diff options
| author | Martin v. Löwis <martin@v.loewis.de> | 2006-06-04 19:36:28 (GMT) |
|---|---|---|
| committer | Martin v. Löwis <martin@v.loewis.de> | 2006-06-04 19:36:28 (GMT) |
| commit | 3f767795f6784ca6bf797b055be67fce5bf2fa06 (patch) | |
| tree | 851eea9180023026ef3ef2319fae871a50bb479c | |
| parent | 67966bed72dc175938c4b971145c4746ac993f19 (diff) | |
| download | cpython-3f767795f6784ca6bf797b055be67fce5bf2fa06.zip cpython-3f767795f6784ca6bf797b055be67fce5bf2fa06.tar.gz cpython-3f767795f6784ca6bf797b055be67fce5bf2fa06.tar.bz2 | |
Patch #1359618: Speed-up charmap encoder.
51 files changed, 510 insertions, 11516 deletions
diff --git a/Include/unicodeobject.h b/Include/unicodeobject.h index 8c39cfe..3177051 100644 --- a/Include/unicodeobject.h +++ b/Include/unicodeobject.h @@ -650,6 +650,11 @@ PyAPI_FUNC(PyObject*) PyUnicode_AsEncodedString( const char *errors /* error handling */ ); +PyAPI_FUNC(PyObject*) PyUnicode_BuildEncodingMap( + PyObject* string /* 256 character map */ + ); + + /* --- UTF-7 Codecs ------------------------------------------------------- */ PyAPI_FUNC(PyObject*) PyUnicode_DecodeUTF7( diff --git a/Lib/encodings/cp037.py b/Lib/encodings/cp037.py index 7e589a9..b73f3f8 100644 --- a/Lib/encodings/cp037.py +++ b/Lib/encodings/cp037.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\x9f' # 0xFF -> CONTROL ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x37, # END OF TRANSMISSION - 0x0005: 0x2D, # ENQUIRY - 0x0006: 0x2E, # ACKNOWLEDGE - 0x0007: 0x2F, # BELL - 0x0008: 0x16, # BACKSPACE - 0x0009: 0x05, # HORIZONTAL TABULATION - 0x000A: 0x25, # 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 - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x3C, # DEVICE CONTROL FOUR - 0x0015: 0x3D, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x32, # SYNCHRONOUS IDLE - 0x0017: 0x26, # END OF TRANSMISSION BLOCK - 0x0018: 0x18, # CANCEL - 0x0019: 0x19, # END OF MEDIUM - 0x001A: 0x3F, # SUBSTITUTE - 0x001B: 0x27, # ESCAPE - 0x001C: 0x1C, # FILE SEPARATOR - 0x001D: 0x1D, # GROUP SEPARATOR - 0x001E: 0x1E, # RECORD SEPARATOR - 0x001F: 0x1F, # UNIT SEPARATOR - 0x0020: 0x40, # SPACE - 0x0021: 0x5A, # EXCLAMATION MARK - 0x0022: 0x7F, # QUOTATION MARK - 0x0023: 0x7B, # NUMBER SIGN - 0x0024: 0x5B, # DOLLAR SIGN - 0x0025: 0x6C, # PERCENT SIGN - 0x0026: 0x50, # AMPERSAND - 0x0027: 0x7D, # APOSTROPHE - 0x0028: 0x4D, # LEFT PARENTHESIS - 0x0029: 0x5D, # RIGHT PARENTHESIS - 0x002A: 0x5C, # ASTERISK - 0x002B: 0x4E, # PLUS SIGN - 0x002C: 0x6B, # COMMA - 0x002D: 0x60, # HYPHEN-MINUS - 0x002E: 0x4B, # FULL STOP - 0x002F: 0x61, # SOLIDUS - 0x0030: 0xF0, # DIGIT ZERO - 0x0031: 0xF1, # DIGIT ONE - 0x0032: 0xF2, # DIGIT TWO - 0x0033: 0xF3, # DIGIT THREE - 0x0034: 0xF4, # DIGIT FOUR - 0x0035: 0xF5, # DIGIT FIVE - 0x0036: 0xF6, # DIGIT SIX - 0x0037: 0xF7, # DIGIT SEVEN - 0x0038: 0xF8, # DIGIT EIGHT - 0x0039: 0xF9, # DIGIT NINE - 0x003A: 0x7A, # COLON - 0x003B: 0x5E, # SEMICOLON - 0x003C: 0x4C, # LESS-THAN SIGN - 0x003D: 0x7E, # EQUALS SIGN - 0x003E: 0x6E, # GREATER-THAN SIGN - 0x003F: 0x6F, # QUESTION MARK - 0x0040: 0x7C, # COMMERCIAL AT - 0x0041: 0xC1, # LATIN CAPITAL LETTER A - 0x0042: 0xC2, # LATIN CAPITAL LETTER B - 0x0043: 0xC3, # LATIN CAPITAL LETTER C - 0x0044: 0xC4, # LATIN CAPITAL LETTER D - 0x0045: 0xC5, # LATIN CAPITAL LETTER E - 0x0046: 0xC6, # LATIN CAPITAL LETTER F - 0x0047: 0xC7, # LATIN CAPITAL LETTER G - 0x0048: 0xC8, # LATIN CAPITAL LETTER H - 0x0049: 0xC9, # LATIN CAPITAL LETTER I - 0x004A: 0xD1, # LATIN CAPITAL LETTER J - 0x004B: 0xD2, # LATIN CAPITAL LETTER K - 0x004C: 0xD3, # LATIN CAPITAL LETTER L - 0x004D: 0xD4, # LATIN CAPITAL LETTER M - 0x004E: 0xD5, # LATIN CAPITAL LETTER N - 0x004F: 0xD6, # LATIN CAPITAL LETTER O - 0x0050: 0xD7, # LATIN CAPITAL LETTER P - 0x0051: 0xD8, # LATIN CAPITAL LETTER Q - 0x0052: 0xD9, # LATIN CAPITAL LETTER R - 0x0053: 0xE2, # LATIN CAPITAL LETTER S - 0x0054: 0xE3, # LATIN CAPITAL LETTER T - 0x0055: 0xE4, # LATIN CAPITAL LETTER U - 0x0056: 0xE5, # LATIN CAPITAL LETTER V - 0x0057: 0xE6, # LATIN CAPITAL LETTER W - 0x0058: 0xE7, # LATIN CAPITAL LETTER X - 0x0059: 0xE8, # LATIN CAPITAL LETTER Y - 0x005A: 0xE9, # LATIN CAPITAL LETTER Z - 0x005B: 0xBA, # LEFT SQUARE BRACKET - 0x005C: 0xE0, # REVERSE SOLIDUS - 0x005D: 0xBB, # RIGHT SQUARE BRACKET - 0x005E: 0xB0, # CIRCUMFLEX ACCENT - 0x005F: 0x6D, # LOW LINE - 0x0060: 0x79, # GRAVE ACCENT - 0x0061: 0x81, # LATIN SMALL LETTER A - 0x0062: 0x82, # LATIN SMALL LETTER B - 0x0063: 0x83, # LATIN SMALL LETTER C - 0x0064: 0x84, # LATIN SMALL LETTER D - 0x0065: 0x85, # LATIN SMALL LETTER E - 0x0066: 0x86, # LATIN SMALL LETTER F - 0x0067: 0x87, # LATIN SMALL LETTER G - 0x0068: 0x88, # LATIN SMALL LETTER H - 0x0069: 0x89, # LATIN SMALL LETTER I - 0x006A: 0x91, # LATIN SMALL LETTER J - 0x006B: 0x92, # LATIN SMALL LETTER K - 0x006C: 0x93, # LATIN SMALL LETTER L - 0x006D: 0x94, # LATIN SMALL LETTER M - 0x006E: 0x95, # LATIN SMALL LETTER N - 0x006F: 0x96, # LATIN SMALL LETTER O - 0x0070: 0x97, # LATIN SMALL LETTER P - 0x0071: 0x98, # LATIN SMALL LETTER Q - 0x0072: 0x99, # LATIN SMALL LETTER R - 0x0073: 0xA2, # LATIN SMALL LETTER S - 0x0074: 0xA3, # LATIN SMALL LETTER T - 0x0075: 0xA4, # LATIN SMALL LETTER U - 0x0076: 0xA5, # LATIN SMALL LETTER V - 0x0077: 0xA6, # LATIN SMALL LETTER W - 0x0078: 0xA7, # LATIN SMALL LETTER X - 0x0079: 0xA8, # LATIN SMALL LETTER Y - 0x007A: 0xA9, # LATIN SMALL LETTER Z - 0x007B: 0xC0, # LEFT CURLY BRACKET - 0x007C: 0x4F, # VERTICAL LINE - 0x007D: 0xD0, # RIGHT CURLY BRACKET - 0x007E: 0xA1, # TILDE - 0x007F: 0x07, # DELETE - 0x0080: 0x20, # CONTROL - 0x0081: 0x21, # CONTROL - 0x0082: 0x22, # CONTROL - 0x0083: 0x23, # CONTROL - 0x0084: 0x24, # CONTROL - 0x0085: 0x15, # CONTROL - 0x0086: 0x06, # CONTROL - 0x0087: 0x17, # CONTROL - 0x0088: 0x28, # CONTROL - 0x0089: 0x29, # CONTROL - 0x008A: 0x2A, # CONTROL - 0x008B: 0x2B, # CONTROL - 0x008C: 0x2C, # CONTROL - 0x008D: 0x09, # CONTROL - 0x008E: 0x0A, # CONTROL - 0x008F: 0x1B, # CONTROL - 0x0090: 0x30, # CONTROL - 0x0091: 0x31, # CONTROL - 0x0092: 0x1A, # CONTROL - 0x0093: 0x33, # CONTROL - 0x0094: 0x34, # CONTROL - 0x0095: 0x35, # CONTROL - 0x0096: 0x36, # CONTROL - 0x0097: 0x08, # CONTROL - 0x0098: 0x38, # CONTROL - 0x0099: 0x39, # CONTROL - 0x009A: 0x3A, # CONTROL - 0x009B: 0x3B, # CONTROL - 0x009C: 0x04, # CONTROL - 0x009D: 0x14, # CONTROL - 0x009E: 0x3E, # CONTROL - 0x009F: 0xFF, # CONTROL - 0x00A0: 0x41, # NO-BREAK SPACE - 0x00A1: 0xAA, # INVERTED EXCLAMATION MARK - 0x00A2: 0x4A, # CENT SIGN - 0x00A3: 0xB1, # POUND SIGN - 0x00A4: 0x9F, # CURRENCY SIGN - 0x00A5: 0xB2, # YEN SIGN - 0x00A6: 0x6A, # BROKEN BAR - 0x00A7: 0xB5, # SECTION SIGN - 0x00A8: 0xBD, # DIAERESIS - 0x00A9: 0xB4, # COPYRIGHT SIGN - 0x00AA: 0x9A, # FEMININE ORDINAL INDICATOR - 0x00AB: 0x8A, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0x5F, # NOT SIGN - 0x00AD: 0xCA, # SOFT HYPHEN - 0x00AE: 0xAF, # REGISTERED SIGN - 0x00AF: 0xBC, # MACRON - 0x00B0: 0x90, # DEGREE SIGN - 0x00B1: 0x8F, # PLUS-MINUS SIGN - 0x00B2: 0xEA, # SUPERSCRIPT TWO - 0x00B3: 0xFA, # SUPERSCRIPT THREE - 0x00B4: 0xBE, # ACUTE ACCENT - 0x00B5: 0xA0, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB3, # MIDDLE DOT - 0x00B8: 0x9D, # CEDILLA - 0x00B9: 0xDA, # SUPERSCRIPT ONE - 0x00BA: 0x9B, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0x8B, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BC: 0xB7, # VULGAR FRACTION ONE QUARTER - 0x00BD: 0xB8, # VULGAR FRACTION ONE HALF - 0x00BE: 0xB9, # VULGAR FRACTION THREE QUARTERS - 0x00BF: 0xAB, # INVERTED QUESTION MARK - 0x00C0: 0x64, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0x65, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0x62, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0x66, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x63, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x67, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0x9E, # LATIN CAPITAL LIGATURE AE - 0x00C7: 0x68, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0x74, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x71, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0x72, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0x73, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0x78, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0x75, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0x76, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0x77, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D0: 0xAC, # LATIN CAPITAL LETTER ETH (ICELANDIC) - 0x00D1: 0x69, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xED, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEB, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xEF, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0xEC, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xBF, # MULTIPLICATION SIGN - 0x00D8: 0x80, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xFD, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xFE, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xFB, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0xFC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xAD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xAE, # LATIN CAPITAL LETTER THORN (ICELANDIC) - 0x00DF: 0x59, # LATIN SMALL LETTER SHARP S (GERMAN) - 0x00E0: 0x44, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x45, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x42, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x46, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x43, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x47, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0x9C, # LATIN SMALL LIGATURE AE - 0x00E7: 0x48, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x54, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x51, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x52, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x53, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x58, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x55, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x56, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x57, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F0: 0x8C, # LATIN SMALL LETTER ETH (ICELANDIC) - 0x00F1: 0x49, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0xCD, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0xCE, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xCB, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0xCF, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0xCC, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xE1, # DIVISION SIGN - 0x00F8: 0x70, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0xDD, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0xDE, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0xDB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xDC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0x8D, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0x8E, # LATIN SMALL LETTER THORN (ICELANDIC) - 0x00FF: 0xDF, # LATIN SMALL LETTER Y WITH DIAERESIS -} diff --git a/Lib/encodings/cp1006.py b/Lib/encodings/cp1006.py index 7829969..584dd31 100644 --- a/Lib/encodings/cp1006.py +++ b/Lib/encodings/cp1006.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,262 +303,6 @@ decoding_table = ( u'\ufe7d' # 0xFF -> ARABIC SHADDA MEDIAL FORM ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00AD: 0xAD, # SOFT HYPHEN - 0x060C: 0xAB, # ARABIC COMMA - 0x061B: 0xAC, # ARABIC SEMICOLON - 0x061F: 0xAE, # ARABIC QUESTION MARK - 0x06F0: 0xA1, # EXTENDED ARABIC-INDIC DIGIT ZERO - 0x06F1: 0xA2, # EXTENDED ARABIC-INDIC DIGIT ONE - 0x06F2: 0xA3, # EXTENDED ARABIC-INDIC DIGIT TWO - 0x06F3: 0xA4, # EXTENDED ARABIC-INDIC DIGIT THREE - 0x06F4: 0xA5, # EXTENDED ARABIC-INDIC DIGIT FOUR - 0x06F5: 0xA6, # EXTENDED ARABIC-INDIC DIGIT FIVE - 0x06F6: 0xA7, # EXTENDED ARABIC-INDIC DIGIT SIX - 0x06F7: 0xA8, # EXTENDED ARABIC-INDIC DIGIT SEVEN - 0x06F8: 0xA9, # EXTENDED ARABIC-INDIC DIGIT EIGHT - 0x06F9: 0xAA, # EXTENDED ARABIC-INDIC DIGIT NINE - 0xFB56: 0xB5, # ARABIC LETTER PEH ISOLATED FORM - 0xFB58: 0xB6, # ARABIC LETTER PEH INITIAL FORM - 0xFB66: 0xBA, # ARABIC LETTER TTEH ISOLATED FORM - 0xFB68: 0xBB, # ARABIC LETTER TTEH INITIAL FORM - 0xFB7A: 0xC0, # ARABIC LETTER TCHEH ISOLATED FORM - 0xFB7C: 0xC1, # ARABIC LETTER TCHEH INITIAL FORM - 0xFB84: 0xC7, # ARABIC LETTER DAHAL ISOLATED FORMN - 0xFB8A: 0xCC, # ARABIC LETTER JEH ISOLATED FORM - 0xFB8C: 0xCA, # ARABIC LETTER RREH ISOLATED FORM - 0xFB92: 0xE5, # ARABIC LETTER GAF ISOLATED FORM - 0xFB94: 0xE6, # ARABIC LETTER GAF INITIAL FORM - 0xFB9E: 0xEC, # ARABIC LETTER NOON GHUNNA ISOLATED FORM - 0xFBA6: 0xF1, # ARABIC LETTER HEH GOAL ISOLATED FORM - 0xFBA8: 0xF2, # ARABIC LETTER HEH GOAL INITIAL FORM - 0xFBA9: 0xF3, # ARABIC LETTER HEH GOAL MEDIAL FORM - 0xFBAA: 0xF4, # ARABIC LETTER HEH DOACHASHMEE ISOLATED FORM - 0xFBAE: 0xFD, # ARABIC LETTER YEH BARREE ISOLATED FORM - 0xFBB0: 0xFC, # ARABIC LETTER YEH BARREE WITH HAMZA ABOVE ISOLATED FORM - 0xFE7C: 0xFE, # ARABIC SHADDA ISOLATED FORM - 0xFE7D: 0xFF, # ARABIC SHADDA MEDIAL FORM - 0xFE80: 0xF5, # ARABIC LETTER HAMZA ISOLATED FORM - 0xFE81: 0xAF, # ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM - 0xFE85: 0xEF, # ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM - 0xFE89: 0xF6, # ARABIC LETTER YEH WITH HAMZA ABOVE ISOLATED FORM - 0xFE8A: 0xF7, # ARABIC LETTER YEH WITH HAMZA ABOVE FINAL FORM - 0xFE8B: 0xF8, # ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM - 0xFE8D: 0xB0, # ARABIC LETTER ALEF ISOLATED FORM - 0xFE8E: None, # ARABIC LETTER ALEF FINAL FORM - 0xFE8F: 0xB3, # ARABIC LETTER BEH ISOLATED FORM - 0xFE91: 0xB4, # ARABIC LETTER BEH INITIAL FORM - 0xFE93: 0xB7, # ARABIC LETTER TEH MARBUTA ISOLATED FORM - 0xFE95: 0xB8, # ARABIC LETTER TEH ISOLATED FORM - 0xFE97: 0xB9, # ARABIC LETTER TEH INITIAL FORM - 0xFE99: 0xBC, # ARABIC LETTER THEH ISOLATED FORM - 0xFE9B: 0xBD, # ARABIC LETTER THEH INITIAL FORM - 0xFE9D: 0xBE, # ARABIC LETTER JEEM ISOLATED FORM - 0xFE9F: 0xBF, # ARABIC LETTER JEEM INITIAL FORM - 0xFEA1: 0xC2, # ARABIC LETTER HAH ISOLATED FORM - 0xFEA3: 0xC3, # ARABIC LETTER HAH INITIAL FORM - 0xFEA5: 0xC4, # ARABIC LETTER KHAH ISOLATED FORM - 0xFEA7: 0xC5, # ARABIC LETTER KHAH INITIAL FORM - 0xFEA9: 0xC6, # ARABIC LETTER DAL ISOLATED FORM - 0xFEAB: 0xC8, # ARABIC LETTER THAL ISOLATED FORM - 0xFEAD: 0xC9, # ARABIC LETTER REH ISOLATED FORM - 0xFEAF: 0xCB, # ARABIC LETTER ZAIN ISOLATED FORM - 0xFEB1: 0xCD, # ARABIC LETTER SEEN ISOLATED FORM - 0xFEB3: 0xCE, # ARABIC LETTER SEEN INITIAL FORM - 0xFEB5: 0xCF, # ARABIC LETTER SHEEN ISOLATED FORM - 0xFEB7: 0xD0, # ARABIC LETTER SHEEN INITIAL FORM - 0xFEB9: 0xD1, # ARABIC LETTER SAD ISOLATED FORM - 0xFEBB: 0xD2, # ARABIC LETTER SAD INITIAL FORM - 0xFEBD: 0xD3, # ARABIC LETTER DAD ISOLATED FORM - 0xFEBF: 0xD4, # ARABIC LETTER DAD INITIAL FORM - 0xFEC1: 0xD5, # ARABIC LETTER TAH ISOLATED FORM - 0xFEC5: 0xD6, # ARABIC LETTER ZAH ISOLATED FORM - 0xFEC9: 0xD7, # ARABIC LETTER AIN ISOLATED FORM - 0xFECA: 0xD8, # ARABIC LETTER AIN FINAL FORM - 0xFECB: 0xD9, # ARABIC LETTER AIN INITIAL FORM - 0xFECC: 0xDA, # ARABIC LETTER AIN MEDIAL FORM - 0xFECD: 0xDB, # ARABIC LETTER GHAIN ISOLATED FORM - 0xFECE: 0xDC, # ARABIC LETTER GHAIN FINAL FORM - 0xFECF: 0xDD, # ARABIC LETTER GHAIN INITIAL FORM - 0xFED0: 0xDE, # ARABIC LETTER GHAIN MEDIAL FORM - 0xFED1: 0xDF, # ARABIC LETTER FEH ISOLATED FORM - 0xFED3: 0xE0, # ARABIC LETTER FEH INITIAL FORM - 0xFED5: 0xE1, # ARABIC LETTER QAF ISOLATED FORM - 0xFED7: 0xE2, # ARABIC LETTER QAF INITIAL FORM - 0xFED9: 0xE3, # ARABIC LETTER KAF ISOLATED FORM - 0xFEDB: 0xE4, # ARABIC LETTER KAF INITIAL FORM - 0xFEDD: 0xE7, # ARABIC LETTER LAM ISOLATED FORM - 0xFEDF: 0xE8, # ARABIC LETTER LAM INITIAL FORM - 0xFEE0: 0xE9, # ARABIC LETTER LAM MEDIAL FORM - 0xFEE1: 0xEA, # ARABIC LETTER MEEM ISOLATED FORM - 0xFEE3: 0xEB, # ARABIC LETTER MEEM INITIAL FORM - 0xFEE5: 0xED, # ARABIC LETTER NOON ISOLATED FORM - 0xFEE7: 0xEE, # ARABIC LETTER NOON INITIAL FORM - 0xFEED: 0xF0, # ARABIC LETTER WAW ISOLATED FORM - 0xFEF1: 0xF9, # ARABIC LETTER YEH ISOLATED FORM - 0xFEF2: 0xFA, # ARABIC LETTER YEH FINAL FORM - 0xFEF3: 0xFB, # ARABIC LETTER YEH INITIAL FORM -} diff --git a/Lib/encodings/cp1026.py b/Lib/encodings/cp1026.py index 01c8804..6d607f9 100644 --- a/Lib/encodings/cp1026.py +++ b/Lib/encodings/cp1026.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\x9f' # 0xFF -> CONTROL ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x37, # END OF TRANSMISSION - 0x0005: 0x2D, # ENQUIRY - 0x0006: 0x2E, # ACKNOWLEDGE - 0x0007: 0x2F, # BELL - 0x0008: 0x16, # BACKSPACE - 0x0009: 0x05, # HORIZONTAL TABULATION - 0x000A: 0x25, # 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 - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x3C, # DEVICE CONTROL FOUR - 0x0015: 0x3D, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x32, # SYNCHRONOUS IDLE - 0x0017: 0x26, # END OF TRANSMISSION BLOCK - 0x0018: 0x18, # CANCEL - 0x0019: 0x19, # END OF MEDIUM - 0x001A: 0x3F, # SUBSTITUTE - 0x001B: 0x27, # ESCAPE - 0x001C: 0x1C, # FILE SEPARATOR - 0x001D: 0x1D, # GROUP SEPARATOR - 0x001E: 0x1E, # RECORD SEPARATOR - 0x001F: 0x1F, # UNIT SEPARATOR - 0x0020: 0x40, # SPACE - 0x0021: 0x4F, # EXCLAMATION MARK - 0x0022: 0xFC, # QUOTATION MARK - 0x0023: 0xEC, # NUMBER SIGN - 0x0024: 0xAD, # DOLLAR SIGN - 0x0025: 0x6C, # PERCENT SIGN - 0x0026: 0x50, # AMPERSAND - 0x0027: 0x7D, # APOSTROPHE - 0x0028: 0x4D, # LEFT PARENTHESIS - 0x0029: 0x5D, # RIGHT PARENTHESIS - 0x002A: 0x5C, # ASTERISK - 0x002B: 0x4E, # PLUS SIGN - 0x002C: 0x6B, # COMMA - 0x002D: 0x60, # HYPHEN-MINUS - 0x002E: 0x4B, # FULL STOP - 0x002F: 0x61, # SOLIDUS - 0x0030: 0xF0, # DIGIT ZERO - 0x0031: 0xF1, # DIGIT ONE - 0x0032: 0xF2, # DIGIT TWO - 0x0033: 0xF3, # DIGIT THREE - 0x0034: 0xF4, # DIGIT FOUR - 0x0035: 0xF5, # DIGIT FIVE - 0x0036: 0xF6, # DIGIT SIX - 0x0037: 0xF7, # DIGIT SEVEN - 0x0038: 0xF8, # DIGIT EIGHT - 0x0039: 0xF9, # DIGIT NINE - 0x003A: 0x7A, # COLON - 0x003B: 0x5E, # SEMICOLON - 0x003C: 0x4C, # LESS-THAN SIGN - 0x003D: 0x7E, # EQUALS SIGN - 0x003E: 0x6E, # GREATER-THAN SIGN - 0x003F: 0x6F, # QUESTION MARK - 0x0040: 0xAE, # COMMERCIAL AT - 0x0041: 0xC1, # LATIN CAPITAL LETTER A - 0x0042: 0xC2, # LATIN CAPITAL LETTER B - 0x0043: 0xC3, # LATIN CAPITAL LETTER C - 0x0044: 0xC4, # LATIN CAPITAL LETTER D - 0x0045: 0xC5, # LATIN CAPITAL LETTER E - 0x0046: 0xC6, # LATIN CAPITAL LETTER F - 0x0047: 0xC7, # LATIN CAPITAL LETTER G - 0x0048: 0xC8, # LATIN CAPITAL LETTER H - 0x0049: 0xC9, # LATIN CAPITAL LETTER I - 0x004A: 0xD1, # LATIN CAPITAL LETTER J - 0x004B: 0xD2, # LATIN CAPITAL LETTER K - 0x004C: 0xD3, # LATIN CAPITAL LETTER L - 0x004D: 0xD4, # LATIN CAPITAL LETTER M - 0x004E: 0xD5, # LATIN CAPITAL LETTER N - 0x004F: 0xD6, # LATIN CAPITAL LETTER O - 0x0050: 0xD7, # LATIN CAPITAL LETTER P - 0x0051: 0xD8, # LATIN CAPITAL LETTER Q - 0x0052: 0xD9, # LATIN CAPITAL LETTER R - 0x0053: 0xE2, # LATIN CAPITAL LETTER S - 0x0054: 0xE3, # LATIN CAPITAL LETTER T - 0x0055: 0xE4, # LATIN CAPITAL LETTER U - 0x0056: 0xE5, # LATIN CAPITAL LETTER V - 0x0057: 0xE6, # LATIN CAPITAL LETTER W - 0x0058: 0xE7, # LATIN CAPITAL LETTER X - 0x0059: 0xE8, # LATIN CAPITAL LETTER Y - 0x005A: 0xE9, # LATIN CAPITAL LETTER Z - 0x005B: 0x68, # LEFT SQUARE BRACKET - 0x005C: 0xDC, # REVERSE SOLIDUS - 0x005D: 0xAC, # RIGHT SQUARE BRACKET - 0x005E: 0x5F, # CIRCUMFLEX ACCENT - 0x005F: 0x6D, # LOW LINE - 0x0060: 0x8D, # GRAVE ACCENT - 0x0061: 0x81, # LATIN SMALL LETTER A - 0x0062: 0x82, # LATIN SMALL LETTER B - 0x0063: 0x83, # LATIN SMALL LETTER C - 0x0064: 0x84, # LATIN SMALL LETTER D - 0x0065: 0x85, # LATIN SMALL LETTER E - 0x0066: 0x86, # LATIN SMALL LETTER F - 0x0067: 0x87, # LATIN SMALL LETTER G - 0x0068: 0x88, # LATIN SMALL LETTER H - 0x0069: 0x89, # LATIN SMALL LETTER I - 0x006A: 0x91, # LATIN SMALL LETTER J - 0x006B: 0x92, # LATIN SMALL LETTER K - 0x006C: 0x93, # LATIN SMALL LETTER L - 0x006D: 0x94, # LATIN SMALL LETTER M - 0x006E: 0x95, # LATIN SMALL LETTER N - 0x006F: 0x96, # LATIN SMALL LETTER O - 0x0070: 0x97, # LATIN SMALL LETTER P - 0x0071: 0x98, # LATIN SMALL LETTER Q - 0x0072: 0x99, # LATIN SMALL LETTER R - 0x0073: 0xA2, # LATIN SMALL LETTER S - 0x0074: 0xA3, # LATIN SMALL LETTER T - 0x0075: 0xA4, # LATIN SMALL LETTER U - 0x0076: 0xA5, # LATIN SMALL LETTER V - 0x0077: 0xA6, # LATIN SMALL LETTER W - 0x0078: 0xA7, # LATIN SMALL LETTER X - 0x0079: 0xA8, # LATIN SMALL LETTER Y - 0x007A: 0xA9, # LATIN SMALL LETTER Z - 0x007B: 0x48, # LEFT CURLY BRACKET - 0x007C: 0xBB, # VERTICAL LINE - 0x007D: 0x8C, # RIGHT CURLY BRACKET - 0x007E: 0xCC, # TILDE - 0x007F: 0x07, # DELETE - 0x0080: 0x20, # CONTROL - 0x0081: 0x21, # CONTROL - 0x0082: 0x22, # CONTROL - 0x0083: 0x23, # CONTROL - 0x0084: 0x24, # CONTROL - 0x0085: 0x15, # CONTROL - 0x0086: 0x06, # CONTROL - 0x0087: 0x17, # CONTROL - 0x0088: 0x28, # CONTROL - 0x0089: 0x29, # CONTROL - 0x008A: 0x2A, # CONTROL - 0x008B: 0x2B, # CONTROL - 0x008C: 0x2C, # CONTROL - 0x008D: 0x09, # CONTROL - 0x008E: 0x0A, # CONTROL - 0x008F: 0x1B, # CONTROL - 0x0090: 0x30, # CONTROL - 0x0091: 0x31, # CONTROL - 0x0092: 0x1A, # CONTROL - 0x0093: 0x33, # CONTROL - 0x0094: 0x34, # CONTROL - 0x0095: 0x35, # CONTROL - 0x0096: 0x36, # CONTROL - 0x0097: 0x08, # CONTROL - 0x0098: 0x38, # CONTROL - 0x0099: 0x39, # CONTROL - 0x009A: 0x3A, # CONTROL - 0x009B: 0x3B, # CONTROL - 0x009C: 0x04, # CONTROL - 0x009D: 0x14, # CONTROL - 0x009E: 0x3E, # CONTROL - 0x009F: 0xFF, # CONTROL - 0x00A0: 0x41, # NO-BREAK SPACE - 0x00A1: 0xAA, # INVERTED EXCLAMATION MARK - 0x00A2: 0xB0, # CENT SIGN - 0x00A3: 0xB1, # POUND SIGN - 0x00A4: 0x9F, # CURRENCY SIGN - 0x00A5: 0xB2, # YEN SIGN - 0x00A6: 0x8E, # BROKEN BAR - 0x00A7: 0xB5, # SECTION SIGN - 0x00A8: 0xBD, # DIAERESIS - 0x00A9: 0xB4, # COPYRIGHT SIGN - 0x00AA: 0x9A, # FEMININE ORDINAL INDICATOR - 0x00AB: 0x8A, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xBA, # NOT SIGN - 0x00AD: 0xCA, # SOFT HYPHEN - 0x00AE: 0xAF, # REGISTERED SIGN - 0x00AF: 0xBC, # MACRON - 0x00B0: 0x90, # DEGREE SIGN - 0x00B1: 0x8F, # PLUS-MINUS SIGN - 0x00B2: 0xEA, # SUPERSCRIPT TWO - 0x00B3: 0xFA, # SUPERSCRIPT THREE - 0x00B4: 0xBE, # ACUTE ACCENT - 0x00B5: 0xA0, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB3, # MIDDLE DOT - 0x00B8: 0x9D, # CEDILLA - 0x00B9: 0xDA, # SUPERSCRIPT ONE - 0x00BA: 0x9B, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0x8B, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BC: 0xB7, # VULGAR FRACTION ONE QUARTER - 0x00BD: 0xB8, # VULGAR FRACTION ONE HALF - 0x00BE: 0xB9, # VULGAR FRACTION THREE QUARTERS - 0x00BF: 0xAB, # INVERTED QUESTION MARK - 0x00C0: 0x64, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0x65, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0x62, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0x66, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x63, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x67, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0x9E, # LATIN CAPITAL LIGATURE AE - 0x00C7: 0x4A, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0x74, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x71, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0x72, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0x73, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0x78, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0x75, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0x76, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0x77, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D1: 0x69, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xED, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEB, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xEF, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x7B, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xBF, # MULTIPLICATION SIGN - 0x00D8: 0x80, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xFD, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xFE, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xFB, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0x7F, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0x59, # LATIN SMALL LETTER SHARP S (GERMAN) - 0x00E0: 0x44, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x45, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x42, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x46, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x43, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x47, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0x9C, # LATIN SMALL LIGATURE AE - 0x00E7: 0xC0, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x54, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x51, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x52, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x53, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x58, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x55, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x56, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x57, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F1: 0x49, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0xCD, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0xCE, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xCB, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0xCF, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0xA1, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xE1, # DIVISION SIGN - 0x00F8: 0x70, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0xDD, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0xDE, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0xDB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xE0, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FF: 0xDF, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x011E: 0x5A, # LATIN CAPITAL LETTER G WITH BREVE - 0x011F: 0xD0, # LATIN SMALL LETTER G WITH BREVE - 0x0130: 0x5B, # LATIN CAPITAL LETTER I WITH DOT ABOVE - 0x0131: 0x79, # LATIN SMALL LETTER DOTLESS I - 0x015E: 0x7C, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015F: 0x6A, # LATIN SMALL LETTER S WITH CEDILLA -} diff --git a/Lib/encodings/cp1140.py b/Lib/encodings/cp1140.py index ac8d41b..fd464ec 100644 --- a/Lib/encodings/cp1140.py +++ b/Lib/encodings/cp1140.