Home | History | Annotate | Download | only in encodings
      1 """ Python Character Mapping Codec mac_latin2 generated from 'MAPPINGS/VENDORS/MICSFT/MAC/LATIN2.TXT' with gencodec.py.
      2 
      3 Written by Marc-Andre Lemburg (mal (at] lemburg.com).
      4 
      5 (c) Copyright CNRI, All Rights Reserved. NO WARRANTY.
      6 (c) Copyright 2000 Guido van Rossum.
      7 
      8 """#"
      9 
     10 import codecs
     11 
     12 ### Codec APIs
     13 
     14 class Codec(codecs.Codec):
     15 
     16     def encode(self,input,errors='strict'):
     17         return codecs.charmap_encode(input,errors,encoding_table)
     18 
     19     def decode(self,input,errors='strict'):
     20         return codecs.charmap_decode(input,errors,decoding_table)
     21 
     22 class IncrementalEncoder(codecs.IncrementalEncoder):
     23     def encode(self, input, final=False):
     24         return codecs.charmap_encode(input,self.errors,encoding_table)[0]
     25 
     26 class IncrementalDecoder(codecs.IncrementalDecoder):
     27     def decode(self, input, final=False):
     28         return codecs.charmap_decode(input,self.errors,decoding_table)[0]
     29 
     30 class StreamWriter(Codec,codecs.StreamWriter):
     31     pass
     32 
     33 class StreamReader(Codec,codecs.StreamReader):
     34     pass
     35 
     36 ### encodings module API
     37 
     38 def getregentry():
     39     return codecs.CodecInfo(
     40         name='mac-latin2',
     41         encode=Codec().encode,
     42         decode=Codec().decode,
     43         incrementalencoder=IncrementalEncoder,
     44         incrementaldecoder=IncrementalDecoder,
     45         streamreader=StreamReader,
     46         streamwriter=StreamWriter,
     47     )
     48 
     49 
     50 ### Decoding Table
     51 
     52 decoding_table = (
     53     '\x00'     #  0x00 -> NULL
     54     '\x01'     #  0x01 -> START OF HEADING
     55     '\x02'     #  0x02 -> START OF TEXT
     56     '\x03'     #  0x03 -> END OF TEXT
     57     '\x04'     #  0x04 -> END OF TRANSMISSION
     58     '\x05'     #  0x05 -> ENQUIRY
     59     '\x06'     #  0x06 -> ACKNOWLEDGE
     60     '\x07'     #  0x07 -> BELL
     61     '\x08'     #  0x08 -> BACKSPACE
     62     '\t'       #  0x09 -> HORIZONTAL TABULATION
     63     '\n'       #  0x0A -> LINE FEED
     64     '\x0b'     #  0x0B -> VERTICAL TABULATION
     65     '\x0c'     #  0x0C -> FORM FEED
     66     '\r'       #  0x0D -> CARRIAGE RETURN
     67     '\x0e'     #  0x0E -> SHIFT OUT
     68     '\x0f'     #  0x0F -> SHIFT IN
     69     '\x10'     #  0x10 -> DATA LINK ESCAPE
     70     '\x11'     #  0x11 -> DEVICE CONTROL ONE
     71     '\x12'     #  0x12 -> DEVICE CONTROL TWO
     72     '\x13'     #  0x13 -> DEVICE CONTROL THREE
     73     '\x14'     #  0x14 -> DEVICE CONTROL FOUR
     74     '\x15'     #  0x15 -> NEGATIVE ACKNOWLEDGE
     75     '\x16'     #  0x16 -> SYNCHRONOUS IDLE
     76     '\x17'     #  0x17 -> END OF TRANSMISSION BLOCK
     77     '\x18'     #  0x18 -> CANCEL
     78     '\x19'     #  0x19 -> END OF MEDIUM
     79     '\x1a'     #  0x1A -> SUBSTITUTE
     80     '\x1b'     #  0x1B -> ESCAPE
     81     '\x1c'     #  0x1C -> FILE SEPARATOR
     82     '\x1d'     #  0x1D -> GROUP SEPARATOR
     83     '\x1e'     #  0x1E -> RECORD SEPARATOR
