diff options
| author | Marc-André Lemburg <mal@egenix.com> | 2005-10-24 12:07:49 (GMT) |
|---|---|---|
| committer | Marc-André Lemburg <mal@egenix.com> | 2005-10-24 12:07:49 (GMT) |
| commit | 3c72ded23dfbeb66072171a5707b27d9984ef332 (patch) | |
| tree | ee435de77abcbe9061c6ec3b3de79e3ab910675c /Lib/encodings/mac_cyrillic.py | |
| parent | 921fa8595ea3d67f657495783fc107a7a4d0ec86 (diff) | |
| download | cpython-3c72ded23dfbeb66072171a5707b27d9984ef332.zip cpython-3c72ded23dfbeb66072171a5707b27d9984ef332.tar.gz cpython-3c72ded23dfbeb66072171a5707b27d9984ef332.tar.bz2 | |
Removed the decoding_map from the codecs where this is possible.
Replaced the tis_620, cp1140 and koi8_u codecs with new ones
based on custom mapping files.
Diffstat (limited to 'Lib/encodings/mac_cyrillic.py')
| -rw-r--r-- | Lib/encodings/mac_cyrillic.py | 1155 |
1 files changed, 513 insertions, 642 deletions
diff --git a/Lib/encodings/mac_cyrillic.py b/Lib/encodings/mac_cyrillic.py index 7dc8e41..e3e5cf2 100644 --- a/Lib/encodings/mac_cyrillic.py +++ b/Lib/encodings/mac_cyrillic.py @@ -1,4 +1,4 @@ -""" Python Character Mapping Codec generated from 'VENDORS/APPLE/CYRILLIC.TXT' with gencodec.py. +""" Python Character Mapping Codec generated from 'MAPPINGS/VENDORS/APPLE/CYRILLIC.TXT' with gencodec.py. """#" @@ -28,654 +28,525 @@ def getregentry(): return (Codec().encode,Codec().decode,StreamReader,StreamWriter) -### Decoding Map - -decoding_map = codecs.make_identity_dict(range(256)) -decoding_map.update({ - 0x0080: 0x0410, # CYRILLIC CAPITAL LETTER A - 0x0081: 0x0411, # CYRILLIC CAPITAL LETTER BE - 0x0082: 0x0412, # CYRILLIC CAPITAL LETTER VE - 0x0083: 0x0413, # CYRILLIC CAPITAL LETTER GHE - 0x0084: 0x0414, # CYRILLIC CAPITAL LETTER DE - 0x0085: 0x0415, # CYRILLIC CAPITAL LETTER IE - 0x0086: 0x0416, # CYRILLIC CAPITAL LETTER ZHE - 0x0087: 0x0417, # CYRILLIC CAPITAL LETTER ZE - 0x0088: 0x0418, # CYRILLIC CAPITAL LETTER I - 0x0089: 0x0419, # CYRILLIC CAPITAL LETTER SHORT I - 0x008a: 0x041a, # CYRILLIC CAPITAL LETTER KA - 0x008b: 0x041b, # CYRILLIC CAPITAL LETTER EL - 0x008c: 0x041c, # CYRILLIC CAPITAL LETTER EM - 0x008d: 0x041d, # CYRILLIC CAPITAL LETTER EN - 0x008e: 0x041e, # CYRILLIC CAPITAL LETTER O - 0x008f: 0x041f, # CYRILLIC CAPITAL LETTER PE - 0x0090: 0x0420, # CYRILLIC CAPITAL LETTER ER - 0x0091: 0x0421, # CYRILLIC CAPITAL LETTER ES - 0x0092: 0x0422, # CYRILLIC CAPITAL LETTER TE - 0x0093: 0x0423, # CYRILLIC CAPITAL LETTER U - 0x0094: 0x0424, # CYRILLIC CAPITAL LETTER EF - 0x0095: 0x0425, # CYRILLIC CAPITAL LETTER HA - 0x0096: 0x0426, # CYRILLIC CAPITAL LETTER TSE - 0x0097: 0x0427, # CYRILLIC CAPITAL LETTER CHE - 0x0098: 0x0428, # CYRILLIC CAPITAL LETTER SHA - 0x0099: 0x0429, # CYRILLIC CAPITAL LETTER SHCHA - 0x009a: 0x042a, # CYRILLIC CAPITAL LETTER HARD SIGN - 0x009b: 0x042b, # CYRILLIC CAPITAL LETTER YERU - 0x009c: 0x042c, # CYRILLIC CAPITAL LETTER SOFT SIGN - 0x009d: 0x042d, # CYRILLIC CAPITAL LETTER E - 0x009e: 0x042e, # CYRILLIC CAPITAL LETTER YU - 0x009f: 0x042f, # CYRILLIC CAPITAL LETTER YA - 0x00a0: 0x2020, # DAGGER - 0x00a1: 0x00b0, # DEGREE SIGN - 0x00a2: 0x0490, # CYRILLIC CAPITAL LETTER GHE WITH UPTURN - 0x00a4: 0x00a7, # SECTION SIGN - 0x00a5: 0x2022, # BULLET - 0x00a6: 0x00b6, # PILCROW SIGN - 0x00a7: 0x0406, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - 0x00a8: 0x00ae, # REGISTERED SIGN - 0x00aa: 0x2122, # TRADE MARK SIGN - 0x00ab: 0x0402, # CYRILLIC CAPITAL LETTER DJE - 0x00ac: 0x0452, # CYRILLIC SMALL LETTER DJE - 0x00ad: 0x2260, # NOT EQUAL TO - 0x00ae: 0x0403, # CYRILLIC CAPITAL LETTER GJE - 0x00af: 0x0453, # CYRILLIC SMALL LETTER GJE - 0x00b0: 0x221e, # INFINITY - 0x00b2: 0x2264, # LESS-THAN OR EQUAL TO - 0x00b3: 0x2265, # GREATER-THAN OR EQUAL TO - 0x00b4: 0x0456, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - 0x00b6: 0x0491, # CYRILLIC SMALL LETTER GHE WITH UPTURN - 0x00b7: 0x0408, # CYRILLIC CAPITAL LETTER JE - 0x00b8: 