Lib/encodings/cp1006.py

   1 """ Python Character Mapping Codec cp1006 generated from 'MAPPINGS/VENDORS/MISC/CP1006.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='cp1006',
  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     '\x80'     #  0x80 -> <control>
 177     '\x81'     #  0x81 -> <control>
 178     '\x82'     #  0x82 -> <control>
 179     '\x83'     #  0x83 -> <control>
 180     '\x84'     #  0x84 -> <control>
 181     '\x85'     #  0x85 -> <control>
 182     '\x86'     #  0x86 -> <control>
 183     '\x87'     #  0x87 -> <control>
 184     '\x88'     #  0x88 -> <control>
 185     '\x89'     #  0x89 -> <control>
 186     '\x8a'     #  0x8A -> <control>
 187     '\x8b'     #  0x8B -> <control>
 188     '\x8c'     #  0x8C -> <control>
 189     '\x8d'     #  0x8D -> <control>
 190     '\x8e'     #  0x8E -> <control>
 191     '\x8f'     #  0x8F -> <control>
 192     '\x90'     #  0x90 -> <control>
 193     '\x91'     #  0x91 -> <control>
 194     '\x92'     #  0x92 -> <control>
 195     '\x93'     #  0x93 -> <control>
 196     '\x94'     #  0x94 -> <control>
 197     '\x95'     #  0x95 -> <control>
 198     '\x96'     #  0x96 -> <control>
 199     '\x97'     #  0x97 -> <control>
 200     '\x98'     #  0x98 -> <control>
 201     '\x99'     #  0x99 -> <control>
 202     '\x9a'     #  0x9A -> <control>
 203     '\x9b'     #  0x9B -> <control>
 204     '\x9c'     #  0x9C -> <control>
 205     '\x9d'     #  0x9D -> <control>
 206     '\x9e'     #  0x9E -> <control>
 207     '\x9f'     #  0x9F -> <control>
 208     '\xa0'     #  0xA0 -> NO-BREAK SPACE
 209     '\u06f0'   #  0xA1 -> EXTENDED ARABIC-INDIC DIGIT ZERO
 210     '\u06f1'   #  0xA2 -> EXTENDED ARABIC-INDIC DIGIT ONE
 211     '\u06f2'   #  0xA3 -> EXTENDED ARABIC-INDIC DIGIT TWO
 212     '\u06f3'   #  0xA4 -> EXTENDED ARABIC-INDIC DIGIT THREE
 213     '\u06f4'   #  0xA5 -> EXTENDED ARABIC-INDIC DIGIT FOUR
 214     '\u06f5'   #  0xA6 -> EXTENDED ARABIC-INDIC DIGIT FIVE
 215     '\u06f6'   #  0xA7 -> EXTENDED ARABIC-INDIC DIGIT SIX
 216     '\u06f7'   #  0xA8 -> EXTENDED ARABIC-INDIC DIGIT SEVEN
 217     '\u06f8'   #  0xA9 -> EXTENDED ARABIC-INDIC DIGIT EIGHT
 218     '\u06f9'   #  0xAA -> EXTENDED ARABIC-INDIC DIGIT NINE
 219     '\u060c'   #  0xAB -> ARABIC COMMA
 220     '\u061b'   #  0xAC -> ARABIC SEMICOLON
 221     '\xad'     #  0xAD -> SOFT HYPHEN
 222     '\u061f'   #  0xAE -> ARABIC QUESTION MARK
 223     '\ufe81'   #  0xAF -> ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM
 224     '\ufe8d'   #  0xB0 -> ARABIC LETTER ALEF ISOLATED FORM
