PageRenderTime 47ms CodeModel.GetById 20ms app.highlight 24ms RepoModel.GetById 1ms app.codeStats 0ms

/Lib/encodings/cp1140.py

http://unladen-swallow.googlecode.com/
Python | 307 lines | 307 code | 0 blank | 0 comment | 0 complexity | e523276b1ad5bd1b1d41eedc71f6de7a MD5 | raw file
  1""" Python Character Mapping Codec cp1140 generated from 'python-mappings/CP1140.TXT' with gencodec.py.
  2
  3"""#"
  4
  5import codecs
  6
  7### Codec APIs
  8
  9class 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
 17class IncrementalEncoder(codecs.IncrementalEncoder):
 18    def encode(self, input, final=False):
 19        return codecs.charmap_encode(input,self.errors,encoding_table)[0]
 20
 21class IncrementalDecoder(codecs.IncrementalDecoder):
 22    def decode(self, input, final=False):
 23        return codecs.charmap_decode(input,self.errors,decoding_table)[0]
 24
 25class StreamWriter(Codec,codecs.StreamWriter):
 26    pass
 27
 28class StreamReader(Codec,codecs.StreamReader):
 29    pass
 30
 31### encodings module API
 32
 33def getregentry():
 34    return codecs.CodecInfo(
 35        name='cp1140',
 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
 47decoding_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'\x9c'     #  0x04 -> CONTROL
 53    u'\t'       #  0x05 -> HORIZONTAL TABULATION
 54    u'\x86'     #  0x06 -> CONTROL
 55    u'\x7f'     #  0x07 -> DELETE
 56    u'\x97'     #  0x08 -> CONTROL
 57    u'\x8d'     #  0x09 -> CONTROL
 58    u'\x8e'     #  0x0A -> CONTROL
 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'\x9d'     #  0x14 -> CONTROL
 69    u'\x85'     #  0x15 -> CONTROL
 70    u'\x08'     #  0x16 -> BACKSPACE
 71    u'\x87'     #  0x17 -> CONTROL
 72    u'\x18'     #  0x18 -> CANCEL
 73    u'\x19'     #  0x19 -> END OF MEDIUM
 74    u'\x92'     #  0x1A -> CONTROL
 75    u'\x8f'     #  0x1B -> CONTROL
 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'\x80'     #  0x20 -> CONTROL
 81    u'\x81'     #  0x21 -> CONTROL
 82    u'\x82'     #  0x22 -> CONTROL
 83    u'\x83'     #  0x23 -> CONTROL
 84    u'\x84'     #  0x24 -> CONTROL
 85    u'\n'       #  0x25 -> LINE FEED
 86    u'\x17'     #  0x26 -> END OF TRANSMISSION BLOCK
 87    u'\x1b'     #  0x27 -> ESCAPE
 88    u'\x88'     #  0x28 -> CONTROL
 89    u'\x89'     #  0x29 -> CONTROL
 90    u'\x8a'     #  0x2A -> CONTROL
 91    u'\x8b'     #  0x2B -> CONTROL
 92    u'\x8c'     #  0x2C -> CONTROL
 93    u'\x05'     #  0x2D -> ENQUIRY
 94    u'\x06'     #  0x2E -> ACKNOWLEDGE
 95    u'\x07'     #  0x2F -> BELL
 96    u'\x90'     #  0x30 -> CONTROL
 97    u'\x91'     #  0x31 -> CONTROL
 98    u'\x16'     #  0x32 -> SYNCHRONOUS IDLE
 99    u'\x93'     #  0x33 -> CONTROL