0x0404, # CYRILLIC CAPITAL LETTER UKRAINIAN IE - 0x00b9: 0x0454, # CYRILLIC SMALL LETTER UKRAINIAN IE - 0x00ba: 0x0407, # CYRILLIC CAPITAL LETTER YI - 0x00bb: 0x0457, # CYRILLIC SMALL LETTER YI - 0x00bc: 0x0409, # CYRILLIC CAPITAL LETTER LJE - 0x00bd: 0x0459, # CYRILLIC SMALL LETTER LJE - 0x00be: 0x040a, # CYRILLIC CAPITAL LETTER NJE - 0x00bf: 0x045a, # CYRILLIC SMALL LETTER NJE - 0x00c0: 0x0458, # CYRILLIC SMALL LETTER JE - 0x00c1: 0x0405, # CYRILLIC CAPITAL LETTER DZE - 0x00c2: 0x00ac, # NOT SIGN - 0x00c3: 0x221a, # SQUARE ROOT - 0x00c4: 0x0192, # LATIN SMALL LETTER F WITH HOOK - 0x00c5: 0x2248, # ALMOST EQUAL TO - 0x00c6: 0x2206, # INCREMENT - 0x00c7: 0x00ab, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00c8: 0x00bb, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00c9: 0x2026, # HORIZONTAL ELLIPSIS - 0x00ca: 0x00a0, # NO-BREAK SPACE - 0x00cb: 0x040b, # CYRILLIC CAPITAL LETTER TSHE - 0x00cc: 0x045b, # CYRILLIC SMALL LETTER TSHE - 0x00cd: 0x040c, # CYRILLIC CAPITAL LETTER KJE - 0x00ce: 0x045c, # CYRILLIC SMALL LETTER KJE - 0x00cf: 0x0455, # CYRILLIC SMALL LETTER DZE - 0x00d0: 0x2013, # EN DASH - 0x00d1: 0x2014, # EM DASH - 0x00d2: 0x201c, # LEFT DOUBLE QUOTATION MARK - 0x00d3: 0x201d, # RIGHT DOUBLE QUOTATION MARK - 0x00d4: 0x2018, # LEFT SINGLE QUOTATION MARK - 0x00d5: 0x2019, # RIGHT SINGLE QUOTATION MARK - 0x00d6: 0x00f7, # DIVISION SIGN - 0x00d7: 0x201e, # DOUBLE LOW-9 QUOTATION MARK - 0x00d8: 0x040e, # CYRILLIC CAPITAL LETTER SHORT U - 0x00d9: 0x045e, # CYRILLIC SMALL LETTER SHORT U - 0x00da: 0x040f, # CYRILLIC CAPITAL LETTER DZHE - 0x00db: 0x045f, # CYRILLIC SMALL LETTER DZHE - 0x00dc: 0x2116, # NUMERO SIGN - 0x00dd: 0x0401, # CYRILLIC CAPITAL LETTER IO - 0x00de: 0x0451, # CYRILLIC SMALL LETTER IO - 0x00df: 0x044f, # CYRILLIC SMALL LETTER YA - 0x00e0: 0x0430, # CYRILLIC SMALL LETTER A - 0x00e1: 0x0431, # CYRILLIC SMALL LETTER BE - 0x00e2: 0x0432, # CYRILLIC SMALL LETTER VE - 0x00e3: 0x0433, # CYRILLIC SMALL LETTER GHE - 0x00e4: 0x0434, # CYRILLIC SMALL LETTER DE - 0x00e5: 0x0435, # CYRILLIC SMALL LETTER IE - 0x00e6: 0x0436, # CYRILLIC SMALL LETTER ZHE - 0x00e7: 0x0437, # CYRILLIC SMALL LETTER ZE - 0x00e8: 0x0438, # CYRILLIC SMALL LETTER I - 0x00e9: 0x0439, # CYRILLIC SMALL LETTER SHORT I - 0x00ea: 0x043a, # CYRILLIC SMALL LETTER KA - 0x00eb: 0x043b, # CYRILLIC SMALL LETTER EL - 0x00ec: 0x043c, # CYRILLIC SMALL LETTER EM - 0x00ed: 0x043d, # CYRILLIC SMALL LETTER EN - 0x00ee: 0x043e, # CYRILLIC SMALL LETTER O - 0x00ef: 0x043f, # CYRILLIC SMALL LETTER PE - 0x00f0: 0x0440, # CYRILLIC SMALL LETTER ER - 0x00f1: 0x0441, # CYRILLIC SMALL LETTER ES - 0x00f2: 0x0442, # CYRILLIC SMALL LETTER TE - 0x00f3: 0x0443, # CYRILLIC SMALL LETTER U - 0x00f4: 0x0444, # CYRILLIC SMALL LETTER EF - 0x00f5: 0x0445, # CYRILLIC SMALL LETTER HA - 0x00f6: 0x0446, # CYRILLIC SMALL LETTER TSE - 0x00f7: 0x0447, # CYRILLIC SMALL LETTER CHE - 0x00f8: 0x0448, # CYRILLIC SMALL LETTER SHA - 0x00f9: 0x0449, # CYRILLIC SMALL LETTER SHCHA - 0x00fa: 0x044a, # CYRILLIC SMALL LETTER HARD SIGN - 0x00fb: 0x044b, # CYRILLIC SMALL LETTER YERU - 0x00fc: 0x044c, # CYRILLIC SMALL LETTER SOFT SIGN - 0x00fd: 0x044d, # CYRILLIC SMALL LETTER E - 0x00fe: 0x044e, # CYRILLIC SMALL LETTER YU - 0x00ff: 0x20ac, # EURO SIGN -}) ### 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 - 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 -> CONTROL CHARACTER - 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'\u2020' # 0x00a0 -> DAGGER - u'\xb0' # 0x00a1 -> DEGREE SIGN - u'\u0490' # 0x00a2 -> CYRILLIC CAPITAL LETTER GHE WITH UPTURN - u'\xa3' # 0x00a3 -> POUND SIGN - u'\xa7' # 0x00a4 -> SECTION SIGN - u'\u2022' # 0x00a5 -> BULLET - u'\xb6' # 0x00a6 -> PILCROW SIGN - u'\u0406' # 0x00a7 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - u'\xae' # 0x00a8 -> REGISTERED SIGN - u'\xa9' # 0x00a9 -> COPYRIGHT SIGN - u'\u2122' # 0x00aa -> TRADE MARK SIGN - u'\u0402' # 0x00ab -> CYRILLIC