|
1 """ Python Character Mapping Codec cp875 generated from 'MAPPINGS/VENDORS/MICSFT/EBCDIC/CP875.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='cp875', |
|
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'\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'\u0391' # 0x41 -> GREEK CAPITAL LETTER ALPHA |
|
114 u'\u0392' # 0x42 -> GREEK CAPITAL LETTER BETA |
|
115 u'\u0393' # 0x43 -> GREEK CAPITAL LETTER GAMMA |
|
116 u'\u0394' # 0x44 -> GREEK CAPITAL LETTER DELTA |
|
117 u'\u0395' # 0x45 -> GREEK CAPITAL LETTER EPSILON |
|
118 u'\u0396' # 0x46 -> GREEK CAPITAL LETTER ZETA |
|
119 u'\u0397' # 0x47 -> GREEK CAPITAL LETTER ETA |
|
120 u'\u0398' # 0x48 -> GREEK CAPITAL LETTER THETA |
|
121 u'\u0399' # 0x49 -> GREEK CAPITAL LETTER IOTA |
|
122 u'[' # 0x4A -> LEFT SQUARE BRACKET |
|
123 u'.' # 0x4B -> FULL STOP |
|
124 u'<' # 0x4C -> LESS-THAN SIGN |
|
125 u'(' # 0x4D -> LEFT PARENTHESIS |
|
126 u'+' # 0x4E -> PLUS SIGN |
|
127 u'!' # 0x4F -> EXCLAMATION MARK |
|
128 u'&' # 0x50 -> AMPERSAND |
|
129 u'\u039a' # 0x51 -> GREEK CAPITAL LETTER KAPPA |
|
130 u'\u039b' # 0x52 -> GREEK CAPITAL LETTER LAMDA |
|
131 u'\u039c' # 0x53 -> GREEK CAPITAL LETTER MU |
|
132 u'\u039d' # 0x54 -> GREEK CAPITAL LETTER NU |
|
133 u'\u039e' # 0x55 -> GREEK CAPITAL LETTER XI |
|
134 u'\u039f' # 0x56 -> GREEK CAPITAL LETTER OMICRON |
|
135 u'\u03a0' # 0x57 -> GREEK CAPITAL LETTER PI |
|
136 u'\u03a1' # 0x58 -> GREEK CAPITAL LETTER RHO |
|
137 u'\u03a3' # 0x59 -> GREEK CAPITAL LETTER SIGMA |
|
138 u']' # 0x5A -> RIGHT SQUARE BRACKET |
|
139 u'$' # 0x5B -> DOLLAR SIGN |
|
140 u'*' # 0x5C -> ASTERISK |
|
141 u')' # 0x5D -> RIGHT PARENTHESIS |
|
142 u';' # 0x5E -> SEMICOLON |
|
143 u'^' # 0x5F -> CIRCUMFLEX ACCENT |
|
144 u'-' # 0x60 -> HYPHEN-MINUS |
|
145 u'/' # 0x61 -> SOLIDUS |
|
146 u'\u03a4' # 0x62 -> GREEK CAPITAL LETTER TAU |
|
147 u'\u03a5' # 0x63 -> GREEK CAPITAL LETTER UPSILON |
|
148 u'\u03a6' # 0x64 -> GREEK CAPITAL LETTER PHI |
|
149 u'\u03a7' # 0x65 -> GREEK CAPITAL LETTER CHI |
|
150 u'\u03a8' # 0x66 -> GREEK CAPITAL LETTER PSI |
|
151 u'\u03a9' # 0x67 -> GREEK CAPITAL LETTER OMEGA |
|
152 u'\u03aa' # 0x68 -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA |
|
153 u'\u03ab' # 0x69 -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA |
|
154 u'|' # 0x6A -> VERTICAL LINE |
|
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'\xa8' # 0x70 -> DIAERESIS |
|
161 u'\u0386' # 0x71 -> GREEK CAPITAL LETTER ALPHA WITH TONOS |
|
162 u'\u0388' # 0x72 -> GREEK CAPITAL LETTER EPSILON WITH TONOS |
|
163 u'\u0389' # 0x73 -> GREEK CAPITAL LETTER ETA WITH TONOS |
|
164 u'\xa0' # 0x74 -> NO-BREAK SPACE |
|
165 u'\u038a' # 0x75 -> GREEK CAPITAL LETTER IOTA WITH TONOS |
|
166 u'\u038c' # 0x76 -> GREEK CAPITAL LETTER OMICRON WITH TONOS |
|
167 u'\u038e' # 0x77 -> GREEK CAPITAL LETTER UPSILON WITH TONOS |
|
168 u'\u038f' # 0x78 -> GREEK CAPITAL LETTER OMEGA WITH TONOS |
|
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'\u0385' # 0x80 -> GREEK DIALYTIKA TONOS |
|
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'\u03b1' # 0x8A -> GREEK SMALL LETTER ALPHA |
|
187 u'\u03b2' # 0x8B -> GREEK SMALL LETTER BETA |
|
188 u'\u03b3' # 0x8C -> GREEK SMALL LETTER GAMMA |
|
189 u'\u03b4' # 0x8D -> GREEK SMALL LETTER DELTA |
