Commit | Line | Data |
---|---|---|
920dae64 AT |
1 | """ Python Character Mapping Codec generated from 'CP1257.TXT' with gencodec.py. |
2 | ||
3 | Written 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 | ||
10 | import codecs | |
11 | ||
12 | ### Codec APIs | |
13 | ||
14 | class Codec(codecs.Codec): | |
15 | ||
16 | def encode(self,input,errors='strict'): | |
17 | ||
18 | return codecs.charmap_encode(input,errors,encoding_map) | |
19 | ||
20 | def decode(self,input,errors='strict'): | |
21 | ||
22 | return codecs.charmap_decode(input,errors,decoding_map) | |
23 | ||
24 | class StreamWriter(Codec,codecs.StreamWriter): | |
25 | pass | |
26 | ||
27 | class StreamReader(Codec,codecs.StreamReader): | |
28 | pass | |
29 | ||
30 | ### encodings module API | |
31 | ||
32 | def getregentry(): | |
33 | ||
34 | return (Codec().encode,Codec().decode,StreamReader,StreamWriter) | |
35 | ||
36 | ### Decoding Map | |
37 | ||
38 | decoding_map = codecs.make_identity_dict(range(256)) | |
39 | decoding_map.update({ | |
40 | 0x0080: 0x20ac, # EURO SIGN | |
41 | 0x0081: None, # UNDEFINED | |
42 | 0x0082: 0x201a, # SINGLE LOW-9 QUOTATION MARK | |
43 | 0x0083: None, # UNDEFINED | |
44 | 0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK | |
45 | 0x0085: 0x2026, # HORIZONTAL ELLIPSIS | |
46 | 0x0086: 0x2020, # DAGGER | |
47 | 0x0087: 0x2021, # DOUBLE DAGGER | |
48 | 0x0088: None, # UNDEFINED | |
49 | 0x0089: 0x2030, # PER MILLE SIGN | |
50 | 0x008a: None, # UNDEFINED | |
51 | 0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK | |
52 | 0x008c: None, # UNDEFINED | |
53 | 0x008d: 0x00a8, # DIAERESIS | |
54 | 0x008e: 0x02c7, # CARON | |
55 | 0x008f: 0x00b8, # CEDILLA | |
56 | 0x0090: None, # UNDEFINED | |
57 | 0x0091: 0x2018, # LEFT SINGLE QUOTATION MARK | |
58 | 0x0092: 0x2019, # RIGHT SINGLE QUOTATION MARK | |
59 | 0x0093: 0x201c, # LEFT DOUBLE QUOTATION MARK | |
60 | 0x0094: 0x201d, # RIGHT DOUBLE QUOTATION MARK | |
61 | 0x0095: 0x2022, # BULLET | |
62 | 0x0096: 0x2013, # EN DASH | |
63 | 0x0097: 0x2014, # EM DASH | |
64 | 0x0098: None, # UNDEFINED | |
65 | 0x0099: 0x2122, # TRADE MARK SIGN | |
66 | 0x009a: None, # UNDEFINED | |
67 | 0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK | |
68 | 0x009c: None, # UNDEFINED | |
69 | 0x009d: 0x00af, # MACRON | |
70 | 0x009e: 0x02db, # OGONEK | |
71 | 0x009f: None, # UNDEFINED | |
72 | 0x00a1: None, # UNDEFINED | |
73 | 0x00a5: None, # UNDEFINED | |
74 | 0x00a8: 0x00d8, # LATIN CAPITAL LETTER O WITH STROKE | |
75 | 0x00aa: 0x0156, # LATIN CAPITAL LETTER R WITH CEDILLA | |
76 | 0x00af: 0x00c6, # LATIN CAPITAL LETTER AE | |
77 | 0x00b8: 0x00f8, # LATIN SMALL LETTER O WITH STROKE | |
78 | 0x00ba: 0x0157, # LATIN SMALL LETTER R WITH CEDILLA | |
79 | 0x00bf: 0x00e6, # LATIN SMALL LETTER AE | |
80 | 0x00c0: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK | |
81 | 0x00c1: 0x012e, # LATIN CAPITAL LETTER I WITH OGONEK | |