100    u'\x94'     #  0x34 -> CONTROL
101    u'\x95'     #  0x35 -> CONTROL
102    u'\x96'     #  0x36 -> CONTROL
103    u'\x04'     #  0x37 -> END OF TRANSMISSION
104    u'\x98'     #  0x38 -> CONTROL
105    u'\x99'     #  0x39 -> CONTROL
106    u'\x9a'     #  0x3A -> CONTROL
107    u'\x9b'     #  0x3B -> CONTROL
108    u'\x14'     #  0x3C -> DEVICE CONTROL FOUR
109    u'\x15'     #  0x3D -> NEGATIVE ACKNOWLEDGE
110    u'\x9e'     #  0x3E -> CONTROL
111    u'\x1a'     #  0x3F -> SUBSTITUTE
112    u' '        #  0x40 -> SPACE
113    u'\xa0'     #  0x41 -> NO-BREAK SPACE
114    u'\xe2'     #  0x42 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
115    u'\xe4'     #  0x43 -> LATIN SMALL LETTER A WITH DIAERESIS
116    u'\xe0'     #  0x44 -> LATIN SMALL LETTER A WITH GRAVE
117    u'\xe1'     #  0x45 -> LATIN SMALL LETTER A WITH ACUTE
118    u'\xe3'     #  0x46 -> LATIN SMALL LETTER A WITH TILDE
119    u'\xe5'     #  0x47 -> LATIN SMALL LETTER A WITH RING ABOVE
120    u'\xe7'     #  0x48 -> LATIN SMALL LETTER C WITH CEDILLA
121    u'\xf1'     #  0x49 -> LATIN SMALL LETTER N WITH TILDE
122    u'\xa2'     #  0x4A -> CENT SIGN
123    u'.'        #  0x4B -> FULL STOP
124    u'<'        #  0x4C -> LESS-THAN SIGN
125    u'('        #  0x4D -> LEFT PARENTHESIS
126    u'+'        #  0x4E -> PLUS SIGN
127    u'|'        #  0x4F -> VERTICAL LINE
128    u'&'        #  0x50 -> AMPERSAND
129    u'\xe9'     #  0x51 -> LATIN SMALL LETTER E WITH ACUTE
130    u'\xea'     #  0x52 -> LATIN SMALL LETTER E WITH CIRCUMFLEX
131    u'\xeb'     #  0x53 -> LATIN SMALL LETTER E WITH DIAERESIS
132    u'\xe8'     #  0x54 -> LATIN SMALL LETTER E WITH GRAVE
133    u'\xed'     #  0x55 -> LATIN SMALL LETTER I WITH ACUTE
134    u'\xee'     #  0x56 -> LATIN SMALL LETTER I WITH CIRCUMFLEX
135    u'\xef'     #  0x57 -> LATIN SMALL LETTER I WITH DIAERESIS
136    u'\xec'     #  0x58 -> LATIN SMALL LETTER I WITH GRAVE
137    u'\xdf'     #  0x59 -> LATIN SMALL LETTER SHARP S (GERMAN)
138    u'!'        #  0x5A -> EXCLAMATION MARK
139    u'$'        #  0x5B -> DOLLAR SIGN
140    u'*'        #  0x5C -> ASTERISK
141    u')'        #  0x5D -> RIGHT PARENTHESIS
142    u';'        #  0x5E -> SEMICOLON
143    u'\xac'     #  0x5F -> NOT SIGN
144    u'-'        #  0x60 -> HYPHEN-MINUS
145    u'/'        #  0x61 -> SOLIDUS
146    u'\xc2'     #  0x62 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX
147    u'\xc4'     #  0x63 -> LATIN CAPITAL LETTER A WITH DIAERESIS
148    u'\xc0'     #  0x64 -> LATIN CAPITAL LETTER A WITH GRAVE
149    u'\xc1'     #  0x65 -> LATIN CAPITAL LETTER A WITH ACUTE
150    u'\xc3'     #  0x66 -> LATIN CAPITAL LETTER A WITH TILDE
151    u'\xc5'     #  0x67 -> LATIN CAPITAL LETTER A WITH RING ABOVE