 225     '\ufe8e'   #  0xB1 -> ARABIC LETTER ALEF FINAL FORM
 226     '\ufe8e'   #  0xB2 -> ARABIC LETTER ALEF FINAL FORM
 227     '\ufe8f'   #  0xB3 -> ARABIC LETTER BEH ISOLATED FORM
 228     '\ufe91'   #  0xB4 -> ARABIC LETTER BEH INITIAL FORM
 229     '\ufb56'   #  0xB5 -> ARABIC LETTER PEH ISOLATED FORM
 230     '\ufb58'   #  0xB6 -> ARABIC LETTER PEH INITIAL FORM
 231     '\ufe93'   #  0xB7 -> ARABIC LETTER TEH MARBUTA ISOLATED FORM
 232     '\ufe95'   #  0xB8 -> ARABIC LETTER TEH ISOLATED FORM
 233     '\ufe97'   #  0xB9 -> ARABIC LETTER TEH INITIAL FORM
 234     '\ufb66'   #  0xBA -> ARABIC LETTER TTEH ISOLATED FORM
 235     '\ufb68'   #  0xBB -> ARABIC LETTER TTEH INITIAL FORM
 236     '\ufe99'   #  0xBC -> ARABIC LETTER THEH ISOLATED FORM
 237     '\ufe9b'   #  0xBD -> ARABIC LETTER THEH INITIAL FORM
 238     '\ufe9d'   #  0xBE -> ARABIC LETTER JEEM ISOLATED FORM
 239     '\ufe9f'   #  0xBF -> ARABIC LETTER JEEM INITIAL FORM
 240     '\ufb7a'   #  0xC0 -> ARABIC LETTER TCHEH ISOLATED FORM
 241     '\ufb7c'   #  0xC1 -> ARABIC LETTER TCHEH INITIAL FORM
 242     '\ufea1'   #  0xC2 -> ARABIC LETTER HAH ISOLATED FORM
 243     '\ufea3'   #  0xC3 -> ARABIC LETTER HAH INITIAL FORM
 244     '\ufea5'   #  0xC4 -> ARABIC LETTER KHAH ISOLATED FORM
 245     '\ufea7'   #  0xC5 -> ARABIC LETTER KHAH INITIAL FORM
 246     '\ufea9'   #  0xC6 -> ARABIC LETTER DAL ISOLATED FORM
 247     '\ufb84'   #  0xC7 -> ARABIC LETTER DAHAL ISOLATED FORMN
 248     '\ufeab'   #  0xC8 -> ARABIC LETTER THAL ISOLATED FORM
 249     '\ufead'   #  0xC9 -> ARABIC LETTER REH ISOLATED FORM
 250     '\ufb8c'   #  0xCA -> ARABIC LETTER RREH ISOLATED FORM
 251     '\ufeaf'   #  0xCB -> ARABIC LETTER ZAIN ISOLATED FORM
 252     '\ufb8a'   #  0xCC -> ARABIC LETTER JEH ISOLATED FORM
 253     '\ufeb1'   #  0xCD -> ARABIC LETTER SEEN ISOLATED FORM
 254     '\ufeb3'   #  0xCE -> ARABIC LETTER SEEN INITIAL FORM
 255     '\ufeb5'   #  0xCF -> ARABIC LETTER SHEEN ISOLATED FORM
 256     '\ufeb7'   #  0xD0 -> ARABIC LETTER SHEEN INITIAL FORM
 257     '\ufeb9'   #  0xD1 -> ARABIC LETTER SAD ISOLATED FORM
 258     '\ufebb'   #  0xD2 -> ARABIC LETTER SAD INITIAL FORM
 259     '\ufebd'   #  0xD3 -> ARABIC LETTER DAD ISOLATED FORM
 260     '\ufebf'   #  0xD4 -> ARABIC LETTER DAD INITIAL FORM
 261     '\ufec1'   #  0xD5 -> ARABIC LETTER TAH ISOLATED FORM
 262     '\ufec5'   #  0xD6 -> ARABIC LETTER ZAH ISOLATED FORM
 263     '\ufec9'   #  0xD7 -> ARABIC LETTER AIN ISOLATED FORM
 264     '\ufeca'   #  0xD8 -> ARABIC LETTER AIN FINAL FORM
 265     '\ufecb'   #  0xD9 -> ARABIC LETTER AIN INITIAL FORM
