|
1 """ Python Character Mapping Codec koi8_r generated from 'MAPPINGS/VENDORS/MISC/KOI8-R.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='koi8-r', |
|
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'\u2500' # 0x80 -> BOX DRAWINGS LIGHT HORIZONTAL |
|
177 u'\u2502' # 0x81 -> BOX DRAWINGS LIGHT VERTICAL |
|
178 u'\u250c' # 0x82 -> BOX DRAWINGS LIGHT DOWN AND RIGHT |
|
179 u'\u2510' # 0x83 -> BOX DRAWINGS LIGHT DOWN AND LEFT |
|
180 u'\u2514' # 0x84 -> BOX DRAWINGS LIGHT UP AND RIGHT |
|
181 u'\u2518' # 0x85 -> BOX DRAWINGS LIGHT UP AND LEFT |
|
182 u'\u251c' # 0x86 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT |
|
183 u'\u2524' # 0x87 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT |
|
184 u'\u252c' # 0x88 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL |
|
185 u'\u2534' # 0x89 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL |
|
186 u'\u253c' # 0x8A -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL |
|
187 u'\u2580' # 0x8B -> UPPER HALF BLOCK |
|
188 u'\u2584' # 0x8C -> LOWER HALF BLOCK |
|
189 u'\u2588' # 0x8D -> FULL BLOCK |
|
190 u'\u258c' # 0x8E -> LEFT HALF BLOCK |
|
191 u'\u2590' # 0x8F -> RIGHT HALF BLOCK |
|
192 u'\u2591' # 0x90 -> LIGHT SHADE |
|
193 u'\u2592' # 0x91 -> MEDIUM SHADE |
|
194 u'\u2593' # 0x92 -> DARK SHADE |
|
195 u'\u2320' # 0x93 -> TOP HALF INTEGRAL |
|
196 u'\u25a0' # 0x94 -> BLACK SQUARE |
|
197 u'\u2219' # 0x95 -> BULLET OPERATOR |
|
198 u'\u221a' # 0x96 -> SQUARE ROOT |
|
199 u'\u2248' # 0x97 -> ALMOST EQUAL TO |
|
200 u'\u2264' # 0x98 -> LESS-THAN OR EQUAL TO |
|
201 u'\u2265' # 0x99 -> GREATER-THAN OR EQUAL TO |
|
202 u'\xa0' # 0x9A -> NO-BREAK SPACE |
|
203 u'\u2321' # 0x9B -> BOTTOM HALF INTEGRAL |
|
204 u'\xb0' # 0x9C -> DEGREE SIGN |
|
205 u'\xb2' # 0x9D -> SUPERSCRIPT TWO |
|
206 u'\xb7' # 0x9E -> MIDDLE DOT |
|
207 u'\xf7' # 0x9F -> DIVISION SIGN |
|
208 u'\u2550' # 0xA0 -> BOX DRAWINGS DOUBLE HORIZONTAL |
|
209 u'\u2551' # 0xA1 -> BOX DRAWINGS DOUBLE VERTICAL |
|
210 u'\u2552' # 0xA2 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE |
|
211 u'\u0451' # 0xA3 -> CYRILLIC SMALL LETTER IO |
|
212 u'\u2553' # 0xA4 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE |
|
213 u'\u2554' # 0xA5 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT |
|
214 u'\u2555' # 0xA6 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE |
|
215 u'\u2556' # 0xA7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE |
|
216 u'\u2557' # 0xA8 -> BOX DRAWINGS DOUBLE DOWN AND LEFT |
|
217 u'\u2558' # 0xA9 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE |
|
218 u'\u2559' # 0xAA -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE |
|
219 u'\u255a' # 0xAB -> BOX DRAWINGS DOUBLE UP AND RIGHT |
|
220 u'\u255b' # 0xAC -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE |
|
221 u'\u255c' # 0xAD -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE |
|
222 u'\u255d' # 0xAE -> BOX DRAWINGS DOUBLE UP AND LEFT |
|
223 u'\u255e' # 0xAF -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE |
|
224 u'\u255f' # 0xB0 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE |
|
225 u'\u2560' # 0xB1 -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT |
|
226 u'\u2561' # 0xB2 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE |
|
227 u'\u0401' # 0xB3 -> CYRILLIC CAPITAL LETTER IO |
|
228 u'\u2562' # 0xB4 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE |
|
229 u'\u2563' # 0xB5 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT |
|
230 u'\u2564' # 0xB6 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE |
|
231 u'\u2565' # 0xB7 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE |
|
232 u'\u2566' # 0xB8 -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL |
|
233 u'\u2567' # 0xB9 -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE |
|
234 u'\u2568' # 0xBA -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE |
|
235 u'\u2569' # 0xBB -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL |
|
236 u'\u256a' # 0xBC -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE |
|
237 u'\u256b' # 0xBD -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE |
|
238 u'\u256c' # 0xBE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL |
|
239 u'\xa9' # 0xBF -> COPYRIGHT SIGN |
|
240 u'\u044e' # 0xC0 -> CYRILLIC SMALL LETTER YU |
|
241 u'\u0430' # 0xC1 -> CYRILLIC SMALL LETTER A |
|
242 u'\u0431' # 0xC2 -> CYRILLIC SMALL LETTER BE |
|
243 u'\u0446' # 0xC3 -> CYRILLIC SMALL LETTER TSE |
|
244 u'\u0434' # 0xC4 -> CYRILLIC SMALL LETTER DE |
|
245 u'\u0435' # 0xC5 -> CYRILLIC SMALL LETTER IE |
|
246 u'\u0444' # 0xC6 -> CYRILLIC SMALL LETTER EF |
|
247 u'\u0433' # 0xC7 -> CYRILLIC SMALL LETTER GHE |
|
248 u'\u0445' # 0xC8 -> CYRILLIC SMALL LETTER HA |
|
249 u'\u0438' # 0xC9 -> CYRILLIC SMALL LETTER I |
|
250 u'\u0439' # 0xCA -> CYRILLIC SMALL LETTER SHORT I |
|
251 u'\u043a' # 0xCB -> CYRILLIC SMALL LETTER KA |
|
252 u'\u043b' # 0xCC -> CYRILLIC SMALL LETTER EL |
|
253 u'\u043c' # 0xCD -> CYRILLIC SMALL LETTER EM |
|
254 u'\u043d' # 0xCE -> CYRILLIC SMALL LETTER EN |
|
255 u'\u043e' # 0xCF -> CYRILLIC SMALL LETTER O |
|
256 u'\u043f' # 0xD0 -> CYRILLIC SMALL LETTER PE |
|
257 u'\u044f' # 0xD1 -> CYRILLIC SMALL LETTER YA |
|
258 u'\u0440' # 0xD2 -> CYRILLIC SMALL LETTER ER |
