Home | History | Annotate | Download | only in encodings
      1 """ Python Character Mapping Codec generated from 'PTCP154.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='ptcp154',
     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 (DEL)
    181     '\u0496'   #  0x80 -> CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER
    182     '\u0492'   #  0x81 -> CYRILLIC CAPITAL LETTER GHE WITH STROKE
    183     '\u04ee'   #  0x82 -> CYRILLIC CAPITAL LETTER U WITH MACRON
    184     '\u0493'   #  0x83 -> CYRILLIC SMALL LETTER GHE WITH STROKE
    185     '\u201e'   #  0x84 -> DOUBLE LOW-9 QUOTATION MARK
    186     '\u2026'   #  0x85 -> HORIZONTAL ELLIPSIS
    187     '\u04b6'   #  0x86 -> CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
    188     '\u04ae'   #  0x87 -> CYRILLIC CAPITAL LETTER STRAIGHT U
    189     '\u04b2'   #  0x88 -> CYRILLIC CAPITAL LETTER HA WITH DESCENDER
    190     '\u04af'   #  0x89 -> CYRILLIC SMALL LETTER STRAIGHT U
    191     '\u04a0'   #  0x8A -> CYRILLIC CAPITAL LETTER BASHKIR KA
    192     '\u04e2'   #  0x8B -> CYRILLIC CAPITAL LETTER I WITH MACRON
    193     '\u04a2'   #  0x8C -> CYRILLIC CAPITAL LETTER EN WITH DESCENDER
    194     '\u049a'   #  0x8D -> CYRILLIC CAPITAL LETTER KA WITH DESCENDER
    195     '\u04ba'   #  0x8E -> CYRILLIC CAPITAL LETTER SHHA
    196     '\u04b8'   #  0x8F -> CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
    197     '\u0497'   #  0x90 -> CYRILLIC SMALL LETTER ZHE WITH DESCENDER
    198     '\u2018'   #  0x91 -> LEFT SINGLE QUOTATION MARK
    199     '\u2019'   #  0x92 -> RIGHT SINGLE QUOTATION MARK
    200     '\u201c'   #  0x93 -> LEFT DOUBLE QUOTATION MARK
    201     '\u201d'   #  0x94 -> RIGHT DOUBLE QUOTATION MARK
    202     '\u2022'   #  0x95 -> BULLET
    203     '\u2013'   #  0x96 -> EN DASH
    204     '\u2014'   #  0x97 -> EM DASH
    205     '\u04b3'   #  0x98 -> CYRILLIC SMALL LETTER HA WITH DESCENDER
    206     '\u04b7'   #  0x99 -> CYRILLIC SMALL LETTER CHE WITH DESCENDER
    207     '\u04a1'   #  0x9A -> CYRILLIC SMALL LETTER BASHKIR KA
    208     '\u04e3'   #  0x9B -> CYRILLIC SMALL LETTER I WITH MACRON
    209     '\u04a3'   #  0x9C -> CYRILLIC SMALL LETTER EN WITH DESCENDER
    210     '\u049b'   #  0x9D -> CYRILLIC SMALL LETTER KA WITH DESCENDER
    211     '\u04bb'   #  0x9E -> CYRILLIC SMALL LETTER SHHA
    212     '\u04b9'   #  0x9F -> CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE
    213     '\xa0'     #  0xA0 -> NO-BREAK SPACE
    214     '\u040e'   #  0xA1 -> CYRILLIC CAPITAL LETTER SHORT U (Byelorussian)
    215     '\u045e'   #  0xA2 -> CYRILLIC SMALL LETTER SHORT U (Byelorussian)
    216     '\u0408'   #  0xA3 -> CYRILLIC CAPITAL LETTER JE
    217     '\u04e8'   #  0xA4 -> CYRILLIC CAPITAL LETTER BARRED O
    218     '\u0498'   #  0xA5 -> CYRILLIC CAPITAL LETTER ZE WITH DESCENDER