CAPITAL LETTER DJE - u'\u0452' # 0x00ac -> CYRILLIC SMALL LETTER DJE - u'\u2260' # 0x00ad -> NOT EQUAL TO - u'\u0403' # 0x00ae -> CYRILLIC CAPITAL LETTER GJE - u'\u0453' # 0x00af -> CYRILLIC SMALL LETTER GJE - u'\u221e' # 0x00b0 -> INFINITY - u'\xb1' # 0x00b1 -> PLUS-MINUS SIGN - u'\u2264' # 0x00b2 -> LESS-THAN OR EQUAL TO - u'\u2265' # 0x00b3 -> GREATER-THAN OR EQUAL TO - u'\u0456' # 0x00b4 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - u'\xb5' # 0x00b5 -> MICRO SIGN - u'\u0491' # 0x00b6 -> CYRILLIC SMALL LETTER GHE WITH UPTURN - u'\u0408' # 0x00b7 -> CYRILLIC CAPITAL LETTER JE - u'\u0404' # 0x00b8 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE - u'\u0454' # 0x00b9 -> CYRILLIC SMALL LETTER UKRAINIAN IE - u'\u0407' # 0x00ba -> CYRILLIC CAPITAL LETTER YI - u'\u0457' # 0x00bb -> CYRILLIC SMALL LETTER YI - u'\u0409' # 0x00bc -> CYRILLIC CAPITAL LETTER LJE - u'\u0459' # 0x00bd -> CYRILLIC SMALL LETTER LJE - u'\u040a' # 0x00be -> CYRILLIC CAPITAL LETTER NJE - u'\u045a' # 0x00bf -> CYRILLIC SMALL LETTER NJE - u'\u0458' # 0x00c0 -> CYRILLIC SMALL LETTER JE - u'\u0405' # 0x00c1 -> CYRILLIC CAPITAL LETTER DZE - u'\xac' # 0x00c2 -> NOT SIGN - u'\u221a' # 0x00c3 -> SQUARE ROOT - u'\u0192' # 0x00c4 -> LATIN SMALL LETTER F WITH HOOK - u'\u2248' # 0x00c5 -> ALMOST EQUAL TO - u'\u2206' # 0x00c6 -> INCREMENT - u'\xab' # 0x00c7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\xbb' # 0x00c8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - u'\u2026' # 0x00c9 -> HORIZONTAL ELLIPSIS - u'\xa0' # 0x00ca -> NO-BREAK SPACE - u'\u040b' # 0x00cb -> CYRILLIC CAPITAL LETTER TSHE - u'\u045b' # 0x00cc -> CYRILLIC SMALL LETTER TSHE - u'\u040c' # 0x00cd -> CYRILLIC CAPITAL LETTER KJE - u'\u045c' # 0x00ce -> CYRILLIC SMALL LETTER KJE - u'\u0455' # 0x00cf -> CYRILLIC SMALL LETTER DZE - u'\u2013' # 0x00d0 -> EN DASH - u'\u2014' # 0x00d1 -> EM DASH - u'\u201c' # 0x00d2 -> LEFT DOUBLE QUOTATION MARK - u'\u201d' # 0x00d3 -> RIGHT DOUBLE QUOTATION MARK - u'\u2018' # 0x00d4 -> LEFT SINGLE QUOTATION MARK - u'\u2019' # 0x00d5 -> RIGHT SINGLE QUOTATION MARK - u'\xf7' # 0x00d6 -> DIVISION SIGN - u'\u201e' # 0x00d7 -> DOUBLE LOW-9 QUOTATION MARK - u'\u040e' # 0x00d8 -> CYRILLIC CAPITAL LETTER SHORT U - u'\u045e' # 0x00d9 -> CYRILLIC SMALL LETTER SHORT U - u'\u040f' # 0x00da -> CYRILLIC CAPITAL LETTER DZHE - u'\u045f' # 0x00db -> CYRILLIC SMALL LETTER DZHE - u'\u2116' # 0x00dc -> NUMERO SIGN - u'\u0401' # 0x00dd -> CYRILLIC CAPITAL LETTER IO - u'\u0451' # 0x00de -> CYRILLIC SMALL LETTER IO - u'\u044f' # 0x00df -> CYRILLIC SMALL LETTER YA - u'\u0430' # 0x00e0 -> CYRILLIC SMALL LETTER A - u'\u0431' # 0x00e1 -> CYRILLIC SMALL LETTER BE - u'\u0432' # 0x00e2 -> CYRILLIC SMALL LETTER VE - u'\u0433' # 0x00e3 -> CYRILLIC SMALL LETTER GHE - u'\u0434' # 0x00e4 -> CYRILLIC SMALL LETTER DE - u'\u0435' # 0x00e5 -> CYRILLIC SMALL LETTER IE - u'\u0436' # 0x00e6 -> CYRILLIC SMALL LETTER ZHE - u'\u0437' # 0x00e7 -> CYRILLIC SMALL LETTER ZE - u'\u0438' # 0x00e8 -> CYRILLIC SMALL LETTER I - u'\u0439' # 0x00e9 -> CYRILLIC SMALL LETTER SHORT I - u'\u043a' # 0x00ea -> CYRILLIC SMALL LETTER KA - u'\u043b' # 0x00eb -> CYRILLIC SMALL LETTER EL - u'\u043c' # 0x00ec -> CYRILLIC SMALL LETTER EM - u'\u043d' # 0x00ed -> CYRILLIC SMALL LETTER EN - u'\u043e' # 0x00ee -> CYRILLIC SMALL LETTER O - u'\u043f' # 0x00ef -> CYRILLIC SMALL LETTER PE - u'\u0440' # 0x00f0 -> CYRILLIC SMALL LETTER ER - u'\u0441' # 0x00f1 -> CYRILLIC SMALL LETTER ES - u'\u0442' # 0x00f2 -> CYRILLIC SMALL LETTER TE - u'\u0443' # 0x00f3 -> CYRILLIC SMALL LETTER U - u'\u0444' # 0x00f4 -> CYRILLIC SMALL LETTER EF - u'\u0445' # 0x00f5 -> CYRILLIC SMALL LETTER HA - u'\u0446' # 0x00f6 -> CYRILLIC SMALL LETTER TSE - u'\u0447' # 0x00f7 -> CYRILLIC SMALL LETTER CHE - u'\u0448' # 0x00f8 -> CYRILLIC SMALL LETTER SHA - u'\u0449' # 0x00f9 -> CYRILLIC SMALL LETTER SHCHA - u'\u044a' # 0x00fa -> CYRILLIC SMALL LETTER HARD SIGN - u'\u044b' # 0x00fb -> CYRILLIC SMALL LETTER YERU - u'\u044c' # 0x00fc -> CYRILLIC SMALL LETTER SOFT SIGN - u'\u044d' # 0x00fd -> CYRILLIC SMALL LETTER E - u'\u044e' # 0x00fe -> CYRILLIC SMALL LETTER YU - u'\u20ac' # 0x00ff -> EURO SIGN + 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 ) ### Encoding Map encoding_map = { - 0x0000: 0x0000, # CONTROL CHARACTER - 0x0001: 0x0001, # CONTROL CHARACTER - 0x0002: 0x0002, # CONTROL CHARACTER - 0x0003: 0x0003, # CONTROL CHARACTER - 0x0004: 0x0004, # CONTROL CHARACTER - 0x0005: 0x0005, # CONTROL CHARACTER - 0x0006: 0x0006, # CONTROL CHARACTER - 0x0007: 0x0007, # CONTROL CHARACTER - 0x0008: 0x0008, # CONTROL CHARACTER - 0x0009: 0x0009, # CONTROL CHARACTER - 0x000a: 0x000a, # CONTROL CHARACTER - 0x000b: 0x000b, # CONTROL CHARACTER - 0x000c: 0x000c, # CONTROL CHARACTER - 0x000d: 0x000d, # CONTROL CHARACTER - 0x000e: 0x000e, # CONTROL CHARACTER - 0x000f: 0x000f, # CONTROL CHARACTER - 0x0010: 0x0010, # CONTROL CHARACTER - 0x0011: 0x0011, # CONTROL CHARACTER - 0x0012: 0x0012, # CONTROL CHARACTER - 0x0013: 0x0013, # CONTROL CHARACTER - 0x0014: 0x0014, # CONTROL CHARACTER - 0x0015: 0x0015, # CONTROL CHARACTER - 0x0016: 0x0016, # CONTROL CHARACTER - 0x0017: 0x0017, # CONTROL CHARACTER - 0x0018: 0x0018, # CONTROL CHARACTER - 0x0019: 0x0019, # CONTROL CHARACTER - 0x001a: 0x001a, # CONTROL CHARACTER - 0x001b: 0x001b, # CONTROL CHARACTER - 0x001c: 0x001c, # CONTROL CHARACTER - 0x001d: 0x001d, # CONTROL CHARACTER - 0x001e: 0x001e, # CONTROL CHARACTER - 0x001f: 0x001f, # CONTROL CHARACTER - 0x0020: 0x0020, # SPACE - 0x0021: 0x0021, # EXCLAMATION MARK - 0x0022: 0x0022, # QUOTATION MARK - 0x0023: 0x0023, # NUMBER SIGN - 0x0024: 0x0024, # DOLLAR SIGN - 0x0025: 0x0025, # PERCENT SIGN - 0x0026: 0x0026, # AMPERSAND - 0x0027: 0x0027, # APOSTROPHE - 0x0028: 0x0028, # LEFT PARENTHESIS - 0x0029: 0x0029, # RIGHT PARENTHESIS - 0x002a: 0x002a, # ASTERISK - 0x002b: 0x002b, # PLUS SIGN - 0x002c: 0x002c, # COMMA - 0x002d: 0x002d, # HYPHEN-MINUS - 0x002e: 0x002e, # FULL STOP - 0x002f: 0x002f, # SOLIDUS - 0x0030: 0x0030, # DIGIT ZERO - 0x0031: 0x0031, # DIGIT ONE - 0x0032: 0x0032, # DIGIT TWO - 0x0033: 0x0033, # DIGIT THREE - 0x0034: 0x0034, # DIGIT FOUR - 0x0035: 0x0035, # DIGIT FIVE - 0x0036: 0x0036, # DIGIT SIX - 0x0037: 0x0037, # DIGIT SEVEN - 0x0038: 0x0038, # DIGIT EIGHT - 0x0039: 0x0039, # DIGIT NINE - 0x003a: 0x003a, # COLON - 0x003b: 0x003b, # SEMICOLON - 0x003c: 0x003c, # LESS-THAN SIGN - 0x003d: 0x003d, # EQUALS SIGN - 0x003e: 0x003e, # GREATER-THAN SIGN - 0x003f: 0x003f, # QUESTION MARK - 0x0040: 0x0040, # COMMERCIAL AT - 0x0041: 0x0041, # LATIN CAPITAL LETTER A - 0x0042: 0x0042, # LATIN CAPITAL LETTER B - 0x0043: 0x0043, # LATIN CAPITAL LETTER C - 0x0044: 0x0044, # LATIN CAPITAL LETTER D - 0x0045: 0x0045, # LATIN CAPITAL LETTER E - 0x0046: 0x0046, # LATIN CAPITAL LETTER F - 0x0047: 0x0047, # LATIN CAPITAL LETTER G - 0x0048: 0x0048, # LATIN CAPITAL LETTER H - 0x0049: 0x0049, # LATIN CAPITAL LETTER I - 0x004a: 0x004a, # LATIN CAPITAL LETTER J - 0x004b: 0x004b, # LATIN CAPITAL LETTER K - 0x004c: 0x004c, # LATIN CAPITAL LETTER L - 0x004d: 0x004d, # LATIN CAPITAL LETTER M - 0x004e: 0x004e, # LATIN CAPITAL LETTER N - 0x004f: 0x004f, # LATIN CAPITAL LETTER O - 0x0050: 0x0050, # LATIN CAPITAL LETTER P - 0x0051: 0x0051, # LATIN CAPITAL LETTER Q - 0x0052: 0x0052, # LATIN CAPITAL LETTER R - 0x0053: 0x0053, # LATIN CAPITAL LETTER S - 0x0054: 0x0054, # LATIN CAPITAL LETTER T - 0x0055: 0x0055, # LATIN CAPITAL LETTER U - 0x0056: 0x0056, # LATIN CAPITAL LETTER V - 0x0057: 0x0057, # LATIN CAPITAL LETTER W - 0x0058: 0x0058, # LATIN CAPITAL LETTER X - 0x0059: 0x0059, # LATIN CAPITAL LETTER Y - 0x005a: 0x005a, # LATIN CAPITAL LETTER Z - 0x005b: 0x005b, # LEFT SQUARE BRACKET - 0x005c: 0x005c, # REVERSE SOLIDUS - 0x005d: 0x005d, # RIGHT SQUARE BRACKET - 0x005e: 0x005e, # CIRCUMFLEX ACCENT - 0x005f: 0x005f, # LOW LINE - 0x0060: 