• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1""" Python Character Mapping Codec generated from 'PTCP154.txt' with gencodec.py.
2
3Written by Marc-Andre Lemburg (mal@lemburg.com).
4
5(c) Copyright CNRI, All Rights Reserved. NO WARRANTY.
6(c) Copyright 2000 Guido van Rossum.
7
8"""#"
9
10import codecs
11
12### Codec APIs
13
14class Codec(codecs.Codec):
15
16    def encode(self,input,errors='strict'):
17        return codecs.charmap_encode(input,errors,encoding_map)
18
19    def decode(self,input,errors='strict'):
20        return codecs.charmap_decode(input,errors,decoding_map)
21
22class IncrementalEncoder(codecs.IncrementalEncoder):
23    def encode(self, input, final=False):
24        return codecs.charmap_encode(input,self.errors,encoding_map)[0]
25
26class IncrementalDecoder(codecs.IncrementalDecoder):
27    def decode(self, input, final=False):
28        return codecs.charmap_decode(input,self.errors,decoding_map)[0]
29
30class StreamWriter(Codec,codecs.StreamWriter):
31    pass
32
33class StreamReader(Codec,codecs.StreamReader):
34    pass
35
36### encodings module API
37
38def getregentry():
39    return codecs.CodecInfo(
40        name='ptcp154',
41        encode=Codec().encode,
42        decode=Codec().decode,
43        incrementalencoder=IncrementalEncoder,
44        incrementaldecoder=IncrementalDecoder,
45        streamreader=StreamReader,
46        streamwriter=StreamWriter,
47    )
48
49### Decoding Map
50
51decoding_map = codecs.make_identity_dict(range(256))
52decoding_map.update({
53        0x0080: 0x0496, #        CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER
54        0x0081: 0x0492, #        CYRILLIC CAPITAL LETTER GHE WITH STROKE
55        0x0082: 0x04ee, #        CYRILLIC CAPITAL LETTER U WITH MACRON
56        0x0083: 0x0493, #        CYRILLIC SMALL LETTER GHE WITH STROKE
57        0x0084: 0x201e, #        DOUBLE LOW-9 QUOTATION MARK
58        0x0085: 0x2026, #        HORIZONTAL ELLIPSIS
59        0x0086: 0x04b6, #        CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
60        0x0087: 0x04ae, #        CYRILLIC CAPITAL LETTER STRAIGHT U
61        0x0088: 0x04b2, #        CYRILLIC CAPITAL LETTER HA WITH DESCENDER
62        0x0089: 0x04af, #        CYRILLIC SMALL LETTER STRAIGHT U
63        0x008a: 0x04a0, #        CYRILLIC CAPITAL LETTER BASHKIR KA
64        0x008b: 0x04e2, #        CYRILLIC CAPITAL LETTER I WITH MACRON
65        0x008c: 0x04a2, #        CYRILLIC CAPITAL LETTER EN WITH DESCENDER
66        0x008d: 0x049a, #        CYRILLIC CAPITAL LETTER KA WITH DESCENDER
67        0x008e: 0x04ba, #        CYRILLIC CAPITAL LETTER SHHA
68        0x008f: 0x04b8, #        CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
69        0x0090: 0x0497, #        CYRILLIC SMALL LETTER ZHE WITH DESCENDER
70        0x0091: 0x2018, #        LEFT SINGLE QUOTATION MARK
71        0x0092: 0x2019, #        RIGHT SINGLE QUOTATION MARK
72        0x0093: 0x201c, #        LEFT DOUBLE QUOTATION MARK
73        0x0094: 0x201d, #        RIGHT DOUBLE QUOTATION MARK
74        0x0095: 0x2022, #        BULLET
75        0x0096: 0x2013, #        EN DASH
76        0x0097: 0x2014, #        EM DASH
