Home | History | Annotate | Download | only in encodings
      1 """ Python Character Mapping Codec koi8_t
      2 """
      3 # http://ru.wikipedia.org/wiki/-8
      4 # http://www.opensource.apple.com/source/libiconv/libiconv-4/libiconv/tests/KOI8-T.TXT
      5 
      6 import codecs
      7 
      8 ### Codec APIs
      9 
     10 class Codec(codecs.Codec):
     11 
     12     def encode(self,input,errors='strict'):
     13         return codecs.charmap_encode(input,errors,encoding_table)
     14 
     15     def decode(self,input,errors='strict'):
     16         return codecs.charmap_decode(input,errors,decoding_table)
     17 
     18 class IncrementalEncoder(codecs.IncrementalEncoder):
     19     def encode(self, input, final=False):
     20         return codecs.charmap_encode(input,self.errors,encoding_table)[0]
     21 
     22 class IncrementalDecoder(codecs.IncrementalDecoder):
     23     def decode(self, input, final=False):
     24         return codecs.charmap_decode(input,self.errors,decoding_table)[0]
     25 
     26 class StreamWriter(Codec,codecs.StreamWriter):
     27     pass
     28 
     29 class StreamReader(Codec,codecs.StreamReader):
     30     pass
     31 
     32 ### encodings module API
     33 
     34 def getregentry():
     35     return codecs.CodecInfo(
     36         name='koi8-t',
     37         encode=Codec().encode,
     38         decode=Codec().decode,
     39         incrementalencoder=IncrementalEncoder,
     40         incrementaldecoder=IncrementalDecoder,
     41         streamreader=StreamReader,
     42         streamwriter=StreamWriter,
     43     )
     44 
     45 
     46 ### Decoding Table
     47 
     48 decoding_table = (
     49     '\x00'     #  0x00 -> NULL
     50     '\x01'     #  0x01 -> START OF HEADING
     51     '\x02'     #  0x02 -> START OF TEXT
     52     '\x03'     #  0x03 -> END OF TEXT
     53     '\x04'     #  0x04 -> END OF TRANSMISSION
     54     '\x05'     #  0x05 -> ENQUIRY
     55     '\x06'     #  0x06 -> ACKNOWLEDGE
     56     '\x07'     #  0x07 -> BELL
     57     '\x08'     #  0x08 -> BACKSPACE
     58     '\t'       #  0x09 -> HORIZONTAL TABULATION
     59     '\n'       #  0x0A -> LINE FEED
     60     '\x0b'     #  0x0B -> VERTICAL TABULATION
     61     '\x0c'     #  0x0C -> FORM FEED
     62     '\r'       #  0x0D -> CARRIAGE RETURN
     63     '\x0e'     #  0x0E -> SHIFT OUT
     64     '\x0f'     #  0x0F -> SHIFT IN
     65     '\x10'     #  0x10 -> DATA LINK ESCAPE
     66     '\x11'     #  0x11 -> DEVICE CONTROL ONE
     67     '\x12'     #  0x12 -> DEVICE CONTROL TWO
     68     '\x13'     #  0x13 -> DEVICE CONTROL THREE
     69     '\x14'     #  0x14 -> DEVICE CONTROL FOUR
     70     '\x15'     #  0x15 -> NEGATIVE ACKNOWLEDGE
     71     '\x16'     #  0x16 -> SYNCHRONOUS IDLE
     72     '\x17'     #  0x17 -> END OF TRANSMISSION BLOCK
     73     '\x18'     #  0x18 -> CANCEL
     74     '\x19'     #  0x19 -> END OF MEDIUM
     75     '\x1a'     #  0x1A -> SUBSTITUTE
     76     '\x1b'     #  0x1B -> ESCAPE
     77     '\x1c'     #  0x1C -> FILE SEPARATOR
     78     '\x1d'     #  0x1D -> GROUP SEPARATOR