     84     '\x1f'     #  0x1F -> UNIT SEPARATOR
     85     ' '        #  0x20 -> SPACE
     86     '!'        #  0x21 -> EXCLAMATION MARK
     87     '"'        #  0x22 -> QUOTATION MARK
     88     '#'        #  0x23 -> NUMBER SIGN
     89     '$'        #  0x24 -> DOLLAR SIGN
     90     '%'        #  0x25 -> PERCENT SIGN
     91     '&'        #  0x26 -> AMPERSAND
     92     "'"        #  0x27 -> APOSTROPHE
     93     '('        #  0x28 -> LEFT PARENTHESIS
     94     ')'        #  0x29 -> RIGHT PARENTHESIS
     95     '*'        #  0x2A -> ASTERISK
     96     '+'        #  0x2B -> PLUS SIGN
     97     ','        #  0x2C -> COMMA
     98     '-'        #  0x2D -> HYPHEN-MINUS
     99     '.'        #  0x2E -> FULL STOP
    100     '/'        #  0x2F -> SOLIDUS
    101     '0'        #  0x30 -> DIGIT ZERO
    102     '1'        #  0x31 -> DIGIT ONE
    103     '2'        #  0x32 -> DIGIT TWO
    104     '3'        #  0x33 -> DIGIT THREE
    105     '4'        #  0x34 -> DIGIT FOUR
    106     '5'        #  0x35 -> DIGIT FIVE
    107     '6'        #  0x36 -> DIGIT SIX
    108     '7'        #  0x37 -> DIGIT SEVEN
    109     '8'        #  0x38 -> DIGIT EIGHT
    110     '9'        #  0x39 -> DIGIT NINE
    111     ':'        #  0x3A -> COLON
    112     ';'        #  0x3B -> SEMICOLON
    113     '<'        #  0x3C -> LESS-THAN SIGN
    114     '='        #  0x3D -> EQUALS SIGN
    115     '>'        #  0x3E -> GREATER-THAN SIGN
    116     '?'        #  0x3F -> QUESTION MARK
    117     '@'        #  0x40 -> COMMERCIAL AT
    118     'A'        #  0x41 -> LATIN CAPITAL LETTER A
    119     'B'        #  0x42 -> LATIN CAPITAL LETTER B
    120     'C'        #  0x43 -> LATIN CAPITAL LETTER C
    121     'D'        #  0x44 -> LATIN CAPITAL LETTER D
    122     'E'        #  0x45 -> LATIN CAPITAL LETTER E
    123     'F'        #  0x46 -> LATIN CAPITAL LETTER F
    124     'G'        #  0x47 -> LATIN CAPITAL LETTER G
    125     'H'        #  0x48 -> LATIN CAPITAL LETTER H
    126     'I'        #  0x49 -> LATIN CAPITAL LETTER I
    127     'J'        #  0x4A -> LATIN CAPITAL LETTER J
    128     'K'        #  0x4B -> LATIN CAPITAL LETTER K
    129     'L'        #  0x4C -> LATIN CAPITAL LETTER L
    130     'M'        #  0x4D -> LATIN CAPITAL LETTER M
    131     'N'        #  0x4E -> LATIN CAPITAL LETTER N
    132     'O'        #  0x4F -> LATIN CAPITAL LETTER O
    133     'P'        #  0x50 -> LATIN CAPITAL LETTER P
    134     'Q'        #  0x51 -> LATIN CAPITAL LETTER Q
    135     'R'        #  0x52 -> LATIN CAPITAL LETTER R
    136     'S'        #  0x53 -> LATIN CAPITAL LETTER S
    137     'T'        #  0x54 -> LATIN CAPITAL LETTER T
    138     'U'        #  0x55 -> LATIN CAPITAL LETTER U
    139     'V'        #  0x56 -> LATIN CAPITAL LETTER V
    140     'W'        #  0x57 -> LATIN CAPITAL LETTER W
    141     'X'        #  0x58 -> LATIN CAPITAL LETTER X
    142     'Y'        #  0x59 -> LATIN CAPITAL LETTER Y
    143     'Z'        #  0x5A -> LATIN CAPITAL LETTER Z
    144     '['        #  0x5B -> LEFT SQUARE BRACKET
    145     '\\'       #  0x5C -> REVERSE SOLIDUS
    146     ']'        #  0x5D -> RIGHT SQUARE BRACKET
    147     '^'        #  0x5E -> CIRCUMFLEX ACCENT
    148     '_'        #  0x5F -> LOW LINE
    149     '`'        #  0x60 -> GRAVE ACCENT
