]>
Commit | Line | Data |
---|---|---|
3257aa99 DM |
1 | """ Python Character Mapping Codec generated from 'PTCP154.txt' with gencodec.py.\r |
2 | \r | |
3 | Written by Marc-Andre Lemburg (mal@lemburg.com).\r | |
4 | \r | |
5 | (c) Copyright CNRI, All Rights Reserved. NO WARRANTY.\r | |
6 | (c) Copyright 2000 Guido van Rossum.\r | |
7 | \r | |
8 | """#"\r | |
9 | \r | |
10 | import codecs\r | |
11 | \r | |
12 | ### Codec APIs\r | |
13 | \r | |
14 | class Codec(codecs.Codec):\r | |
15 | \r | |
16 | def encode(self,input,errors='strict'):\r | |
17 | return codecs.charmap_encode(input,errors,encoding_map)\r | |
18 | \r | |
19 | def decode(self,input,errors='strict'):\r | |
20 | return codecs.charmap_decode(input,errors,decoding_map)\r | |
21 | \r | |
22 | class IncrementalEncoder(codecs.IncrementalEncoder):\r | |
23 | def encode(self, input, final=False):\r | |
24 | return codecs.charmap_encode(input,self.errors,encoding_map)[0]\r | |
25 | \r | |
26 | class IncrementalDecoder(codecs.IncrementalDecoder):\r | |
27 | def decode(self, input, final=False):\r | |
28 | return codecs.charmap_decode(input,self.errors,decoding_map)[0]\r | |
29 | \r | |
30 | class StreamWriter(Codec,codecs.StreamWriter):\r | |
31 | pass\r | |
32 | \r | |
33 | class StreamReader(Codec,codecs.StreamReader):\r | |
34 | pass\r | |
35 | \r | |
36 | ### encodings module API\r | |
37 | \r | |
38 | def getregentry():\r | |
39 | return codecs.CodecInfo(\r | |
40 | name='ptcp154',\r | |
41 | encode=Codec().encode,\r | |
42 | decode=Codec().decode,\r | |
43 | incrementalencoder=IncrementalEncoder,\r | |
44 | incrementaldecoder=IncrementalDecoder,\r | |
45 | streamreader=StreamReader,\r | |
46 | streamwriter=StreamWriter,\r | |
47 | )\r | |
48 | \r | |
49 | ### Decoding Map\r | |
50 | \r | |
51 | decoding_map = codecs.make_identity_dict(range(256))\r | |
52 | decoding_map.update({\r | |
53 | 0x0080: 0x0496, # CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER\r | |
54 | 0x0081: 0x0492, # CYRILLIC CAPITAL LETTER GHE WITH STROKE\r | |
55 | 0x0082: 0x04ee, # CYRILLIC CAPITAL LETTER U WITH MACRON\r | |
56 | 0x0083: 0x0493, # CYRILLIC SMALL LETTER GHE WITH STROKE\r | |
57 | 0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK\r | |
58 | 0x0085: 0x2026, # HORIZONTAL ELLIPSIS\r | |
59 | 0x0086: 0x04b6, # CYRILLIC CAPITAL LETTER CHE WITH DESCENDER\r | |
60 | 0x0087: 0x04ae, # CYRILLIC CAPITAL LETTER STRAIGHT U\r | |
61 | 0x0088: 0x04b2, # CYRILLIC CAPITAL LETTER HA WITH DESCENDER\r | |
62 | 0x0089: 0x04af, # CYRILLIC SMALL LETTER STRAIGHT U\r | |
63 | 0x008a: 0x04a0, # CYRILLIC CAPITAL LETTER BASHKIR KA\r | |
64 | 0x008b: 0x04e2, # CYRILLIC CAPITAL LETTER I WITH MACRON\r | |
65 | 0x008c: 0x04a2, # CYRILLIC CAPITAL LETTER EN WITH DESCENDER\r | |
66 | 0x008d: 0x049a, # CYRILLIC CAPITAL LETTER KA WITH DESCENDER\r | |
67 | 0x008e: 0x04ba, # CYRILLIC CAPITAL LETTER SHHA\r | |
68 | 0x008f: 0x04b8, # CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE\r | |
69 | 0x0090: 0x0497, # CYRILLIC SMALL LETTER ZHE WITH DESCENDER\r | |
