Home | History | Annotate | Download | only in encodings
      1 """ Python Character Mapping Codec cp1251 generated from 'MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1251.TXT' with gencodec.py.
      2 
      3 """#"

      4 
      5 import codecs
      6 
      7 ### Codec APIs

      8 
      9 class Codec(codecs.Codec):
     10 
     11     def encode(self,input,errors='strict'):
     12         return codecs.charmap_encode(input,errors,encoding_table)
     13 
     14     def decode(self,input,errors='strict'):
     15         return codecs.charmap_decode(input,errors,decoding_table)
     16 
     17 class IncrementalEncoder(codecs.IncrementalEncoder):
     18     def encode(self, input, final=False):
     19         return codecs.charmap_encode(input,self.errors,encoding_table)[0]
     20 
     21 class IncrementalDecoder(codecs.IncrementalDecoder):
     22     def decode(self, input, final=False):
     23         return codecs.charmap_decode(input,self.errors,decoding_table)[0]
     24 
     25 class StreamWriter(Codec,codecs.StreamWriter):
     26     pass
     27 
     28 class StreamReader(Codec,codecs.StreamReader):
     29     pass
     30 
     31 ### encodings module API

     32 
     33 def getregentry():
     34     return codecs.CodecInfo(
     35         name='cp1251',
     36         encode=Codec().encode,
     37         decode=Codec().decode,
     38         incrementalencoder=IncrementalEncoder,
     39         incrementaldecoder=IncrementalDecoder,
     40         streamreader=StreamReader,
     41         streamwriter=StreamWriter,
     42     )
     43 
     44 
     45 ### Decoding Table