77        0x0098: 0x04b3, #        CYRILLIC SMALL LETTER HA WITH DESCENDER
78        0x0099: 0x04b7, #        CYRILLIC SMALL LETTER CHE WITH DESCENDER
79        0x009a: 0x04a1, #        CYRILLIC SMALL LETTER BASHKIR KA
80        0x009b: 0x04e3, #        CYRILLIC SMALL LETTER I WITH MACRON
81        0x009c: 0x04a3, #        CYRILLIC SMALL LETTER EN WITH DESCENDER
82        0x009d: 0x049b, #        CYRILLIC SMALL LETTER KA WITH DESCENDER
83        0x009e: 0x04bb, #        CYRILLIC SMALL LETTER SHHA
84        0x009f: 0x04b9, #        CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE
85        0x00a1: 0x040e, #        CYRILLIC CAPITAL LETTER SHORT U (Byelorussian)
86        0x00a2: 0x045e, #        CYRILLIC SMALL LETTER SHORT U (Byelorussian)
87        0x00a3: 0x0408, #        CYRILLIC CAPITAL LETTER JE
88        0x00a4: 0x04e8, #        CYRILLIC CAPITAL LETTER BARRED O
89        0x00a5: 0x0498, #        CYRILLIC CAPITAL LETTER ZE WITH DESCENDER
90        0x00a6: 0x04b0, #        CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE
91        0x00a8: 0x0401, #        CYRILLIC CAPITAL LETTER IO
92        0x00aa: 0x04d8, #        CYRILLIC CAPITAL LETTER SCHWA
93        0x00ad: 0x04ef, #        CYRILLIC SMALL LETTER U WITH MACRON
94        0x00af: 0x049c, #        CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
95        0x00b1: 0x04b1, #        CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE
96        0x00b2: 0x0406, #        CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
97        0x00b3: 0x0456, #        CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
98        0x00b4: 0x0499, #        CYRILLIC SMALL LETTER ZE WITH DESCENDER
99        0x00b5: 0x04e9, #        CYRILLIC SMALL LETTER BARRED O
100        0x00b8: 0x0451, #        CYRILLIC SMALL LETTER IO
101        0x00b9: 0x2116, #        NUMERO SIGN
102        0x00ba: 0x04d9, #        CYRILLIC SMALL LETTER SCHWA
103        0x00bc: 0x0458, #        CYRILLIC SMALL LETTER JE
104        0x00bd: 0x04aa, #        CYRILLIC CAPITAL LETTER ES WITH DESCENDER
105        0x00be: 0x04ab, #        CYRILLIC SMALL LETTER ES WITH DESCENDER
106        0x00bf: 0x049d, #        CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE
107        0x00c0: 0x0410, #        CYRILLIC CAPITAL LETTER A
108        0x00c1: 0x0411, #        CYRILLIC CAPITAL LETTER BE
109        0x00c2: 0x0412, #        CYRILLIC CAPITAL LETTER VE
110        0x00c3: 0x0413, #        CYRILLIC CAPITAL LETTER GHE
111        0x00c4: 0x0414, #        CYRILLIC CAPITAL LETTER DE
112        0x00c5: 0x0415, #        CYRILLIC CAPITAL LETTER IE
113        0x00c6: 0x0416, #        CYRILLIC CAPITAL LETTER ZHE
114        0x00c7: 0x0417, #        CYRILLIC CAPITAL LETTER ZE
115        0x00c8: 0x0418, #        CYRILLIC CAPITAL LETTER I
116        0x00c9: 0x0419, #        CYRILLIC CAPITAL LETTER SHORT I
117        0x00ca: 0x041a, #        CYRILLIC CAPITAL LETTER KA
118        0x00cb: 0x041b, #        CYRILLIC CAPITAL LETTER EL
119        0x00cc: 0x041c, #        CYRILLIC CAPITAL LETTER EM
120        0x00cd: 0x041d, #        CYRILLIC CAPITAL LETTER EN
121        0x00ce: 0x041e, #        CYRILLIC CAPITAL LETTER O