152    u'\xc7'     #  0x68 -> LATIN CAPITAL LETTER C WITH CEDILLA
153    u'\xd1'     #  0x69 -> LATIN CAPITAL LETTER N WITH TILDE
154    u'\xa6'     #  0x6A -> BROKEN BAR
155    u','        #  0x6B -> COMMA
156    u'%'        #  0x6C -> PERCENT SIGN
157    u'_'        #  0x6D -> LOW LINE
158    u'>'        #  0x6E -> GREATER-THAN SIGN
159    u'?'        #  0x6F -> QUESTION MARK
160    u'\xf8'     #  0x70 -> LATIN SMALL LETTER O WITH STROKE
161    u'\xc9'     #  0x71 -> LATIN CAPITAL LETTER E WITH ACUTE
162    u'\xca'     #  0x72 -> LATIN CAPITAL LETTER E WITH CIRCUMFLEX
163    u'\xcb'     #  0x73 -> LATIN CAPITAL LETTER E WITH DIAERESIS
164    u'\xc8'     #  0x74 -> LATIN CAPITAL LETTER E WITH GRAVE
165    u'\xcd'     #  0x75 -> LATIN CAPITAL LETTER I WITH ACUTE
166    u'\xce'     #  0x76 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX
167    u'\xcf'     #  0x77 -> LATIN CAPITAL LETTER I WITH DIAERESIS
168    u'\xcc'     #  0x78 -> LATIN CAPITAL LETTER I WITH GRAVE
169    u'`'        #  0x79 -> GRAVE ACCENT
170    u':'        #  0x7A -> COLON
171    u'#'        #  0x7B -> NUMBER SIGN
172    u'@'        #  0x7C -> COMMERCIAL AT
173    u"'"        #  0x7D -> APOSTROPHE
174    u'='        #  0x7E -> EQUALS SIGN
175    u'"'        #  0x7F -> QUOTATION MARK
176    u'\xd8'     #  0x80 -> LATIN CAPITAL LETTER O WITH STROKE
177    u'a'        #  0x81 -> LATIN SMALL LETTER A
178    u'b'        #  0x82 -> LATIN SMALL LETTER B
179    u'c'        #  0x83 -> LATIN SMALL LETTER C
180    u'd'        #  0x84 -> LATIN SMALL LETTER D
181    u'e'        #  0x85 -> LATIN SMALL LETTER E
182    u'f'        #  0x86 -> LATIN SMALL LETTER F
183    u'g'        #  0x87 -> LATIN SMALL LETTER G
184    u'h'        #  0x88 -> LATIN SMALL LETTER H
185    u'i'        #  0x89 -> LATIN SMALL LETTER I
186    u'\xab'     #  0x8A -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
187    u'\xbb'     #  0x8B -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
188    u'\xf0'     #  0x8C -> LATIN SMALL LETTER ETH (ICELANDIC)
189    u'\xfd'     #  0x8D -> LATIN SMALL LETTER Y WITH ACUTE
190    u'\xfe'     #  0x8E -> LATIN SMALL LETTER THORN (ICELANDIC)
191    u'\xb1'     #  0x8F -> PLUS-MINUS SIGN
192    u'\xb0'     #  0x90 -> DEGREE SIGN
193    u'j'        #  0x91 -> LATIN SMALL LETTER J
194    u'k'        #  0x92 -> LATIN SMALL LETTER K
195    u'l'        #  0x93 -> LATIN SMALL LETTER L
196    u'm'        #  0x94 -> LATIN SMALL LETTER M
197    u'n'        #  0x95 -> LATIN SMALL LETTER N
198    u'o'        #  0x96 -> LATIN SMALL LETTER O
199    u'p'        #  0x97 -> LATIN SMALL LETTER P
200    u'q'        #  0x98 -> LATIN SMALL LETTER Q
201    u'r'        #  0x99 -> LATIN SMALL LETTER R
