\\' # 0x5C -> REVERSE SOLIDUS, left-right - u']' # 0x5D -> RIGHT SQUARE BRACKET, left-right - u'^' # 0x5E -> CIRCUMFLEX ACCENT, left-right - u'_' # 0x5F -> LOW LINE, left-right - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET, left-right - u'|' # 0x7C -> VERTICAL LINE, left-right - u'}' # 0x7D -> RIGHT CURLY BRACKET, left-right - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xa0' # 0x81 -> NO-BREAK SPACE, right-left - u'\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u06ba' # 0x8B -> ARABIC LETTER NOON GHUNNA - u'\xab' # 0x8C -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\u2026' # 0x93 -> HORIZONTAL ELLIPSIS, right-left - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\xbb' # 0x98 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf7' # 0x9B -> DIVISION SIGN, right-left - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u' ' # 0xA0 -> SPACE, right-left - u'!' # 0xA1 -> EXCLAMATION MARK, right-left - u'"' # 0xA2 -> QUOTATION MARK, right-left - u'#' # 0xA3 -> NUMBER SIGN, right-left - u'$' # 0xA4 -> DOLLAR SIGN, right-left - u'\u066a' # 0xA5 -> ARABIC PERCENT SIGN - u'&' # 0xA6 -> AMPERSAND, right-left - u"'" # 0xA7 -> APOSTROPHE, right-left - u'(' # 0xA8 -> LEFT PARENTHESIS, right-left - u')' # 0xA9 -> RIGHT PARENTHESIS, right-left - u'*' # 0xAA -> ASTERISK, right-left - u'+' # 0xAB -> PLUS SIGN, right-left - u'\u060c' # 0xAC -> ARABIC COMMA - u'-' # 0xAD -> HYPHEN-MINUS, right-left - u'.' # 0xAE -> FULL STOP, right-left - u'/' # 0xAF -> SOLIDUS, right-left - u'\u06f0' # 0xB0 -> EXTENDED ARABIC-INDIC DIGIT ZERO, right-left (need override) - u'\u06f1' # 0xB1 -> EXTENDED ARABIC-INDIC DIGIT ONE, right-left (need override) - u'\u06f2' # 0xB2 -> EXTENDED ARABIC-INDIC DIGIT TWO, right-left (need override) - u'\u06f3' # 0xB3 -> EXTENDED ARABIC-INDIC DIGIT THREE, right-left (need override) - u'\u06f4' # 0xB4 -> EXTENDED ARABIC-INDIC DIGIT FOUR, right-left (need override) - u'\u06f5' # 0xB5 -> EXTENDED ARABIC-INDIC DIGIT FIVE, right-left (need override) - u'\u06f6' # 0xB6 -> EXTENDED ARABIC-INDIC DIGIT SIX, right-left (need override) - u'\u06f7' # 0xB7 -> EXTENDED ARABIC-INDIC DIGIT SEVEN, right-left (need override) - u'\u06f8' # 0xB8 -> EXTENDED ARABIC-INDIC DIGIT EIGHT, right-left (need override) - u'\u06f9' # 0xB9 -> EXTENDED ARABIC-INDIC DIGIT NINE, right-left (need override) - u':' # 0xBA -> COLON, right-left - u'\u061b' # 0xBB -> ARABIC SEMICOLON - u'<' # 0xBC -> LESS-THAN SIGN, right-left - u'=' # 0xBD -> EQUALS SIGN, right-left - u'>' # 0xBE -> GREATER-THAN SIGN, right-left - u'\u061f' # 0xBF -> ARABIC QUESTION MARK - u'\u274a' # 0xC0 -> EIGHT TEARDROP-SPOKED PROPELLER ASTERISK, right-left - u'\u0621' # 0xC1 -> ARABIC LETTER HAMZA - u'\u0622' # 0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE - u'\u0623' # 0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE - u'\u0624' # 0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE - u'\u0625' # 0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW - u'\u0626' # 0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE - u'\u0627' # 0xC7 -> ARABIC LETTER ALEF - u'\u0628' # 0xC8 -> ARABIC LETTER BEH - u'\u0629' # 0xC9 -> ARABIC LETTER TEH MARBUTA - u'\u062a' # 0xCA -> ARABIC LETTER TEH - u'\u062b' # 0xCB -> ARABIC LETTER THEH - u'\u062c' # 0xCC -> ARABIC LETTER JEEM - u'\u062d' # 0xCD -> ARABIC LETTER HAH - u'\u062e' # 0xCE -> ARABIC LETTER KHAH - u'\u062f' # 0xCF -> ARABIC LETTER DAL - u'\u0630' # 0xD0 -> ARABIC LETTER THAL - u'\u0631' # 0xD1 -> ARABIC LETTER REH - u'\u0632' # 0xD2 -> ARABIC LETTER ZAIN - u'\u0633' # 0xD3 -> ARABIC LETTER SEEN - u'\u0634' # 0xD4 -> ARABIC LETTER SHEEN - u'\u0635' # 0xD5 -> ARABIC LETTER SAD - u'\u0636' # 0xD6 -> ARABIC LETTER DAD - u'\u0637' # 0xD7 -> ARABIC LETTER TAH - u'\u0638' # 0xD8 -> ARABIC LETTER ZAH - u'\u0639' # 0xD9 -> ARABIC LETTER AIN - u'\u063a' # 0xDA -> ARABIC LETTER GHAIN - u'[' # 0xDB -> LEFT SQUARE BRACKET, right-left - u'\\' # 0xDC -> REVERSE SOLIDUS, right-left - u']' # 0xDD -> RIGHT SQUARE BRACKET, right-left - u'^' # 0xDE -> CIRCUMFLEX ACCENT, right-left - u'_' # 0xDF -> LOW LINE, right-left - u'\u0640' # 0xE0 -> ARABIC TATWEEL - u'\u0641' # 0xE1 -> ARABIC LETTER FEH - u'\u0642' # 0xE2 -> ARABIC LETTER QAF - u'\u0643' # 0xE3 -> ARABIC LETTER KAF - u'\u0644' # 0xE4 -> ARABIC LETTER LAM - u'\u0645' # 0xE5 -> ARABIC LETTER MEEM - u'\u0646' # 0xE6 -> ARABIC LETTER NOON - u'\u0647' # 0xE7 -> ARABIC LETTER HEH - u'\u0648' # 0xE8 -> ARABIC LETTER WAW - u'\u0649' # 0xE9 -> ARABIC LETTER ALEF MAKSURA - u'\u064a' # 0xEA -> ARABIC LETTER YEH - u'\u064b' # 0xEB -> ARABIC FATHATAN - u'\u064c' # 0xEC -> ARABIC DAMMATAN - u'\u064d' # 0xED -> ARABIC KASRATAN - u'\u064e' # 0xEE -> ARABIC FATHA - u'\u064f' # 0xEF -> ARABIC DAMMA - u'\u0650' # 0xF0 -> ARABIC KASRA - u'\u0651' # 0xF1 -> ARABIC SHADDA - u'\u0652' # 0xF2 -> ARABIC SUKUN - u'\u067e' # 0xF3 -> ARABIC LETTER PEH - u'\u0679' # 0xF4 -> ARABIC LETTER TTEH - u'\u0686' # 0xF5 -> ARABIC LETTER TCHEH - u'\u06d5' # 0xF6 -> ARABIC LETTER AE - u'\u06a4' # 0xF7 -> ARABIC LETTER VEH - u'\u06af' # 0xF8 -> ARABIC LETTER GAF - u'\u0688' # 0xF9 -> ARABIC LETTER DDAL - u'\u0691' # 0xFA -> ARABIC LETTER RREH - u'{' # 0xFB -> LEFT CURLY BRACKET, right-left - u'|' # 0xFC -> VERTICAL LINE, right-left - u'}' # 0xFD -> RIGHT CURLY BRACKET, right-left - u'\u0698' # 0xFE -> ARABIC LETTER JEH - u'\u06d2' # 0xFF -> ARABIC LETTER YEH BARREE + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE, left-right + '!' # 0x21 -> EXCLAMATION MARK, left-right + '"' # 0x22 -> QUOTATION MARK, left-right + '#' # 0x23 -> NUMBER SIGN, left-right + '$' # 0x24 -> DOLLAR SIGN, left-right + '%' # 0x25 -> PERCENT SIGN, left-right + '&' # 0x26 -> AMPERSAND, left-right + "'" # 0x27 -> APOSTROPHE, left-right + '(' # 0x28 -> LEFT PARENTHESIS, left-right + ')' # 0x29 -> RIGHT PARENTHESIS, left-right + '*' # 0x2A -> ASTERISK, left-right + '+' # 0x2B -> PLUS SIGN, left-right + ',' # 0x2C -> COMMA, left-right; in Arabic-script context, displayed as 0x066C ARABIC THOUSANDS SEPARATOR + '-' # 0x2D -> HYPHEN-MINUS, left-right + '.' # 0x2E -> FULL STOP, left-right; in Arabic-script context, displayed as 0x066B ARABIC DECIMAL SEPARATOR + '/' # 0x2F -> SOLIDUS, left-right + '0' # 0x30 -> DIGIT ZERO; in Arabic-script context, displayed as 0x06F0 EXTENDED ARABIC-INDIC DIGIT ZERO + '1' # 0x31 -> DIGIT ONE; in Arabic-script context, displayed as 0x06F1 EXTENDED ARABIC-INDIC DIGIT ONE + '2' # 0x32 -> DIGIT TWO; in Arabic-script context, displayed as 0x06F2 EXTENDED ARABIC-INDIC DIGIT TWO + '3' # 0x33 -> DIGIT THREE; in Arabic-script context, displayed as 0x06F3 EXTENDED ARABIC-INDIC DIGIT THREE + '4' # 0x34 -> DIGIT FOUR; in Arabic-script context, displayed as 0x06F4 EXTENDED ARABIC-INDIC DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE; in Arabic-script context, displayed as 0x06F5 EXTENDED ARABIC-INDIC DIGIT FIVE + '6' # 0x36 -> DIGIT SIX; in Arabic-script context, displayed as 0x06F6 EXTENDED ARABIC-INDIC DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN; in Arabic-script context, displayed as 0x06F7 EXTENDED ARABIC-INDIC DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT; in Arabic-script context, displayed as 0x06F8 EXTENDED ARABIC-INDIC DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE; in Arabic-script context, displayed as 0x06F9 EXTENDED ARABIC-INDIC DIGIT NINE + ':' # 0x3A -> COLON, left-right + ';' # 0x3B -> SEMICOLON, left-right + '<' # 0x3C -> LESS-THAN SIGN, left-right + '=' # 0x3D -> EQUALS SIGN, left-right + '>' # 0x3E -> GREATER-THAN SIGN, left-right + '?' # 0x3F -> QUESTION MARK, left-right + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET, left-right + '\\' # 0x5C -> REVERSE SOLIDUS, left-right + ']' # 0x5D -> RIGHT SQUARE BRACKET, left-right + '^' # 0x5E -> CIRCUMFLEX ACCENT, left-right + '_' # 0x5F -> LOW LINE, left-right + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET, left-right + '|' # 0x7C -> VERTICAL LINE, left-right + '}' # 0x7D -> RIGHT CURLY BRACKET, left-right + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xa0' # 0x81 -> NO-BREAK SPACE, right-left + '\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\u06ba' # 0x8B -> ARABIC LETTER NOON GHUNNA + '\xab' # 0x8C -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\u2026' # 0x93 -> HORIZONTAL ELLIPSIS, right-left + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\xbb' # 0x98 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf7' # 0x9B -> DIVISION SIGN, right-left + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + ' ' # 0xA0 -> SPACE, right-left + '!' # 0xA1 -> EXCLAMATION MARK, right-left + '"' # 0xA2 -> QUOTATION MARK, right-left + '#' # 0xA3 -> NUMBER SIGN, right-left + '$' # 0xA4 -> DOLLAR SIGN, right-left + '\u066a' # 0xA5 -> ARABIC PERCENT SIGN + '&' # 0xA6 -> AMPERSAND, right-left + "'" # 0xA7 -> APOSTROPHE, right-left + '(' # 0xA8 -> LEFT PARENTHESIS, right-left + ')' # 0xA9 -> RIGHT PARENTHESIS, right-left + '*' # 0xAA -> ASTERISK, right-left + '+' # 0xAB -> PLUS SIGN, right-left + '\u060c' # 0xAC -> ARABIC COMMA + '-' # 0xAD -> HYPHEN-MINUS, right-left + '.' # 0xAE -> FULL STOP, right-left + '/' # 0xAF -> SOLIDUS, right-left + '\u06f0' # 0xB0 -> EXTENDED ARABIC-INDIC DIGIT ZERO, right-left (need override) + '\u06f1' # 0xB1 -> EXTENDED ARABIC-INDIC DIGIT ONE, right-left (need override) + '\u06f2' # 0xB2 -> EXTENDED