|
1 """ Python Character Mapping Codec cp874 generated from 'MAPPINGS/VENDORS/MICSFT/WINDOWS/CP874.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='cp874', |
|
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'\u20ac' # 0x80 -> EURO SIGN |
|
177 u'\ufffe' # 0x81 -> UNDEFINED |
|
178 u'\ufffe' # 0x82 -> UNDEFINED |
|
179 u'\ufffe' # 0x83 -> UNDEFINED |
|
180 u'\ufffe' # 0x84 -> UNDEFINED |
|
181 u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS |
|
182 u'\ufffe' # 0x86 -> UNDEFINED |
|
183 u'\ufffe' # 0x87 -> UNDEFINED |
|
184 u'\ufffe' # 0x88 -> UNDEFINED |
|
185 u'\ufffe' # 0x89 -> UNDEFINED |
|
186 u'\ufffe' # 0x8A -> UNDEFINED |
|
187 u'\ufffe' # 0x8B -> UNDEFINED |
|
188 u'\ufffe' # 0x8C -> UNDEFINED |
|
189 u'\ufffe' # 0x8D -> UNDEFINED |
|
190 u'\ufffe' # 0x8E -> UNDEFINED |
|
191 u'\ufffe' # 0x8F -> UNDEFINED |
|
192 u'\ufffe' # 0x90 -> UNDEFINED |
|
193 u'\u2018' # 0x91 -> LEFT SINGLE QUOTATION MARK |
|
194 u'\u2019' # 0x92 -> RIGHT SINGLE QUOTATION MARK |
|
195 u'\u201c' # 0x93 -> LEFT DOUBLE QUOTATION MARK |
|
196 u'\u201d' # 0x94 -> RIGHT DOUBLE QUOTATION MARK |
|
197 u'\u2022' # 0x95 -> BULLET |
|
198 u'\u2013' # 0x96 -> EN DASH |
|
199 u'\u2014' # 0x97 -> EM DASH |
|
200 u'\ufffe' # 0x98 -> UNDEFINED |
|
201 u'\ufffe' # 0x99 -> UNDEFINED |
|
202 u'\ufffe' # 0x9A -> UNDEFINED |
|
203 u'\ufffe' # 0x9B -> UNDEFINED |
|
204 u'\ufffe' # 0x9C -> UNDEFINED |
|
205 u'\ufffe' # 0x9D -> UNDEFINED |
|
206 u'\ufffe' # 0x9E -> UNDEFINED |
|
207 u'\ufffe' # 0x9F -> UNDEFINED |
|
208 u'\xa0' # 0xA0 -> NO-BREAK SPACE |
|
209 u'\u0e01' # 0xA1 -> THAI CHARACTER KO KAI |
|
210 u'\u0e02' # 0xA2 -> THAI CHARACTER KHO KHAI |
|
211 u'\u0e03' # 0xA3 -> THAI CHARACTER KHO KHUAT |
|
212 u'\u0e04' # 0xA4 -> THAI CHARACTER KHO KHWAI |
|
213 u'\u0e05' # 0xA5 -> THAI CHARACTER KHO KHON |
|
214 u'\u0e06' # 0xA6 -> THAI CHARACTER KHO RAKHANG |
|
215 u'\u0e07' # 0xA7 -> THAI CHARACTER NGO NGU |
|
216 u'\u0e08' # 0xA8 -> THAI CHARACTER CHO CHAN |
|
217 u'\u0e09' # 0xA9 -> THAI CHARACTER CHO CHING |
|
218 u'\u0e0a' # 0xAA -> THAI CHARACTER CHO CHANG |
|
219 u'\u0e0b' # 0xAB -> THAI CHARACTER SO SO |