     46 
     47 decoding_table = (
     48     u'\x00'     #  0x00 -> NULL

     49     u'\x01'     #  0x01 -> START OF HEADING

     50     u'\x02'     #  0x02 -> START OF TEXT

     51     u'\x03'     #  0x03 -> END OF TEXT

     52     u'\x04'     #  0x04 -> END OF TRANSMISSION

     53     u'\x05'     #  0x05 -> ENQUIRY

     54     u'\x06'     #  0x06 -> ACKNOWLEDGE

     55     u'\x07'     #  0x07 -> BELL

     56     u'\x08'     #  0x08 -> BACKSPACE

     57     u'\t'       #  0x09 -> HORIZONTAL TABULATION

     58     u'\n'       #  0x0A -> LINE FEED

     59     u'\x0b'     #  0x0B -> VERTICAL TABULATION

     60     u'\x0c'     #  0x0C -> FORM FEED

     61     u'\r'       #  0x0D -> CARRIAGE RETURN

     62     u'\x0e'     #  0x0E -> SHIFT OUT

     63     u'\x0f'     #  0x0F -> SHIFT IN

     64     u'\x10'     #  0x10 -> DATA LINK ESCAPE

     65     u'\x11'     #  0x11 -> DEVICE CONTROL ONE

     66     u'\x12'     #  0x12 -> DEVICE CONTROL TWO

     67     u'\x13'     #  0x13 -> DEVICE CONTROL THREE

     68     u'\x14'     #  0x14 -> DEVICE CONTROL FOUR

     69     u'\x15'     #  0x15 -> NEGATIVE ACKNOWLEDGE

     70     u'\x16'     #  0x16 -> SYNCHRONOUS IDLE

     71     u'\x17'     #  0x17 -> END OF TRANSMISSION BLOCK

     72     u'\x18'     #  0x18 -> CANCEL

     73     u'\x19'     #  0x19 -> END OF MEDIUM

     74     u'\x1a'     #  0x1A -> SUBSTITUTE

     75     u'\x1b'     #  0x1B -> ESCAPE

     76     u'\x1c'     #  0x1C -> FILE SEPARATOR

     77     u'\x1d'     #  0x1D -> GROUP SEPARATOR

     78     u'\x1e'     #  0x1E -> RECORD SEPARATOR

     79     u'\x1f'     #  0x1F -> UNIT SEPARATOR

     80     u' '        #  0x20 -> SPACE

     81     u'!'        #  0x21 -> EXCLAMATION MARK

     82     u'"'        #  0x22 -> QUOTATION MARK

     83     u'#'        #  0x23 -> NUMBER SIGN

     84     u'$'        #  0x24 -> DOLLAR SIGN

     85     u'%'        #  0x25 -> PERCENT SIGN

     86     u'&'        #  0x26 -> AMPERSAND

     87     u"'"        #  0x27 -> APOSTROPHE

     88     u'('        #  0x28 -> LEFT PARENTHESIS

     89     u')'        #  0x29 -> RIGHT PARENTHESIS

     90     u'*'        #  0x2A -> ASTERISK

     91     u'+'        #  0x2B -> PLUS SIGN

     92     u','        #  0x2C -> COMMA

     93     u'-'        #  0x2D -> HYPHEN-MINUS

     94     u'.'        #  0x2E -> FULL STOP

     95     u'/'        #  0x2F -> SOLIDUS

     96     u'0'        #  0x30 -> DIGIT ZERO

     97     u'1'        #  0x31 -> DIGIT ONE

     98     u'2'        #  0x32 -> DIGIT TWO

     99     u'3'        #  0x33 -> DIGIT THREE

    100     u'4'        #  0x34 -> DIGIT FOUR

    101     u'5'        #  0x35 -> DIGIT FIVE

    102     u'6'        #  0x36 -> DIGIT SIX

    103     u'7'        #  0x37 -> DIGIT SEVEN

    104     u'8'        #  0x38 -> DIGIT EIGHT

    105     u'9'        #  0x39 -> DIGIT NINE

    106     u':'        #  0x3A -> COLON

    107     u';'        #  0x3B -> SEMICOLON

    108     u'<'        #  0x3C -> LESS-THAN SIGN

    109     u'='        #  0x3D -> EQUALS SIGN

    110     u'>'        #  0x3E -> GREATER-THAN SIGN

    111     u'?'        #  0x3F -> QUESTION MARK

    112     u'@'        #  0x40 -> COMMERCIAL AT

    113     u'A'        #  0x41 -> LATIN CAPITAL LETTER A

    114     u'B'        #  0x42 -> LATIN CAPITAL LETTER B

    115     u'C'        #  0x43 -> LATIN CAPITAL LETTER C

    116     u'D'        #  0x44 -> LATIN CAPITAL LETTER D

    117     u'E'        #  0x45 -> LATIN CAPITAL LETTER E

    118     u'F'        #  0x46 -> LATIN CAPITAL LETTER F

    119     u'G'        #  0x47 -> LATIN CAPITAL LETTER G