ARABIC-INDIC DIGIT TWO, right-left (need override) + '\u06f3' # 0xB3 -> EXTENDED ARABIC-INDIC DIGIT THREE, right-left (need override) + '\u06f4' # 0xB4 -> EXTENDED ARABIC-INDIC DIGIT FOUR, right-left (need override) + '\u06f5' # 0xB5 -> EXTENDED ARABIC-INDIC DIGIT FIVE, right-left (need override) + '\u06f6' # 0xB6 -> EXTENDED ARABIC-INDIC DIGIT SIX, right-left (need override) + '\u06f7' # 0xB7 -> EXTENDED ARABIC-INDIC DIGIT SEVEN, right-left (need override) + '\u06f8' # 0xB8 -> EXTENDED ARABIC-INDIC DIGIT EIGHT, right-left (need override) + '\u06f9' # 0xB9 -> EXTENDED ARABIC-INDIC DIGIT NINE, right-left (need override) + ':' # 0xBA -> COLON, right-left + '\u061b' # 0xBB -> ARABIC SEMICOLON + '<' # 0xBC -> LESS-THAN SIGN, right-left + '=' # 0xBD -> EQUALS SIGN, right-left + '>' # 0xBE -> GREATER-THAN SIGN, right-left + '\u061f' # 0xBF -> ARABIC QUESTION MARK + '\u274a' # 0xC0 -> EIGHT TEARDROP-SPOKED PROPELLER ASTERISK, right-left + '\u0621' # 0xC1 -> ARABIC LETTER HAMZA + '\u0622' # 0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE + '\u0623' # 0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE + '\u0624' # 0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE + '\u0625' # 0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW + '\u0626' # 0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE + '\u0627' # 0xC7 -> ARABIC LETTER ALEF + '\u0628' # 0xC8 -> ARABIC LETTER BEH + '\u0629' # 0xC9 -> ARABIC LETTER TEH MARBUTA + '\u062a' # 0xCA -> ARABIC LETTER TEH + '\u062b' # 0xCB -> ARABIC LETTER THEH + '\u062c' # 0xCC -> ARABIC LETTER JEEM + '\u062d' # 0xCD -> ARABIC LETTER HAH + '\u062e' # 0xCE -> ARABIC LETTER KHAH + '\u062f' # 0xCF -> ARABIC LETTER DAL + '\u0630' # 0xD0 -> ARABIC LETTER THAL + '\u0631' # 0xD1 -> ARABIC LETTER REH + '\u0632' # 0xD2 -> ARABIC LETTER ZAIN + '\u0633' # 0xD3 -> ARABIC LETTER SEEN + '\u0634' # 0xD4 -> ARABIC LETTER SHEEN + '\u0635' # 0xD5 -> ARABIC LETTER SAD + '\u0636' # 0xD6 -> ARABIC LETTER DAD + '\u0637' # 0xD7 -> ARABIC LETTER TAH + '\u0638' # 0xD8 -> ARABIC LETTER ZAH + '\u0639' # 0xD9 -> ARABIC LETTER AIN + '\u063a' # 0xDA -> ARABIC LETTER GHAIN + '[' # 0xDB -> LEFT SQUARE BRACKET, right-left + '\\' # 0xDC -> REVERSE SOLIDUS, right-left + ']' # 0xDD -> RIGHT SQUARE BRACKET, right-left + '^' # 0xDE -> CIRCUMFLEX ACCENT, right-left + '_' # 0xDF -> LOW LINE, right-left + '\u0640' # 0xE0 -> ARABIC TATWEEL + '\u0641' # 0xE1 -> ARABIC LETTER FEH + '\u0642' # 0xE2 -> ARABIC LETTER QAF + '\u0643' # 0xE3 -> ARABIC LETTER KAF + '\u0644' # 0xE4 -> ARABIC LETTER LAM + '\u0645' # 0xE5 -> ARABIC LETTER MEEM + '\u0646' # 0xE6 -> ARABIC LETTER NOON + '\u0647' # 0xE7 -> ARABIC LETTER HEH + '\u0648' # 0xE8 -> ARABIC LETTER WAW + '\u0649' # 0xE9 -> ARABIC LETTER ALEF MAKSURA + '\u064a' # 0xEA -> ARABIC LETTER YEH + '\u064b' # 0xEB -> ARABIC FATHATAN + '\u064c' # 0xEC -> ARABIC DAMMATAN + '\u064d' # 0xED -> ARABIC KASRATAN + '\u064e' # 0xEE -> ARABIC FATHA + '\u064f' # 0xEF -> ARABIC DAMMA + '\u0650' # 0xF0 -> ARABIC KASRA + '\u0651' # 0xF1 -> ARABIC SHADDA + '\u0652' # 0xF2 -> ARABIC SUKUN + '\u067e' # 0xF3 -> ARABIC LETTER PEH + '\u0679' # 0xF4 -> ARABIC LETTER TTEH + '\u0686' # 0xF5 -> ARABIC LETTER TCHEH + '\u06d5' # 0xF6 -> ARABIC LETTER AE + '\u06a4' # 0xF7 -> ARABIC LETTER VEH + '\u06af' # 0xF8 -> ARABIC LETTER GAF + '\u0688' # 0xF9 -> ARABIC LETTER DDAL + '\u0691' # 0xFA -> ARABIC LETTER RREH + '{' # 0xFB -> LEFT CURLY BRACKET, right-left + '|' # 0xFC -> VERTICAL LINE, right-left + '}' # 0xFD -> RIGHT CURLY BRACKET, right-left + '\u0698' # 0xFE -> ARABIC LETTER JEH + '\u06d2' # 0xFF -> ARABIC LETTER YEH BARREE ) ### Encoding table diff --git a/Lib/encodings/mac_greek.py b/Lib/encodings/mac_greek.py index 68f4fff..d3d0c4f 100644 --- a/Lib/encodings/mac_greek.py +++ b/Lib/encodings/mac_greek.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xb9' # 0x81 -> SUPERSCRIPT ONE - u'\xb2' # 0x82 -> SUPERSCRIPT TWO - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xb3' # 0x84 -> SUPERSCRIPT THREE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\u0385' # 0x87 -> GREEK DIALYTIKA TONOS - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\u0384' # 0x8B -> GREEK TONOS - u'\xa8' # 0x8C -> DIAERESIS - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xa3' # 0x92 -> POUND SIGN - u'\u2122' # 0x93 -> TRADE MARK SIGN - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\u2022' # 0x96 -> BULLET - u'\xbd' # 0x97 -> VULGAR FRACTION ONE HALF - u'\u2030' # 0x98 -> PER MILLE SIGN - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xa6' # 0x9B -> BROKEN BAR - u'\u20ac' # 0x9C -> EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u2020' # 0xA0 -> DAGGER - u'\u0393' # 0xA1 -> GREEK CAPITAL LETTER GAMMA - u'\u0394' # 0xA2 -> GREEK CAPITAL LETTER DELTA - u'\u0398' # 0xA3 -> GREEK CAPITAL LETTER THETA - u'\u039b' # 0xA4 -> GREEK CAPITAL LETTER LAMDA - u'\u039e' # 0xA5 -> GREEK CAPITAL LETTER XI - u'\u03a0' # 0xA6 -> GREEK CAPITAL LETTER PI - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u03a3' # 0xAA -> GREEK CAPITAL LETTER SIGMA - u'\u03aa' # 0xAB -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - u'\xa7' # 0xAC -> SECTION SIGN - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\xb0' # 0xAE -> DEGREE SIGN - u'\xb7' # 0xAF -> MIDDLE DOT - u'\u0391' # 0xB0 -> GREEK CAPITAL LETTER ALPHA - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\xa5' # 0xB4 -> YEN SIGN - u'\u0392' # 0xB5 -> GREEK CAPITAL LETTER BETA - u'\u0395' # 0xB6 -> GREEK CAPITAL LETTER EPSILON - u'\u0396' # 0xB7 -> GREEK CAPITAL LETTER ZETA - u'\u0397' # 0xB8 -> GREEK CAPITAL LETTER ETA - u'\u0399' # 0xB9 -> GREEK CAPITAL LETTER IOTA - u'\u039a' # 0xBA -> GREEK CAPITAL LETTER KAPPA - u'\u039c' # 0xBB -> GREEK CAPITAL LETTER MU - u'\u03a6' # 0xBC -> GREEK CAPITAL LETTER PHI - u'\u03ab' # 0xBD -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - u'\u03a8' # 0xBE -> GREEK CAPITAL LETTER PSI - u'\u03a9' # 0xBF -> GREEK CAPITAL LETTER OMEGA - u'\u03ac' # 0xC0 -> GREEK SMALL LETTER ALPHA WITH TONOS - u'\u039d' # 0xC1 -> GREEK CAPITAL LETTER NU - u'\xac' # 0xC2 -> NOT SIGN - u'\u039f' # 0xC3 -> GREEK CAPITAL LETTER OMICRON - u'\u03a1' # 0xC4 -> GREEK CAPITAL LETTER RHO - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u03a4' # 0xC6 -> GREEK CAPITAL LETTER TAU - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\u03a5' # 0xCB -> GREEK CAPITAL LETTER UPSILON - u'\u03a7' # 0xCC -> GREEK CAPITAL LETTER CHI - u'\u0386' # 0xCD -> GREEK CAPITAL LETTER ALPHA WITH TONOS - u'\u0388' # 0xCE -> GREEK CAPITAL LETTER EPSILON WITH TONOS - u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE - u'\u2013' # 0xD0 -> EN DASH - u'\u2015' # 0xD1 -> HORIZONTAL BAR - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u0389' # 0xD7 -> GREEK CAPITAL LETTER ETA WITH TONOS - u'\u038a' # 0xD8 -> GREEK CAPITAL LETTER IOTA WITH TONOS - u'\u038c' # 0xD9 -> GREEK CAPITAL LETTER OMICRON WITH TONOS - u'\u038e' # 0xDA -> GREEK CAPITAL LETTER UPSILON WITH TONOS - u'\u03ad' # 0xDB -> GREEK SMALL LETTER EPSILON WITH TONOS - u'\u03ae' # 0xDC -> GREEK SMALL LETTER ETA WITH TONOS - u'\u03af' # 0xDD -> GREEK SMALL LETTER IOTA WITH TONOS - u'\u03cc' # 0xDE -> GREEK SMALL LETTER OMICRON WITH TONOS - u'\u038f' # 0xDF -> GREEK CAPITAL LETTER OMEGA WITH TONOS - u'\u03cd' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH TONOS - u'\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA - u'\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA - u'\u03c8' # 0xE3 -> GREEK SMALL LETTER PSI - u'\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA - u'\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON - u'\u03c6' # 0xE6 -> GREEK SMALL LETTER PHI - u'\u03b3' # 0xE7 -> GREEK SMALL LETTER GAMMA - u'\u03b7' # 0xE8 -> GREEK SMALL LETTER ETA - u'\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA - u'\u03be' # 0xEA -> GREEK SMALL LETTER XI - u'\u03ba' # 0xEB -> GREEK SMALL LETTER KAPPA - u'\u03bb' # 0xEC -> GREEK SMALL LETTER LAMDA - u'\u03bc' # 0xED -> GREEK SMALL LETTER MU - u'\u03bd' # 0xEE -> GREEK SMALL LETTER NU - u'\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON - u'\u03c0' # 0xF0 -> GREEK SMALL LETTER PI - u'\u03ce' # 0xF1 -> GREEK SMALL LETTER OMEGA WITH TONOS - u'\u03c1' # 0xF2 -> GREEK SMALL LETTER RHO - u'\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA - u'\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU - u'\u03b8' # 0xF5 -> GREEK SMALL LETTER THETA - u'\u03c9' # 0xF6 -> GREEK SMALL LETTER OMEGA - u'\u03c2' # 0xF7 -> GREEK SMALL LETTER FINAL SIGMA - u'\u03c7' # 0xF8 -> GREEK SMALL LETTER CHI - u'\u03c5' # 0xF9 -> GREEK SMALL LETTER UPSILON - u'\u03b6' # 0xFA -> GREEK SMALL LETTER ZETA - u'\u03ca' # 0xFB -> GREEK SMALL LETTER IOTA WITH DIALYTIKA - u'\u03cb' # 0xFC -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA - u'\u0390' # 0xFD -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - u'\u03b0' # 0xFE -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - u'\xad' # 0xFF -> SOFT HYPHEN # before Mac OS 9.2.2, was undefined + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xb9' # 0x81 -> SUPERSCRIPT ONE + '\xb2' # 0x82 -> SUPERSCRIPT TWO + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xb3' # 0x84 -> SUPERSCRIPT THREE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\u0385' # 0x87 -> GREEK DIALYTIKA TONOS + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\u0384' # 