Lib/encodings/cp1254.py

   1 """ Python Character Mapping Codec cp1254 generated from 'MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1254.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='cp1254',
  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'\u20ac'   #  0x80 -> EURO SIGN
 177     u'\ufffe'   #  0x81 -> UNDEFINED
 178     u'\u201a'   #  0x82 -> SINGLE LOW-9 QUOTATION MARK
 179     u'\u0192'   #  0x83 -> LATIN SMALL LETTER F WITH HOOK
 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'\u02c6'   #  0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT
 185     u'\u2030'   #  0x89 -> PER MILLE SIGN
 186     u'\u0160'   #  0x8A -> LATIN CAPITAL LETTER S WITH CARON
 187     u'\u2039'   #  0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
 188     u'\u0152'   #  0x8C -> LATIN CAPITAL LIGATURE OE
 189     u'\ufffe'   #  0x8D -> UNDEFINED
 190     u'\ufffe'   #  0x8E -> UNDEFINED
 191     u'\ufffe'   #  0x8F -> UNDEFINED
 192     u'\ufffe'   #  0x90 -> UNDEFINED
 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'\u02dc'   #  0x98 -> SMALL TILDE
 201     u'\u2122'   #  0x99 -> TRADE MARK SIGN
 202     u'\u0161'   #  0x9A -> LATIN SMALL LETTER S WITH CARON
 203     u'\u203a'   #  0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
 204     u'\u0153'   #  0x9C -> LATIN SMALL LIGATURE OE
 205     u'\ufffe'   #  0x9D -> UNDEFINED
 206     u'\ufffe'   #  0x9E -> UNDEFINED
 207     u'\u0178'   #  0x9F -> LATIN CAPITAL LETTER Y WITH DIAERESIS
 208     u'\xa0'     #  0xA0 -> NO-BREAK SPACE
 209     u'\xa1'     #  0xA1 -> INVERTED EXCLAMATION MARK
 210     u'\xa2'     #  0xA2 -> CENT SIGN
 211     u'\xa3'     #  0xA3 -> POUND SIGN
 212     u'\xa4'     #  0xA4 -> CURRENCY SIGN
 213     u'\xa5'     #  0xA5 -> YEN SIGN
 214     u'\xa6'     #  0xA6 -> BROKEN BAR
 215     u'\xa7'     #  0xA7 -> SECTION SIGN
 216     u'\xa8'     #  0xA8 -> DIAERESIS
 217     u'\xa9'     #  0xA9 -> COPYRIGHT SIGN
 218     u'\xaa'     #  0xAA -> FEMININE ORDINAL INDICATOR
 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'\xaf'     #  0xAF -> MACRON
 224     u'\xb0'     #  0xB0 -> DEGREE SIGN
 225     u'\xb1'     #  0xB1 -> PLUS-MINUS SIGN
 226     u'\xb2'     #  0xB2 -> SUPERSCRIPT TWO
 227     u'\xb3'     #  0xB3 -> SUPERSCRIPT THREE
 228     u'\xb4'     #  0xB4 -> ACUTE ACCENT
 229     u'\xb5'     #  0xB5 -> MICRO SIGN
 230     u'\xb6'     #  0xB6 -> PILCROW SIGN
 231     u'\xb7'     #  0xB7 -> MIDDLE DOT
 232     u'\xb8'     #  0xB8 -> CEDILLA
 233     u'\xb9'     #  0xB9 -> SUPERSCRIPT ONE
 234     u'\xba'     #  0xBA -> MASCULINE ORDINAL INDICATOR
 235     u'\xbb'     #  0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
 236     u'\xbc'     #  0xBC -> VULGAR FRACTION ONE QUARTER
 237     u'\xbd'     #  0xBD -> VULGAR FRACTION ONE HALF
 238     u'\xbe'     #  0xBE -> VULGAR FRACTION THREE QUARTERS
 239     u'\xbf'     #  0xBF -> INVERTED QUESTION MARK
 240     u'\xc0'     #  0xC0 -> LATIN CAPITAL LETTER A WITH GRAVE
 241     u'\xc1'     #  0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE
 242     u'\xc2'     #  0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX
 243     u'\xc3'     #  0xC3 -> LATIN CAPITAL LETTER A WITH TILDE
 244     u'\xc4'     #  0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS
 245     u'\xc5'     #  0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE
 246     u'\xc6'     #  0xC6 -> LATIN CAPITAL LETTER AE
 247     u'\xc7'     #  0xC7 -> LATIN CAPITAL LETTER C WITH CEDILLA
 248     u'\xc8'     #  0xC8 -> LATIN CAPITAL LETTER E WITH GRAVE
 249     u'\xc9'     #  0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE
 250     u'\xca'     #  0xCA -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX
 251     u'\xcb'     #  0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS
 252     u'\xcc'     #  0xCC -> LATIN CAPITAL LETTER I WITH GRAVE
 253     u'\xcd'     #  0xCD -> LATIN CAPITAL LETTER I WITH ACUTE
 254     u'\xce'     #  0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX
 255     u'\xcf'     #  0xCF -> LATIN CAPITAL LETTER I WITH DIAERESIS
 256     u'\u011e'   #  0xD0 -> LATIN CAPITAL LETTER G WITH BREVE
 257     u'\xd1'     #  0xD1 -> LATIN CAPITAL LETTER N WITH TILDE
 258     u'\xd2'     #  0xD2 -> LATIN CAPITAL LETTER O WITH GRAVE
 259     u'\xd3'     #  0xD3 -> LATIN CAPITAL LETTER O WITH ACUTE
 260     u'\xd4'     #  0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX
 261     u'\xd5'     #  0xD5 -> LATIN CAPITAL LETTER O WITH TILDE
 262     u'\xd6'     #  0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS
 263     u'\xd7'     #  0xD7 -> MULTIPLICATION SIGN
 264     u'\xd8'     #  0xD8 -> LATIN CAPITAL LETTER O WITH STROKE
 265     u'\xd9'     #  0xD9 -> LATIN CAPITAL LETTER U WITH GRAVE
 266     u'\xda'     #  0xDA -> LATIN CAPITAL LETTER U WITH ACUTE
 267     u'\xdb'     #  0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX
 268     u'\xdc'     #  0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS
 269     u'\u0130'   #  0xDD -> LATIN CAPITAL LETTER I WITH DOT ABOVE
 270     u'\u015e'   #  0xDE -> LATIN CAPITAL LETTER S WITH CEDILLA
 271     u'\xdf'     #  0xDF -> LATIN SMALL LETTER SHARP S
 272     u'\xe0'     #  0xE0 -> LATIN SMALL LETTER A WITH GRAVE
 273     u'\xe1'     #  0xE1 -> LATIN SMALL LETTER A WITH ACUTE
 274     u'\xe2'     #  0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
 275     u'\xe3'     #  0xE3 -> LATIN SMALL LETTER A WITH TILDE
 276     u'\xe4'     #  0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS
 277     u'\xe5'     #  0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE
 278     u'\xe6'     #  0xE6 -> LATIN SMALL LETTER AE
 279     u'\xe7'     #  0xE7 -> LATIN SMALL LETTER C WITH CEDILLA
 280     u'\xe8'     #  0xE8 -> LATIN SMALL LETTER E WITH GRAVE
 281     u'\xe9'     #  0xE9 -> LATIN SMALL LETTER E WITH ACUTE
 282     u'\xea'     #  0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX
 283     u'\xeb'     #  0xEB -> LATIN SMALL LETTER E WITH DIAERESIS
 284     u'\xec'     #  0xEC -> LATIN SMALL LETTER I WITH GRAVE
 285     u'\xed'     #  0xED -> LATIN SMALL LETTER I WITH ACUTE
 286     u'\xee'     #  0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX
 287     u'\xef'     #  0xEF -> LATIN SMALL LETTER I WITH DIAERESIS
 288     u'\u011f'   #  0xF0 -> LATIN SMALL LETTER G WITH BREVE
 289     u'\xf1'     #  0xF1 -> LATIN SMALL LETTER N WITH TILDE
 290     u'\xf2'     #  0xF2 -> LATIN SMALL LETTER O WITH GRAVE
 291     u'\xf3'     #  0xF3 -> LATIN SMALL LETTER O WITH ACUTE
 292     u'\xf4'     #  0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
 293     u'\xf5'     #  0xF5 -> LATIN SMALL LETTER O WITH TILDE
 294     u'\xf6'     #  0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS
 295     u'\xf7'     #  0xF7 -> DIVISION SIGN
 296     u'\xf8'     #  0xF8 -> LATIN SMALL LETTER O WITH STROKE
 297     u'\xf9'     #  0xF9 -> LATIN SMALL LETTER U WITH GRAVE
 298     u'\xfa'     #  0xFA -> LATIN SMALL LETTER U WITH ACUTE
 299     u'\xfb'     #  0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX
 300     u'\xfc'     #  0xFC -> LATIN SMALL LETTER U WITH DIAERESIS
 301     u'\u0131'   #  0xFD -> LATIN SMALL LETTER DOTLESS I
 302     u'\u015f'   #  0xFE -> LATIN SMALL LETTER S WITH CEDILLA
 303     u'\xff'     #  0xFF -> LATIN SMALL LETTER Y WITH DIAERESIS
 304 )
 305
 306 ### Encoding table
 307 encoding_table=codecs.charmap_build(decoding_table)