|
259 u'\u0441' # 0xD3 -> CYRILLIC SMALL LETTER ES |
|
260 u'\u0442' # 0xD4 -> CYRILLIC SMALL LETTER TE |
|
261 u'\u0443' # 0xD5 -> CYRILLIC SMALL LETTER U |
|
262 u'\u0436' # 0xD6 -> CYRILLIC SMALL LETTER ZHE |
|
263 u'\u0432' # 0xD7 -> CYRILLIC SMALL LETTER VE |
|
264 u'\u044c' # 0xD8 -> CYRILLIC SMALL LETTER SOFT SIGN |
|
265 u'\u044b' # 0xD9 -> CYRILLIC SMALL LETTER YERU |
|
266 u'\u0437' # 0xDA -> CYRILLIC SMALL LETTER ZE |
|
267 u'\u0448' # 0xDB -> CYRILLIC SMALL LETTER SHA |
|
268 u'\u044d' # 0xDC -> CYRILLIC SMALL LETTER E |
|
269 u'\u0449' # 0xDD -> CYRILLIC SMALL LETTER SHCHA |
|
270 u'\u0447' # 0xDE -> CYRILLIC SMALL LETTER CHE |
|
271 u'\u044a' # 0xDF -> CYRILLIC SMALL LETTER HARD SIGN |
|
272 u'\u042e' # 0xE0 -> CYRILLIC CAPITAL LETTER YU |
|
273 u'\u0410' # 0xE1 -> CYRILLIC CAPITAL LETTER A |
|
274 u'\u0411' # 0xE2 -> CYRILLIC CAPITAL LETTER BE |
|
275 u'\u0426' # 0xE3 -> CYRILLIC CAPITAL LETTER TSE |
|
276 u'\u0414' # 0xE4 -> CYRILLIC CAPITAL LETTER DE |
|
277 u'\u0415' # 0xE5 -> CYRILLIC CAPITAL LETTER IE |
|
278 u'\u0424' # 0xE6 -> CYRILLIC CAPITAL LETTER EF |
|
279 u'\u0413' # 0xE7 -> CYRILLIC CAPITAL LETTER GHE |
|
280 u'\u0425' # 0xE8 -> CYRILLIC CAPITAL LETTER HA |
|
281 u'\u0418' # 0xE9 -> CYRILLIC CAPITAL LETTER I |
|
282 u'\u0419' # 0xEA -> CYRILLIC CAPITAL LETTER SHORT I |
|
283 u'\u041a' # 0xEB -> CYRILLIC CAPITAL LETTER KA |
|
284 u'\u041b' # 0xEC -> CYRILLIC CAPITAL LETTER EL |
|
285 u'\u041c' # 0xED -> CYRILLIC CAPITAL LETTER EM |
|
286 u'\u041d' # 0xEE -> CYRILLIC CAPITAL LETTER EN |
|
287 u'\u041e' # 0xEF -> CYRILLIC CAPITAL LETTER O |
|
288 u'\u041f' # 0xF0 -> CYRILLIC CAPITAL LETTER PE |
|
289 u'\u042f' # 0xF1 -> CYRILLIC CAPITAL LETTER YA |
|
290 u'\u0420' # 0xF2 -> CYRILLIC CAPITAL LETTER ER |
|
291 u'\u0421' # 0xF3 -> CYRILLIC CAPITAL LETTER ES |
|
292 u'\u0422' # 0xF4 -> CYRILLIC CAPITAL LETTER TE |
|
293 u'\u0423' # 0xF5 -> CYRILLIC CAPITAL LETTER U |
|
294 u'\u0416' # 0xF6 -> CYRILLIC CAPITAL LETTER ZHE |
|
295 u'\u0412' # 0xF7 -> CYRILLIC CAPITAL LETTER VE |
|
296 u'\u042c' # 0xF8 -> CYRILLIC CAPITAL LETTER SOFT SIGN |
|
297 u'\u042b' # 0xF9 -> CYRILLIC CAPITAL LETTER YERU |
|
298 u'\u0417' # 0xFA -> CYRILLIC CAPITAL LETTER ZE |
|
299 u'\u0428' # 0xFB -> CYRILLIC CAPITAL LETTER SHA |
|
300 u'\u042d' # 0xFC -> CYRILLIC CAPITAL LETTER E |
|
301 u'\u0429' # 0xFD -> CYRILLIC CAPITAL LETTER SHCHA |
|
302 u'\u0427' # 0xFE -> CYRILLIC CAPITAL LETTER CHE |
|
303 u'\u042a' # 0xFF -> CYRILLIC CAPITAL LETTER HARD SIGN |
|
304 ) |
|
305 |
|
306 ### Encoding table |
|
307 encoding_table=codecs.charmap_build(decoding_table) |