| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame^] | 1 | /* |
| 2 | * _cp932.c: the CP932 codec |
| 3 | * |
| 4 | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
| 5 | * $CJKCodecs: _cp932.c,v 1.2 2003/12/31 05:46:55 perky Exp $ |
| 6 | */ |
| 7 | |
| 8 | #include "codeccommon.h" |
| 9 | |
| 10 | ENCMAP(jisxcommon) |
| 11 | ENCMAP(cp932ext) |
| 12 | DECMAP(jisx0208) |
| 13 | DECMAP(cp932ext) |
| 14 | |
| 15 | ENCODER(cp932) |
| 16 | { |
| 17 | while (inleft > 0) { |
| 18 | Py_UNICODE c = IN1; |
| 19 | DBCHAR code; |
| 20 | unsigned char c1, c2; |
| 21 | |
| 22 | if (c <= 0x80) { |
| 23 | WRITE1(c) |
| 24 | NEXT(1, 1) |
| 25 | continue; |
| 26 | } else if (c >= 0xff61 && c <= 0xff9f) { |
| 27 | WRITE1(c - 0xfec0) |
| 28 | NEXT(1, 1) |
| 29 | continue; |
| 30 | } else if (c >= 0xf8f0 && c <= 0xf8f3) { |
| 31 | /* Windows compatability */ |
| 32 | RESERVE_OUTBUF(1) |
| 33 | if (c == 0xf8f0) |
| 34 | OUT1(0xa0) |
| 35 | else |
| 36 | OUT1(c - 0xfef1 + 0xfd) |
| 37 | NEXT(1, 1) |
| 38 | continue; |
| 39 | } |
| 40 | |
| 41 | UCS4INVALID(c) |
| 42 | RESERVE_OUTBUF(2) |
| 43 | |
| 44 | TRYMAP_ENC(cp932ext, code, c) { |
| 45 | OUT1(code >> 8) |
| 46 | OUT2(code & 0xff) |
| 47 | } else TRYMAP_ENC(jisxcommon, code, c) { |
| 48 | if (code & 0x8000) /* MSB set: JIS X 0212 */ |
| 49 | return 1; |
| 50 | |
| 51 | /* JIS X 0208 */ |
| 52 | c1 = code >> 8; |
| 53 | c2 = code & 0xff; |
| 54 | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); |
| 55 | c1 = (c1 - 0x21) >> 1; |
| 56 | OUT1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1) |
| 57 | OUT2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41) |
| 58 | } else if (c >= 0xe000 && c < 0xe758) { |
| 59 | /* User-defined area */ |
| 60 | c1 = (Py_UNICODE)(c - 0xe000) / 188; |
| 61 | c2 = (Py_UNICODE)(c - 0xe000) % 188; |
| 62 | OUT1(c1 + 0xf0) |
| 63 | OUT2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41) |
| 64 | } else |
| 65 | return 1; |
| 66 | |
| 67 | NEXT(1, 2) |
| 68 | } |
| 69 | |
| 70 | return 0; |
| 71 | } |
| 72 | |
| 73 | DECODER(cp932) |
| 74 | { |
| 75 | while (inleft > 0) { |
| 76 | unsigned char c = IN1, c2; |
| 77 | |
| 78 | RESERVE_OUTBUF(1) |
| 79 | if (c <= 0x80) { |
| 80 | OUT1(c) |
| 81 | NEXT(1, 1) |
| 82 | continue; |
| 83 | } else if (c >= 0xa0 && c <= 0xdf) { |
| 84 | if (c == 0xa0) |
| 85 | OUT1(0xf8f0) /* half-width katakana */ |
| 86 | else |
| 87 | OUT1(0xfec0 + c) |
| 88 | NEXT(1, 1) |
| 89 | continue; |
| 90 | } else if (c >= 0xfd/* && c <= 0xff*/) { |
| 91 | /* Windows compatibility */ |
| 92 | OUT1(0xf8f1 - 0xfd + c) |
| 93 | NEXT(1, 1) |
| 94 | continue; |
| 95 | } |
| 96 | |
| 97 | RESERVE_INBUF(2) |
| 98 | c2 = IN2; |
| 99 | |
| 100 | TRYMAP_DEC(cp932ext, **outbuf, c, c2); |
| 101 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)) { |
| 102 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
| 103 | return 2; |
| 104 | |
| 105 | c = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
| 106 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
| 107 | c = (2 * c + (c2 < 0x5e ? 0 : 1) + 0x21); |
| 108 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
| 109 | |
| 110 | TRYMAP_DEC(jisx0208, **outbuf, c, c2); |
| 111 | else return 2; |
| 112 | } else if (c >= 0xf0 && c <= 0xf9) { |
| 113 | if ((c2 >= 0x40 && c2 <= 0x7e) || (c2 >= 0x80 && c2 <= 0xfc)) |
| 114 | OUT1(0xe000 + 188 * (c - 0xf0) + |
| 115 | (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41)) |
| 116 | else |
| 117 | return 2; |
| 118 | } else |
| 119 | return 2; |
| 120 | |
| 121 | NEXT(2, 1) |
| 122 | } |
| 123 | |
| 124 | return 0; |
| 125 | } |
| 126 | |
| 127 | #include "codecentry.h" |
| 128 | BEGIN_CODEC_REGISTRY(cp932) |
| 129 | MAPOPEN(ja_JP) |
| 130 | IMPORTMAP_DEC(jisx0208) |
| 131 | IMPORTMAP_ENCDEC(cp932ext) |
| 132 | IMPORTMAP_ENC(jisxcommon) |
| 133 | MAPCLOSE() |
| 134 | END_CODEC_REGISTRY(cp932) |