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\x9f' # 0xFF -> CONTROL ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x37, # END OF TRANSMISSION - 0x0005: 0x2D, # ENQUIRY - 0x0006: 0x2E, # ACKNOWLEDGE - 0x0007: 0x2F, # BELL - 0x0008: 0x16, # BACKSPACE - 0x0009: 0x05, # HORIZONTAL TABULATION - 0x000A: 0x25, # 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 - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x3C, # DEVICE CONTROL FOUR - 0x0015: 0x3D, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x32, # SYNCHRONOUS IDLE - 0x0017: 0x26, # END OF TRANSMISSION BLOCK - 0x0018: 0x18, # CANCEL - 0x0019: 0x19, # END OF MEDIUM - 0x001A: 0x3F, # SUBSTITUTE - 0x001B: 0x27, # ESCAPE - 0x001C: 0x1C, # FILE SEPARATOR - 0x001D: 0x1D, # GROUP SEPARATOR - 0x001E: 0x1E, # RECORD SEPARATOR - 0x001F: 0x1F, # UNIT SEPARATOR - 0x0020: 0x40, # SPACE - 0x0021: 0x5A, # EXCLAMATION MARK - 0x0022: 0x7F, # QUOTATION MARK - 0x0023: 0x7B, # NUMBER SIGN - 0x0024: 0x5B, # DOLLAR SIGN - 0x0025: 0x6C, # PERCENT SIGN - 0x0026: 0x50, # AMPERSAND - 0x0027: 0x7D, # APOSTROPHE - 0x0028: 0x4D, # LEFT PARENTHESIS - 0x0029: 0x5D, # RIGHT PARENTHESIS - 0x002A: 0x5C, # ASTERISK - 0x002B: 0x4E, # PLUS SIGN - 0x002C: 0x6B, # COMMA - 0x002D: 0x60, # HYPHEN-MINUS - 0x002E: 0x4B, # FULL STOP - 0x002F: 0x61, # SOLIDUS - 0x0030: 0xF0, # DIGIT ZERO - 0x0031: 0xF1, # DIGIT ONE - 0x0032: 0xF2, # DIGIT TWO - 0x0033: 0xF3, # DIGIT THREE - 0x0034: 0xF4, # DIGIT FOUR - 0x0035: 0xF5, # DIGIT FIVE - 0x0036: 0xF6, # DIGIT SIX - 0x0037: 0xF7, # DIGIT SEVEN - 0x0038: 0xF8, # DIGIT EIGHT - 0x0039: 0xF9, # DIGIT NINE - 0x003A: 0x7A, # COLON - 0x003B: 0x5E, # SEMICOLON - 0x003C: 0x4C, # LESS-THAN SIGN - 0x003D: 0x7E, # EQUALS SIGN - 0x003E: 0x6E, # GREATER-THAN SIGN - 0x003F: 0x6F, # QUESTION MARK - 0x0040: 0x7C, # COMMERCIAL AT - 0x0041: 0xC1, # LATIN CAPITAL LETTER A - 0x0042: 0xC2, # LATIN CAPITAL LETTER B - 0x0043: 0xC3, # LATIN CAPITAL LETTER C - 0x0044: 0xC4, # LATIN CAPITAL LETTER D - 0x0045: 0xC5, # LATIN CAPITAL LETTER E - 0x0046: 0xC6, # LATIN CAPITAL LETTER F - 0x0047: 0xC7, # LATIN CAPITAL LETTER G - 0x0048: 0xC8, # LATIN CAPITAL LETTER H - 0x0049: 0xC9, # LATIN CAPITAL LETTER I - 0x004A: 0xD1, # LATIN CAPITAL LETTER J - 0x004B: 0xD2, # LATIN CAPITAL LETTER K - 0x004C: 0xD3, # LATIN CAPITAL LETTER L - 0x004D: 0xD4, # LATIN CAPITAL LETTER M - 0x004E: 0xD5, # LATIN CAPITAL LETTER N - 0x004F: 0xD6, # LATIN CAPITAL LETTER O - 0x0050: 0xD7, # LATIN CAPITAL LETTER P - 0x0051: 0xD8, # LATIN CAPITAL LETTER Q - 0x0052: 0xD9, # LATIN CAPITAL LETTER R - 0x0053: 0xE2, # LATIN CAPITAL LETTER S - 0x0054: 0xE3, # LATIN CAPITAL LETTER T - 0x0055: 0xE4, # LATIN CAPITAL LETTER U - 0x0056: 0xE5, # LATIN CAPITAL LETTER V - 0x0057: 0xE6, # LATIN CAPITAL LETTER W - 0x0058: 0xE7, # LATIN CAPITAL LETTER X - 0x0059: 0xE8, # LATIN CAPITAL LETTER Y - 0x005A: 0xE9, # LATIN CAPITAL LETTER Z - 0x005B: 0xBA, # LEFT SQUARE BRACKET - 0x005C: 0xE0, # REVERSE SOLIDUS - 0x005D: 0xBB, # RIGHT SQUARE BRACKET - 0x005E: 0xB0, # CIRCUMFLEX ACCENT - 0x005F: 0x6D, # LOW LINE - 0x0060: 0x79, # GRAVE ACCENT - 0x0061: 0x81, # LATIN SMALL LETTER A - 0x0062: 0x82, # LATIN SMALL LETTER B - 0x0063: 0x83, # LATIN SMALL LETTER C - 0x0064: 0x84, # LATIN SMALL LETTER D - 0x0065: 0x85, # LATIN SMALL LETTER E - 0x0066: 0x86, # LATIN SMALL LETTER F - 0x0067: 0x87, # LATIN SMALL LETTER G - 0x0068: 0x88, # LATIN SMALL LETTER H - 0x0069: 0x89, # LATIN SMALL LETTER I - 0x006A: 0x91, # LATIN SMALL LETTER J - 0x006B: 0x92, # LATIN SMALL LETTER K - 0x006C: 0x93, # LATIN SMALL LETTER L - 0x006D: 0x94, # LATIN SMALL LETTER M - 0x006E: 0x95, # LATIN SMALL LETTER N - 0x006F: 0x96, # LATIN SMALL LETTER O - 0x0070: 0x97, # LATIN SMALL LETTER P - 0x0071: 0x98, # LATIN SMALL LETTER Q - 0x0072: 0x99, # LATIN SMALL LETTER R - 0x0073: 0xA2, # LATIN SMALL LETTER S - 0x0074: 0xA3, # LATIN SMALL LETTER T - 0x0075: 0xA4, # LATIN SMALL LETTER U - 0x0076: 0xA5, # LATIN SMALL LETTER V - 0x0077: 0xA6, # LATIN SMALL LETTER W - 0x0078: 0xA7, # LATIN SMALL LETTER X - 0x0079: 0xA8, # LATIN SMALL LETTER Y - 0x007A: 0xA9, # LATIN SMALL LETTER Z - 0x007B: 0xC0, # LEFT CURLY BRACKET - 0x007C: 0x4F, # VERTICAL LINE - 0x007D: 0xD0, # RIGHT CURLY BRACKET - 0x007E: 0xA1, # TILDE - 0x007F: 0x07, # DELETE - 0x0080: 0x20, # CONTROL - 0x0081: 0x21, # CONTROL - 0x0082: 0x22, # CONTROL - 0x0083: 0x23, # CONTROL - 0x0084: 0x24, # CONTROL - 0x0085: 0x15, # CONTROL - 0x0086: 0x06, # CONTROL - 0x0087: 0x17, # CONTROL - 0x0088: 0x28, # CONTROL - 0x0089: 0x29, # CONTROL - 0x008A: 0x2A, # CONTROL - 0x008B: 0x2B, # CONTROL - 0x008C: 0x2C, # CONTROL - 0x008D: 0x09, # CONTROL - 0x008E: 0x0A, # CONTROL - 0x008F: 0x1B, # CONTROL - 0x0090: 0x30, # CONTROL - 0x0091: 0x31, # CONTROL - 0x0092: 0x1A, # CONTROL - 0x0093: 0x33, # CONTROL - 0x0094: 0x34, # CONTROL - 0x0095: 0x35, # CONTROL - 0x0096: 0x36, # CONTROL - 0x0097: 0x08, # CONTROL - 0x0098: 0x38, # CONTROL - 0x0099: 0x39, # CONTROL - 0x009A: 0x3A, # CONTROL - 0x009B: 0x3B, # CONTROL - 0x009C: 0x04, # CONTROL - 0x009D: 0x14, # CONTROL - 0x009E: 0x3E, # CONTROL - 0x009F: 0xFF, # CONTROL - 0x00A0: 0x41, # NO-BREAK SPACE - 0x00A1: 0xAA, # INVERTED EXCLAMATION MARK - 0x00A2: 0x4A, # CENT SIGN - 0x00A3: 0xB1, # POUND SIGN - 0x00A5: 0xB2, # YEN SIGN - 0x00A6: 0x6A, # BROKEN BAR - 0x00A7: 0xB5, # SECTION SIGN - 0x00A8: 0xBD, # DIAERESIS - 0x00A9: 0xB4, # COPYRIGHT SIGN - 0x00AA: 0x9A, # FEMININE ORDINAL INDICATOR - 0x00AB: 0x8A, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0x5F, # NOT SIGN - 0x00AD: 0xCA, # SOFT HYPHEN - 0x00AE: 0xAF, # REGISTERED SIGN - 0x00AF: 0xBC, # MACRON - 0x00B0: 0x90, # DEGREE SIGN - 0x00B1: 0x8F, # PLUS-MINUS SIGN - 0x00B2: 0xEA, # SUPERSCRIPT TWO - 0x00B3: 0xFA, # SUPERSCRIPT THREE - 0x00B4: 0xBE, # ACUTE ACCENT - 0x00B5: 0xA0, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB3, # MIDDLE DOT - 0x00B8: 0x9D, # CEDILLA - 0x00B9: 0xDA, # SUPERSCRIPT ONE - 0x00BA: 0x9B, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0x8B, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BC: 0xB7, # VULGAR FRACTION ONE QUARTER - 0x00BD: 0xB8, # VULGAR FRACTION ONE HALF - 0x00BE: 0xB9, # VULGAR FRACTION THREE QUARTERS - 0x00BF: 0xAB, # INVERTED QUESTION MARK - 0x00C0: 0x64, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0x65, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0x62, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0x66, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x63, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x67, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0x9E, # LATIN CAPITAL LIGATURE AE - 0x00C7: 0x68, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0x74, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x71, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0x72, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0x73, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0x78, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0x75, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0x76, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0x77, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D0: 0xAC, # LATIN CAPITAL LETTER ETH (ICELANDIC) - 0x00D1: 0x69, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xED, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEB, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xEF, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0xEC, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xBF, # MULTIPLICATION SIGN - 0x00D8: 0x80, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xFD, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xFE, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xFB, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0xFC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xAD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xAE, # LATIN CAPITAL LETTER THORN (ICELANDIC) - 0x00DF: 0x59, # LATIN SMALL LETTER SHARP S (GERMAN) - 0x00E0: 0x44, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x45, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x42, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x46, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x43, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x47, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0x9C, # LATIN SMALL LIGATURE AE - 0x00E7: 0x48, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x54, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x51, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x52, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x53, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x58, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x55, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x56, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x57, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F0: 0x8C, # LATIN SMALL LETTER ETH (ICELANDIC) - 0x00F1: 0x49, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0xCD, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0xCE, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xCB, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0xCF, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0xCC, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xE1, # DIVISION SIGN - 0x00F8: 0x70, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0xDD, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0xDE, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0xDB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xDC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0x8D, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0x8E, # LATIN SMALL LETTER THORN (ICELANDIC) - 0x00FF: 0xDF, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x20AC: 0x9F, # EURO SIGN -} diff --git a/Lib/encodings/cp1250.py b/Lib/encodings/cp1250.py index 6e6f57c..9685ed7 100644 --- a/Lib/encodings/cp1250.py +++ b/Lib/encodings/cp1250.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,258 +303,6 @@ decoding_table = ( u'\u02d9' # 0xFF -> DOT ABOVE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A6: 0xA6, # BROKEN BAR - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0xA8, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAC, # NOT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AE: 0xAE, # REGISTERED SIGN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B4: 0xB4, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00B8: 0xB8, # CEDILLA - 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00C1: 0xC1, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xC2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C7: 0xC7, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CB: 0xCB, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CD: 0xCD, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xCE, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xD4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xD7, # MULTIPLICATION SIGN - 0x00DA: 0xDA, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DC: 0xDC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S - 0x00E1: 0xE1, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0xE2, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E7: 0xE7, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE - 0x00EB: 0xEB, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00ED: 0xED, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0xEE, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xF4, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xF7, # DIVISION SIGN - 0x00FA: 0xFA, # LATIN SMALL LETTER U WITH ACUTE - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x0102: 0xC3, # LATIN CAPITAL LETTER A WITH BREVE - 0x0103: 0xE3, # LATIN SMALL LETTER A WITH BREVE - 0x0104: 0xA5, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xB9, # LATIN SMALL LETTER A WITH OGONEK - 0x0106: 0xC6, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0xE6, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x010E: 0xCF, # LATIN CAPITAL LETTER D WITH CARON - 0x010F: 0xEF, # LATIN SMALL LETTER D WITH CARON - 0x0110: 0xD0, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xF0, # LATIN SMALL LETTER D WITH STROKE - 0x0118: 0xCA, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xEA, # LATIN SMALL LETTER E WITH OGONEK - 0x011A: 0xCC, # LATIN CAPITAL LETTER E WITH CARON - 0x011B: 0xEC, # LATIN SMALL LETTER E WITH CARON - 0x0139: 0xC5, # LATIN CAPITAL LETTER L WITH ACUTE - 0x013A: 0xE5, # LATIN SMALL LETTER L WITH ACUTE - 0x013D: 0xBC, # LATIN CAPITAL LETTER L WITH CARON - 0x013E: 0xBE, # LATIN SMALL LETTER L WITH CARON - 0x0141: 0xA3, # LATIN CAPITAL LETTER L WITH STROKE - 0x0142: 0xB3, # LATIN SMALL LETTER L WITH STROKE - 0x0143: 0xD1, # LATIN CAPITAL LETTER N WITH ACUTE - 0x0144: 0xF1, # LATIN SMALL LETTER N WITH ACUTE - 0x0147: 0xD2, # LATIN CAPITAL LETTER N WITH CARON - 0x0148: 0xF2, # LATIN SMALL LETTER N WITH CARON - 0x0150: 0xD5, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - 0x0151: 0xF5, # LATIN SMALL LETTER O WITH DOUBLE ACUTE - 0x0154: 0xC0, # LATIN CAPITAL LETTER R WITH ACUTE - 0x0155: 0xE0, # LATIN SMALL LETTER R WITH ACUTE - 0x0158: 0xD8, # LATIN CAPITAL LETTER R WITH CARON - 0x0159: 0xF8, # LATIN SMALL LETTER R WITH CARON - 0x015A: 0x8C, # LATIN CAPITAL LETTER S WITH ACUTE - 0x015B: 0x9C, # LATIN SMALL LETTER S WITH ACUTE - 0x015E: 0xAA, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015F: 0xBA, # LATIN SMALL LETTER S WITH CEDILLA - 0x0160: 0x8A, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0x9A, # LATIN SMALL LETTER S WITH CARON - 0x0162: 0xDE, # LATIN CAPITAL LETTER T WITH CEDILLA - 0x0163: 0xFE, # LATIN SMALL LETTER T WITH CEDILLA - 0x0164: 0x8D, # LATIN CAPITAL LETTER T WITH CARON - 0x0165: 0x9D, # LATIN SMALL LETTER T WITH CARON - 0x016E: 0xD9, # LATIN CAPITAL LETTER U WITH RING ABOVE - 0x016F: 0xF9, # LATIN SMALL LETTER U WITH RING ABOVE - 0x0170: 0xDB, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - 0x0171: 0xFB, # LATIN SMALL LETTER U WITH DOUBLE ACUTE - 0x0179: 0x8F, # LATIN CAPITAL LETTER Z WITH ACUTE - 0x017A: 0x9F, # LATIN SMALL LETTER Z WITH ACUTE - 0x017B: 0xAF, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xBF, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x017D: 0x8E, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0x9E, # LATIN SMALL LETTER Z WITH CARON - 0x02C7: 0xA1, # CARON - 0x02D8: 0xA2, # BREVE - 0x02D9: 0xFF, # DOT ABOVE - 0x02DB: 0xB2, # OGONEK - 0x02DD: 0xBD, # DOUBLE ACUTE ACCENT - 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 - 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 - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1251.py b/Lib/encodings/cp1251.py index ed835fe..dd21975 100644 --- a/Lib/encodings/cp1251.py +++ b/Lib/encodings/cp1251.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,262 +303,6 @@ decoding_table = ( u'\u044f' # 0xFF -> CYRILLIC SMALL LETTER YA ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A6: 0xA6, # BROKEN BAR - 0x00A7: 0xA7, # SECTION SIGN - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAC, # NOT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AE: 0xAE, # REGISTERED SIGN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x0401: 0xA8, # CYRILLIC CAPITAL LETTER IO - 0x0402: 0x80, # CYRILLIC CAPITAL LETTER DJE - 0x0403: 0x81, # CYRILLIC CAPITAL LETTER GJE - 0x0404: 0xAA, # CYRILLIC CAPITAL LETTER UKRAINIAN IE - 0x0405: 0xBD, # CYRILLIC CAPITAL LETTER DZE - 0x0406: 0xB2, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0407: 0xAF, # CYRILLIC CAPITAL LETTER YI - 0x0408: 0xA3, # CYRILLIC CAPITAL LETTER JE - 0x0409: 0x8A, # CYRILLIC CAPITAL LETTER LJE - 0x040A: 0x8C, # CYRILLIC CAPITAL LETTER NJE - 0x040B: 0x8E, # CYRILLIC CAPITAL LETTER TSHE - 0x040C: 0x8D, # CYRILLIC CAPITAL LETTER KJE - 0x040E: 0xA1, # CYRILLIC CAPITAL LETTER SHORT U - 0x040F: 0x8F, # CYRILLIC CAPITAL LETTER DZHE - 0x0410: 0xC0, # CYRILLIC CAPITAL LETTER A - 0x0411: 0xC1, # CYRILLIC CAPITAL LETTER BE - 0x0412: 0xC2, # CYRILLIC CAPITAL LETTER VE - 0x0413: 0xC3, # CYRILLIC CAPITAL LETTER GHE - 0x0414: 0xC4, # CYRILLIC CAPITAL LETTER DE - 0x0415: 0xC5, # CYRILLIC CAPITAL LETTER IE - 0x0416: 0xC6, # CYRILLIC CAPITAL LETTER ZHE - 0x0417: 0xC7, # CYRILLIC CAPITAL LETTER ZE - 0x0418: 0xC8, # CYRILLIC CAPITAL LETTER I - 0x0419: 0xC9, # CYRILLIC CAPITAL LETTER SHORT I - 0x041A: 0xCA, # CYRILLIC CAPITAL LETTER KA - 0x041B: 0xCB, # CYRILLIC CAPITAL LETTER EL - 0x041C: 0xCC, # CYRILLIC CAPITAL LETTER EM - 0x041D: 0xCD, # CYRILLIC CAPITAL LETTER EN - 0x041E: 0xCE, # CYRILLIC CAPITAL LETTER O - 0x041F: 0xCF, # CYRILLIC CAPITAL LETTER PE - 0x0420: 0xD0, # CYRILLIC CAPITAL LETTER ER - 0x0421: 0xD1, # CYRILLIC CAPITAL LETTER ES - 0x0422: 0xD2, # CYRILLIC CAPITAL LETTER TE - 0x0423: 0xD3, # CYRILLIC CAPITAL LETTER U - 0x0424: 0xD4, # CYRILLIC CAPITAL LETTER EF - 0x0425: 0xD5, # CYRILLIC CAPITAL LETTER HA - 0x0426: 0xD6, # CYRILLIC CAPITAL LETTER TSE - 0x0427: 0xD7, # CYRILLIC CAPITAL LETTER CHE - 0x0428: 0xD8, # CYRILLIC CAPITAL LETTER SHA - 0x0429: 0xD9, # CYRILLIC CAPITAL LETTER SHCHA - 0x042A: 0xDA, # CYRILLIC CAPITAL LETTER HARD SIGN - 0x042B: 0xDB, # CYRILLIC CAPITAL LETTER YERU - 0x042C: 0xDC, # CYRILLIC CAPITAL LETTER SOFT SIGN - 0x042D: 0xDD, # CYRILLIC CAPITAL LETTER E - 0x042E: 0xDE, # CYRILLIC CAPITAL LETTER YU - 0x042F: 0xDF, # 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: 0xFF, # CYRILLIC SMALL LETTER YA - 0x0451: 0xB8, # CYRILLIC SMALL LETTER IO - 0x0452: 0x90, # CYRILLIC SMALL LETTER DJE - 0x0453: 0x83, # CYRILLIC SMALL LETTER GJE - 0x0454: 0xBA, # CYRILLIC SMALL LETTER UKRAINIAN IE - 0x0455: 0xBE, # CYRILLIC SMALL LETTER DZE - 0x0456: 0xB3, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0457: 0xBF, # CYRILLIC SMALL LETTER YI - 0x0458: 0xBC, # CYRILLIC SMALL LETTER JE - 0x0459: 0x9A, # CYRILLIC SMALL LETTER LJE - 0x045A: 0x9C, # CYRILLIC SMALL LETTER NJE - 0x045B: 0x9E, # CYRILLIC SMALL LETTER TSHE - 0x045C: 0x9D, # CYRILLIC SMALL LETTER KJE - 0x045E: 0xA2, # CYRILLIC SMALL LETTER SHORT U - 0x045F: 0x9F, # CYRILLIC SMALL LETTER DZHE - 0x0490: 0xA5, # CYRILLIC CAPITAL LETTER GHE WITH UPTURN - 0x0491: 0xB4, # CYRILLIC SMALL LETTER GHE WITH UPTURN - 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 - 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: 0x88, # EURO SIGN - 0x2116: 0xB9, # NUMERO SIGN - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1252.py b/Lib/encodings/cp1252.py index e5b6905..d98644a 100644 --- a/Lib/encodings/cp1252.py +++ b/Lib/encodings/cp1252.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,258 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # 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 - 0x00D0: 0xD0, # LATIN CAPITAL LETTER ETH - 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 - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xDE, # LATIN CAPITAL LETTER THORN - 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 - 0x00F0: 0xF0, # LATIN SMALL LETTER ETH - 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 - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0xFE, # LATIN SMALL LETTER THORN - 0x00FF: 0xFF, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x0152: 0x8C, # LATIN CAPITAL LIGATURE OE - 0x0153: 0x9C, # LATIN SMALL LIGATURE OE - 0x0160: 0x8A, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0x9A, # LATIN SMALL LETTER S WITH CARON - 0x0178: 0x9F, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x017D: 0x8E, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0x9E, # LATIN SMALL LETTER Z WITH CARON - 0x0192: 0x83, # LATIN SMALL LETTER F WITH HOOK - 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 - 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 - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1253.py b/Lib/encodings/cp1253.py index 3ce70b25..60e6663 100644 --- a/Lib/encodings/cp1253.py +++ b/Lib/encodings/cp1253.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,246 +303,6 @@ decoding_table = ( u'\ufffe' # 0xFF -> UNDEFINED ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 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 - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAC, # NOT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AE: 0xAE, # REGISTERED SIGN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B2: 0xB2, # SUPERSCRIPT TWO - 0x00B3: 0xB3, # SUPERSCRIPT THREE - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BD: 0xBD, # VULGAR FRACTION ONE HALF - 0x0192: 0x83, # LATIN SMALL LETTER F WITH HOOK - 0x0384: 0xB4, # GREEK TONOS - 0x0385: 0xA1, # GREEK DIALYTIKA TONOS - 0x0386: 0xA2, # GREEK CAPITAL LETTER ALPHA WITH TONOS - 0x0388: 0xB8, # GREEK CAPITAL LETTER EPSILON WITH TONOS - 0x0389: 0xB9, # GREEK CAPITAL LETTER ETA WITH TONOS - 0x038A: 0xBA, # GREEK CAPITAL LETTER IOTA WITH TONOS - 0x038C: 0xBC, # GREEK CAPITAL LETTER OMICRON WITH TONOS - 0x038E: 0xBE, # GREEK CAPITAL LETTER UPSILON WITH TONOS - 0x038F: 0xBF, # GREEK CAPITAL LETTER OMEGA WITH TONOS - 0x0390: 0xC0, # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - 0x0391: 0xC1, # GREEK CAPITAL LETTER ALPHA - 0x0392: 0xC2, # GREEK CAPITAL LETTER BETA - 0x0393: 0xC3, # GREEK CAPITAL LETTER GAMMA - 0x0394: 0xC4, # GREEK CAPITAL LETTER DELTA - 0x0395: 0xC5, # GREEK CAPITAL LETTER EPSILON - 0x0396: 0xC6, # GREEK CAPITAL LETTER ZETA - 0x0397: 0xC7, # GREEK CAPITAL LETTER ETA - 0x0398: 0xC8, # GREEK CAPITAL LETTER THETA - 0x0399: 0xC9, # GREEK CAPITAL LETTER IOTA - 0x039A: 0xCA, # GREEK CAPITAL LETTER KAPPA - 0x039B: 0xCB, # GREEK CAPITAL LETTER LAMDA - 0x039C: 0xCC, # GREEK CAPITAL LETTER MU - 0x039D: 0xCD, # GREEK CAPITAL LETTER NU - 0x039E: 0xCE, # GREEK CAPITAL LETTER XI - 0x039F: 0xCF, # GREEK CAPITAL LETTER OMICRON - 0x03A0: 0xD0, # GREEK CAPITAL LETTER PI - 0x03A1: 0xD1, # GREEK CAPITAL LETTER RHO - 0x03A3: 0xD3, # GREEK CAPITAL LETTER SIGMA - 0x03A4: 0xD4, # GREEK CAPITAL LETTER TAU - 0x03A5: 0xD5, # GREEK CAPITAL LETTER UPSILON - 0x03A6: 0xD6, # GREEK CAPITAL LETTER PHI - 0x03A7: 0xD7, # GREEK CAPITAL LETTER CHI - 0x03A8: 0xD8, # GREEK CAPITAL LETTER PSI - 0x03A9: 0xD9, # GREEK CAPITAL LETTER OMEGA - 0x03AA: 0xDA, # GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - 0x03AB: 0xDB, # GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - 0x03AC: 0xDC, # GREEK SMALL LETTER ALPHA WITH TONOS - 0x03AD: 0xDD, # GREEK SMALL LETTER EPSILON WITH TONOS - 0x03AE: 0xDE, # GREEK SMALL LETTER ETA WITH TONOS - 0x03AF: 0xDF, # GREEK SMALL LETTER IOTA WITH TONOS - 0x03B0: 0xE0, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - 0x03B1: 0xE1, # GREEK SMALL LETTER ALPHA - 0x03B2: 0xE2, # GREEK SMALL LETTER BETA - 0x03B3: 0xE3, # GREEK SMALL LETTER GAMMA - 0x03B4: 0xE4, # GREEK SMALL LETTER DELTA - 0x03B5: 0xE5, # GREEK SMALL LETTER EPSILON - 0x03B6: 0xE6, # GREEK SMALL LETTER ZETA - 0x03B7: 0xE7, # GREEK SMALL LETTER ETA - 0x03B8: 0xE8, # GREEK SMALL LETTER THETA - 0x03B9: 0xE9, # GREEK SMALL LETTER IOTA - 0x03BA: 0xEA, # GREEK SMALL LETTER KAPPA - 0x03BB: 0xEB, # GREEK SMALL LETTER LAMDA - 0x03BC: 0xEC, # GREEK SMALL LETTER MU - 0x03BD: 0xED, # GREEK SMALL LETTER NU - 0x03BE: 0xEE, # GREEK SMALL LETTER XI - 0x03BF: 0xEF, # GREEK SMALL LETTER OMICRON - 0x03C0: 0xF0, # GREEK SMALL LETTER PI - 0x03C1: 0xF1, # GREEK SMALL LETTER RHO - 0x03C2: 0xF2, # GREEK SMALL LETTER FINAL SIGMA - 0x03C3: 0xF3, # GREEK SMALL LETTER SIGMA - 0x03C4: 0xF4, # GREEK SMALL LETTER TAU - 0x03C5: 0xF5, # GREEK SMALL LETTER UPSILON - 0x03C6: 0xF6, # GREEK SMALL LETTER PHI - 0x03C7: 0xF7, # GREEK SMALL LETTER CHI - 0x03C8: 0xF8, # GREEK SMALL LETTER PSI - 0x03C9: 0xF9, # GREEK SMALL LETTER OMEGA - 0x03CA: 0xFA, # GREEK SMALL LETTER IOTA WITH DIALYTIKA - 0x03CB: 0xFB, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA - 0x03CC: 0xFC, # GREEK SMALL LETTER OMICRON WITH TONOS - 0x03CD: 0xFD, # GREEK SMALL LETTER UPSILON WITH TONOS - 0x03CE: 0xFE, # GREEK SMALL LETTER OMEGA WITH TONOS - 0x2013: 0x96, # EN DASH - 0x2014: 0x97, # EM DASH - 0x2015: 0xAF, # HORIZONTAL BAR - 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 - 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 - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1254.py b/Lib/encodings/cp1254.py index 31cd48c..04872c1 100644 --- a/Lib/encodings/cp1254.py +++ b/Lib/encodings/cp1254.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,256 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # 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 - 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 - 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 - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1255.py b/Lib/encodings/cp1255.py index 47c43ce..cb59e3f 100644 --- a/Lib/encodings/cp1255.py +++ b/Lib/encodings/cp1255.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,240 +303,6 @@ decoding_table = ( u'\ufffe' # 0xFF -> UNDEFINED ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A1: 0xA1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A5: 0xA5, # YEN SIGN - 0x00A6: 0xA6, # BROKEN BAR - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0xA8, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 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 - 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 - 0x00D7: 0xAA, # MULTIPLICATION SIGN - 0x00F7: 0xBA, # DIVISION SIGN - 0x0192: 0x83, # LATIN SMALL LETTER F WITH HOOK - 0x02C6: 0x88, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02DC: 0x98, # SMALL TILDE - 0x05B0: 0xC0, # HEBREW POINT SHEVA - 0x05B1: 0xC1, # HEBREW POINT HATAF SEGOL - 0x05B2: 0xC2, # HEBREW POINT HATAF PATAH - 0x05B3: 0xC3, # HEBREW POINT HATAF QAMATS - 0x05B4: 0xC4, # HEBREW POINT HIRIQ - 0x05B5: 0xC5, # HEBREW POINT TSERE - 0x05B6: 0xC6, # HEBREW POINT SEGOL - 0x05B7: 0xC7, # HEBREW POINT PATAH - 0x05B8: 0xC8, # HEBREW POINT QAMATS - 0x05B9: 0xC9, # HEBREW POINT HOLAM - 0x05BB: 0xCB, # HEBREW POINT QUBUTS - 0x05BC: 0xCC, # HEBREW POINT DAGESH OR MAPIQ - 0x05BD: 0xCD, # HEBREW POINT METEG - 0x05BE: 0xCE, # HEBREW PUNCTUATION MAQAF - 0x05BF: 0xCF, # HEBREW POINT RAFE - 0x05C0: 0xD0, # HEBREW PUNCTUATION PASEQ - 0x05C1: 0xD1, # HEBREW POINT SHIN DOT - 0x05C2: 0xD2, # HEBREW POINT SIN DOT - 0x05C3: 0xD3, # HEBREW PUNCTUATION SOF PASUQ - 0x05D0: 0xE0, # HEBREW LETTER ALEF - 0x05D1: 0xE1, # HEBREW LETTER BET - 0x05D2: 0xE2, # HEBREW LETTER GIMEL - 0x05D3: 0xE3, # HEBREW LETTER DALET - 0x05D4: 0xE4, # HEBREW LETTER HE - 0x05D5: 0xE5, # HEBREW LETTER VAV - 0x05D6: 0xE6, # HEBREW LETTER ZAYIN - 0x05D7: 0xE7, # HEBREW LETTER HET - 0x05D8: 0xE8, # HEBREW LETTER TET - 0x05D9: 0xE9, # HEBREW LETTER YOD - 0x05DA: 0xEA, # HEBREW LETTER FINAL KAF - 0x05DB: 0xEB, # HEBREW LETTER KAF - 0x05DC: 0xEC, # HEBREW LETTER LAMED - 0x05DD: 0xED, # HEBREW LETTER FINAL MEM - 0x05DE: 0xEE, # HEBREW LETTER MEM - 0x05DF: 0xEF, # HEBREW LETTER FINAL NUN - 0x05E0: 0xF0, # HEBREW LETTER NUN - 0x05E1: 0xF1, # HEBREW LETTER SAMEKH - 0x05E2: 0xF2, # HEBREW LETTER AYIN - 0x05E3: 0xF3, # HEBREW LETTER FINAL PE - 0x05E4: 0xF4, # HEBREW LETTER PE - 0x05E5: 0xF5, # HEBREW LETTER FINAL TSADI - 0x05E6: 0xF6, # HEBREW LETTER TSADI - 0x05E7: 0xF7, # HEBREW LETTER QOF - 0x05E8: 0xF8, # HEBREW LETTER RESH - 0x05E9: 0xF9, # HEBREW LETTER SHIN - 0x05EA: 0xFA, # HEBREW LETTER TAV - 0x05F0: 0xD4, # HEBREW LIGATURE YIDDISH DOUBLE VAV - 0x05F1: 0xD5, # HEBREW LIGATURE YIDDISH VAV YOD - 0x05F2: 0xD6, # HEBREW LIGATURE YIDDISH DOUBLE YOD - 0x05F3: 0xD7, # HEBREW PUNCTUATION GERESH - 0x05F4: 0xD8, # HEBREW PUNCTUATION GERSHAYIM - 0x200E: 0xFD, # LEFT-TO-RIGHT MARK - 0x200F: 0xFE, # RIGHT-TO-LEFT MARK - 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 - 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 - 0x20AA: 0xA4, # NEW SHEQEL SIGN - 0x20AC: 0x80, # EURO SIGN - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1256.py b/Lib/encodings/cp1256.py index e90393b..9ebb1df 100644 --- a/Lib/encodings/cp1256.py +++ b/Lib/encodings/cp1256.