     79     '\x1e'     #  0x1E -> RECORD SEPARATOR
     80     '\x1f'     #  0x1F -> UNIT SEPARATOR
     81     ' '        #  0x20 -> SPACE
     82     '!'        #  0x21 -> EXCLAMATION MARK
     83     '"'        #  0x22 -> QUOTATION MARK
     84     '#'        #  0x23 -> NUMBER SIGN
     85     '$'        #  0x24 -> DOLLAR SIGN
     86     '%'        #  0x25 -> PERCENT SIGN
     87     '&'        #  0x26 -> AMPERSAND
     88     "'"        #  0x27 -> APOSTROPHE
     89     '('        #  0x28 -> LEFT PARENTHESIS
     90     ')'        #  0x29 -> RIGHT PARENTHESIS
     91     '*'        #  0x2A -> ASTERISK
     92     '+'        #  0x2B -> PLUS SIGN
     93     ','        #  0x2C -> COMMA
     94     '-'        #  0x2D -> HYPHEN-MINUS
     95     '.'        #  0x2E -> FULL STOP
     96     '/'        #  0x2F -> SOLIDUS
     97     '0'        #  0x30 -> DIGIT ZERO
     98     '1'        #  0x31 -> DIGIT ONE
     99     '2'        #  0x32 -> DIGIT TWO
    100     '3'        #  0x33 -> DIGIT THREE
    101     '4'        #  0x34 -> DIGIT FOUR
    102     '5'        #  0x35 -> DIGIT FIVE
    103     '6'        #  0x36 -> DIGIT SIX
    104     '7'        #  0x37 -> DIGIT SEVEN
    105     '8'        #  0x38 -> DIGIT EIGHT
    106     '9'        #  0x39 -> DIGIT NINE
    107     ':'        #  0x3A -> COLON
    108     ';'        #  0x3B -> SEMICOLON
    109     '<'        #  0x3C -> LESS-THAN SIGN
    110     '='        #  0x3D -> EQUALS SIGN
    111     '>'        #  0x3E -> GREATER-THAN SIGN
    112     '?'        #  0x3F -> QUESTION MARK
    113     '@'        #  0x40 -> COMMERCIAL AT
    114     'A'        #  0x41 -> LATIN CAPITAL LETTER A
    115     'B'        #  0x42 -> LATIN CAPITAL LETTER B
    116     'C'        #  0x43 -> LATIN CAPITAL LETTER C
    117     'D'        #  0x44 -> LATIN CAPITAL LETTER D
    118     'E'        #  0x45 -> LATIN CAPITAL LETTER E
    119     'F'        #  0x46 -> LATIN CAPITAL LETTER F
    120     'G'        #  0x47 -> LATIN CAPITAL LETTER G
    121     'H'        #  0x48 -> LATIN CAPITAL LETTER H
    122     'I'        #  0x49 -> LATIN CAPITAL LETTER I
    123     'J'        #  0x4A -> LATIN CAPITAL LETTER J
    124     'K'        #  0x4B -> LATIN CAPITAL LETTER K
    125     'L'        #  0x4C -> LATIN CAPITAL LETTER L
    126     'M'        #  0x4D -> LATIN CAPITAL LETTER M
    127     'N'        #  0x4E -> LATIN CAPITAL LETTER N
    128     'O'        #  0x4F -> LATIN CAPITAL LETTER O
    129     'P'        #  0x50 -> LATIN CAPITAL LETTER P
    130     'Q'        #  0x51 -> LATIN CAPITAL LETTER Q
    131     'R'        #  0x52 -> LATIN CAPITAL LETTER R
    132     'S'        #  0x53 -> LATIN CAPITAL LETTER S
    133     'T'        #  0x54 -> LATIN CAPITAL LETTER T
    134     'U'        #  0x55 -> LATIN CAPITAL LETTER U
    135     'V'        #  0x56 -> LATIN CAPITAL LETTER V
    136     'W'        #  0x57 -> LATIN CAPITAL LETTER W
    137     'X'        #  0x58 -> LATIN CAPITAL LETTER X
    138     'Y'        #  0x59 -> LATIN CAPITAL LETTER Y
    139     'Z'        #  0x5A -> LATIN CAPITAL LETTER Z
    140     '['        #  0x5B -> LEFT SQUARE BRACKET