    120     u'H'        #  0x48 -> LATIN CAPITAL LETTER H

    121     u'I'        #  0x49 -> LATIN CAPITAL LETTER I

    122     u'J'        #  0x4A -> LATIN CAPITAL LETTER J

    123     u'K'        #  0x4B -> LATIN CAPITAL LETTER K

    124     u'L'        #  0x4C -> LATIN CAPITAL LETTER L

    125     u'M'        #  0x4D -> LATIN CAPITAL LETTER M

    126     u'N'        #  0x4E -> LATIN CAPITAL LETTER N

    127     u'O'        #  0x4F -> LATIN CAPITAL LETTER O

    128     u'P'        #  0x50 -> LATIN CAPITAL LETTER P

    129     u'Q'        #  0x51 -> LATIN CAPITAL LETTER Q

    130     u'R'        #  0x52 -> LATIN CAPITAL LETTER R

    131     u'S'        #  0x53 -> LATIN CAPITAL LETTER S

    132     u'T'        #  0x54 -> LATIN CAPITAL LETTER T

    133     u'U'        #  0x55 -> LATIN CAPITAL LETTER U

    134     u'V'        #  0x56 -> LATIN CAPITAL LETTER V

    135     u'W'        #  0x57 -> LATIN CAPITAL LETTER W

    136     u'X'        #  0x58 -> LATIN CAPITAL LETTER X

    137     u'Y'        #  0x59 -> LATIN CAPITAL LETTER Y

    138     u'Z'        #  0x5A -> LATIN CAPITAL LETTER Z

    139     u'['        #  0x5B -> LEFT SQUARE BRACKET

    140     u'\\'       #  0x5C -> REVERSE SOLIDUS
    141     u']'        #  0x5D -> RIGHT SQUARE BRACKET
    142     u'^'        #  0x5E -> CIRCUMFLEX ACCENT
    143     u'_'        #  0x5F -> LOW LINE
    144     u'`'        #  0x60 -> GRAVE ACCENT
    145     u'a'        #  0x61 -> LATIN SMALL LETTER A
    146     u'b'        #  0x62 -> LATIN SMALL LETTER B
    147     u'c'        #  0x63 -> LATIN SMALL LETTER C
    148     u'd'        #  0x64 -> LATIN SMALL LETTER D
    149     u'e'        #  0x65 -> LATIN SMALL LETTER E
    150     u'f'        #  0x66 -> LATIN SMALL LETTER F
    151     u'g'        #  0x67 -> LATIN SMALL LETTER G
    152     u'h'        #  0x68 -> LATIN SMALL LETTER H
    153     u'i'        #  0x69 -> LATIN SMALL LETTER I
    154     u'j'        #  0x6A -> LATIN SMALL LETTER J
    155     u'k'        #  0x6B -> LATIN SMALL LETTER K
    156     u'l'        #  0x6C -> LATIN SMALL LETTER L
    157     u'm'        #  0x6D -> LATIN SMALL LETTER M
    158     u'n'        #  0x6E -> LATIN SMALL LETTER N
    159     u'o'        #  0x6F -> LATIN SMALL LETTER O
    160     u'p'        #  0x70 -> LATIN SMALL LETTER P
    161     u'q'        #  0x71 -> LATIN SMALL LETTER Q
    162     u'r'        #  0x72 -> LATIN SMALL LETTER R
    163     u's'        #  0x73 -> LATIN SMALL LETTER S
    164     u't'        #  0x74 -> LATIN SMALL LETTER T
    165     u'u'        #  0x75 -> LATIN SMALL LETTER U
    166     u'v'        #  0x76 -> LATIN SMALL LETTER V
    167     u'w'        #  0x77 -> LATIN SMALL LETTER W
    168     u'x'        #  0x78 -> LATIN SMALL LETTER X
    169     u'y'        #  0x79 -> LATIN SMALL LETTER Y
    170     u'z'        #  0x7A -> LATIN SMALL LETTER Z
    171     u'{'        #  0x7B -> LEFT CURLY BRACKET
    172     u'|'        #  0x7C -> VERTICAL LINE
    173     u'}'        #  0x7D -> RIGHT CURLY BRACKET
    174     u'~'        #  0x7E -> TILDE
    175     u'\x7f'     #  0x7F -> DELETE
    176     u'\u0402'   #  0x80 -> CYRILLIC CAPITAL LETTER DJE
    177     u'\u0403'   #  0x81 -> CYRILLIC CAPITAL LETTER GJE
    178     u'\u201a'   #  0x82 -> SINGLE LOW-9 QUOTATION MARK
    179     u'\u0453'   #  0x83 -> CYRILLIC SMALL LETTER GJE
    180     u'\u201e'   #  0x84 -> DOUBLE LOW-9 QUOTATION MARK
    181     u'\u2026'   #  0x85 -> HORIZONTAL ELLIPSIS
    182     u'\u2020'   #  0x86 -> DAGGER
    183     u'\u2021'   #  0x87 -> DOUBLE DAGGER
    184     u'\u20ac'   #  0x88 -> EURO SIGN
    185     u'\u2030'   #  0x89 -> PER MILLE SIGN
