diff options
author | Marc-André Lemburg <mal@egenix.com> | 2005-10-24 12:14:59 (GMT) |
---|---|---|
committer | Marc-André Lemburg <mal@egenix.com> | 2005-10-24 12:14:59 (GMT) |
commit | d9cf593b496f4441990eda7680518418eb80d62e (patch) | |
tree | 125b8b8859087c99e8c3be1d753289db6f058189 /Lib/encodings/cp1254.py | |
parent | 3c72ded23dfbeb66072171a5707b27d9984ef332 (diff) | |
download | cpython-d9cf593b496f4441990eda7680518418eb80d62e.zip cpython-d9cf593b496f4441990eda7680518418eb80d62e.tar.gz cpython-d9cf593b496f4441990eda7680518418eb80d62e.tar.bz2 |
Cosmetic change: make all hex literals use upper case hex so that they
look more like the Unicode Consortium files.
Add ending new-line to all source files.
Diffstat (limited to 'Lib/encodings/cp1254.py')
-rw-r--r-- | Lib/encodings/cp1254.py | 631 |
1 files changed, 316 insertions, 315 deletions
diff --git a/Lib/encodings/cp1254.py b/Lib/encodings/cp1254.py index 2dd48c1..e6242cb 100644 --- a/Lib/encodings/cp1254.py +++ b/Lib/encodings/cp1254.py @@ -42,12 +42,12 @@ decoding_table = ( 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'\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 @@ -58,12 +58,12 @@ decoding_table = ( 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'\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 @@ -74,12 +74,12 @@ decoding_table = ( 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'*' # 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 @@ -90,12 +90,12 @@ decoding_table = ( 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':' # 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 @@ -106,12 +106,12 @@ decoding_table = ( 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'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 @@ -122,12 +122,12 @@ decoding_table = ( 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'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 @@ -138,12 +138,12 @@ decoding_table = ( 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'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 @@ -154,12 +154,12 @@ decoding_table = ( 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'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 @@ -170,12 +170,12 @@ decoding_table = ( 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'\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 @@ -186,108 +186,108 @@ decoding_table = ( 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 + 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 ) ### Encoding Map @@ -303,12 +303,12 @@ encoding_map = { 0x0007: 0x07, # BELL 0x0008: 0x08, # BACKSPACE 0x0009: 0x09, # HORIZONTAL TABULATION - 0x000a: 0x0a, # LINE FEED - 0x000b: 0x0b, # VERTICAL TABULATION - 0x000c: 0x0c, # FORM FEED - 0x000d: 0x0d, # CARRIAGE RETURN - 0x000e: 0x0e, # SHIFT OUT - 0x000f: 0x0f, # SHIFT IN + 0x000A: 0x0A, # LINE FEED + 0x000B: 0x0B, # VERTICAL TABULATION + 0x000C: 0x0C, # FORM FEED + 0x000D: 0x0D, # CARRIAGE RETURN + 0x000E: 0x0E, # SHIFT OUT + 0x000F: 0x0F, # SHIFT IN 0x0010: 0x10, # DATA LINK ESCAPE 0x0011: 0x11, # DEVICE CONTROL ONE 0x0012: 0x12, # DEVICE CONTROL