Lib/encodings/koi8_u.py

   1 """ Python Character Mapping Codec koi8_u generated from 'python-mappings/KOI8-U.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='koi8-u',
  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     '\x00'     #  0x00 -> NULL
  49     '\x01'     #  0x01 -> START OF HEADING
  50     '\x02'     #  0x02 -> START OF TEXT
  51     '\x03'     #  0x03 -> END OF TEXT
  52     '\x04'     #  0x04 -> END OF TRANSMISSION
  53     '\x05'     #  0x05 -> ENQUIRY
  54     '\x06'     #  0x06 -> ACKNOWLEDGE
  55     '\x07'     #  0x07 -> BELL
  56     '\x08'     #  0x08 -> BACKSPACE
  57     '\t'       #  0x09 -> HORIZONTAL TABULATION
  58     '\n'       #  0x0A -> LINE FEED
  59     '\x0b'     #  0x0B -> VERTICAL TABULATION
  60     '\x0c'     #  0x0C -> FORM FEED
  61     '\r'       #  0x0D -> CARRIAGE RETURN
  62     '\x0e'     #  0x0E -> SHIFT OUT
  63     '\x0f'     #  0x0F -> SHIFT IN
  64     '\x10'     #  0x10 -> DATA LINK ESCAPE
  65     '\x11'     #  0x11 -> DEVICE CONTROL ONE
  66     '\x12'     #  0x12 -> DEVICE CONTROL TWO
  67     '\x13'     #  0x13 -> DEVICE CONTROL THREE
  68     '\x14'     #  0x14 -> DEVICE CONTROL FOUR
  69     '\x15'     #  0x15 -> NEGATIVE ACKNOWLEDGE
  70     '\x16'     #  0x16 -> SYNCHRONOUS IDLE
  71     '\x17'     #  0x17 -> END OF TRANSMISSION BLOCK
  72     '\x18'     #  0x18 -> CANCEL
  73     '\x19'     #  0x19 -> END OF MEDIUM
  74     '\x1a'     #  0x1A -> SUBSTITUTE
  75     '\x1b'     #  0x1B -> ESCAPE
  76     '\x1c'     #  0x1C -> FILE SEPARATOR
  77     '\x1d'     #  0x1D -> GROUP SEPARATOR
  78     '\x1e'     #  0x1E -> RECORD SEPARATOR
  79     '\x1f'     #  0x1F -> UNIT SEPARATOR
  80     ' '        #  0x20 -> SPACE
  81     '!'        #  0x21 -> EXCLAMATION MARK
  82     '"'        #  0x22 -> QUOTATION MARK
  83     '#'        #  0x23 -> NUMBER SIGN
  84     '$'        #  0x24 -> DOLLAR SIGN
  85     '%'        #  0x25 -> PERCENT SIGN
  86     '&'        #  0x26 -> AMPERSAND
  87     "'"        #  0x27 -> APOSTROPHE
  88     '('        #  0x28 -> LEFT PARENTHESIS
  89     ')'        #  0x29 -> RIGHT PARENTHESIS
  90     '*'        #  0x2A -> ASTERISK
  91     '+'        #  0x2B -> PLUS SIGN
  92     ','        #  0x2C -> COMMA
  93     '-'        #  0x2D -> HYPHEN-MINUS
  94     '.'        #  0x2E -> FULL STOP
  95     '/'        #  0x2F -> SOLIDUS
  96     '0'        #  0x30 -> DIGIT ZERO
  97     '1'        #  0x31 -> DIGIT ONE
  98     '2'        #  0x32 -> DIGIT TWO
  99     '3'        #  0x33 -> DIGIT THREE
 100     '4'        #  0x34 -> DIGIT FOUR
 101     '5'        #  0x35 -> DIGIT FIVE
 102     '6'        #  0x36 -> DIGIT SIX
 103     '7'        #  0x37 -> DIGIT SEVEN
 104     '8'        #  0x38 -> DIGIT EIGHT
 105     '9'        #  0x39 -> DIGIT NINE
 106     ':'        #  0x3A -> COLON
 107     ';'        #  0x3B -> SEMICOLON