0x0060, # GRAVE ACCENT - 0x0061: 0x0061, # LATIN SMALL LETTER A - 0x0062: 0x0062, # LATIN SMALL LETTER B - 0x0063: 0x0063, # LATIN SMALL LETTER C - 0x0064: 0x0064, # LATIN SMALL LETTER D - 0x0065: 0x0065, # LATIN SMALL LETTER E - 0x0066: 0x0066, # LATIN SMALL LETTER F - 0x0067: 0x0067, # LATIN SMALL LETTER G - 0x0068: 0x0068, # LATIN SMALL LETTER H - 0x0069: 0x0069, # LATIN SMALL LETTER I - 0x006a: 0x006a, # LATIN SMALL LETTER J - 0x006b: 0x006b, # LATIN SMALL LETTER K - 0x006c: 0x006c, # LATIN SMALL LETTER L - 0x006d: 0x006d, # LATIN SMALL LETTER M - 0x006e: 0x006e, # LATIN SMALL LETTER N - 0x006f: 0x006f, # LATIN SMALL LETTER O - 0x0070: 0x0070, # LATIN SMALL LETTER P - 0x0071: 0x0071, # LATIN SMALL LETTER Q - 0x0072: 0x0072, # LATIN SMALL LETTER R - 0x0073: 0x0073, # LATIN SMALL LETTER S - 0x0074: 0x0074, # LATIN SMALL LETTER T - 0x0075: 0x0075, # LATIN SMALL LETTER U - 0x0076: 0x0076, # LATIN SMALL LETTER V - 0x0077: 0x0077, # LATIN SMALL LETTER W - 0x0078: 0x0078, # LATIN SMALL LETTER X - 0x0079: 0x0079, # LATIN SMALL LETTER Y - 0x007a: 0x007a, # LATIN SMALL LETTER Z - 0x007b: 0x007b, # LEFT CURLY BRACKET - 0x007c: 0x007c, # VERTICAL LINE - 0x007d: 0x007d, # RIGHT CURLY BRACKET - 0x007e: 0x007e, # TILDE - 0x007f: 0x007f, # CONTROL CHARACTER - 0x00a0: 0x00ca, # NO-BREAK SPACE - 0x00a3: 0x00a3, # POUND SIGN - 0x00a7: 0x00a4, # SECTION SIGN - 0x00a9: 0x00a9, # COPYRIGHT SIGN - 0x00ab: 0x00c7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00ac: 0x00c2, # NOT SIGN - 0x00ae: 0x00a8, # REGISTERED SIGN - 0x00b0: 0x00a1, # DEGREE SIGN - 0x00b1: 0x00b1, # PLUS-MINUS SIGN - 0x00b5: 0x00b5, # MICRO SIGN - 0x00b6: 0x00a6, # PILCROW SIGN - 0x00bb: 0x00c8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00f7: 0x00d6, # DIVISION SIGN - 0x0192: 0x00c4, # LATIN SMALL LETTER F WITH HOOK - 0x0401: 0x00dd, # CYRILLIC CAPITAL LETTER IO - 0x0402: 0x00ab, # CYRILLIC CAPITAL LETTER DJE - 0x0403: 0x00ae, # CYRILLIC CAPITAL LETTER GJE - 0x0404: 0x00b8, # CYRILLIC CAPITAL LETTER UKRAINIAN IE - 0x0405: 0x00c1, # CYRILLIC CAPITAL LETTER DZE - 0x0406: 0x00a7, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0407: 0x00ba, # CYRILLIC CAPITAL LETTER YI - 0x0408: 0x00b7, # CYRILLIC CAPITAL LETTER JE - 0x0409: 0x00bc, # CYRILLIC CAPITAL LETTER LJE - 0x040a: 0x00be, # CYRILLIC CAPITAL LETTER NJE - 0x040b: 0x00cb, # CYRILLIC CAPITAL LETTER TSHE - 0x040c: 0x00cd, # CYRILLIC CAPITAL LETTER KJE - 0x040e: 0x00d8, # CYRILLIC CAPITAL LETTER SHORT U - 0x040f: 0x00da, # CYRILLIC CAPITAL LETTER DZHE - 0x0410: 0x0080, # CYRILLIC CAPITAL LETTER A - 0x0411: 0x0081, # CYRILLIC CAPITAL LETTER BE - 0x0412: 0x0082, # CYRILLIC CAPITAL LETTER VE - 0x0413: 0x0083, # CYRILLIC CAPITAL LETTER GHE - 0x0414: 0x0084, # CYRILLIC CAPITAL LETTER DE - 0x0415: 0x0085, # CYRILLIC CAPITAL LETTER IE - 0x0416: 0x0086, # CYRILLIC CAPITAL LETTER ZHE - 0x0417: 0x0087, # CYRILLIC CAPITAL LETTER ZE - 0x0418: 0x0088, # CYRILLIC CAPITAL LETTER I - 0x0419: 0x0089, # CYRILLIC CAPITAL LETTER SHORT I - 0x041a: 0x008a, # CYRILLIC CAPITAL LETTER KA - 0x041b: 0x008b, # CYRILLIC CAPITAL LETTER EL - 0x041c: 0x008c, # CYRILLIC CAPITAL LETTER EM - 0x041d: 0x008d, # CYRILLIC CAPITAL LETTER EN - 0x041e: 0x008e, # CYRILLIC CAPITAL LETTER O - 0x041f: 0x008f, # CYRILLIC CAPITAL LETTER PE - 0x0420: 0x0090, # CYRILLIC CAPITAL LETTER ER - 0x0421: 0x0091, # CYRILLIC CAPITAL LETTER ES - 0x0422: 0x0092, # CYRILLIC CAPITAL LETTER TE - 0x0423: 0x0093, # CYRILLIC CAPITAL LETTER U - 0x0424: 0x0094, # CYRILLIC CAPITAL LETTER EF - 0x0425: 0x0095, # CYRILLIC CAPITAL LETTER HA - 0x0426: 0x0096, # CYRILLIC CAPITAL LETTER TSE - 0x0427: 0x0097, # CYRILLIC CAPITAL LETTER CHE - 0x0428: 0x0098, # CYRILLIC CAPITAL LETTER SHA - 0x0429: 0x0099, # CYRILLIC CAPITAL LETTER SHCHA - 0x042a: 0x009a, # CYRILLIC CAPITAL LETTER HARD SIGN - 0x042b: 0x009b, # CYRILLIC CAPITAL LETTER YERU - 0x042c: 0x009c, # CYRILLIC CAPITAL LETTER SOFT SIGN - 0x042d: 