    150     'a'        #  0x61 -> LATIN SMALL LETTER A
    151     'b'        #  0x62 -> LATIN SMALL LETTER B
    152     'c'        #  0x63 -> LATIN SMALL LETTER C
    153     'd'        #  0x64 -> LATIN SMALL LETTER D
    154     'e'        #  0x65 -> LATIN SMALL LETTER E
    155     'f'        #  0x66 -> LATIN SMALL LETTER F
    156     'g'        #  0x67 -> LATIN SMALL LETTER G
    157     'h'        #  0x68 -> LATIN SMALL LETTER H
    158     'i'        #  0x69 -> LATIN SMALL LETTER I
    159     'j'        #  0x6A -> LATIN SMALL LETTER J
    160     'k'        #  0x6B -> LATIN SMALL LETTER K
    161     'l'        #  0x6C -> LATIN SMALL LETTER L
    162     'm'        #  0x6D -> LATIN SMALL LETTER M
    163     'n'        #  0x6E -> LATIN SMALL LETTER N
    164     'o'        #  0x6F -> LATIN SMALL LETTER O
    165     'p'        #  0x70 -> LATIN SMALL LETTER P
    166     'q'        #  0x71 -> LATIN SMALL LETTER Q
    167     'r'        #  0x72 -> LATIN SMALL LETTER R
    168     's'        #  0x73 -> LATIN SMALL LETTER S
    169     't'        #  0x74 -> LATIN SMALL LETTER T
    170     'u'        #  0x75 -> LATIN SMALL LETTER U
    171     'v'        #  0x76 -> LATIN SMALL LETTER V
    172     'w'        #  0x77 -> LATIN SMALL LETTER W
    173     'x'        #  0x78 -> LATIN SMALL LETTER X
    174     'y'        #  0x79 -> LATIN SMALL LETTER Y
    175     'z'        #  0x7A -> LATIN SMALL LETTER Z
    176     '{'        #  0x7B -> LEFT CURLY BRACKET
    177     '|'        #  0x7C -> VERTICAL LINE
    178     '}'        #  0x7D -> RIGHT CURLY BRACKET
    179     '~'        #  0x7E -> TILDE
    180     '\x7f'     #  0x7F -> DELETE
    181     '\xc4'     #  0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS
    182     '\u0100'   #  0x81 -> LATIN CAPITAL LETTER A WITH MACRON
    183     '\u0101'   #  0x82 -> LATIN SMALL LETTER A WITH MACRON
    184     '\xc9'     #  0x83 -> LATIN CAPITAL LETTER E WITH ACUTE
    185     '\u0104'   #  0x84 -> LATIN CAPITAL LETTER A WITH OGONEK
    186     '\xd6'     #  0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS
    187     '\xdc'     #  0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS
    188     '\xe1'     #  0x87 -> LATIN SMALL LETTER A WITH ACUTE
    189     '\u0105'   #  0x88 -> LATIN SMALL LETTER A WITH OGONEK
    190     '\u010c'   #  0x89 -> LATIN CAPITAL LETTER C WITH CARON
    191     '\xe4'     #  0x8A -> LATIN SMALL LETTER A WITH DIAERESIS
    192     '\u010d'   #  0x8B -> LATIN SMALL LETTER C WITH CARON
    193     '\u0106'   #  0x8C -> LATIN CAPITAL LETTER C WITH ACUTE
    194     '\u0107'   #  0x8D -> LATIN SMALL LETTER C WITH ACUTE
    195     '\xe9'     #  0x8E -> LATIN SMALL LETTER E WITH ACUTE
    196     '\u0179'   #  0x8F -> LATIN CAPITAL LETTER Z WITH ACUTE
    197     '\u017a'   #  0x90 -> LATIN SMALL LETTER Z WITH ACUTE
    198     '\u010e'   #  0x91 -> LATIN CAPITAL LETTER D WITH CARON
    199     '\xed'     #  0x92 -> LATIN SMALL LETTER I WITH ACUTE
    200     '\u010f'   #  0x93 -> LATIN SMALL LETTER D WITH CARON
    201     '\u0112'   #  0x94 -> LATIN CAPITAL LETTER E WITH MACRON
    202     '\u0113'   #  0x95 -> LATIN SMALL LETTER E WITH MACRON
    203     '\u0116'   #  0x96 -> LATIN CAPITAL LETTER E WITH DOT ABOVE
    204     '\xf3'     #  0x97 -> LATIN SMALL LETTER O WITH ACUTE
    205     '\u0117'   #  0x98 -> LATIN SMALL LETTER E WITH DOT ABOVE