 108     '<'        #  0x3C -> LESS-THAN SIGN
 109     '='        #  0x3D -> EQUALS SIGN
 110     '>'        #  0x3E -> GREATER-THAN SIGN
 111     '?'        #  0x3F -> QUESTION MARK
 112     '@'        #  0x40 -> COMMERCIAL AT
 113     'A'        #  0x41 -> LATIN CAPITAL LETTER A
 114     'B'        #  0x42 -> LATIN CAPITAL LETTER B
 115     'C'        #  0x43 -> LATIN CAPITAL LETTER C
 116     'D'        #  0x44 -> LATIN CAPITAL LETTER D
 117     'E'        #  0x45 -> LATIN CAPITAL LETTER E
 118     'F'        #  0x46 -> LATIN CAPITAL LETTER F
 119     'G'        #  0x47 -> LATIN CAPITAL LETTER G
 120     'H'        #  0x48 -> LATIN CAPITAL LETTER H
 121     'I'        #  0x49 -> LATIN CAPITAL LETTER I
 122     'J'        #  0x4A -> LATIN CAPITAL LETTER J
 123     'K'        #  0x4B -> LATIN CAPITAL LETTER K
 124     'L'        #  0x4C -> LATIN CAPITAL LETTER L
 125     'M'        #  0x4D -> LATIN CAPITAL LETTER M
 126     'N'        #  0x4E -> LATIN CAPITAL LETTER N
 127     'O'        #  0x4F -> LATIN CAPITAL LETTER O
 128     'P'        #  0x50 -> LATIN CAPITAL LETTER P
 129     'Q'        #  0x51 -> LATIN CAPITAL LETTER Q
 130     'R'        #  0x52 -> LATIN CAPITAL LETTER R
 131     'S'        #  0x53 -> LATIN CAPITAL LETTER S
 132     'T'        #  0x54 -> LATIN CAPITAL LETTER T
 133     'U'        #  0x55 -> LATIN CAPITAL LETTER U
 134     'V'        #  0x56 -> LATIN CAPITAL LETTER V
 135     'W'        #  0x57 -> LATIN CAPITAL LETTER W
 136     'X'        #  0x58 -> LATIN CAPITAL LETTER X
 137     'Y'        #  0x59 -> LATIN CAPITAL LETTER Y
 138     'Z'        #  0x5A -> LATIN CAPITAL LETTER Z
 139     '['        #  0x5B -> LEFT SQUARE BRACKET
 140     '\\'       #  0x5C -> REVERSE SOLIDUS
 141     ']'        #  0x5D -> RIGHT SQUARE BRACKET
 142     '^'        #  0x5E -> CIRCUMFLEX ACCENT
 143     '_'        #  0x5F -> LOW LINE
 144     '`'        #  0x60 -> GRAVE ACCENT
 145     'a'        #  0x61 -> LATIN SMALL LETTER A
 146     'b'        #  0x62 -> LATIN SMALL LETTER B
 147     'c'        #  0x63 -> LATIN SMALL LETTER C
 148     'd'        #  0x64 -> LATIN SMALL LETTER D
 149     'e'        #  0x65 -> LATIN SMALL LETTER E
 150     'f'        #  0x66 -> LATIN SMALL LETTER F
 151     'g'        #  0x67 -> LATIN SMALL LETTER G
 152     'h'        #  0x68 -> LATIN SMALL LETTER H
 153     'i'        #  0x69 -> LATIN SMALL LETTER I
 154     'j'        #  0x6A -> LATIN SMALL LETTER J
 155     'k'        #  0x6B -> LATIN SMALL LETTER K
 156     'l'        #  0x6C -> LATIN SMALL LETTER L
 157     'm'        #  0x6D -> LATIN SMALL LETTER M
 158     'n'        #  0x6E -> LATIN SMALL LETTER N
 159     'o'        #  0x6F -> LATIN SMALL LETTER O
 160     'p'        #  0x70 -> LATIN SMALL LETTER P
 161     'q'        #  0x71 -> LATIN SMALL LETTER Q
 162     'r'        #  0x72 -> LATIN SMALL LETTER R
 163     's'        #  0x73 -> LATIN SMALL LETTER S