TWO @@ -319,12 +319,12 @@ encoding_map = { 0x0017: 0x17, # END OF TRANSMISSION BLOCK 0x0018: 0x18, # CANCEL 0x0019: 0x19, # END OF MEDIUM - 0x001a: 0x1a, # SUBSTITUTE - 0x001b: 0x1b, # ESCAPE - 0x001c: 0x1c, # FILE SEPARATOR - 0x001d: 0x1d, # GROUP SEPARATOR - 0x001e: 0x1e, # RECORD SEPARATOR - 0x001f: 0x1f, # UNIT SEPARATOR + 0x001A: 0x1A, # SUBSTITUTE + 0x001B: 0x1B, # ESCAPE + 0x001C: 0x1C, # FILE SEPARATOR + 0x001D: 0x1D, # GROUP SEPARATOR + 0x001E: 0x1E, # RECORD SEPARATOR + 0x001F: 0x1F, # UNIT SEPARATOR 0x0020: 0x20, # SPACE 0x0021: 0x21, # EXCLAMATION MARK 0x0022: 0x22, # QUOTATION MARK @@ -335,12 +335,12 @@ encoding_map = { 0x0027: 0x27, # APOSTROPHE 0x0028: 0x28, # LEFT PARENTHESIS 0x0029: 0x29, # RIGHT PARENTHESIS - 0x002a: 0x2a, # ASTERISK - 0x002b: 0x2b, # PLUS SIGN - 0x002c: 0x2c, # COMMA - 0x002d: 0x2d, # HYPHEN-MINUS - 0x002e: 0x2e, # FULL STOP - 0x002f: 0x2f, # SOLIDUS + 0x002A: 0x2A, # ASTERISK + 0x002B: 0x2B, # PLUS SIGN + 0x002C: 0x2C, # COMMA + 0x002D: 0x2D, # HYPHEN-MINUS + 0x002E: 0x2E, # FULL STOP + 0x002F: 0x2F, # SOLIDUS 0x0030: 0x30, # DIGIT ZERO 0x0031: 0x31, # DIGIT ONE 0x0032: 0x32, # DIGIT TWO @@ -351,12 +351,12 @@ encoding_map = { 0x0037: 0x37, # DIGIT SEVEN 0x0038: 0x38, # DIGIT EIGHT 0x0039: 0x39, # DIGIT NINE - 0x003a: 0x3a, # COLON - 0x003b: 0x3b, # SEMICOLON - 0x003c: 0x3c, # LESS-THAN SIGN - 0x003d: 0x3d, # EQUALS SIGN - 0x003e: 0x3e, # GREATER-THAN SIGN - 0x003f: 0x3f, # QUESTION MARK + 0x003A: 0x3A, # COLON + 0x003B: 0x3B, # SEMICOLON + 0x003C: 0x3C, # LESS-THAN SIGN + 0x003D: 0x3D, # EQUALS SIGN + 0x003E: 0x3E, # GREATER-THAN SIGN + 0x003F: 0x3F, # QUESTION MARK 0x0040: 0x40, # COMMERCIAL AT 0x0041: 0x41, # LATIN CAPITAL LETTER A 0x0042: 0x42, # LATIN CAPITAL LETTER B @@ -367,12 +367,12 @@ encoding_map = { 0x0047: 0x47, # LATIN CAPITAL LETTER G 0x0048: 0x48, # LATIN CAPITAL LETTER H 0x0049: 0x49, # LATIN CAPITAL LETTER I - 0x004a: 0x4a, # LATIN CAPITAL LETTER J - 0x004b: 0x4b, # LATIN CAPITAL LETTER K - 0x004c: 0x4c, # LATIN CAPITAL LETTER L - 0x004d: 0x4d, # LATIN CAPITAL LETTER M - 0x004e: 0x4e, # LATIN CAPITAL LETTER N - 0x004f: 0x4f, # LATIN CAPITAL LETTER O + 0x004A: 0x4A, # LATIN CAPITAL LETTER J + 0x004B: 0x4B, # LATIN CAPITAL LETTER K + 0x004C: 0x4C, # LATIN CAPITAL LETTER L + 0x004D: 0x4D, # LATIN CAPITAL LETTER M + 0x004E: 0x4E, # LATIN CAPITAL LETTER N + 0x004F: 0x4F, # LATIN CAPITAL LETTER O 0x0050: 0x50, # LATIN CAPITAL LETTER P 0x0051: 0x51, # LATIN CAPITAL LETTER Q 0x0052: 0x52, # LATIN CAPITAL LETTER R @@ -383,12 +383,12 @@ encoding_map = { 0x0057: 0x57, # LATIN CAPITAL LETTER W 0x0058: 0x58, # LATIN CAPITAL LETTER X 0x0059: 0x59, # LATIN CAPITAL LETTER Y - 0x005a: 0x5a, # LATIN CAPITAL LETTER Z - 0x005b: 0x5b, # LEFT SQUARE BRACKET - 0x005c: 0x5c, # REVERSE SOLIDUS - 0x005d: 0x5d, # RIGHT SQUARE BRACKET - 0x005e: 0x5e, # CIRCUMFLEX ACCENT - 0x005f: 0x5f, # LOW LINE + 0x005A: 0x5A, # LATIN CAPITAL LETTER Z + 0x005B: 0x5B, # LEFT SQUARE BRACKET + 0x005C: 0x5C, # REVERSE SOLIDUS + 0x005D: 0x5D, # RIGHT SQUARE BRACKET + 0x005E: 0x5E, # CIRCUMFLEX ACCENT + 0x005F: 0x5F, # LOW LINE 0x0060: 0x60, # GRAVE