    186     u'\u0409'   #  0x8A -> CYRILLIC CAPITAL LETTER LJE
    187     u'\u2039'   #  0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
    188     u'\u040a'   #  0x8C -> CYRILLIC CAPITAL LETTER NJE
    189     u'\u040c'   #  0x8D -> CYRILLIC CAPITAL LETTER KJE
    190     u'\u040b'   #  0x8E -> CYRILLIC CAPITAL LETTER TSHE
    191     u'\u040f'   #  0x8F -> CYRILLIC CAPITAL LETTER DZHE
    192     u'\u0452'   #  0x90 -> CYRILLIC SMALL LETTER DJE
    193     u'\u2018'   #  0x91 -> LEFT SINGLE QUOTATION MARK
    194     u'\u2019'   #  0x92 -> RIGHT SINGLE QUOTATION MARK
    195     u'\u201c'   #  0x93 -> LEFT DOUBLE QUOTATION MARK
    196     u'\u201d'   #  0x94 -> RIGHT DOUBLE QUOTATION MARK
    197     u'\u2022'   #  0x95 -> BULLET
    198     u'\u2013'   #  0x96 -> EN DASH
    199     u'\u2014'   #  0x97 -> EM DASH
    200     u'\ufffe'   #  0x98 -> UNDEFINED
    201     u'\u2122'   #  0x99 -> TRADE MARK SIGN
    202     u'\u0459'   #  0x9A -> CYRILLIC SMALL LETTER LJE
    203     u'\u203a'   #  0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
    204     u'\u045a'   #  0x9C -> CYRILLIC SMALL LETTER NJE
    205     u'\u045c'   #  0x9D -> CYRILLIC SMALL LETTER KJE
    206     u'\u045b'   #  0x9E -> CYRILLIC SMALL LETTER TSHE
    207     u'\u045f'   #  0x9F -> CYRILLIC SMALL LETTER DZHE
    208     u'\xa0'     #  0xA0 -> NO-BREAK SPACE
    209     u'\u040e'   #  0xA1 -> CYRILLIC CAPITAL LETTER SHORT U
    210     u'\u045e'   #  0xA2 -> CYRILLIC SMALL LETTER SHORT U
    211     u'\u0408'   #  0xA3 -> CYRILLIC CAPITAL LETTER JE
    212     u'\xa4'     #  0xA4 -> CURRENCY SIGN
    213     u'\u0490'   #  0xA5 -> CYRILLIC CAPITAL LETTER GHE WITH UPTURN
    214     u'\xa6'     #  0xA6 -> BROKEN BAR
    215     u'\xa7'     #  0xA7 -> SECTION SIGN
    216     u'\u0401'   #  0xA8 -> CYRILLIC CAPITAL LETTER IO
    217     u'\xa9'     #  0xA9 -> COPYRIGHT SIGN
    218     u'\u0404'   #  0xAA -> CYRILLIC CAPITAL LETTER UKRAINIAN IE
    219     u'\xab'     #  0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
    220     u'\xac'     #  0xAC -> NOT SIGN
    221     u'\xad'     #  0xAD -> SOFT HYPHEN
    222     u'\xae'     #  0xAE -> REGISTERED SIGN
    223     u'\u0407'   #  0xAF -> CYRILLIC CAPITAL LETTER YI
    224     u'\xb0'     #  0xB0 -> DEGREE SIGN
    225     u'\xb1'     #  0xB1 -> PLUS-MINUS SIGN
    226     u'\u0406'   #  0xB2 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
    227     u'\u0456'   #  0xB3 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
    228     u'\u0491'   #  0xB4 -> CYRILLIC SMALL LETTER GHE WITH UPTURN
    229     u'\xb5'     #  0xB5 -> MICRO SIGN
    230     u'\xb6'     #  0xB6 -> PILCROW SIGN
    231     u'\xb7'     #  0xB7 -> MIDDLE DOT
    232     u'\u0451'   #  0xB8 -> CYRILLIC SMALL LETTER IO
    233     u'\u2116'   #  0xB9 -> NUMERO SIGN
    234     u'\u0454'   #  0xBA -> CYRILLIC SMALL LETTER UKRAINIAN IE
    235     u'\xbb'     #  0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
    236     u'\u0458'   #  0xBC -> CYRILLIC SMALL LETTER JE
    237     u'\u0405'   #  0xBD -> CYRILLIC CAPITAL LETTER DZE
    238     u'\u0455'   #  0xBE -> CYRILLIC SMALL LETTER DZE
    239     u'\u0457'   #  0xBF -> CYRILLIC SMALL LETTER YI
    240     u'\u0410'   #  0xC0 -> CYRILLIC CAPITAL LETTER A
    241     u'\u0411'   #  0xC1 -> CYRILLIC CAPITAL LETTER BE
    242     u'\u0412'   #  0xC2 -> CYRILLIC CAPITAL LETTER VE
    243     u'\u0413'   #  0xC3 -> CYRILLIC CAPITAL LETTER GHE
    244     u'\u0414'   #  0xC4 -> CYRILLIC CAPITAL LETTER DE
    245     u'\u0415'   #  0xC5 -> CYRILLIC CAPITAL LETTER IE
    246     u'\u0416'   #  0xC6 -> CYRILLIC CAPITAL LETTER ZHE