 164     't'        #  0x74 -> LATIN SMALL LETTER T
 165     'u'        #  0x75 -> LATIN SMALL LETTER U
 166     'v'        #  0x76 -> LATIN SMALL LETTER V
 167     'w'        #  0x77 -> LATIN SMALL LETTER W
 168     'x'        #  0x78 -> LATIN SMALL LETTER X
 169     'y'        #  0x79 -> LATIN SMALL LETTER Y
 170     'z'        #  0x7A -> LATIN SMALL LETTER Z
 171     '{'        #  0x7B -> LEFT CURLY BRACKET
 172     '|'        #  0x7C -> VERTICAL LINE
 173     '}'        #  0x7D -> RIGHT CURLY BRACKET
 174     '~'        #  0x7E -> TILDE
 175     '\x7f'     #  0x7F -> DELETE
 176     '\u2500'   #  0x80 -> BOX DRAWINGS LIGHT HORIZONTAL
 177     '\u2502'   #  0x81 -> BOX DRAWINGS LIGHT VERTICAL
 178     '\u250c'   #  0x82 -> BOX DRAWINGS LIGHT DOWN AND RIGHT
 179     '\u2510'   #  0x83 -> BOX DRAWINGS LIGHT DOWN AND LEFT
 180     '\u2514'   #  0x84 -> BOX DRAWINGS LIGHT UP AND RIGHT
 181     '\u2518'   #  0x85 -> BOX DRAWINGS LIGHT UP AND LEFT
 182     '\u251c'   #  0x86 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT
 183     '\u2524'   #  0x87 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT
 184     '\u252c'   #  0x88 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL
 185     '\u2534'   #  0x89 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL
 186     '\u253c'   #  0x8A -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
 187     '\u2580'   #  0x8B -> UPPER HALF BLOCK
 188     '\u2584'   #  0x8C -> LOWER HALF BLOCK
 189     '\u2588'   #  0x8D -> FULL BLOCK
 190     '\u258c'   #  0x8E -> LEFT HALF BLOCK
 191     '\u2590'   #  0x8F -> RIGHT HALF BLOCK
 192     '\u2591'   #  0x90 -> LIGHT SHADE
 193     '\u2592'   #  0x91 -> MEDIUM SHADE
 194     '\u2593'   #  0x92 -> DARK SHADE
 195     '\u2320'   #  0x93 -> TOP HALF INTEGRAL
 196     '\u25a0'   #  0x94 -> BLACK SQUARE
 197     '\u2219'   #  0x95 -> BULLET OPERATOR
 198     '\u221a'   #  0x96 -> SQUARE ROOT
 199     '\u2248'   #  0x97 -> ALMOST EQUAL TO
 200     '\u2264'   #  0x98 -> LESS-THAN OR EQUAL TO
 201     '\u2265'   #  0x99 -> GREATER-THAN OR EQUAL TO
 202     '\xa0'     #  0x9A -> NO-BREAK SPACE
 203     '\u2321'   #  0x9B -> BOTTOM HALF INTEGRAL
 204     '\xb0'     #  0x9C -> DEGREE SIGN
 205     '\xb2'     #  0x9D -> SUPERSCRIPT TWO
 206     '\xb7'     #  0x9E -> MIDDLE DOT
 207     '\xf7'     #  0x9F -> DIVISION SIGN
 208     '\u2550'   #  0xA0 -> BOX DRAWINGS DOUBLE HORIZONTAL
 209     '\u2551'   #  0xA1 -> BOX DRAWINGS DOUBLE VERTICAL
 210     '\u2552'   #  0xA2 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE
 211     '\u0451'   #  0xA3 -> CYRILLIC SMALL LETTER IO
 212     '\u0454'   #  0xA4 -> CYRILLIC SMALL LETTER UKRAINIAN IE
 213     '\u2554'   #  0xA5 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT
 214     '\u0456'   #  0xA6 -> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