ACCENT 0x0061: 0x61, # LATIN SMALL LETTER A 0x0062: 0x62, # LATIN SMALL LETTER B @@ -399,12 +399,12 @@ encoding_map = { 0x0067: 0x67, # LATIN SMALL LETTER G 0x0068: 0x68, # LATIN SMALL LETTER H 0x0069: 0x69, # LATIN SMALL LETTER I - 0x006a: 0x6a, # LATIN SMALL LETTER J - 0x006b: 0x6b, # LATIN SMALL LETTER K - 0x006c: 0x6c, # LATIN SMALL LETTER L - 0x006d: 0x6d, # LATIN SMALL LETTER M - 0x006e: 0x6e, # LATIN SMALL LETTER N - 0x006f: 0x6f, # LATIN SMALL LETTER O + 0x006A: 0x6A, # LATIN SMALL LETTER J + 0x006B: 0x6B, # LATIN SMALL LETTER K + 0x006C: 0x6C, # LATIN SMALL LETTER L + 0x006D: 0x6D, # LATIN SMALL LETTER M + 0x006E: 0x6E, # LATIN SMALL LETTER N + 0x006F: 0x6F, # LATIN SMALL LETTER O 0x0070: 0x70, # LATIN SMALL LETTER P 0x0071: 0x71, # LATIN SMALL LETTER Q 0x0072: 0x72, # LATIN SMALL LETTER R @@ -415,131 +415,132 @@ encoding_map = { 0x0077: 0x77, # LATIN SMALL LETTER W 0x0078: 0x78, # LATIN SMALL LETTER X 0x0079: 0x79, # LATIN SMALL LETTER Y - 0x007a: 0x7a, # LATIN SMALL LETTER Z - 0x007b: 0x7b, # LEFT CURLY BRACKET - 0x007c: 0x7c, # VERTICAL LINE - 0x007d: 0x7d, # RIGHT CURLY BRACKET - 0x007e: 0x7e, # TILDE - 0x007f: 0x7f, # DELETE - 0x00a0: 0xa0, # NO-BREAK SPACE - 0x00a1: 0xa1, # INVERTED EXCLAMATION MARK - 0x00a2: 0xa2, # CENT SIGN - 0x00a3: 0xa3, # POUND SIGN - 0x00a4: 0xa4, # CURRENCY SIGN - 0x00a5: 0xa5, # YEN SIGN - 0x00a6: 0xa6, # BROKEN BAR - 0x00a7: 0xa7, # SECTION SIGN - 0x00a8: 0xa8, # DIAERESIS - 0x00a9: 0xa9, # COPYRIGHT SIGN - 0x00aa: 0xaa, # FEMININE ORDINAL INDICATOR - 0x00ab: 0xab, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00ac: 0xac, # NOT SIGN - 0x00ad: 0xad, # SOFT HYPHEN - 0x00ae: 0xae, # REGISTERED SIGN - 0x00af: 0xaf, # MACRON - 0x00b0: 0xb0, # DEGREE SIGN - 0x00b1: 0xb1, # PLUS-MINUS SIGN - 0x00b2: 0xb2, # SUPERSCRIPT TWO - 0x00b3: 0xb3, # SUPERSCRIPT THREE - 0x00b4: 0xb4, # ACUTE ACCENT - 0x00b5: 0xb5, # MICRO SIGN - 0x00b6: 0xb6, # PILCROW SIGN - 0x00b7: 0xb7, # MIDDLE DOT - 0x00b8: 0xb8, # CEDILLA - 0x00b9: 0xb9, # SUPERSCRIPT ONE - 0x00ba: 0xba, # MASCULINE ORDINAL INDICATOR - 0x00bb: 0xbb, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00bc: 0xbc, # VULGAR FRACTION ONE QUARTER - 0x00bd: 0xbd, # VULGAR FRACTION ONE HALF - 0x00be: 0xbe, # VULGAR FRACTION THREE QUARTERS - 0x00bf: 0xbf, # INVERTED QUESTION MARK - 0x00c0: 0xc0, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00c1: 0xc1, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00c2: 0xc2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00c3: 0xc3, # LATIN CAPITAL LETTER A WITH TILDE - 0x00c4: 0xc4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00c5: 0xc5, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00c6: 0xc6, # LATIN CAPITAL LETTER AE - 0x00c7: 0xc7, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00c8: 0xc8, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00c9: 0xc9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00ca: 0xca, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00cb: 0xcb, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00cc: 0xcc, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00cd: 