projects / OpenSPARC-T2-DV / blame
| Commit | Line | Data |
|---|---|---|
| 86530b38 AT |
1 | """ Python Character Mapping Codec for PalmOS 3.5. |
| 2 | ||
| 3 | Written by Sjoerd Mullender (sjoerd@acm.org); based on iso8859_15.py. | |
| 4 | ||
| 5 | """#" | |
| 6 | ||
| 7 | import codecs | |
| 8 | ||
| 9 | ### Codec APIs | |
| 10 | ||
| 11 | class Codec(codecs.Codec): | |
| 12 | def encode(self,input,errors='strict'): | |
| 13 | return codecs.charmap_encode(input,errors,encoding_map) | |
| 14 | ||
| 15 | def decode(self,input,errors='strict'): | |
| 16 | return codecs.charmap_decode(input,errors,decoding_map) | |
| 17 | ||
| 18 | class StreamWriter(Codec,codecs.StreamWriter): | |
| 19 | pass | |
| 20 | ||
| 21 | class StreamReader(Codec,codecs.StreamReader): | |
| 22 | pass | |
| 23 | ||
| 24 | ### encodings module API | |
| 25 | ||
| 26 | def getregentry(): | |
| 27 | return (Codec().encode,Codec().decode,StreamReader,StreamWriter) | |
| 28 | ||
| 29 | ### Decoding Map | |
| 30 | ||
| 31 | decoding_map = codecs.make_identity_dict(range(256)) | |
| 32 | ||
| 33 | # The PalmOS character set is mostly iso-8859-1 with some differences. | |
| 34 | decoding_map.update({ | |
| 35 | 0x0080: 0x20ac, # EURO SIGN | |
| 36 | 0x0082: 0x201a, # SINGLE LOW-9 QUOTATION MARK | |
| 37 | 0x0083: 0x0192, # LATIN SMALL LETTER F WITH HOOK | |
| 38 | 0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK | |
| 39 | 0x0085: 0x2026, # HORIZONTAL ELLIPSIS | |
| 40 | 0x0086: 0x2020, # DAGGER | |
| 41 | 0x0087: 0x2021, # DOUBLE DAGGER | |
| 42 | 0x0088: 0x02c6, # MODIFIER LETTER CIRCUMFLEX ACCENT | |
| 43 | 0x0089: 0x2030, # PER MILLE SIGN | |
| 44 | 0x008a: 0x0160, # LATIN CAPITAL LETTER S WITH CARON | |
| 45 | 0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK | |
| 46 | 0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE | |
| 47 | 0x008d: 0x2666, # BLACK DIAMOND SUIT | |
| 48 | 0x008e: 0x2663, # BLACK CLUB SUIT | |
| 49 | 0x008f: 0x2665, # BLACK HEART SUIT | |
| 50 | 0x0090: 0x2660, # BLACK SPADE SUIT | |
| 51 | 0x0091: 0x2018, # LEFT SINGLE QUOTATION MARK | |
| 52 | 0x0092: 0x2019, # RIGHT SINGLE QUOTATION MARK | |
| 53 | 0x0093: 0x201c, # LEFT DOUBLE QUOTATION MARK | |
| 54 | 0x0094: 0x201d, # RIGHT DOUBLE QUOTATION MARK | |
| 55 | 0x0095: 0x2022, # BULLET | |
| 56 | 0x0096: 0x2013, # EN DASH | |
| 57 | 0x0097: 0x2014, # EM DASH | |
| 58 | 0x0098: 0x02dc, # SMALL TILDE | |
| 59 | 0x0099: 0x2122, # TRADE MARK SIGN | |
| 60 | 0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON | |
| 61 | 0x009c: 0x0153, # LATIN SMALL LIGATURE OE | |
| 62 | 0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS | |
| 63 | }) | |
| 64 | ||
| 65 | ### Encoding Map | |
| 66 | ||
| 67 | encoding_map = codecs.make_encoding_map(decoding_map) |