    247     u'\u0417'   #  0xC7 -> CYRILLIC CAPITAL LETTER ZE
    248     u'\u0418'   #  0xC8 -> CYRILLIC CAPITAL LETTER I
    249     u'\u0419'   #  0xC9 -> CYRILLIC CAPITAL LETTER SHORT I
    250     u'\u041a'   #  0xCA -> CYRILLIC CAPITAL LETTER KA
    251     u'\u041b'   #  0xCB -> CYRILLIC CAPITAL LETTER EL
    252     u'\u041c'   #  0xCC -> CYRILLIC CAPITAL LETTER EM
    253     u'\u041d'   #  0xCD -> CYRILLIC CAPITAL LETTER EN
    254     u'\u041e'   #  0xCE -> CYRILLIC CAPITAL LETTER O
    255     u'\u041f'   #  0xCF -> CYRILLIC CAPITAL LETTER PE
    256     u'\u0420'   #  0xD0 -> CYRILLIC CAPITAL LETTER ER
    257     u'\u0421'   #  0xD1 -> CYRILLIC CAPITAL LETTER ES
    258     u'\u0422'   #  0xD2 -> CYRILLIC CAPITAL LETTER TE
    259     u'\u0423'   #  0xD3 -> CYRILLIC CAPITAL LETTER U
    260     u'\u0424'   #  0xD4 -> CYRILLIC CAPITAL LETTER EF
    261     u'\u0425'   #  0xD5 -> CYRILLIC CAPITAL LETTER HA
    262     u'\u0426'   #  0xD6 -> CYRILLIC CAPITAL LETTER TSE
    263     u'\u0427'   #  0xD7 -> CYRILLIC CAPITAL LETTER CHE
    264     u'\u0428'   #  0xD8 -> CYRILLIC CAPITAL LETTER SHA
    265     u'\u0429'   #  0xD9 -> CYRILLIC CAPITAL LETTER SHCHA
    266     u'\u042a'   #  0xDA -> CYRILLIC CAPITAL LETTER HARD SIGN
    267     u'\u042b'   #  0xDB -> CYRILLIC CAPITAL LETTER YERU
    268     u'\u042c'   #  0xDC -> CYRILLIC CAPITAL LETTER SOFT SIGN
    269     u'\u042d'   #  0xDD -> CYRILLIC CAPITAL LETTER E
    270     u'\u042e'   #  0xDE -> CYRILLIC CAPITAL LETTER YU
    271     u'\u042f'   #  0xDF -> CYRILLIC CAPITAL LETTER YA
    272     u'\u0430'   #  0xE0 -> CYRILLIC SMALL LETTER A
    273     u'\u0431'   #  0xE1 -> CYRILLIC SMALL LETTER BE
    274     u'\u0432'   #  0xE2 -> CYRILLIC SMALL LETTER VE
    275     u'\u0433'   #  0xE3 -> CYRILLIC SMALL LETTER GHE
    276     u'\u0434'   #  0xE4 -> CYRILLIC SMALL LETTER DE
    277     u'\u0435'   #  0xE5 -> CYRILLIC SMALL LETTER IE
    278     u'\u0436'   #  0xE6 -> CYRILLIC SMALL LETTER ZHE
    279     u'\u0437'   #  0xE7 -> CYRILLIC SMALL LETTER ZE
    280     u'\u0438'   #  0xE8 -> CYRILLIC SMALL LETTER I
    281     u'\u0439'   #  0xE9 -> CYRILLIC SMALL LETTER SHORT I
    282     u'\u043a'   #  0xEA -> CYRILLIC SMALL LETTER KA
    283     u'\u043b'   #  0xEB -> CYRILLIC SMALL LETTER EL
    284     u'\u043c'   #  0xEC -> CYRILLIC SMALL LETTER EM
    285     u'\u043d'   #  0xED -> CYRILLIC SMALL LETTER EN
    286     u'\u043e'   #  0xEE -> CYRILLIC SMALL LETTER O
    287     u'\u043f'   #  0xEF -> CYRILLIC SMALL LETTER PE
    288     u'\u0440'   #  0xF0 -> CYRILLIC SMALL LETTER ER
    289     u'\u0441'   #  0xF1 -> CYRILLIC SMALL LETTER ES
    290     u'\u0442'   #  0xF2 -> CYRILLIC SMALL LETTER TE
    291     u'\u0443'   #  0xF3 -> CYRILLIC SMALL LETTER U
    292     u'\u0444'   #  0xF4 -> CYRILLIC SMALL LETTER EF
    293     u'\u0445'   #  0xF5 -> CYRILLIC SMALL LETTER HA
    294     u'\u0446'   #  0xF6 -> CYRILLIC SMALL LETTER TSE
    295     u'\u0447'   #  0xF7 -> CYRILLIC SMALL LETTER CHE
    296     u'\u0448'   #  0xF8 -> CYRILLIC SMALL LETTER SHA
    297     u'\u0449'   #  0xF9 -> CYRILLIC SMALL LETTER SHCHA
    298     u'\u044a'   #  0xFA -> CYRILLIC SMALL LETTER HARD SIGN
    299     u'\u044b'   #  0xFB -> CYRILLIC SMALL LETTER YERU
    300     u'\u044c'   #  0xFC -> CYRILLIC SMALL LETTER SOFT SIGN
    301     u'\u044d'   #  0xFD -> CYRILLIC SMALL LETTER E
    302     u'\u044e'   #  0xFE -> CYRILLIC SMALL LETTER YU
    303     u'\u044f'   #  0xFF -> CYRILLIC SMALL LETTER YA
    304 )
    305 
    306 ### Encoding table
    307 encoding_table=codecs.charmap_build(decoding_table)
    308