0x8B -> GREEK TONOS + '\xa8' # 0x8C -> DIAERESIS + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xa3' # 0x92 -> POUND SIGN + '\u2122' # 0x93 -> TRADE MARK SIGN + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\u2022' # 0x96 -> BULLET + '\xbd' # 0x97 -> VULGAR FRACTION ONE HALF + '\u2030' # 0x98 -> PER MILLE SIGN + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xa6' # 0x9B -> BROKEN BAR + '\u20ac' # 0x9C -> EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\u2020' # 0xA0 -> DAGGER + '\u0393' # 0xA1 -> GREEK CAPITAL LETTER GAMMA + '\u0394' # 0xA2 -> GREEK CAPITAL LETTER DELTA + '\u0398' # 0xA3 -> GREEK CAPITAL LETTER THETA + '\u039b' # 0xA4 -> GREEK CAPITAL LETTER LAMDA + '\u039e' # 0xA5 -> GREEK CAPITAL LETTER XI + '\u03a0' # 0xA6 -> GREEK CAPITAL LETTER PI + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u03a3' # 0xAA -> GREEK CAPITAL LETTER SIGMA + '\u03aa' # 0xAB -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA + '\xa7' # 0xAC -> SECTION SIGN + '\u2260' # 0xAD -> NOT EQUAL TO + '\xb0' # 0xAE -> DEGREE SIGN + '\xb7' # 0xAF -> MIDDLE DOT + '\u0391' # 0xB0 -> GREEK CAPITAL LETTER ALPHA + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\xa5' # 0xB4 -> YEN SIGN + '\u0392' # 0xB5 -> GREEK CAPITAL LETTER BETA + '\u0395' # 0xB6 -> GREEK CAPITAL LETTER EPSILON + '\u0396' # 0xB7 -> GREEK CAPITAL LETTER ZETA + '\u0397' # 0xB8 -> GREEK CAPITAL LETTER ETA + '\u0399' # 0xB9 -> GREEK CAPITAL LETTER IOTA + '\u039a' # 0xBA -> GREEK CAPITAL LETTER KAPPA + '\u039c' # 0xBB -> GREEK CAPITAL LETTER MU + '\u03a6' # 0xBC -> GREEK CAPITAL LETTER PHI + '\u03ab' # 0xBD -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA + '\u03a8' # 0xBE -> GREEK CAPITAL LETTER PSI + '\u03a9' # 0xBF -> GREEK CAPITAL LETTER OMEGA + '\u03ac' # 0xC0 -> GREEK SMALL LETTER ALPHA WITH TONOS + '\u039d' # 0xC1 -> GREEK CAPITAL LETTER NU + '\xac' # 0xC2 -> NOT SIGN + '\u039f' # 0xC3 -> GREEK CAPITAL LETTER OMICRON + '\u03a1' # 0xC4 -> GREEK CAPITAL LETTER RHO + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u03a4' # 0xC6 -> GREEK CAPITAL LETTER TAU + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\u03a5' # 0xCB -> GREEK CAPITAL LETTER UPSILON + '\u03a7' # 0xCC -> GREEK CAPITAL LETTER CHI + '\u0386' # 0xCD -> GREEK CAPITAL LETTER ALPHA WITH TONOS + '\u0388' # 0xCE -> GREEK CAPITAL LETTER EPSILON WITH TONOS + '\u0153' # 0xCF -> LATIN SMALL LIGATURE OE + '\u2013' # 0xD0 -> EN DASH + '\u2015' # 0xD1 -> HORIZONTAL BAR + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u0389' # 0xD7 -> GREEK CAPITAL LETTER ETA WITH TONOS + '\u038a' # 0xD8 -> GREEK CAPITAL LETTER IOTA WITH TONOS + '\u038c' # 0xD9 -> GREEK CAPITAL LETTER OMICRON WITH TONOS + '\u038e' # 0xDA -> GREEK CAPITAL LETTER UPSILON WITH TONOS + '\u03ad' # 0xDB -> GREEK SMALL LETTER EPSILON WITH TONOS + '\u03ae' # 0xDC -> GREEK SMALL LETTER ETA WITH TONOS + '\u03af' # 0xDD -> GREEK SMALL LETTER IOTA WITH TONOS + '\u03cc' # 0xDE -> GREEK SMALL LETTER OMICRON WITH TONOS + '\u038f' # 0xDF -> GREEK CAPITAL LETTER OMEGA WITH TONOS + '\u03cd' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH TONOS + '\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA + '\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA + '\u03c8' # 0xE3 -> GREEK SMALL LETTER PSI + '\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA + '\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON + '\u03c6' # 0xE6 -> GREEK SMALL LETTER PHI + '\u03b3' # 0xE7 -> GREEK SMALL LETTER GAMMA + '\u03b7' # 0xE8 -> GREEK SMALL LETTER ETA + '\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA + '\u03be' # 0xEA -> GREEK SMALL LETTER XI + '\u03ba' # 0xEB -> GREEK SMALL LETTER KAPPA + '\u03bb' # 0xEC -> GREEK SMALL LETTER LAMDA + '\u03bc' # 0xED -> GREEK SMALL LETTER MU + '\u03bd' # 0xEE -> GREEK SMALL LETTER NU + '\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON + '\u03c0' # 0xF0 -> GREEK SMALL LETTER PI + '\u03ce' # 0xF1 -> GREEK SMALL LETTER OMEGA WITH TONOS + '\u03c1' # 0xF2 -> GREEK SMALL LETTER RHO + '\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA + '\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU + '\u03b8' # 0xF5 -> GREEK SMALL LETTER THETA + '\u03c9' # 0xF6 -> GREEK SMALL LETTER OMEGA + '\u03c2' # 0xF7 -> GREEK SMALL LETTER FINAL SIGMA + '\u03c7' # 0xF8 -> GREEK SMALL LETTER CHI + '\u03c5' # 0xF9 -> GREEK SMALL LETTER UPSILON + '\u03b6' # 0xFA -> GREEK SMALL LETTER ZETA + '\u03ca' # 0xFB -> GREEK SMALL LETTER IOTA WITH DIALYTIKA + '\u03cb' # 0xFC -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA + '\u0390' # 0xFD -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS + '\u03b0' # 0xFE -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS + '\xad' # 0xFF -> SOFT HYPHEN # before Mac OS 9.2.2, was undefined ) ### Encoding table diff --git a/Lib/encodings/mac_iceland.py b/Lib/encodings/mac_iceland.py index c24add2..add10f4 100644 --- a/Lib/encodings/mac_iceland.py +++ b/Lib/encodings/mac_iceland.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\xdd' # 0xA0 -> LATIN CAPITAL LETTER Y WITH ACUTE - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\xb4' # 0xAB -> ACUTE ACCENT - u'\xa8' # 0xAC -> DIAERESIS - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\xc6' # 0xAE -> LATIN CAPITAL LETTER AE - u'\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE - u'\u221e' # 0xB0 -> INFINITY - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\xa5' # 0xB4 -> YEN SIGN - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL - u'\u2211' # 0xB7 -> N-ARY SUMMATION - u'\u220f' # 0xB8 -> N-ARY PRODUCT - u'\u03c0' # 0xB9 -> GREEK SMALL LETTER PI - u'\u222b' # 0xBA -> INTEGRAL - u'\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR - u'\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR - u'\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA - u'\xe6' # 0xBE -> LATIN SMALL LETTER AE - u'\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE - u'\xbf' # 0xC0 -> INVERTED QUESTION MARK - u'\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u2206' # 0xC6 -> INCREMENT - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE - u'\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE - u'\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE - u'\u2013' # 0xD0 -> EN DASH - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u25ca' # 0xD7 -> LOZENGE - u'\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\u2044' # 0xDA -> FRACTION SLASH - u'\u20ac' # 0xDB -> EURO SIGN - u'\xd0' # 0xDC -> LATIN CAPITAL LETTER ETH - u'\xf0' # 0xDD -> LATIN SMALL LETTER ETH - u'\xde' # 0xDE -> LATIN CAPITAL LETTER THORN - u'\xfe' # 0xDF -> LATIN SMALL LETTER THORN - u'\xfd' # 0xE0 -> LATIN SMALL LETTER Y WITH ACUTE - u'\xb7' # 0xE1 -> MIDDLE DOT - u'\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK - u'\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2030' # 0xE4 -> PER MILLE SIGN - u'\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\uf8ff' # 0xF0 -> Apple logo - u'\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I - u'\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u02dc' # 0xF7 -> SMALL TILDE - u'\xaf' # 0xF8 -> MACRON - u'\u02d8' # 0xF9 -> BREVE - u'\u02d9' # 0xFA -> DOT ABOVE - u'\u02da' # 0xFB -> RING ABOVE - u'\xb8' # 0xFC -> CEDILLA - u'\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT - u'\u02db' # 0xFE -> OGONEK - u'\u02c7' # 0xFF -> CARON + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\xdd' # 0xA0 -> LATIN CAPITAL LETTER Y WITH ACUTE + '\xb0' # 0xA1 -> DEGREE SIGN + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u2122' # 0xAA -> TRADE MARK SIGN + '\xb4' # 0xAB -> ACUTE ACCENT + '\xa8' # 0xAC -> DIAERESIS + '\u2260' # 0xAD -> NOT EQUAL TO + '\xc6' # 0xAE -> LATIN CAPITAL LETTER AE + '\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE + '\u221e' # 0xB0 -> INFINITY + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\xa5' # 0xB4 -> YEN SIGN + '\xb5' # 0xB5 -> MICRO SIGN + '\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL + '\u2211' # 0xB7 -> N-ARY SUMMATION + '\u220f' # 0xB8 -> N-ARY PRODUCT + '\u03c0' # 0xB9 -> GREEK SMALL LETTER PI + '\u222b' # 0xBA -> INTEGRAL + '\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR + '\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR + '\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA + '\xe6' # 0xBE -> LATIN SMALL LETTER AE + '\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE + '\xbf' # 0xC0 -> INVERTED QUESTION MARK + '\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u2206' # 0xC6 -> INCREMENT + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE + '\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE + '\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xCF -> LATIN SMALL LIGATURE OE + '\u2013' # 0xD0 -> EN DASH + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u25ca' # 0xD7 -> LOZENGE + '\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\u2044' # 0xDA -> FRACTION SLASH + '\u20ac' # 0xDB -> EURO SIGN + '\xd0' # 0xDC -> LATIN CAPITAL LETTER ETH + '\xf0' # 0xDD -> LATIN SMALL LETTER ETH + '\xde' # 0xDE -> LATIN CAPITAL LETTER THORN + '\xfe' # 0xDF -> LATIN SMALL LETTER THORN + '\xfd' # 0xE0 -> LATIN SMALL LETTER Y WITH ACUTE + '\xb7' # 0xE1 -> MIDDLE DOT + '\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK + '\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK + '\u2030' # 0xE4 -> PER MILLE SIGN + '\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\uf8ff' # 0xF0 -> Apple logo + '\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE + '\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I + '\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u02dc' # 0xF7 -> SMALL TILDE + '\xaf' # 0xF8 -> MACRON + '\u02d8' # 0xF9 -> BREVE + '\u02d9' # 0xFA -> DOT ABOVE + '\u02da' # 0xFB -> RING ABOVE + '\xb8' # 0xFC -> CEDILLA + '\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT + '\u02db' # 0xFE -> OGONEK + '\u02c7' # 0xFF -> CARON ) ### Encoding table diff --git a/Lib/encodings/mac_roman.py b/Lib/encodings/mac_roman.py index 62605ec..b74b002 100644 --- a/Lib/encodings/mac_roman.py +++ b/Lib/encodings/mac_roman.