202    u'\xaa'     #  0x9A -> FEMININE ORDINAL INDICATOR
203    u'\xba'     #  0x9B -> MASCULINE ORDINAL INDICATOR
204    u'\xe6'     #  0x9C -> LATIN SMALL LIGATURE AE
205    u'\xb8'     #  0x9D -> CEDILLA
206    u'\xc6'     #  0x9E -> LATIN CAPITAL LIGATURE AE
207    u'\u20ac'   #  0x9F -> EURO SIGN
208    u'\xb5'     #  0xA0 -> MICRO SIGN
209    u'~'        #  0xA1 -> TILDE
210    u's'        #  0xA2 -> LATIN SMALL LETTER S
211    u't'        #  0xA3 -> LATIN SMALL LETTER T
212    u'u'        #  0xA4 -> LATIN SMALL LETTER U
213    u'v'        #  0xA5 -> LATIN SMALL LETTER V
214    u'w'        #  0xA6 -> LATIN SMALL LETTER W
215    u'x'        #  0xA7 -> LATIN SMALL LETTER X
216    u'y'        #  0xA8 -> LATIN SMALL LETTER Y
217    u'z'        #  0xA9 -> LATIN SMALL LETTER Z
218    u'\xa1'     #  0xAA -> INVERTED EXCLAMATION MARK
219    u'\xbf'     #  0xAB -> INVERTED QUESTION MARK
220    u'\xd0'     #  0xAC -> LATIN CAPITAL LETTER ETH (ICELANDIC)
221    u'\xdd'     #  0xAD -> LATIN CAPITAL LETTER Y WITH ACUTE
222    u'\xde'     #  0xAE -> LATIN CAPITAL LETTER THORN (ICELANDIC)
223    u'\xae'     #  0xAF -> REGISTERED SIGN
224    u'^'        #  0xB0 -> CIRCUMFLEX ACCENT
225    u'\xa3'     #  0xB1 -> POUND SIGN
226    u'\xa5'     #  0xB2 -> YEN SIGN
227    u'\xb7'     #  0xB3 -> MIDDLE DOT
228    u'\xa9'     #  0xB4 -> COPYRIGHT SIGN
229    u'\xa7'     #  0xB5 -> SECTION SIGN
230    u'\xb6'     #  0xB6 -> PILCROW SIGN
231    u'\xbc'     #  0xB7 -> VULGAR FRACTION ONE QUARTER
232    u'\xbd'     #  0xB8 -> VULGAR FRACTION ONE HALF
233    u'\xbe'     #  0xB9 -> VULGAR FRACTION THREE QUARTERS
234    u'['        #  0xBA -> LEFT SQUARE BRACKET
235    u']'        #  0xBB -> RIGHT SQUARE BRACKET
236    u'\xaf'     #  0xBC -> MACRON
237    u'\xa8'     #  0xBD -> DIAERESIS
238    u'\xb4'     #  0xBE -> ACUTE ACCENT
239    u'\xd7'     #  0xBF -> MULTIPLICATION SIGN
240    u'{'        #  0xC0 -> LEFT CURLY BRACKET
241    u'A'        #  0xC1 -> LATIN CAPITAL LETTER A
242    u'B'        #  0xC2 -> LATIN CAPITAL LETTER B
243    u'C'        #  0xC3 -> LATIN CAPITAL LETTER C
244    u'D'        #  0xC4 -> LATIN CAPITAL LETTER D
245    u'E'        #  0xC5 -> LATIN CAPITAL LETTER E
246    u'F'        #  0xC6 -> LATIN CAPITAL LETTER F
247    u'G'        #  0xC7 -> LATIN CAPITAL LETTER G
248    u'H'        #  0xC8 -> LATIN CAPITAL LETTER H
249    u'I'        #  0xC9 -> LATIN CAPITAL LETTER I
250    u'\xad'     #  0xCA -> SOFT HYPHEN
251    u'\xf4'     #  0xCB -> LATIN SMALL LETTER O WITH CIRCUMFLEX
252    u'\xf6'     #  0xCC -> LATIN SMALL LETTER O WITH DIAERESIS
253    u'\xf2'     #  0xCD -> LATIN SMALL LETTER O WITH GRAVE
254    u'\xf3'     #  0xCE -> LATIN SMALL LETTER O WITH ACUTE
