Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 1 | /* |
| 2 | * _codecs_jp.c: Codecs collection for Japanese encodings |
| 3 | * |
| 4 | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 5 | */ |
| 6 | |
| 7 | #define USING_BINARY_PAIR_SEARCH |
| 8 | #define EMPBASE 0x20000 |
| 9 | |
| 10 | #include "cjkcodecs.h" |
| 11 | #include "mappings_jp.h" |
| 12 | #include "mappings_jisx0213_pair.h" |
| 13 | #include "alg_jisx0201.h" |
| 14 | #include "emu_jisx0213_2000.h" |
| 15 | |
| 16 | /* |
| 17 | * CP932 codec |
| 18 | */ |
| 19 | |
| 20 | ENCODER(cp932) |
| 21 | { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 22 | while (*inpos < inlen) { |
| 23 | Py_UCS4 c = INCHAR1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 24 | DBCHAR code; |
| 25 | unsigned char c1, c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 26 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 27 | if (c <= 0x80) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 28 | WRITEBYTE1((unsigned char)c) |
| 29 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 30 | continue; |
| 31 | } |
| 32 | else if (c >= 0xff61 && c <= 0xff9f) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 33 | WRITEBYTE1(c - 0xfec0) |
| 34 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 35 | continue; |
| 36 | } |
| 37 | else if (c >= 0xf8f0 && c <= 0xf8f3) { |
| 38 | /* Windows compatibility */ |
| 39 | REQUIRE_OUTBUF(1) |
| 40 | if (c == 0xf8f0) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 41 | OUTBYTE1(0xa0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 42 | else |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 43 | OUTBYTE1(c - 0xfef1 + 0xfd) |
| 44 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 45 | continue; |
| 46 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 47 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 48 | if (c > 0xFFFF) |
| 49 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 50 | REQUIRE_OUTBUF(2) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 51 | |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 52 | if (TRYMAP_ENC(cp932ext, code, c)) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 53 | OUTBYTE1(code >> 8) |
| 54 | OUTBYTE2(code & 0xff) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 55 | } |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 56 | else if (TRYMAP_ENC(jisxcommon, code, c)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 57 | if (code & 0x8000) /* MSB set: JIS X 0212 */ |
| 58 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 59 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 60 | /* JIS X 0208 */ |
| 61 | c1 = code >> 8; |
| 62 | c2 = code & 0xff; |
| 63 | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); |
| 64 | c1 = (c1 - 0x21) >> 1; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 65 | OUTBYTE1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1) |
| 66 | OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 67 | } |
| 68 | else if (c >= 0xe000 && c < 0xe758) { |
| 69 | /* User-defined area */ |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 70 | c1 = (Py_UCS4)(c - 0xe000) / 188; |
| 71 | c2 = (Py_UCS4)(c - 0xe000) % 188; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 72 | OUTBYTE1(c1 + 0xf0) |
| 73 | OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 74 | } |
| 75 | else |
| 76 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 77 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 78 | NEXT(1, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 79 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 80 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 81 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 82 | } |
| 83 | |
| 84 | DECODER(cp932) |
| 85 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 86 | while (inleft > 0) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 87 | unsigned char c = INBYTE1, c2; |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 88 | Py_UCS4 decoded; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 89 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 90 | if (c <= 0x80) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 91 | OUTCHAR(c); |
| 92 | NEXT_IN(1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 93 | continue; |
| 94 | } |
| 95 | else if (c >= 0xa0 && c <= 0xdf) { |
| 96 | if (c == 0xa0) |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 97 | OUTCHAR(0xf8f0); /* half-width katakana */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 98 | else |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 99 | OUTCHAR(0xfec0 + c); |
| 100 | NEXT_IN(1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 101 | continue; |
| 102 | } |
| 103 | else if (c >= 0xfd/* && c <= 0xff*/) { |
| 104 | /* Windows compatibility */ |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 105 | OUTCHAR(0xf8f1 - 0xfd + c); |
| 106 | NEXT_IN(1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 107 | continue; |
| 108 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 109 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 110 | REQUIRE_INBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 111 | c2 = INBYTE2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 112 | |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 113 | if (TRYMAP_DEC(cp932ext, decoded, c, c2)) |
| 114 | OUTCHAR(decoded); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 115 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){ |
| 116 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 117 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 118 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 119 | c = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
| 120 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
| 121 | c = (2 * c + (c2 < 0x5e ? 0 : 1) + 0x21); |