    206     '\xf4'     #  0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
    207     '\xf6'     #  0x9A -> LATIN SMALL LETTER O WITH DIAERESIS
    208     '\xf5'     #  0x9B -> LATIN SMALL LETTER O WITH TILDE
    209     '\xfa'     #  0x9C -> LATIN SMALL LETTER U WITH ACUTE
    210     '\u011a'   #  0x9D -> LATIN CAPITAL LETTER E WITH CARON
    211     '\u011b'   #  0x9E -> LATIN SMALL LETTER E WITH CARON
    212     '\xfc'     #  0x9F -> LATIN SMALL LETTER U WITH DIAERESIS
    213     '\u2020'   #  0xA0 -> DAGGER
    214     '\xb0'     #  0xA1 -> DEGREE SIGN
    215     '\u0118'   #  0xA2 -> LATIN CAPITAL LETTER E WITH OGONEK
    216     '\xa3'     #  0xA3 -> POUND SIGN
    217     '\xa7'     #  0xA4 -> SECTION SIGN
    218     '\u2022'   #  0xA5 -> BULLET
    219     '\xb6'     #  0xA6 -> PILCROW SIGN
    220     '\xdf'     #  0xA7 -> LATIN SMALL LETTER SHARP S
    221     '\xae'     #  0xA8 -> REGISTERED SIGN
    222     '\xa9'     #  0xA9 -> COPYRIGHT SIGN
    223     '\u2122'   #  0xAA -> TRADE MARK SIGN
    224     '\u0119'   #  0xAB -> LATIN SMALL LETTER E WITH OGONEK
    225     '\xa8'     #  0xAC -> DIAERESIS
    226     '\u2260'   #  0xAD -> NOT EQUAL TO
    227     '\u0123'   #  0xAE -> LATIN SMALL LETTER G WITH CEDILLA
    228     '\u012e'   #  0xAF -> LATIN CAPITAL LETTER I WITH OGONEK
    229     '\u012f'   #  0xB0 -> LATIN SMALL LETTER I WITH OGONEK
    230     '\u012a'   #  0xB1 -> LATIN CAPITAL LETTER I WITH MACRON
    231     '\u2264'   #  0xB2 -> LESS-THAN OR EQUAL TO
    232     '\u2265'   #  0xB3 -> GREATER-THAN OR EQUAL TO
    233     '\u012b'   #  0xB4 -> LATIN SMALL LETTER I WITH MACRON
    234     '\u0136'   #  0xB5 -> LATIN CAPITAL LETTER K WITH CEDILLA
    235     '\u2202'   #  0xB6 -> PARTIAL DIFFERENTIAL
    236     '\u2211'   #  0xB7 -> N-ARY SUMMATION
    237     '\u0142'   #  0xB8 -> LATIN SMALL LETTER L WITH STROKE
    238     '\u013b'   #  0xB9 -> LATIN CAPITAL LETTER L WITH CEDILLA
    239     '\u013c'   #  0xBA -> LATIN SMALL LETTER L WITH CEDILLA
    240     '\u013d'   #  0xBB -> LATIN CAPITAL LETTER L WITH CARON
    241     '\u013e'   #  0xBC -> LATIN SMALL LETTER L WITH CARON
    242     '\u0139'   #  0xBD -> LATIN CAPITAL LETTER L WITH ACUTE
    243     '\u013a'   #  0xBE -> LATIN SMALL LETTER L WITH ACUTE
    244     '\u0145'   #  0xBF -> LATIN CAPITAL LETTER N WITH CEDILLA
    245     '\u0146'   #  0xC0 -> LATIN SMALL LETTER N WITH CEDILLA
    246     '\u0143'   #  0xC1 -> LATIN CAPITAL LETTER N WITH ACUTE
    247     '\xac'     #  0xC2 -> NOT SIGN
    248     '\u221a'   #  0xC3 -> SQUARE ROOT
    249     '\u0144'   #  0xC4 -> LATIN SMALL LETTER N WITH ACUTE
    250     '\u0147'   #  0xC5 -> LATIN CAPITAL LETTER N WITH CARON
    251     '\u2206'   #  0xC6 -> INCREMENT
    252     '\xab'     #  0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
    253     '\xbb'     #  0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
    254     '\u2026'   #  0xC9 -> HORIZONTAL ELLIPSIS
    255     '\xa0'     #  0xCA -> NO-BREAK SPACE
    256     '\u0148'   #  0xCB -> LATIN SMALL LETTER N WITH CARON
    257     '\u0150'   #  0xCC -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
    258     '\xd5'     #  0xCD -> LATIN CAPITAL LETTER O WITH TILDE
    259     '\u0151'   #  0xCE -> LATIN SMALL LETTER O WITH DOUBLE ACUTE