0xcd, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00ce: 0xce, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00cf: 0xcf, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00d1: 0xd1, # LATIN CAPITAL LETTER N WITH TILDE - 0x00d2: 0xd2, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00d3: 0xd3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00d4: 0xd4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00d5: 0xd5, # LATIN CAPITAL LETTER O WITH TILDE - 0x00d6: 0xd6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00d7: 0xd7, # MULTIPLICATION SIGN - 0x00d8: 0xd8, # LATIN CAPITAL LETTER O WITH STROKE - 0x00d9: 0xd9, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00da: 0xda, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00db: 0xdb, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00dc: 0xdc, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00df: 0xdf, # LATIN SMALL LETTER SHARP S - 0x00e0: 0xe0, # LATIN SMALL LETTER A WITH GRAVE - 0x00e1: 0xe1, # LATIN SMALL LETTER A WITH ACUTE - 0x00e2: 0xe2, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00e3: 0xe3, # LATIN SMALL LETTER A WITH TILDE - 0x00e4: 0xe4, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00e5: 0xe5, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00e6: 0xe6, # LATIN SMALL LETTER AE - 0x00e7: 0xe7, # LATIN SMALL LETTER C WITH CEDILLA - 0x00e8: 0xe8, # LATIN SMALL LETTER E WITH GRAVE - 0x00e9: 0xe9, # LATIN SMALL LETTER E WITH ACUTE - 0x00ea: 0xea, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00eb: 0xeb, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00ec: 0xec, # LATIN SMALL LETTER I WITH GRAVE - 0x00ed: 0xed, # LATIN SMALL LETTER I WITH ACUTE - 0x00ee: 0xee, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00ef: 0xef, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00f1: 0xf1, # LATIN SMALL LETTER N WITH TILDE - 0x00f2: 0xf2, # LATIN SMALL LETTER O WITH GRAVE - 0x00f3: 0xf3, # LATIN SMALL LETTER O WITH ACUTE - 0x00f4: 0xf4, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00f5: 0xf5, # LATIN SMALL LETTER O WITH TILDE - 0x00f6: 0xf6, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00f7: 0xf7, # DIVISION SIGN - 0x00f8: 0xf8, # LATIN SMALL LETTER O WITH STROKE - 0x00f9: 0xf9, # LATIN SMALL LETTER U WITH GRAVE - 0x00fa: 0xfa, # LATIN SMALL LETTER U WITH ACUTE - 0x00fb: 0xfb, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00fc: 0xfc, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00ff: 0xff, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x011e: 0xd0, # LATIN CAPITAL LETTER G WITH BREVE - 0x011f: 0xf0, # LATIN SMALL LETTER G WITH BREVE - 0x0130: 0xdd, # LATIN CAPITAL LETTER I WITH DOT ABOVE - 0x0131: 0xfd, # LATIN SMALL LETTER DOTLESS I - 0x0152: 0x8c, # LATIN CAPITAL LIGATURE OE - 0x0153: 0x9c, # LATIN SMALL LIGATURE OE - 0x015e: 0xde, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015f: 0xfe, # LATIN SMALL LETTER S WITH CEDILLA - 0x0160: 0x8a, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0x9a, # LATIN SMALL LETTER S WITH CARON - 0x0178: 0x9f, # LATIN CAPITAL LETTER Y WITH DIAERESIS + 0x007A: 0x7A, # LATIN SMALL