| 122 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 123 | |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 124 | if (TRYMAP_DEC(jisx0208, decoded, c, c2)) |
| 125 | OUTCHAR(decoded); |
| 126 | else |
| 127 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 128 | } |
| 129 | else if (c >= 0xf0 && c <= 0xf9) { |
| 130 | if ((c2 >= 0x40 && c2 <= 0x7e) || |
| 131 | (c2 >= 0x80 && c2 <= 0xfc)) |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 132 | OUTCHAR(0xe000 + 188 * (c - 0xf0) + |
| 133 | (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41)); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 134 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 135 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 136 | } |
| 137 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 138 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 139 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 140 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 141 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 142 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 143 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 144 | } |
| 145 | |
| 146 | |
| 147 | /* |
| 148 | * EUC-JIS-2004 codec |
| 149 | */ |
| 150 | |
| 151 | ENCODER(euc_jis_2004) |
| 152 | { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 153 | while (*inpos < inlen) { |
| 154 | Py_UCS4 c = INCHAR1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 155 | DBCHAR code; |
| 156 | Py_ssize_t insize; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 157 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 158 | if (c < 0x80) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 159 | WRITEBYTE1(c) |
| 160 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 161 | continue; |
| 162 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 163 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 164 | insize = 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 165 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 166 | if (c <= 0xFFFF) { |
| 167 | EMULATE_JISX0213_2000_ENCODE_BMP(code, c) |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 168 | else if (TRYMAP_ENC(jisx0213_bmp, code, c)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 169 | if (code == MULTIC) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 170 | if (inlen - *inpos < 2) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 171 | if (flags & MBENC_FLUSH) { |
| 172 | code = find_pairencmap( |
| 173 | (ucs2_t)c, 0, |
| 174 | jisx0213_pair_encmap, |
| 175 | JISX0213_ENCPAIRS); |
| 176 | if (code == DBCINV) |
| 177 | return 1; |
| 178 | } |
| 179 | else |
| 180 | return MBERR_TOOFEW; |
| 181 | } |
| 182 | else { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 183 | Py_UCS4 c2 = INCHAR2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 184 | code = find_pairencmap( |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 185 | (ucs2_t)c, c2, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 186 | jisx0213_pair_encmap, |
| 187 | JISX0213_ENCPAIRS); |
| 188 | if (code == DBCINV) { |
| 189 | code = find_pairencmap( |
| 190 | (ucs2_t)c, 0, |
| 191 | jisx0213_pair_encmap, |
| 192 | JISX0213_ENCPAIRS); |
| 193 | if (code == DBCINV) |
| 194 | return 1; |
| 195 | } else |
| 196 | insize = 2; |
| 197 | } |
| 198 | } |
| 199 | } |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 200 | else if (TRYMAP_ENC(jisxcommon, code, c)); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 201 | else if (c >= 0xff61 && c <= 0xff9f) { |
| 202 | /* JIS X 0201 half-width katakana */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 203 | WRITEBYTE2(0x8e, c - 0xfec0) |
| 204 | NEXT(1, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 205 | continue; |
| 206 | } |
| 207 | else if (c == 0xff3c) |
| 208 | /* F/W REVERSE SOLIDUS (see NOTES) */ |
| 209 | code = 0x2140; |
| 210 | else if (c == 0xff5e) |
| 211 | /* F/W TILDE (see NOTES) */ |
| 212 | code = 0x2232; |
| 213 | else |
| 214 | return 1; |
| 215 | } |
| 216 | else if (c >> 16 == EMPBASE >> 16) { |
| 217 | EMULATE_JISX0213_2000_ENCODE_EMP(code, c) |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 218 | else if (TRYMAP_ENC(jisx0213_emp, code, c & 0xffff)); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 219 | else return insize; |
| 220 | } |
| 221 | else |
| 222 | return insize; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 223 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 224 | if (code & 0x8000) { |
| 225 | /* Codeset 2 */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 226 | WRITEBYTE3(0x8f, code >> 8, (code & 0xFF) | 0x80) |
| 227 | NEXT(insize, 3); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 228 | } else { |
| 229 | /* Codeset 1 */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 230 | WRITEBYTE2((code >> 8) | 0x80, (code & 0xFF) | 0x80) |
| 231 | NEXT(insize, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 232 | } |
| 233 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 234 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 235 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 236 | } |
| 237 | |
| 238 | DECODER(euc_jis_2004) |
| 239 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 240 | while (inleft > 0) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 241 | unsigned char c = INBYTE1; |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 242 | Py_UCS4 code, decoded; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 243 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 244 | if (c < 0x80) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 245 | OUTCHAR(c); |
| 246 | NEXT_IN(1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 247 | continue; |