 215     '\u0457'   #  0xA7 -> CYRILLIC SMALL LETTER YI (UKRAINIAN)
 216     '\u2557'   #  0xA8 -> BOX DRAWINGS DOUBLE DOWN AND LEFT
 217     '\u2558'   #  0xA9 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE
 218     '\u2559'   #  0xAA -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE
 219     '\u255a'   #  0xAB -> BOX DRAWINGS DOUBLE UP AND RIGHT
 220     '\u255b'   #  0xAC -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE
 221     '\u0491'   #  0xAD -> CYRILLIC SMALL LETTER UKRAINIAN GHE WITH UPTURN
 222     '\u255d'   #  0xAE -> BOX DRAWINGS DOUBLE UP AND LEFT
 223     '\u255e'   #  0xAF -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE
 224     '\u255f'   #  0xB0 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE
 225     '\u2560'   #  0xB1 -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
 226     '\u2561'   #  0xB2 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE
 227     '\u0401'   #  0xB3 -> CYRILLIC CAPITAL LETTER IO
 228     '\u0404'   #  0xB4 -> CYRILLIC CAPITAL LETTER UKRAINIAN IE
 229     '\u2563'   #  0xB5 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT
 230     '\u0406'   #  0xB6 -> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
 231     '\u0407'   #  0xB7 -> CYRILLIC CAPITAL LETTER YI (UKRAINIAN)
 232     '\u2566'   #  0xB8 -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
 233     '\u2567'   #  0xB9 -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE
 234     '\u2568'   #  0xBA -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE
 235     '\u2569'   #  0xBB -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL
 236     '\u256a'   #  0xBC -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE
 237     '\u0490'   #  0xBD -> CYRILLIC CAPITAL LETTER UKRAINIAN GHE WITH UPTURN
 238     '\u256c'   #  0xBE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
 239     '\xa9'     #  0xBF -> COPYRIGHT SIGN
 240     '\u044e'   #  0xC0 -> CYRILLIC SMALL LETTER YU
 241     '\u0430'   #  0xC1 -> CYRILLIC SMALL LETTER A
 242     '\u0431'   #  0xC2 -> CYRILLIC SMALL LETTER BE
 243     '\u0446'   #  0xC3 -> CYRILLIC SMALL LETTER TSE
 244     '\u0434'   #  0xC4 -> CYRILLIC SMALL LETTER DE
 245     '\u0435'   #  0xC5 -> CYRILLIC SMALL LETTER IE
 246     '\u0444'   #  0xC6 -> CYRILLIC SMALL LETTER EF
 247     '\u0433'   #  0xC7 -> CYRILLIC SMALL LETTER GHE
 248     '\u0445'   #  0xC8 -> CYRILLIC SMALL LETTER HA
 249     '\u0438'   #  0xC9 -> CYRILLIC SMALL LETTER I
 250     '\u0439'   #  0xCA -> CYRILLIC SMALL LETTER SHORT I
 251     '\u043a'   #  0xCB -> CYRILLIC SMALL LETTER KA
 252     '\u043b'   #  0xCC -> CYRILLIC SMALL LETTER EL
 253     '\u043c'   #  0xCD -> CYRILLIC SMALL LETTER EM
 254     '\u043d'   #  0xCE -> CYRILLIC SMALL LETTER EN
 255     '\u043e'   #  0xCF -> CYRILLIC SMALL LETTER O
 256     '\u043f'   #  0xD0 -> CYRILLIC SMALL LETTER PE