LETTER Z + 0x007B: 0x7B, # LEFT CURLY BRACKET + 0x007C: 0x7C, # VERTICAL LINE + 0x007D: 0x7D, # RIGHT CURLY BRACKET + 0x007E: 0x7E, # TILDE + 0x007F: 0x7F, # DELETE + 0x00A0: 0xA0, # NO-BREAK SPACE + 0x00A1: 0xA1, # INVERTED EXCLAMATION MARK + 0x00A2: 0xA2, # CENT SIGN + 0x00A3: 0xA3, # POUND SIGN + 0x00A4: 0xA4, # CURRENCY SIGN + 0x00A5: 0xA5, # YEN SIGN + 0x00A6: 0xA6, # BROKEN BAR + 0x00A7: 0xA7, # SECTION SIGN + 0x00A8: 0xA8, # DIAERESIS + 0x00A9: 0xA9, # COPYRIGHT SIGN + 0x00AA: 0xAA, # FEMININE ORDINAL INDICATOR + 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + 0x00AC: 0xAC, # NOT SIGN + 0x00AD: 0xAD, # SOFT HYPHEN + 0x00AE: 0xAE, # REGISTERED SIGN + 0x00AF: 0xAF, # MACRON + 0x00B0: 0xB0, # DEGREE SIGN + 0x00B1: 0xB1, # PLUS-MINUS SIGN + 0x00B2: 0xB2, # SUPERSCRIPT TWO + 0x00B3: 0xB3, # SUPERSCRIPT THREE + 0x00B4: 0xB4, # ACUTE ACCENT + 0x00B5: 0xB5, # MICRO SIGN + 0x00B6: 0xB6, # PILCROW SIGN + 0x00B7: 0xB7, # MIDDLE DOT + 0x00B8: 0xB8, # CEDILLA + 0x00B9: 0xB9, # SUPERSCRIPT ONE + 0x00BA: 0xBA, # MASCULINE ORDINAL INDICATOR + 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + 0x00BC: 0xBC, # VULGAR FRACTION ONE QUARTER + 0x00BD: 0xBD, # VULGAR FRACTION ONE HALF + 0x00BE: 0xBE, # VULGAR FRACTION THREE QUARTERS + 0x00BF: 0xBF, # INVERTED QUESTION MARK + 0x00C0: 0xC0, # LATIN CAPITAL LETTER A WITH GRAVE + 0x00C1: 0xC1, # LATIN CAPITAL LETTER A WITH ACUTE + 0x00C2: 0xC2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX + 0x00C3: 0xC3, # LATIN CAPITAL LETTER A WITH TILDE + 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS + 0x00C5: 0xC5, # LATIN CAPITAL LETTER A WITH RING ABOVE + 0x00C6: 0xC6, # LATIN CAPITAL LETTER AE + 0x00C7: 0xC7, # LATIN CAPITAL LETTER C WITH CEDILLA + 0x00C8: 0xC8, # LATIN CAPITAL LETTER E WITH GRAVE + 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE + 0x00CA: 0xCA, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX + 0x00CB: 0xCB, # LATIN CAPITAL LETTER E WITH DIAERESIS + 0x00CC: 0xCC, # LATIN CAPITAL LETTER I WITH GRAVE + 0x00CD: 0xCD, # LATIN CAPITAL LETTER I WITH ACUTE + 0x00CE: 0xCE, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX + 0x00CF: 0xCF, # LATIN CAPITAL LETTER I WITH DIAERESIS + 0x00D1: 0xD1, # LATIN CAPITAL LETTER N WITH TILDE + 0x00D2: 0xD2, # LATIN CAPITAL LETTER O WITH GRAVE + 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE + 0x00D4: 0xD4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX + 0x00D5: 0xD5, # LATIN CAPITAL LETTER O WITH TILDE + 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS + 0x00D7: 0xD7, # MULTIPLICATION SIGN + 0x00D8: 0xD8, # LATIN CAPITAL LETTER O WITH STROKE + 0x00D9: 0xD9, # LATIN CAPITAL LETTER U WITH GRAVE + 0x00DA: 0xDA, # LATIN CAPITAL LETTER U WITH ACUTE + 0x00DB: 0xDB, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX + 0x00DC: 0xDC, # LATIN CAPITAL LETTER U WITH DIAERESIS + 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S + 0x00E0: 0xE0, # LATIN SMALL LETTER A WITH GRAVE + 0x00E1: 0xE1, # LATIN SMALL LETTER A WITH ACUTE + 0x00E2: 0xE2, # LATIN SMALL LETTER A WITH CIRCUMFLEX + 0x00E3: 0xE3, # LATIN SMALL LETTER A WITH TILDE + 