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u06d2' # 0xFF -> ARABIC LETTER YEH BARREE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 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 - 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 - 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 - 0x00D7: 0xD7, # MULTIPLICATION SIGN - 0x00E0: 0xE0, # LATIN SMALL LETTER A WITH GRAVE - 0x00E2: 0xE2, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 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 - 0x00EE: 0xEE, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0xEF, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F4: 0xF4, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F7: 0xF7, # DIVISION SIGN - 0x00F9: 0xF9, # LATIN SMALL LETTER U WITH GRAVE - 0x00FB: 0xFB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x0152: 0x8C, # LATIN CAPITAL LIGATURE OE - 0x0153: 0x9C, # LATIN SMALL LIGATURE OE - 0x0192: 0x83, # LATIN SMALL LETTER F WITH HOOK - 0x02C6: 0x88, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x060C: 0xA1, # ARABIC COMMA - 0x061B: 0xBA, # ARABIC SEMICOLON - 0x061F: 0xBF, # ARABIC QUESTION MARK - 0x0621: 0xC1, # ARABIC LETTER HAMZA - 0x0622: 0xC2, # ARABIC LETTER ALEF WITH MADDA ABOVE - 0x0623: 0xC3, # ARABIC LETTER ALEF WITH HAMZA ABOVE - 0x0624: 0xC4, # ARABIC LETTER WAW WITH HAMZA ABOVE - 0x0625: 0xC5, # ARABIC LETTER ALEF WITH HAMZA BELOW - 0x0626: 0xC6, # ARABIC LETTER YEH WITH HAMZA ABOVE - 0x0627: 0xC7, # ARABIC LETTER ALEF - 0x0628: 0xC8, # ARABIC LETTER BEH - 0x0629: 0xC9, # ARABIC LETTER TEH MARBUTA - 0x062A: 0xCA, # ARABIC LETTER TEH - 0x062B: 0xCB, # ARABIC LETTER THEH - 0x062C: 0xCC, # ARABIC LETTER JEEM - 0x062D: 0xCD, # ARABIC LETTER HAH - 0x062E: 0xCE, # ARABIC LETTER KHAH - 0x062F: 0xCF, # ARABIC LETTER DAL - 0x0630: 0xD0, # ARABIC LETTER THAL - 0x0631: 0xD1, # ARABIC LETTER REH - 0x0632: 0xD2, # ARABIC LETTER ZAIN - 0x0633: 0xD3, # ARABIC LETTER SEEN - 0x0634: 0xD4, # ARABIC LETTER SHEEN - 0x0635: 0xD5, # ARABIC LETTER SAD - 0x0636: 0xD6, # ARABIC LETTER DAD - 0x0637: 0xD8, # ARABIC LETTER TAH - 0x0638: 0xD9, # ARABIC LETTER ZAH - 0x0639: 0xDA, # ARABIC LETTER AIN - 0x063A: 0xDB, # ARABIC LETTER GHAIN - 0x0640: 0xDC, # ARABIC TATWEEL - 0x0641: 0xDD, # ARABIC LETTER FEH - 0x0642: 0xDE, # ARABIC LETTER QAF - 0x0643: 0xDF, # ARABIC LETTER KAF - 0x0644: 0xE1, # ARABIC LETTER LAM - 0x0645: 0xE3, # ARABIC LETTER MEEM - 0x0646: 0xE4, # ARABIC LETTER NOON - 0x0647: 0xE5, # ARABIC LETTER HEH - 0x0648: 0xE6, # ARABIC LETTER WAW - 0x0649: 0xEC, # ARABIC LETTER ALEF MAKSURA - 0x064A: 0xED, # ARABIC LETTER YEH - 0x064B: 0xF0, # ARABIC FATHATAN - 0x064C: 0xF1, # ARABIC DAMMATAN - 0x064D: 0xF2, # ARABIC KASRATAN - 0x064E: 0xF3, # ARABIC FATHA - 0x064F: 0xF5, # ARABIC DAMMA - 0x0650: 0xF6, # ARABIC KASRA - 0x0651: 0xF8, # ARABIC SHADDA - 0x0652: 0xFA, # ARABIC SUKUN - 0x0679: 0x8A, # ARABIC LETTER TTEH - 0x067E: 0x81, # ARABIC LETTER PEH - 0x0686: 0x8D, # ARABIC LETTER TCHEH - 0x0688: 0x8F, # ARABIC LETTER DDAL - 0x0691: 0x9A, # ARABIC LETTER RREH - 0x0698: 0x8E, # ARABIC LETTER JEH - 0x06A9: 0x98, # ARABIC LETTER KEHEH - 0x06AF: 0x90, # ARABIC LETTER GAF - 0x06BA: 0x9F, # ARABIC LETTER NOON GHUNNA - 0x06BE: 0xAA, # ARABIC LETTER HEH DOACHASHMEE - 0x06C1: 0xC0, # ARABIC LETTER HEH GOAL - 0x06D2: 0xFF, # ARABIC LETTER YEH BARREE - 0x200C: 0x9D, # ZERO WIDTH NON-JOINER - 0x200D: 0x9E, # ZERO WIDTH JOINER - 0x200E: 0xFD, # LEFT-TO-RIGHT MARK - 0x200F: 0xFE, # RIGHT-TO-LEFT MARK - 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 - 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 - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1257.py b/Lib/encodings/cp1257.py index dcc81c0..57a86ed 100644 --- a/Lib/encodings/cp1257.py +++ b/Lib/encodings/cp1257.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,251 +303,6 @@ decoding_table = ( u'\u02d9' # 0xFF -> DOT ABOVE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A6: 0xA6, # BROKEN BAR - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0x8D, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAC, # NOT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AE: 0xAE, # REGISTERED SIGN - 0x00AF: 0x9D, # 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: 0x8F, # CEDILLA - 0x00B9: 0xB9, # SUPERSCRIPT ONE - 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 - 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0xC5, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0xAF, # LATIN CAPITAL LETTER AE - 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D5: 0xD5, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xD7, # MULTIPLICATION SIGN - 0x00D8: 0xA8, # LATIN CAPITAL LETTER O WITH STROKE - 0x00DC: 0xDC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S - 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0xE5, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0xBF, # LATIN SMALL LETTER AE - 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE - 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE - 0x00F5: 0xF5, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xF7, # DIVISION SIGN - 0x00F8: 0xB8, # LATIN SMALL LETTER O WITH STROKE - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x0100: 0xC2, # LATIN CAPITAL LETTER A WITH MACRON - 0x0101: 0xE2, # LATIN SMALL LETTER A WITH MACRON - 0x0104: 0xC0, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xE0, # LATIN SMALL LETTER A WITH OGONEK - 0x0106: 0xC3, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0xE3, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x0112: 0xC7, # LATIN CAPITAL LETTER E WITH MACRON - 0x0113: 0xE7, # LATIN SMALL LETTER E WITH MACRON - 0x0116: 0xCB, # LATIN CAPITAL LETTER E WITH DOT ABOVE - 0x0117: 0xEB, # LATIN SMALL LETTER E WITH DOT ABOVE - 0x0118: 0xC6, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xE6, # LATIN SMALL LETTER E WITH OGONEK - 0x0122: 0xCC, # LATIN CAPITAL LETTER G WITH CEDILLA - 0x0123: 0xEC, # LATIN SMALL LETTER G WITH CEDILLA - 0x012A: 0xCE, # LATIN CAPITAL LETTER I WITH MACRON - 0x012B: 0xEE, # LATIN SMALL LETTER I WITH MACRON - 0x012E: 0xC1, # LATIN CAPITAL LETTER I WITH OGONEK - 0x012F: 0xE1, # LATIN SMALL LETTER I WITH OGONEK - 0x0136: 0xCD, # LATIN CAPITAL LETTER K WITH CEDILLA - 0x0137: 0xED, # LATIN SMALL LETTER K WITH CEDILLA - 0x013B: 0xCF, # LATIN CAPITAL LETTER L WITH CEDILLA - 0x013C: 0xEF, # LATIN SMALL LETTER L WITH CEDILLA - 0x0141: 0xD9, # LATIN CAPITAL LETTER L WITH STROKE - 0x0142: 0xF9, # LATIN SMALL LETTER L WITH STROKE - 0x0143: 0xD1, # LATIN CAPITAL LETTER N WITH ACUTE - 0x0144: 0xF1, # LATIN SMALL LETTER N WITH ACUTE - 0x0145: 0xD2, # LATIN CAPITAL LETTER N WITH CEDILLA - 0x0146: 0xF2, # LATIN SMALL LETTER N WITH CEDILLA - 0x014C: 0xD4, # LATIN CAPITAL LETTER O WITH MACRON - 0x014D: 0xF4, # LATIN SMALL LETTER O WITH MACRON - 0x0156: 0xAA, # LATIN CAPITAL LETTER R WITH CEDILLA - 0x0157: 0xBA, # LATIN SMALL LETTER R WITH CEDILLA - 0x015A: 0xDA, # LATIN CAPITAL LETTER S WITH ACUTE - 0x015B: 0xFA, # LATIN SMALL LETTER S WITH ACUTE - 0x0160: 0xD0, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xF0, # LATIN SMALL LETTER S WITH CARON - 0x016A: 0xDB, # LATIN CAPITAL LETTER U WITH MACRON - 0x016B: 0xFB, # LATIN SMALL LETTER U WITH MACRON - 0x0172: 0xD8, # LATIN CAPITAL LETTER U WITH OGONEK - 0x0173: 0xF8, # LATIN SMALL LETTER U WITH OGONEK - 0x0179: 0xCA, # LATIN CAPITAL LETTER Z WITH ACUTE - 0x017A: 0xEA, # LATIN SMALL LETTER Z WITH ACUTE - 0x017B: 0xDD, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xFD, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x017D: 0xDE, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xFE, # LATIN SMALL LETTER Z WITH CARON - 0x02C7: 0x8E, # CARON - 0x02D9: 0xFF, # DOT ABOVE - 0x02DB: 0x9E, # OGONEK - 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 - 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 - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp1258.py b/Lib/encodings/cp1258.py index d4d2271..bd1728c 100644 --- a/Lib/encodings/cp1258.py +++ b/Lib/encodings/cp1258.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,254 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # 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 - 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 - 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 - 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xD4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 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 - 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 - 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 - 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xF4, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 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 - 0x0102: 0xC3, # LATIN CAPITAL LETTER A WITH BREVE - 0x0103: 0xE3, # LATIN SMALL LETTER A WITH BREVE - 0x0110: 0xD0, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xF0, # LATIN SMALL LETTER D WITH STROKE - 0x0152: 0x8C, # LATIN CAPITAL LIGATURE OE - 0x0153: 0x9C, # LATIN SMALL LIGATURE OE - 0x0178: 0x9F, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x0192: 0x83, # LATIN SMALL LETTER F WITH HOOK - 0x01A0: 0xD5, # LATIN CAPITAL LETTER O WITH HORN - 0x01A1: 0xF5, # LATIN SMALL LETTER O WITH HORN - 0x01AF: 0xDD, # LATIN CAPITAL LETTER U WITH HORN - 0x01B0: 0xFD, # LATIN SMALL LETTER U WITH HORN - 0x02C6: 0x88, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02DC: 0x98, # SMALL TILDE - 0x0300: 0xCC, # COMBINING GRAVE ACCENT - 0x0301: 0xEC, # COMBINING ACUTE ACCENT - 0x0303: 0xDE, # COMBINING TILDE - 0x0309: 0xD2, # COMBINING HOOK ABOVE - 0x0323: 0xF2, # COMBINING DOT BELOW - 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 - 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 - 0x20AB: 0xFE, # DONG SIGN - 0x20AC: 0x80, # EURO SIGN - 0x2122: 0x99, # TRADE MARK SIGN -} diff --git a/Lib/encodings/cp424.py b/Lib/encodings/cp424.py index 966aecb..b83787e 100644 --- a/Lib/encodings/cp424.py +++ b/Lib/encodings/cp424.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,225 +303,6 @@ decoding_table = ( u'\x9f' # 0xFF -> EIGHT ONES ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x37, # END OF TRANSMISSION - 0x0005: 0x2D, # ENQUIRY - 0x0006: 0x2E, # ACKNOWLEDGE - 0x0007: 0x2F, # BELL - 0x0008: 0x16, # BACKSPACE - 0x0009: 0x05, # HORIZONTAL TABULATION - 0x000A: 0x25, # 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 - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x3C, # DEVICE CONTROL FOUR - 0x0015: 0x3D, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x32, # SYNCHRONOUS IDLE - 0x0017: 0x26, # END OF TRANSMISSION BLOCK - 0x0018: 0x18, # CANCEL - 0x0019: 0x19, # END OF MEDIUM - 0x001A: 0x3F, # SUBSTITUTE - 0x001B: 0x27, # ESCAPE - 0x001C: 0x1C, # FILE SEPARATOR - 0x001D: 0x1D, # GROUP SEPARATOR - 0x001E: 0x1E, # RECORD SEPARATOR - 0x001F: 0x1F, # UNIT SEPARATOR - 0x0020: 0x40, # SPACE - 0x0021: 0x5A, # EXCLAMATION MARK - 0x0022: 0x7F, # QUOTATION MARK - 0x0023: 0x7B, # NUMBER SIGN - 0x0024: 0x5B, # DOLLAR SIGN - 0x0025: 0x6C, # PERCENT SIGN - 0x0026: 0x50, # AMPERSAND - 0x0027: 0x7D, # APOSTROPHE - 0x0028: 0x4D, # LEFT PARENTHESIS - 0x0029: 0x5D, # RIGHT PARENTHESIS - 0x002A: 0x5C, # ASTERISK - 0x002B: 0x4E, # PLUS SIGN - 0x002C: 0x6B, # COMMA - 0x002D: 0x60, # HYPHEN-MINUS - 0x002E: 0x4B, # FULL STOP - 0x002F: 0x61, # SOLIDUS - 0x0030: 0xF0, # DIGIT ZERO - 0x0031: 0xF1, # DIGIT ONE - 0x0032: 0xF2, # DIGIT TWO - 0x0033: 0xF3, # DIGIT THREE - 0x0034: 0xF4, # DIGIT FOUR - 0x0035: 0xF5, # DIGIT FIVE - 0x0036: 0xF6, # DIGIT SIX - 0x0037: 0xF7, # DIGIT SEVEN - 0x0038: 0xF8, # DIGIT EIGHT - 0x0039: 0xF9, # DIGIT NINE - 0x003A: 0x7A, # COLON - 0x003B: 0x5E, # SEMICOLON - 0x003C: 0x4C, # LESS-THAN SIGN - 0x003D: 0x7E, # EQUALS SIGN - 0x003E: 0x6E, # GREATER-THAN SIGN - 0x003F: 0x6F, # QUESTION MARK - 0x0040: 0x7C, # COMMERCIAL AT - 0x0041: 0xC1, # LATIN CAPITAL LETTER A - 0x0042: 0xC2, # LATIN CAPITAL LETTER B - 0x0043: 0xC3, # LATIN CAPITAL LETTER C - 0x0044: 0xC4, # LATIN CAPITAL LETTER D - 0x0045: 0xC5, # LATIN CAPITAL LETTER E - 0x0046: 0xC6, # LATIN CAPITAL LETTER F - 0x0047: 0xC7, # LATIN CAPITAL LETTER G - 0x0048: 0xC8, # LATIN CAPITAL LETTER H - 0x0049: 0xC9, # LATIN CAPITAL LETTER I - 0x004A: 0xD1, # LATIN CAPITAL LETTER J - 0x004B: 0xD2, # LATIN CAPITAL LETTER K - 0x004C: 0xD3, # LATIN CAPITAL LETTER L - 0x004D: 0xD4, # LATIN CAPITAL LETTER M - 0x004E: 0xD5, # LATIN CAPITAL LETTER N - 0x004F: 0xD6, # LATIN CAPITAL LETTER O - 0x0050: 0xD7, # LATIN CAPITAL LETTER P - 0x0051: 0xD8, # LATIN CAPITAL LETTER Q - 0x0052: 0xD9, # LATIN CAPITAL LETTER R - 0x0053: 0xE2, # LATIN CAPITAL LETTER S - 0x0054: 0xE3, # LATIN CAPITAL LETTER T - 0x0055: 0xE4, # LATIN CAPITAL LETTER U - 0x0056: 0xE5, # LATIN CAPITAL LETTER V - 0x0057: 0xE6, # LATIN CAPITAL LETTER W - 0x0058: 0xE7, # LATIN CAPITAL LETTER X - 0x0059: 0xE8, # LATIN CAPITAL LETTER Y - 0x005A: 0xE9, # LATIN CAPITAL LETTER Z - 0x005B: 0xBA, # LEFT SQUARE BRACKET - 0x005C: 0xE0, # REVERSE SOLIDUS - 0x005D: 0xBB, # RIGHT SQUARE BRACKET - 0x005E: 0xB0, # CIRCUMFLEX ACCENT - 0x005F: 0x6D, # LOW LINE - 0x0060: 0x79, # GRAVE ACCENT - 0x0061: 0x81, # LATIN SMALL LETTER A - 0x0062: 0x82, # LATIN SMALL LETTER B - 0x0063: 0x83, # LATIN SMALL LETTER C - 0x0064: 0x84, # LATIN SMALL LETTER D - 0x0065: 0x85, # LATIN SMALL LETTER E - 0x0066: 0x86, # LATIN SMALL LETTER F - 0x0067: 0x87, # LATIN SMALL LETTER G - 0x0068: 0x88, # LATIN SMALL LETTER H - 0x0069: 0x89, # LATIN SMALL LETTER I - 0x006A: 0x91, # LATIN SMALL LETTER J - 0x006B: 0x92, # LATIN SMALL LETTER K - 0x006C: 0x93, # LATIN SMALL LETTER L - 0x006D: 0x94, # LATIN SMALL LETTER M - 0x006E: 0x95, # LATIN SMALL LETTER N - 0x006F: 0x96, # LATIN SMALL LETTER O - 0x0070: 0x97, # LATIN SMALL LETTER P - 0x0071: 0x98, # LATIN SMALL LETTER Q - 0x0072: 0x99, # LATIN SMALL LETTER R - 0x0073: 0xA2, # LATIN SMALL LETTER S - 0x0074: 0xA3, # LATIN SMALL LETTER T - 0x0075: 0xA4, # LATIN SMALL LETTER U - 0x0076: 0xA5, # LATIN SMALL LETTER V - 0x0077: 0xA6, # LATIN SMALL LETTER W - 0x0078: 0xA7, # LATIN SMALL LETTER X - 0x0079: 0xA8, # LATIN SMALL LETTER Y - 0x007A: 0xA9, # LATIN SMALL LETTER Z - 0x007B: 0xC0, # LEFT CURLY BRACKET - 0x007C: 0x4F, # VERTICAL LINE - 0x007D: 0xD0, # RIGHT CURLY BRACKET - 0x007E: 0xA1, # TILDE - 0x007F: 0x07, # DELETE - 0x0080: 0x20, # DIGIT SELECT - 0x0081: 0x21, # START OF SIGNIFICANCE - 0x0082: 0x22, # FIELD SEPARATOR - 0x0083: 0x23, # WORD UNDERSCORE - 0x0084: 0x24, # BYPASS OR INHIBIT PRESENTATION - 0x0085: 0x15, # NEW LINE - 0x0086: 0x06, # REQUIRED NEW LINE - 0x0087: 0x17, # PROGRAM OPERATOR COMMUNICATION - 0x0088: 0x28, # SET ATTRIBUTE - 0x0089: 0x29, # START FIELD EXTENDED - 0x008A: 0x2A, # SET MODE OR SWITCH - 0x008B: 0x2B, # CONTROL SEQUENCE PREFIX - 0x008C: 0x2C, # MODIFY FIELD ATTRIBUTE - 0x008D: 0x09, # SUPERSCRIPT - 0x008E: 0x0A, # REPEAT - 0x008F: 0x1B, # CUSTOMER USE ONE - 0x0090: 0x30, # <reserved> - 0x0091: 0x31, # <reserved> - 0x0092: 0x1A, # UNIT BACK SPACE - 0x0093: 0x33, # INDEX RETURN - 0x0094: 0x34, # PRESENTATION POSITION - 0x0095: 0x35, # TRANSPARENT - 0x0096: 0x36, # NUMERIC BACKSPACE - 0x0097: 0x08, # GRAPHIC ESCAPE - 0x0098: 0x38, # SUBSCRIPT - 0x0099: 0x39, # INDENT TABULATION - 0x009A: 0x3A, # REVERSE FORM FEED - 0x009B: 0x3B, # CUSTOMER USE THREE - 0x009C: 0x04, # SELECT - 0x009D: 0x14, # RESTORE/ENABLE PRESENTATION - 0x009E: 0x3E, # <reserved> - 0x009F: 0xFF, # EIGHT ONES - 0x00A0: 0x74, # NO-BREAK SPACE - 0x00A2: 0x4A, # CENT SIGN - 0x00A3: 0xB1, # POUND SIGN - 0x00A4: 0x9F, # CURRENCY SIGN - 0x00A5: 0xB2, # YEN SIGN - 0x00A6: 0x6A, # BROKEN BAR - 0x00A7: 0xB5, # SECTION SIGN - 0x00A8: 0xBD, # DIAERESIS - 0x00A9: 0xB4, # COPYRIGHT SIGN - 0x00AB: 0x8A, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0x5F, # NOT SIGN - 0x00AD: 0xCA, # SOFT HYPHEN - 0x00AE: 0xAF, # REGISTERED SIGN - 0x00AF: 0xBC, # MACRON - 0x00B0: 0x90, # DEGREE SIGN - 0x00B1: 0x8F, # PLUS-MINUS SIGN - 0x00B2: 0xEA, # SUPERSCRIPT TWO - 0x00B3: 0xFA, # SUPERSCRIPT THREE - 0x00B4: 0xBE, # ACUTE ACCENT - 0x00B5: 0xA0, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB3, # MIDDLE DOT - 0x00B8: 0x9D, # CEDILLA - 0x00B9: 0xDA, # SUPERSCRIPT ONE - 0x00BB: 0x8B, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BC: 0xB7, # VULGAR FRACTION ONE QUARTER - 0x00BD: 0xB8, # VULGAR FRACTION ONE HALF - 0x00BE: 0xB9, # VULGAR FRACTION THREE QUARTERS - 0x00D7: 0xBF, # MULTIPLICATION SIGN - 0x00F7: 0xE1, # DIVISION SIGN - 0x05D0: 0x41, # HEBREW LETTER ALEF - 0x05D1: 0x42, # HEBREW LETTER BET - 0x05D2: 0x43, # HEBREW LETTER GIMEL - 0x05D3: 0x44, # HEBREW LETTER DALET - 0x05D4: 0x45, # HEBREW LETTER HE - 0x05D5: 0x46, # HEBREW LETTER VAV - 0x05D6: 0x47, # HEBREW LETTER ZAYIN - 0x05D7: 0x48, # HEBREW LETTER HET - 0x05D8: 0x49, # HEBREW LETTER TET - 0x05D9: 0x51, # HEBREW LETTER YOD - 0x05DA: 0x52, # HEBREW LETTER FINAL KAF - 0x05DB: 0x53, # HEBREW LETTER KAF - 0x05DC: 0x54, # HEBREW LETTER LAMED - 0x05DD: 0x55, # HEBREW LETTER FINAL MEM - 0x05DE: 0x56, # HEBREW LETTER MEM - 0x05DF: 0x57, # HEBREW LETTER FINAL NUN - 0x05E0: 0x58, # HEBREW LETTER NUN - 0x05E1: 0x59, # HEBREW LETTER SAMEKH - 0x05E2: 0x62, # HEBREW LETTER AYIN - 0x05E3: 0x63, # HEBREW LETTER FINAL PE - 0x05E4: 0x64, # HEBREW LETTER PE - 0x05E5: 0x65, # HEBREW LETTER FINAL TSADI - 0x05E6: 0x66, # HEBREW LETTER TSADI - 0x05E7: 0x67, # HEBREW LETTER QOF - 0x05E8: 0x68, # HEBREW LETTER RESH - 0x05E9: 0x69, # HEBREW LETTER SHIN - 0x05EA: 0x71, # HEBREW LETTER TAV - 0x2017: 0x78, # DOUBLE LOW LINE -} diff --git a/Lib/encodings/cp500.py b/Lib/encodings/cp500.py index 83af090..0f48437 100644 --- a/Lib/encodings/cp500.py +++ b/Lib/encodings/cp500.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\x9f' # 0xFF -> CONTROL ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x37, # END OF TRANSMISSION - 0x0005: 0x2D, # ENQUIRY - 0x0006: 0x2E, # ACKNOWLEDGE - 0x0007: 0x2F, # BELL - 0x0008: 0x16, # BACKSPACE - 0x0009: 0x05, # HORIZONTAL TABULATION - 0x000A: 0x25, # 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 - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x3C, # DEVICE CONTROL FOUR - 0x0015: 0x3D, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x32, # SYNCHRONOUS IDLE - 0x0017: 0x26, # END OF TRANSMISSION BLOCK - 0x0018: 0x18, # CANCEL - 0x0019: 0x19, # END OF MEDIUM - 0x001A: 0x3F, # SUBSTITUTE - 0x001B: 0x27, # ESCAPE - 0x001C: 0x1C, # FILE SEPARATOR - 0x001D: 0x1D, # GROUP SEPARATOR - 0x001E: 0x1E, # RECORD SEPARATOR - 0x001F: 0x1F, # UNIT SEPARATOR - 0x0020: 0x40, # SPACE - 0x0021: 0x4F, # EXCLAMATION MARK - 0x0022: 0x7F, # QUOTATION MARK - 0x0023: 0x7B, # NUMBER SIGN - 0x0024: 0x5B, # DOLLAR SIGN - 0x0025: 0x6C, # PERCENT SIGN - 0x0026: 0x50, # AMPERSAND - 0x0027: 0x7D, # APOSTROPHE - 0x0028: 0x4D, # LEFT PARENTHESIS - 0x0029: 0x5D, # RIGHT PARENTHESIS - 0x002A: 0x5C, # ASTERISK - 0x002B: 0x4E, # PLUS SIGN - 0x002C: 0x6B, # COMMA - 0x002D: 0x60, # HYPHEN-MINUS - 0x002E: 0x4B, # FULL STOP - 0x002F: 0x61, # SOLIDUS - 0x0030: 0xF0, # DIGIT ZERO - 0x0031: 0xF1, # DIGIT ONE - 0x0032: 0xF2, # DIGIT TWO - 0x0033: 0xF3, # DIGIT THREE - 0x0034: 0xF4, # DIGIT FOUR - 0x0035: 0xF5, # DIGIT FIVE - 0x0036: 0xF6, # DIGIT SIX - 0x0037: 0xF7, # DIGIT SEVEN - 0x0038: 0xF8, # DIGIT EIGHT - 0x0039: 0xF9, # DIGIT NINE - 0x003A: 0x7A, # COLON - 0x003B: 0x5E, # SEMICOLON - 0x003C: 0x4C, # LESS-THAN SIGN - 0x003D: 0x7E, # EQUALS SIGN - 0x003E: 0x6E, # GREATER-THAN SIGN - 0x003F: 0x6F, # QUESTION MARK - 0x0040: 0x7C, # COMMERCIAL AT - 0x0041: 0xC1, # LATIN CAPITAL LETTER A - 0x0042: 0xC2, # LATIN CAPITAL LETTER B - 0x0043: 0xC3, # LATIN CAPITAL LETTER C - 0x0044: 0xC4, # LATIN CAPITAL LETTER D - 0x0045: 0xC5, # LATIN CAPITAL LETTER E - 0x0046: 0xC6, # LATIN CAPITAL LETTER F - 0x0047: 0xC7, # LATIN CAPITAL LETTER G - 0x0048: 0xC8, # LATIN CAPITAL LETTER H - 0x0049: 0xC9, # LATIN CAPITAL LETTER I - 0x004A: 0xD1, # LATIN CAPITAL LETTER J - 0x004B: 0xD2, # LATIN CAPITAL LETTER K - 0x004C: 0xD3, # LATIN CAPITAL LETTER L - 0x004D: 0xD4, # LATIN CAPITAL LETTER M - 0x004E: 0xD5, # LATIN CAPITAL LETTER N - 0x004F: 0xD6, # LATIN CAPITAL LETTER O - 0x0050: 0xD7, # LATIN CAPITAL LETTER P - 0x0051: 0xD8, # LATIN CAPITAL LETTER Q - 0x0052: 0xD9, # LATIN CAPITAL LETTER R - 0x0053: 0xE2, # LATIN CAPITAL LETTER S - 0x0054: 0xE3, # LATIN CAPITAL LETTER T - 0x0055: 0xE4, # LATIN CAPITAL LETTER U - 0x0056: 0xE5, # LATIN CAPITAL LETTER V - 0x0057: 0xE6, # LATIN CAPITAL LETTER W - 0x0058: 0xE7, # LATIN CAPITAL LETTER X - 0x0059: 0xE8, # LATIN CAPITAL LETTER Y - 0x005A: 0xE9, # LATIN CAPITAL LETTER Z - 0x005B: 0x4A, # LEFT SQUARE BRACKET - 0x005C: 0xE0, # REVERSE SOLIDUS - 0x005D: 0x5A, # RIGHT SQUARE BRACKET - 0x005E: 0x5F, # CIRCUMFLEX ACCENT - 0x005F: 0x6D, # LOW LINE - 0x0060: 0x79, # GRAVE ACCENT - 0x0061: 0x81, # LATIN SMALL LETTER A - 0x0062: 0x82, # LATIN SMALL LETTER B - 0x0063: 0x83, # LATIN SMALL LETTER C - 0x0064: 0x84, # LATIN SMALL LETTER D - 0x0065: 0x85, # LATIN SMALL LETTER E - 0x0066: 0x86, # LATIN SMALL LETTER F - 0x0067: 0x87, # LATIN SMALL LETTER G - 0x0068: 0x88, # LATIN SMALL LETTER H - 0x0069: 0x89, # LATIN SMALL LETTER I - 0x006A: 0x91, # LATIN SMALL LETTER J - 0x006B: 0x92, # LATIN SMALL LETTER K - 0x006C: 0x93, # LATIN SMALL LETTER L - 0x006D: 0x94, # LATIN SMALL LETTER M - 0x006E: 0x95, # LATIN SMALL LETTER N - 0x006F: 0x96, # LATIN SMALL LETTER O - 0x0070: 0x97, # LATIN SMALL LETTER P - 0x0071: 0x98, # LATIN SMALL LETTER Q - 0x0072: 0x99, # LATIN SMALL LETTER R - 0x0073: 0xA2, # LATIN SMALL LETTER S - 0x0074: 0xA3, # LATIN SMALL LETTER T - 0x0075: 0xA4, # LATIN SMALL LETTER U - 0x0076: 0xA5, # LATIN SMALL LETTER V - 0x0077: 0xA6, # LATIN SMALL LETTER W - 0x0078: 0xA7, # LATIN SMALL LETTER X - 0x0079: 0xA8, # LATIN SMALL LETTER Y - 0x007A: 0xA9, # LATIN SMALL LETTER Z - 0x007B: 0xC0, # LEFT CURLY BRACKET - 0x007C: 0xBB, # VERTICAL LINE - 0x007D: 0xD0, # RIGHT CURLY BRACKET - 0x007E: 0xA1, # TILDE - 0x007F: 0x07, # DELETE - 0x0080: 0x20, # CONTROL - 0x0081: 0x21, # CONTROL - 0x0082: 0x22, # CONTROL - 0x0083: 0x23, # CONTROL - 0x0084: 0x24, # CONTROL - 0x0085: 0x15, # CONTROL - 0x0086: 0x06, # CONTROL - 0x0087: 0x17, # CONTROL - 0x0088: 0x28, # CONTROL - 0x0089: 0x29, # CONTROL - 0x008A: 0x2A, # CONTROL - 0x008B: 0x2B, # CONTROL - 0x008C: 0x2C, # CONTROL - 0x008D: 0x09, # CONTROL - 0x008E: 0x0A, # CONTROL - 0x008F: 0x1B, # CONTROL - 0x0090: 0x30, # CONTROL - 0x0091: 0x31, # CONTROL - 0x0092: 0x1A, # CONTROL - 0x0093: 0x33, # CONTROL - 0x0094: 0x34, # CONTROL - 0x0095: 0x35, # CONTROL - 0x0096: 0x36, # CONTROL - 0x0097: 0x08, # CONTROL - 0x0098: 0x38, # CONTROL - 0x0099: 0x39, # CONTROL - 0x009A: 0x3A, # CONTROL - 0x009B: 0x3B, # CONTROL - 0x009C: 0x04, # CONTROL - 0x009D: 0x14, # CONTROL - 0x009E: 0x3E, # CONTROL - 0x009F: 0xFF, # CONTROL - 0x00A0: 0x41, # NO-BREAK SPACE - 0x00A1: 0xAA, # INVERTED EXCLAMATION MARK - 0x00A2: 0xB0, # CENT SIGN - 0x00A3: 0xB1, # POUND SIGN - 0x00A4: 0x9F, # CURRENCY SIGN - 0x00A5: 0xB2, # YEN SIGN - 0x00A6: 0x6A, # BROKEN BAR - 0x00A7: 0xB5, # SECTION SIGN - 0x00A8: 0xBD, # DIAERESIS - 0x00A9: 0xB4, # COPYRIGHT SIGN - 0x00AA: 0x9A, # FEMININE ORDINAL INDICATOR - 0x00AB: 0x8A, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xBA, # NOT SIGN - 0x00AD: 0xCA, # SOFT HYPHEN - 0x00AE: 0xAF, # REGISTERED SIGN - 0x00AF: 0xBC, # MACRON - 0x00B0: 0x90, # DEGREE SIGN - 0x00B1: 0x8F, # PLUS-MINUS SIGN - 0x00B2: 0xEA, # SUPERSCRIPT TWO - 0x00B3: 0xFA, # SUPERSCRIPT THREE - 0x00B4: 0xBE, # ACUTE ACCENT - 0x00B5: 0xA0, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB3, # MIDDLE DOT - 0x00B8: 0x9D, # CEDILLA - 0x00B9: 0xDA, # SUPERSCRIPT ONE - 0x00BA: 0x9B, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0x8B, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BC: 0xB7, # VULGAR FRACTION ONE QUARTER - 0x00BD: 0xB8, # VULGAR FRACTION ONE HALF - 0x00BE: 0xB9, # VULGAR FRACTION THREE QUARTERS - 0x00BF: 0xAB, # INVERTED QUESTION MARK - 0x00C0: 0x64, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0x65, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0x62, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0x66, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x63, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x67, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0x9E, # LATIN CAPITAL LIGATURE AE - 0x00C7: 0x68, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0x74, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x71, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0x72, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0x73, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0x78, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0x75, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0x76, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0x77, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D0: 0xAC, # LATIN CAPITAL LETTER ETH (ICELANDIC) - 0x00D1: 0x69, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xED, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEB, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xEF, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0xEC, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xBF, # MULTIPLICATION SIGN - 0x00D8: 0x80, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xFD, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xFE, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xFB, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0xFC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xAD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xAE, # LATIN CAPITAL LETTER THORN (ICELANDIC) - 0x00DF: 0x59, # LATIN SMALL LETTER SHARP S (GERMAN) - 0x00E0: 0x44, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x45, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x42, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x46, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x43, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x47, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0x9C, # LATIN SMALL LIGATURE AE - 0x00E7: 0x48, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x54, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x51, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x52, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x53, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x58, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x55, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x56, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x57, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F0: 0x8C, # LATIN SMALL LETTER ETH (ICELANDIC) - 0x00F1: 0x49, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0xCD, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0xCE, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xCB, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0xCF, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0xCC, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xE1, # DIVISION SIGN - 0x00F8: 0x70, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0xDD, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0xDE, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0xDB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xDC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0x8D, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0x8E, # LATIN SMALL LETTER THORN (ICELANDIC) - 0x00FF: 0xDF, # LATIN SMALL LETTER Y WITH DIAERESIS -} diff --git a/Lib/encodings/cp856.py b/Lib/encodings/cp856.py index c72fcad..ef017e4 100644 --- a/Lib/encodings/cp856.py +++ b/Lib/encodings/cp856.