255    u'\xf5'     #  0xCF -> LATIN SMALL LETTER O WITH TILDE
256    u'}'        #  0xD0 -> RIGHT CURLY BRACKET
257    u'J'        #  0xD1 -> LATIN CAPITAL LETTER J
258    u'K'        #  0xD2 -> LATIN CAPITAL LETTER K
259    u'L'        #  0xD3 -> LATIN CAPITAL LETTER L
260    u'M'        #  0xD4 -> LATIN CAPITAL LETTER M
261    u'N'        #  0xD5 -> LATIN CAPITAL LETTER N
262    u'O'        #  0xD6 -> LATIN CAPITAL LETTER O
263    u'P'        #  0xD7 -> LATIN CAPITAL LETTER P
264    u'Q'        #  0xD8 -> LATIN CAPITAL LETTER Q
265    u'R'        #  0xD9 -> LATIN CAPITAL LETTER R
266    u'\xb9'     #  0xDA -> SUPERSCRIPT ONE
267    u'\xfb'     #  0xDB -> LATIN SMALL LETTER U WITH CIRCUMFLEX
268    u'\xfc'     #  0xDC -> LATIN SMALL LETTER U WITH DIAERESIS
269    u'\xf9'     #  0xDD -> LATIN SMALL LETTER U WITH GRAVE
270    u'\xfa'     #  0xDE -> LATIN SMALL LETTER U WITH ACUTE
271    u'\xff'     #  0xDF -> LATIN SMALL LETTER Y WITH DIAERESIS
272    u'\\'       #  0xE0 -> REVERSE SOLIDUS
273    u'\xf7'     #  0xE1 -> DIVISION SIGN
274    u'S'        #  0xE2 -> LATIN CAPITAL LETTER S
275    u'T'        #  0xE3 -> LATIN CAPITAL LETTER T
276    u'U'        #  0xE4 -> LATIN CAPITAL LETTER U
277    u'V'        #  0xE5 -> LATIN CAPITAL LETTER V
278    u'W'        #  0xE6 -> LATIN CAPITAL LETTER W
279    u'X'        #  0xE7 -> LATIN CAPITAL LETTER X
280    u'Y'        #  0xE8 -> LATIN CAPITAL LETTER Y
281    u'Z'        #  0xE9 -> LATIN CAPITAL LETTER Z
282    u'\xb2'     #  0xEA -> SUPERSCRIPT TWO
283    u'\xd4'     #  0xEB -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX
284    u'\xd6'     #  0xEC -> LATIN CAPITAL LETTER O WITH DIAERESIS
285    u'\xd2'     #  0xED -> LATIN CAPITAL LETTER O WITH GRAVE
286    u'\xd3'     #  0xEE -> LATIN CAPITAL LETTER O WITH ACUTE
287    u'\xd5'     #  0xEF -> LATIN CAPITAL LETTER O WITH TILDE
288    u'0'        #  0xF0 -> DIGIT ZERO
289    u'1'        #  0xF1 -> DIGIT ONE
290    u'2'        #  0xF2 -> DIGIT TWO
291    u'3'        #  0xF3 -> DIGIT THREE
292    u'4'        #  0xF4 -> DIGIT FOUR
293    u'5'        #  0xF5 -> DIGIT FIVE
294    u'6'        #  0xF6 -> DIGIT SIX
295    u'7'        #  0xF7 -> DIGIT SEVEN
296    u'8'        #  0xF8 -> DIGIT EIGHT
297    u'9'        #  0xF9 -> DIGIT NINE
298    u'\xb3'     #  0xFA -> SUPERSCRIPT THREE
299    u'\xdb'     #  0xFB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX
300    u'\xdc'     #  0xFC -> LATIN CAPITAL LETTER U WITH DIAERESIS
301    u'\xd9'     #  0xFD -> LATIN CAPITAL LETTER U WITH GRAVE
302    u'\xda'     #  0xFE -> LATIN CAPITAL LETTER U WITH ACUTE
303    u'\x9f'     #  0xFF -> CONTROL
304)
305
306### Encoding table
307encoding_table=codecs.charmap_build(decoding_table)