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS + 0x00E5: 0xE5, # LATIN SMALL LETTER A WITH RING ABOVE + 0x00E6: 0xE6, # LATIN SMALL LETTER AE + 0x00E7: 0xE7, # LATIN SMALL LETTER C WITH CEDILLA + 0x00E8: 0xE8, # LATIN SMALL LETTER E WITH GRAVE + 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE + 0x00EA: 0xEA, # LATIN SMALL LETTER E WITH CIRCUMFLEX + 0x00EB: 0xEB, # LATIN SMALL LETTER E WITH DIAERESIS + 0x00EC: 0xEC, # LATIN SMALL LETTER I WITH GRAVE + 0x00ED: 0xED, # LATIN SMALL LETTER I WITH ACUTE + 0x00EE: 0xEE, # LATIN SMALL LETTER I WITH CIRCUMFLEX + 0x00EF: 0xEF, # LATIN SMALL LETTER I WITH DIAERESIS + 0x00F1: 0xF1, # LATIN SMALL LETTER N WITH TILDE + 0x00F2: 0xF2, # LATIN SMALL LETTER O WITH GRAVE + 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE + 0x00F4: 0xF4, # LATIN SMALL LETTER O WITH CIRCUMFLEX + 0x00F5: 0xF5, # LATIN SMALL LETTER O WITH TILDE + 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS + 0x00F7: 0xF7, # DIVISION SIGN + 0x00F8: 0xF8, # LATIN SMALL LETTER O WITH STROKE + 0x00F9: 0xF9, # LATIN SMALL LETTER U WITH GRAVE + 0x00FA: 0xFA, # LATIN SMALL LETTER U WITH ACUTE + 0x00FB: 0xFB, # LATIN SMALL LETTER U WITH CIRCUMFLEX + 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS + 0x00FF: 0xFF, # LATIN SMALL LETTER Y WITH DIAERESIS + 0x011E: 0xD0, # LATIN CAPITAL LETTER G WITH BREVE + 0x011F: 0xF0, # LATIN SMALL LETTER G WITH BREVE + 0x0130: 0xDD, # LATIN CAPITAL LETTER I WITH DOT ABOVE + 0x0131: 0xFD, # LATIN SMALL LETTER DOTLESS I + 0x0152: 0x8C, # LATIN CAPITAL LIGATURE OE + 0x0153: 0x9C, # LATIN SMALL LIGATURE OE + 0x015E: 0xDE, # LATIN CAPITAL LETTER S WITH CEDILLA + 0x015F: 0xFE, # LATIN SMALL LETTER S WITH CEDILLA + 0x0160: 0x8A, # LATIN CAPITAL LETTER S WITH CARON + 0x0161: 0x9A, # LATIN SMALL LETTER S WITH CARON + 0x0178: 0x9F, # LATIN CAPITAL LETTER Y WITH DIAERESIS 0x0192: 0x83, # LATIN SMALL LETTER F WITH HOOK - 0x02c6: 0x88, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02dc: 0x98, # SMALL TILDE + 0x02C6: 0x88, # MODIFIER LETTER CIRCUMFLEX ACCENT + 0x02DC: 0x98, # SMALL TILDE 0x2013: 0x96, # EN DASH 0x2014: 0x97, # EM DASH 0x2018: 0x91, # LEFT SINGLE QUOTATION MARK 0x2019: 0x92, # RIGHT SINGLE QUOTATION MARK - 0x201a: 0x82, # SINGLE LOW-9 QUOTATION MARK - 0x201c: 0x93, # LEFT DOUBLE QUOTATION MARK - 0x201d: 0x94, # RIGHT DOUBLE QUOTATION MARK - 0x201e: 0x84, # DOUBLE LOW-9 QUOTATION MARK + 0x201A: 0x82, # SINGLE LOW-9 QUOTATION MARK + 0x201C: 0x93, # LEFT DOUBLE QUOTATION MARK + 0x201D: 0x94, # RIGHT DOUBLE QUOTATION MARK + 0x201E: 0x84, # DOUBLE LOW-9 QUOTATION MARK 0x2020: 0x86, # DAGGER 0x2021: 0x87, # DOUBLE DAGGER 0x2022: 0x95, # BULLET 0x2026: 0x85, # HORIZONTAL ELLIPSIS 0x2030: 0x89, # PER MILLE SIGN - 0x2039: 0x8b, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK - 0x203a: 0x9b, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - 0x20ac: 0x80, # EURO SIGN + 0x2039: 0x8B, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK + 0x203A: 0x9B, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + 0x20AC: 0x80, # EURO SIGN 0x2122: 0x99, # TRADE MARK SIGN -}
\ No newline at end of file +} + |