Lib/encodings/cp1256.py

   1 """ Python Character Mapping Codec cp1256 generated from 'MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1256.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='cp1256',
  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     '\u20ac'   #  0x80 -> EURO SIGN
 177     '\u067e'   #  0x81 -> ARABIC LETTER PEH
 178     '\u201a'   #  0x82 -> SINGLE LOW-9 QUOTATION MARK
 179     '\u0192'   #  0x83 -> LATIN SMALL LETTER F WITH HOOK
 180     '\u201e'   #  0x84 -> DOUBLE LOW-9 QUOTATION MARK
 181     '\u2026'   #  0x85 -> HORIZONTAL ELLIPSIS
 182     '\u2020'   #  0x86 -> DAGGER
 183     '\u2021'   #  0x87 -> DOUBLE DAGGER
 184     '\u02c6'   #  0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT
 185     '\u2030'   #  0x89 -> PER MILLE SIGN
 186     '\u0679'   #  0x8A -> ARABIC LETTER TTEH
 187     '\u2039'   #  0x8B -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
 188     '\u0152'   #  0x8C -> LATIN CAPITAL LIGATURE OE
 189     '\u0686'   #  0x8D -> ARABIC LETTER TCHEH
 190     '\u0698'   #  0x8E -> ARABIC LETTER JEH
 191     '\u0688'   #  0x8F -> ARABIC LETTER DDAL
 192     '\u06af'   #  0x90 -> ARABIC LETTER GAF
 193     '\u2018'   #  0x91 -> LEFT SINGLE QUOTATION MARK
 194     '\u2019'   #  0x92 -> RIGHT SINGLE QUOTATION MARK
 195     '\u201c'   #  0x93 -> LEFT DOUBLE QUOTATION MARK
 196     '\u201d'   #  0x94 -> RIGHT DOUBLE QUOTATION MARK
 197     '\u2022'   #  0x95 -> BULLET
 198     '\u2013'   #  0x96 -> EN DASH
 199     '\u2014'   #  0x97 -> EM DASH
 200     '\u06a9'   #  0x98 -> ARABIC LETTER KEHEH
 201     '\u2122'   #  0x99 -> TRADE MARK SIGN
 202     '\u0691'   #  0x9A -> ARABIC LETTER RREH
 203     '\u203a'   #  0x9B -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
 204     '\u0153'   #  0x9C -> LATIN SMALL LIGATURE OE
 205     '\u200c'   #  0x9D -> ZERO WIDTH NON-JOINER
 206     '\u200d'   #  0x9E -> ZERO WIDTH JOINER
 207     '\u06ba'   #  0x9F -> ARABIC LETTER NOON GHUNNA
 208     '\xa0'     #  0xA0 -> NO-BREAK SPACE
 209     '\u060c'   #  0xA1 -> ARABIC COMMA
 210     '\xa2'     #  0xA2 -> CENT SIGN
 211     '\xa3'     #  0xA3 -> POUND SIGN
 212     '\xa4'     #  0xA4 -> CURRENCY SIGN
 213     '\xa5'     #  0xA5 -> YEN SIGN
 214     '\xa6'     #  0xA6 -> BROKEN BAR
 215     '\xa7'     #  0xA7 -> SECTION SIGN
 216     '\xa8'     #  0xA8 -> DIAERESIS
 217     '\xa9'     #  0xA9 -> COPYRIGHT SIGN
 218     '\u06be'   #  0xAA -> ARABIC LETTER HEH DOACHASHMEE
 219     '\xab'     #  0xAB -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
 220     '\xac'     #  0xAC -> NOT SIGN
 221     '\xad'     #  0xAD -> SOFT HYPHEN
 222     '\xae'     #  0xAE -> REGISTERED SIGN
 223     '\xaf'     #  0xAF -> MACRON
 224     '\xb0'     #  0xB0 -> DEGREE SIGN
 225     '\xb1'     #  0xB1 -> PLUS-MINUS SIGN
 226     '\xb2'     #  0xB2 -> SUPERSCRIPT TWO
 227     '\xb3'     #  0xB3 -> SUPERSCRIPT THREE
 228     '\xb4'     #  0xB4 -> ACUTE ACCENT
 229     '\xb5'     #  0xB5 -> MICRO SIGN
 230     '\xb6'     #  0xB6 -> PILCROW SIGN
 231     '\xb7'     #  0xB7 -> MIDDLE DOT
 232     '\xb8'     #  0xB8 -> CEDILLA
 233     '\xb9'     #  0xB9 -> SUPERSCRIPT ONE
 234     '\u061b'   #  0xBA -> ARABIC SEMICOLON
 235     '\xbb'     #  0xBB -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
 236     '\xbc'     #  0xBC -> VULGAR FRACTION ONE QUARTER
 237     '\xbd'     #  0xBD -> VULGAR FRACTION ONE HALF
 238     '\xbe'     #  0xBE -> VULGAR FRACTION THREE QUARTERS
 239     '\u061f'   #  0xBF -> ARABIC QUESTION MARK
 240     '\u06c1'   #  0xC0 -> ARABIC LETTER HEH GOAL