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u2020' # 0xA0 -> DAGGER - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\xb4' # 0xAB -> ACUTE ACCENT - u'\xa8' # 0xAC -> DIAERESIS - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\xc6' # 0xAE -> LATIN CAPITAL LETTER AE - u'\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE - u'\u221e' # 0xB0 -> INFINITY - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\xa5' # 0xB4 -> YEN SIGN - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL - u'\u2211' # 0xB7 -> N-ARY SUMMATION - u'\u220f' # 0xB8 -> N-ARY PRODUCT - u'\u03c0' # 0xB9 -> GREEK SMALL LETTER PI - u'\u222b' # 0xBA -> INTEGRAL - u'\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR - u'\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR - u'\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA - u'\xe6' # 0xBE -> LATIN SMALL LETTER AE - u'\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE - u'\xbf' # 0xC0 -> INVERTED QUESTION MARK - u'\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u2206' # 0xC6 -> INCREMENT - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE - u'\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE - u'\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE - u'\u2013' # 0xD0 -> EN DASH - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u25ca' # 0xD7 -> LOZENGE - u'\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\u2044' # 0xDA -> FRACTION SLASH - u'\u20ac' # 0xDB -> EURO SIGN - u'\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\ufb01' # 0xDE -> LATIN SMALL LIGATURE FI - u'\ufb02' # 0xDF -> LATIN SMALL LIGATURE FL - u'\u2021' # 0xE0 -> DOUBLE DAGGER - u'\xb7' # 0xE1 -> MIDDLE DOT - u'\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK - u'\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2030' # 0xE4 -> PER MILLE SIGN - u'\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\uf8ff' # 0xF0 -> Apple logo - u'\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I - u'\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u02dc' # 0xF7 -> SMALL TILDE - u'\xaf' # 0xF8 -> MACRON - u'\u02d8' # 0xF9 -> BREVE - u'\u02d9' # 0xFA -> DOT ABOVE - u'\u02da' # 0xFB -> RING ABOVE - u'\xb8' # 0xFC -> CEDILLA - u'\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT - u'\u02db' # 0xFE -> OGONEK - u'\u02c7' # 0xFF -> CARON + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\u2020' # 0xA0 -> DAGGER + '\xb0' # 0xA1 -> DEGREE SIGN + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u2122' # 0xAA -> TRADE MARK SIGN + '\xb4' # 0xAB -> ACUTE ACCENT + '\xa8' # 0xAC -> DIAERESIS + '\u2260' # 0xAD -> NOT EQUAL TO + '\xc6' # 0xAE -> LATIN CAPITAL LETTER AE + '\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE + '\u221e' # 0xB0 -> INFINITY + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\xa5' # 0xB4 -> YEN SIGN + '\xb5' # 0xB5 -> MICRO SIGN + '\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL + '\u2211' # 0xB7 -> N-ARY SUMMATION + '\u220f' # 0xB8 -> N-ARY PRODUCT + '\u03c0' # 0xB9 -> GREEK SMALL LETTER PI + '\u222b' # 0xBA -> INTEGRAL + '\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR + '\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR + '\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA + '\xe6' # 0xBE -> LATIN SMALL LETTER AE + '\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE + '\xbf' # 0xC0 -> INVERTED QUESTION MARK + '\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u2206' # 0xC6 -> INCREMENT + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE + '\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE + '\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xCF -> LATIN SMALL LIGATURE OE + '\u2013' # 0xD0 -> EN DASH + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u25ca' # 0xD7 -> LOZENGE + '\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\u2044' # 0xDA -> FRACTION SLASH + '\u20ac' # 0xDB -> EURO SIGN + '\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\ufb01' # 0xDE -> LATIN SMALL LIGATURE FI + '\ufb02' # 0xDF -> LATIN SMALL LIGATURE FL + '\u2021' # 0xE0 -> DOUBLE DAGGER + '\xb7' # 0xE1 -> MIDDLE DOT + '\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK + '\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK + '\u2030' # 0xE4 -> PER MILLE SIGN + '\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\uf8ff' # 0xF0 -> Apple logo + '\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE + '\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I + '\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u02dc' # 0xF7 -> SMALL TILDE + '\xaf' # 0xF8 -> MACRON + '\u02d8' # 0xF9 -> BREVE + '\u02d9' # 0xFA -> DOT ABOVE + '\u02da' # 0xFB -> RING ABOVE + '\xb8' # 0xFC -> CEDILLA + '\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT + '\u02db' # 0xFE -> OGONEK + '\u02c7' # 0xFF -> CARON ) ### Encoding table diff --git a/Lib/encodings/mac_romanian.py b/Lib/encodings/mac_romanian.py index 5bd5ae8..d141b4c 100644 --- a/Lib/encodings/mac_romanian.py +++ b/Lib/encodings/mac_romanian.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u2020' # 0xA0 -> DAGGER - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\xb4' # 0xAB -> ACUTE ACCENT - u'\xa8' # 0xAC -> DIAERESIS - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\u0102' # 0xAE -> LATIN CAPITAL LETTER A WITH BREVE - u'\u0218' # 0xAF -> LATIN CAPITAL LETTER S WITH COMMA BELOW # for Unicode 3.0 and later - u'\u221e' # 0xB0 -> INFINITY - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\xa5' # 0xB4 -> YEN SIGN - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL - u'\u2211' # 0xB7 -> N-ARY SUMMATION - u'\u220f' # 0xB8 -> N-ARY PRODUCT - u'\u03c0' # 0xB9 -> GREEK SMALL LETTER PI - u'\u222b' # 0xBA -> INTEGRAL - u'\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR - u'\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR - u'\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA - u'\u0103' # 0xBE -> LATIN SMALL LETTER A WITH BREVE - u'\u0219' # 0xBF -> LATIN SMALL LETTER S WITH COMMA BELOW # for Unicode 3.0 and later - u'\xbf' # 0xC0 -> INVERTED QUESTION MARK - u'\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u2206' # 0xC6 -> INCREMENT - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE - u'\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE - u'\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE - u'\u2013' # 0xD0 -> EN DASH - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u25ca' # 0xD7 -> LOZENGE - u'\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\u2044' # 0xDA -> FRACTION SLASH - u'\u20ac' # 0xDB -> EURO SIGN - u'\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK - u'\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - u'\u021a' # 0xDE -> LATIN CAPITAL LETTER T WITH COMMA BELOW # for Unicode 3.0 and later - u'\u021b' # 0xDF -> LATIN SMALL LETTER T WITH COMMA BELOW # for Unicode 3.0 and later - u'\u2021' # 0xE0 -> DOUBLE DAGGER - u'\xb7' # 0xE1 -> MIDDLE DOT - u'\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK - u'\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2030' # 0xE4 -> PER MILLE SIGN - u'\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\uf8ff' # 0xF0 -> Apple logo - u'\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I - u'\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u02dc' # 0xF7 -> SMALL TILDE - u'\xaf' # 0xF8 -> MACRON - u'\u02d8' # 0xF9 -> BREVE - u'\u02d9' # 0xFA -> DOT ABOVE - u'\u02da' # 0xFB -> RING ABOVE - u'\xb8' # 0xFC -> CEDILLA - u'\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT - u'\u02db' # 0xFE -> OGONEK - u'\u02c7' # 0xFF -> CARON + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\u2020' # 0xA0 -> DAGGER + '\xb0' # 0xA1 -> DEGREE SIGN + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u2122' # 0xAA -> TRADE MARK SIGN + '\xb4' # 0xAB -> ACUTE ACCENT + '\xa8' # 0xAC -> DIAERESIS + '\u2260' # 0xAD -> NOT EQUAL TO + '\u0102' # 0xAE -> LATIN CAPITAL LETTER A WITH BREVE + '\u0218' # 0xAF -> LATIN CAPITAL LETTER S WITH COMMA BELOW # for Unicode 3.0 and later + '\u221e' # 0xB0 -> INFINITY + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\xa5' # 0xB4 -> YEN SIGN + '\xb5' # 0xB5 -> MICRO SIGN + '\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL + '\u2211' # 0xB7 -> N-ARY SUMMATION + '\u220f' # 0xB8 -> N-ARY PRODUCT + '\u03c0' # 0xB9 -> GREEK SMALL LETTER PI + '\u222b' # 0xBA -> INTEGRAL + '\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR + '\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR + '\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA + '\u0103' # 0xBE -> LATIN SMALL LETTER A WITH BREVE + '\u0219' # 0xBF -> LATIN SMALL LETTER S WITH COMMA BELOW # for Unicode 3.0 and later + '\xbf' # 0xC0 -> INVERTED QUESTION MARK + '\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u2206' # 0xC6 -> INCREMENT + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE + '\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE + '\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xCF -> LATIN SMALL LIGATURE OE + '\u2013' # 0xD0 -> EN DASH + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u25ca' # 0xD7 -> LOZENGE + '\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\u2044' # 0xDA -> FRACTION SLASH + '\u20ac' # 0xDB -> EURO SIGN + '\u2039' # 0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK + '\u203a' # 0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + '\u021a' # 0xDE -> LATIN CAPITAL LETTER T WITH COMMA BELOW # for Unicode 3.0 and later + '\u021b' # 0xDF -> LATIN SMALL LETTER T WITH COMMA BELOW # for Unicode 3.0 and later + '\u2021' # 0xE0 -> DOUBLE DAGGER + '\xb7' # 0xE1 -> MIDDLE DOT + '\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK + '\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK + '\u2030' # 0xE4 -> PER MILLE SIGN + '\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\uf8ff' # 0xF0 -> Apple logo + '\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE + '\u0131' # 0xF5 -> LATIN SMALL LETTER DOTLESS I + '\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u02dc' # 0xF7 -> SMALL TILDE + '\xaf' # 0xF8 -> MACRON + '\u02d8' # 0xF9 -> BREVE + '\u02d9' # 0xFA -> DOT ABOVE + '\u02da' # 0xFB -> RING ABOVE + '\xb8' # 0xFC -> CEDILLA + '\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT + '\u02db' # 0xFE -> OGONEK + '\u02c7' # 0xFF -> CARON ) ### Encoding table diff --git a/Lib/encodings/mac_turkish.py b/Lib/encodings/mac_turkish.py index 0787f49..044d4cb 100644 --- a/Lib/encodings/mac_turkish.py +++ b/Lib/encodings/mac_turkish.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> CONTROL CHARACTER - u'\x01' # 0x01 -> CONTROL CHARACTER - u'\x02' # 0x02 -> CONTROL CHARACTER - u'\x03' # 0x03 -> CONTROL CHARACTER - u'\x04' # 0x04 -> CONTROL CHARACTER - u'\x05' # 0x05 -> CONTROL CHARACTER - u'\x06' # 0x06 -> CONTROL CHARACTER - u'\x07' # 0x07 -> CONTROL CHARACTER - u'\x08' # 0x08 -> CONTROL CHARACTER - u'\t' # 0x09 -> CONTROL CHARACTER - u'\n' # 0x0A -> CONTROL CHARACTER - u'\x0b' # 0x0B -> CONTROL CHARACTER - u'\x0c' # 0x0C -> CONTROL CHARACTER - u'\r' # 0x0D -> CONTROL CHARACTER - u'\x0e' # 0x0E -> CONTROL CHARACTER - u'\x0f' # 0x0F -> CONTROL CHARACTER - u'\x10' # 0x10 -> CONTROL CHARACTER - u'\x11' # 0x11 -> CONTROL CHARACTER - u'\x12' # 0x12 -> CONTROL CHARACTER - u'\x13' # 0x13 -> CONTROL CHARACTER - u'\x14' # 0x14 -> CONTROL CHARACTER - u'\x15' # 0x15 -> CONTROL CHARACTER - u'\x16' # 0x16 -> CONTROL CHARACTER - u'\x17' # 0x17 -> CONTROL CHARACTER - u'\x18' # 0x18 -> CONTROL CHARACTER - u'\x19' # 0x19 -> CONTROL CHARACTER - u'\x1a' # 0x1A -> CONTROL CHARACTER - u'\x1b' # 0x1B -> CONTROL CHARACTER - u'\x1c' # 0x1C -> CONTROL CHARACTER - u'\x1d' # 0x1D -> CONTROL CHARACTER - u'\x1e' # 0x1E -> CONTROL CHARACTER - u'\x1f' # 0x1F -> CONTROL CHARACTER - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> CONTROL CHARACTER - u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS - u'\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE - u'\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA - u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE - u'\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE - u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS - u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS - u'\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE - u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE - u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX - u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS - u'\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE - u'\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE - u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA - u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE - u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE - u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX - u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS - u'\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE - u'\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE - u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX - u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS - u'\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE - u'\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE - u'\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE - u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX - u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS - u'\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE - u'\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE - u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE - u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX - u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS - u'\u2020' # 0xA0 -> DAGGER - u'\xb0' # 0xA1 -> DEGREE SIGN - u'\xa2' # 0xA2 -> CENT SIGN - u'\xa3' # 0xA3 -> POUND SIGN - u'\xa7' # 0xA4 -> SECTION SIGN - u'\u2022' # 0xA5 -> BULLET - u'\xb6' # 0xA6 -> PILCROW SIGN - u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S - u'\xae' # 0xA8 -> REGISTERED SIGN - u'\xa9' # 0xA9 -> COPYRIGHT SIGN - u'\u2122' # 0xAA -> TRADE MARK SIGN - u'\xb4' # 0xAB -> ACUTE ACCENT - u'\xa8' # 0xAC -> DIAERESIS - u'\u2260' # 0xAD -> NOT EQUAL TO - u'\xc6' # 0xAE -> LATIN CAPITAL LETTER AE - u'\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE - u'\u221e' # 0xB0 -> INFINITY - u'\xb1' # 0xB1 -> PLUS-MINUS SIGN - u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO - u'\xa5' # 0xB4 -> YEN SIGN - u'\xb5' # 0xB5 -> MICRO SIGN - u'\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL - u'\u2211' # 0xB7 -> N-ARY SUMMATION - u'\u220f' # 0xB8 -> N-ARY PRODUCT - u'\u03c0' # 0xB9 -> GREEK SMALL LETTER PI - u'\u222b' # 0xBA -> INTEGRAL - u'\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR - u'\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR - u'\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA - u'\xe6' # 0xBE -> LATIN SMALL LETTER AE - u'\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE - u'\xbf' # 0xC0 -> INVERTED QUESTION MARK - u'\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK - u'\xac' # 0xC2 -> NOT SIGN - u'\u221a' # 0xC3 -> SQUARE ROOT - u'\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0xC5 -> ALMOST EQUAL TO - u'\u2206' # 0xC6 -> INCREMENT - u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0xCA -> NO-BREAK SPACE - u'\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE - u'\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE - u'\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE - u'\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE - u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE - u'\u2013' # 0xD0 -> EN DASH - u'\u2014' # 0xD1 -> EM DASH - u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0xD6 -> DIVISION SIGN - u'\u25ca' # 0xD7 -> LOZENGE - u'\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS - u'\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS - u'\u011e' # 0xDA -> LATIN CAPITAL LETTER G WITH BREVE - u'\u011f' # 0xDB -> LATIN SMALL LETTER G WITH BREVE - u'\u0130' # 0xDC -> LATIN CAPITAL LETTER I WITH DOT ABOVE - u'\u0131' # 0xDD -> LATIN SMALL LETTER DOTLESS I - u'\u015e' # 0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA - u'\u015f' # 0xDF -> LATIN SMALL LETTER S WITH CEDILLA - u'\u2021' # 0xE0 -> DOUBLE DAGGER - u'\xb7' # 0xE1 -> MIDDLE DOT - u'\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK - u'\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK - u'\u2030' # 0xE4 -> PER MILLE SIGN - u'\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX - u'\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX - u'\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE - u'\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS - u'\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE - u'\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE - u'\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX - u'\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS - u'\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE - u'\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE - u'\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX - u'\uf8ff' # 0xF0 -> Apple logo - u'\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE - u'\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE - u'\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX - u'\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE - u'\uf8a0' # 0xF5 -> undefined1 - u'\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT - u'\u02dc' # 0xF7 -> SMALL TILDE - u'\xaf' # 0xF8 -> MACRON - u'\u02d8' # 0xF9 -> BREVE - u'\u02d9' # 0xFA -> DOT ABOVE - u'\u02da' # 0xFB -> RING ABOVE - u'\xb8' # 0xFC -> CEDILLA - u'\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT - u'\u02db' # 0xFE -> OGONEK - u'\u02c7' # 0xFF -> CARON + '\x00' # 0x00 -> CONTROL CHARACTER + '\x01' # 0x01 -> CONTROL CHARACTER + '\x02' # 0x02 -> CONTROL CHARACTER + '\x03' # 0x03 -> CONTROL CHARACTER + '\x04' # 0x04 -> CONTROL