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,222 +303,6 @@ decoding_table = ( u'\xa0' # 0xFF -> NO-BREAK SPACE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xFF, # NO-BREAK SPACE - 0x00A2: 0xBD, # CENT SIGN - 0x00A3: 0x9C, # POUND SIGN - 0x00A4: 0xCF, # CURRENCY SIGN - 0x00A5: 0xBE, # YEN SIGN - 0x00A6: 0xDD, # BROKEN BAR - 0x00A7: 0xF5, # SECTION SIGN - 0x00A8: 0xF9, # DIAERESIS - 0x00A9: 0xB8, # COPYRIGHT SIGN - 0x00AB: 0xAE, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAA, # NOT SIGN - 0x00AD: 0xF0, # SOFT HYPHEN - 0x00AE: 0xA9, # REGISTERED SIGN - 0x00AF: 0xEE, # MACRON - 0x00B0: 0xF8, # DEGREE SIGN - 0x00B1: 0xF1, # PLUS-MINUS SIGN - 0x00B2: 0xFD, # SUPERSCRIPT TWO - 0x00B3: 0xFC, # SUPERSCRIPT THREE - 0x00B4: 0xEF, # ACUTE ACCENT - 0x00B5: 0xE6, # MICRO SIGN - 0x00B6: 0xF4, # PILCROW SIGN - 0x00B7: 0xFA, # MIDDLE DOT - 0x00B8: 0xF7, # CEDILLA - 0x00B9: 0xFB, # SUPERSCRIPT ONE - 0x00BB: 0xAF, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BC: 0xAC, # VULGAR FRACTION ONE QUARTER - 0x00BD: 0xAB, # VULGAR FRACTION ONE HALF - 0x00BE: 0xF3, # VULGAR FRACTION THREE QUARTERS - 0x00D7: 0x9E, # MULTIPLICATION SIGN - 0x00F7: 0xF6, # DIVISION SIGN - 0x05D0: 0x80, # HEBREW LETTER ALEF - 0x05D1: 0x81, # HEBREW LETTER BET - 0x05D2: 0x82, # HEBREW LETTER GIMEL - 0x05D3: 0x83, # HEBREW LETTER DALET - 0x05D4: 0x84, # HEBREW LETTER HE - 0x05D5: 0x85, # HEBREW LETTER VAV - 0x05D6: 0x86, # HEBREW LETTER ZAYIN - 0x05D7: 0x87, # HEBREW LETTER HET - 0x05D8: 0x88, # HEBREW LETTER TET - 0x05D9: 0x89, # HEBREW LETTER YOD - 0x05DA: 0x8A, # HEBREW LETTER FINAL KAF - 0x05DB: 0x8B, # HEBREW LETTER KAF - 0x05DC: 0x8C, # HEBREW LETTER LAMED - 0x05DD: 0x8D, # HEBREW LETTER FINAL MEM - 0x05DE: 0x8E, # HEBREW LETTER MEM - 0x05DF: 0x8F, # HEBREW LETTER FINAL NUN - 0x05E0: 0x90, # HEBREW LETTER NUN - 0x05E1: 0x91, # HEBREW LETTER SAMEKH - 0x05E2: 0x92, # HEBREW LETTER AYIN - 0x05E3: 0x93, # HEBREW LETTER FINAL PE - 0x05E4: 0x94, # HEBREW LETTER PE - 0x05E5: 0x95, # HEBREW LETTER FINAL TSADI - 0x05E6: 0x96, # HEBREW LETTER TSADI - 0x05E7: 0x97, # HEBREW LETTER QOF - 0x05E8: 0x98, # HEBREW LETTER RESH - 0x05E9: 0x99, # HEBREW LETTER SHIN - 0x05EA: 0x9A, # HEBREW LETTER TAV - 0x2017: 0xF2, # DOUBLE LOW LINE - 0x2500: 0xC4, # BOX DRAWINGS LIGHT HORIZONTAL - 0x2502: 0xB3, # BOX DRAWINGS LIGHT VERTICAL - 0x250C: 0xDA, # BOX DRAWINGS LIGHT DOWN AND RIGHT - 0x2510: 0xBF, # BOX DRAWINGS LIGHT DOWN AND LEFT - 0x2514: 0xC0, # BOX DRAWINGS LIGHT UP AND RIGHT - 0x2518: 0xD9, # BOX DRAWINGS LIGHT UP AND LEFT - 0x251C: 0xC3, # BOX DRAWINGS LIGHT VERTICAL AND RIGHT - 0x2524: 0xB4, # BOX DRAWINGS LIGHT VERTICAL AND LEFT - 0x252C: 0xC2, # BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - 0x2534: 0xC1, # BOX DRAWINGS LIGHT UP AND HORIZONTAL - 0x253C: 0xC5, # BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - 0x2550: 0xCD, # BOX DRAWINGS DOUBLE HORIZONTAL - 0x2551: 0xBA, # BOX DRAWINGS DOUBLE VERTICAL - 0x2554: 0xC9, # BOX DRAWINGS DOUBLE DOWN AND RIGHT - 0x2557: 0xBB, # BOX DRAWINGS DOUBLE DOWN AND LEFT - 0x255A: 0xC8, # BOX DRAWINGS DOUBLE UP AND RIGHT - 0x255D: 0xBC, # BOX DRAWINGS DOUBLE UP AND LEFT - 0x2560: 0xCC, # BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - 0x2563: 0xB9, # BOX DRAWINGS DOUBLE VERTICAL AND LEFT - 0x2566: 0xCB, # BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - 0x2569: 0xCA, # BOX DRAWINGS DOUBLE UP AND HORIZONTAL - 0x256C: 0xCE, # BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - 0x2580: 0xDF, # UPPER HALF BLOCK - 0x2584: 0xDC, # LOWER HALF BLOCK - 0x2588: 0xDB, # FULL BLOCK - 0x2591: 0xB0, # LIGHT SHADE - 0x2592: 0xB1, # MEDIUM SHADE - 0x2593: 0xB2, # DARK SHADE - 0x25A0: 0xFE, # BLACK SQUARE -} diff --git a/Lib/encodings/cp874.py b/Lib/encodings/cp874.py index 591e8aa..684d3bf 100644 --- a/Lib/encodings/cp874.py +++ b/Lib/encodings/cp874.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,232 +303,6 @@ decoding_table = ( u'\ufffe' # 0xFF -> UNDEFINED ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x0E01: 0xA1, # THAI CHARACTER KO KAI - 0x0E02: 0xA2, # THAI CHARACTER KHO KHAI - 0x0E03: 0xA3, # THAI CHARACTER KHO KHUAT - 0x0E04: 0xA4, # THAI CHARACTER KHO KHWAI - 0x0E05: 0xA5, # THAI CHARACTER KHO KHON - 0x0E06: 0xA6, # THAI CHARACTER KHO RAKHANG - 0x0E07: 0xA7, # THAI CHARACTER NGO NGU - 0x0E08: 0xA8, # THAI CHARACTER CHO CHAN - 0x0E09: 0xA9, # THAI CHARACTER CHO CHING - 0x0E0A: 0xAA, # THAI CHARACTER CHO CHANG - 0x0E0B: 0xAB, # THAI CHARACTER SO SO - 0x0E0C: 0xAC, # THAI CHARACTER CHO CHOE - 0x0E0D: 0xAD, # THAI CHARACTER YO YING - 0x0E0E: 0xAE, # THAI CHARACTER DO CHADA - 0x0E0F: 0xAF, # THAI CHARACTER TO PATAK - 0x0E10: 0xB0, # THAI CHARACTER THO THAN - 0x0E11: 0xB1, # THAI CHARACTER THO NANGMONTHO - 0x0E12: 0xB2, # THAI CHARACTER THO PHUTHAO - 0x0E13: 0xB3, # THAI CHARACTER NO NEN - 0x0E14: 0xB4, # THAI CHARACTER DO DEK - 0x0E15: 0xB5, # THAI CHARACTER TO TAO - 0x0E16: 0xB6, # THAI CHARACTER THO THUNG - 0x0E17: 0xB7, # THAI CHARACTER THO THAHAN - 0x0E18: 0xB8, # THAI CHARACTER THO THONG - 0x0E19: 0xB9, # THAI CHARACTER NO NU - 0x0E1A: 0xBA, # THAI CHARACTER BO BAIMAI - 0x0E1B: 0xBB, # THAI CHARACTER PO PLA - 0x0E1C: 0xBC, # THAI CHARACTER PHO PHUNG - 0x0E1D: 0xBD, # THAI CHARACTER FO FA - 0x0E1E: 0xBE, # THAI CHARACTER PHO PHAN - 0x0E1F: 0xBF, # THAI CHARACTER FO FAN - 0x0E20: 0xC0, # THAI CHARACTER PHO SAMPHAO - 0x0E21: 0xC1, # THAI CHARACTER MO MA - 0x0E22: 0xC2, # THAI CHARACTER YO YAK - 0x0E23: 0xC3, # THAI CHARACTER RO RUA - 0x0E24: 0xC4, # THAI CHARACTER RU - 0x0E25: 0xC5, # THAI CHARACTER LO LING - 0x0E26: 0xC6, # THAI CHARACTER LU - 0x0E27: 0xC7, # THAI CHARACTER WO WAEN - 0x0E28: 0xC8, # THAI CHARACTER SO SALA - 0x0E29: 0xC9, # THAI CHARACTER SO RUSI - 0x0E2A: 0xCA, # THAI CHARACTER SO SUA - 0x0E2B: 0xCB, # THAI CHARACTER HO HIP - 0x0E2C: 0xCC, # THAI CHARACTER LO CHULA - 0x0E2D: 0xCD, # THAI CHARACTER O ANG - 0x0E2E: 0xCE, # THAI CHARACTER HO NOKHUK - 0x0E2F: 0xCF, # THAI CHARACTER PAIYANNOI - 0x0E30: 0xD0, # THAI CHARACTER SARA A - 0x0E31: 0xD1, # THAI CHARACTER MAI HAN-AKAT - 0x0E32: 0xD2, # THAI CHARACTER SARA AA - 0x0E33: 0xD3, # THAI CHARACTER SARA AM - 0x0E34: 0xD4, # THAI CHARACTER SARA I - 0x0E35: 0xD5, # THAI CHARACTER SARA II - 0x0E36: 0xD6, # THAI CHARACTER SARA UE - 0x0E37: 0xD7, # THAI CHARACTER SARA UEE - 0x0E38: 0xD8, # THAI CHARACTER SARA U - 0x0E39: 0xD9, # THAI CHARACTER SARA UU - 0x0E3A: 0xDA, # THAI CHARACTER PHINTHU - 0x0E3F: 0xDF, # THAI CURRENCY SYMBOL BAHT - 0x0E40: 0xE0, # THAI CHARACTER SARA E - 0x0E41: 0xE1, # THAI CHARACTER SARA AE - 0x0E42: 0xE2, # THAI CHARACTER SARA O - 0x0E43: 0xE3, # THAI CHARACTER SARA AI MAIMUAN - 0x0E44: 0xE4, # THAI CHARACTER SARA AI MAIMALAI - 0x0E45: 0xE5, # THAI CHARACTER LAKKHANGYAO - 0x0E46: 0xE6, # THAI CHARACTER MAIYAMOK - 0x0E47: 0xE7, # THAI CHARACTER MAITAIKHU - 0x0E48: 0xE8, # THAI CHARACTER MAI EK - 0x0E49: 0xE9, # THAI CHARACTER MAI THO - 0x0E4A: 0xEA, # THAI CHARACTER MAI TRI - 0x0E4B: 0xEB, # THAI CHARACTER MAI CHATTAWA - 0x0E4C: 0xEC, # THAI CHARACTER THANTHAKHAT - 0x0E4D: 0xED, # THAI CHARACTER NIKHAHIT - 0x0E4E: 0xEE, # THAI CHARACTER YAMAKKAN - 0x0E4F: 0xEF, # THAI CHARACTER FONGMAN - 0x0E50: 0xF0, # THAI DIGIT ZERO - 0x0E51: 0xF1, # THAI DIGIT ONE - 0x0E52: 0xF2, # THAI DIGIT TWO - 0x0E53: 0xF3, # THAI DIGIT THREE - 0x0E54: 0xF4, # THAI DIGIT FOUR - 0x0E55: 0xF5, # THAI DIGIT FIVE - 0x0E56: 0xF6, # THAI DIGIT SIX - 0x0E57: 0xF7, # THAI DIGIT SEVEN - 0x0E58: 0xF8, # THAI DIGIT EIGHT - 0x0E59: 0xF9, # THAI DIGIT NINE - 0x0E5A: 0xFA, # THAI CHARACTER ANGKHANKHU - 0x0E5B: 0xFB, # THAI CHARACTER KHOMUT - 0x2013: 0x96, # EN DASH - 0x2014: 0x97, # EM DASH - 0x2018: 0x91, # LEFT SINGLE QUOTATION MARK - 0x2019: 0x92, # RIGHT SINGLE QUOTATION MARK - 0x201C: 0x93, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0x94, # RIGHT DOUBLE QUOTATION MARK - 0x2022: 0x95, # BULLET - 0x2026: 0x85, # HORIZONTAL ELLIPSIS - 0x20AC: 0x80, # EURO SIGN -} diff --git a/Lib/encodings/cp875.py b/Lib/encodings/cp875.py index 879d5a4..07e1c08 100644 --- a/Lib/encodings/cp875.py +++ b/Lib/encodings/cp875.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,257 +303,6 @@ decoding_table = ( u'\x9f' # 0xFF -> CONTROL ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x37, # END OF TRANSMISSION - 0x0005: 0x2D, # ENQUIRY - 0x0006: 0x2E, # ACKNOWLEDGE - 0x0007: 0x2F, # BELL - 0x0008: 0x16, # BACKSPACE - 0x0009: 0x05, # HORIZONTAL TABULATION - 0x000A: 0x25, # 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 - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x3C, # DEVICE CONTROL FOUR - 0x0015: 0x3D, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x32, # SYNCHRONOUS IDLE - 0x0017: 0x26, # END OF TRANSMISSION BLOCK - 0x0018: 0x18, # CANCEL - 0x0019: 0x19, # END OF MEDIUM - 0x001A: None, # SUBSTITUTE - 0x001B: 0x27, # ESCAPE - 0x001C: 0x1C, # FILE SEPARATOR - 0x001D: 0x1D, # GROUP SEPARATOR - 0x001E: 0x1E, # RECORD SEPARATOR - 0x001F: 0x1F, # UNIT SEPARATOR - 0x0020: 0x40, # SPACE - 0x0021: 0x4F, # EXCLAMATION MARK - 0x0022: 0x7F, # QUOTATION MARK - 0x0023: 0x7B, # NUMBER SIGN - 0x0024: 0x5B, # DOLLAR SIGN - 0x0025: 0x6C, # PERCENT SIGN - 0x0026: 0x50, # AMPERSAND - 0x0027: 0x7D, # APOSTROPHE - 0x0028: 0x4D, # LEFT PARENTHESIS - 0x0029: 0x5D, # RIGHT PARENTHESIS - 0x002A: 0x5C, # ASTERISK - 0x002B: 0x4E, # PLUS SIGN - 0x002C: 0x6B, # COMMA - 0x002D: 0x60, # HYPHEN-MINUS - 0x002E: 0x4B, # FULL STOP - 0x002F: 0x61, # SOLIDUS - 0x0030: 0xF0, # DIGIT ZERO - 0x0031: 0xF1, # DIGIT ONE - 0x0032: 0xF2, # DIGIT TWO - 0x0033: 0xF3, # DIGIT THREE - 0x0034: 0xF4, # DIGIT FOUR - 0x0035: 0xF5, # DIGIT FIVE - 0x0036: 0xF6, # DIGIT SIX - 0x0037: 0xF7, # DIGIT SEVEN - 0x0038: 0xF8, # DIGIT EIGHT - 0x0039: 0xF9, # DIGIT NINE - 0x003A: 0x7A, # COLON - 0x003B: 0x5E, # SEMICOLON - 0x003C: 0x4C, # LESS-THAN SIGN - 0x003D: 0x7E, # EQUALS SIGN - 0x003E: 0x6E, # GREATER-THAN SIGN - 0x003F: 0x6F, # QUESTION MARK - 0x0040: 0x7C, # COMMERCIAL AT - 0x0041: 0xC1, # LATIN CAPITAL LETTER A - 0x0042: 0xC2, # LATIN CAPITAL LETTER B - 0x0043: 0xC3, # LATIN CAPITAL LETTER C - 0x0044: 0xC4, # LATIN CAPITAL LETTER D - 0x0045: 0xC5, # LATIN CAPITAL LETTER E - 0x0046: 0xC6, # LATIN CAPITAL LETTER F - 0x0047: 0xC7, # LATIN CAPITAL LETTER G - 0x0048: 0xC8, # LATIN CAPITAL LETTER H - 0x0049: 0xC9, # LATIN CAPITAL LETTER I - 0x004A: 0xD1, # LATIN CAPITAL LETTER J - 0x004B: 0xD2, # LATIN CAPITAL LETTER K - 0x004C: 0xD3, # LATIN CAPITAL LETTER L - 0x004D: 0xD4, # LATIN CAPITAL LETTER M - 0x004E: 0xD5, # LATIN CAPITAL LETTER N - 0x004F: 0xD6, # LATIN CAPITAL LETTER O - 0x0050: 0xD7, # LATIN CAPITAL LETTER P - 0x0051: 0xD8, # LATIN CAPITAL LETTER Q - 0x0052: 0xD9, # LATIN CAPITAL LETTER R - 0x0053: 0xE2, # LATIN CAPITAL LETTER S - 0x0054: 0xE3, # LATIN CAPITAL LETTER T - 0x0055: 0xE4, # LATIN CAPITAL LETTER U - 0x0056: 0xE5, # LATIN CAPITAL LETTER V - 0x0057: 0xE6, # LATIN CAPITAL LETTER W - 0x0058: 0xE7, # LATIN CAPITAL LETTER X - 0x0059: 0xE8, # LATIN CAPITAL LETTER Y - 0x005A: 0xE9, # LATIN CAPITAL LETTER Z - 0x005B: 0x4A, # LEFT SQUARE BRACKET - 0x005C: 0xE0, # REVERSE SOLIDUS - 0x005D: 0x5A, # RIGHT SQUARE BRACKET - 0x005E: 0x5F, # CIRCUMFLEX ACCENT - 0x005F: 0x6D, # LOW LINE - 0x0060: 0x79, # GRAVE ACCENT - 0x0061: 0x81, # LATIN SMALL LETTER A - 0x0062: 0x82, # LATIN SMALL LETTER B - 0x0063: 0x83, # LATIN SMALL LETTER C - 0x0064: 0x84, # LATIN SMALL LETTER D - 0x0065: 0x85, # LATIN SMALL LETTER E - 0x0066: 0x86, # LATIN SMALL LETTER F - 0x0067: 0x87, # LATIN SMALL LETTER G - 0x0068: 0x88, # LATIN SMALL LETTER H - 0x0069: 0x89, # LATIN SMALL LETTER I - 0x006A: 0x91, # LATIN SMALL LETTER J - 0x006B: 0x92, # LATIN SMALL LETTER K - 0x006C: 0x93, # LATIN SMALL LETTER L - 0x006D: 0x94, # LATIN SMALL LETTER M - 0x006E: 0x95, # LATIN SMALL LETTER N - 0x006F: 0x96, # LATIN SMALL LETTER O - 0x0070: 0x97, # LATIN SMALL LETTER P - 0x0071: 0x98, # LATIN SMALL LETTER Q - 0x0072: 0x99, # LATIN SMALL LETTER R - 0x0073: 0xA2, # LATIN SMALL LETTER S - 0x0074: 0xA3, # LATIN SMALL LETTER T - 0x0075: 0xA4, # LATIN SMALL LETTER U - 0x0076: 0xA5, # LATIN SMALL LETTER V - 0x0077: 0xA6, # LATIN SMALL LETTER W - 0x0078: 0xA7, # LATIN SMALL LETTER X - 0x0079: 0xA8, # LATIN SMALL LETTER Y - 0x007A: 0xA9, # LATIN SMALL LETTER Z - 0x007B: 0xC0, # LEFT CURLY BRACKET - 0x007C: 0x6A, # VERTICAL LINE - 0x007D: 0xD0, # RIGHT CURLY BRACKET - 0x007E: 0xA1, # TILDE - 0x007F: 0x07, # DELETE - 0x0080: 0x20, # CONTROL - 0x0081: 0x21, # CONTROL - 0x0082: 0x22, # CONTROL - 0x0083: 0x23, # CONTROL - 0x0084: 0x24, # CONTROL - 0x0085: 0x15, # CONTROL - 0x0086: 0x06, # CONTROL - 0x0087: 0x17, # CONTROL - 0x0088: 0x28, # CONTROL - 0x0089: 0x29, # CONTROL - 0x008A: 0x2A, # CONTROL - 0x008B: 0x2B, # CONTROL - 0x008C: 0x2C, # CONTROL - 0x008D: 0x09, # CONTROL - 0x008E: 0x0A, # CONTROL - 0x008F: 0x1B, # CONTROL - 0x0090: 0x30, # CONTROL - 0x0091: 0x31, # CONTROL - 0x0092: 0x1A, # CONTROL - 0x0093: 0x33, # CONTROL - 0x0094: 0x34, # CONTROL - 0x0095: 0x35, # CONTROL - 0x0096: 0x36, # CONTROL - 0x0097: 0x08, # CONTROL - 0x0098: 0x38, # CONTROL - 0x0099: 0x39, # CONTROL - 0x009A: 0x3A, # CONTROL - 0x009B: 0x3B, # CONTROL - 0x009C: 0x04, # CONTROL - 0x009D: 0x14, # CONTROL - 0x009E: 0x3E, # CONTROL - 0x009F: 0xFF, # CONTROL - 0x00A0: 0x74, # NO-BREAK SPACE - 0x00A3: 0xB0, # POUND SIGN - 0x00A6: 0xDF, # BROKEN BAR - 0x00A7: 0xEB, # SECTION SIGN - 0x00A8: 0x70, # DIAERESIS - 0x00A9: 0xFB, # COPYRIGHT SIGN - 0x00AB: 0xEE, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xEF, # NOT SIGN - 0x00AD: 0xCA, # SOFT HYPHEN - 0x00B0: 0x90, # DEGREE SIGN - 0x00B1: 0xDA, # PLUS-MINUS SIGN - 0x00B2: 0xEA, # SUPERSCRIPT TWO - 0x00B3: 0xFA, # SUPERSCRIPT THREE - 0x00B4: 0xA0, # ACUTE ACCENT - 0x00BB: 0xFE, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BD: 0xDB, # VULGAR FRACTION ONE HALF - 0x0385: 0x80, # GREEK DIALYTIKA TONOS - 0x0386: 0x71, # GREEK CAPITAL LETTER ALPHA WITH TONOS - 0x0387: 0xDD, # GREEK ANO TELEIA - 0x0388: 0x72, # GREEK CAPITAL LETTER EPSILON WITH TONOS - 0x0389: 0x73, # GREEK CAPITAL LETTER ETA WITH TONOS - 0x038A: 0x75, # GREEK CAPITAL LETTER IOTA WITH TONOS - 0x038C: 0x76, # GREEK CAPITAL LETTER OMICRON WITH TONOS - 0x038E: 0x77, # GREEK CAPITAL LETTER UPSILON WITH TONOS - 0x038F: 0x78, # GREEK CAPITAL LETTER OMEGA WITH TONOS - 0x0390: 0xCC, # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - 0x0391: 0x41, # GREEK CAPITAL LETTER ALPHA - 0x0392: 0x42, # GREEK CAPITAL LETTER BETA - 0x0393: 0x43, # GREEK CAPITAL LETTER GAMMA - 0x0394: 0x44, # GREEK CAPITAL LETTER DELTA - 0x0395: 0x45, # GREEK CAPITAL LETTER EPSILON - 0x0396: 0x46, # GREEK CAPITAL LETTER ZETA - 0x0397: 0x47, # GREEK CAPITAL LETTER ETA - 0x0398: 0x48, # GREEK CAPITAL LETTER THETA - 0x0399: 0x49, # GREEK CAPITAL LETTER IOTA - 0x039A: 0x51, # GREEK CAPITAL LETTER KAPPA - 0x039B: 0x52, # GREEK CAPITAL LETTER LAMDA - 0x039C: 0x53, # GREEK CAPITAL LETTER MU - 0x039D: 0x54, # GREEK CAPITAL LETTER NU - 0x039E: 0x55, # GREEK CAPITAL LETTER XI - 0x039F: 0x56, # GREEK CAPITAL LETTER OMICRON - 0x03A0: 0x57, # GREEK CAPITAL LETTER PI - 0x03A1: 0x58, # GREEK CAPITAL LETTER RHO - 0x03A3: 0x59, # GREEK CAPITAL LETTER SIGMA - 0x03A4: 0x62, # GREEK CAPITAL LETTER TAU - 0x03A5: 0x63, # GREEK CAPITAL LETTER UPSILON - 0x03A6: 0x64, # GREEK CAPITAL LETTER PHI - 0x03A7: 0x65, # GREEK CAPITAL LETTER CHI - 0x03A8: 0x66, # GREEK CAPITAL LETTER PSI - 0x03A9: 0x67, # GREEK CAPITAL LETTER OMEGA - 0x03AA: 0x68, # GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - 0x03AB: 0x69, # GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - 0x03AC: 0xB1, # GREEK SMALL LETTER ALPHA WITH TONOS - 0x03AD: 0xB2, # GREEK SMALL LETTER EPSILON WITH TONOS - 0x03AE: 0xB3, # GREEK SMALL LETTER ETA WITH TONOS - 0x03AF: 0xB5, # GREEK SMALL LETTER IOTA WITH TONOS - 0x03B0: 0xCD, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - 0x03B1: 0x8A, # GREEK SMALL LETTER ALPHA - 0x03B2: 0x8B, # GREEK SMALL LETTER BETA - 0x03B3: 0x8C, # GREEK SMALL LETTER GAMMA - 0x03B4: 0x8D, # GREEK SMALL LETTER DELTA - 0x03B5: 0x8E, # GREEK SMALL LETTER EPSILON - 0x03B6: 0x8F, # GREEK SMALL LETTER ZETA - 0x03B7: 0x9A, # GREEK SMALL LETTER ETA - 0x03B8: 0x9B, # GREEK SMALL LETTER THETA - 0x03B9: 0x9C, # GREEK SMALL LETTER IOTA - 0x03BA: 0x9D, # GREEK SMALL LETTER KAPPA - 0x03BB: 0x9E, # GREEK SMALL LETTER LAMDA - 0x03BC: 0x9F, # GREEK SMALL LETTER MU - 0x03BD: 0xAA, # GREEK SMALL LETTER NU - 0x03BE: 0xAB, # GREEK SMALL LETTER XI - 0x03BF: 0xAC, # GREEK SMALL LETTER OMICRON - 0x03C0: 0xAD, # GREEK SMALL LETTER PI - 0x03C1: 0xAE, # GREEK SMALL LETTER RHO - 0x03C2: 0xBA, # GREEK SMALL LETTER FINAL SIGMA - 0x03C3: 0xAF, # GREEK SMALL LETTER SIGMA - 0x03C4: 0xBB, # GREEK SMALL LETTER TAU - 0x03C5: 0xBC, # GREEK SMALL LETTER UPSILON - 0x03C6: 0xBD, # GREEK SMALL LETTER PHI - 0x03C7: 0xBE, # GREEK SMALL LETTER CHI - 0x03C8: 0xBF, # GREEK SMALL LETTER PSI - 0x03C9: 0xCB, # GREEK SMALL LETTER OMEGA - 0x03CA: 0xB4, # GREEK SMALL LETTER IOTA WITH DIALYTIKA - 0x03CB: 0xB8, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA - 0x03CC: 0xB6, # GREEK SMALL LETTER OMICRON WITH TONOS - 0x03CD: 0xB7, # GREEK SMALL LETTER UPSILON WITH TONOS - 0x03CE: 0xB9, # GREEK SMALL LETTER OMEGA WITH TONOS - 0x2015: 0xCF, # HORIZONTAL BAR - 0x2018: 0xCE, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xDE, # RIGHT SINGLE QUOTATION MARK -} diff --git a/Lib/encodings/iso8859_1.py b/Lib/encodings/iso8859_1.py index b985585..cef5930 100644 --- a/Lib/encodings/iso8859_1.py +++ b/Lib/encodings/iso8859_1.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 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 - 0x00D0: 0xD0, # LATIN CAPITAL LETTER ETH (Icelandic) - 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 - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xDE, # LATIN CAPITAL LETTER THORN (Icelandic) - 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S (German) - 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 - 0x00F0: 0xF0, # LATIN SMALL LETTER ETH (Icelandic) - 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 - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0xFE, # LATIN SMALL LETTER THORN (Icelandic) - 0x00FF: 0xFF, # LATIN SMALL LETTER Y WITH DIAERESIS -} diff --git a/Lib/encodings/iso8859_10.py b/Lib/encodings/iso8859_10.py index 8588430..9c972b8 100644 --- a/Lib/encodings/iso8859_10.py +++ b/Lib/encodings/iso8859_10.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u0138' # 0xFF -> LATIN SMALL LETTER KRA ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A7: 0xA7, # SECTION SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 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 - 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CB: 0xCB, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CD: 0xCD, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xCE, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0xCF, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D0: 0xD0, # LATIN CAPITAL LETTER ETH (Icelandic) - 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 - 0x00D8: 0xD8, # LATIN CAPITAL LETTER O WITH STROKE - 0x00DA: 0xDA, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xDB, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0xDC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xDE, # LATIN CAPITAL LETTER THORN (Icelandic) - 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S (German) - 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 - 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE - 0x00EB: 0xEB, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00ED: 0xED, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0xEE, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0xEF, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F0: 0xF0, # LATIN SMALL LETTER ETH (Icelandic) - 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 - 0x00F8: 0xF8, # LATIN SMALL LETTER O WITH STROKE - 0x00FA: 0xFA, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0xFB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0xFE, # LATIN SMALL LETTER THORN (Icelandic) - 0x0100: 0xC0, # LATIN CAPITAL LETTER A WITH MACRON - 0x0101: 0xE0, # LATIN SMALL LETTER A WITH MACRON - 0x0104: 0xA1, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xB1, # LATIN SMALL LETTER A WITH OGONEK - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x0110: 0xA9, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xB9, # LATIN SMALL LETTER D WITH STROKE - 0x0112: 0xA2, # LATIN CAPITAL LETTER E WITH MACRON - 0x0113: 0xB2, # LATIN SMALL LETTER E WITH MACRON - 0x0116: 0xCC, # LATIN CAPITAL LETTER E WITH DOT ABOVE - 0x0117: 0xEC, # LATIN SMALL LETTER E WITH DOT ABOVE - 0x0118: 0xCA, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xEA, # LATIN SMALL LETTER E WITH OGONEK - 0x0122: 0xA3, # LATIN CAPITAL LETTER G WITH CEDILLA - 0x0123: 0xB3, # LATIN SMALL LETTER G WITH CEDILLA - 0x0128: 0xA5, # LATIN CAPITAL LETTER I WITH TILDE - 0x0129: 0xB5, # LATIN SMALL LETTER I WITH TILDE - 0x012A: 0xA4, # LATIN CAPITAL LETTER I WITH MACRON - 0x012B: 0xB4, # LATIN SMALL LETTER I WITH MACRON - 0x012E: 0xC7, # LATIN CAPITAL LETTER I WITH OGONEK - 0x012F: 0xE7, # LATIN SMALL LETTER I WITH OGONEK - 0x0136: 0xA6, # LATIN CAPITAL LETTER K WITH CEDILLA - 0x0137: 0xB6, # LATIN SMALL LETTER K WITH CEDILLA - 0x0138: 0xFF, # LATIN SMALL LETTER KRA - 0x013B: 0xA8, # LATIN CAPITAL LETTER L WITH CEDILLA - 0x013C: 0xB8, # LATIN SMALL LETTER L WITH CEDILLA - 0x0145: 0xD1, # LATIN CAPITAL LETTER N WITH CEDILLA - 0x0146: 0xF1, # LATIN SMALL LETTER N WITH CEDILLA - 0x014A: 0xAF, # LATIN CAPITAL LETTER ENG - 0x014B: 0xBF, # LATIN SMALL LETTER ENG - 0x014C: 0xD2, # LATIN CAPITAL LETTER O WITH MACRON - 0x014D: 0xF2, # LATIN SMALL LETTER O WITH MACRON - 0x0160: 0xAA, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xBA, # LATIN SMALL LETTER S WITH CARON - 0x0166: 0xAB, # LATIN CAPITAL LETTER T WITH STROKE - 0x0167: 0xBB, # LATIN SMALL LETTER T WITH STROKE - 0x0168: 0xD7, # LATIN CAPITAL LETTER U WITH TILDE - 0x0169: 0xF7, # LATIN SMALL LETTER U WITH TILDE - 0x016A: 0xAE, # LATIN CAPITAL LETTER U WITH MACRON - 0x016B: 0xBE, # LATIN SMALL LETTER U WITH MACRON - 0x0172: 0xD9, # LATIN CAPITAL LETTER U WITH OGONEK - 0x0173: 0xF9, # LATIN SMALL LETTER U WITH OGONEK - 0x017D: 0xAC, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xBC, # LATIN SMALL LETTER Z WITH CARON - 0x2015: 0xBD, # HORIZONTAL BAR -} diff --git a/Lib/encodings/iso8859_11.py b/Lib/encodings/iso8859_11.py index fffe692..7688224 100644 --- a/Lib/encodings/iso8859_11.py +++ b/Lib/encodings/iso8859_11.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,255 +303,6 @@ decoding_table = ( u'\ufffe' ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x0E01: 0xA1, # THAI CHARACTER KO KAI - 0x0E02: 0xA2, # THAI CHARACTER KHO KHAI - 0x0E03: 0xA3, # THAI CHARACTER KHO KHUAT - 0x0E04: 0xA4, # THAI CHARACTER KHO KHWAI - 0x0E05: 0xA5, # THAI CHARACTER KHO KHON - 0x0E06: 0xA6, # THAI CHARACTER KHO RAKHANG - 0x0E07: 0xA7, # THAI CHARACTER NGO NGU - 0x0E08: 0xA8, # THAI CHARACTER CHO CHAN - 0x0E09: 0xA9, # THAI CHARACTER CHO CHING - 0x0E0A: 0xAA, # THAI CHARACTER CHO CHANG - 0x0E0B: 0xAB, # THAI CHARACTER SO SO - 0x0E0C: 0xAC, # THAI CHARACTER CHO CHOE - 0x0E0D: 0xAD, # THAI CHARACTER YO YING - 0x0E0E: 0xAE, # THAI CHARACTER DO CHADA - 0x0E0F: 0xAF, # THAI CHARACTER TO PATAK - 0x0E10: 0xB0, # THAI CHARACTER THO THAN - 0x0E11: 0xB1, # THAI CHARACTER THO NANGMONTHO - 0x0E12: 0xB2, # THAI CHARACTER THO PHUTHAO - 0x0E13: 0xB3, # THAI CHARACTER NO NEN - 0x0E14: 0xB4, # THAI CHARACTER DO DEK - 0x0E15: 0xB5, # THAI CHARACTER TO TAO - 0x0E16: 0xB6, # THAI CHARACTER THO THUNG - 0x0E17: 0xB7, # THAI CHARACTER THO THAHAN - 0x0E18: 0xB8, # THAI CHARACTER THO THONG - 0x0E19: 0xB9, # THAI CHARACTER NO NU - 0x0E1A: 0xBA, # THAI CHARACTER BO BAIMAI - 0x0E1B: 0xBB, # THAI CHARACTER PO PLA - 0x0E1C: 0xBC, # THAI CHARACTER PHO PHUNG - 0x0E1D: 0xBD, # THAI CHARACTER FO FA - 0x0E1E: 0xBE, # THAI CHARACTER PHO PHAN - 0x0E1F: 0xBF, # THAI CHARACTER FO FAN - 0x0E20: 0xC0, # THAI CHARACTER PHO SAMPHAO - 0x0E21: 0xC1, # THAI CHARACTER MO MA - 0x0E22: 0xC2, # THAI CHARACTER YO YAK - 0x0E23: 0xC3, # THAI CHARACTER RO RUA - 0x0E24: 0xC4, # THAI CHARACTER RU - 0x0E25: 0xC5, # THAI CHARACTER LO LING - 0x0E26: 0xC6, # THAI CHARACTER LU - 0x0E27: 0xC7, # THAI CHARACTER WO WAEN - 0x0E28: 0xC8, # THAI CHARACTER SO SALA - 0x0E29: 0xC9, # THAI CHARACTER SO RUSI - 0x0E2A: 0xCA, # THAI CHARACTER SO SUA - 0x0E2B: 0xCB, # THAI CHARACTER HO HIP - 0x0E2C: 0xCC, # THAI CHARACTER LO CHULA - 0x0E2D: 0xCD, # THAI CHARACTER O ANG - 0x0E2E: 0xCE, # THAI CHARACTER HO NOKHUK - 0x0E2F: 0xCF, # THAI CHARACTER PAIYANNOI - 0x0E30: 0xD0, # THAI CHARACTER SARA A - 0x0E31: 0xD1, # THAI CHARACTER MAI HAN-AKAT - 0x0E32: 0xD2, # THAI CHARACTER SARA AA - 0x0E33: 0xD3, # THAI CHARACTER SARA AM - 0x0E34: 0xD4, # THAI CHARACTER SARA I - 0x0E35: 0xD5, # THAI CHARACTER SARA II - 0x0E36: 0xD6, # THAI CHARACTER SARA UE - 0x0E37: 0xD7, # THAI CHARACTER SARA UEE - 0x0E38: 0xD8, # THAI CHARACTER SARA U - 0x0E39: 0xD9, # THAI CHARACTER SARA UU - 0x0E3A: 0xDA, # THAI CHARACTER PHINTHU - 0x0E3F: 0xDF, # THAI CURRENCY SYMBOL BAHT - 0x0E40: 0xE0, # THAI CHARACTER SARA E - 0x0E41: 0xE1, # THAI CHARACTER SARA AE - 0x0E42: 0xE2, # THAI CHARACTER SARA O - 0x0E43: 0xE3, # THAI CHARACTER SARA AI MAIMUAN - 0x0E44: 0xE4, # THAI CHARACTER SARA AI MAIMALAI - 0x0E45: 0xE5, # THAI CHARACTER LAKKHANGYAO - 0x0E46: 0xE6, # THAI CHARACTER MAIYAMOK - 0x0E47: 0xE7, # THAI CHARACTER MAITAIKHU - 0x0E48: 0xE8, # THAI CHARACTER MAI EK - 0x0E49: 0xE9, # THAI CHARACTER MAI THO - 0x0E4A: 0xEA, # THAI CHARACTER MAI TRI - 0x0E4B: 0xEB, # THAI CHARACTER MAI CHATTAWA - 0x0E4C: 0xEC, # THAI CHARACTER THANTHAKHAT - 0x0E4D: 0xED, # THAI CHARACTER NIKHAHIT - 0x0E4E: 0xEE, # THAI CHARACTER YAMAKKAN - 0x0E4F: 0xEF, # THAI CHARACTER FONGMAN - 0x0E50: 0xF0, # THAI DIGIT ZERO - 0x0E51: 0xF1, # THAI DIGIT ONE - 0x0E52: 0xF2, # THAI DIGIT TWO - 0x0E53: 0xF3, # THAI DIGIT THREE - 0x0E54: 0xF4, # THAI DIGIT FOUR - 0x0E55: 0xF5, # THAI DIGIT FIVE - 0x0E56: 0xF6, # THAI DIGIT SIX - 0x0E57: 0xF7, # THAI DIGIT SEVEN - 0x0E58: 0xF8, # THAI DIGIT EIGHT - 0x0E59: 0xF9, # THAI DIGIT NINE - 0x0E5A: 0xFA, # THAI CHARACTER ANGKHANKHU - 0x0E5B: 0xFB, # THAI CHARACTER KHOMUT -} diff --git a/Lib/encodings/iso8859_13.py b/Lib/encodings/iso8859_13.py index a890580..571e14b 100644 --- a/Lib/encodings/iso8859_13.py +++ b/Lib/encodings/iso8859_13.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u2019' # 0xFF -> RIGHT SINGLE QUOTATION MARK ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A6: 0xA6, # BROKEN BAR - 0x00A7: 0xA7, # SECTION SIGN - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAC, # NOT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AE: 0xAE, # REGISTERED SIGN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B2: 0xB2, # SUPERSCRIPT TWO - 0x00B3: 0xB3, # SUPERSCRIPT THREE - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00B9: 0xB9, # SUPERSCRIPT ONE - 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 - 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0xC5, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0xAF, # LATIN CAPITAL LETTER AE - 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D5: 0xD5, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xD7, # MULTIPLICATION SIGN - 0x00D8: 0xA8, # LATIN CAPITAL LETTER O WITH STROKE - 0x00DC: 0xDC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S (German) - 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0xE5, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0xBF, # LATIN SMALL LETTER AE - 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE - 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE - 0x00F5: 0xF5, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xF7, # DIVISION SIGN - 0x00F8: 0xB8, # LATIN SMALL LETTER O WITH STROKE - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x0100: 0xC2, # LATIN CAPITAL LETTER A WITH MACRON - 0x0101: 0xE2, # LATIN SMALL LETTER A WITH MACRON - 0x0104: 0xC0, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xE0, # LATIN SMALL LETTER A WITH OGONEK - 0x0106: 0xC3, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0xE3, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x0112: 0xC7, # LATIN CAPITAL LETTER E WITH MACRON - 0x0113: 0xE7, # LATIN SMALL LETTER E WITH MACRON - 0x0116: 0xCB, # LATIN CAPITAL LETTER E WITH DOT ABOVE - 0x0117: 0xEB, # LATIN SMALL LETTER E WITH DOT ABOVE - 0x0118: 0xC6, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xE6, # LATIN SMALL LETTER E WITH OGONEK - 0x0122: 0xCC, # LATIN CAPITAL LETTER G WITH CEDILLA - 0x0123: 0xEC, # LATIN SMALL LETTER G WITH CEDILLA - 0x012A: 0xCE, # LATIN CAPITAL LETTER I WITH MACRON - 0x012B: 0xEE, # LATIN SMALL LETTER I WITH MACRON - 0x012E: 0xC1, # LATIN CAPITAL LETTER I WITH OGONEK - 0x012F: 0xE1, # LATIN SMALL LETTER I WITH OGONEK - 0x0136: 0xCD, # LATIN CAPITAL LETTER K WITH CEDILLA - 0x0137: 0xED, # LATIN SMALL LETTER K WITH CEDILLA - 0x013B: 0xCF, # LATIN CAPITAL LETTER L WITH CEDILLA - 0x013C: 0xEF, # LATIN SMALL LETTER L WITH CEDILLA - 0x0141: 0xD9, # LATIN CAPITAL LETTER L WITH STROKE - 0x0142: 0xF9, # LATIN SMALL LETTER L WITH STROKE - 0x0143: 0xD1, # LATIN CAPITAL LETTER N WITH ACUTE - 0x0144: 0xF1, # LATIN SMALL LETTER N WITH ACUTE - 0x0145: 0xD2, # LATIN CAPITAL LETTER N WITH CEDILLA - 0x0146: 0xF2, # LATIN SMALL LETTER N WITH CEDILLA - 0x014C: 0xD4, # LATIN CAPITAL LETTER O WITH MACRON - 0x014D: 0xF4, # LATIN SMALL LETTER O WITH MACRON - 0x0156: 0xAA, # LATIN CAPITAL LETTER R WITH CEDILLA - 0x0157: 0xBA, # LATIN SMALL LETTER R WITH CEDILLA - 0x015A: 0xDA, # LATIN CAPITAL LETTER S WITH ACUTE - 0x015B: 0xFA, # LATIN SMALL LETTER S WITH ACUTE - 0x0160: 0xD0, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xF0, # LATIN SMALL LETTER S WITH CARON - 0x016A: 0xDB, # LATIN CAPITAL LETTER U WITH MACRON - 0x016B: 0xFB, # LATIN SMALL LETTER U WITH MACRON - 0x0172: 0xD8, # LATIN CAPITAL LETTER U WITH OGONEK - 0x0173: 0xF8, # LATIN SMALL LETTER U WITH OGONEK - 0x0179: 0xCA, # LATIN CAPITAL LETTER Z WITH ACUTE - 0x017A: 0xEA, # LATIN SMALL LETTER Z WITH ACUTE - 0x017B: 0xDD, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xFD, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x017D: 0xDE, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xFE, # LATIN SMALL LETTER Z WITH CARON - 0x2019: 0xFF, # RIGHT SINGLE QUOTATION MARK - 0x201C: 0xB4, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xA1, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xA5, # DOUBLE LOW-9 QUOTATION MARK -} diff --git a/Lib/encodings/iso8859_14.py b/Lib/encodings/iso8859_14.py index afa458c..022063d 100644 --- a/Lib/encodings/iso8859_14.py +++ b/Lib/encodings/iso8859_14.