| 248 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 249 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 250 | if (c == 0x8e) { |
| 251 | /* JIS X 0201 half-width katakana */ |
| 252 | unsigned char c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 253 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 254 | REQUIRE_INBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 255 | c2 = INBYTE2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 256 | if (c2 >= 0xa1 && c2 <= 0xdf) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 257 | OUTCHAR(0xfec0 + c2); |
| 258 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 259 | } |
| 260 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 261 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 262 | } |
| 263 | else if (c == 0x8f) { |
| 264 | unsigned char c2, c3; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 265 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 266 | REQUIRE_INBUF(3) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 267 | c2 = INBYTE2 ^ 0x80; |
| 268 | c3 = INBYTE3 ^ 0x80; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 269 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 270 | /* JIS X 0213 Plane 2 or JIS X 0212 (see NOTES) */ |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 271 | EMULATE_JISX0213_2000_DECODE_PLANE2(writer, c2, c3) |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 272 | else if (TRYMAP_DEC(jisx0213_2_bmp, decoded, c2, c3)) |
| 273 | OUTCHAR(decoded); |
| 274 | else if (TRYMAP_DEC(jisx0213_2_emp, code, c2, c3)) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 275 | OUTCHAR(EMPBASE | code); |
| 276 | NEXT_IN(3); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 277 | continue; |
| 278 | } |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 279 | else if (TRYMAP_DEC(jisx0212, decoded, c2, c3)) |
| 280 | OUTCHAR(decoded); |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 281 | else return 1; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 282 | NEXT_IN(3); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 283 | } |
| 284 | else { |
| 285 | unsigned char c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 286 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 287 | REQUIRE_INBUF(2) |
| 288 | c ^= 0x80; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 289 | c2 = INBYTE2 ^ 0x80; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 290 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 291 | /* JIS X 0213 Plane 1 */ |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 292 | EMULATE_JISX0213_2000_DECODE_PLANE1(writer, c, c2) |
| 293 | else if (c == 0x21 && c2 == 0x40) OUTCHAR(0xff3c); |
| 294 | else if (c == 0x22 && c2 == 0x32) OUTCHAR(0xff5e); |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 295 | else if (TRYMAP_DEC(jisx0208, decoded, c, c2)) |
| 296 | OUTCHAR(decoded); |
| 297 | else if (TRYMAP_DEC(jisx0213_1_bmp, decoded, c, c2)) |
| 298 | OUTCHAR(decoded); |
| 299 | else if (TRYMAP_DEC(jisx0213_1_emp, code, c, c2)) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 300 | OUTCHAR(EMPBASE | code); |
| 301 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 302 | continue; |
| 303 | } |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 304 | else if (TRYMAP_DEC(jisx0213_pair, code, c, c2)) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 305 | OUTCHAR2(code >> 16, code & 0xffff); |
| 306 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 307 | continue; |
| 308 | } |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 309 | else return 1; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 310 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 311 | } |
| 312 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 313 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 314 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 315 | } |
| 316 | |
| 317 | |
| 318 | /* |
| 319 | * EUC-JP codec |
| 320 | */ |
| 321 | |
| 322 | ENCODER(euc_jp) |
| 323 | { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 324 | while (*inpos < inlen) { |
| 325 | Py_UCS4 c = INCHAR1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 326 | DBCHAR code; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 327 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 328 | if (c < 0x80) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 329 | WRITEBYTE1((unsigned char)c) |
| 330 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 331 | continue; |
| 332 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 333 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 334 | if (c > 0xFFFF) |
| 335 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 336 | |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 337 | if (TRYMAP_ENC(jisxcommon, code, c)); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 338 | else if (c >= 0xff61 && c <= 0xff9f) { |
| 339 | /* JIS X 0201 half-width katakana */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 340 | WRITEBYTE2(0x8e, c - 0xfec0) |
| 341 | NEXT(1, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 342 | continue; |
| 343 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 344 | #ifndef STRICT_BUILD |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 345 | else if (c == 0xff3c) /* FULL-WIDTH REVERSE SOLIDUS */ |
| 346 | code = 0x2140; |
| 347 | else if (c == 0xa5) { /* YEN SIGN */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 348 | WRITEBYTE1(0x5c); |
| 349 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 350 | continue; |
| 351 | } else if (c == 0x203e) { /* OVERLINE */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 352 | WRITEBYTE1(0x7e); |
| 353 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 354 | continue; |
| 355 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 356 | #endif |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 357 | else |