|
220 u'\u0e0c' # 0xAC -> THAI CHARACTER CHO CHOE |
|
221 u'\u0e0d' # 0xAD -> THAI CHARACTER YO YING |
|
222 u'\u0e0e' # 0xAE -> THAI CHARACTER DO CHADA |
|
223 u'\u0e0f' # 0xAF -> THAI CHARACTER TO PATAK |
|
224 u'\u0e10' # 0xB0 -> THAI CHARACTER THO THAN |
|
225 u'\u0e11' # 0xB1 -> THAI CHARACTER THO NANGMONTHO |
|
226 u'\u0e12' # 0xB2 -> THAI CHARACTER THO PHUTHAO |
|
227 u'\u0e13' # 0xB3 -> THAI CHARACTER NO NEN |
|
228 u'\u0e14' # 0xB4 -> THAI CHARACTER DO DEK |
|
229 u'\u0e15' # 0xB5 -> THAI CHARACTER TO TAO |
|
230 u'\u0e16' # 0xB6 -> THAI CHARACTER THO THUNG |
|
231 u'\u0e17' # 0xB7 -> THAI CHARACTER THO THAHAN |
|
232 u'\u0e18' # 0xB8 -> THAI CHARACTER THO THONG |
|
233 u'\u0e19' # 0xB9 -> THAI CHARACTER NO NU |
|
234 u'\u0e1a' # 0xBA -> THAI CHARACTER BO BAIMAI |
|
235 u'\u0e1b' # 0xBB -> THAI CHARACTER PO PLA |
|
236 u'\u0e1c' # 0xBC -> THAI CHARACTER PHO PHUNG |
|
237 u'\u0e1d' # 0xBD -> THAI CHARACTER FO FA |
|
238 u'\u0e1e' # 0xBE -> THAI CHARACTER PHO PHAN |
|
239 u'\u0e1f' # 0xBF -> THAI CHARACTER FO FAN |
|
240 u'\u0e20' # 0xC0 -> THAI CHARACTER PHO SAMPHAO |
|
241 u'\u0e21' # 0xC1 -> THAI CHARACTER MO MA |
|
242 u'\u0e22' # 0xC2 -> THAI CHARACTER YO YAK |
|
243 u'\u0e23' # 0xC3 -> THAI CHARACTER RO RUA |
|
244 u'\u0e24' # 0xC4 -> THAI CHARACTER RU |
|
245 u'\u0e25' # 0xC5 -> THAI CHARACTER LO LING |
|
246 u'\u0e26' # 0xC6 -> THAI CHARACTER LU |
|
247 u'\u0e27' # 0xC7 -> THAI CHARACTER WO WAEN |
|
248 u'\u0e28' # 0xC8 -> THAI CHARACTER SO SALA |
|
249 u'\u0e29' # 0xC9 -> THAI CHARACTER SO RUSI |
|
250 u'\u0e2a' # 0xCA -> THAI CHARACTER SO SUA |
|
251 u'\u0e2b' # 0xCB -> THAI CHARACTER HO HIP |
|
252 u'\u0e2c' # 0xCC -> THAI CHARACTER LO CHULA |
|
253 u'\u0e2d' # 0xCD -> THAI CHARACTER O ANG |
|
254 u'\u0e2e' # 0xCE -> THAI CHARACTER HO NOKHUK |
|
255 u'\u0e2f' # 0xCF -> THAI CHARACTER PAIYANNOI |
|
256 u'\u0e30' # 0xD0 -> THAI CHARACTER SARA A |
|
257 u'\u0e31' # 0xD1 -> THAI CHARACTER MAI HAN-AKAT |
|
258 u'\u0e32' # 0xD2 -> THAI CHARACTER SARA AA |
|
259 u'\u0e33' # 0xD3 -> THAI CHARACTER SARA AM |
|
260 u'\u0e34' # 0xD4 -> THAI CHARACTER SARA I |
|
261 u'\u0e35' # 0xD5 -> THAI CHARACTER SARA II |