CHARACTER + '\x05' # 0x05 -> CONTROL CHARACTER + '\x06' # 0x06 -> CONTROL CHARACTER + '\x07' # 0x07 -> CONTROL CHARACTER + '\x08' # 0x08 -> CONTROL CHARACTER + '\t' # 0x09 -> CONTROL CHARACTER + '\n' # 0x0A -> CONTROL CHARACTER + '\x0b' # 0x0B -> CONTROL CHARACTER + '\x0c' # 0x0C -> CONTROL CHARACTER + '\r' # 0x0D -> CONTROL CHARACTER + '\x0e' # 0x0E -> CONTROL CHARACTER + '\x0f' # 0x0F -> CONTROL CHARACTER + '\x10' # 0x10 -> CONTROL CHARACTER + '\x11' # 0x11 -> CONTROL CHARACTER + '\x12' # 0x12 -> CONTROL CHARACTER + '\x13' # 0x13 -> CONTROL CHARACTER + '\x14' # 0x14 -> CONTROL CHARACTER + '\x15' # 0x15 -> CONTROL CHARACTER + '\x16' # 0x16 -> CONTROL CHARACTER + '\x17' # 0x17 -> CONTROL CHARACTER + '\x18' # 0x18 -> CONTROL CHARACTER + '\x19' # 0x19 -> CONTROL CHARACTER + '\x1a' # 0x1A -> CONTROL CHARACTER + '\x1b' # 0x1B -> CONTROL CHARACTER + '\x1c' # 0x1C -> CONTROL CHARACTER + '\x1d' # 0x1D -> CONTROL CHARACTER + '\x1e' # 0x1E -> CONTROL CHARACTER + '\x1f' # 0x1F -> CONTROL CHARACTER + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> CONTROL CHARACTER + '\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS + '\xc5' # 0x81 -> LATIN CAPITAL LETTER A WITH RING ABOVE + '\xc7' # 0x82 -> LATIN CAPITAL LETTER C WITH CEDILLA + '\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE + '\xd1' # 0x84 -> LATIN CAPITAL LETTER N WITH TILDE + '\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS + '\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS + '\xe1' # 0x87 -> LATIN SMALL LETTER A WITH ACUTE + '\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE + '\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX + '\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS + '\xe3' # 0x8B -> LATIN SMALL LETTER A WITH TILDE + '\xe5' # 0x8C -> LATIN SMALL LETTER A WITH RING ABOVE + '\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA + '\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE + '\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE + '\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX + '\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS + '\xed' # 0x92 -> LATIN SMALL LETTER I WITH ACUTE + '\xec' # 0x93 -> LATIN SMALL LETTER I WITH GRAVE + '\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX + '\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS + '\xf1' # 0x96 -> LATIN SMALL LETTER N WITH TILDE + '\xf3' # 0x97 -> LATIN SMALL LETTER O WITH ACUTE + '\xf2' # 0x98 -> LATIN SMALL LETTER O WITH GRAVE + '\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX + '\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS + '\xf5' # 0x9B -> LATIN SMALL LETTER O WITH TILDE + '\xfa' # 0x9C -> LATIN SMALL LETTER U WITH ACUTE + '\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE + '\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX + '\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS + '\u2020' # 0xA0 -> DAGGER + '\xb0' # 0xA1 -> DEGREE SIGN + '\xa2' # 0xA2 -> CENT SIGN + '\xa3' # 0xA3 -> POUND SIGN + '\xa7' # 0xA4 -> SECTION SIGN + '\u2022' # 0xA5 -> BULLET + '\xb6' # 0xA6 -> PILCROW SIGN + '\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S + '\xae' # 0xA8 -> REGISTERED SIGN + '\xa9' # 0xA9 -> COPYRIGHT SIGN + '\u2122' # 0xAA -> TRADE MARK SIGN + '\xb4' # 0xAB -> ACUTE ACCENT + '\xa8' # 0xAC -> DIAERESIS + '\u2260' # 0xAD -> NOT EQUAL TO + '\xc6' # 0xAE -> LATIN CAPITAL LETTER AE + '\xd8' # 0xAF -> LATIN CAPITAL LETTER O WITH STROKE + '\u221e' # 0xB0 -> INFINITY + '\xb1' # 0xB1 -> PLUS-MINUS SIGN + '\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO + '\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO + '\xa5' # 0xB4 -> YEN SIGN + '\xb5' # 0xB5 -> MICRO SIGN + '\u2202' # 0xB6 -> PARTIAL DIFFERENTIAL + '\u2211' # 0xB7 -> N-ARY SUMMATION + '\u220f' # 0xB8 -> N-ARY PRODUCT + '\u03c0' # 0xB9 -> GREEK SMALL LETTER PI + '\u222b' # 0xBA -> INTEGRAL + '\xaa' # 0xBB -> FEMININE ORDINAL INDICATOR + '\xba' # 0xBC -> MASCULINE ORDINAL INDICATOR + '\u03a9' # 0xBD -> GREEK CAPITAL LETTER OMEGA + '\xe6' # 0xBE -> LATIN SMALL LETTER AE + '\xf8' # 0xBF -> LATIN SMALL LETTER O WITH STROKE + '\xbf' # 0xC0 -> INVERTED QUESTION MARK + '\xa1' # 0xC1 -> INVERTED EXCLAMATION MARK + '\xac' # 0xC2 -> NOT SIGN + '\u221a' # 0xC3 -> SQUARE ROOT + '\u0192' # 0xC4 -> LATIN SMALL LETTER F WITH HOOK + '\u2248' # 0xC5 -> ALMOST EQUAL TO + '\u2206' # 0xC6 -> INCREMENT + '\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + '\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + '\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS + '\xa0' # 0xCA -> NO-BREAK SPACE + '\xc0' # 0xCB -> LATIN CAPITAL LETTER A WITH GRAVE + '\xc3' # 0xCC -> LATIN CAPITAL LETTER A WITH TILDE + '\xd5' # 0xCD -> LATIN CAPITAL LETTER O WITH TILDE + '\u0152' # 0xCE -> LATIN CAPITAL LIGATURE OE + '\u0153' # 0xCF -> LATIN SMALL LIGATURE OE + '\u2013' # 0xD0 -> EN DASH + '\u2014' # 0xD1 -> EM DASH + '\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK + '\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK + '\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK + '\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK + '\xf7' # 0xD6 -> DIVISION SIGN + '\u25ca' # 0xD7 -> LOZENGE + '\xff' # 0xD8 -> LATIN SMALL LETTER Y WITH DIAERESIS + '\u0178' # 0xD9 -> LATIN CAPITAL LETTER Y WITH DIAERESIS + '\u011e' # 0xDA -> LATIN CAPITAL LETTER G WITH BREVE + '\u011f' # 0xDB -> LATIN SMALL LETTER G WITH BREVE + '\u0130' # 0xDC -> LATIN CAPITAL LETTER I WITH DOT ABOVE + '\u0131' # 0xDD -> LATIN SMALL LETTER DOTLESS I + '\u015e' # 0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA + '\u015f' # 0xDF -> LATIN SMALL LETTER S WITH CEDILLA + '\u2021' # 0xE0 -> DOUBLE DAGGER + '\xb7' # 0xE1 -> MIDDLE DOT + '\u201a' # 0xE2 -> SINGLE LOW-9 QUOTATION MARK + '\u201e' # 0xE3 -> DOUBLE LOW-9 QUOTATION MARK + '\u2030' # 0xE4 -> PER MILLE SIGN + '\xc2' # 0xE5 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX + '\xca' # 0xE6 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX + '\xc1' # 0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE + '\xcb' # 0xE8 -> LATIN CAPITAL LETTER E WITH DIAERESIS + '\xc8' # 0xE9 -> LATIN CAPITAL LETTER E WITH GRAVE + '\xcd' # 0xEA -> LATIN CAPITAL LETTER I WITH ACUTE + '\xce' # 0xEB -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX + '\xcf' # 0xEC -> LATIN CAPITAL LETTER I WITH DIAERESIS + '\xcc' # 0xED -> LATIN CAPITAL LETTER I WITH GRAVE + '\xd3' # 0xEE -> LATIN CAPITAL LETTER O WITH ACUTE + '\xd4' # 0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX + '\uf8ff' # 0xF0 -> Apple logo + '\xd2' # 0xF1 -> LATIN CAPITAL LETTER O WITH GRAVE + '\xda' # 0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE + '\xdb' # 0xF3 -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX + '\xd9' # 0xF4 -> LATIN CAPITAL LETTER U WITH GRAVE + '\uf8a0' # 0xF5 -> undefined1 + '\u02c6' # 0xF6 -> MODIFIER LETTER CIRCUMFLEX ACCENT + '\u02dc' # 0xF7 -> SMALL TILDE + '\xaf' # 0xF8 -> MACRON + '\u02d8' # 0xF9 -> BREVE + '\u02d9' # 0xFA -> DOT ABOVE + '\u02da' # 0xFB -> RING ABOVE + '\xb8' # 0xFC -> CEDILLA + '\u02dd' # 0xFD -> DOUBLE ACUTE ACCENT + '\u02db' # 0xFE -> OGONEK + '\u02c7' # 0xFF -> CARON ) ### Encoding table diff --git a/Lib/encodings/punycode.py b/Lib/encodings/punycode.py index 0e4aa0b..c86aad7 100644 --- a/Lib/encodings/punycode.py +++ b/Lib/encodings/punycode.py @@ -189,7 +189,7 @@ def punycode_decode(text, errors): else: base = text[:pos] extended = text[pos+1:] - base = unicode(base, "ascii", errors) + base = str(base, "ascii", errors) extended = extended.upper() return insertion_sort(base, extended, errors) diff --git a/Lib/encodings/tis_620.py b/Lib/encodings/tis_620.py index b2cd22b..e288386 100644 --- a/Lib/encodings/tis_620.py +++ b/Lib/encodings/tis_620.py @@ -45,262 +45,262 @@ def getregentry(): ### Decoding Table decoding_table = ( - u'\x00' # 0x00 -> NULL - u'\x01' # 0x01 -> START OF HEADING - u'\x02' # 0x02 -> START OF TEXT - u'\x03' # 0x03 -> END OF TEXT - u'\x04' # 0x04 -> END OF TRANSMISSION - u'\x05' # 0x05 -> ENQUIRY - u'\x06' # 0x06 -> ACKNOWLEDGE - u'\x07' # 0x07 -> BELL - u'\x08' # 0x08 -> BACKSPACE - u'\t' # 0x09 -> HORIZONTAL TABULATION - u'\n' # 0x0A -> LINE FEED - u'\x0b' # 0x0B -> VERTICAL TABULATION - u'\x0c' # 0x0C -> FORM FEED - u'\r' # 0x0D -> CARRIAGE RETURN - u'\x0e' # 0x0E -> SHIFT OUT - u'\x0f' # 0x0F -> SHIFT IN - u'\x10' # 0x10 -> DATA LINK ESCAPE - u'\x11' # 0x11 -> DEVICE CONTROL ONE - u'\x12' # 0x12 -> DEVICE CONTROL TWO - u'\x13' # 0x13 -> DEVICE CONTROL THREE - u'\x14' # 0x14 -> DEVICE CONTROL FOUR - u'\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE - u'\x16' # 0x16 -> SYNCHRONOUS IDLE - u'\x17' # 0x17 -> END OF TRANSMISSION BLOCK - u'\x18' # 0x18 -> CANCEL - u'\x19' # 0x19 -> END OF MEDIUM - u'\x1a' # 0x1A -> SUBSTITUTE - u'\x1b' # 0x1B -> ESCAPE - u'\x1c' # 0x1C -> FILE SEPARATOR - u'\x1d' # 0x1D -> GROUP SEPARATOR - u'\x1e' # 0x1E -> RECORD SEPARATOR - u'\x1f' # 0x1F -> UNIT SEPARATOR - u' ' # 0x20 -> SPACE - u'!' # 0x21 -> EXCLAMATION MARK - u'"' # 0x22 -> QUOTATION MARK - u'#' # 0x23 -> NUMBER SIGN - u'$' # 0x24 -> DOLLAR SIGN - u'%' # 0x25 -> PERCENT SIGN - u'&' # 0x26 -> AMPERSAND - u"'" # 0x27 -> APOSTROPHE - u'(' # 0x28 -> LEFT PARENTHESIS - u')' # 0x29 -> RIGHT PARENTHESIS - u'*' # 0x2A -> ASTERISK - u'+' # 0x2B -> PLUS SIGN - u',' # 0x2C -> COMMA - u'-' # 0x2D -> HYPHEN-MINUS - u'.' # 