    141     '\\'       #  0x5C -> REVERSE SOLIDUS
    142     ']'        #  0x5D -> RIGHT SQUARE BRACKET
    143     '^'        #  0x5E -> CIRCUMFLEX ACCENT
    144     '_'        #  0x5F -> LOW LINE
    145     '`'        #  0x60 -> GRAVE ACCENT
    146     'a'        #  0x61 -> LATIN SMALL LETTER A
    147     'b'        #  0x62 -> LATIN SMALL LETTER B
    148     'c'        #  0x63 -> LATIN SMALL LETTER C
    149     'd'        #  0x64 -> LATIN SMALL LETTER D
    150     'e'        #  0x65 -> LATIN SMALL LETTER E
    151     'f'        #  0x66 -> LATIN SMALL LETTER F
    152     'g'        #  0x67 -> LATIN SMALL LETTER G
    153     'h'        #  0x68 -> LATIN SMALL LETTER H
    154     'i'        #  0x69 -> LATIN SMALL LETTER I
    155     'j'        #  0x6A -> LATIN SMALL LETTER J
    156     'k'        #  0x6B -> LATIN SMALL LETTER K
    157     'l'        #  0x6C -> LATIN SMALL LETTER L
    158     'm'        #  0x6D -> LATIN SMALL LETTER M
    159     'n'        #  0x6E -> LATIN SMALL LETTER N
    160     'o'        #  0x6F -> LATIN SMALL LETTER O
    161     'p'        #  0x70 -> LATIN SMALL LETTER P
    162     'q'        #  0x71 -> LATIN SMALL LETTER Q
    163     'r'        #  0x72 -> LATIN SMALL LETTER R
    164     's'        #  0x73 -> LATIN SMALL LETTER S
    165     't'        #  0x74 -> LATIN SMALL LETTER T
    166     'u'        #  0x75 -> LATIN SMALL LETTER U
    167     'v'        #  0x76 -> LATIN SMALL LETTER V
    168     'w'        #  0x77 -> LATIN SMALL LETTER W
    169     'x'        #  0x78 -> LATIN SMALL LETTER X
    170     'y'        #  0x79 -> LATIN SMALL LETTER Y
    171     'z'        #  0x7A -> LATIN SMALL LETTER Z
    172     '{'        #  0x7B -> LEFT CURLY BRACKET
    173     '|'        #  0x7C -> VERTICAL LINE
    174     '}'        #  0x7D -> RIGHT CURLY BRACKET
    175     '~'        #  0x7E -> TILDE
    176     '\x7f'     #  0x7F -> DELETE
    177     '\u049b'   #  0x80 -> CYRILLIC SMALL LETTER KA WITH DESCENDER
    178     '\u0493'   #  0x81 -> CYRILLIC SMALL LETTER GHE WITH STROKE
    179     '\u201a'   #  0x82 -> SINGLE LOW-9 QUOTATION MARK
    180     '\u0492'   #  0x83 -> CYRILLIC CAPITAL LETTER GHE WITH STROKE
    181     '\u201e'   #  0x84 -> DOUBLE LOW-9 QUOTATION MARK
    182     '\u2026'   #  0x85 -> HORIZONTAL ELLIPSIS
    183     '\u2020'   #  0x86 -> DAGGER
    184     '\u2021'   #  0x87 -> DOUBLE DAGGER
    185     '\ufffe'   #  0x88 -> UNDEFINED
    186     '\u2030'   #  0x89 -> PER MILLE SIGN
    187     '\u04b3'   #  0x8A -> CYRILLIC SMALL LETTER HA WITH DESCENDER
    188     '\u2039'   #  0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
    189     '\u04b2'   #  0x8C -> CYRILLIC CAPITAL LETTER HA WITH DESCENDER
    190     '\u04b7'   #  0x8D -> CYRILLIC SMALL LETTER CHE WITH DESCENDER
    191     '\u04b6'   #  0x8E -> CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
    192     '\ufffe'   #  0x8F -> UNDEFINED
    193     '\u049a'   #  0x90 -> CYRILLIC CAPITAL LETTER KA WITH DESCENDER
    194     '\u2018'   #  0x91 -> LEFT SINGLE QUOTATION MARK
    195     '\u2019'   #  0x92 -> RIGHT SINGLE QUOTATION MARK
    196     '\u201c'   #  0x93 -> LEFT DOUBLE QUOTATION MARK
