Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1 | /* |
| 2 | * _gb18030.c: the GB18030 codec |
| 3 | * |
| 4 | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
| 5 | * $CJKCodecs: _gb18030.c,v 1.2 2003/12/31 05:46:55 perky Exp $ |
| 6 | */ |
| 7 | |
| 8 | #include "codeccommon.h" |
| 9 | #include "tweak_gbk.h" |
| 10 | #include "map_gb18030uni.h" |
| 11 | |
| 12 | ENCMAP(gbcommon) |
| 13 | ENCMAP(gb18030ext) |
| 14 | DECMAP(gb2312) |
| 15 | DECMAP(gbkext) |
| 16 | DECMAP(gb18030ext) |
| 17 | |
| 18 | ENCODER(gb18030) |
| 19 | { |
| 20 | while (inleft > 0) { |
| 21 | ucs4_t c = IN1; |
| 22 | DBCHAR code; |
| 23 | |
| 24 | if (c < 0x80) { |
| 25 | WRITE1(c) |
| 26 | NEXT(1, 1) |
| 27 | continue; |
| 28 | } |
| 29 | |
| 30 | DECODE_SURROGATE(c) |
| 31 | if (c > 0x10FFFF) |
| 32 | #if Py_UNICODE_SIZE == 2 |
| 33 | return 2; /* surrogates pair */ |
| 34 | #else |
| 35 | return 1; |
| 36 | #endif |
| 37 | else if (c >= 0x10000) { |
| 38 | ucs4_t tc = c - 0x10000; |
| 39 | |
| 40 | RESERVE_OUTBUF(4) |
| 41 | |
| 42 | OUT4((unsigned char)(tc % 10) + 0x30) |
| 43 | tc /= 10; |
| 44 | OUT3((unsigned char)(tc % 126) + 0x81) |
| 45 | tc /= 126; |
| 46 | OUT2((unsigned char)(tc % 10) + 0x30) |
| 47 | tc /= 10; |
| 48 | OUT1((unsigned char)(tc + 0x90)) |
| 49 | |
| 50 | #if Py_UNICODE_SIZE == 2 |
| 51 | NEXT(2, 4) /* surrogates pair */ |
| 52 | #else |
| 53 | NEXT(1, 4) |
| 54 | #endif |
| 55 | continue; |
| 56 | } |
| 57 | |
| 58 | RESERVE_OUTBUF(2) |
| 59 | |
| 60 | GBK_PREENCODE(c, code) |
| 61 | else TRYMAP_ENC(gbcommon, code, c); |
| 62 | else TRYMAP_ENC(gb18030ext, code, c); |
| 63 | else { |
| 64 | const struct _gb18030_to_unibmp_ranges *utrrange; |
| 65 | |
| 66 | RESERVE_OUTBUF(4) |
| 67 | |
| 68 | for (utrrange = gb18030_to_unibmp_ranges; |
| 69 | utrrange->first != 0; |
| 70 | utrrange++) |
| 71 | if (utrrange->first <= c && c <= utrrange->last) { |
| 72 | Py_UNICODE tc; |
| 73 | |
| 74 | tc = c - utrrange->first + utrrange->base; |
| 75 | |
| 76 | OUT4((unsigned char)(tc % 10) + 0x30) |
| 77 | tc /= 10; |
| 78 | OUT3((unsigned char)(tc % 126) + 0x81) |
| 79 | tc /= 126; |
| 80 | OUT2((unsigned char)(tc % 10) + 0x30) |
| 81 | tc /= 10; |
| 82 | OUT1((unsigned char)tc + 0x81) |
| 83 | |
| 84 | NEXT(1, 4) |
| 85 | break; |
| 86 | } |
| 87 | |
| 88 | if (utrrange->first == 0) { |
| 89 | PyErr_SetString(PyExc_RuntimeError, |
| 90 | "unicode mapping invalid"); |