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A3: 0xA3, # POUND SIGN - 0x00A7: 0xA7, # SECTION SIGN - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AE: 0xAE, # REGISTERED SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 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 - 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 - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 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 - 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 - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FF: 0xFF, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x010A: 0xA4, # LATIN CAPITAL LETTER C WITH DOT ABOVE - 0x010B: 0xA5, # LATIN SMALL LETTER C WITH DOT ABOVE - 0x0120: 0xB2, # LATIN CAPITAL LETTER G WITH DOT ABOVE - 0x0121: 0xB3, # LATIN SMALL LETTER G WITH DOT ABOVE - 0x0174: 0xD0, # LATIN CAPITAL LETTER W WITH CIRCUMFLEX - 0x0175: 0xF0, # LATIN SMALL LETTER W WITH CIRCUMFLEX - 0x0176: 0xDE, # LATIN CAPITAL LETTER Y WITH CIRCUMFLEX - 0x0177: 0xFE, # LATIN SMALL LETTER Y WITH CIRCUMFLEX - 0x0178: 0xAF, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x1E02: 0xA1, # LATIN CAPITAL LETTER B WITH DOT ABOVE - 0x1E03: 0xA2, # LATIN SMALL LETTER B WITH DOT ABOVE - 0x1E0A: 0xA6, # LATIN CAPITAL LETTER D WITH DOT ABOVE - 0x1E0B: 0xAB, # LATIN SMALL LETTER D WITH DOT ABOVE - 0x1E1E: 0xB0, # LATIN CAPITAL LETTER F WITH DOT ABOVE - 0x1E1F: 0xB1, # LATIN SMALL LETTER F WITH DOT ABOVE - 0x1E40: 0xB4, # LATIN CAPITAL LETTER M WITH DOT ABOVE - 0x1E41: 0xB5, # LATIN SMALL LETTER M WITH DOT ABOVE - 0x1E56: 0xB7, # LATIN CAPITAL LETTER P WITH DOT ABOVE - 0x1E57: 0xB9, # LATIN SMALL LETTER P WITH DOT ABOVE - 0x1E60: 0xBB, # LATIN CAPITAL LETTER S WITH DOT ABOVE - 0x1E61: 0xBF, # LATIN SMALL LETTER S WITH DOT ABOVE - 0x1E6A: 0xD7, # LATIN CAPITAL LETTER T WITH DOT ABOVE - 0x1E6B: 0xF7, # LATIN SMALL LETTER T WITH DOT ABOVE - 0x1E80: 0xA8, # LATIN CAPITAL LETTER W WITH GRAVE - 0x1E81: 0xB8, # LATIN SMALL LETTER W WITH GRAVE - 0x1E82: 0xAA, # LATIN CAPITAL LETTER W WITH ACUTE - 0x1E83: 0xBA, # LATIN SMALL LETTER W WITH ACUTE - 0x1E84: 0xBD, # LATIN CAPITAL LETTER W WITH DIAERESIS - 0x1E85: 0xBE, # LATIN SMALL LETTER W WITH DIAERESIS - 0x1EF2: 0xAC, # LATIN CAPITAL LETTER Y WITH GRAVE - 0x1EF3: 0xBC, # LATIN SMALL LETTER Y WITH GRAVE -} diff --git a/Lib/encodings/iso8859_15.py b/Lib/encodings/iso8859_15.py index 4a8334e..25deaf6 100644 --- a/Lib/encodings/iso8859_15.py +++ b/Lib/encodings/iso8859_15.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A1: 0xA1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A5: 0xA5, # YEN SIGN - 0x00A7: 0xA7, # SECTION SIGN - 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 - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00B9: 0xB9, # SUPERSCRIPT ONE - 0x00BA: 0xBA, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 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 - 0x00D0: 0xD0, # LATIN CAPITAL LETTER ETH - 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 - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xDE, # LATIN CAPITAL LETTER THORN - 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 - 0x00F0: 0xF0, # LATIN SMALL LETTER ETH - 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 - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0xFE, # LATIN SMALL LETTER THORN - 0x00FF: 0xFF, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x0152: 0xBC, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xBD, # LATIN SMALL LIGATURE OE - 0x0160: 0xA6, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xA8, # LATIN SMALL LETTER S WITH CARON - 0x0178: 0xBE, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x017D: 0xB4, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xB8, # LATIN SMALL LETTER Z WITH CARON - 0x20AC: 0xA4, # EURO SIGN -} diff --git a/Lib/encodings/iso8859_16.py b/Lib/encodings/iso8859_16.py index aeebfb6..a4ed7f2 100644 --- a/Lib/encodings/iso8859_16.py +++ b/Lib/encodings/iso8859_16.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A7: 0xA7, # SECTION SIGN - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B6: 0xB6, # PILCROW SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION 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 - 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 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 - 0x00D2: 0xD2, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xD4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 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 - 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS - 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 - 0x00F2: 0xF2, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xF4, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS - 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 - 0x0102: 0xC3, # LATIN CAPITAL LETTER A WITH BREVE - 0x0103: 0xE3, # LATIN SMALL LETTER A WITH BREVE - 0x0104: 0xA1, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xA2, # LATIN SMALL LETTER A WITH OGONEK - 0x0106: 0xC5, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0xE5, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0xB2, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xB9, # LATIN SMALL LETTER C WITH CARON - 0x0110: 0xD0, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xF0, # LATIN SMALL LETTER D WITH STROKE - 0x0118: 0xDD, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xFD, # LATIN SMALL LETTER E WITH OGONEK - 0x0141: 0xA3, # LATIN CAPITAL LETTER L WITH STROKE - 0x0142: 0xB3, # LATIN SMALL LETTER L WITH STROKE - 0x0143: 0xD1, # LATIN CAPITAL LETTER N WITH ACUTE - 0x0144: 0xF1, # LATIN SMALL LETTER N WITH ACUTE - 0x0150: 0xD5, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - 0x0151: 0xF5, # LATIN SMALL LETTER O WITH DOUBLE ACUTE - 0x0152: 0xBC, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xBD, # LATIN SMALL LIGATURE OE - 0x015A: 0xD7, # LATIN CAPITAL LETTER S WITH ACUTE - 0x015B: 0xF7, # LATIN SMALL LETTER S WITH ACUTE - 0x0160: 0xA6, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xA8, # LATIN SMALL LETTER S WITH CARON - 0x0170: 0xD8, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - 0x0171: 0xF8, # LATIN SMALL LETTER U WITH DOUBLE ACUTE - 0x0178: 0xBE, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x0179: 0xAC, # LATIN CAPITAL LETTER Z WITH ACUTE - 0x017A: 0xAE, # LATIN SMALL LETTER Z WITH ACUTE - 0x017B: 0xAF, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xBF, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x017D: 0xB4, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xB8, # LATIN SMALL LETTER Z WITH CARON - 0x0218: 0xAA, # LATIN CAPITAL LETTER S WITH COMMA BELOW - 0x0219: 0xBA, # LATIN SMALL LETTER S WITH COMMA BELOW - 0x021A: 0xDE, # LATIN CAPITAL LETTER T WITH COMMA BELOW - 0x021B: 0xFE, # LATIN SMALL LETTER T WITH COMMA BELOW - 0x201D: 0xB5, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xA5, # DOUBLE LOW-9 QUOTATION MARK - 0x20AC: 0xA4, # EURO SIGN -} diff --git a/Lib/encodings/iso8859_2.py b/Lib/encodings/iso8859_2.py index 845f322..eb2bc97 100644 --- a/Lib/encodings/iso8859_2.py +++ b/Lib/encodings/iso8859_2.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02d9' # 0xFF -> DOT ABOVE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0xA8, # DIAERESIS - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B4: 0xB4, # ACUTE ACCENT - 0x00B8: 0xB8, # CEDILLA - 0x00C1: 0xC1, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xC2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C7: 0xC7, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CB: 0xCB, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CD: 0xCD, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xCE, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00D3: 0xD3, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xD4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xD7, # MULTIPLICATION SIGN - 0x00DA: 0xDA, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DC: 0xDC, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xDD, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DF: 0xDF, # LATIN SMALL LETTER SHARP S - 0x00E1: 0xE1, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0xE2, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E7: 0xE7, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE - 0x00EB: 0xEB, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00ED: 0xED, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0xEE, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00F3: 0xF3, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0xF4, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xF7, # DIVISION SIGN - 0x00FA: 0xFA, # LATIN SMALL LETTER U WITH ACUTE - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0xFD, # LATIN SMALL LETTER Y WITH ACUTE - 0x0102: 0xC3, # LATIN CAPITAL LETTER A WITH BREVE - 0x0103: 0xE3, # LATIN SMALL LETTER A WITH BREVE - 0x0104: 0xA1, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xB1, # LATIN SMALL LETTER A WITH OGONEK - 0x0106: 0xC6, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0xE6, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x010E: 0xCF, # LATIN CAPITAL LETTER D WITH CARON - 0x010F: 0xEF, # LATIN SMALL LETTER D WITH CARON - 0x0110: 0xD0, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xF0, # LATIN SMALL LETTER D WITH STROKE - 0x0118: 0xCA, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xEA, # LATIN SMALL LETTER E WITH OGONEK - 0x011A: 0xCC, # LATIN CAPITAL LETTER E WITH CARON - 0x011B: 0xEC, # LATIN SMALL LETTER E WITH CARON - 0x0139: 0xC5, # LATIN CAPITAL LETTER L WITH ACUTE - 0x013A: 0xE5, # LATIN SMALL LETTER L WITH ACUTE - 0x013D: 0xA5, # LATIN CAPITAL LETTER L WITH CARON - 0x013E: 0xB5, # LATIN SMALL LETTER L WITH CARON - 0x0141: 0xA3, # LATIN CAPITAL LETTER L WITH STROKE - 0x0142: 0xB3, # LATIN SMALL LETTER L WITH STROKE - 0x0143: 0xD1, # LATIN CAPITAL LETTER N WITH ACUTE - 0x0144: 0xF1, # LATIN SMALL LETTER N WITH ACUTE - 0x0147: 0xD2, # LATIN CAPITAL LETTER N WITH CARON - 0x0148: 0xF2, # LATIN SMALL LETTER N WITH CARON - 0x0150: 0xD5, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - 0x0151: 0xF5, # LATIN SMALL LETTER O WITH DOUBLE ACUTE - 0x0154: 0xC0, # LATIN CAPITAL LETTER R WITH ACUTE - 0x0155: 0xE0, # LATIN SMALL LETTER R WITH ACUTE - 0x0158: 0xD8, # LATIN CAPITAL LETTER R WITH CARON - 0x0159: 0xF8, # LATIN SMALL LETTER R WITH CARON - 0x015A: 0xA6, # LATIN CAPITAL LETTER S WITH ACUTE - 0x015B: 0xB6, # LATIN SMALL LETTER S WITH ACUTE - 0x015E: 0xAA, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015F: 0xBA, # LATIN SMALL LETTER S WITH CEDILLA - 0x0160: 0xA9, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xB9, # LATIN SMALL LETTER S WITH CARON - 0x0162: 0xDE, # LATIN CAPITAL LETTER T WITH CEDILLA - 0x0163: 0xFE, # LATIN SMALL LETTER T WITH CEDILLA - 0x0164: 0xAB, # LATIN CAPITAL LETTER T WITH CARON - 0x0165: 0xBB, # LATIN SMALL LETTER T WITH CARON - 0x016E: 0xD9, # LATIN CAPITAL LETTER U WITH RING ABOVE - 0x016F: 0xF9, # LATIN SMALL LETTER U WITH RING ABOVE - 0x0170: 0xDB, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - 0x0171: 0xFB, # LATIN SMALL LETTER U WITH DOUBLE ACUTE - 0x0179: 0xAC, # LATIN CAPITAL LETTER Z WITH ACUTE - 0x017A: 0xBC, # LATIN SMALL LETTER Z WITH ACUTE - 0x017B: 0xAF, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xBF, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x017D: 0xAE, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xBE, # LATIN SMALL LETTER Z WITH CARON - 0x02C7: 0xB7, # CARON - 0x02D8: 0xA2, # BREVE - 0x02D9: 0xFF, # DOT ABOVE - 0x02DB: 0xB2, # OGONEK - 0x02DD: 0xBD, # DOUBLE ACUTE ACCENT -} diff --git a/Lib/encodings/iso8859_3.py b/Lib/encodings/iso8859_3.py index fbc8775..3922229 100644 --- a/Lib/encodings/iso8859_3.py +++ b/Lib/encodings/iso8859_3.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,256 +303,6 @@ decoding_table = ( u'\u02d9' # 0xFF -> DOT ABOVE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A3: 0xA3, # POUND SIGN - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0xA8, # DIAERESIS - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B2: 0xB2, # SUPERSCRIPT TWO - 0x00B3: 0xB3, # SUPERSCRIPT THREE - 0x00B4: 0xB4, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B7: 0xB7, # MIDDLE DOT - 0x00B8: 0xB8, # CEDILLA - 0x00BD: 0xBD, # VULGAR FRACTION ONE HALF - 0x00C0: 0xC0, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0xC1, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xC2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C4: 0xC4, # LATIN CAPITAL LETTER A WITH DIAERESIS - 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 - 0x00D6: 0xD6, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D7: 0xD7, # MULTIPLICATION SIGN - 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 - 0x00E4: 0xE4, # LATIN SMALL LETTER A WITH DIAERESIS - 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 - 0x00F6: 0xF6, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xF7, # DIVISION SIGN - 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 - 0x0108: 0xC6, # LATIN CAPITAL LETTER C WITH CIRCUMFLEX - 0x0109: 0xE6, # LATIN SMALL LETTER C WITH CIRCUMFLEX - 0x010A: 0xC5, # LATIN CAPITAL LETTER C WITH DOT ABOVE - 0x010B: 0xE5, # LATIN SMALL LETTER C WITH DOT ABOVE - 0x011C: 0xD8, # LATIN CAPITAL LETTER G WITH CIRCUMFLEX - 0x011D: 0xF8, # LATIN SMALL LETTER G WITH CIRCUMFLEX - 0x011E: 0xAB, # LATIN CAPITAL LETTER G WITH BREVE - 0x011F: 0xBB, # LATIN SMALL LETTER G WITH BREVE - 0x0120: 0xD5, # LATIN CAPITAL LETTER G WITH DOT ABOVE - 0x0121: 0xF5, # LATIN SMALL LETTER G WITH DOT ABOVE - 0x0124: 0xA6, # LATIN CAPITAL LETTER H WITH CIRCUMFLEX - 0x0125: 0xB6, # LATIN SMALL LETTER H WITH CIRCUMFLEX - 0x0126: 0xA1, # LATIN CAPITAL LETTER H WITH STROKE - 0x0127: 0xB1, # LATIN SMALL LETTER H WITH STROKE - 0x0130: 0xA9, # LATIN CAPITAL LETTER I WITH DOT ABOVE - 0x0131: 0xB9, # LATIN SMALL LETTER DOTLESS I - 0x0134: 0xAC, # LATIN CAPITAL LETTER J WITH CIRCUMFLEX - 0x0135: 0xBC, # LATIN SMALL LETTER J WITH CIRCUMFLEX - 0x015C: 0xDE, # LATIN CAPITAL LETTER S WITH CIRCUMFLEX - 0x015D: 0xFE, # LATIN SMALL LETTER S WITH CIRCUMFLEX - 0x015E: 0xAA, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015F: 0xBA, # LATIN SMALL LETTER S WITH CEDILLA - 0x016C: 0xDD, # LATIN CAPITAL LETTER U WITH BREVE - 0x016D: 0xFD, # LATIN SMALL LETTER U WITH BREVE - 0x017B: 0xAF, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xBF, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x02D8: 0xA2, # BREVE - 0x02D9: 0xFF, # DOT ABOVE -} diff --git a/Lib/encodings/iso8859_4.py b/Lib/encodings/iso8859_4.py index e705954..abdd7cf 100644 --- a/Lib/encodings/iso8859_4.py +++ b/Lib/encodings/iso8859_4.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02d9' # 0xFF -> DOT ABOVE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0xA8, # DIAERESIS - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00AF: 0xAF, # MACRON - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B4: 0xB4, # ACUTE ACCENT - 0x00B8: 0xB8, # CEDILLA - 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 - 0x00C9: 0xC9, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CB: 0xCB, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CD: 0xCD, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xCE, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 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 - 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 - 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 - 0x00E9: 0xE9, # LATIN SMALL LETTER E WITH ACUTE - 0x00EB: 0xEB, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00ED: 0xED, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0xEE, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 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 - 0x00FA: 0xFA, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0xFB, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0xFC, # LATIN SMALL LETTER U WITH DIAERESIS - 0x0100: 0xC0, # LATIN CAPITAL LETTER A WITH MACRON - 0x0101: 0xE0, # LATIN SMALL LETTER A WITH MACRON - 0x0104: 0xA1, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0xB1, # LATIN SMALL LETTER A WITH OGONEK - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x0110: 0xD0, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xF0, # LATIN SMALL LETTER D WITH STROKE - 0x0112: 0xAA, # LATIN CAPITAL LETTER E WITH MACRON - 0x0113: 0xBA, # LATIN SMALL LETTER E WITH MACRON - 0x0116: 0xCC, # LATIN CAPITAL LETTER E WITH DOT ABOVE - 0x0117: 0xEC, # LATIN SMALL LETTER E WITH DOT ABOVE - 0x0118: 0xCA, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xEA, # LATIN SMALL LETTER E WITH OGONEK - 0x0122: 0xAB, # LATIN CAPITAL LETTER G WITH CEDILLA - 0x0123: 0xBB, # LATIN SMALL LETTER G WITH CEDILLA - 0x0128: 0xA5, # LATIN CAPITAL LETTER I WITH TILDE - 0x0129: 0xB5, # LATIN SMALL LETTER I WITH TILDE - 0x012A: 0xCF, # LATIN CAPITAL LETTER I WITH MACRON - 0x012B: 0xEF, # LATIN SMALL LETTER I WITH MACRON - 0x012E: 0xC7, # LATIN CAPITAL LETTER I WITH OGONEK - 0x012F: 0xE7, # LATIN SMALL LETTER I WITH OGONEK - 0x0136: 0xD3, # LATIN CAPITAL LETTER K WITH CEDILLA - 0x0137: 0xF3, # LATIN SMALL LETTER K WITH CEDILLA - 0x0138: 0xA2, # LATIN SMALL LETTER KRA - 0x013B: 0xA6, # LATIN CAPITAL LETTER L WITH CEDILLA - 0x013C: 0xB6, # LATIN SMALL LETTER L WITH CEDILLA - 0x0145: 0xD1, # LATIN CAPITAL LETTER N WITH CEDILLA - 0x0146: 0xF1, # LATIN SMALL LETTER N WITH CEDILLA - 0x014A: 0xBD, # LATIN CAPITAL LETTER ENG - 0x014B: 0xBF, # LATIN SMALL LETTER ENG - 0x014C: 0xD2, # LATIN CAPITAL LETTER O WITH MACRON - 0x014D: 0xF2, # LATIN SMALL LETTER O WITH MACRON - 0x0156: 0xA3, # LATIN CAPITAL LETTER R WITH CEDILLA - 0x0157: 0xB3, # LATIN SMALL LETTER R WITH CEDILLA - 0x0160: 0xA9, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xB9, # LATIN SMALL LETTER S WITH CARON - 0x0166: 0xAC, # LATIN CAPITAL LETTER T WITH STROKE - 0x0167: 0xBC, # LATIN SMALL LETTER T WITH STROKE - 0x0168: 0xDD, # LATIN CAPITAL LETTER U WITH TILDE - 0x0169: 0xFD, # LATIN SMALL LETTER U WITH TILDE - 0x016A: 0xDE, # LATIN CAPITAL LETTER U WITH MACRON - 0x016B: 0xFE, # LATIN SMALL LETTER U WITH MACRON - 0x0172: 0xD9, # LATIN CAPITAL LETTER U WITH OGONEK - 0x0173: 0xF9, # LATIN SMALL LETTER U WITH OGONEK - 0x017D: 0xAE, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xBE, # LATIN SMALL LETTER Z WITH CARON - 0x02C7: 0xB7, # CARON - 0x02D9: 0xFF, # DOT ABOVE - 0x02DB: 0xB2, # OGONEK -} diff --git a/Lib/encodings/iso8859_5.py b/Lib/encodings/iso8859_5.py index 93a4e90..698f879 100644 --- a/Lib/encodings/iso8859_5.py +++ b/Lib/encodings/iso8859_5.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u045f' # 0xFF -> CYRILLIC SMALL LETTER DZHE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A7: 0xFD, # SECTION SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x0401: 0xA1, # CYRILLIC CAPITAL LETTER IO - 0x0402: 0xA2, # CYRILLIC CAPITAL LETTER DJE - 0x0403: 0xA3, # CYRILLIC CAPITAL LETTER GJE - 0x0404: 0xA4, # CYRILLIC CAPITAL LETTER UKRAINIAN IE - 0x0405: 0xA5, # CYRILLIC CAPITAL LETTER DZE - 0x0406: 0xA6, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0407: 0xA7, # CYRILLIC CAPITAL LETTER YI - 0x0408: 0xA8, # CYRILLIC CAPITAL LETTER JE - 0x0409: 0xA9, # CYRILLIC CAPITAL LETTER LJE - 0x040A: 0xAA, # CYRILLIC CAPITAL LETTER NJE - 0x040B: 0xAB, # CYRILLIC CAPITAL LETTER TSHE - 0x040C: 0xAC, # CYRILLIC CAPITAL LETTER KJE - 0x040E: 0xAE, # CYRILLIC CAPITAL LETTER SHORT U - 0x040F: 0xAF, # CYRILLIC CAPITAL LETTER DZHE - 0x0410: 0xB0, # CYRILLIC CAPITAL LETTER A - 0x0411: 0xB1, # CYRILLIC CAPITAL LETTER BE - 0x0412: 0xB2, # CYRILLIC CAPITAL LETTER VE - 0x0413: 0xB3, # CYRILLIC CAPITAL LETTER GHE - 0x0414: 0xB4, # CYRILLIC CAPITAL LETTER DE - 0x0415: 0xB5, # CYRILLIC CAPITAL LETTER IE - 0x0416: 0xB6, # CYRILLIC CAPITAL LETTER ZHE - 0x0417: 0xB7, # CYRILLIC CAPITAL LETTER ZE - 0x0418: 0xB8, # CYRILLIC CAPITAL LETTER I - 0x0419: 0xB9, # CYRILLIC CAPITAL LETTER SHORT I - 0x041A: 0xBA, # CYRILLIC CAPITAL LETTER KA - 0x041B: 0xBB, # CYRILLIC CAPITAL LETTER EL - 0x041C: 0xBC, # CYRILLIC CAPITAL LETTER EM - 0x041D: 0xBD, # CYRILLIC CAPITAL LETTER EN - 0x041E: 0xBE, # CYRILLIC CAPITAL LETTER O - 0x041F: 0xBF, # CYRILLIC CAPITAL LETTER PE - 0x0420: 0xC0, # CYRILLIC CAPITAL LETTER ER - 0x0421: 0xC1, # CYRILLIC CAPITAL LETTER ES - 0x0422: 0xC2, # CYRILLIC CAPITAL LETTER TE - 0x0423: 0xC3, # CYRILLIC CAPITAL LETTER U - 0x0424: 0xC4, # CYRILLIC CAPITAL LETTER EF - 0x0425: 0xC5, # CYRILLIC CAPITAL LETTER HA - 0x0426: 0xC6, # CYRILLIC CAPITAL LETTER TSE - 0x0427: 0xC7, # CYRILLIC CAPITAL LETTER CHE - 0x0428: 0xC8, # CYRILLIC CAPITAL LETTER SHA - 0x0429: 0xC9, # CYRILLIC CAPITAL LETTER SHCHA - 0x042A: 0xCA, # CYRILLIC CAPITAL LETTER HARD SIGN - 0x042B: 0xCB, # CYRILLIC CAPITAL LETTER YERU - 0x042C: 0xCC, # CYRILLIC CAPITAL LETTER SOFT SIGN - 0x042D: 0xCD, # CYRILLIC CAPITAL LETTER E - 0x042E: 0xCE, # CYRILLIC CAPITAL LETTER YU - 0x042F: 0xCF, # CYRILLIC CAPITAL LETTER YA - 0x0430: 0xD0, # CYRILLIC SMALL LETTER A - 0x0431: 0xD1, # CYRILLIC SMALL LETTER BE - 0x0432: 0xD2, # CYRILLIC SMALL LETTER VE - 0x0433: 0xD3, # CYRILLIC SMALL LETTER GHE - 0x0434: 0xD4, # CYRILLIC SMALL LETTER DE - 0x0435: 0xD5, # CYRILLIC SMALL LETTER IE - 0x0436: 0xD6, # CYRILLIC SMALL LETTER ZHE - 0x0437: 0xD7, # CYRILLIC SMALL LETTER ZE - 0x0438: 0xD8, # CYRILLIC SMALL LETTER I - 0x0439: 0xD9, # CYRILLIC SMALL LETTER SHORT I - 0x043A: 0xDA, # CYRILLIC SMALL LETTER KA - 0x043B: 0xDB, # CYRILLIC SMALL LETTER EL - 0x043C: 0xDC, # CYRILLIC SMALL LETTER EM - 0x043D: 0xDD, # CYRILLIC SMALL LETTER EN - 0x043E: 0xDE, # CYRILLIC SMALL LETTER O - 0x043F: 0xDF, # CYRILLIC SMALL LETTER PE - 0x0440: 0xE0, # CYRILLIC SMALL LETTER ER - 0x0441: 0xE1, # CYRILLIC SMALL LETTER ES - 0x0442: 0xE2, # CYRILLIC SMALL LETTER TE - 0x0443: 0xE3, # CYRILLIC SMALL LETTER U - 0x0444: 0xE4, # CYRILLIC SMALL LETTER EF - 0x0445: 0xE5, # CYRILLIC SMALL LETTER HA - 0x0446: 0xE6, # CYRILLIC SMALL LETTER TSE - 0x0447: 0xE7, # CYRILLIC SMALL LETTER CHE - 0x0448: 0xE8, # CYRILLIC SMALL LETTER SHA - 0x0449: 0xE9, # CYRILLIC SMALL LETTER SHCHA - 0x044A: 0xEA, # CYRILLIC SMALL LETTER HARD SIGN - 0x044B: 0xEB, # CYRILLIC SMALL LETTER YERU - 0x044C: 0xEC, # CYRILLIC SMALL LETTER SOFT SIGN - 0x044D: 0xED, # CYRILLIC SMALL LETTER E - 0x044E: 0xEE, # CYRILLIC SMALL LETTER YU - 0x044F: 0xEF, # CYRILLIC SMALL LETTER YA - 0x0451: 0xF1, # CYRILLIC SMALL LETTER IO - 0x0452: 0xF2, # CYRILLIC SMALL LETTER DJE - 0x0453: 0xF3, # CYRILLIC SMALL LETTER GJE - 0x0454: 0xF4, # CYRILLIC SMALL LETTER UKRAINIAN IE - 0x0455: 0xF5, # CYRILLIC SMALL LETTER DZE - 0x0456: 0xF6, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0457: 0xF7, # CYRILLIC SMALL LETTER YI - 0x0458: 0xF8, # CYRILLIC SMALL LETTER JE - 0x0459: 0xF9, # CYRILLIC SMALL LETTER LJE - 0x045A: 0xFA, # CYRILLIC SMALL LETTER NJE - 0x045B: 0xFB, # CYRILLIC SMALL LETTER TSHE - 0x045C: 0xFC, # CYRILLIC SMALL LETTER KJE - 0x045E: 0xFE, # CYRILLIC SMALL LETTER SHORT U - 0x045F: 0xFF, # CYRILLIC SMALL LETTER DZHE - 0x2116: 0xF0, # NUMERO SIGN -} diff --git a/Lib/encodings/iso8859_6.py b/Lib/encodings/iso8859_6.py index f911cc4..060822a 100644 --- a/Lib/encodings/iso8859_6.py +++ b/Lib/encodings/iso8859_6.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,218 +303,6 @@ decoding_table = ( u'\ufffe' ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A4: 0xA4, # CURRENCY SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x060C: 0xAC, # ARABIC COMMA - 0x061B: 0xBB, # ARABIC SEMICOLON - 0x061F: 0xBF, # ARABIC QUESTION MARK - 0x0621: 0xC1, # ARABIC LETTER HAMZA - 0x0622: 0xC2, # ARABIC LETTER ALEF WITH MADDA ABOVE - 0x0623: 0xC3, # ARABIC LETTER ALEF WITH HAMZA ABOVE - 0x0624: 0xC4, # ARABIC LETTER WAW WITH HAMZA ABOVE - 0x0625: 0xC5, # ARABIC LETTER ALEF WITH HAMZA BELOW - 0x0626: 0xC6, # ARABIC LETTER YEH WITH HAMZA ABOVE - 0x0627: 0xC7, # ARABIC LETTER ALEF - 0x0628: 0xC8, # ARABIC LETTER BEH - 0x0629: 0xC9, # ARABIC LETTER TEH MARBUTA - 0x062A: 0xCA, # ARABIC LETTER TEH - 0x062B: 0xCB, # ARABIC LETTER THEH - 0x062C: 0xCC, # ARABIC LETTER JEEM - 0x062D: 0xCD, # ARABIC LETTER HAH - 0x062E: 0xCE, # ARABIC LETTER KHAH - 0x062F: 0xCF, # ARABIC LETTER DAL - 0x0630: 0xD0, # ARABIC LETTER THAL - 0x0631: 0xD1, # ARABIC LETTER REH - 0x0632: 0xD2, # ARABIC LETTER ZAIN - 0x0633: 0xD3, # ARABIC LETTER SEEN - 0x0634: 0xD4, # ARABIC LETTER SHEEN - 0x0635: 0xD5, # ARABIC LETTER SAD - 0x0636: 0xD6, # ARABIC LETTER DAD - 0x0637: 0xD7, # ARABIC LETTER TAH - 0x0638: 0xD8, # ARABIC LETTER ZAH - 0x0639: 0xD9, # ARABIC LETTER AIN - 0x063A: 0xDA, # ARABIC LETTER GHAIN - 0x0640: 0xE0, # ARABIC TATWEEL - 0x0641: 0xE1, # ARABIC LETTER FEH - 0x0642: 0xE2, # ARABIC LETTER QAF - 0x0643: 0xE3, # ARABIC LETTER KAF - 0x0644: 0xE4, # ARABIC LETTER LAM - 0x0645: 0xE5, # ARABIC LETTER MEEM - 0x0646: 0xE6, # ARABIC LETTER NOON - 0x0647: 0xE7, # ARABIC LETTER HEH - 0x0648: 0xE8, # ARABIC LETTER WAW - 0x0649: 0xE9, # ARABIC LETTER ALEF MAKSURA - 0x064A: 0xEA, # ARABIC LETTER YEH - 0x064B: 0xEB, # ARABIC FATHATAN - 0x064C: 0xEC, # ARABIC DAMMATAN - 0x064D: 0xED, # ARABIC KASRATAN - 0x064E: 0xEE, # ARABIC FATHA - 0x064F: 0xEF, # ARABIC DAMMA - 0x0650: 0xF0, # ARABIC KASRA - 0x0651: 0xF1, # ARABIC SHADDA - 0x0652: 0xF2, # ARABIC SUKUN -} diff --git a/Lib/encodings/iso8859_7.py b/Lib/encodings/iso8859_7.py index 4cce6e2..2152194 100644 --- a/Lib/encodings/iso8859_7.py +++ b/Lib/encodings/iso8859_7.