    197     '\u201d'   #  0x94 -> RIGHT DOUBLE QUOTATION MARK
    198     '\u2022'   #  0x95 -> BULLET
    199     '\u2013'   #  0x96 -> EN DASH
    200     '\u2014'   #  0x97 -> EM DASH
    201     '\ufffe'   #  0x98 -> UNDEFINED
    202     '\u2122'   #  0x99 -> TRADE MARK SIGN
    203     '\ufffe'   #  0x9A -> UNDEFINED
    204     '\u203a'   #  0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
    205     '\ufffe'   #  0x9C -> UNDEFINED
    206     '\ufffe'   #  0x9D -> UNDEFINED
    207     '\ufffe'   #  0x9E -> UNDEFINED
    208     '\ufffe'   #  0x9F -> UNDEFINED
    209     '\ufffe'   #  0xA0 -> UNDEFINED
    210     '\u04ef'   #  0xA1 -> CYRILLIC SMALL LETTER U WITH MACRON
    211     '\u04ee'   #  0xA2 -> CYRILLIC CAPITAL LETTER U WITH MACRON
    212     '\u0451'   #  0xA3 -> CYRILLIC SMALL LETTER IO
    213     '\xa4'     #  0xA4 -> CURRENCY SIGN
    214     '\u04e3'   #  0xA5 -> CYRILLIC SMALL LETTER I WITH MACRON
    215     '\xa6'     #  0xA6 -> BROKEN BAR
    216     '\xa7'     #  0xA7 -> SECTION SIGN
    217     '\ufffe'   #  0xA8 -> UNDEFINED
    218     '\ufffe'   #  0xA9 -> UNDEFINED
    219     '\ufffe'   #  0xAA -> UNDEFINED
    220     '\xab'     #  0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
    221     '\xac'     #  0xAC -> NOT SIGN
    222     '\xad'     #  0xAD -> SOFT HYPHEN
    223     '\xae'     #  0xAE -> REGISTERED SIGN
    224     '\ufffe'   #  0xAF -> UNDEFINED
    225     '\xb0'     #  0xB0 -> DEGREE SIGN
    226     '\xb1'     #  0xB1 -> PLUS-MINUS SIGN
    227     '\xb2'     #  0xB2 -> SUPERSCRIPT TWO
    228     '\u0401'   #  0xB3 -> CYRILLIC CAPITAL LETTER IO
    229     '\ufffe'   #  0xB4 -> UNDEFINED
    230     '\u04e2'   #  0xB5 -> CYRILLIC CAPITAL LETTER I WITH MACRON
    231     '\xb6'     #  0xB6 -> PILCROW SIGN
    232     '\xb7'     #  0xB7 -> MIDDLE DOT
    233     '\ufffe'   #  0xB8 -> UNDEFINED
    234     '\u2116'   #  0xB9 -> NUMERO SIGN
    235     '\ufffe'   #  0xBA -> UNDEFINED
    236     '\xbb'     #  0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
    237     '\ufffe'   #  0xBC -> UNDEFINED
    238     '\ufffe'   #  0xBD -> UNDEFINED
    239     '\ufffe'   #  0xBE -> UNDEFINED
    240     '\xa9'     #  0xBF -> COPYRIGHT SIGN
    241     '\u044e'   #  0xC0 -> CYRILLIC SMALL LETTER YU
    242     '\u0430'   #  0xC1 -> CYRILLIC SMALL LETTER A
    243     '\u0431'   #  0xC2 -> CYRILLIC SMALL LETTER BE
    244     '\u0446'   #  0xC3 -> CYRILLIC SMALL LETTER TSE
    245     '\u0434'   #  0xC4 -> CYRILLIC SMALL LETTER DE
    246     '\u0435'   #  0xC5 -> CYRILLIC SMALL LETTER IE
    247     '\u0444'   #  0xC6 -> CYRILLIC SMALL LETTER EF
    248     '\u0433'   #  0xC7 -> CYRILLIC SMALL LETTER GHE
    249     '\u0445'   #  0xC8 -> CYRILLIC SMALL LETTER HA
    250     '\u0438'   #  0xC9 -> CYRILLIC SMALL LETTER I
    251     '\u0439'   #  0xCA -> CYRILLIC SMALL LETTER SHORT I
    252     '\u043a'   #  0xCB -> CYRILLIC SMALL LETTER KA
    253     '\u043b'   #  0xCC -> CYRILLIC SMALL LETTER EL
    254     '\u043c'   #  0xCD -> CYRILLIC SMALL LETTER EM
    255     '\u043d'   #  0xCE -> CYRILLIC SMALL LETTER EN