0x2E -> FULL STOP - u'/' # 0x2F -> SOLIDUS - u'0' # 0x30 -> DIGIT ZERO - u'1' # 0x31 -> DIGIT ONE - u'2' # 0x32 -> DIGIT TWO - u'3' # 0x33 -> DIGIT THREE - u'4' # 0x34 -> DIGIT FOUR - u'5' # 0x35 -> DIGIT FIVE - u'6' # 0x36 -> DIGIT SIX - u'7' # 0x37 -> DIGIT SEVEN - u'8' # 0x38 -> DIGIT EIGHT - u'9' # 0x39 -> DIGIT NINE - u':' # 0x3A -> COLON - u';' # 0x3B -> SEMICOLON - u'<' # 0x3C -> LESS-THAN SIGN - u'=' # 0x3D -> EQUALS SIGN - u'>' # 0x3E -> GREATER-THAN SIGN - u'?' # 0x3F -> QUESTION MARK - u'@' # 0x40 -> COMMERCIAL AT - u'A' # 0x41 -> LATIN CAPITAL LETTER A - u'B' # 0x42 -> LATIN CAPITAL LETTER B - u'C' # 0x43 -> LATIN CAPITAL LETTER C - u'D' # 0x44 -> LATIN CAPITAL LETTER D - u'E' # 0x45 -> LATIN CAPITAL LETTER E - u'F' # 0x46 -> LATIN CAPITAL LETTER F - u'G' # 0x47 -> LATIN CAPITAL LETTER G - u'H' # 0x48 -> LATIN CAPITAL LETTER H - u'I' # 0x49 -> LATIN CAPITAL LETTER I - u'J' # 0x4A -> LATIN CAPITAL LETTER J - u'K' # 0x4B -> LATIN CAPITAL LETTER K - u'L' # 0x4C -> LATIN CAPITAL LETTER L - u'M' # 0x4D -> LATIN CAPITAL LETTER M - u'N' # 0x4E -> LATIN CAPITAL LETTER N - u'O' # 0x4F -> LATIN CAPITAL LETTER O - u'P' # 0x50 -> LATIN CAPITAL LETTER P - u'Q' # 0x51 -> LATIN CAPITAL LETTER Q - u'R' # 0x52 -> LATIN CAPITAL LETTER R - u'S' # 0x53 -> LATIN CAPITAL LETTER S - u'T' # 0x54 -> LATIN CAPITAL LETTER T - u'U' # 0x55 -> LATIN CAPITAL LETTER U - u'V' # 0x56 -> LATIN CAPITAL LETTER V - u'W' # 0x57 -> LATIN CAPITAL LETTER W - u'X' # 0x58 -> LATIN CAPITAL LETTER X - u'Y' # 0x59 -> LATIN CAPITAL LETTER Y - u'Z' # 0x5A -> LATIN CAPITAL LETTER Z - u'[' # 0x5B -> LEFT SQUARE BRACKET - u'\\' # 0x5C -> REVERSE SOLIDUS - u']' # 0x5D -> RIGHT SQUARE BRACKET - u'^' # 0x5E -> CIRCUMFLEX ACCENT - u'_' # 0x5F -> LOW LINE - u'`' # 0x60 -> GRAVE ACCENT - u'a' # 0x61 -> LATIN SMALL LETTER A - u'b' # 0x62 -> LATIN SMALL LETTER B - u'c' # 0x63 -> LATIN SMALL LETTER C - u'd' # 0x64 -> LATIN SMALL LETTER D - u'e' # 0x65 -> LATIN SMALL LETTER E - u'f' # 0x66 -> LATIN SMALL LETTER F - u'g' # 0x67 -> LATIN SMALL LETTER G - u'h' # 0x68 -> LATIN SMALL LETTER H - u'i' # 0x69 -> LATIN SMALL LETTER I - u'j' # 0x6A -> LATIN SMALL LETTER J - u'k' # 0x6B -> LATIN SMALL LETTER K - u'l' # 0x6C -> LATIN SMALL LETTER L - u'm' # 0x6D -> LATIN SMALL LETTER M - u'n' # 0x6E -> LATIN SMALL LETTER N - u'o' # 0x6F -> LATIN SMALL LETTER O - u'p' # 0x70 -> LATIN SMALL LETTER P - u'q' # 0x71 -> LATIN SMALL LETTER Q - u'r' # 0x72 -> LATIN SMALL LETTER R - u's' # 0x73 -> LATIN SMALL LETTER S - u't' # 0x74 -> LATIN SMALL LETTER T - u'u' # 0x75 -> LATIN SMALL LETTER U - u'v' # 0x76 -> LATIN SMALL LETTER V - u'w' # 0x77 -> LATIN SMALL LETTER W - u'x' # 0x78 -> LATIN SMALL LETTER X - u'y' # 0x79 -> LATIN SMALL LETTER Y - u'z' # 0x7A -> LATIN SMALL LETTER Z - u'{' # 0x7B -> LEFT CURLY BRACKET - u'|' # 0x7C -> VERTICAL LINE - u'}' # 0x7D -> RIGHT CURLY BRACKET - u'~' # 0x7E -> TILDE - u'\x7f' # 0x7F -> DELETE - u'\x80' # 0x80 -> <control> - u'\x81' # 0x81 -> <control> - u'\x82' # 0x82 -> <control> - u'\x83' # 0x83 -> <control> - u'\x84' # 0x84 -> <control> - u'\x85' # 0x85 -> <control> - u'\x86' # 0x86 -> <control> - u'\x87' # 0x87 -> <control> - u'\x88' # 0x88 -> <control> - u'\x89' # 0x89 -> <control> - u'\x8a' # 0x8A -> <control> - u'\x8b' # 0x8B -> <control> - u'\x8c' # 0x8C -> <control> - u'\x8d' # 0x8D -> <control> - u'\x8e' # 0x8E -> <control> - u'\x8f' # 0x8F -> <control> - u'\x90' # 0x90 -> <control> - u'\x91' # 0x91 -> <control> - u'\x92' # 0x92 -> <control> - u'\x93' # 0x93 -> <control> - u'\x94' # 0x94 -> <control> - u'\x95' # 0x95 -> <control> - u'\x96' # 0x96 -> <control> - u'\x97' # 0x97 -> <control> - u'\x98' # 0x98 -> <control> - u'\x99' # 0x99 -> <control> - u'\x9a' # 0x9A -> <control> - u'\x9b' # 0x9B -> <control> - u'\x9c' # 0x9C -> <control> - u'\x9d' # 0x9D -> <control> - u'\x9e' # 0x9E -> <control> - u'\x9f' # 0x9F -> <control> - u'\ufffe' - u'\u0e01' # 0xA1 -> THAI CHARACTER KO KAI - u'\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI - u'\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT - u'\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI - u'\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON - u'\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG - u'\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU - u'\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN - u'\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING - u'\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG - u'\u0e0b' # 0xAB -> THAI CHARACTER SO SO - u'\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE - u'\u0e0d' # 0xAD -> THAI CHARACTER YO YING - u'\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA - u'\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK - u'\u0e10' # 0xB0 -> THAI CHARACTER THO THAN - u'\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO - u'\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO - u'\u0e13' # 0xB3 -> THAI CHARACTER NO NEN - u'\u0e14' # 0xB4 -> THAI CHARACTER DO DEK - u'\u0e15' # 0xB5 -> THAI CHARACTER TO TAO - u'\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG - u'\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN - u'\u0e18' # 0xB8 -> THAI CHARACTER THO THONG - u'\u0e19' # 0xB9 -> THAI CHARACTER NO NU - u'\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI - u'\u0e1b' # 0xBB -> THAI CHARACTER PO PLA - u'\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG - u'\u0e1d' # 0xBD -> THAI CHARACTER FO FA - u'\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN - u'\u0e1f' # 0xBF -> THAI CHARACTER FO FAN - u'\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO - u'\u0e21' # 0xC1 -> THAI CHARACTER MO MA - u'\u0e22' # 0xC2 -> THAI CHARACTER YO YAK - u'\u0e23' # 0xC3 -> THAI CHARACTER RO RUA - u'\u0e24' # 0xC4 -> THAI CHARACTER RU - u'\u0e25' # 0xC5 -> THAI CHARACTER LO LING - u'\u0e26' # 0xC6 -> THAI CHARACTER LU - u'\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN - u'\u0e28' # 0xC8 -> THAI CHARACTER SO SALA - u'\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI - u'\u0e2a' # 0xCA -> THAI CHARACTER SO SUA - u'\u0e2b' # 0xCB -> THAI CHARACTER HO HIP - u'\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA - u'\u0e2d' # 0xCD -> THAI CHARACTER O ANG - u'\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK - u'\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI - u'\u0e30' # 0xD0 -> THAI CHARACTER SARA A - u'\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT - u'\u0e32' # 0xD2 -> THAI CHARACTER SARA AA - u'\u0e33' # 0xD3 -> THAI CHARACTER SARA AM - u'\u0e34' # 0xD4 -> THAI CHARACTER SARA I - u'\u0e35' # 0xD5 -> THAI CHARACTER SARA II - u'\u0e36' # 0xD6 -> THAI CHARACTER SARA UE - u'\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE - u'\u0e38' # 0xD8 -> THAI CHARACTER SARA U - u'\u0e39' # 0xD9 -> THAI CHARACTER SARA UU - u'\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT - u'\u0e40' # 0xE0 -> THAI CHARACTER SARA E - u'\u0e41' # 0xE1 -> THAI CHARACTER SARA AE - u'\u0e42' # 0xE2 -> THAI CHARACTER SARA O - u'\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN - u'\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI - u'\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO - u'\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK - u'\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU - u'\u0e48' # 0xE8 -> THAI CHARACTER MAI EK - u'\u0e49' # 0xE9 -> THAI CHARACTER MAI THO - u'\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI - u'\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA - u'\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT - u'\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT - u'\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN - u'\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN - u'\u0e50' # 0xF0 -> THAI DIGIT ZERO - u'\u0e51' # 0xF1 -> THAI DIGIT ONE - u'\u0e52' # 0xF2 -> THAI DIGIT TWO - u'\u0e53' # 0xF3 -> THAI DIGIT THREE - u'\u0e54' # 0xF4 -> THAI DIGIT FOUR - u'\u0e55' # 0xF5 -> THAI DIGIT FIVE - u'\u0e56' # 0xF6 -> THAI DIGIT SIX - u'\u0e57' # 0xF7 -> THAI DIGIT SEVEN - u'\u0e58' # 0xF8 -> THAI DIGIT EIGHT - u'\u0e59' # 0xF9 -> THAI DIGIT NINE - u'\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU - u'\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT - u'\ufffe' - u'\ufffe' - u'\ufffe' - u'\ufffe' + '\x00' # 0x00 -> NULL + '\x01' # 0x01 -> START OF HEADING + '\x02' # 0x02 -> START OF TEXT + '\x03' # 0x03 -> END OF TEXT + '\x04' # 0x04 -> END OF TRANSMISSION + '\x05' # 0x05 -> ENQUIRY + '\x06' # 0x06 -> ACKNOWLEDGE + '\x07' # 0x07 -> BELL + '\x08' # 0x08 -> BACKSPACE + '\t' # 0x09 -> HORIZONTAL TABULATION + '\n' # 0x0A -> LINE FEED + '\x0b' # 0x0B -> VERTICAL TABULATION + '\x0c' # 0x0C -> FORM FEED + '\r' # 0x0D -> CARRIAGE RETURN + '\x0e' # 0x0E -> SHIFT OUT + '\x0f' # 0x0F -> SHIFT IN + '\x10' # 0x10 -> DATA LINK ESCAPE + '\x11' # 0x11 -> DEVICE CONTROL ONE + '\x12' # 0x12 -> DEVICE CONTROL TWO + '\x13' # 0x13 -> DEVICE CONTROL THREE + '\x14' # 0x14 -> DEVICE CONTROL FOUR + '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE + '\x16' # 0x16 -> SYNCHRONOUS IDLE + '\x17' # 0x17 -> END OF TRANSMISSION BLOCK + '\x18' # 0x18 -> CANCEL + '\x19' # 0x19 -> END OF MEDIUM + '\x1a' # 0x1A -> SUBSTITUTE + '\x1b' # 0x1B -> ESCAPE + '\x1c' # 0x1C -> FILE SEPARATOR + '\x1d' # 0x1D -> GROUP SEPARATOR + '\x1e' # 0x1E -> RECORD