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,260 +303,6 @@ decoding_table = ( u'\ufffe' ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 0x00A3: 0xA3, # POUND SIGN - 0x00A6: 0xA6, # BROKEN BAR - 0x00A7: 0xA7, # SECTION SIGN - 0x00A8: 0xA8, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xAB, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xAC, # NOT SIGN - 0x00AD: 0xAD, # SOFT HYPHEN - 0x00B0: 0xB0, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B2: 0xB2, # SUPERSCRIPT TWO - 0x00B3: 0xB3, # SUPERSCRIPT THREE - 0x00B7: 0xB7, # MIDDLE DOT - 0x00BB: 0xBB, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BD: 0xBD, # VULGAR FRACTION ONE HALF - 0x037A: 0xAA, # GREEK YPOGEGRAMMENI - 0x0384: 0xB4, # GREEK TONOS - 0x0385: 0xB5, # GREEK DIALYTIKA TONOS - 0x0386: 0xB6, # GREEK CAPITAL LETTER ALPHA WITH TONOS - 0x0388: 0xB8, # GREEK CAPITAL LETTER EPSILON WITH TONOS - 0x0389: 0xB9, # GREEK CAPITAL LETTER ETA WITH TONOS - 0x038A: 0xBA, # GREEK CAPITAL LETTER IOTA WITH TONOS - 0x038C: 0xBC, # GREEK CAPITAL LETTER OMICRON WITH TONOS - 0x038E: 0xBE, # GREEK CAPITAL LETTER UPSILON WITH TONOS - 0x038F: 0xBF, # GREEK CAPITAL LETTER OMEGA WITH TONOS - 0x0390: 0xC0, # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - 0x0391: 0xC1, # GREEK CAPITAL LETTER ALPHA - 0x0392: 0xC2, # GREEK CAPITAL LETTER BETA - 0x0393: 0xC3, # GREEK CAPITAL LETTER GAMMA - 0x0394: 0xC4, # GREEK CAPITAL LETTER DELTA - 0x0395: 0xC5, # GREEK CAPITAL LETTER EPSILON - 0x0396: 0xC6, # GREEK CAPITAL LETTER ZETA - 0x0397: 0xC7, # GREEK CAPITAL LETTER ETA - 0x0398: 0xC8, # GREEK CAPITAL LETTER THETA - 0x0399: 0xC9, # GREEK CAPITAL LETTER IOTA - 0x039A: 0xCA, # GREEK CAPITAL LETTER KAPPA - 0x039B: 0xCB, # GREEK CAPITAL LETTER LAMDA - 0x039C: 0xCC, # GREEK CAPITAL LETTER MU - 0x039D: 0xCD, # GREEK CAPITAL LETTER NU - 0x039E: 0xCE, # GREEK CAPITAL LETTER XI - 0x039F: 0xCF, # GREEK CAPITAL LETTER OMICRON - 0x03A0: 0xD0, # GREEK CAPITAL LETTER PI - 0x03A1: 0xD1, # GREEK CAPITAL LETTER RHO - 0x03A3: 0xD3, # GREEK CAPITAL LETTER SIGMA - 0x03A4: 0xD4, # GREEK CAPITAL LETTER TAU - 0x03A5: 0xD5, # GREEK CAPITAL LETTER UPSILON - 0x03A6: 0xD6, # GREEK CAPITAL LETTER PHI - 0x03A7: 0xD7, # GREEK CAPITAL LETTER CHI - 0x03A8: 0xD8, # GREEK CAPITAL LETTER PSI - 0x03A9: 0xD9, # GREEK CAPITAL LETTER OMEGA - 0x03AA: 0xDA, # GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - 0x03AB: 0xDB, # GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - 0x03AC: 0xDC, # GREEK SMALL LETTER ALPHA WITH TONOS - 0x03AD: 0xDD, # GREEK SMALL LETTER EPSILON WITH TONOS - 0x03AE: 0xDE, # GREEK SMALL LETTER ETA WITH TONOS - 0x03AF: 0xDF, # GREEK SMALL LETTER IOTA WITH TONOS - 0x03B0: 0xE0, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - 0x03B1: 0xE1, # GREEK SMALL LETTER ALPHA - 0x03B2: 0xE2, # GREEK SMALL LETTER BETA - 0x03B3: 0xE3, # GREEK SMALL LETTER GAMMA - 0x03B4: 0xE4, # GREEK SMALL LETTER DELTA - 0x03B5: 0xE5, # GREEK SMALL LETTER EPSILON - 0x03B6: 0xE6, # GREEK SMALL LETTER ZETA - 0x03B7: 0xE7, # GREEK SMALL LETTER ETA - 0x03B8: 0xE8, # GREEK SMALL LETTER THETA - 0x03B9: 0xE9, # GREEK SMALL LETTER IOTA - 0x03BA: 0xEA, # GREEK SMALL LETTER KAPPA - 0x03BB: 0xEB, # GREEK SMALL LETTER LAMDA - 0x03BC: 0xEC, # GREEK SMALL LETTER MU - 0x03BD: 0xED, # GREEK SMALL LETTER NU - 0x03BE: 0xEE, # GREEK SMALL LETTER XI - 0x03BF: 0xEF, # GREEK SMALL LETTER OMICRON - 0x03C0: 0xF0, # GREEK SMALL LETTER PI - 0x03C1: 0xF1, # GREEK SMALL LETTER RHO - 0x03C2: 0xF2, # GREEK SMALL LETTER FINAL SIGMA - 0x03C3: 0xF3, # GREEK SMALL LETTER SIGMA - 0x03C4: 0xF4, # GREEK SMALL LETTER TAU - 0x03C5: 0xF5, # GREEK SMALL LETTER UPSILON - 0x03C6: 0xF6, # GREEK SMALL LETTER PHI - 0x03C7: 0xF7, # GREEK SMALL LETTER CHI - 0x03C8: 0xF8, # GREEK SMALL LETTER PSI - 0x03C9: 0xF9, # GREEK SMALL LETTER OMEGA - 0x03CA: 0xFA, # GREEK SMALL LETTER IOTA WITH DIALYTIKA - 0x03CB: 0xFB, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA - 0x03CC: 0xFC, # GREEK SMALL LETTER OMICRON WITH TONOS - 0x03CD: 0xFD, # GREEK SMALL LETTER UPSILON WITH TONOS - 0x03CE: 0xFE, # GREEK SMALL LETTER OMEGA WITH TONOS - 0x2015: 0xAF, # HORIZONTAL BAR - 0x2018: 0xA1, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xA2, # RIGHT SINGLE QUOTATION MARK - 0x20AC: 0xA4, # EURO SIGN - 0x20AF: 0xA5, # DRACHMA SIGN -} diff --git a/Lib/encodings/iso8859_8.py b/Lib/encodings/iso8859_8.py index 8c29a87..053a5f5 100644 --- a/Lib/encodings/iso8859_8.py +++ b/Lib/encodings/iso8859_8.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,227 +303,6 @@ decoding_table = ( u'\ufffe' ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x00A0: 0xA0, # NO-BREAK SPACE - 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 - 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 - 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 - 0x00D7: 0xAA, # MULTIPLICATION SIGN - 0x00F7: 0xBA, # DIVISION SIGN - 0x05D0: 0xE0, # HEBREW LETTER ALEF - 0x05D1: 0xE1, # HEBREW LETTER BET - 0x05D2: 0xE2, # HEBREW LETTER GIMEL - 0x05D3: 0xE3, # HEBREW LETTER DALET - 0x05D4: 0xE4, # HEBREW LETTER HE - 0x05D5: 0xE5, # HEBREW LETTER VAV - 0x05D6: 0xE6, # HEBREW LETTER ZAYIN - 0x05D7: 0xE7, # HEBREW LETTER HET - 0x05D8: 0xE8, # HEBREW LETTER TET - 0x05D9: 0xE9, # HEBREW LETTER YOD - 0x05DA: 0xEA, # HEBREW LETTER FINAL KAF - 0x05DB: 0xEB, # HEBREW LETTER KAF - 0x05DC: 0xEC, # HEBREW LETTER LAMED - 0x05DD: 0xED, # HEBREW LETTER FINAL MEM - 0x05DE: 0xEE, # HEBREW LETTER MEM - 0x05DF: 0xEF, # HEBREW LETTER FINAL NUN - 0x05E0: 0xF0, # HEBREW LETTER NUN - 0x05E1: 0xF1, # HEBREW LETTER SAMEKH - 0x05E2: 0xF2, # HEBREW LETTER AYIN - 0x05E3: 0xF3, # HEBREW LETTER FINAL PE - 0x05E4: 0xF4, # HEBREW LETTER PE - 0x05E5: 0xF5, # HEBREW LETTER FINAL TSADI - 0x05E6: 0xF6, # HEBREW LETTER TSADI - 0x05E7: 0xF7, # HEBREW LETTER QOF - 0x05E8: 0xF8, # HEBREW LETTER RESH - 0x05E9: 0xF9, # HEBREW LETTER SHIN - 0x05EA: 0xFA, # HEBREW LETTER TAV - 0x200E: 0xFD, # LEFT-TO-RIGHT MARK - 0x200F: 0xFE, # RIGHT-TO-LEFT MARK - 0x2017: 0xDF, # DOUBLE LOW LINE -} diff --git a/Lib/encodings/iso8859_9.py b/Lib/encodings/iso8859_9.py index 9648e9f..8c3b3bc 100644 --- a/Lib/encodings/iso8859_9.py +++ b/Lib/encodings/iso8859_9.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\xff' # 0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 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 - 0x015E: 0xDE, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015F: 0xFE, # LATIN SMALL LETTER S WITH CEDILLA -} diff --git a/Lib/encodings/koi8_r.py b/Lib/encodings/koi8_r.py index 3efeb56..7295260 100644 --- a/Lib/encodings/koi8_r.py +++ b/Lib/encodings/koi8_r.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0x9A, # NO-BREAK SPACE - 0x00A9: 0xBF, # COPYRIGHT SIGN - 0x00B0: 0x9C, # DEGREE SIGN - 0x00B2: 0x9D, # SUPERSCRIPT TWO - 0x00B7: 0x9E, # MIDDLE DOT - 0x00F7: 0x9F, # DIVISION SIGN - 0x0401: 0xB3, # CYRILLIC CAPITAL LETTER IO - 0x0410: 0xE1, # CYRILLIC CAPITAL LETTER A - 0x0411: 0xE2, # CYRILLIC CAPITAL LETTER BE - 0x0412: 0xF7, # CYRILLIC CAPITAL LETTER VE - 0x0413: 0xE7, # CYRILLIC CAPITAL LETTER GHE - 0x0414: 0xE4, # CYRILLIC CAPITAL LETTER DE - 0x0415: 0xE5, # CYRILLIC CAPITAL LETTER IE - 0x0416: 0xF6, # CYRILLIC CAPITAL LETTER ZHE - 0x0417: 0xFA, # CYRILLIC CAPITAL LETTER ZE - 0x0418: 0xE9, # CYRILLIC CAPITAL LETTER I - 0x0419: 0xEA, # CYRILLIC CAPITAL LETTER SHORT I - 0x041A: 0xEB, # CYRILLIC CAPITAL LETTER KA - 0x041B: 0xEC, # CYRILLIC CAPITAL LETTER EL - 0x041C: 0xED, # CYRILLIC CAPITAL LETTER EM - 0x041D: 0xEE, # CYRILLIC CAPITAL LETTER EN - 0x041E: 0xEF, # CYRILLIC CAPITAL LETTER O - 0x041F: 0xF0, # CYRILLIC CAPITAL LETTER PE - 0x0420: 0xF2, # CYRILLIC CAPITAL LETTER ER - 0x0421: 0xF3, # CYRILLIC CAPITAL LETTER ES - 0x0422: 0xF4, # CYRILLIC CAPITAL LETTER TE - 0x0423: 0xF5, # CYRILLIC CAPITAL LETTER U - 0x0424: 0xE6, # CYRILLIC CAPITAL LETTER EF - 0x0425: 0xE8, # CYRILLIC CAPITAL LETTER HA - 0x0426: 0xE3, # CYRILLIC CAPITAL LETTER TSE - 0x0427: 0xFE, # CYRILLIC CAPITAL LETTER CHE - 0x0428: 0xFB, # CYRILLIC CAPITAL LETTER SHA - 0x0429: 0xFD, # CYRILLIC CAPITAL LETTER SHCHA - 0x042A: 0xFF, # CYRILLIC CAPITAL LETTER HARD SIGN - 0x042B: 0xF9, # CYRILLIC CAPITAL LETTER YERU - 0x042C: 0xF8, # CYRILLIC CAPITAL LETTER SOFT SIGN - 0x042D: 0xFC, # CYRILLIC CAPITAL LETTER E - 0x042E: 0xE0, # CYRILLIC CAPITAL LETTER YU - 0x042F: 0xF1, # CYRILLIC CAPITAL LETTER YA - 0x0430: 0xC1, # CYRILLIC SMALL LETTER A - 0x0431: 0xC2, # CYRILLIC SMALL LETTER BE - 0x0432: 0xD7, # CYRILLIC SMALL LETTER VE - 0x0433: 0xC7, # CYRILLIC SMALL LETTER GHE - 0x0434: 0xC4, # CYRILLIC SMALL LETTER DE - 0x0435: 0xC5, # CYRILLIC SMALL LETTER IE - 0x0436: 0xD6, # CYRILLIC SMALL LETTER ZHE - 0x0437: 0xDA, # CYRILLIC SMALL LETTER ZE - 0x0438: 0xC9, # CYRILLIC SMALL LETTER I - 0x0439: 0xCA, # CYRILLIC SMALL LETTER SHORT I - 0x043A: 0xCB, # CYRILLIC SMALL LETTER KA - 0x043B: 0xCC, # CYRILLIC SMALL LETTER EL - 0x043C: 0xCD, # CYRILLIC SMALL LETTER EM - 0x043D: 0xCE, # CYRILLIC SMALL LETTER EN - 0x043E: 0xCF, # CYRILLIC SMALL LETTER O - 0x043F: 0xD0, # CYRILLIC SMALL LETTER PE - 0x0440: 0xD2, # CYRILLIC SMALL LETTER ER - 0x0441: 0xD3, # CYRILLIC SMALL LETTER ES - 0x0442: 0xD4, # CYRILLIC SMALL LETTER TE - 0x0443: 0xD5, # CYRILLIC SMALL LETTER U - 0x0444: 0xC6, # CYRILLIC SMALL LETTER EF - 0x0445: 0xC8, # CYRILLIC SMALL LETTER HA - 0x0446: 0xC3, # CYRILLIC SMALL LETTER TSE - 0x0447: 0xDE, # CYRILLIC SMALL LETTER CHE - 0x0448: 0xDB, # CYRILLIC SMALL LETTER SHA - 0x0449: 0xDD, # CYRILLIC SMALL LETTER SHCHA - 0x044A: 0xDF, # CYRILLIC SMALL LETTER HARD SIGN - 0x044B: 0xD9, # CYRILLIC SMALL LETTER YERU - 0x044C: 0xD8, # CYRILLIC SMALL LETTER SOFT SIGN - 0x044D: 0xDC, # CYRILLIC SMALL LETTER E - 0x044E: 0xC0, # CYRILLIC SMALL LETTER YU - 0x044F: 0xD1, # CYRILLIC SMALL LETTER YA - 0x0451: 0xA3, # CYRILLIC SMALL LETTER IO - 0x2219: 0x95, # BULLET OPERATOR - 0x221A: 0x96, # SQUARE ROOT - 0x2248: 0x97, # ALMOST EQUAL TO - 0x2264: 0x98, # LESS-THAN OR EQUAL TO - 0x2265: 0x99, # GREATER-THAN OR EQUAL TO - 0x2320: 0x93, # TOP HALF INTEGRAL - 0x2321: 0x9B, # BOTTOM HALF INTEGRAL - 0x2500: 0x80, # BOX DRAWINGS LIGHT HORIZONTAL - 0x2502: 0x81, # BOX DRAWINGS LIGHT VERTICAL - 0x250C: 0x82, # BOX DRAWINGS LIGHT DOWN AND RIGHT - 0x2510: 0x83, # BOX DRAWINGS LIGHT DOWN AND LEFT - 0x2514: 0x84, # BOX DRAWINGS LIGHT UP AND RIGHT - 0x2518: 0x85, # BOX DRAWINGS LIGHT UP AND LEFT - 0x251C: 0x86, # BOX DRAWINGS LIGHT VERTICAL AND RIGHT - 0x2524: 0x87, # BOX DRAWINGS LIGHT VERTICAL AND LEFT - 0x252C: 0x88, # BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - 0x2534: 0x89, # BOX DRAWINGS LIGHT UP AND HORIZONTAL - 0x253C: 0x8A, # BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - 0x2550: 0xA0, # BOX DRAWINGS DOUBLE HORIZONTAL - 0x2551: 0xA1, # BOX DRAWINGS DOUBLE VERTICAL - 0x2552: 0xA2, # BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - 0x2553: 0xA4, # BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE - 0x2554: 0xA5, # BOX DRAWINGS DOUBLE DOWN AND RIGHT - 0x2555: 0xA6, # BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE - 0x2556: 0xA7, # BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE - 0x2557: 0xA8, # BOX DRAWINGS DOUBLE DOWN AND LEFT - 0x2558: 0xA9, # BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - 0x2559: 0xAA, # BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - 0x255A: 0xAB, # BOX DRAWINGS DOUBLE UP AND RIGHT - 0x255B: 0xAC, # BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - 0x255C: 0xAD, # BOX DRAWINGS UP DOUBLE AND LEFT SINGLE - 0x255D: 0xAE, # BOX DRAWINGS DOUBLE UP AND LEFT - 0x255E: 0xAF, # BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - 0x255F: 0xB0, # BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - 0x2560: 0xB1, # BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - 0x2561: 0xB2, # BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - 0x2562: 0xB4, # BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE - 0x2563: 0xB5, # BOX DRAWINGS DOUBLE VERTICAL AND LEFT - 0x2564: 0xB6, # BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE - 0x2565: 0xB7, # BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE - 0x2566: 0xB8, # BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - 0x2567: 0xB9, # BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - 0x2568: 0xBA, # BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - 0x2569: 0xBB, # BOX DRAWINGS DOUBLE UP AND HORIZONTAL - 0x256A: 0xBC, # BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - 0x256B: 0xBD, # BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE - 0x256C: 0xBE, # BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - 0x2580: 0x8B, # UPPER HALF BLOCK - 0x2584: 0x8C, # LOWER HALF BLOCK - 0x2588: 0x8D, # FULL BLOCK - 0x258C: 0x8E, # LEFT HALF BLOCK - 0x2590: 0x8F, # RIGHT HALF BLOCK - 0x2591: 0x90, # LIGHT SHADE - 0x2592: 0x91, # MEDIUM SHADE - 0x2593: 0x92, # DARK SHADE - 0x25A0: 0x94, # BLACK SQUARE -} diff --git a/Lib/encodings/koi8_u.py b/Lib/encodings/koi8_u.py index 5f46db1..ce63d97 100644 --- a/Lib/encodings/koi8_u.py +++ b/Lib/encodings/koi8_u.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x00A0: 0x9A, # NO-BREAK SPACE - 0x00A9: 0xBF, # COPYRIGHT SIGN - 0x00B0: 0x9C, # DEGREE SIGN - 0x00B2: 0x9D, # SUPERSCRIPT TWO - 0x00B7: 0x9E, # MIDDLE DOT - 0x00F7: 0x9F, # DIVISION SIGN - 0x0401: 0xB3, # CYRILLIC CAPITAL LETTER IO - 0x0404: 0xB4, # CYRILLIC CAPITAL LETTER UKRAINIAN IE - 0x0406: 0xB6, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0407: 0xB7, # CYRILLIC CAPITAL LETTER YI (UKRAINIAN) - 0x0410: 0xE1, # CYRILLIC CAPITAL LETTER A - 0x0411: 0xE2, # CYRILLIC CAPITAL LETTER BE - 0x0412: 0xF7, # CYRILLIC CAPITAL LETTER VE - 0x0413: 0xE7, # CYRILLIC CAPITAL LETTER GHE - 0x0414: 0xE4, # CYRILLIC CAPITAL LETTER DE - 0x0415: 0xE5, # CYRILLIC CAPITAL LETTER IE - 0x0416: 0xF6, # CYRILLIC CAPITAL LETTER ZHE - 0x0417: 0xFA, # CYRILLIC CAPITAL LETTER ZE - 0x0418: 0xE9, # CYRILLIC CAPITAL LETTER I - 0x0419: 0xEA, # CYRILLIC CAPITAL LETTER SHORT I - 0x041A: 0xEB, # CYRILLIC CAPITAL LETTER KA - 0x041B: 0xEC, # CYRILLIC CAPITAL LETTER EL - 0x041C: 0xED, # CYRILLIC CAPITAL LETTER EM - 0x041D: 0xEE, # CYRILLIC CAPITAL LETTER EN - 0x041E: 0xEF, # CYRILLIC CAPITAL LETTER O - 0x041F: 0xF0, # CYRILLIC CAPITAL LETTER PE - 0x0420: 0xF2, # CYRILLIC CAPITAL LETTER ER - 0x0421: 0xF3, # CYRILLIC CAPITAL LETTER ES - 0x0422: 0xF4, # CYRILLIC CAPITAL LETTER TE - 0x0423: 0xF5, # CYRILLIC CAPITAL LETTER U - 0x0424: 0xE6, # CYRILLIC CAPITAL LETTER EF - 0x0425: 0xE8, # CYRILLIC CAPITAL LETTER HA - 0x0426: 0xE3, # CYRILLIC CAPITAL LETTER TSE - 0x0427: 0xFE, # CYRILLIC CAPITAL LETTER CHE - 0x0428: 0xFB, # CYRILLIC CAPITAL LETTER SHA - 0x0429: 0xFD, # CYRILLIC CAPITAL LETTER SHCHA - 0x042A: 0xFF, # CYRILLIC CAPITAL LETTER HARD SIGN - 0x042B: 0xF9, # CYRILLIC CAPITAL LETTER YERU - 0x042C: 0xF8, # CYRILLIC CAPITAL LETTER SOFT SIGN - 0x042D: 0xFC, # CYRILLIC CAPITAL LETTER E - 0x042E: 0xE0, # CYRILLIC CAPITAL LETTER YU - 0x042F: 0xF1, # CYRILLIC CAPITAL LETTER YA - 0x0430: 0xC1, # CYRILLIC SMALL LETTER A - 0x0431: 0xC2, # CYRILLIC SMALL LETTER BE - 0x0432: 0xD7, # CYRILLIC SMALL LETTER VE - 0x0433: 0xC7, # CYRILLIC SMALL LETTER GHE - 0x0434: 0xC4, # CYRILLIC SMALL LETTER DE - 0x0435: 0xC5, # CYRILLIC SMALL LETTER IE - 0x0436: 0xD6, # CYRILLIC SMALL LETTER ZHE - 0x0437: 0xDA, # CYRILLIC SMALL LETTER ZE - 0x0438: 0xC9, # CYRILLIC SMALL LETTER I - 0x0439: 0xCA, # CYRILLIC SMALL LETTER SHORT I - 0x043A: 0xCB, # CYRILLIC SMALL LETTER KA - 0x043B: 0xCC, # CYRILLIC SMALL LETTER EL - 0x043C: 0xCD, # CYRILLIC SMALL LETTER EM - 0x043D: 0xCE, # CYRILLIC SMALL LETTER EN - 0x043E: 0xCF, # CYRILLIC SMALL LETTER O - 0x043F: 0xD0, # CYRILLIC SMALL LETTER PE - 0x0440: 0xD2, # CYRILLIC SMALL LETTER ER - 0x0441: 0xD3, # CYRILLIC SMALL LETTER ES - 0x0442: 0xD4, # CYRILLIC SMALL LETTER TE - 0x0443: 0xD5, # CYRILLIC SMALL LETTER U - 0x0444: 0xC6, # CYRILLIC SMALL LETTER EF - 0x0445: 0xC8, # CYRILLIC SMALL LETTER HA - 0x0446: 0xC3, # CYRILLIC SMALL LETTER TSE - 0x0447: 0xDE, # CYRILLIC SMALL LETTER CHE - 0x0448: 0xDB, # CYRILLIC SMALL LETTER SHA - 0x0449: 0xDD, # CYRILLIC SMALL LETTER SHCHA - 0x044A: 0xDF, # CYRILLIC SMALL LETTER HARD SIGN - 0x044B: 0xD9, # CYRILLIC SMALL LETTER YERU - 0x044C: 0xD8, # CYRILLIC SMALL LETTER SOFT SIGN - 0x044D: 0xDC, # CYRILLIC SMALL LETTER E - 0x044E: 0xC0, # CYRILLIC SMALL LETTER YU - 0x044F: 0xD1, # CYRILLIC SMALL LETTER YA - 0x0451: 0xA3, # CYRILLIC SMALL LETTER IO - 0x0454: 0xA4, # CYRILLIC SMALL LETTER UKRAINIAN IE - 0x0456: 0xA6, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I - 0x0457: 0xA7, # CYRILLIC SMALL LETTER YI (UKRAINIAN) - 0x0490: 0xBD, # CYRILLIC CAPITAL LETTER UKRAINIAN GHE WITH UPTURN - 0x0491: 0xAD, # CYRILLIC SMALL LETTER UKRAINIAN GHE WITH UPTURN - 0x2219: 0x95, # BULLET OPERATOR - 0x221A: 0x96, # SQUARE ROOT - 0x2248: 0x97, # ALMOST EQUAL TO - 0x2264: 0x98, # LESS-THAN OR EQUAL TO - 0x2265: 0x99, # GREATER-THAN OR EQUAL TO - 0x2320: 0x93, # TOP HALF INTEGRAL - 0x2321: 0x9B, # BOTTOM HALF INTEGRAL - 0x2500: 0x80, # BOX DRAWINGS LIGHT HORIZONTAL - 0x2502: 0x81, # BOX DRAWINGS LIGHT VERTICAL - 0x250C: 0x82, # BOX DRAWINGS LIGHT DOWN AND RIGHT - 0x2510: 0x83, # BOX DRAWINGS LIGHT DOWN AND LEFT - 0x2514: 0x84, # BOX DRAWINGS LIGHT UP AND RIGHT - 0x2518: 0x85, # BOX DRAWINGS LIGHT UP AND LEFT - 0x251C: 0x86, # BOX DRAWINGS LIGHT VERTICAL AND RIGHT - 0x2524: 0x87, # BOX DRAWINGS LIGHT VERTICAL AND LEFT - 0x252C: 0x88, # BOX DRAWINGS LIGHT DOWN AND HORIZONTAL - 0x2534: 0x89, # BOX DRAWINGS LIGHT UP AND HORIZONTAL - 0x253C: 0x8A, # BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL - 0x2550: 0xA0, # BOX DRAWINGS DOUBLE HORIZONTAL - 0x2551: 0xA1, # BOX DRAWINGS DOUBLE VERTICAL - 0x2552: 0xA2, # BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE - 0x2554: 0xA5, # BOX DRAWINGS DOUBLE DOWN AND RIGHT - 0x2557: 0xA8, # BOX DRAWINGS DOUBLE DOWN AND LEFT - 0x2558: 0xA9, # BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE - 0x2559: 0xAA, # BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE - 0x255A: 0xAB, # BOX DRAWINGS DOUBLE UP AND RIGHT - 0x255B: 0xAC, # BOX DRAWINGS UP SINGLE AND LEFT DOUBLE - 0x255D: 0xAE, # BOX DRAWINGS DOUBLE UP AND LEFT - 0x255E: 0xAF, # BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE - 0x255F: 0xB0, # BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE - 0x2560: 0xB1, # BOX DRAWINGS DOUBLE VERTICAL AND RIGHT - 0x2561: 0xB2, # BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE - 0x2563: 0xB5, # BOX DRAWINGS DOUBLE VERTICAL AND LEFT - 0x2566: 0xB8, # BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL - 0x2567: 0xB9, # BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE - 0x2568: 0xBA, # BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE - 0x2569: 0xBB, # BOX DRAWINGS DOUBLE UP AND HORIZONTAL - 0x256A: 0xBC, # BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE - 0x256C: 0xBE, # BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL - 0x2580: 0x8B, # UPPER HALF BLOCK - 0x2584: 0x8C, # LOWER HALF BLOCK - 0x2588: 0x8D, # FULL BLOCK - 0x258C: 0x8E, # LEFT HALF BLOCK - 0x2590: 0x8F, # RIGHT HALF BLOCK - 0x2591: 0x90, # LIGHT SHADE - 0x2592: 0x91, # MEDIUM SHADE - 0x2593: 0x92, # DARK SHADE - 0x25A0: 0x94, # BLACK SQUARE -} diff --git a/Lib/encodings/mac_centeuro.py b/Lib/encodings/mac_centeuro.py index 54a1510..0ceba55 100644 --- a/Lib/encodings/mac_centeuro.py +++ b/Lib/encodings/mac_centeuro.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02c7' # 0xFF -> CARON ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 - 0x00A8: 0xAC, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00B0: 0xA1, # DEGREE SIGN - 0x00B6: 0xA6, # PILCROW SIGN - 0x00BB: 0xC8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00C1: 0xE7, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CD: 0xEA, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEF, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xCD, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00DA: 0xF2, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xF8, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0x9B, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0xF9, # LATIN SMALL LETTER Y WITH ACUTE - 0x0100: 0x81, # LATIN CAPITAL LETTER A WITH MACRON - 0x0101: 0x82, # LATIN SMALL LETTER A WITH MACRON - 0x0104: 0x84, # LATIN CAPITAL LETTER A WITH OGONEK - 0x0105: 0x88, # LATIN SMALL LETTER A WITH OGONEK - 0x0106: 0x8C, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0x8D, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0x89, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0x8B, # LATIN SMALL LETTER C WITH CARON - 0x010E: 0x91, # LATIN CAPITAL LETTER D WITH CARON - 0x010F: 0x93, # LATIN SMALL LETTER D WITH CARON - 0x0112: 0x94, # LATIN CAPITAL LETTER E WITH MACRON - 0x0113: 0x95, # LATIN SMALL LETTER E WITH MACRON - 0x0116: 0x96, # LATIN CAPITAL LETTER E WITH DOT ABOVE - 0x0117: 0x98, # LATIN SMALL LETTER E WITH DOT ABOVE - 0x0118: 0xA2, # LATIN CAPITAL LETTER E WITH OGONEK - 0x0119: 0xAB, # LATIN SMALL LETTER E WITH OGONEK - 0x011A: 0x9D, # LATIN CAPITAL LETTER E WITH CARON - 0x011B: 0x9E, # LATIN SMALL LETTER E WITH CARON - 0x0122: 0xFE, # LATIN CAPITAL LETTER G WITH CEDILLA - 0x0123: 0xAE, # LATIN SMALL LETTER G WITH CEDILLA - 0x012A: 0xB1, # LATIN CAPITAL LETTER I WITH MACRON - 0x012B: 0xB4, # LATIN SMALL LETTER I WITH MACRON - 0x012E: 0xAF, # LATIN CAPITAL LETTER I WITH OGONEK - 0x012F: 0xB0, # LATIN SMALL LETTER I WITH OGONEK - 0x0136: 0xB5, # LATIN CAPITAL LETTER K WITH CEDILLA - 0x0137: 0xFA, # LATIN SMALL LETTER K WITH CEDILLA - 0x0139: 0xBD, # LATIN CAPITAL LETTER L WITH ACUTE - 0x013A: 0xBE, # LATIN SMALL LETTER L WITH ACUTE - 0x013B: 0xB9, # LATIN CAPITAL LETTER L WITH CEDILLA - 0x013C: 0xBA, # LATIN SMALL LETTER L WITH CEDILLA - 0x013D: 0xBB, # LATIN CAPITAL LETTER L WITH CARON - 0x013E: 0xBC, # LATIN SMALL LETTER L WITH CARON - 0x0141: 0xFC, # LATIN CAPITAL LETTER L WITH STROKE - 0x0142: 0xB8, # LATIN SMALL LETTER L WITH STROKE - 0x0143: 0xC1, # LATIN CAPITAL LETTER N WITH ACUTE - 0x0144: 0xC4, # LATIN SMALL LETTER N WITH ACUTE - 0x0145: 0xBF, # LATIN CAPITAL LETTER N WITH CEDILLA - 0x0146: 0xC0, # LATIN SMALL LETTER N WITH CEDILLA - 0x0147: 0xC5, # LATIN CAPITAL LETTER N WITH CARON - 0x0148: 0xCB, # LATIN SMALL LETTER N WITH CARON - 0x014C: 0xCF, # LATIN CAPITAL LETTER O WITH MACRON - 0x014D: 0xD8, # LATIN SMALL LETTER O WITH MACRON - 0x0150: 0xCC, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE - 0x0151: 0xCE, # LATIN SMALL LETTER O WITH DOUBLE ACUTE - 0x0154: 0xD9, # LATIN CAPITAL LETTER R WITH ACUTE - 0x0155: 0xDA, # LATIN SMALL LETTER R WITH ACUTE - 0x0156: 0xDF, # LATIN CAPITAL LETTER R WITH CEDILLA - 0x0157: 0xE0, # LATIN SMALL LETTER R WITH CEDILLA - 0x0158: 0xDB, # LATIN CAPITAL LETTER R WITH CARON - 0x0159: 0xDE, # LATIN SMALL LETTER R WITH CARON - 0x015A: 0xE5, # LATIN CAPITAL LETTER S WITH ACUTE - 0x015B: 0xE6, # LATIN SMALL LETTER S WITH ACUTE - 0x0160: 0xE1, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xE4, # LATIN SMALL LETTER S WITH CARON - 0x0164: 0xE8, # LATIN CAPITAL LETTER T WITH CARON - 0x0165: 0xE9, # LATIN SMALL LETTER T WITH CARON - 0x016A: 0xED, # LATIN CAPITAL LETTER U WITH MACRON - 0x016B: 0xF0, # LATIN SMALL LETTER U WITH MACRON - 0x016E: 0xF1, # LATIN CAPITAL LETTER U WITH RING ABOVE - 0x016F: 0xF3, # LATIN SMALL LETTER U WITH RING ABOVE - 0x0170: 0xF4, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE - 0x0171: 0xF5, # LATIN SMALL LETTER U WITH DOUBLE ACUTE - 0x0172: 0xF6, # LATIN CAPITAL LETTER U WITH OGONEK - 0x0173: 0xF7, # LATIN SMALL LETTER U WITH OGONEK - 0x0179: 0x8F, # LATIN CAPITAL LETTER Z WITH ACUTE - 0x017A: 0x90, # LATIN SMALL LETTER Z WITH ACUTE - 0x017B: 0xFB, # LATIN CAPITAL LETTER Z WITH DOT ABOVE - 0x017C: 0xFD, # LATIN SMALL LETTER Z WITH DOT ABOVE - 0x017D: 0xEB, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xEC, # LATIN SMALL LETTER Z WITH CARON - 0x02C7: 0xFF, # CARON - 0x2013: 0xD0, # EN DASH - 0x2014: 0xD1, # EM DASH - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201A: 0xE2, # SINGLE LOW-9 QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xE3, # DOUBLE LOW-9 QUOTATION MARK - 0x2020: 0xA0, # DAGGER - 0x2022: 0xA5, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2039: 0xDC, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK - 0x203A: 0xDD, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - 0x2122: 0xAA, # TRADE MARK SIGN - 0x2202: 0xB6, # PARTIAL DIFFERENTIAL - 0x2206: 0xC6, # INCREMENT - 0x2211: 0xB7, # N-ARY SUMMATION - 0x221A: 0xC3, # SQUARE ROOT - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO - 0x25CA: 0xD7, # LOZENGE -} diff --git a/Lib/encodings/mac_croatian.py b/Lib/encodings/mac_croatian.py index 9e93cdd..535c5ff 100644 --- a/Lib/encodings/mac_croatian.py +++ b/Lib/encodings/mac_croatian.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02c7' # 0xFF -> CARON ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 - 0x00A1: 0xC1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A7: 0xA4, # SECTION SIGN - 0x00A8: 0xAC, # DIAERESIS - 0x00A9: 0xD9, # COPYRIGHT SIGN - 0x00AA: 0xBB, # FEMININE ORDINAL INDICATOR - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00AF: 0xF8, # MACRON - 0x00B0: 0xA1, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B4: 0xAB, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xA6, # PILCROW SIGN - 0x00B7: 0xE1, # MIDDLE DOT - 0x00B8: 0xFC, # CEDILLA - 0x00BA: 0xBC, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0xDF, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BF: 0xC0, # INVERTED QUESTION MARK - 0x00C0: 0xCB, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0xE7, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xE5, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0xCC, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x81, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0xDE, # LATIN CAPITAL LETTER AE - 0x00C7: 0x82, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0xE9, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0xFD, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0xFA, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0xED, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0xEA, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xEB, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0xEC, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D1: 0x84, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xF1, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEF, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xCD, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D8: 0xAF, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xF4, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xF2, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xF3, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x8B, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x8C, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0xFE, # LATIN SMALL LETTER AE - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x93, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F1: 0x96, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0x98, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0x9B, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00F8: 0xBF, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x0106: 0xC6, # LATIN CAPITAL LETTER C WITH ACUTE - 0x0107: 0xE6, # LATIN SMALL LETTER C WITH ACUTE - 0x010C: 0xC8, # LATIN CAPITAL LETTER C WITH CARON - 0x010D: 0xE8, # LATIN SMALL LETTER C WITH CARON - 0x0110: 0xD0, # LATIN CAPITAL LETTER D WITH STROKE - 0x0111: 0xF0, # LATIN SMALL LETTER D WITH STROKE - 0x0131: 0xF5, # LATIN SMALL LETTER DOTLESS I - 0x0152: 0xCE, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xCF, # LATIN SMALL LIGATURE OE - 0x0160: 0xA9, # LATIN CAPITAL LETTER S WITH CARON - 0x0161: 0xB9, # LATIN SMALL LETTER S WITH CARON - 0x017D: 0xAE, # LATIN CAPITAL LETTER Z WITH CARON - 0x017E: 0xBE, # LATIN SMALL LETTER Z WITH CARON - 0x0192: 0xC4, # LATIN SMALL LETTER F WITH HOOK - 0x02C6: 0xF6, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02C7: 0xFF, # CARON - 0x02DA: 0xFB, # RING ABOVE - 0x02DC: 0xF7, # SMALL TILDE - 0x03A9: 0xBD, # GREEK CAPITAL LETTER OMEGA - 0x03C0: 0xF9, # GREEK SMALL LETTER PI - 0x2013: 0xE0, # EN DASH - 0x2014: 0xD1, # EM DASH - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201A: 0xE2, # SINGLE LOW-9 QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xE3, # DOUBLE LOW-9 QUOTATION MARK - 0x2020: 0xA0, # DAGGER - 0x2022: 0xA5, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2030: 0xE4, # PER MILLE SIGN - 0x2039: 0xDC, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK - 0x203A: 0xDD, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - 0x2044: 0xDA, # FRACTION SLASH - 0x20AC: 0xDB, # EURO SIGN - 0x2122: 0xAA, # TRADE MARK SIGN - 0x2202: 0xB6, # PARTIAL DIFFERENTIAL - 0x2206: 0xB4, # INCREMENT - 0x220F: 0xB8, # N-ARY PRODUCT - 0x2211: 0xB7, # N-ARY SUMMATION - 0x221A: 0xC3, # SQUARE ROOT - 0x221E: 0xB0, # INFINITY - 0x222B: 0xBA, # INTEGRAL - 0x2248: 0xC5, # ALMOST EQUAL TO - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO - 0x25CA: 0xD7, # LOZENGE - 0xF8FF: 0xD8, # Apple logo -} diff --git a/Lib/encodings/mac_cyrillic.py b/Lib/encodings/mac_cyrillic.py index 8ffd715..a56b8c4 100644 --- a/Lib/encodings/mac_cyrillic.py +++ b/Lib/encodings/mac_cyrillic.