    219     '\u04b0'   #  0xA6 -> CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE
    220     '\xa7'     #  0xA7 -> SECTION SIGN
    221     '\u0401'   #  0xA8 -> CYRILLIC CAPITAL LETTER IO
    222     '\xa9'     #  0xA9 -> COPYRIGHT SIGN
    223     '\u04d8'   #  0xAA -> CYRILLIC CAPITAL LETTER SCHWA
    224     '\xab'     #  0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
    225     '\xac'     #  0xAC -> NOT SIGN
    226     '\u04ef'   #  0xAD -> CYRILLIC SMALL LETTER U WITH MACRON
    227     '\xae'     #  0xAE -> REGISTERED SIGN
    228     '\u049c'   #  0xAF -> CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
    229     '\xb0'     #  0xB0 -> DEGREE SIGN
    230     '\u04b1'   #  0xB1 -> CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE
    231     '\u0406'   #  0xB2 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
    232     '\u0456'   #  0xB3 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
    233     '\u0499'   #  0xB4 -> CYRILLIC SMALL LETTER ZE WITH DESCENDER
    234     '\u04e9'   #  0xB5 -> CYRILLIC SMALL LETTER BARRED O
    235     '\xb6'     #  0xB6 -> PILCROW SIGN
    236     '\xb7'     #  0xB7 -> MIDDLE DOT
    237     '\u0451'   #  0xB8 -> CYRILLIC SMALL LETTER IO
    238     '\u2116'   #  0xB9 -> NUMERO SIGN
    239     '\u04d9'   #  0xBA -> CYRILLIC SMALL LETTER SCHWA
    240     '\xbb'     #  0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
    241     '\u0458'   #  0xBC -> CYRILLIC SMALL LETTER JE
    242     '\u04aa'   #  0xBD -> CYRILLIC CAPITAL LETTER ES WITH DESCENDER
    243     '\u04ab'   #  0xBE -> CYRILLIC SMALL LETTER ES WITH DESCENDER
    244     '\u049d'   #  0xBF -> CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE
    245     '\u0410'   #  0xC0 -> CYRILLIC CAPITAL LETTER A
    246     '\u0411'   #  0xC1 -> CYRILLIC CAPITAL LETTER BE
    247     '\u0412'   #  0xC2 -> CYRILLIC CAPITAL LETTER VE
    248     '\u0413'   #  0xC3 -> CYRILLIC CAPITAL LETTER GHE
    249     '\u0414'   #  0xC4 -> CYRILLIC CAPITAL LETTER DE
    250     '\u0415'   #  0xC5 -> CYRILLIC CAPITAL LETTER IE
    251     '\u0416'   #  0xC6 -> CYRILLIC CAPITAL LETTER ZHE
    252     '\u0417'   #  0xC7 -> CYRILLIC CAPITAL LETTER ZE
    253     '\u0418'   #  0xC8 -> CYRILLIC CAPITAL LETTER I
    254     '\u0419'   #  0xC9 -> CYRILLIC CAPITAL LETTER SHORT I
    255     '\u041a'   #  0xCA -> CYRILLIC CAPITAL LETTER KA
    256     '\u041b'   #  0xCB -> CYRILLIC CAPITAL LETTER EL
    257     '\u041c'   #  0xCC -> CYRILLIC CAPITAL LETTER EM
    258     '\u041d'   #  0xCD -> CYRILLIC CAPITAL LETTER EN
    259     '\u041e'   #  0xCE -> CYRILLIC CAPITAL LETTER O
    260     '\u041f'   #  0xCF -> CYRILLIC CAPITAL LETTER PE
    261     '\u0420'   #  0xD0 -> CYRILLIC CAPITAL LETTER ER
    262     '\u0421'   #  0xD1 -> CYRILLIC CAPITAL LETTER ES
    263     '\u0422'   #  0xD2 -> CYRILLIC CAPITAL LETTER TE