0x009d, # CYRILLIC CAPITAL LETTER E - 0x042e: 0x009e, # CYRILLIC CAPITAL LETTER YU - 0x042f: 0x009f, # CYRILLIC CAPITAL LETTER YA - 0x0430: 0x00e0, # CYRILLIC SMALL LETTER A - 0x0431: 0x00e1, # CYRILLIC SMALL LETTER BE - 0x0432: 0x00e2, # CYRILLIC SMALL LETTER VE - 0x0433: 0x00e3, # CYRILLIC SMALL LETTER GHE - 0x0434: 0x00e4, # CYRILLIC SMALL LETTER DE - 0x0435: 0x00e5, # CYRILLIC SMALL LETTER IE - 0x0436: 0x00e6, # CYRILLIC SMALL LETTER ZHE - 0x0437: 0x00e7, # CYRILLIC SMALL LETTER ZE - 0x0438: 0x00e8, # CYRILLIC SMALL LETTER I - 0x0439: 0x00e9, # CYRILLIC SMALL LETTER SHORT I - 0x043a: 0x00ea, # CYRILLIC SMALL LETTER KA - 0x043b: 0x00eb, # CYRILLIC SMALL LETTER EL - 0x043c: 0x00ec, # CYRILLIC SMALL LETTER EM - 0x043d: 0x00ed, # CYRILLIC SMALL LETTER EN - 0x043e: 0x00ee, # CYRILLIC SMALL LETTER O - 0x043f: 0x00ef, # CYRILLIC SMALL LETTER PE - 0x0440: 0x00f0, # CYRILLIC SMALL LETTER ER - 0x0441: 0x00f1, # CYRILLIC SMALL LETTER ES - 0x0442: 0x00f2, # CYRILLIC SMALL LETTER TE - 0x0443: 0x00f3, # CYRILLIC SMALL LETTER U - 0x0444: 0x00f4, # CYRILLIC SMALL LETTER EF - 0x0445: 0x00f5, # CYRILLIC SMALL LETTER HA - 0x0446: 0x00f6, # CYRILLIC SMALL LETTER TSE - 0x0447: 0x00f7, # CYRILLIC SMALL LETTER CHE - 0x0448: 0x00f8, # CYRILLIC SMALL LETTER SHA - 0x0449: 0x00f9, # CYRILLIC SMALL LETTER SHCHA - 0x044a: 0x00fa, # CYRILLIC SMALL LETTER HARD SIGN - 0x044b: 0x00fb, # CYRILLIC SMALL LETTER YERU - 0x044c: 0x00fc, # CYRILLIC SMALL LETTER SOFT SIGN - 0x044d: 0x00fd, # CYRILLIC SMALL LETTER E - 0x044e: 0x00fe, # CYRILLIC SMALL LETTER YU - 0x044f: 0x00df, # CYRILLIC SMALL LETTER YA - 0x0451: 0x00de, # CYRILLIC SMALL LETTER IO - 0x0452: 0x00ac, # CYRILLIC SMALL LETTER DJE - 0x0453: 0x00af, # CYRILLIC SMALL LETTER GJE - 0x0454: 0x00b9, # CYRILLIC SMALL LETTER UKRAINIAN IE - 0x0455: 0x00cf, # CYRILLIC SMALL LETTER DZE - 0x0456: 0x00b4, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0457: 0x00bb, # CYRILLIC SMALL LETTER YI - 0x0458: 0x00c0, # CYRILLIC SMALL LETTER JE - 0x0459: 0x00bd, # CYRILLIC SMALL LETTER LJE - 0x045a: 0x00bf, # CYRILLIC SMALL LETTER NJE - 0x045b: 0x00cc, # CYRILLIC SMALL LETTER TSHE - 0x045c: 0x00ce, # CYRILLIC SMALL LETTER KJE - 0x045e: 0x00d9, # CYRILLIC SMALL LETTER SHORT U - 0x045f: 0x00db, # CYRILLIC SMALL LETTER DZHE - 0x0490: 0x00a2, # CYRILLIC CAPITAL LETTER GHE WITH UPTURN - 0x0491: 0x00b6, # CYRILLIC SMALL LETTER GHE WITH UPTURN - 0x2013: 0x00d0, # EN DASH - 0x2014: 0x00d1, # EM DASH - 0x2018: 0x00d4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0x00d5, # RIGHT SINGLE QUOTATION MARK - 0x201c: 0x00d2, # LEFT DOUBLE QUOTATION MARK - 0x201d: 0x00d3, # RIGHT DOUBLE QUOTATION MARK - 0x201e: 0x00d7, # DOUBLE LOW-9 QUOTATION MARK - 0x2020: 0x00a0, # DAGGER - 0x2022: 0x00a5, # BULLET - 0x2026: 0x00c9, # HORIZONTAL ELLIPSIS - 0x20ac: 0x00ff, # EURO SIGN - 0x2116: 0x00dc, # NUMERO SIGN - 0x2122: 0x00aa, # TRADE MARK SIGN - 0x2206: 0x00c6, # INCREMENT - 0x221a: 0x00c3, # SQUARE ROOT - 0x221e: 0x00b0, # INFINITY - 0x2248: 0x00c5, # ALMOST EQUAL TO - 0x2260: 0x00ad, # NOT EQUAL TO - 0x2264: 0x00b2, # LESS-THAN OR EQUAL TO - 0x2265: 0x00b3, # GREATER-THAN OR EQUAL TO + 0x0000: 0x00, # CONTROL CHARACTER + 0x0001: 0x01, # CONTROL CHARACTER + 0x0002: 0x02, # CONTROL CHARACTER + 0x0003: 0x03, # CONTROL CHARACTER + 0x0004: 0x04, # CONTROL CHARACTER + 0x0005: 0x05, # CONTROL CHARACTER + 0x0006: 0x06, # CONTROL CHARACTER + 0x0007: 0x07, # CONTROL CHARACTER + 0x0008: 0x08, # CONTROL CHARACTER + 0x0009: 0x09, # CONTROL CHARACTER + 0x000a: 0x0a, # CONTROL CHARACTER + 0x000b: 0x0b, # CONTROL CHARACTER + 0x000c: 0x0c, # CONTROL CHARACTER + 0x000d: 0x0d, # CONTROL CHARACTER + 0x000e: 0x0e, # CONTROL CHARACTER + 0x000f: 0x0f, # CONTROL CHARACTER + 0x0010: 0x10, # CONTROL CHARACTER + 0x0011: 0x11, # CONTROL CHARACTER + 0x0012: 0x12, # CONTROL CHARACTER + 0x0013: 0x13, # CONTROL CHARACTER + 0x0014: 0x14, # CONTROL CHARACTER + 0x0015: 