70 | 0x0091: 0x2018, # LEFT SINGLE QUOTATION MARK\r | |
71 | 0x0092: 0x2019, # RIGHT SINGLE QUOTATION MARK\r | |
72 | 0x0093: 0x201c, # LEFT DOUBLE QUOTATION MARK\r | |
73 | 0x0094: 0x201d, # RIGHT DOUBLE QUOTATION MARK\r | |
74 | 0x0095: 0x2022, # BULLET\r | |
75 | 0x0096: 0x2013, # EN DASH\r | |
76 | 0x0097: 0x2014, # EM DASH\r | |
77 | 0x0098: 0x04b3, # CYRILLIC SMALL LETTER HA WITH DESCENDER\r | |
78 | 0x0099: 0x04b7, # CYRILLIC SMALL LETTER CHE WITH DESCENDER\r | |
79 | 0x009a: 0x04a1, # CYRILLIC SMALL LETTER BASHKIR KA\r | |
80 | 0x009b: 0x04e3, # CYRILLIC SMALL LETTER I WITH MACRON\r | |
81 | 0x009c: 0x04a3, # CYRILLIC SMALL LETTER EN WITH DESCENDER\r | |
82 | 0x009d: 0x049b, # CYRILLIC SMALL LETTER KA WITH DESCENDER\r | |
83 | 0x009e: 0x04bb, # CYRILLIC SMALL LETTER SHHA\r | |
84 | 0x009f: 0x04b9, # CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE\r | |
85 | 0x00a1: 0x040e, # CYRILLIC CAPITAL LETTER SHORT U (Byelorussian)\r | |
86 | 0x00a2: 0x045e, # CYRILLIC SMALL LETTER SHORT U (Byelorussian)\r | |
87 | 0x00a3: 0x0408, # CYRILLIC CAPITAL LETTER JE\r | |
88 | 0x00a4: 0x04e8, # CYRILLIC CAPITAL LETTER BARRED O\r | |
89 | 0x00a5: 0x0498, # CYRILLIC CAPITAL LETTER ZE WITH DESCENDER\r | |
90 | 0x00a6: 0x04b0, # CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE\r | |
91 | 0x00a8: 0x0401, # CYRILLIC CAPITAL LETTER IO\r | |
92 | 0x00aa: 0x04d8, # CYRILLIC CAPITAL LETTER SCHWA\r | |
93 | 0x00ad: 0x04ef, # CYRILLIC SMALL LETTER U WITH MACRON\r | |
94 | 0x00af: 0x049c, # CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE\r | |
95 | 0x00b1: 0x04b1, # CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE\r | |
96 | 0x00b2: 0x0406, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I\r | |
97 | 0x00b3: 0x0456, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I\r | |
98 | 0x00b4: 0x0499, # CYRILLIC SMALL LETTER ZE WITH DESCENDER\r | |
99 | 0x00b5: 0x04e9, # CYRILLIC SMALL LETTER BARRED O\r | |
100 | 0x00b8: 0x0451, # CYRILLIC SMALL LETTER IO\r | |
101 | 0x00b9: 0x2116, # NUMERO SIGN\r | |
102 | 0x00ba: 0x04d9, # CYRILLIC SMALL LETTER SCHWA\r | |
103 | 0x00bc: 0x0458, # CYRILLIC SMALL LETTER JE\r | |
104 | 0x00bd: 0x04aa, # CYRILLIC CAPITAL LETTER ES WITH DESCENDER\r | |
105 | 0x00be: 0x04ab, # CYRILLIC SMALL LETTER ES WITH DESCENDER\r | |
106 | 0x00bf: 0x049d, # CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE\r | |
107 | 0x00c0: 0x0410, # CYRILLIC CAPITAL LETTER A\r | |
108 | 0x00c1: 0x0411, # CYRILLIC CAPITAL LETTER BE\r | |
109 | 0x00c2: 0x0412, # CYRILLIC CAPITAL LETTER VE\r | |
110 | 0x00c3: 0x0413, # CYRILLIC CAPITAL LETTER GHE\r | |
111 | 0x00c4: 0x0414, # CYRILLIC CAPITAL LETTER DE\r | |
112 | 0x00c5: 0x0415, # CYRILLIC CAPITAL LETTER IE\r | |
113 | 0x00c6: 0x0416, # CYRILLIC CAPITAL LETTER ZHE\r | |
114 | 0x00c7: 0x0417, # CYRILLIC CAPITAL LETTER ZE\r | |
115 | 0x00c8: 0x0418, # CYRILLIC CAPITAL LETTER I\r | |
116 | 0x00c9: 0x0419, # CYRILLIC CAPITAL LETTER SHORT I\r | |
117 | 0x00ca: 0x041a, # CYRILLIC CAPITAL LETTER KA\r | |
118 | 0x00cb: 0x041b, # CYRILLIC CAPITAL LETTER EL\r | |