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u20ac' # 0xFF -> EURO SIGN ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 -} diff --git a/Lib/encodings/mac_farsi.py b/Lib/encodings/mac_farsi.py index 6d26a42..a6afd14 100644 --- a/Lib/encodings/mac_farsi.py +++ b/Lib/encodings/mac_farsi.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u06d2' # 0xFF -> ARABIC LETTER YEH BARREE ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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, left-right - 0x0020: 0xA0, # SPACE, right-left - 0x0021: 0x21, # EXCLAMATION MARK, left-right - 0x0021: 0xA1, # EXCLAMATION MARK, right-left - 0x0022: 0x22, # QUOTATION MARK, left-right - 0x0022: 0xA2, # QUOTATION MARK, right-left - 0x0023: 0x23, # NUMBER SIGN, left-right - 0x0023: 0xA3, # NUMBER SIGN, right-left - 0x0024: 0x24, # DOLLAR SIGN, left-right - 0x0024: 0xA4, # DOLLAR SIGN, right-left - 0x0025: 0x25, # PERCENT SIGN, left-right - 0x0026: 0x26, # AMPERSAND, left-right - 0x0026: 0xA6, # AMPERSAND, right-left - 0x0027: 0x27, # APOSTROPHE, left-right - 0x0027: 0xA7, # APOSTROPHE, right-left - 0x0028: 0x28, # LEFT PARENTHESIS, left-right - 0x0028: 0xA8, # LEFT PARENTHESIS, right-left - 0x0029: 0x29, # RIGHT PARENTHESIS, left-right - 0x0029: 0xA9, # RIGHT PARENTHESIS, right-left - 0x002A: 0x2A, # ASTERISK, left-right - 0x002A: 0xAA, # ASTERISK, right-left - 0x002B: 0x2B, # PLUS SIGN, left-right - 0x002B: 0xAB, # PLUS SIGN, right-left - 0x002C: 0x2C, # COMMA, left-right; in Arabic-script context, displayed as 0x066C ARABIC THOUSANDS SEPARATOR - 0x002D: 0x2D, # HYPHEN-MINUS, left-right - 0x002D: 0xAD, # HYPHEN-MINUS, right-left - 0x002E: 0x2E, # FULL STOP, left-right; in Arabic-script context, displayed as 0x066B ARABIC DECIMAL SEPARATOR - 0x002E: 0xAE, # FULL STOP, right-left - 0x002F: 0x2F, # SOLIDUS, left-right - 0x002F: 0xAF, # SOLIDUS, right-left - 0x0030: 0x30, # DIGIT ZERO; in Arabic-script context, displayed as 0x06F0 EXTENDED ARABIC-INDIC DIGIT ZERO - 0x0031: 0x31, # DIGIT ONE; in Arabic-script context, displayed as 0x06F1 EXTENDED ARABIC-INDIC DIGIT ONE - 0x0032: 0x32, # DIGIT TWO; in Arabic-script context, displayed as 0x06F2 EXTENDED ARABIC-INDIC DIGIT TWO - 0x0033: 0x33, # DIGIT THREE; in Arabic-script context, displayed as 0x06F3 EXTENDED ARABIC-INDIC DIGIT THREE - 0x0034: 0x34, # DIGIT FOUR; in Arabic-script context, displayed as 0x06F4 EXTENDED ARABIC-INDIC DIGIT FOUR - 0x0035: 0x35, # DIGIT FIVE; in Arabic-script context, displayed as 0x06F5 EXTENDED ARABIC-INDIC DIGIT FIVE - 0x0036: 0x36, # DIGIT SIX; in Arabic-script context, displayed as 0x06F6 EXTENDED ARABIC-INDIC DIGIT SIX - 0x0037: 0x37, # DIGIT SEVEN; in Arabic-script context, displayed as 0x06F7 EXTENDED ARABIC-INDIC DIGIT SEVEN - 0x0038: 0x38, # DIGIT EIGHT; in Arabic-script context, displayed as 0x06F8 EXTENDED ARABIC-INDIC DIGIT EIGHT - 0x0039: 0x39, # DIGIT NINE; in Arabic-script context, displayed as 0x06F9 EXTENDED ARABIC-INDIC DIGIT NINE - 0x003A: 0x3A, # COLON, left-right - 0x003A: 0xBA, # COLON, right-left - 0x003B: 0x3B, # SEMICOLON, left-right - 0x003C: 0x3C, # LESS-THAN SIGN, left-right - 0x003C: 0xBC, # LESS-THAN SIGN, right-left - 0x003D: 0x3D, # EQUALS SIGN, left-right - 0x003D: 0xBD, # EQUALS SIGN, right-left - 0x003E: 0x3E, # GREATER-THAN SIGN, left-right - 0x003E: 0xBE, # GREATER-THAN SIGN, right-left - 0x003F: 0x3F, # QUESTION MARK, left-right - 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, left-right - 0x005B: 0xDB, # LEFT SQUARE BRACKET, right-left - 0x005C: 0x5C, # REVERSE SOLIDUS, left-right - 0x005C: 0xDC, # REVERSE SOLIDUS, right-left - 0x005D: 0x5D, # RIGHT SQUARE BRACKET, left-right - 0x005D: 0xDD, # RIGHT SQUARE BRACKET, right-left - 0x005E: 0x5E, # CIRCUMFLEX ACCENT, left-right - 0x005E: 0xDE, # CIRCUMFLEX ACCENT, right-left - 0x005F: 0x5F, # LOW LINE, left-right - 0x005F: 0xDF, # LOW LINE, right-left - 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, left-right - 0x007B: 0xFB, # LEFT CURLY BRACKET, right-left - 0x007C: 0x7C, # VERTICAL LINE, left-right - 0x007C: 0xFC, # VERTICAL LINE, right-left - 0x007D: 0x7D, # RIGHT CURLY BRACKET, left-right - 0x007D: 0xFD, # RIGHT CURLY BRACKET, right-left - 0x007E: 0x7E, # TILDE - 0x007F: 0x7F, # CONTROL CHARACTER - 0x00A0: 0x81, # NO-BREAK SPACE, right-left - 0x00AB: 0x8C, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left - 0x00BB: 0x98, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK, right-left - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C7: 0x82, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00D1: 0x84, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F1: 0x96, # LATIN SMALL LETTER N WITH TILDE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0x9B, # DIVISION SIGN, right-left - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x060C: 0xAC, # ARABIC COMMA - 0x061B: 0xBB, # ARABIC SEMICOLON - 0x061F: 0xBF, # ARABIC QUESTION MARK - 0x0621: 0xC1, # ARABIC LETTER HAMZA - 0x0622: 0xC2, # ARABIC LETTER ALEF WITH MADDA ABOVE - 0x0623: 0xC3, # ARABIC LETTER ALEF WITH HAMZA ABOVE - 0x0624: 0xC4, # ARABIC LETTER WAW WITH HAMZA ABOVE - 0x0625: 0xC5, # ARABIC LETTER ALEF WITH HAMZA BELOW - 0x0626: 0xC6, # ARABIC LETTER YEH WITH HAMZA ABOVE - 0x0627: 0xC7, # ARABIC LETTER ALEF - 0x0628: 0xC8, # ARABIC LETTER BEH - 0x0629: 0xC9, # ARABIC LETTER TEH MARBUTA - 0x062A: 0xCA, # ARABIC LETTER TEH - 0x062B: 0xCB, # ARABIC LETTER THEH - 0x062C: 0xCC, # ARABIC LETTER JEEM - 0x062D: 0xCD, # ARABIC LETTER HAH - 0x062E: 0xCE, # ARABIC LETTER KHAH - 0x062F: 0xCF, # ARABIC LETTER DAL - 0x0630: 0xD0, # ARABIC LETTER THAL - 0x0631: 0xD1, # ARABIC LETTER REH - 0x0632: 0xD2, # ARABIC LETTER ZAIN - 0x0633: 0xD3, # ARABIC LETTER SEEN - 0x0634: 0xD4, # ARABIC LETTER SHEEN - 0x0635: 0xD5, # ARABIC LETTER SAD - 0x0636: 0xD6, # ARABIC LETTER DAD - 0x0637: 0xD7, # ARABIC LETTER TAH - 0x0638: 0xD8, # ARABIC LETTER ZAH - 0x0639: 0xD9, # ARABIC LETTER AIN - 0x063A: 0xDA, # ARABIC LETTER GHAIN - 0x0640: 0xE0, # ARABIC TATWEEL - 0x0641: 0xE1, # ARABIC LETTER FEH - 0x0642: 0xE2, # ARABIC LETTER QAF - 0x0643: 0xE3, # ARABIC LETTER KAF - 0x0644: 0xE4, # ARABIC LETTER LAM - 0x0645: 0xE5, # ARABIC LETTER MEEM - 0x0646: 0xE6, # ARABIC LETTER NOON - 0x0647: 0xE7, # ARABIC LETTER HEH - 0x0648: 0xE8, # ARABIC LETTER WAW - 0x0649: 0xE9, # ARABIC LETTER ALEF MAKSURA - 0x064A: 0xEA, # ARABIC LETTER YEH - 0x064B: 0xEB, # ARABIC FATHATAN - 0x064C: 0xEC, # ARABIC DAMMATAN - 0x064D: 0xED, # ARABIC KASRATAN - 0x064E: 0xEE, # ARABIC FATHA - 0x064F: 0xEF, # ARABIC DAMMA - 0x0650: 0xF0, # ARABIC KASRA - 0x0651: 0xF1, # ARABIC SHADDA - 0x0652: 0xF2, # ARABIC SUKUN - 0x066A: 0xA5, # ARABIC PERCENT SIGN - 0x0679: 0xF4, # ARABIC LETTER TTEH - 0x067E: 0xF3, # ARABIC LETTER PEH - 0x0686: 0xF5, # ARABIC LETTER TCHEH - 0x0688: 0xF9, # ARABIC LETTER DDAL - 0x0691: 0xFA, # ARABIC LETTER RREH - 0x0698: 0xFE, # ARABIC LETTER JEH - 0x06A4: 0xF7, # ARABIC LETTER VEH - 0x06AF: 0xF8, # ARABIC LETTER GAF - 0x06BA: 0x8B, # ARABIC LETTER NOON GHUNNA - 0x06D2: 0xFF, # ARABIC LETTER YEH BARREE - 0x06D5: 0xF6, # ARABIC LETTER AE - 0x06F0: 0xB0, # EXTENDED ARABIC-INDIC DIGIT ZERO, right-left (need override) - 0x06F1: 0xB1, # EXTENDED ARABIC-INDIC DIGIT ONE, right-left (need override) - 0x06F2: 0xB2, # EXTENDED ARABIC-INDIC DIGIT TWO, right-left (need override) - 0x06F3: 0xB3, # EXTENDED ARABIC-INDIC DIGIT THREE, right-left (need override) - 0x06F4: 0xB4, # EXTENDED ARABIC-INDIC DIGIT FOUR, right-left (need override) - 0x06F5: 0xB5, # EXTENDED ARABIC-INDIC DIGIT FIVE, right-left (need override) - 0x06F6: 0xB6, # EXTENDED ARABIC-INDIC DIGIT SIX, right-left (need override) - 0x06F7: 0xB7, # EXTENDED ARABIC-INDIC DIGIT SEVEN, right-left (need override) - 0x06F8: 0xB8, # EXTENDED ARABIC-INDIC DIGIT EIGHT, right-left (need override) - 0x06F9: 0xB9, # EXTENDED ARABIC-INDIC DIGIT NINE, right-left (need override) - 0x2026: 0x93, # HORIZONTAL ELLIPSIS, right-left - 0x274A: 0xC0, # EIGHT TEARDROP-SPOKED PROPELLER ASTERISK, right-left -} diff --git a/Lib/encodings/mac_greek.py b/Lib/encodings/mac_greek.py index 7264f9a..526cd9e 100644 --- a/Lib/encodings/mac_greek.py +++ b/Lib/encodings/mac_greek.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\xad' # 0xFF -> SOFT HYPHEN # before Mac OS 9.2.2, was undefined ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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: 0x92, # POUND SIGN - 0x00A5: 0xB4, # YEN SIGN - 0x00A6: 0x9B, # BROKEN BAR - 0x00A7: 0xAC, # SECTION SIGN - 0x00A8: 0x8C, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AD: 0xFF, # SOFT HYPHEN # before Mac OS 9.2.2, was undefined - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00B0: 0xAE, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B2: 0x82, # SUPERSCRIPT TWO - 0x00B3: 0x84, # SUPERSCRIPT THREE - 0x00B7: 0xAF, # MIDDLE DOT - 0x00B9: 0x81, # SUPERSCRIPT ONE - 0x00BB: 0xC8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BD: 0x97, # VULGAR FRACTION ONE HALF - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x0153: 0xCF, # LATIN SMALL LIGATURE OE - 0x0384: 0x8B, # GREEK TONOS - 0x0385: 0x87, # GREEK DIALYTIKA TONOS - 0x0386: 0xCD, # GREEK CAPITAL LETTER ALPHA WITH TONOS - 0x0388: 0xCE, # GREEK CAPITAL LETTER EPSILON WITH TONOS - 0x0389: 0xD7, # GREEK CAPITAL LETTER ETA WITH TONOS - 0x038A: 0xD8, # GREEK CAPITAL LETTER IOTA WITH TONOS - 0x038C: 0xD9, # GREEK CAPITAL LETTER OMICRON WITH TONOS - 0x038E: 0xDA, # GREEK CAPITAL LETTER UPSILON WITH TONOS - 0x038F: 0xDF, # GREEK CAPITAL LETTER OMEGA WITH TONOS - 0x0390: 0xFD, # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS - 0x0391: 0xB0, # GREEK CAPITAL LETTER ALPHA - 0x0392: 0xB5, # GREEK CAPITAL LETTER BETA - 0x0393: 0xA1, # GREEK CAPITAL LETTER GAMMA - 0x0394: 0xA2, # GREEK CAPITAL LETTER DELTA - 0x0395: 0xB6, # GREEK CAPITAL LETTER EPSILON - 0x0396: 0xB7, # GREEK CAPITAL LETTER ZETA - 0x0397: 0xB8, # GREEK CAPITAL LETTER ETA - 0x0398: 0xA3, # GREEK CAPITAL LETTER THETA - 0x0399: 0xB9, # GREEK CAPITAL LETTER IOTA - 0x039A: 0xBA, # GREEK CAPITAL LETTER KAPPA - 0x039B: 0xA4, # GREEK CAPITAL LETTER LAMDA - 0x039C: 0xBB, # GREEK CAPITAL LETTER MU - 0x039D: 0xC1, # GREEK CAPITAL LETTER NU - 0x039E: 0xA5, # GREEK CAPITAL LETTER XI - 0x039F: 0xC3, # GREEK CAPITAL LETTER OMICRON - 0x03A0: 0xA6, # GREEK CAPITAL LETTER PI - 0x03A1: 0xC4, # GREEK CAPITAL LETTER RHO - 0x03A3: 0xAA, # GREEK CAPITAL LETTER SIGMA - 0x03A4: 0xC6, # GREEK CAPITAL LETTER TAU - 0x03A5: 0xCB, # GREEK CAPITAL LETTER UPSILON - 0x03A6: 0xBC, # GREEK CAPITAL LETTER PHI - 0x03A7: 0xCC, # GREEK CAPITAL LETTER CHI - 0x03A8: 0xBE, # GREEK CAPITAL LETTER PSI - 0x03A9: 0xBF, # GREEK CAPITAL LETTER OMEGA - 0x03AA: 0xAB, # GREEK CAPITAL LETTER IOTA WITH DIALYTIKA - 0x03AB: 0xBD, # GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA - 0x03AC: 0xC0, # GREEK SMALL LETTER ALPHA WITH TONOS - 0x03AD: 0xDB, # GREEK SMALL LETTER EPSILON WITH TONOS - 0x03AE: 0xDC, # GREEK SMALL LETTER ETA WITH TONOS - 0x03AF: 0xDD, # GREEK SMALL LETTER IOTA WITH TONOS - 0x03B0: 0xFE, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS - 0x03B1: 0xE1, # GREEK SMALL LETTER ALPHA - 0x03B2: 0xE2, # GREEK SMALL LETTER BETA - 0x03B3: 0xE7, # GREEK SMALL LETTER GAMMA - 0x03B4: 0xE4, # GREEK SMALL LETTER DELTA - 0x03B5: 0xE5, # GREEK SMALL LETTER EPSILON - 0x03B6: 0xFA, # GREEK SMALL LETTER ZETA - 0x03B7: 0xE8, # GREEK SMALL LETTER ETA - 0x03B8: 0xF5, # GREEK SMALL LETTER THETA - 0x03B9: 0xE9, # GREEK SMALL LETTER IOTA - 0x03BA: 0xEB, # GREEK SMALL LETTER KAPPA - 0x03BB: 0xEC, # GREEK SMALL LETTER LAMDA - 0x03BC: 0xED, # GREEK SMALL LETTER MU - 0x03BD: 0xEE, # GREEK SMALL LETTER NU - 0x03BE: 0xEA, # GREEK SMALL LETTER XI - 0x03BF: 0xEF, # GREEK SMALL LETTER OMICRON - 0x03C0: 0xF0, # GREEK SMALL LETTER PI - 0x03C1: 0xF2, # GREEK SMALL LETTER RHO - 0x03C2: 0xF7, # GREEK SMALL LETTER FINAL SIGMA - 0x03C3: 0xF3, # GREEK SMALL LETTER SIGMA - 0x03C4: 0xF4, # GREEK SMALL LETTER TAU - 0x03C5: 0xF9, # GREEK SMALL LETTER UPSILON - 0x03C6: 0xE6, # GREEK SMALL LETTER PHI - 0x03C7: 0xF8, # GREEK SMALL LETTER CHI - 0x03C8: 0xE3, # GREEK SMALL LETTER PSI - 0x03C9: 0xF6, # GREEK SMALL LETTER OMEGA - 0x03CA: 0xFB, # GREEK SMALL LETTER IOTA WITH DIALYTIKA - 0x03CB: 0xFC, # GREEK SMALL LETTER UPSILON WITH DIALYTIKA - 0x03CC: 0xDE, # GREEK SMALL LETTER OMICRON WITH TONOS - 0x03CD: 0xE0, # GREEK SMALL LETTER UPSILON WITH TONOS - 0x03CE: 0xF1, # GREEK SMALL LETTER OMEGA WITH TONOS - 0x2013: 0xD0, # EN DASH - 0x2015: 0xD1, # HORIZONTAL BAR - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x2020: 0xA0, # DAGGER - 0x2022: 0x96, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2030: 0x98, # PER MILLE SIGN - 0x20AC: 0x9C, # EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN - 0x2122: 0x93, # TRADE MARK SIGN - 0x2248: 0xC5, # ALMOST EQUAL TO - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO -} diff --git a/Lib/encodings/mac_iceland.py b/Lib/encodings/mac_iceland.py index 5d8d9ad..dea5220 100644 --- a/Lib/encodings/mac_iceland.py +++ b/Lib/encodings/mac_iceland.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02c7' # 0xFF -> CARON ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 - 0x00A1: 0xC1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A5: 0xB4, # YEN SIGN - 0x00A7: 0xA4, # SECTION SIGN - 0x00A8: 0xAC, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AA: 0xBB, # FEMININE ORDINAL INDICATOR - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00AF: 0xF8, # MACRON - 0x00B0: 0xA1, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B4: 0xAB, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xA6, # PILCROW SIGN - 0x00B7: 0xE1, # MIDDLE DOT - 0x00B8: 0xFC, # CEDILLA - 0x00BA: 0xBC, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0xC8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BF: 0xC0, # INVERTED QUESTION MARK - 0x00C0: 0xCB, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0xE7, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xE5, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0xCC, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x81, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0xAE, # LATIN CAPITAL LETTER AE - 0x00C7: 0x82, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0xE9, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0xE6, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0xE8, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0xED, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0xEA, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xEB, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0xEC, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D0: 0xDC, # LATIN CAPITAL LETTER ETH - 0x00D1: 0x84, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xF1, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEF, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xCD, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D8: 0xAF, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xF4, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xF2, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xF3, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DD: 0xA0, # LATIN CAPITAL LETTER Y WITH ACUTE - 0x00DE: 0xDE, # LATIN CAPITAL LETTER THORN - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x8B, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x8C, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0xBE, # LATIN SMALL LETTER AE - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x93, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F0: 0xDD, # LATIN SMALL LETTER ETH - 0x00F1: 0x96, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0x98, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0x9B, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00F8: 0xBF, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FD: 0xE0, # LATIN SMALL LETTER Y WITH ACUTE - 0x00FE: 0xDF, # LATIN SMALL LETTER THORN - 0x00FF: 0xD8, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x0131: 0xF5, # LATIN SMALL LETTER DOTLESS I - 0x0152: 0xCE, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xCF, # LATIN SMALL LIGATURE OE - 0x0178: 0xD9, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x0192: 0xC4, # LATIN SMALL LETTER F WITH HOOK - 0x02C6: 0xF6, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02C7: 0xFF, # CARON - 0x02D8: 0xF9, # BREVE - 0x02D9: 0xFA, # DOT ABOVE - 0x02DA: 0xFB, # RING ABOVE - 0x02DB: 0xFE, # OGONEK - 0x02DC: 0xF7, # SMALL TILDE - 0x02DD: 0xFD, # DOUBLE ACUTE ACCENT - 0x03A9: 0xBD, # GREEK CAPITAL LETTER OMEGA - 0x03C0: 0xB9, # GREEK SMALL LETTER PI - 0x2013: 0xD0, # EN DASH - 0x2014: 0xD1, # EM DASH - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201A: 0xE2, # SINGLE LOW-9 QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xE3, # DOUBLE LOW-9 QUOTATION MARK - 0x2022: 0xA5, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2030: 0xE4, # PER MILLE SIGN - 0x2044: 0xDA, # FRACTION SLASH - 0x20AC: 0xDB, # EURO SIGN - 0x2122: 0xAA, # TRADE MARK SIGN - 0x2202: 0xB6, # PARTIAL DIFFERENTIAL - 0x2206: 0xC6, # INCREMENT - 0x220F: 0xB8, # N-ARY PRODUCT - 0x2211: 0xB7, # N-ARY SUMMATION - 0x221A: 0xC3, # SQUARE ROOT - 0x221E: 0xB0, # INFINITY - 0x222B: 0xBA, # INTEGRAL - 0x2248: 0xC5, # ALMOST EQUAL TO - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO - 0x25CA: 0xD7, # LOZENGE - 0xF8FF: 0xF0, # Apple logo -} diff --git a/Lib/encodings/mac_roman.py b/Lib/encodings/mac_roman.py index 9552e53..aa99e5e 100644 --- a/Lib/encodings/mac_roman.py +++ b/Lib/encodings/mac_roman.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02c7' # 0xFF -> CARON ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 - 0x00A1: 0xC1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A5: 0xB4, # YEN SIGN - 0x00A7: 0xA4, # SECTION SIGN - 0x00A8: 0xAC, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AA: 0xBB, # FEMININE ORDINAL INDICATOR - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00AF: 0xF8, # MACRON - 0x00B0: 0xA1, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B4: 0xAB, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xA6, # PILCROW SIGN - 0x00B7: 0xE1, # MIDDLE DOT - 0x00B8: 0xFC, # CEDILLA - 0x00BA: 0xBC, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0xC8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BF: 0xC0, # INVERTED QUESTION MARK - 0x00C0: 0xCB, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0xE7, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xE5, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0xCC, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x81, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0xAE, # LATIN CAPITAL LETTER AE - 0x00C7: 0x82, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0xE9, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0xE6, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0xE8, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0xED, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0xEA, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xEB, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0xEC, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D1: 0x84, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xF1, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEF, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xCD, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D8: 0xAF, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xF4, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xF2, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xF3, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x8B, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x8C, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0xBE, # LATIN SMALL LETTER AE - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x93, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F1: 0x96, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0x98, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0x9B, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00F8: 0xBF, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FF: 0xD8, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x0131: 0xF5, # LATIN SMALL LETTER DOTLESS I - 0x0152: 0xCE, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xCF, # LATIN SMALL LIGATURE OE - 0x0178: 0xD9, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x0192: 0xC4, # LATIN SMALL LETTER F WITH HOOK - 0x02C6: 0xF6, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02C7: 0xFF, # CARON - 0x02D8: 0xF9, # BREVE - 0x02D9: 0xFA, # DOT ABOVE - 0x02DA: 0xFB, # RING ABOVE - 0x02DB: 0xFE, # OGONEK - 0x02DC: 0xF7, # SMALL TILDE - 0x02DD: 0xFD, # DOUBLE ACUTE ACCENT - 0x03A9: 0xBD, # GREEK CAPITAL LETTER OMEGA - 0x03C0: 0xB9, # GREEK SMALL LETTER PI - 0x2013: 0xD0, # EN DASH - 0x2014: 0xD1, # EM DASH - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201A: 0xE2, # SINGLE LOW-9 QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xE3, # DOUBLE LOW-9 QUOTATION MARK - 0x2020: 0xA0, # DAGGER - 0x2021: 0xE0, # DOUBLE DAGGER - 0x2022: 0xA5, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2030: 0xE4, # PER MILLE SIGN - 0x2039: 0xDC, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK - 0x203A: 0xDD, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - 0x2044: 0xDA, # FRACTION SLASH - 0x20AC: 0xDB, # EURO SIGN - 0x2122: 0xAA, # TRADE MARK SIGN - 0x2202: 0xB6, # PARTIAL DIFFERENTIAL - 0x2206: 0xC6, # INCREMENT - 0x220F: 0xB8, # N-ARY PRODUCT - 0x2211: 0xB7, # N-ARY SUMMATION - 0x221A: 0xC3, # SQUARE ROOT - 0x221E: 0xB0, # INFINITY - 0x222B: 0xBA, # INTEGRAL - 0x2248: 0xC5, # ALMOST EQUAL TO - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO - 0x25CA: 0xD7, # LOZENGE - 0xF8FF: 0xF0, # Apple logo - 0xFB01: 0xDE, # LATIN SMALL LIGATURE FI - 0xFB02: 0xDF, # LATIN SMALL LIGATURE FL -} diff --git a/Lib/encodings/mac_romanian.py b/Lib/encodings/mac_romanian.py index 51282c3..b8ada81 100644 --- a/Lib/encodings/mac_romanian.py +++ b/Lib/encodings/mac_romanian.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02c7' # 0xFF -> CARON ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 - 0x00A1: 0xC1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A5: 0xB4, # YEN SIGN - 0x00A7: 0xA4, # SECTION SIGN - 0x00A8: 0xAC, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AA: 0xBB, # FEMININE ORDINAL INDICATOR - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00AF: 0xF8, # MACRON - 0x00B0: 0xA1, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B4: 0xAB, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xA6, # PILCROW SIGN - 0x00B7: 0xE1, # MIDDLE DOT - 0x00B8: 0xFC, # CEDILLA - 0x00BA: 0xBC, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0xC8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BF: 0xC0, # INVERTED QUESTION MARK - 0x00C0: 0xCB, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0xE7, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xE5, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0xCC, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x81, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C7: 0x82, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0xE9, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0xE6, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0xE8, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0xED, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0xEA, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xEB, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0xEC, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D1: 0x84, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xF1, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEF, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xCD, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D9: 0xF4, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xF2, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xF3, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x8B, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x8C, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x93, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F1: 0x96, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0x98, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0x9B, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FF: 0xD8, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x0102: 0xAE, # LATIN CAPITAL LETTER A WITH BREVE - 0x0103: 0xBE, # LATIN SMALL LETTER A WITH BREVE - 0x0131: 0xF5, # LATIN SMALL LETTER DOTLESS I - 0x0152: 0xCE, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xCF, # LATIN SMALL LIGATURE OE - 0x0178: 0xD9, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x0192: 0xC4, # LATIN SMALL LETTER F WITH HOOK - 0x0218: 0xAF, # LATIN CAPITAL LETTER S WITH COMMA BELOW # for Unicode 3.0 and later - 0x0219: 0xBF, # LATIN SMALL LETTER S WITH COMMA BELOW # for Unicode 3.0 and later - 0x021A: 0xDE, # LATIN CAPITAL LETTER T WITH COMMA BELOW # for Unicode 3.0 and later - 0x021B: 0xDF, # LATIN SMALL LETTER T WITH COMMA BELOW # for Unicode 3.0 and later - 0x02C6: 0xF6, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02C7: 0xFF, # CARON - 0x02D8: 0xF9, # BREVE - 0x02D9: 0xFA, # DOT ABOVE - 0x02DA: 0xFB, # RING ABOVE - 0x02DB: 0xFE, # OGONEK - 0x02DC: 0xF7, # SMALL TILDE - 0x02DD: 0xFD, # DOUBLE ACUTE ACCENT - 0x03A9: 0xBD, # GREEK CAPITAL LETTER OMEGA - 0x03C0: 0xB9, # GREEK SMALL LETTER PI - 0x2013: 0xD0, # EN DASH - 0x2014: 0xD1, # EM DASH - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201A: 0xE2, # SINGLE LOW-9 QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xE3, # DOUBLE LOW-9 QUOTATION MARK - 0x2020: 0xA0, # DAGGER - 0x2021: 0xE0, # DOUBLE DAGGER - 0x2022: 0xA5, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2030: 0xE4, # PER MILLE SIGN - 0x2039: 0xDC, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK - 0x203A: 0xDD, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK - 0x2044: 0xDA, # FRACTION SLASH - 0x20AC: 0xDB, # EURO SIGN - 0x2122: 0xAA, # TRADE MARK SIGN - 0x2202: 0xB6, # PARTIAL DIFFERENTIAL - 0x2206: 0xC6, # INCREMENT - 0x220F: 0xB8, # N-ARY PRODUCT - 0x2211: 0xB7, # N-ARY SUMMATION - 0x221A: 0xC3, # SQUARE ROOT - 0x221E: 0xB0, # INFINITY - 0x222B: 0xBA, # INTEGRAL - 0x2248: 0xC5, # ALMOST EQUAL TO - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO - 0x25CA: 0xD7, # LOZENGE - 0xF8FF: 0xF0, # Apple logo -} diff --git a/Lib/encodings/mac_turkish.py b/Lib/encodings/mac_turkish.py index 4e5641f..9a747ee 100644 --- a/Lib/encodings/mac_turkish.py +++ b/Lib/encodings/mac_turkish.