| 358 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 359 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 360 | if (code & 0x8000) { |
| 361 | /* JIS X 0212 */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 362 | WRITEBYTE3(0x8f, code >> 8, (code & 0xFF) | 0x80) |
| 363 | NEXT(1, 3); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 364 | } else { |
| 365 | /* JIS X 0208 */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 366 | WRITEBYTE2((code >> 8) | 0x80, (code & 0xFF) | 0x80) |
| 367 | NEXT(1, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 368 | } |
| 369 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 370 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 371 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 372 | } |
| 373 | |
| 374 | DECODER(euc_jp) |
| 375 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 376 | while (inleft > 0) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 377 | unsigned char c = INBYTE1; |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 378 | Py_UCS4 decoded; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 379 | |
Victor Stinner | 5dfe3bb | 2011-06-03 23:34:09 +0200 | [diff] [blame] | 380 | if (c < 0x80) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 381 | OUTCHAR(c); |
| 382 | NEXT_IN(1); |
Victor Stinner | 5dfe3bb | 2011-06-03 23:34:09 +0200 | [diff] [blame] | 383 | continue; |
| 384 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 385 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 386 | if (c == 0x8e) { |
| 387 | /* JIS X 0201 half-width katakana */ |
| 388 | unsigned char c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 389 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 390 | REQUIRE_INBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 391 | c2 = INBYTE2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 392 | if (c2 >= 0xa1 && c2 <= 0xdf) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 393 | OUTCHAR(0xfec0 + c2); |
| 394 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 395 | } |
| 396 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 397 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 398 | } |
| 399 | else if (c == 0x8f) { |
| 400 | unsigned char c2, c3; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 401 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 402 | REQUIRE_INBUF(3) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 403 | c2 = INBYTE2; |
| 404 | c3 = INBYTE3; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 405 | /* JIS X 0212 */ |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 406 | if (TRYMAP_DEC(jisx0212, decoded, c2 ^ 0x80, c3 ^ 0x80)) { |
| 407 | OUTCHAR(decoded); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 408 | NEXT_IN(3); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 409 | } |
| 410 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 411 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 412 | } |
| 413 | else { |
| 414 | unsigned char c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 415 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 416 | REQUIRE_INBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 417 | c2 = INBYTE2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 418 | /* JIS X 0208 */ |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 419 | #ifndef STRICT_BUILD |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 420 | if (c == 0xa1 && c2 == 0xc0) |
| 421 | /* FULL-WIDTH REVERSE SOLIDUS */ |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 422 | OUTCHAR(0xff3c); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 423 | else |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 424 | #endif |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 425 | if (TRYMAP_DEC(jisx0208, decoded, c ^ 0x80, c2 ^ 0x80)) |
| 426 | OUTCHAR(decoded); |
| 427 | else |
| 428 | return 1; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 429 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 430 | } |
| 431 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 432 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 433 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 434 | } |
| 435 | |
| 436 | |
| 437 | /* |
| 438 | * SHIFT_JIS codec |
| 439 | */ |
| 440 | |
| 441 | ENCODER(shift_jis) |
| 442 | { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 443 | while (*inpos < inlen) { |
| 444 | Py_UCS4 c = INCHAR1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 445 | DBCHAR code; |
| 446 | unsigned char c1, c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 447 | |
| 448 | #ifdef STRICT_BUILD |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 449 | JISX0201_R_ENCODE(c, code) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 450 | #else |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 451 | if (c < 0x80) code = c; |
| 452 | else if (c == 0x00a5) code = 0x5c; /* YEN SIGN */ |
| 453 | else if (c == 0x203e) code = 0x7e; /* OVERLINE */ |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 454 | #endif |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 455 | else JISX0201_K_ENCODE(c, code) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 456 | else if (c > 0xFFFF) |
| 457 | return 1; |
| 458 | else |
| 459 | code = NOCHAR; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 460 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 461 | if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) { |
| 462 | REQUIRE_OUTBUF(1) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 463 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 464 | OUTBYTE1((unsigned char)code) |
| 465 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 466 | continue; |