|
262 u'\u0e36' # 0xD6 -> THAI CHARACTER SARA UE |
|
263 u'\u0e37' # 0xD7 -> THAI CHARACTER SARA UEE |
|
264 u'\u0e38' # 0xD8 -> THAI CHARACTER SARA U |
|
265 u'\u0e39' # 0xD9 -> THAI CHARACTER SARA UU |
|
266 u'\u0e3a' # 0xDA -> THAI CHARACTER PHINTHU |
|
267 u'\ufffe' # 0xDB -> UNDEFINED |
|
268 u'\ufffe' # 0xDC -> UNDEFINED |
|
269 u'\ufffe' # 0xDD -> UNDEFINED |
|
270 u'\ufffe' # 0xDE -> UNDEFINED |
|
271 u'\u0e3f' # 0xDF -> THAI CURRENCY SYMBOL BAHT |
|
272 u'\u0e40' # 0xE0 -> THAI CHARACTER SARA E |
|
273 u'\u0e41' # 0xE1 -> THAI CHARACTER SARA AE |
|
274 u'\u0e42' # 0xE2 -> THAI CHARACTER SARA O |
|
275 u'\u0e43' # 0xE3 -> THAI CHARACTER SARA AI MAIMUAN |
|
276 u'\u0e44' # 0xE4 -> THAI CHARACTER SARA AI MAIMALAI |
|
277 u'\u0e45' # 0xE5 -> THAI CHARACTER LAKKHANGYAO |
|
278 u'\u0e46' # 0xE6 -> THAI CHARACTER MAIYAMOK |
|
279 u'\u0e47' # 0xE7 -> THAI CHARACTER MAITAIKHU |
|
280 u'\u0e48' # 0xE8 -> THAI CHARACTER MAI EK |
|
281 u'\u0e49' # 0xE9 -> THAI CHARACTER MAI THO |
|
282 u'\u0e4a' # 0xEA -> THAI CHARACTER MAI TRI |
|
283 u'\u0e4b' # 0xEB -> THAI CHARACTER MAI CHATTAWA |
|
284 u'\u0e4c' # 0xEC -> THAI CHARACTER THANTHAKHAT |
|
285 u'\u0e4d' # 0xED -> THAI CHARACTER NIKHAHIT |
|
286 u'\u0e4e' # 0xEE -> THAI CHARACTER YAMAKKAN |
|
287 u'\u0e4f' # 0xEF -> THAI CHARACTER FONGMAN |
|
288 u'\u0e50' # 0xF0 -> THAI DIGIT ZERO |
|
289 u'\u0e51' # 0xF1 -> THAI DIGIT ONE |
|
290 u'\u0e52' # 0xF2 -> THAI DIGIT TWO |
|
291 u'\u0e53' # 0xF3 -> THAI DIGIT THREE |
|
292 u'\u0e54' # 0xF4 -> THAI DIGIT FOUR |
|
293 u'\u0e55' # 0xF5 -> THAI DIGIT FIVE |
|
294 u'\u0e56' # 0xF6 -> THAI DIGIT SIX |
|
295 u'\u0e57' # 0xF7 -> THAI DIGIT SEVEN |
|
296 u'\u0e58' # 0xF8 -> THAI DIGIT EIGHT |
|
297 u'\u0e59' # 0xF9 -> THAI DIGIT NINE |
|
298 u'\u0e5a' # 0xFA -> THAI CHARACTER ANGKHANKHU |
|
299 u'\u0e5b' # 0xFB -> THAI CHARACTER KHOMUT |
|
300 u'\ufffe' # 0xFC -> UNDEFINED |
|
301 u'\ufffe' # 0xFD -> UNDEFINED |
|
302 u'\ufffe' # 0xFE -> UNDEFINED |
|
303 u'\ufffe' # 0xFF -> UNDEFINED |
|
304 ) |
|
305 |
|
306 ### Encoding table |
|
307 encoding_table=codecs.charmap_build(decoding_table) |