    256     '\u043e'   #  0xCF -> CYRILLIC SMALL LETTER O
    257     '\u043f'   #  0xD0 -> CYRILLIC SMALL LETTER PE
    258     '\u044f'   #  0xD1 -> CYRILLIC SMALL LETTER YA
    259     '\u0440'   #  0xD2 -> CYRILLIC SMALL LETTER ER
    260     '\u0441'   #  0xD3 -> CYRILLIC SMALL LETTER ES
    261     '\u0442'   #  0xD4 -> CYRILLIC SMALL LETTER TE
    262     '\u0443'   #  0xD5 -> CYRILLIC SMALL LETTER U
    263     '\u0436'   #  0xD6 -> CYRILLIC SMALL LETTER ZHE
    264     '\u0432'   #  0xD7 -> CYRILLIC SMALL LETTER VE
    265     '\u044c'   #  0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN
    266     '\u044b'   #  0xD9 -> CYRILLIC SMALL LETTER YERU
    267     '\u0437'   #  0xDA -> CYRILLIC SMALL LETTER ZE
    268     '\u0448'   #  0xDB -> CYRILLIC SMALL LETTER SHA
    269     '\u044d'   #  0xDC -> CYRILLIC SMALL LETTER E
    270     '\u0449'   #  0xDD -> CYRILLIC SMALL LETTER SHCHA
    271     '\u0447'   #  0xDE -> CYRILLIC SMALL LETTER CHE
    272     '\u044a'   #  0xDF -> CYRILLIC SMALL LETTER HARD SIGN
    273     '\u042e'   #  0xE0 -> CYRILLIC CAPITAL LETTER YU
    274     '\u0410'   #  0xE1 -> CYRILLIC CAPITAL LETTER A
    275     '\u0411'   #  0xE2 -> CYRILLIC CAPITAL LETTER BE
    276     '\u0426'   #  0xE3 -> CYRILLIC CAPITAL LETTER TSE
    277     '\u0414'   #  0xE4 -> CYRILLIC CAPITAL LETTER DE
    278     '\u0415'   #  0xE5 -> CYRILLIC CAPITAL LETTER IE
    279     '\u0424'   #  0xE6 -> CYRILLIC CAPITAL LETTER EF
    280     '\u0413'   #  0xE7 -> CYRILLIC CAPITAL LETTER GHE
    281     '\u0425'   #  0xE8 -> CYRILLIC CAPITAL LETTER HA
    282     '\u0418'   #  0xE9 -> CYRILLIC CAPITAL LETTER I
    283     '\u0419'   #  0xEA -> CYRILLIC CAPITAL LETTER SHORT I
    284     '\u041a'   #  0xEB -> CYRILLIC CAPITAL LETTER KA
    285     '\u041b'   #  0xEC -> CYRILLIC CAPITAL LETTER EL
    286     '\u041c'   #  0xED -> CYRILLIC CAPITAL LETTER EM
    287     '\u041d'   #  0xEE -> CYRILLIC CAPITAL LETTER EN
    288     '\u041e'   #  0xEF -> CYRILLIC CAPITAL LETTER O
    289     '\u041f'   #  0xF0 -> CYRILLIC CAPITAL LETTER PE
    290     '\u042f'   #  0xF1 -> CYRILLIC CAPITAL LETTER YA
    291     '\u0420'   #  0xF2 -> CYRILLIC CAPITAL LETTER ER
    292     '\u0421'   #  0xF3 -> CYRILLIC CAPITAL LETTER ES
    293     '\u0422'   #  0xF4 -> CYRILLIC CAPITAL LETTER TE
    294     '\u0423'   #  0xF5 -> CYRILLIC CAPITAL LETTER U
    295     '\u0416'   #  0xF6 -> CYRILLIC CAPITAL LETTER ZHE
    296     '\u0412'   #  0xF7 -> CYRILLIC CAPITAL LETTER VE
    297     '\u042c'   #  0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN
    298     '\u042b'   #  0xF9 -> CYRILLIC CAPITAL LETTER YERU
    299     '\u0417'   #  0xFA -> CYRILLIC CAPITAL LETTER ZE
    300     '\u0428'   #  0xFB -> CYRILLIC CAPITAL LETTER SHA
    301     '\u042d'   #  0xFC -> CYRILLIC CAPITAL LETTER E
    302     '\u0429'   #  0xFD -> CYRILLIC CAPITAL LETTER SHCHA
    303     '\u0427'   #  0xFE -> CYRILLIC CAPITAL LETTER CHE
    304     '\u042a'   #  0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN
    305 )
    306 
    307 ### Encoding table
    308 encoding_table=codecs.charmap_build(decoding_table)
    309