| 467 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 468 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 469 | REQUIRE_OUTBUF(2) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 470 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 471 | if (code == NOCHAR) { |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 472 | if (TRYMAP_ENC(jisxcommon, code, c)); |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 473 | #ifndef STRICT_BUILD |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 474 | else if (c == 0xff3c) |
| 475 | code = 0x2140; /* FULL-WIDTH REVERSE SOLIDUS */ |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 476 | #endif |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 477 | else |
| 478 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 479 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 480 | if (code & 0x8000) /* MSB set: JIS X 0212 */ |
| 481 | return 1; |
| 482 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 483 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 484 | c1 = code >> 8; |
| 485 | c2 = code & 0xff; |
| 486 | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); |
| 487 | c1 = (c1 - 0x21) >> 1; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 488 | OUTBYTE1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1) |
| 489 | OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41) |
| 490 | NEXT(1, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 491 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 492 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 493 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 494 | } |
| 495 | |
| 496 | DECODER(shift_jis) |
| 497 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 498 | while (inleft > 0) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 499 | unsigned char c = INBYTE1; |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 500 | Py_UCS4 decoded; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 501 | |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 502 | #ifdef STRICT_BUILD |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 503 | JISX0201_R_DECODE(c, writer) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 504 | #else |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 505 | if (c < 0x80) OUTCHAR(c); |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 506 | #endif |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 507 | else JISX0201_K_DECODE(c, writer) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 508 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){ |
| 509 | unsigned char c1, c2; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 510 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 511 | REQUIRE_INBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 512 | c2 = INBYTE2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 513 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 514 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 515 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 516 | c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
| 517 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
| 518 | c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1) + 0x21); |
| 519 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 520 | |
| 521 | #ifndef STRICT_BUILD |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 522 | if (c1 == 0x21 && c2 == 0x40) { |
| 523 | /* FULL-WIDTH REVERSE SOLIDUS */ |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 524 | OUTCHAR(0xff3c); |
| 525 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 526 | continue; |
| 527 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 528 | #endif |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 529 | if (TRYMAP_DEC(jisx0208, decoded, c1, c2)) { |
| 530 | OUTCHAR(decoded); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 531 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 532 | continue; |
| 533 | } |
| 534 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 535 | return 1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 536 | } |
| 537 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 538 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 539 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 540 | NEXT_IN(1); /* JIS X 0201 */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 541 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 542 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 543 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 544 | } |
| 545 | |
| 546 | |
| 547 | /* |
| 548 | * SHIFT_JIS-2004 codec |
| 549 | */ |
| 550 | |
| 551 | ENCODER(shift_jis_2004) |
| 552 | { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 553 | while (*inpos < inlen) { |
| 554 | Py_UCS4 c = INCHAR1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 555 | DBCHAR code = NOCHAR; |
| 556 | int c1, c2; |
| 557 | Py_ssize_t insize; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 558 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 559 | JISX0201_ENCODE(c, code) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 560 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 561 | if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 562 | WRITEBYTE1((unsigned char)code) |
| 563 | NEXT(1, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 564 | continue; |
| 565 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 566 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 567 | REQUIRE_OUTBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 568 | insize = 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 569 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 570 | if (code == NOCHAR) { |
| 571 | if (c <= 0xffff) { |