82 | 0x00c2: 0x0100, # LATIN CAPITAL LETTER A WITH MACRON | |
83 | 0x00c3: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE | |
84 | 0x00c6: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK | |
85 | 0x00c7: 0x0112, # LATIN CAPITAL LETTER E WITH MACRON | |
86 | 0x00c8: 0x010c, # LATIN CAPITAL LETTER C WITH CARON | |
87 | 0x00ca: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE | |
88 | 0x00cb: 0x0116, # LATIN CAPITAL LETTER E WITH DOT ABOVE | |
89 | 0x00cc: 0x0122, # LATIN CAPITAL LETTER G WITH CEDILLA | |
90 | 0x00cd: 0x0136, # LATIN CAPITAL LETTER K WITH CEDILLA | |
91 | 0x00ce: 0x012a, # LATIN CAPITAL LETTER I WITH MACRON | |
92 | 0x00cf: 0x013b, # LATIN CAPITAL LETTER L WITH CEDILLA | |
93 | 0x00d0: 0x0160, # LATIN CAPITAL LETTER S WITH CARON | |
94 | 0x00d1: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE | |
95 | 0x00d2: 0x0145, # LATIN CAPITAL LETTER N WITH CEDILLA | |
96 | 0x00d4: 0x014c, # LATIN CAPITAL LETTER O WITH MACRON | |
97 | 0x00d8: 0x0172, # LATIN CAPITAL LETTER U WITH OGONEK | |
98 | 0x00d9: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE | |
99 | 0x00da: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE | |
100 | 0x00db: 0x016a, # LATIN CAPITAL LETTER U WITH MACRON | |
101 | 0x00dd: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE | |
102 | 0x00de: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON | |
103 | 0x00e0: 0x0105, # LATIN SMALL LETTER A WITH OGONEK | |
104 | 0x00e1: 0x012f, # LATIN SMALL LETTER I WITH OGONEK | |
105 | 0x00e2: 0x0101, # LATIN SMALL LETTER A WITH MACRON | |
106 | 0x00e3: 0x0107, # LATIN SMALL LETTER C WITH ACUTE | |
107 | 0x00e6: 0x0119, # LATIN SMALL LETTER E WITH OGONEK | |
108 | 0x00e7: 0x0113, # LATIN SMALL LETTER E WITH MACRON | |
109 | 0x00e8: 0x010d, # LATIN SMALL LETTER C WITH CARON | |
110 | 0x00ea: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE | |
111 | 0x00eb: 0x0117, # LATIN SMALL LETTER E WITH DOT ABOVE | |
112 | 0x00ec: 0x0123, # LATIN SMALL LETTER G WITH CEDILLA | |
113 | 0x00ed: 0x0137, # LATIN SMALL LETTER K WITH CEDILLA | |
114 | 0x00ee: 0x012b, # LATIN SMALL LETTER I WITH MACRON | |
115 | 0x00ef: 0x013c, # LATIN SMALL LETTER L WITH CEDILLA | |
116 | 0x00f0: 0x0161, # LATIN SMALL LETTER S WITH CARON | |
117 | 0x00f1: 0x0144, # LATIN SMALL LETTER N WITH ACUTE | |
118 | 0x00f2: 0x0146, # LATIN SMALL LETTER N WITH CEDILLA | |
119 | 0x00f4: 0x014d, # LATIN SMALL LETTER O WITH MACRON | |
120 | 0x00f8: 0x0173, # LATIN SMALL LETTER U WITH OGONEK | |
121 | 0x00f9: 0x0142, # LATIN SMALL LETTER L WITH STROKE | |
122 | 0x00fa: 0x015b, # LATIN SMALL LETTER S WITH ACUTE | |
123 | 0x00fb: 0x016b, # LATIN SMALL LETTER U WITH MACRON | |
124 | 0x00fd: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE | |
125 | 0x00fe: 0x017e, # LATIN SMALL LETTER Z WITH CARON | |
126 | 0x00ff: 0x02d9, # DOT ABOVE | |
127 | }) | |
128 | ||
129 | ### Encoding Map | |
130 | ||
131 | encoding_map = codecs.make_encoding_map(decoding_map) |