122        0x00cf: 0x041f, #        CYRILLIC CAPITAL LETTER PE
123        0x00d0: 0x0420, #        CYRILLIC CAPITAL LETTER ER
124        0x00d1: 0x0421, #        CYRILLIC CAPITAL LETTER ES
125        0x00d2: 0x0422, #        CYRILLIC CAPITAL LETTER TE
126        0x00d3: 0x0423, #        CYRILLIC CAPITAL LETTER U
127        0x00d4: 0x0424, #        CYRILLIC CAPITAL LETTER EF
128        0x00d5: 0x0425, #        CYRILLIC CAPITAL LETTER HA
129        0x00d6: 0x0426, #        CYRILLIC CAPITAL LETTER TSE
130        0x00d7: 0x0427, #        CYRILLIC CAPITAL LETTER CHE
131        0x00d8: 0x0428, #        CYRILLIC CAPITAL LETTER SHA
132        0x00d9: 0x0429, #        CYRILLIC CAPITAL LETTER SHCHA
133        0x00da: 0x042a, #        CYRILLIC CAPITAL LETTER HARD SIGN
134        0x00db: 0x042b, #        CYRILLIC CAPITAL LETTER YERU
135        0x00dc: 0x042c, #        CYRILLIC CAPITAL LETTER SOFT SIGN
136        0x00dd: 0x042d, #        CYRILLIC CAPITAL LETTER E
137        0x00de: 0x042e, #        CYRILLIC CAPITAL LETTER YU
138        0x00df: 0x042f, #        CYRILLIC CAPITAL LETTER YA
139        0x00e0: 0x0430, #        CYRILLIC SMALL LETTER A
140        0x00e1: 0x0431, #        CYRILLIC SMALL LETTER BE
141        0x00e2: 0x0432, #        CYRILLIC SMALL LETTER VE
142        0x00e3: 0x0433, #        CYRILLIC SMALL LETTER GHE
143        0x00e4: 0x0434, #        CYRILLIC SMALL LETTER DE
144        0x00e5: 0x0435, #        CYRILLIC SMALL LETTER IE
145        0x00e6: 0x0436, #        CYRILLIC SMALL LETTER ZHE
146        0x00e7: 0x0437, #        CYRILLIC SMALL LETTER ZE
147        0x00e8: 0x0438, #        CYRILLIC SMALL LETTER I
148        0x00e9: 0x0439, #        CYRILLIC SMALL LETTER SHORT I
149        0x00ea: 0x043a, #        CYRILLIC SMALL LETTER KA
150        0x00eb: 0x043b, #        CYRILLIC SMALL LETTER EL
151        0x00ec: 0x043c, #        CYRILLIC SMALL LETTER EM
152        0x00ed: 0x043d, #        CYRILLIC SMALL LETTER EN
153        0x00ee: 0x043e, #        CYRILLIC SMALL LETTER O
154        0x00ef: 0x043f, #        CYRILLIC SMALL LETTER PE
155        0x00f0: 0x0440, #        CYRILLIC SMALL LETTER ER
156        0x00f1: 0x0441, #        CYRILLIC SMALL LETTER ES
157        0x00f2: 0x0442, #        CYRILLIC SMALL LETTER TE
158        0x00f3: 0x0443, #        CYRILLIC SMALL LETTER U
159        0x00f4: 0x0444, #        CYRILLIC SMALL LETTER EF
160        0x00f5: 0x0445, #        CYRILLIC SMALL LETTER HA
161        0x00f6: 0x0446, #        CYRILLIC SMALL LETTER TSE
162        0x00f7: 0x0447, #        CYRILLIC SMALL LETTER CHE
163        0x00f8: 0x0448, #        CYRILLIC SMALL LETTER SHA
164        0x00f9: 0x0449, #        CYRILLIC SMALL LETTER SHCHA
165        0x00fa: 0x044a, #        CYRILLIC SMALL LETTER HARD SIGN
166        0x00fb: 0x044b, #        CYRILLIC SMALL LETTER YERU
167        0x00fc: 0x044c, #        CYRILLIC SMALL LETTER SOFT SIGN
168        0x00fd: 0x044d, #        CYRILLIC SMALL LETTER E
169        0x00fe: 0x044e, #        CYRILLIC SMALL LETTER YU
170        0x00ff: 0x044f, #        CYRILLIC SMALL LETTER YA
171})
172
173### Encoding Map
174
175encoding_map = codecs.make_encoding_map(decoding_map)
176