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,263 +303,6 @@ decoding_table = ( u'\u02c7' # 0xFF -> CARON ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 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 - 0x00A1: 0xC1, # INVERTED EXCLAMATION MARK - 0x00A2: 0xA2, # CENT SIGN - 0x00A3: 0xA3, # POUND SIGN - 0x00A5: 0xB4, # YEN SIGN - 0x00A7: 0xA4, # SECTION SIGN - 0x00A8: 0xAC, # DIAERESIS - 0x00A9: 0xA9, # COPYRIGHT SIGN - 0x00AA: 0xBB, # FEMININE ORDINAL INDICATOR - 0x00AB: 0xC7, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00AC: 0xC2, # NOT SIGN - 0x00AE: 0xA8, # REGISTERED SIGN - 0x00AF: 0xF8, # MACRON - 0x00B0: 0xA1, # DEGREE SIGN - 0x00B1: 0xB1, # PLUS-MINUS SIGN - 0x00B4: 0xAB, # ACUTE ACCENT - 0x00B5: 0xB5, # MICRO SIGN - 0x00B6: 0xA6, # PILCROW SIGN - 0x00B7: 0xE1, # MIDDLE DOT - 0x00B8: 0xFC, # CEDILLA - 0x00BA: 0xBC, # MASCULINE ORDINAL INDICATOR - 0x00BB: 0xC8, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK - 0x00BF: 0xC0, # INVERTED QUESTION MARK - 0x00C0: 0xCB, # LATIN CAPITAL LETTER A WITH GRAVE - 0x00C1: 0xE7, # LATIN CAPITAL LETTER A WITH ACUTE - 0x00C2: 0xE5, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX - 0x00C3: 0xCC, # LATIN CAPITAL LETTER A WITH TILDE - 0x00C4: 0x80, # LATIN CAPITAL LETTER A WITH DIAERESIS - 0x00C5: 0x81, # LATIN CAPITAL LETTER A WITH RING ABOVE - 0x00C6: 0xAE, # LATIN CAPITAL LETTER AE - 0x00C7: 0x82, # LATIN CAPITAL LETTER C WITH CEDILLA - 0x00C8: 0xE9, # LATIN CAPITAL LETTER E WITH GRAVE - 0x00C9: 0x83, # LATIN CAPITAL LETTER E WITH ACUTE - 0x00CA: 0xE6, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX - 0x00CB: 0xE8, # LATIN CAPITAL LETTER E WITH DIAERESIS - 0x00CC: 0xED, # LATIN CAPITAL LETTER I WITH GRAVE - 0x00CD: 0xEA, # LATIN CAPITAL LETTER I WITH ACUTE - 0x00CE: 0xEB, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX - 0x00CF: 0xEC, # LATIN CAPITAL LETTER I WITH DIAERESIS - 0x00D1: 0x84, # LATIN CAPITAL LETTER N WITH TILDE - 0x00D2: 0xF1, # LATIN CAPITAL LETTER O WITH GRAVE - 0x00D3: 0xEE, # LATIN CAPITAL LETTER O WITH ACUTE - 0x00D4: 0xEF, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX - 0x00D5: 0xCD, # LATIN CAPITAL LETTER O WITH TILDE - 0x00D6: 0x85, # LATIN CAPITAL LETTER O WITH DIAERESIS - 0x00D8: 0xAF, # LATIN CAPITAL LETTER O WITH STROKE - 0x00D9: 0xF4, # LATIN CAPITAL LETTER U WITH GRAVE - 0x00DA: 0xF2, # LATIN CAPITAL LETTER U WITH ACUTE - 0x00DB: 0xF3, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX - 0x00DC: 0x86, # LATIN CAPITAL LETTER U WITH DIAERESIS - 0x00DF: 0xA7, # LATIN SMALL LETTER SHARP S - 0x00E0: 0x88, # LATIN SMALL LETTER A WITH GRAVE - 0x00E1: 0x87, # LATIN SMALL LETTER A WITH ACUTE - 0x00E2: 0x89, # LATIN SMALL LETTER A WITH CIRCUMFLEX - 0x00E3: 0x8B, # LATIN SMALL LETTER A WITH TILDE - 0x00E4: 0x8A, # LATIN SMALL LETTER A WITH DIAERESIS - 0x00E5: 0x8C, # LATIN SMALL LETTER A WITH RING ABOVE - 0x00E6: 0xBE, # LATIN SMALL LETTER AE - 0x00E7: 0x8D, # LATIN SMALL LETTER C WITH CEDILLA - 0x00E8: 0x8F, # LATIN SMALL LETTER E WITH GRAVE - 0x00E9: 0x8E, # LATIN SMALL LETTER E WITH ACUTE - 0x00EA: 0x90, # LATIN SMALL LETTER E WITH CIRCUMFLEX - 0x00EB: 0x91, # LATIN SMALL LETTER E WITH DIAERESIS - 0x00EC: 0x93, # LATIN SMALL LETTER I WITH GRAVE - 0x00ED: 0x92, # LATIN SMALL LETTER I WITH ACUTE - 0x00EE: 0x94, # LATIN SMALL LETTER I WITH CIRCUMFLEX - 0x00EF: 0x95, # LATIN SMALL LETTER I WITH DIAERESIS - 0x00F1: 0x96, # LATIN SMALL LETTER N WITH TILDE - 0x00F2: 0x98, # LATIN SMALL LETTER O WITH GRAVE - 0x00F3: 0x97, # LATIN SMALL LETTER O WITH ACUTE - 0x00F4: 0x99, # LATIN SMALL LETTER O WITH CIRCUMFLEX - 0x00F5: 0x9B, # LATIN SMALL LETTER O WITH TILDE - 0x00F6: 0x9A, # LATIN SMALL LETTER O WITH DIAERESIS - 0x00F7: 0xD6, # DIVISION SIGN - 0x00F8: 0xBF, # LATIN SMALL LETTER O WITH STROKE - 0x00F9: 0x9D, # LATIN SMALL LETTER U WITH GRAVE - 0x00FA: 0x9C, # LATIN SMALL LETTER U WITH ACUTE - 0x00FB: 0x9E, # LATIN SMALL LETTER U WITH CIRCUMFLEX - 0x00FC: 0x9F, # LATIN SMALL LETTER U WITH DIAERESIS - 0x00FF: 0xD8, # LATIN SMALL LETTER Y WITH DIAERESIS - 0x011E: 0xDA, # LATIN CAPITAL LETTER G WITH BREVE - 0x011F: 0xDB, # LATIN SMALL LETTER G WITH BREVE - 0x0130: 0xDC, # LATIN CAPITAL LETTER I WITH DOT ABOVE - 0x0131: 0xDD, # LATIN SMALL LETTER DOTLESS I - 0x0152: 0xCE, # LATIN CAPITAL LIGATURE OE - 0x0153: 0xCF, # LATIN SMALL LIGATURE OE - 0x015E: 0xDE, # LATIN CAPITAL LETTER S WITH CEDILLA - 0x015F: 0xDF, # LATIN SMALL LETTER S WITH CEDILLA - 0x0178: 0xD9, # LATIN CAPITAL LETTER Y WITH DIAERESIS - 0x0192: 0xC4, # LATIN SMALL LETTER F WITH HOOK - 0x02C6: 0xF6, # MODIFIER LETTER CIRCUMFLEX ACCENT - 0x02C7: 0xFF, # CARON - 0x02D8: 0xF9, # BREVE - 0x02D9: 0xFA, # DOT ABOVE - 0x02DA: 0xFB, # RING ABOVE - 0x02DB: 0xFE, # OGONEK - 0x02DC: 0xF7, # SMALL TILDE - 0x02DD: 0xFD, # DOUBLE ACUTE ACCENT - 0x03A9: 0xBD, # GREEK CAPITAL LETTER OMEGA - 0x03C0: 0xB9, # GREEK SMALL LETTER PI - 0x2013: 0xD0, # EN DASH - 0x2014: 0xD1, # EM DASH - 0x2018: 0xD4, # LEFT SINGLE QUOTATION MARK - 0x2019: 0xD5, # RIGHT SINGLE QUOTATION MARK - 0x201A: 0xE2, # SINGLE LOW-9 QUOTATION MARK - 0x201C: 0xD2, # LEFT DOUBLE QUOTATION MARK - 0x201D: 0xD3, # RIGHT DOUBLE QUOTATION MARK - 0x201E: 0xE3, # DOUBLE LOW-9 QUOTATION MARK - 0x2020: 0xA0, # DAGGER - 0x2021: 0xE0, # DOUBLE DAGGER - 0x2022: 0xA5, # BULLET - 0x2026: 0xC9, # HORIZONTAL ELLIPSIS - 0x2030: 0xE4, # PER MILLE SIGN - 0x2122: 0xAA, # TRADE MARK SIGN - 0x2202: 0xB6, # PARTIAL DIFFERENTIAL - 0x2206: 0xC6, # INCREMENT - 0x220F: 0xB8, # N-ARY PRODUCT - 0x2211: 0xB7, # N-ARY SUMMATION - 0x221A: 0xC3, # SQUARE ROOT - 0x221E: 0xB0, # INFINITY - 0x222B: 0xBA, # INTEGRAL - 0x2248: 0xC5, # ALMOST EQUAL TO - 0x2260: 0xAD, # NOT EQUAL TO - 0x2264: 0xB2, # LESS-THAN OR EQUAL TO - 0x2265: 0xB3, # GREATER-THAN OR EQUAL TO - 0x25CA: 0xD7, # LOZENGE - 0xF8A0: 0xF5, # undefined1 - 0xF8FF: 0xF0, # Apple logo -} diff --git a/Lib/encodings/tis_620.py b/Lib/encodings/tis_620.py index 166d932..6fd8deb 100644 --- a/Lib/encodings/tis_620.py +++ b/Lib/encodings/tis_620.py @@ -9,14 +9,14 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) + return codecs.charmap_encode(input,errors,encoding_table) def decode(self,input,errors='strict'): return codecs.charmap_decode(input,errors,decoding_table) class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_table)[0] class IncrementalDecoder(codecs.IncrementalDecoder): def decode(self, input, final=False): @@ -303,254 +303,6 @@ decoding_table = ( u'\ufffe' ) -### Encoding Map +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) -encoding_map = { - 0x0000: 0x00, # NULL - 0x0001: 0x01, # START OF HEADING - 0x0002: 0x02, # START OF TEXT - 0x0003: 0x03, # END OF TEXT - 0x0004: 0x04, # END OF TRANSMISSION - 0x0005: 0x05, # ENQUIRY - 0x0006: 0x06, # ACKNOWLEDGE - 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 - 0x0010: 0x10, # DATA LINK ESCAPE - 0x0011: 0x11, # DEVICE CONTROL ONE - 0x0012: 0x12, # DEVICE CONTROL TWO - 0x0013: 0x13, # DEVICE CONTROL THREE - 0x0014: 0x14, # DEVICE CONTROL FOUR - 0x0015: 0x15, # NEGATIVE ACKNOWLEDGE - 0x0016: 0x16, # SYNCHRONOUS IDLE - 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 - 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, # DELETE - 0x0080: 0x80, # <control> - 0x0081: 0x81, # <control> - 0x0082: 0x82, # <control> - 0x0083: 0x83, # <control> - 0x0084: 0x84, # <control> - 0x0085: 0x85, # <control> - 0x0086: 0x86, # <control> - 0x0087: 0x87, # <control> - 0x0088: 0x88, # <control> - 0x0089: 0x89, # <control> - 0x008A: 0x8A, # <control> - 0x008B: 0x8B, # <control> - 0x008C: 0x8C, # <control> - 0x008D: 0x8D, # <control> - 0x008E: 0x8E, # <control> - 0x008F: 0x8F, # <control> - 0x0090: 0x90, # <control> - 0x0091: 0x91, # <control> - 0x0092: 0x92, # <control> - 0x0093: 0x93, # <control> - 0x0094: 0x94, # <control> - 0x0095: 0x95, # <control> - 0x0096: 0x96, # <control> - 0x0097: 0x97, # <control> - 0x0098: 0x98, # <control> - 0x0099: 0x99, # <control> - 0x009A: 0x9A, # <control> - 0x009B: 0x9B, # <control> - 0x009C: 0x9C, # <control> - 0x009D: 0x9D, # <control> - 0x009E: 0x9E, # <control> - 0x009F: 0x9F, # <control> - 0x0E01: 0xA1, # THAI CHARACTER KO KAI - 0x0E02: 0xA2, # THAI CHARACTER KHO KHAI - 0x0E03: 0xA3, # THAI CHARACTER KHO KHUAT - 0x0E04: 0xA4, # THAI CHARACTER KHO KHWAI - 0x0E05: 0xA5, # THAI CHARACTER KHO KHON - 0x0E06: 0xA6, # THAI CHARACTER KHO RAKHANG - 0x0E07: 0xA7, # THAI CHARACTER NGO NGU - 0x0E08: 0xA8, # THAI CHARACTER CHO CHAN - 0x0E09: 0xA9, # THAI CHARACTER CHO CHING - 0x0E0A: 0xAA, # THAI CHARACTER CHO CHANG - 0x0E0B: 0xAB, # THAI CHARACTER SO SO - 0x0E0C: 0xAC, # THAI CHARACTER CHO CHOE - 0x0E0D: 0xAD, # THAI CHARACTER YO YING - 0x0E0E: 0xAE, # THAI CHARACTER DO CHADA - 0x0E0F: 0xAF, # THAI CHARACTER TO PATAK - 0x0E10: 0xB0, # THAI CHARACTER THO THAN - 0x0E11: 0xB1, # THAI CHARACTER THO NANGMONTHO - 0x0E12: 0xB2, # THAI CHARACTER THO PHUTHAO - 0x0E13: 0xB3, # THAI CHARACTER NO NEN - 0x0E14: 0xB4, # THAI CHARACTER DO DEK - 0x0E15: 0xB5, # THAI CHARACTER TO TAO - 0x0E16: 0xB6, # THAI CHARACTER THO THUNG - 0x0E17: 0xB7, # THAI CHARACTER THO THAHAN - 0x0E18: 0xB8, # THAI CHARACTER THO THONG - 0x0E19: 0xB9, # THAI CHARACTER NO NU - 0x0E1A: 0xBA, # THAI CHARACTER BO BAIMAI - 0x0E1B: 0xBB, # THAI CHARACTER PO PLA - 0x0E1C: 0xBC, # THAI CHARACTER PHO PHUNG - 0x0E1D: 0xBD, # THAI CHARACTER FO FA - 0x0E1E: 0xBE, # THAI CHARACTER PHO PHAN - 0x0E1F: 0xBF, # THAI CHARACTER FO FAN - 0x0E20: 0xC0, # THAI CHARACTER PHO SAMPHAO - 0x0E21: 0xC1, # THAI CHARACTER MO MA - 0x0E22: 0xC2, # THAI CHARACTER YO YAK - 0x0E23: 0xC3, # THAI CHARACTER RO RUA - 0x0E24: 0xC4, # THAI CHARACTER RU - 0x0E25: 0xC5, # THAI CHARACTER LO LING - 0x0E26: 0xC6, # THAI CHARACTER LU - 0x0E27: 0xC7, # THAI CHARACTER WO WAEN - 0x0E28: 0xC8, # THAI CHARACTER SO SALA - 0x0E29: 0xC9, # THAI CHARACTER SO RUSI - 0x0E2A: 0xCA, # THAI CHARACTER SO SUA - 0x0E2B: 0xCB, # THAI CHARACTER HO HIP - 0x0E2C: 0xCC, # THAI CHARACTER LO CHULA - 0x0E2D: 0xCD, # THAI CHARACTER O ANG - 0x0E2E: 0xCE, # THAI CHARACTER HO NOKHUK - 0x0E2F: 0xCF, # THAI CHARACTER PAIYANNOI - 0x0E30: 0xD0, # THAI CHARACTER SARA A - 0x0E31: 0xD1, # THAI CHARACTER MAI HAN-AKAT - 0x0E32: 0xD2, # THAI CHARACTER SARA AA - 0x0E33: 0xD3, # THAI CHARACTER SARA AM - 0x0E34: 0xD4, # THAI CHARACTER SARA I - 0x0E35: 0xD5, # THAI CHARACTER SARA II - 0x0E36: 0xD6, # THAI CHARACTER SARA UE - 0x0E37: 0xD7, # THAI CHARACTER SARA UEE - 0x0E38: 0xD8, # THAI CHARACTER SARA U - 0x0E39: 0xD9, # THAI CHARACTER SARA UU - 0x0E3A: 0xDA, # THAI CHARACTER PHINTHU - 0x0E3F: 0xDF, # THAI CURRENCY SYMBOL BAHT - 0x0E40: 0xE0, # THAI CHARACTER SARA E - 0x0E41: 0xE1, # THAI CHARACTER SARA AE - 0x0E42: 0xE2, # THAI CHARACTER SARA O - 0x0E43: 0xE3, # THAI CHARACTER SARA AI MAIMUAN - 0x0E44: 0xE4, # THAI CHARACTER SARA AI MAIMALAI - 0x0E45: 0xE5, # THAI CHARACTER LAKKHANGYAO - 0x0E46: 0xE6, # THAI CHARACTER MAIYAMOK - 0x0E47: 0xE7, # THAI CHARACTER MAITAIKHU - 0x0E48: 0xE8, # THAI CHARACTER MAI EK - 0x0E49: 0xE9, # THAI CHARACTER MAI THO - 0x0E4A: 0xEA, # THAI CHARACTER MAI TRI - 0x0E4B: 0xEB, # THAI CHARACTER MAI CHATTAWA - 0x0E4C: 0xEC, # THAI CHARACTER THANTHAKHAT - 0x0E4D: 0xED, # THAI CHARACTER NIKHAHIT - 0x0E4E: 0xEE, # THAI CHARACTER YAMAKKAN - 0x0E4F: 0xEF, # THAI CHARACTER FONGMAN - 0x0E50: 0xF0, # THAI DIGIT ZERO - 0x0E51: 0xF1, # THAI DIGIT ONE - 0x0E52: 0xF2, # THAI DIGIT TWO - 0x0E53: 0xF3, # THAI DIGIT THREE - 0x0E54: 0xF4, # THAI DIGIT FOUR - 0x0E55: 0xF5, # THAI DIGIT FIVE - 0x0E56: 0xF6, # THAI DIGIT SIX - 0x0E57: 0xF7, # THAI DIGIT SEVEN - 0x0E58: 0xF8, # THAI DIGIT EIGHT - 0x0E59: 0xF9, # THAI DIGIT NINE - 0x0E5A: 0xFA, # THAI CHARACTER ANGKHANKHU - 0x0E5B: 0xFB, # THAI CHARACTER KHOMUT -} @@ -104,6 +104,9 @@ Extension Modules Library ------- +- Patch #1359618: Speed up charmap encoder by using a trie structure + for lookup. + - The functions in the ``pprint`` module now sort dictionaries by key before computing the display. Before 2.5, ``pprint`` sorted a dictionary if and only if its display required more than one line, although that diff --git a/Modules/_codecsmodule.c b/Modules/_codecsmodule.c index 080fa74..32fa82f 100644 --- a/Modules/_codecsmodule.c +++ b/Modules/_codecsmodule.c @@ -792,6 +792,15 @@ charmap_encode(PyObject *self, return v; } +static PyObject* +charmap_build(PyObject *self, PyObject *args) +{ + PyObject *map; + if (!PyArg_ParseTuple(args, "U:charmap_build", &map)) + return NULL; + return PyUnicode_BuildEncodingMap(map); +} + #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) static PyObject * @@ -897,6 +906,7 @@ static PyMethodDef _codecs_functions[] = { {"ascii_decode", ascii_decode, METH_VARARGS}, {"charmap_encode", charmap_encode, METH_VARARGS}, {"charmap_decode", charmap_decode, METH_VARARGS}, + {"charmap_build", charmap_build, METH_VARARGS}, {"readbuffer_encode", readbuffer_encode, METH_VARARGS}, {"charbuffer_encode", charbuffer_encode, METH_VARARGS}, #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c index f93cfa5..eb5bdd8 100644 --- a/Objects/unicodeobject.c +++ b/Objects/unicodeobject.c @@ -3057,6 +3057,219 @@ PyObject *PyUnicode_DecodeCharmap(const char *s, return NULL; } +/* Charmap encoding: the lookup table */ + +struct encoding_map{ + PyObject_HEAD + unsigned char level1[32]; + int count2, count3; + unsigned char level23[1]; +}; + +static PyObject* +encoding_map_size(PyObject *obj, PyObject* args) +{ + struct encoding_map *map = (struct encoding_map*)obj; + return PyInt_FromLong(sizeof(*map) - 1 + 16*map->count2 + + 128*map->count3); +} + +static PyMethodDef encoding_map_methods[] = { + {"size", encoding_map_size, METH_NOARGS, + PyDoc_STR("Return the size (in bytes) of this object") }, + { 0 } +}; + +static void +encoding_map_dealloc(PyObject* o) +{ + PyObject_FREE(o); +} + +static PyTypeObject EncodingMapType = { + PyObject_HEAD_INIT(NULL) + 0, /*ob_size*/ + "EncodingMap", /*tp_name*/ + sizeof(struct encoding_map), /*tp_basicsize*/ + 0, /*tp_itemsize*/ + /* methods */ + encoding_map_dealloc, /*tp_dealloc*/ + 0, /*tp_print*/ + 0, /*tp_getattr*/ + 0, /*tp_setattr*/ + 0, /*tp_compare*/ + 0, /*tp_repr*/ + 0, /*tp_as_number*/ + 0, /*tp_as_sequence*/ + 0, /*tp_as_mapping*/ + 0, /*tp_hash*/ + 0, /*tp_call*/ + 0, /*tp_str*/ + 0, /*tp_getattro*/ + 0, /*tp_setattro*/ + 0, /*tp_as_buffer*/ + Py_TPFLAGS_DEFAULT, /*tp_flags*/ + 0, /*tp_doc*/ + 0, /*tp_traverse*/ + 0, /*tp_clear*/ + 0, /*tp_richcompare*/ + 0, /*tp_weaklistoffset*/ + 0, /*tp_iter*/ + 0, /*tp_iternext*/ + encoding_map_methods, /*tp_methods*/ + 0, /*tp_members*/ + 0, /*tp_getset*/ + 0, /*tp_base*/ + 0, /*tp_dict*/ + 0, /*tp_descr_get*/ + 0, /*tp_descr_set*/ + 0, /*tp_dictoffset*/ + 0, /*tp_init*/ + 0, /*tp_alloc*/ + 0, /*tp_new*/ + 0, /*tp_free*/ + 0, /*tp_is_gc*/ +}; + +PyObject* +PyUnicode_BuildEncodingMap(PyObject* string) +{ + Py_UNICODE *decode; + PyObject *result; + struct encoding_map *mresult; + int i; + int need_dict = 0; + unsigned char level1[32]; + unsigned char level2[512]; + unsigned char *mlevel1, *mlevel2, *mlevel3; + int count2 = 0, count3 = 0; + + if (!PyUnicode_Check(string) || PyUnicode_GetSize(string) != 256) { + PyErr_BadArgument(); + return NULL; + } + decode = PyUnicode_AS_UNICODE(string); + memset(level1, 0xFF, sizeof level1); + memset(level2, 0xFF, sizeof level2); + + /* If there isn't a one-to-one mapping of NULL to \0, + or if there are non-BMP characters, we need to use + a mapping dictionary. */ + if (decode[0] != 0) + need_dict = 1; + for (i = 1; i < 256; i++) { + int l1, l2; + if (decode[i] == 0 + #ifdef Py_UNICODE_WIDE + || decode[i] > 0xFFFF + #endif + ) { + need_dict = 1; + break; + } + if (decode[i] == 0xFFFE) + /* unmapped character */ + continue; + l1 = decode[i] >> 11; + l2 = decode[i] >> 7; + if (level1[l1] == 0xFF) + level1[l1] = count2++; + if (level2[l2] == 0xFF) + level2[l2] = count3++; + } + + if (count2 >= 0xFF || count3 >= 0xFF) + need_dict = 1; + + if (need_dict) { + PyObject *result = PyDict_New(); + PyObject *key, *value; + if (!result) + return NULL; + for (i = 0; i < 256; i++) { + key = value = NULL; + key = PyInt_FromLong(decode[i]); + value = PyInt_FromLong(i); + if (!key || !value) + goto failed1; + if (PyDict_SetItem(result, key, value) == -1) + goto failed1; + } + return result; + failed1: + Py_XDECREF(key); + Py_XDECREF(value); + Py_DECREF(result); + return NULL; + } + + /* Create a three-level trie */ + result = PyObject_MALLOC(sizeof(struct encoding_map) + + 16*count2 + 128*count3 - 1); + if (!result) + return PyErr_NoMemory(); + PyObject_Init(result, &EncodingMapType); + mresult = (struct encoding_map*)result; + mresult->count2 = count2; + mresult->count3 = count3; + mlevel1 = mresult->level1; + mlevel2 = mresult->level23; + mlevel3 = mresult->level23 + 16*count2; + memcpy(mlevel1, level1, 32); + memset(mlevel2, 0xFF, 16*count2); + memset(mlevel3, 0, 128*count3); + count3 = 0; + for (i = 1; i < 256; i++) { + int o1, o2, o3, i2, i3; + if (decode[i] == 0xFFFE) + /* unmapped character */ + continue; + o1 = decode[i]>>11; + o2 = (decode[i]>>7) & 0xF; + i2 = 16*mlevel1[o1] + o2; + if (mlevel2[i2] == 0xFF) + mlevel2[i2] = count3++; + o3 = decode[i] & 0x7F; + i3 = 128*mlevel2[i2] + o3; + mlevel3[i3] = i; + } + return result; +} + +static int +encoding_map_lookup(Py_UNICODE c, PyObject *mapping) +{ + struct encoding_map *map = (struct encoding_map*)mapping; + int l1 = c>>11; + int l2 = (c>>7) & 0xF; + int l3 = c & 0x7F; + int i; + +#ifdef Py_UNICODE_WIDE + if (c > 0xFFFF) { + return -1; + } +#endif + if (c == 0) + return 0; + /* level 1*/ + i = map->level1[l1]; + if (i == 0xFF) { + return -1; + } + /* level 2*/ + i = map->level23[16*i+l2]; + if (i == 0xFF) { + return -1; + } + /* level 3 */ + i = map->level23[16*map->count2 + 128*i + l3]; + if (i == 0) { + return -1; + } + return i; +} + /* Lookup the character ch in the mapping. If the character can't be found, Py_None is returned (or NULL, if another error occurred). */ @@ -3102,6 +3315,22 @@ static PyObject *charmapencode_lookup(Py_UNICODE c, PyObject *mapping) } } +static int +charmapencode_resize(PyObject **outobj, Py_ssize_t *outpos, Py_ssize_t requiredsize) +{ + Py_ssize_t outsize = PyString_GET_SIZE(*outobj); + /* exponentially overallocate to minimize reallocations */ + if (requiredsize < 2*outsize) + requiredsize = 2*outsize; + if (_PyString_Resize(outobj, requiredsize)) { + return 0; + } + return 1; +} + +typedef enum charmapencode_result { + enc_SUCCESS, enc_FAILED, enc_EXCEPTION +}charmapencode_result; /* lookup the character, put the result in the output string and adjust various state variables. Reallocate the output string if not enough space is available. Return a new reference to the object that @@ -3109,51 +3338,58 @@ static PyObject *charmapencode_lookup(Py_UNICODE c, PyObject *mapping) (in which case no character was written) or NULL, if a reallocation error occurred. The caller must decref the result */ static -PyObject *charmapencode_output(Py_UNICODE c, PyObject *mapping, +charmapencode_result charmapencode_output(Py_UNICODE c, PyObject *mapping, PyObject **outobj, Py_ssize_t *outpos) { - PyObject *rep = charmapencode_lookup(c, mapping); + PyObject *rep; + char *outstart; + Py_ssize_t outsize = PyString_GET_SIZE(*outobj); + if (mapping->ob_type == &EncodingMapType) { + int res = encoding_map_lookup(c, mapping); + Py_ssize_t requiredsize = *outpos+1; + if (res == -1) + return enc_FAILED; + if (outsize<requiredsize) + if (!charmapencode_resize(outobj, outpos, requiredsize)) + return enc_EXCEPTION; + outstart = PyString_AS_STRING(*outobj); + outstart[(*outpos)++] = (char)res; + return enc_SUCCESS; + } + + rep = charmapencode_lookup(c, mapping); if (rep==NULL) - return NULL; - else if (rep==Py_None) - return rep; - else { - char *outstart = PyString_AS_STRING(*outobj); - Py_ssize_t outsize = PyString_GET_SIZE(*outobj); + return enc_EXCEPTION; + else if (rep==Py_None) { + Py_DECREF(rep); + return enc_FAILED; + } else { if (PyInt_Check(rep)) { Py_ssize_t requiredsize = *outpos+1; - if (outsize<requiredsize) { - /* exponentially overallocate to minimize reallocations */ - if (requiredsize < 2*outsize) - requiredsize = 2*outsize; - if (_PyString_Resize(outobj, requiredsize)) { + if (outsize<requiredsize) + if (!charmapencode_resize(outobj, outpos, requiredsize)) { Py_DECREF(rep); - return NULL; + return enc_EXCEPTION; } - outstart = PyString_AS_STRING(*outobj); - } + outstart = PyString_AS_STRING(*outobj); outstart[(*outpos)++] = (char)PyInt_AS_LONG(rep); } else { const char *repchars = PyString_AS_STRING(rep); Py_ssize_t repsize = PyString_GET_SIZE(rep); Py_ssize_t requiredsize = *outpos+repsize; - if (outsize<requiredsize) { - /* exponentially overallocate to minimize reallocations */ - if (requiredsize < 2*outsize) - requiredsize = 2*outsize; - if (_PyString_Resize(outobj, requiredsize)) { + if (outsize<requiredsize) + if (!charmapencode_resize(outobj, outpos, requiredsize)) { Py_DECREF(rep); - return NULL; + return enc_EXCEPTION; } - outstart = PyString_AS_STRING(*outobj); - } + outstart = PyString_AS_STRING(*outobj); memcpy(outstart + *outpos, repchars, repsize); *outpos += repsize; } } - return rep; + return enc_SUCCESS; } /* handle an error in PyUnicode_EncodeCharmap @@ -3175,18 +3411,27 @@ int charmap_encoding_error( Py_ssize_t collpos; char *encoding = "charmap"; char *reason = "character maps to <undefined>"; + charmapencode_result x; - PyObject *x; /* find all unencodable characters */ while (collendpos < size) { - x = charmapencode_lookup(p[collendpos], mapping); - if (x==NULL) + PyObject *rep; + if (mapping->ob_type == &EncodingMapType) { + int res = encoding_map_lookup(p[collendpos], mapping); + if (res != -1) + break; + ++collendpos; + continue; + } + + rep = charmapencode_lookup(p[collendpos], mapping); + if (rep==NULL) return -1; - else if (x!=Py_None) { - Py_DECREF(x); + else if (rep!=Py_None) { + Py_DECREF(rep); break; } - Py_DECREF(x); + Py_DECREF(rep); ++collendpos; } /* cache callback name lookup @@ -3210,15 +3455,13 @@ int charmap_encoding_error( case 2: /* replace */ for (collpos = collstartpos; collpos<collendpos; ++collpos) { x = charmapencode_output('?', mapping, res, respos); - if (x==NULL) { + if (x==enc_EXCEPTION) { return -1; } - else if (x==Py_None) { - Py_DECREF(x); + else if (x==enc_FAILED) { raise_encode_exception(exceptionObject, encoding, p, size, collstartpos, collendpos, reason); return -1; } - Py_DECREF(x); } /* fall through */ case 3: /* ignore */ @@ -3232,14 +3475,12 @@ int charmap_encoding_error( sprintf(buffer, "&#%d;", (int)p[collpos]); for (cp = buffer; *cp; ++cp) { x = charmapencode_output(*cp, mapping, res, respos); - if (x==NULL) + if (x==enc_EXCEPTION) return -1; - else if (x==Py_None) { - Py_DECREF(x); + else if (x==enc_FAILED) { raise_encode_exception(exceptionObject, encoding, p, size, collstartpos, collendpos, reason); return -1; } - Py_DECREF(x); } } *inpos = collendpos; @@ -3254,17 +3495,14 @@ int charmap_encoding_error( repsize = PyUnicode_GET_SIZE(repunicode); for (uni2 = PyUnicode_AS_UNICODE(repunicode); repsize-->0; ++uni2) { x = charmapencode_output(*uni2, mapping, res, respos); - if (x==NULL) { - Py_DECREF(repunicode); + if (x==enc_EXCEPTION) { return -1; } - else if (x==Py_None) { + else if (x==enc_FAILED) { Py_DECREF(repunicode); - Py_DECREF(x); raise_encode_exception(exceptionObject, encoding, p, size, collstartpos, collendpos, reason); return -1; } - Py_DECREF(x); } *inpos = newpos; Py_DECREF(repunicode); @@ -3304,22 +3542,20 @@ PyObject *PyUnicode_EncodeCharmap(const Py_UNICODE *p, while (inpos<size) { /* try to encode it */ - PyObject *x = charmapencode_output(p[inpos], mapping, &res, &respos); - if (x==NULL) /* error */ + charmapencode_result x = charmapencode_output(p[inpos], mapping, &res, &respos); + if (x==enc_EXCEPTION) /* error */ goto onError; - if (x==Py_None) { /* unencodable character */ + if (x==enc_FAILED) { /* unencodable character */ if (charmap_encoding_error(p, size, &inpos, mapping, &exc, &known_errorHandler, &errorHandler, errors, &res, &respos)) { - Py_DECREF(x); goto onError; } } else /* done with this character => adjust input position */ ++inpos; - Py_DECREF(x); } /* Resize if we allocated to much */ diff --git a/Tools/unicode/Makefile b/Tools/unicode/Makefile index fbd3557..35744ad 100644 --- a/Tools/unicode/Makefile +++ b/Tools/unicode/Makefile @@ -78,7 +78,7 @@ cjk: build/ ### Cleanup clean: - $(RM) build/* + $(RM) -f build/* distclean: clean $(RM) -rf MAPPINGS/ diff --git a/Tools/unicode/gencodec.py b/Tools/unicode/gencodec.py index 0c55de6..8a2ca64 100644 --- a/Tools/unicode/gencodec.py +++ b/Tools/unicode/gencodec.py @@ -270,6 +270,11 @@ def codegen(name, map, encodingname, comments=1): comments=comments, precisions=(4, 2)) + if decoding_table_code: + suffix = 'table' + else: + suffix = 'map' + l = [ '''\ """ Python Character Mapping Codec %s generated from '%s' with gencodec.py. @@ -283,30 +288,20 @@ import codecs class Codec(codecs.Codec): def encode(self,input,errors='strict'): - return codecs.charmap_encode(input,errors,encoding_map) - - def decode(self,input,errors='strict'):''' % (encodingname, name) - ] - if decoding_table_code: - l.append('''\ - return codecs.charmap_decode(input,errors,decoding_table)''') - else: - l.append('''\ - return codecs.charmap_decode(input,errors,decoding_map)''') + return codecs.charmap_encode(input,errors,encoding_%s) - l.append(''' + def decode(self,input,errors='strict'): + return codecs.charmap_decode(input,errors,decoding_%s) +''' % (encodingname, name, suffix, suffix)] + l.append('''\ class IncrementalEncoder(codecs.IncrementalEncoder): def encode(self, input, final=False): - return codecs.charmap_encode(input,self.errors,encoding_map)[0] + return codecs.charmap_encode(input,self.errors,encoding_%s)[0] class IncrementalDecoder(codecs.IncrementalDecoder): - def decode(self, input, final=False):''') - if decoding_table_code: - l.append('''\ - return codecs.charmap_decode(input,self.errors,decoding_table)[0]''') - else: - l.append('''\ - return codecs.charmap_decode(input,self.errors,decoding_map)[0]''') + def decode(self, input, final=False): + return codecs.charmap_decode(input,self.errors,decoding_%s)[0]''' % + (suffix, suffix)) l.append(''' class StreamWriter(Codec,codecs.StreamWriter): @@ -319,13 +314,13 @@ class StreamReader(Codec,codecs.StreamReader): def getregentry(): return codecs.CodecInfo( - Codec().encode, - Codec().decode, name=%r, - streamwriter=StreamWriter, - streamreader=StreamReader, + encode=Codec().encode, + decode=Codec().decode, incrementalencoder=IncrementalEncoder, incrementaldecoder=IncrementalDecoder, + streamreader=StreamReader, + streamwriter=StreamWriter, ) ''' % encodingname.replace('_', '-')) @@ -342,10 +337,16 @@ def getregentry(): l.extend(decoding_table_code) # Add encoding map - l.append(''' + if decoding_table_code: + l.append(''' +### Encoding table +encoding_table=codecs.charmap_build(decoding_table) +''') + else: + l.append(''' ### Encoding Map ''') - l.extend(encoding_map_code) + l.extend(encoding_map_code) # Final new-line l.append('') |