0x15, # CONTROL CHARACTER + 0x0016: 0x16, # CONTROL CHARACTER + 0x0017: 0x17, # CONTROL CHARACTER + 0x0018: 0x18, # CONTROL CHARACTER + 0x0019: 0x19, # CONTROL CHARACTER + 0x001a: 0x1a, # CONTROL CHARACTER + 0x001b: 0x1b, # CONTROL CHARACTER + 0x001c: 0x1c, # CONTROL CHARACTER + 0x001d: 0x1d, # CONTROL CHARACTER + 0x001e: 0x1e, # CONTROL CHARACTER + 0x001f: 0x1f, # CONTROL CHARACTER + 0x0020: 0x20, # SPACE + 0x0021: 0x21, # EXCLAMATION MARK + 0x0022: 0x22, # QUOTATION MARK + 0x0023: 0x23, # NUMBER SIGN + 0x0024: 0x24, # DOLLAR SIGN + 0x0025: 0x25, # PERCENT SIGN + 0x0026: 0x26, # AMPERSAND + 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 + 0x0030: 0x30, # DIGIT ZERO + 0x0031: 0x31, # DIGIT ONE + 0x0032: 0x32, # DIGIT TWO + 0x0033: 0x33, # DIGIT THREE + 0x0034: 0x34, # DIGIT FOUR + 0x0035: 0x35, # DIGIT FIVE + 0x0036: 0x36, # DIGIT SIX + 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 + 0x0040: 0x40, # COMMERCIAL AT + 0x0041: 0x41, # LATIN CAPITAL LETTER A + 0x0042: 0x42, # LATIN CAPITAL LETTER B + 0x0043: 0x43, # LATIN CAPITAL LETTER C + 0x0044: 0x44, # LATIN CAPITAL LETTER D + 0x0045: 0x45, # LATIN CAPITAL LETTER E + 0x0046: 0x46, # LATIN CAPITAL LETTER F + 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 + 0x0050: 0x50, # LATIN CAPITAL LETTER P + 0x0051: 0x51, # LATIN CAPITAL LETTER Q + 0x0052: 0x52, # LATIN CAPITAL LETTER R + 0x0053: 0x53, # LATIN CAPITAL LETTER S + 0x0054: 0x54, # LATIN CAPITAL LETTER T + 0x0055: 0x55, # LATIN CAPITAL LETTER U + 0x0056: 0x56, # LATIN CAPITAL LETTER V + 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 + 0x0060: 0x60, # GRAVE ACCENT + 0x0061: 0x61, # LATIN SMALL LETTER A + 0x0062: 0x62, # LATIN SMALL LETTER B + 0x0063: 0x63, # LATIN SMALL LETTER C + 0x0064: 0x64, # LATIN SMALL LETTER D + 0x0065: 0x65, # LATIN SMALL LETTER E + 0x0066: 0x66, # LATIN SMALL LETTER F + 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 + 0x0070: 0x70, # LATIN SMALL LETTER P + 0x0071: 0x71, # LATIN SMALL LETTER Q + 0x0072: 0x72, # LATIN SMALL LETTER R + 0x0073: 0x73, # LATIN SMALL LETTER S + 0x0074: 0x74, # LATIN SMALL LETTER T + 0x0075: 0x75, # LATIN SMALL LETTER U + 0x0076: 0x76, # LATIN SMALL LETTER V + 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, # CONTROL CHARACTER + 0x00a0: 0xca, # NO-BREAK SPACE + 0x00a3: 0xa3, # POUND SIGN + 0x00a7: 0xa4, # SECTION SIGN + 0x00a9: 0xa9, # COPYRIGHT SIGN + 0x00ab: 0xc7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + 0x00ac: 0xc2, # NOT SIGN + 0x00ae: 0xa8, # REGISTERED SIGN + 0x00b0: 0xa1, # DEGREE SIGN + 0x00b1: 0xb1, # PLUS-MINUS SIGN + 0x00b5: 0xb5, # MICRO SIGN + 0x00b6: 0xa6, # PILCROW SIGN + 0x00bb: 0xc8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + 0x00f7: 0xd6, # DIVISION SIGN + 0x0192: 0xc4, # LATIN SMALL LETTER F WITH HOOK + 0x0401: 0xdd, # CYRILLIC CAPITAL LETTER IO + 0x0402: 0xab, # CYRILLIC CAPITAL LETTER DJE + 0x0403: 0xae, # CYRILLIC CAPITAL LETTER GJE + 0x0404: 0xb8, # CYRILLIC CAPITAL LETTER UKRAINIAN IE + 0x0405: 0xc1, # CYRILLIC CAPITAL LETTER DZE + 0x0406: 0xa7, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + 0x0407: 0xba, # CYRILLIC CAPITAL LETTER YI + 0x0408: 0xb7, # CYRILLIC CAPITAL LETTER JE + 0x0409: 0xbc, # CYRILLIC CAPITAL LETTER LJE + 0x040a: 0xbe, # CYRILLIC CAPITAL LETTER NJE + 0x040b: 0xcb, # CYRILLIC CAPITAL LETTER TSHE + 0x040c: 0xcd, # CYRILLIC CAPITAL LETTER KJE + 0x040e: 0xd8, # CYRILLIC CAPITAL LETTER SHORT U + 0x040f: 0xda, # CYRILLIC CAPITAL LETTER DZHE + 0x0410: 0x80, # CYRILLIC CAPITAL LETTER A + 0x0411: 0x81, # CYRILLIC CAPITAL LETTER BE + 0x0412: 0x82, # CYRILLIC CAPITAL LETTER VE + 0x0413: 0x83, # CYRILLIC CAPITAL LETTER GHE + 0x0414: 0x84, # CYRILLIC CAPITAL LETTER DE + 0x0415: 0x85, # CYRILLIC CAPITAL LETTER IE + 0x0416: 0x86, # CYRILLIC CAPITAL LETTER ZHE + 0x0417: 0x87, # CYRILLIC CAPITAL LETTER ZE + 0x0418: 0x88, # CYRILLIC CAPITAL LETTER I + 0x0419: 0x89, # CYRILLIC CAPITAL LETTER