 266     '\ufecc'   #  0xDA -> ARABIC LETTER AIN MEDIAL FORM
 267     '\ufecd'   #  0xDB -> ARABIC LETTER GHAIN ISOLATED FORM
 268     '\ufece'   #  0xDC -> ARABIC LETTER GHAIN FINAL FORM
 269     '\ufecf'   #  0xDD -> ARABIC LETTER GHAIN INITIAL FORM
 270     '\ufed0'   #  0xDE -> ARABIC LETTER GHAIN MEDIAL FORM
 271     '\ufed1'   #  0xDF -> ARABIC LETTER FEH ISOLATED FORM
 272     '\ufed3'   #  0xE0 -> ARABIC LETTER FEH INITIAL FORM
 273     '\ufed5'   #  0xE1 -> ARABIC LETTER QAF ISOLATED FORM
 274     '\ufed7'   #  0xE2 -> ARABIC LETTER QAF INITIAL FORM
 275     '\ufed9'   #  0xE3 -> ARABIC LETTER KAF ISOLATED FORM
 276     '\ufedb'   #  0xE4 -> ARABIC LETTER KAF INITIAL FORM
 277     '\ufb92'   #  0xE5 -> ARABIC LETTER GAF ISOLATED FORM
 278     '\ufb94'   #  0xE6 -> ARABIC LETTER GAF INITIAL FORM
 279     '\ufedd'   #  0xE7 -> ARABIC LETTER LAM ISOLATED FORM
 280     '\ufedf'   #  0xE8 -> ARABIC LETTER LAM INITIAL FORM
 281     '\ufee0'   #  0xE9 -> ARABIC LETTER LAM MEDIAL FORM
 282     '\ufee1'   #  0xEA -> ARABIC LETTER MEEM ISOLATED FORM
 283     '\ufee3'   #  0xEB -> ARABIC LETTER MEEM INITIAL FORM
 284     '\ufb9e'   #  0xEC -> ARABIC LETTER NOON GHUNNA ISOLATED FORM
 285     '\ufee5'   #  0xED -> ARABIC LETTER NOON ISOLATED FORM
 286     '\ufee7'   #  0xEE -> ARABIC LETTER NOON INITIAL FORM
 287     '\ufe85'   #  0xEF -> ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM
 288     '\ufeed'   #  0xF0 -> ARABIC LETTER WAW ISOLATED FORM
 289     '\ufba6'   #  0xF1 -> ARABIC LETTER HEH GOAL ISOLATED FORM
 290     '\ufba8'   #  0xF2 -> ARABIC LETTER HEH GOAL INITIAL FORM
 291     '\ufba9'   #  0xF3 -> ARABIC LETTER HEH GOAL MEDIAL FORM
 292     '\ufbaa'   #  0xF4 -> ARABIC LETTER HEH DOACHASHMEE ISOLATED FORM
 293     '\ufe80'   #  0xF5 -> ARABIC LETTER HAMZA ISOLATED FORM
 294     '\ufe89'   #  0xF6 -> ARABIC LETTER YEH WITH HAMZA ABOVE ISOLATED FORM
 295     '\ufe8a'   #  0xF7 -> ARABIC LETTER YEH WITH HAMZA ABOVE FINAL FORM
 296     '\ufe8b'   #  0xF8 -> ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM
 297     '\ufef1'   #  0xF9 -> ARABIC LETTER YEH ISOLATED FORM
 298     '\ufef2'   #  0xFA -> ARABIC LETTER YEH FINAL FORM
 299     '\ufef3'   #  0xFB -> ARABIC LETTER YEH INITIAL FORM
 300     '\ufbb0'   #  0xFC -> ARABIC LETTER YEH BARREE WITH HAMZA ABOVE ISOLATED FORM
 301     '\ufbae'   #  0xFD -> ARABIC LETTER YEH BARREE ISOLATED FORM
 302     '\ufe7c'   #  0xFE -> ARABIC SHADDA ISOLATED FORM
 303     '\ufe7d'   #  0xFF -> ARABIC SHADDA MEDIAL FORM
 304 )
 305
 306 ### Encoding table
 307 encoding_table=codecs.charmap_build(decoding_table)