    260     '\u014c'   #  0xCF -> LATIN CAPITAL LETTER O WITH MACRON
    261     '\u2013'   #  0xD0 -> EN DASH
    262     '\u2014'   #  0xD1 -> EM DASH
    263     '\u201c'   #  0xD2 -> LEFT DOUBLE QUOTATION MARK
    264     '\u201d'   #  0xD3 -> RIGHT DOUBLE QUOTATION MARK
    265     '\u2018'   #  0xD4 -> LEFT SINGLE QUOTATION MARK
    266     '\u2019'   #  0xD5 -> RIGHT SINGLE QUOTATION MARK
    267     '\xf7'     #  0xD6 -> DIVISION SIGN
    268     '\u25ca'   #  0xD7 -> LOZENGE
    269     '\u014d'   #  0xD8 -> LATIN SMALL LETTER O WITH MACRON
    270     '\u0154'   #  0xD9 -> LATIN CAPITAL LETTER R WITH ACUTE
    271     '\u0155'   #  0xDA -> LATIN SMALL LETTER R WITH ACUTE
    272     '\u0158'   #  0xDB -> LATIN CAPITAL LETTER R WITH CARON
    273     '\u2039'   #  0xDC -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
    274     '\u203a'   #  0xDD -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
    275     '\u0159'   #  0xDE -> LATIN SMALL LETTER R WITH CARON
    276     '\u0156'   #  0xDF -> LATIN CAPITAL LETTER R WITH CEDILLA
    277     '\u0157'   #  0xE0 -> LATIN SMALL LETTER R WITH CEDILLA
    278     '\u0160'   #  0xE1 -> LATIN CAPITAL LETTER S WITH CARON
    279     '\u201a'   #  0xE2 -> SINGLE LOW-9 QUOTATION MARK
    280     '\u201e'   #  0xE3 -> DOUBLE LOW-9 QUOTATION MARK
    281     '\u0161'   #  0xE4 -> LATIN SMALL LETTER S WITH CARON
    282     '\u015a'   #  0xE5 -> LATIN CAPITAL LETTER S WITH ACUTE
    283     '\u015b'   #  0xE6 -> LATIN SMALL LETTER S WITH ACUTE
    284     '\xc1'     #  0xE7 -> LATIN CAPITAL LETTER A WITH ACUTE
    285     '\u0164'   #  0xE8 -> LATIN CAPITAL LETTER T WITH CARON
    286     '\u0165'   #  0xE9 -> LATIN SMALL LETTER T WITH CARON
    287     '\xcd'     #  0xEA -> LATIN CAPITAL LETTER I WITH ACUTE
    288     '\u017d'   #  0xEB -> LATIN CAPITAL LETTER Z WITH CARON
    289     '\u017e'   #  0xEC -> LATIN SMALL LETTER Z WITH CARON
    290     '\u016a'   #  0xED -> LATIN CAPITAL LETTER U WITH MACRON
    291     '\xd3'     #  0xEE -> LATIN CAPITAL LETTER O WITH ACUTE
    292     '\xd4'     #  0xEF -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX
    293     '\u016b'   #  0xF0 -> LATIN SMALL LETTER U WITH MACRON
    294     '\u016e'   #  0xF1 -> LATIN CAPITAL LETTER U WITH RING ABOVE
    295     '\xda'     #  0xF2 -> LATIN CAPITAL LETTER U WITH ACUTE
    296     '\u016f'   #  0xF3 -> LATIN SMALL LETTER U WITH RING ABOVE
    297     '\u0170'   #  0xF4 -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
    298     '\u0171'   #  0xF5 -> LATIN SMALL LETTER U WITH DOUBLE ACUTE
    299     '\u0172'   #  0xF6 -> LATIN CAPITAL LETTER U WITH OGONEK
    300     '\u0173'   #  0xF7 -> LATIN SMALL LETTER U WITH OGONEK
    301     '\xdd'     #  0xF8 -> LATIN CAPITAL LETTER Y WITH ACUTE
    302     '\xfd'     #  0xF9 -> LATIN SMALL LETTER Y WITH ACUTE
    303     '\u0137'   #  0xFA -> LATIN SMALL LETTER K WITH CEDILLA
    304     '\u017b'   #  0xFB -> LATIN CAPITAL LETTER Z WITH DOT ABOVE
    305     '\u0141'   #  0xFC -> LATIN CAPITAL LETTER L WITH STROKE
    306     '\u017c'   #  0xFD -> LATIN SMALL LETTER Z WITH DOT ABOVE
    307     '\u0122'   #  0xFE -> LATIN CAPITAL LETTER G WITH CEDILLA
    308     '\u02c7'   #  0xFF -> CARON
    309 )
    310 
    311 ### Encoding table
    312 encoding_table=codecs.charmap_build(decoding_table)
    313