SHORT I + 0x041a: 0x8a, # CYRILLIC CAPITAL LETTER KA + 0x041b: 0x8b, # CYRILLIC CAPITAL LETTER EL + 0x041c: 0x8c, # CYRILLIC CAPITAL LETTER EM + 0x041d: 0x8d, # CYRILLIC CAPITAL LETTER EN + 0x041e: 0x8e, # CYRILLIC CAPITAL LETTER O + 0x041f: 0x8f, # CYRILLIC CAPITAL LETTER PE + 0x0420: 0x90, # CYRILLIC CAPITAL LETTER ER + 0x0421: 0x91, # CYRILLIC CAPITAL LETTER ES + 0x0422: 0x92, # CYRILLIC CAPITAL LETTER TE + 0x0423: 0x93, # CYRILLIC CAPITAL LETTER U + 0x0424: 0x94, # CYRILLIC CAPITAL LETTER EF + 0x0425: 0x95, # CYRILLIC CAPITAL LETTER HA + 0x0426: 0x96, # CYRILLIC CAPITAL LETTER TSE + 0x0427: 0x97, # CYRILLIC CAPITAL LETTER CHE + 0x0428: 0x98, # CYRILLIC CAPITAL LETTER SHA + 0x0429: 0x99, # CYRILLIC CAPITAL LETTER SHCHA + 0x042a: 0x9a, # CYRILLIC CAPITAL LETTER HARD SIGN + 0x042b: 0x9b, # CYRILLIC CAPITAL LETTER YERU + 0x042c: 0x9c, # CYRILLIC CAPITAL LETTER SOFT SIGN + 0x042d: 0x9d, # CYRILLIC CAPITAL LETTER E + 0x042e: 0x9e, # CYRILLIC CAPITAL LETTER YU + 0x042f: 0x9f, # CYRILLIC CAPITAL LETTER YA + 0x0430: 0xe0, # CYRILLIC SMALL LETTER A + 0x0431: 0xe1, # CYRILLIC SMALL LETTER BE + 0x0432: 0xe2, # CYRILLIC SMALL LETTER VE + 0x0433: 0xe3, # CYRILLIC SMALL LETTER GHE + 0x0434: 0xe4, # CYRILLIC SMALL LETTER DE + 0x0435: 0xe5, # CYRILLIC SMALL LETTER IE + 0x0436: 0xe6, # CYRILLIC SMALL LETTER ZHE + 0x0437: 0xe7, # CYRILLIC SMALL LETTER ZE + 0x0438: 0xe8, # CYRILLIC SMALL LETTER I + 0x0439: 0xe9, # CYRILLIC SMALL LETTER SHORT I + 0x043a: 0xea, # CYRILLIC SMALL LETTER KA + 0x043b: 0xeb, # CYRILLIC SMALL LETTER EL + 0x043c: 0xec, # CYRILLIC SMALL LETTER EM + 0x043d: 0xed, # CYRILLIC SMALL LETTER EN + 0x043e: 0xee, # CYRILLIC SMALL LETTER O + 0x043f: 0xef, # CYRILLIC SMALL LETTER PE + 0x0440: 0xf0, # CYRILLIC SMALL LETTER ER + 0x0441: 0xf1, # CYRILLIC SMALL LETTER ES + 0x0442: 0xf2, # CYRILLIC SMALL LETTER TE + 0x0443: 0xf3, # CYRILLIC SMALL LETTER U + 0x0444: 0xf4, # CYRILLIC SMALL LETTER EF + 0x0445: 0xf5, # CYRILLIC SMALL LETTER HA + 0x0446: 0xf6, # CYRILLIC SMALL LETTER TSE + 0x0447: 0xf7, # CYRILLIC SMALL LETTER CHE + 0x0448: 0xf8, # CYRILLIC SMALL LETTER SHA + 0x0449: 0xf9, # CYRILLIC SMALL LETTER SHCHA + 0x044a: 0xfa, # CYRILLIC SMALL LETTER HARD SIGN + 0x044b: 0xfb, # CYRILLIC SMALL LETTER YERU + 0x044c: 0xfc, # CYRILLIC SMALL LETTER SOFT SIGN + 0x044d: 0xfd, # CYRILLIC SMALL LETTER E + 0x044e: 0xfe, # CYRILLIC SMALL LETTER YU + 0x044f: 0xdf, # CYRILLIC SMALL LETTER YA + 0x0451: 0xde, # CYRILLIC SMALL LETTER IO + 0x0452: 0xac, # CYRILLIC SMALL LETTER DJE + 0x0453: 0xaf, # CYRILLIC SMALL LETTER GJE + 0x0454: 0xb9, # CYRILLIC SMALL LETTER UKRAINIAN IE + 0x0455: 0xcf, # CYRILLIC SMALL LETTER DZE + 0x0456: 0xb4, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + 0x0457: 0xbb, # CYRILLIC SMALL LETTER YI + 0x0458: 0xc0, # CYRILLIC SMALL LETTER JE + 0x0459: 0xbd, # CYRILLIC SMALL LETTER LJE + 0x045a: 0xbf, # CYRILLIC SMALL LETTER NJE + 0x045b: 0xcc, # CYRILLIC SMALL LETTER TSHE + 0x045c: 0xce, # CYRILLIC SMALL LETTER KJE + 0x045e: 0xd9, # CYRILLIC SMALL LETTER SHORT U + 0x045f: 0xdb, # CYRILLIC SMALL LETTER DZHE + 0x0490: 0xa2, # CYRILLIC CAPITAL LETTER GHE WITH UPTURN + 0x0491: 0xb6, # CYRILLIC SMALL LETTER GHE WITH UPTURN + 0x2013: 0xd0, # EN DASH + 0x2014: 0xd1, # EM DASH + 0x2018: 0xd4, # LEFT SINGLE QUOTATION MARK + 0x2019: 0xd5, # RIGHT SINGLE QUOTATION MARK + 0x201c: 0xd2, # LEFT DOUBLE QUOTATION MARK + 0x201d: 0xd3, # RIGHT DOUBLE QUOTATION MARK + 0x201e: 0xd7, # DOUBLE LOW-9 QUOTATION MARK + 0x2020: 0xa0, # DAGGER + 0x2022: 0xa5, # BULLET + 0x2026: 0xc9, # HORIZONTAL ELLIPSIS + 0x20ac: 0xff, # EURO SIGN + 0x2116: 0xdc, # NUMERO SIGN + 0x2122: 0xaa, # TRADE MARK SIGN + 0x2206: 0xc6, # INCREMENT + 0x221a: 0xc3, # SQUARE ROOT + 0x221e: 0xb0, # INFINITY + 0x2248: 0xc5, # ALMOST EQUAL TO + 0x2260: 0xad, # NOT EQUAL TO + 0x2264: 0xb2, # LESS-THAN OR EQUAL TO + 0x2265: 0xb3, # GREATER-THAN OR EQUAL TO }
\ No newline at end of file |