119 | 0x00cc: 0x041c, # CYRILLIC CAPITAL LETTER EM\r | |
120 | 0x00cd: 0x041d, # CYRILLIC CAPITAL LETTER EN\r | |
121 | 0x00ce: 0x041e, # CYRILLIC CAPITAL LETTER O\r | |
122 | 0x00cf: 0x041f, # CYRILLIC CAPITAL LETTER PE\r | |
123 | 0x00d0: 0x0420, # CYRILLIC CAPITAL LETTER ER\r | |
124 | 0x00d1: 0x0421, # CYRILLIC CAPITAL LETTER ES\r | |
125 | 0x00d2: 0x0422, # CYRILLIC CAPITAL LETTER TE\r | |
126 | 0x00d3: 0x0423, # CYRILLIC CAPITAL LETTER U\r | |
127 | 0x00d4: 0x0424, # CYRILLIC CAPITAL LETTER EF\r | |
128 | 0x00d5: 0x0425, # CYRILLIC CAPITAL LETTER HA\r | |
129 | 0x00d6: 0x0426, # CYRILLIC CAPITAL LETTER TSE\r | |
130 | 0x00d7: 0x0427, # CYRILLIC CAPITAL LETTER CHE\r | |
131 | 0x00d8: 0x0428, # CYRILLIC CAPITAL LETTER SHA\r | |
132 | 0x00d9: 0x0429, # CYRILLIC CAPITAL LETTER SHCHA\r | |
133 | 0x00da: 0x042a, # CYRILLIC CAPITAL LETTER HARD SIGN\r | |
134 | 0x00db: 0x042b, # CYRILLIC CAPITAL LETTER YERU\r | |
135 | 0x00dc: 0x042c, # CYRILLIC CAPITAL LETTER SOFT SIGN\r | |
136 | 0x00dd: 0x042d, # CYRILLIC CAPITAL LETTER E\r | |
137 | 0x00de: 0x042e, # CYRILLIC CAPITAL LETTER YU\r | |
138 | 0x00df: 0x042f, # CYRILLIC CAPITAL LETTER YA\r | |
139 | 0x00e0: 0x0430, # CYRILLIC SMALL LETTER A\r | |
140 | 0x00e1: 0x0431, # CYRILLIC SMALL LETTER BE\r | |
141 | 0x00e2: 0x0432, # CYRILLIC SMALL LETTER VE\r | |
142 | 0x00e3: 0x0433, # CYRILLIC SMALL LETTER GHE\r | |
143 | 0x00e4: 0x0434, # CYRILLIC SMALL LETTER DE\r | |
144 | 0x00e5: 0x0435, # CYRILLIC SMALL LETTER IE\r | |
145 | 0x00e6: 0x0436, # CYRILLIC SMALL LETTER ZHE\r | |
146 | 0x00e7: 0x0437, # CYRILLIC SMALL LETTER ZE\r | |
147 | 0x00e8: 0x0438, # CYRILLIC SMALL LETTER I\r | |
148 | 0x00e9: 0x0439, # CYRILLIC SMALL LETTER SHORT I\r | |
149 | 0x00ea: 0x043a, # CYRILLIC SMALL LETTER KA\r | |
150 | 0x00eb: 0x043b, # CYRILLIC SMALL LETTER EL\r | |
151 | 0x00ec: 0x043c, # CYRILLIC SMALL LETTER EM\r | |
152 | 0x00ed: 0x043d, # CYRILLIC SMALL LETTER EN\r | |
153 | 0x00ee: 0x043e, # CYRILLIC SMALL LETTER O\r | |
154 | 0x00ef: 0x043f, # CYRILLIC SMALL LETTER PE\r | |
155 | 0x00f0: 0x0440, # CYRILLIC SMALL LETTER ER\r | |
156 | 0x00f1: 0x0441, # CYRILLIC SMALL LETTER ES\r | |
157 | 0x00f2: 0x0442, # CYRILLIC SMALL LETTER TE\r | |
158 | 0x00f3: 0x0443, # CYRILLIC SMALL LETTER U\r | |
159 | 0x00f4: 0x0444, # CYRILLIC SMALL LETTER EF\r | |
160 | 0x00f5: 0x0445, # CYRILLIC SMALL LETTER HA\r | |
161 | 0x00f6: 0x0446, # CYRILLIC SMALL LETTER TSE\r | |
162 | 0x00f7: 0x0447, # CYRILLIC SMALL LETTER CHE\r | |
163 | 0x00f8: 0x0448, # CYRILLIC SMALL LETTER SHA\r | |
164 | 0x00f9: 0x0449, # CYRILLIC SMALL LETTER SHCHA\r | |
165 | 0x00fa: 0x044a, # CYRILLIC SMALL LETTER HARD SIGN\r | |
166 | 0x00fb: 0x044b, # CYRILLIC SMALL LETTER YERU\r | |
167 | 0x00fc: 0x044c, # CYRILLIC SMALL LETTER SOFT SIGN\r | |
168 | 0x00fd: 0x044d, # CYRILLIC SMALL LETTER E\r | |
169 | 0x00fe: 0x044e, # CYRILLIC SMALL LETTER YU\r | |
170 | 0x00ff: 0x044f, # CYRILLIC SMALL LETTER YA\r | |
171 | })\r | |
172 | \r | |
173 | ### Encoding Map\r | |
174 | \r | |
175 | encoding_map = codecs.make_encoding_map(decoding_map)\r |