| 91 | return 1; |
| 92 | } |
| 93 | continue; |
| 94 | } |
| 95 | |
| 96 | OUT1((code >> 8) | 0x80) |
| 97 | if (code & 0x8000) |
| 98 | OUT2((code & 0xFF)) /* MSB set: GBK or GB18030ext */ |
| 99 | else |
| 100 | OUT2((code & 0xFF) | 0x80) /* MSB unset: GB2312 */ |
| 101 | |
| 102 | NEXT(1, 2) |
| 103 | } |
| 104 | |
| 105 | return 0; |
| 106 | } |
| 107 | |
| 108 | DECODER(gb18030) |
| 109 | { |
| 110 | while (inleft > 0) { |
| 111 | unsigned char c = IN1, c2; |
| 112 | |
| 113 | RESERVE_OUTBUF(1) |
| 114 | |
| 115 | if (c < 0x80) { |
| 116 | OUT1(c) |
| 117 | NEXT(1, 1) |
| 118 | continue; |
| 119 | } |
| 120 | |
| 121 | RESERVE_INBUF(2) |
| 122 | |
| 123 | c2 = IN2; |
| 124 | if (c2 >= 0x30 && c2 <= 0x39) { /* 4 bytes seq */ |
| 125 | const struct _gb18030_to_unibmp_ranges *utr; |
| 126 | unsigned char c3, c4; |
| 127 | ucs4_t lseq; |
| 128 | |
| 129 | RESERVE_INBUF(4) |
| 130 | c3 = IN3; |
| 131 | c4 = IN4; |
| 132 | if (c < 0x81 || c3 < 0x81 || c4 < 0x30 || c4 > 0x39) |
| 133 | return 4; |
| 134 | c -= 0x81; c2 -= 0x30; |
| 135 | c3 -= 0x81; c4 -= 0x30; |
| 136 | |
| 137 | if (c < 4) { /* U+0080 - U+FFFF */ |
| 138 | lseq = ((ucs4_t)c * 10 + c2) * 1260 + |
| 139 | (ucs4_t)c3 * 10 + c4; |
| 140 | if (lseq < 39420) { |
| 141 | for (utr = gb18030_to_unibmp_ranges; |
| 142 | lseq >= (utr + 1)->base; |
| 143 | utr++) ; |
| 144 | OUT1(utr->first - utr->base + lseq) |
| 145 | NEXT(4, 1) |
| 146 | continue; |
| 147 | } |
| 148 | } |
| 149 | else if (c >= 15) { /* U+10000 - U+10FFFF */ |
| 150 | lseq = 0x10000 + (((ucs4_t)c-15) * 10 + c2) * 1260 + |
| 151 | (ucs4_t)c3 * 10 + c4; |
| 152 | if (lseq <= 0x10FFFF) { |
| 153 | PUTUCS4(lseq); |
| 154 | NEXT_IN(4) |
| 155 | continue; |
| 156 | } |
| 157 | } |
| 158 | return 4; |
| 159 | } |
| 160 | |
| 161 | GBK_PREDECODE(c, c2, **outbuf) |
| 162 | else TRYMAP_DEC(gb2312, **outbuf, c ^ 0x80, c2 ^ 0x80); |
| 163 | else TRYMAP_DEC(gbkext, **outbuf, c, c2); |
| 164 | else TRYMAP_DEC(gb18030ext, **outbuf, c, c2); |
| 165 | else return 2; |
| 166 | |
| 167 | NEXT(2, 1) |
| 168 | } |
| 169 | |
| 170 | return 0; |
| 171 | } |
| 172 | |
| 173 | #include "codecentry.h" |
| 174 | BEGIN_CODEC_REGISTRY(gb18030) |
| 175 | MAPOPEN(zh_CN) |
| 176 | IMPORTMAP_DEC(gb2312) |
| 177 | IMPORTMAP_DEC(gbkext) |
| 178 | IMPORTMAP_ENC(gbcommon) |
| 179 | IMPORTMAP_ENCDEC(gb18030ext) |
| 180 | MAPCLOSE() |
| 181 | END_CODEC_REGISTRY(gb18030) |