 241     '\u0621'   #  0xC1 -> ARABIC LETTER HAMZA
 242     '\u0622'   #  0xC2 -> ARABIC LETTER ALEF WITH MADDA ABOVE
 243     '\u0623'   #  0xC3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE
 244     '\u0624'   #  0xC4 -> ARABIC LETTER WAW WITH HAMZA ABOVE
 245     '\u0625'   #  0xC5 -> ARABIC LETTER ALEF WITH HAMZA BELOW
 246     '\u0626'   #  0xC6 -> ARABIC LETTER YEH WITH HAMZA ABOVE
 247     '\u0627'   #  0xC7 -> ARABIC LETTER ALEF
 248     '\u0628'   #  0xC8 -> ARABIC LETTER BEH
 249     '\u0629'   #  0xC9 -> ARABIC LETTER TEH MARBUTA
 250     '\u062a'   #  0xCA -> ARABIC LETTER TEH
 251     '\u062b'   #  0xCB -> ARABIC LETTER THEH
 252     '\u062c'   #  0xCC -> ARABIC LETTER JEEM
 253     '\u062d'   #  0xCD -> ARABIC LETTER HAH
 254     '\u062e'   #  0xCE -> ARABIC LETTER KHAH
 255     '\u062f'   #  0xCF -> ARABIC LETTER DAL
 256     '\u0630'   #  0xD0 -> ARABIC LETTER THAL
 257     '\u0631'   #  0xD1 -> ARABIC LETTER REH
 258     '\u0632'   #  0xD2 -> ARABIC LETTER ZAIN
 259     '\u0633'   #  0xD3 -> ARABIC LETTER SEEN
 260     '\u0634'   #  0xD4 -> ARABIC LETTER SHEEN
 261     '\u0635'   #  0xD5 -> ARABIC LETTER SAD
 262     '\u0636'   #  0xD6 -> ARABIC LETTER DAD
 263     '\xd7'     #  0xD7 -> MULTIPLICATION SIGN
 264     '\u0637'   #  0xD8 -> ARABIC LETTER TAH
 265     '\u0638'   #  0xD9 -> ARABIC LETTER ZAH
 266     '\u0639'   #  0xDA -> ARABIC LETTER AIN
 267     '\u063a'   #  0xDB -> ARABIC LETTER GHAIN
 268     '\u0640'   #  0xDC -> ARABIC TATWEEL
 269     '\u0641'   #  0xDD -> ARABIC LETTER FEH
 270     '\u0642'   #  0xDE -> ARABIC LETTER QAF
 271     '\u0643'   #  0xDF -> ARABIC LETTER KAF
 272     '\xe0'     #  0xE0 -> LATIN SMALL LETTER A WITH GRAVE
 273     '\u0644'   #  0xE1 -> ARABIC LETTER LAM
 274     '\xe2'     #  0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
 275     '\u0645'   #  0xE3 -> ARABIC LETTER MEEM
 276     '\u0646'   #  0xE4 -> ARABIC LETTER NOON
 277     '\u0647'   #  0xE5 -> ARABIC LETTER HEH
 278     '\u0648'   #  0xE6 -> ARABIC LETTER WAW
 279     '\xe7'     #  0xE7 -> LATIN SMALL LETTER C WITH CEDILLA
 280     '\xe8'     #  0xE8 -> LATIN SMALL LETTER E WITH GRAVE
 281     '\xe9'     #  0xE9 -> LATIN SMALL LETTER E WITH ACUTE
 282     '\xea'     #  0xEA -> LATIN SMALL LETTER E WITH CIRCUMFLEX
 283     '\xeb'     #  0xEB -> LATIN SMALL LETTER E WITH DIAERESIS
 284     '\u0649'   #  0xEC -> ARABIC LETTER ALEF MAKSURA
 285     '\u064a'   #  0xED -> ARABIC LETTER YEH
 286     '\xee'     #  0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX
 287     '\xef'     #  0xEF -> LATIN SMALL LETTER I WITH DIAERESIS
 288     '\u064b'   #  0xF0 -> ARABIC FATHATAN
 289     '\u064c'   #  0xF1 -> ARABIC DAMMATAN
 290     '\u064d'   #  0xF2 -> ARABIC KASRATAN
 291     '\u064e'   #  0xF3 -> ARABIC FATHA
 292     '\xf4'     #  0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
 293     '\u064f'   #  0xF5 -> ARABIC DAMMA
 294     '\u0650'   #  0xF6 -> ARABIC KASRA
 295     '\xf7'     #  0xF7 -> DIVISION SIGN
 296     '\u0651'   #  0xF8 -> ARABIC SHADDA
 297     '\xf9'     #  0xF9 -> LATIN SMALL LETTER U WITH GRAVE
 298     '\u0652'   #  0xFA -> ARABIC SUKUN
 299     '\xfb'     #  0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX
 300     '\xfc'     #  0xFC -> LATIN SMALL LETTER U WITH DIAERESIS
 301     '\u200e'   #  0xFD -> LEFT-TO-RIGHT MARK
 302     '\u200f'   #  0xFE -> RIGHT-TO-LEFT MARK
 303     '\u06d2'   #  0xFF -> ARABIC LETTER YEH BARREE
 304 )
 305
 306 ### Encoding table
 307 encoding_table=codecs.charmap_build(decoding_table)