SEPARATOR + '\x1f' # 0x1F -> UNIT SEPARATOR + ' ' # 0x20 -> SPACE + '!' # 0x21 -> EXCLAMATION MARK + '"' # 0x22 -> QUOTATION MARK + '#' # 0x23 -> NUMBER SIGN + '$' # 0x24 -> DOLLAR SIGN + '%' # 0x25 -> PERCENT SIGN + '&' # 0x26 -> AMPERSAND + "'" # 0x27 -> APOSTROPHE + '(' # 0x28 -> LEFT PARENTHESIS + ')' # 0x29 -> RIGHT PARENTHESIS + '*' # 0x2A -> ASTERISK + '+' # 0x2B -> PLUS SIGN + ',' # 0x2C -> COMMA + '-' # 0x2D -> HYPHEN-MINUS + '.' # 0x2E -> FULL STOP + '/' # 0x2F -> SOLIDUS + '0' # 0x30 -> DIGIT ZERO + '1' # 0x31 -> DIGIT ONE + '2' # 0x32 -> DIGIT TWO + '3' # 0x33 -> DIGIT THREE + '4' # 0x34 -> DIGIT FOUR + '5' # 0x35 -> DIGIT FIVE + '6' # 0x36 -> DIGIT SIX + '7' # 0x37 -> DIGIT SEVEN + '8' # 0x38 -> DIGIT EIGHT + '9' # 0x39 -> DIGIT NINE + ':' # 0x3A -> COLON + ';' # 0x3B -> SEMICOLON + '<' # 0x3C -> LESS-THAN SIGN + '=' # 0x3D -> EQUALS SIGN + '>' # 0x3E -> GREATER-THAN SIGN + '?' # 0x3F -> QUESTION MARK + '@' # 0x40 -> COMMERCIAL AT + 'A' # 0x41 -> LATIN CAPITAL LETTER A + 'B' # 0x42 -> LATIN CAPITAL LETTER B + 'C' # 0x43 -> LATIN CAPITAL LETTER C + 'D' # 0x44 -> LATIN CAPITAL LETTER D + 'E' # 0x45 -> LATIN CAPITAL LETTER E + 'F' # 0x46 -> LATIN CAPITAL LETTER F + 'G' # 0x47 -> LATIN CAPITAL LETTER G + 'H' # 0x48 -> LATIN CAPITAL LETTER H + 'I' # 0x49 -> LATIN CAPITAL LETTER I + 'J' # 0x4A -> LATIN CAPITAL LETTER J + 'K' # 0x4B -> LATIN CAPITAL LETTER K + 'L' # 0x4C -> LATIN CAPITAL LETTER L + 'M' # 0x4D -> LATIN CAPITAL LETTER M + 'N' # 0x4E -> LATIN CAPITAL LETTER N + 'O' # 0x4F -> LATIN CAPITAL LETTER O + 'P' # 0x50 -> LATIN CAPITAL LETTER P + 'Q' # 0x51 -> LATIN CAPITAL LETTER Q + 'R' # 0x52 -> LATIN CAPITAL LETTER R + 'S' # 0x53 -> LATIN CAPITAL LETTER S + 'T' # 0x54 -> LATIN CAPITAL LETTER T + 'U' # 0x55 -> LATIN CAPITAL LETTER U + 'V' # 0x56 -> LATIN CAPITAL LETTER V + 'W' # 0x57 -> LATIN CAPITAL LETTER W + 'X' # 0x58 -> LATIN CAPITAL LETTER X + 'Y' # 0x59 -> LATIN CAPITAL LETTER Y + 'Z' # 0x5A -> LATIN CAPITAL LETTER Z + '[' # 0x5B -> LEFT SQUARE BRACKET + '\\' # 0x5C -> REVERSE SOLIDUS + ']' # 0x5D -> RIGHT SQUARE BRACKET + '^' # 0x5E -> CIRCUMFLEX ACCENT + '_' # 0x5F -> LOW LINE + '`' # 0x60 -> GRAVE ACCENT + 'a' # 0x61 -> LATIN SMALL LETTER A + 'b' # 0x62 -> LATIN SMALL LETTER B + 'c' # 0x63 -> LATIN SMALL LETTER C + 'd' # 0x64 -> LATIN SMALL LETTER D + 'e' # 0x65 -> LATIN SMALL LETTER E + 'f' # 0x66 -> LATIN SMALL LETTER F + 'g' # 0x67 -> LATIN SMALL LETTER G + 'h' # 0x68 -> LATIN SMALL LETTER H + 'i' # 0x69 -> LATIN SMALL LETTER I + 'j' # 0x6A -> LATIN SMALL LETTER J + 'k' # 0x6B -> LATIN SMALL LETTER K + 'l' # 0x6C -> LATIN SMALL LETTER L + 'm' # 0x6D -> LATIN SMALL LETTER M + 'n' # 0x6E -> LATIN SMALL LETTER N + 'o' # 0x6F -> LATIN SMALL LETTER O + 'p' # 0x70 -> LATIN SMALL LETTER P + 'q' # 0x71 -> LATIN SMALL LETTER Q + 'r' # 0x72 -> LATIN SMALL LETTER R + 's' # 0x73 -> LATIN SMALL LETTER S + 't' # 0x74 -> LATIN SMALL LETTER T + 'u' # 0x75 -> LATIN SMALL LETTER U + 'v' # 0x76 -> LATIN SMALL LETTER V + 'w' # 0x77 -> LATIN SMALL LETTER W + 'x' # 0x78 -> LATIN SMALL LETTER X + 'y' # 0x79 -> LATIN SMALL LETTER Y + 'z' # 0x7A -> LATIN SMALL LETTER Z + '{' # 0x7B -> LEFT CURLY BRACKET + '|' # 0x7C -> VERTICAL LINE + '}' # 0x7D -> RIGHT CURLY BRACKET + '~' # 0x7E -> TILDE + '\x7f' # 0x7F -> DELETE + '\x80' # 0x80 -> <control> + '\x81' # 0x81 -> <control> + '\x82' # 0x82 -> <control> + '\x83' # 0x83 -> <control> + '\x84' # 0x84 -> <control> + '\x85' # 0x85 -> <control> + '\x86' # 0x86 -> <control> + '\x87' # 0x87 -> <control> + '\x88' # 0x88 -> <control> + '\x89' # 0x89 -> <control> + '\x8a' # 0x8A -> <control> + '\x8b' # 0x8B -> <control> + '\x8c' # 0x8C -> <control> + '\x8d' # 0x8D -> <control> + '\x8e' # 0x8E -> <control> + '\x8f' # 0x8F -> <control> + '\x90' # 0x90 -> <control> + '\x91' # 0x91 -> <control> + '\x92' # 0x92 -> <control> + '\x93' # 0x93 -> <control> + '\x94' # 0x94 -> <control> + '\x95' # 0x95 -> <control> + '\x96' # 0x96 -> <control> + '\x97' # 0x97 -> <control> + '\x98' # 0x98 -> <control> + '\x99' # 0x99 -> <control> + '\x9a' # 0x9A -> <control> + '\x9b' # 0x9B -> <control> + '\x9c' # 0x9C -> <control> + '\x9d' # 0x9D -> <control> + '\x9e' # 0x9E -> <control> + '\x9f' # 0x9F -> <control> + '\ufffe' + '\u0e01' # 0xA1 -> THAI CHARACTER KO KAI + '\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI + '\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT + '\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI + '\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON + '\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG + '\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU + '\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN + '\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING + '\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG + '\u0e0b' # 0xAB -> THAI CHARACTER SO SO + '\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE + '\u0e0d' # 0xAD -> THAI CHARACTER YO YING + '\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA + '\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK + '\u0e10' # 0xB0 -> THAI CHARACTER THO THAN + '\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO + '\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO + '\u0e13' # 0xB3 -> THAI CHARACTER NO NEN + '\u0e14' # 0xB4 -> THAI CHARACTER DO DEK + '\u0e15' # 0xB5 -> THAI CHARACTER TO TAO + '\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG + '\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN + '\u0e18' # 0xB8 -> THAI CHARACTER THO THONG + '\u0e19' # 0xB9 -> THAI CHARACTER NO NU + '\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI + '\u0e1b' # 0xBB -> THAI CHARACTER PO PLA + '\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG + '\u0e1d' # 0xBD -> THAI CHARACTER FO FA + '\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN + '\u0e1f' # 0xBF -> THAI CHARACTER FO FAN + '\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO + '\u0e21' # 0xC1 -> THAI CHARACTER MO MA + '\u0e22' # 0xC2 -> THAI CHARACTER YO YAK + '\u0e23' # 0xC3 -> THAI CHARACTER RO RUA + '\u0e24' # 0xC4 -> THAI CHARACTER RU + '\u0e25' # 0xC5 -> THAI CHARACTER LO LING + '\u0e26' # 0xC6 -> THAI CHARACTER LU + '\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN + '\u0e28' # 0xC8 -> THAI CHARACTER SO SALA + '\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI + '\u0e2a' # 0xCA -> THAI CHARACTER SO SUA + '\u0e2b' # 0xCB -> THAI CHARACTER HO HIP + '\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA + '\u0e2d' # 0xCD -> THAI CHARACTER O ANG + '\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK + '\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI + '\u0e30' # 0xD0 -> THAI CHARACTER SARA A + '\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT + '\u0e32' # 0xD2 -> THAI CHARACTER SARA AA + '\u0e33' # 0xD3 -> THAI CHARACTER SARA AM + '\u0e34' # 0xD4 -> THAI CHARACTER SARA I + '\u0e35' # 0xD5 -> THAI CHARACTER SARA II + '\u0e36' # 0xD6 -> THAI CHARACTER SARA UE + '\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE + '\u0e38' # 0xD8 -> THAI CHARACTER SARA U + '\u0e39' # 0xD9 -> THAI CHARACTER SARA UU + '\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' + '\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT + '\u0e40' # 0xE0 -> THAI CHARACTER SARA E + '\u0e41' # 0xE1 -> THAI CHARACTER SARA AE + '\u0e42' # 0xE2 -> THAI CHARACTER SARA O + '\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN + '\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI + '\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO + '\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK + '\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU + '\u0e48' # 0xE8 -> THAI CHARACTER MAI EK + '\u0e49' # 0xE9 -> THAI CHARACTER MAI THO + '\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI + '\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA + '\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT + '\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT + '\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN + '\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN + '\u0e50' # 0xF0 -> THAI DIGIT ZERO + '\u0e51' # 0xF1 -> THAI DIGIT ONE + '\u0e52' # 0xF2 -> THAI DIGIT TWO + '\u0e53' # 0xF3 -> THAI DIGIT THREE + '\u0e54' # 0xF4 -> THAI DIGIT FOUR + '\u0e55' # 0xF5 -> THAI DIGIT FIVE + '\u0e56' # 0xF6 -> THAI DIGIT SIX + '\u0e57' # 0xF7 -> THAI DIGIT SEVEN + '\u0e58' # 0xF8 -> THAI DIGIT EIGHT + '\u0e59' # 0xF9 -> THAI DIGIT NINE + '\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU + '\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT + '\ufffe' + '\ufffe' + '\ufffe' + '\ufffe' ) ### Encoding table diff --git a/Lib/encodings/utf_8_sig.py b/Lib/encodings/utf_8_sig.py index a0cc1af..07cd5ee 100644 --- a/Lib/encodings/utf_8_sig.py +++ b/Lib/encodings/utf_8_sig.py @@ -57,7 +57,7 @@ class IncrementalDecoder(codecs.BufferedIncrementalDecoder): if codecs.BOM_UTF8.startswith(input): # not enough data to decide if this really is a BOM # => try again on the next call - return (u"", 0) + return ("", 0) else: self.first = 0 else: @@ -106,7 +106,7 @@ class StreamReader(codecs.StreamReader): if len(input) < 3 and codecs.BOM_UTF8.startswith(input): # not enough data to decide if this is a BOM # => try again on the next call - return (u"", 0) + return ("", 0) self.decode = codecs.utf_8_decode return decode(input, errors) |