 257     '\u044f'   #  0xD1 -> CYRILLIC SMALL LETTER YA
 258     '\u0440'   #  0xD2 -> CYRILLIC SMALL LETTER ER
 259     '\u0441'   #  0xD3 -> CYRILLIC SMALL LETTER ES
 260     '\u0442'   #  0xD4 -> CYRILLIC SMALL LETTER TE
 261     '\u0443'   #  0xD5 -> CYRILLIC SMALL LETTER U
 262     '\u0436'   #  0xD6 -> CYRILLIC SMALL LETTER ZHE
 263     '\u0432'   #  0xD7 -> CYRILLIC SMALL LETTER VE
 264     '\u044c'   #  0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN
 265     '\u044b'   #  0xD9 -> CYRILLIC SMALL LETTER YERU
 266     '\u0437'   #  0xDA -> CYRILLIC SMALL LETTER ZE
 267     '\u0448'   #  0xDB -> CYRILLIC SMALL LETTER SHA
 268     '\u044d'   #  0xDC -> CYRILLIC SMALL LETTER E
 269     '\u0449'   #  0xDD -> CYRILLIC SMALL LETTER SHCHA
 270     '\u0447'   #  0xDE -> CYRILLIC SMALL LETTER CHE
 271     '\u044a'   #  0xDF -> CYRILLIC SMALL LETTER HARD SIGN
 272     '\u042e'   #  0xE0 -> CYRILLIC CAPITAL LETTER YU
 273     '\u0410'   #  0xE1 -> CYRILLIC CAPITAL LETTER A
 274     '\u0411'   #  0xE2 -> CYRILLIC CAPITAL LETTER BE
 275     '\u0426'   #  0xE3 -> CYRILLIC CAPITAL LETTER TSE
 276     '\u0414'   #  0xE4 -> CYRILLIC CAPITAL LETTER DE
 277     '\u0415'   #  0xE5 -> CYRILLIC CAPITAL LETTER IE
 278     '\u0424'   #  0xE6 -> CYRILLIC CAPITAL LETTER EF
 279     '\u0413'   #  0xE7 -> CYRILLIC CAPITAL LETTER GHE
 280     '\u0425'   #  0xE8 -> CYRILLIC CAPITAL LETTER HA
 281     '\u0418'   #  0xE9 -> CYRILLIC CAPITAL LETTER I
 282     '\u0419'   #  0xEA -> CYRILLIC CAPITAL LETTER SHORT I
 283     '\u041a'   #  0xEB -> CYRILLIC CAPITAL LETTER KA
 284     '\u041b'   #  0xEC -> CYRILLIC CAPITAL LETTER EL
 285     '\u041c'   #  0xED -> CYRILLIC CAPITAL LETTER EM
 286     '\u041d'   #  0xEE -> CYRILLIC CAPITAL LETTER EN
 287     '\u041e'   #  0xEF -> CYRILLIC CAPITAL LETTER O
 288     '\u041f'   #  0xF0 -> CYRILLIC CAPITAL LETTER PE
 289     '\u042f'   #  0xF1 -> CYRILLIC CAPITAL LETTER YA
 290     '\u0420'   #  0xF2 -> CYRILLIC CAPITAL LETTER ER
 291     '\u0421'   #  0xF3 -> CYRILLIC CAPITAL LETTER ES
 292     '\u0422'   #  0xF4 -> CYRILLIC CAPITAL LETTER TE
 293     '\u0423'   #  0xF5 -> CYRILLIC CAPITAL LETTER U
 294     '\u0416'   #  0xF6 -> CYRILLIC CAPITAL LETTER ZHE
 295     '\u0412'   #  0xF7 -> CYRILLIC CAPITAL LETTER VE
 296     '\u042c'   #  0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN
 297     '\u042b'   #  0xF9 -> CYRILLIC CAPITAL LETTER YERU
 298     '\u0417'   #  0xFA -> CYRILLIC CAPITAL LETTER ZE
 299     '\u0428'   #  0xFB -> CYRILLIC CAPITAL LETTER SHA
 300     '\u042d'   #  0xFC -> CYRILLIC CAPITAL LETTER E
 301     '\u0429'   #  0xFD -> CYRILLIC CAPITAL LETTER SHCHA
 302     '\u0427'   #  0xFE -> CYRILLIC CAPITAL LETTER CHE
 303     '\u042a'   #  0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN
 304 )
 305
 306 ### Encoding table
 307 encoding_table=codecs.charmap_build(decoding_table)