|
190 u'\u03b5' # 0x8E -> GREEK SMALL LETTER EPSILON |
|
191 u'\u03b6' # 0x8F -> GREEK SMALL LETTER ZETA |
|
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'\u03b7' # 0x9A -> GREEK SMALL LETTER ETA |
|
203 u'\u03b8' # 0x9B -> GREEK SMALL LETTER THETA |
|
204 u'\u03b9' # 0x9C -> GREEK SMALL LETTER IOTA |
|
205 u'\u03ba' # 0x9D -> GREEK SMALL LETTER KAPPA |
|
206 u'\u03bb' # 0x9E -> GREEK SMALL LETTER LAMDA |
|
207 u'\u03bc' # 0x9F -> GREEK SMALL LETTER MU |
|
208 u'\xb4' # 0xA0 -> ACUTE ACCENT |
|
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'\u03bd' # 0xAA -> GREEK SMALL LETTER NU |
|
219 u'\u03be' # 0xAB -> GREEK SMALL LETTER XI |
|
220 u'\u03bf' # 0xAC -> GREEK SMALL LETTER OMICRON |
|
221 u'\u03c0' # 0xAD -> GREEK SMALL LETTER PI |
|
222 u'\u03c1' # 0xAE -> GREEK SMALL LETTER RHO |
|
223 u'\u03c3' # 0xAF -> GREEK SMALL LETTER SIGMA |
|
224 u'\xa3' # 0xB0 -> POUND SIGN |
|
225 u'\u03ac' # 0xB1 -> GREEK SMALL LETTER ALPHA WITH TONOS |
|
226 u'\u03ad' # 0xB2 -> GREEK SMALL LETTER EPSILON WITH TONOS |
|
227 u'\u03ae' # 0xB3 -> GREEK SMALL LETTER ETA WITH TONOS |
|
228 u'\u03ca' # 0xB4 -> GREEK SMALL LETTER IOTA WITH DIALYTIKA |
|
229 u'\u03af' # 0xB5 -> GREEK SMALL LETTER IOTA WITH TONOS |
|
230 u'\u03cc' # 0xB6 -> GREEK SMALL LETTER OMICRON WITH TONOS |
|
231 u'\u03cd' # 0xB7 -> GREEK SMALL LETTER UPSILON WITH TONOS |
|
232 u'\u03cb' # 0xB8 -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA |
|
233 u'\u03ce' # 0xB9 -> GREEK SMALL LETTER OMEGA WITH TONOS |
|
234 u'\u03c2' # 0xBA -> GREEK SMALL LETTER FINAL SIGMA |
|
235 u'\u03c4' # 0xBB -> GREEK SMALL LETTER TAU |
|
236 u'\u03c5' # 0xBC -> GREEK SMALL LETTER UPSILON |
|
237 u'\u03c6' # 0xBD -> GREEK SMALL LETTER PHI |
|
238 u'\u03c7' # 0xBE -> GREEK SMALL LETTER CHI |
|
239 u'\u03c8' # 0xBF -> GREEK SMALL LETTER PSI |
|
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'\u03c9' # 0xCB -> GREEK SMALL LETTER OMEGA |
|
252 u'\u0390' # 0xCC -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS |
|
253 u'\u03b0' # 0xCD -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS |
|
254 u'\u2018' # 0xCE -> LEFT SINGLE QUOTATION MARK |
|
255 u'\u2015' # 0xCF -> HORIZONTAL BAR |
|
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'\xb1' # 0xDA -> PLUS-MINUS SIGN |
|
267 u'\xbd' # 0xDB -> VULGAR FRACTION ONE HALF |
|
268 u'\x1a' # 0xDC -> SUBSTITUTE |
|
269 u'\u0387' # 0xDD -> GREEK ANO TELEIA |
|
270 u'\u2019' # 0xDE -> RIGHT SINGLE QUOTATION MARK |
|
271 u'\xa6' # 0xDF -> BROKEN BAR |
|
272 u'\\' # 0xE0 -> REVERSE SOLIDUS |
|
273 u'\x1a' # 0xE1 -> SUBSTITUTE |
|
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'\xa7' # 0xEB -> SECTION SIGN |
|
284 u'\x1a' # 0xEC -> SUBSTITUTE |
|
285 u'\x1a' # 0xED -> SUBSTITUTE |
|
286 u'\xab' # 0xEE -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK |
|
287 u'\xac' # 0xEF -> NOT SIGN |
|
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'\xa9' # 0xFB -> COPYRIGHT SIGN |
|
300 u'\x1a' # 0xFC -> SUBSTITUTE |
|
301 u'\x1a' # 0xFD -> SUBSTITUTE |
|
302 u'\xbb' # 0xFE -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK |
|
303 u'\x9f' # 0xFF -> CONTROL |
|
304 ) |
|
305 |
|
306 ### Encoding table |
|
307 encoding_table=codecs.charmap_build(decoding_table) |