    264     '\u0423'   #  0xD3 -> CYRILLIC CAPITAL LETTER U
    265     '\u0424'   #  0xD4 -> CYRILLIC CAPITAL LETTER EF
    266     '\u0425'   #  0xD5 -> CYRILLIC CAPITAL LETTER HA
    267     '\u0426'   #  0xD6 -> CYRILLIC CAPITAL LETTER TSE
    268     '\u0427'   #  0xD7 -> CYRILLIC CAPITAL LETTER CHE
    269     '\u0428'   #  0xD8 -> CYRILLIC CAPITAL LETTER SHA
    270     '\u0429'   #  0xD9 -> CYRILLIC CAPITAL LETTER SHCHA
    271     '\u042a'   #  0xDA -> CYRILLIC CAPITAL LETTER HARD SIGN
    272     '\u042b'   #  0xDB -> CYRILLIC CAPITAL LETTER YERU
    273     '\u042c'   #  0xDC -> CYRILLIC CAPITAL LETTER SOFT SIGN
    274     '\u042d'   #  0xDD -> CYRILLIC CAPITAL LETTER E
    275     '\u042e'   #  0xDE -> CYRILLIC CAPITAL LETTER YU
    276     '\u042f'   #  0xDF -> CYRILLIC CAPITAL LETTER YA
    277     '\u0430'   #  0xE0 -> CYRILLIC SMALL LETTER A
    278     '\u0431'   #  0xE1 -> CYRILLIC SMALL LETTER BE
    279     '\u0432'   #  0xE2 -> CYRILLIC SMALL LETTER VE
    280     '\u0433'   #  0xE3 -> CYRILLIC SMALL LETTER GHE
    281     '\u0434'   #  0xE4 -> CYRILLIC SMALL LETTER DE
    282     '\u0435'   #  0xE5 -> CYRILLIC SMALL LETTER IE
    283     '\u0436'   #  0xE6 -> CYRILLIC SMALL LETTER ZHE
    284     '\u0437'   #  0xE7 -> CYRILLIC SMALL LETTER ZE
    285     '\u0438'   #  0xE8 -> CYRILLIC SMALL LETTER I
    286     '\u0439'   #  0xE9 -> CYRILLIC SMALL LETTER SHORT I
    287     '\u043a'   #  0xEA -> CYRILLIC SMALL LETTER KA
    288     '\u043b'   #  0xEB -> CYRILLIC SMALL LETTER EL
    289     '\u043c'   #  0xEC -> CYRILLIC SMALL LETTER EM
    290     '\u043d'   #  0xED -> CYRILLIC SMALL LETTER EN
    291     '\u043e'   #  0xEE -> CYRILLIC SMALL LETTER O
    292     '\u043f'   #  0xEF -> CYRILLIC SMALL LETTER PE
    293     '\u0440'   #  0xF0 -> CYRILLIC SMALL LETTER ER
    294     '\u0441'   #  0xF1 -> CYRILLIC SMALL LETTER ES
    295     '\u0442'   #  0xF2 -> CYRILLIC SMALL LETTER TE
    296     '\u0443'   #  0xF3 -> CYRILLIC SMALL LETTER U
    297     '\u0444'   #  0xF4 -> CYRILLIC SMALL LETTER EF
    298     '\u0445'   #  0xF5 -> CYRILLIC SMALL LETTER HA
    299     '\u0446'   #  0xF6 -> CYRILLIC SMALL LETTER TSE
    300     '\u0447'   #  0xF7 -> CYRILLIC SMALL LETTER CHE
    301     '\u0448'   #  0xF8 -> CYRILLIC SMALL LETTER SHA
    302     '\u0449'   #  0xF9 -> CYRILLIC SMALL LETTER SHCHA
    303     '\u044a'   #  0xFA -> CYRILLIC SMALL LETTER HARD SIGN
    304     '\u044b'   #  0xFB -> CYRILLIC SMALL LETTER YERU
    305     '\u044c'   #  0xFC -> CYRILLIC SMALL LETTER SOFT SIGN
    306     '\u044d'   #  0xFD -> CYRILLIC SMALL LETTER E
    307     '\u044e'   #  0xFE -> CYRILLIC SMALL LETTER YU
    308     '\u044f'   #  0xFF -> CYRILLIC SMALL LETTER YA
    309 )
    310 
    311 ### Encoding table
    312 encoding_table=codecs.charmap_build(decoding_table)
    313