| 572 | EMULATE_JISX0213_2000_ENCODE_BMP(code, c) |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 573 | else if (TRYMAP_ENC(jisx0213_bmp, code, c)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 574 | if (code == MULTIC) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 575 | if (inlen - *inpos < 2) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 576 | if (flags & MBENC_FLUSH) { |
| 577 | code = find_pairencmap |
| 578 | ((ucs2_t)c, 0, |
| 579 | jisx0213_pair_encmap, |
| 580 | JISX0213_ENCPAIRS); |
| 581 | if (code == DBCINV) |
| 582 | return 1; |
| 583 | } |
| 584 | else |
| 585 | return MBERR_TOOFEW; |
| 586 | } |
| 587 | else { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 588 | Py_UCS4 ch2 = INCHAR2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 589 | code = find_pairencmap( |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 590 | (ucs2_t)c, ch2, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 591 | jisx0213_pair_encmap, |
| 592 | JISX0213_ENCPAIRS); |
| 593 | if (code == DBCINV) { |
| 594 | code = find_pairencmap( |
| 595 | (ucs2_t)c, 0, |
| 596 | jisx0213_pair_encmap, |
| 597 | JISX0213_ENCPAIRS); |
| 598 | if (code == DBCINV) |
| 599 | return 1; |
| 600 | } |
| 601 | else |
| 602 | insize = 2; |
| 603 | } |
| 604 | } |
| 605 | } |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 606 | else if (TRYMAP_ENC(jisxcommon, code, c)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 607 | /* abandon JIS X 0212 codes */ |
| 608 | if (code & 0x8000) |
| 609 | return 1; |
| 610 | } |
| 611 | else return 1; |
| 612 | } |
| 613 | else if (c >> 16 == EMPBASE >> 16) { |
| 614 | EMULATE_JISX0213_2000_ENCODE_EMP(code, c) |
Victor Stinner | bd97ac3 | 2013-10-28 23:54:13 +0100 | [diff] [blame^] | 615 | else if (TRYMAP_ENC(jisx0213_emp, code, c&0xffff)); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 616 | else return insize; |
| 617 | } |
| 618 | else |
| 619 | return insize; |
| 620 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 621 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 622 | c1 = code >> 8; |
| 623 | c2 = (code & 0xff) - 0x21; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 624 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 625 | if (c1 & 0x80) { /* Plane 2 */ |
| 626 | if (c1 >= 0xee) c1 -= 0x87; |
| 627 | else if (c1 >= 0xac || c1 == 0xa8) c1 -= 0x49; |
| 628 | else c1 -= 0x43; |
| 629 | } |
| 630 | else /* Plane 1 */ |
| 631 | c1 -= 0x21; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 632 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 633 | if (c1 & 1) c2 += 0x5e; |
| 634 | c1 >>= 1; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 635 | OUTBYTE1(c1 + (c1 < 0x1f ? 0x81 : 0xc1)) |
| 636 | OUTBYTE2(c2 + (c2 < 0x3f ? 0x40 : 0x41)) |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 637 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 638 | NEXT(insize, 2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 639 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 640 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 641 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 642 | } |
| 643 | |
| 644 | DECODER(shift_jis_2004) |
| 645 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 646 | while (inleft > 0) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 647 | unsigned char c = INBYTE1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 648 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 649 | JISX0201_DECODE(c, writer) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 650 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc)){ |
| 651 | unsigned char c1, c2; |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 652 | Py_UCS4 code, decoded; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 653 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 654 | REQUIRE_INBUF(2) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 655 | c2 = INBYTE2; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 656 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 657 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 658 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 659 | c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
| 660 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
| 661 | c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1)); |
| 662 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 663 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 664 | if (c1 < 0x5e) { /* Plane 1 */ |
| 665 | c1 += 0x21; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 666 | EMULATE_JISX0213_2000_DECODE_PLANE1(writer, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 667 | c1, c2) |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 668 | else if (TRYMAP_DEC(jisx0208, decoded, c1, c2)) |
| 669 | OUTCHAR(decoded); |
| 670 | else if (TRYMAP_DEC(jisx0213_1_bmp, decoded, c1, c2)) |
| 671 | OUTCHAR(decoded); |
| 672 | else if (TRYMAP_DEC(jisx0213_1_emp, code, c1, c2)) |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 673 | OUTCHAR(EMPBASE | code); |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 674 | else if (TRYMAP_DEC(jisx0213_pair, code, c1, c2)) |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 675 | OUTCHAR2(code >> 16, code & 0xffff); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 676 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 677 | return 1; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 678 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 679 | } |
| 680 | else { /* Plane 2 */ |
| 681 | if (c1 >= 0x67) c1 += 0x07; |
| 682 | else if (c1 >= 0x63 || c1 == 0x5f) c1 -= 0x37; |
| 683 | else c1 -= 0x3d; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 684 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 685 | EMULATE_JISX0213_2000_DECODE_PLANE2(writer, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 686 | c1, c2) |
Victor Stinner | 11bdf91 | 2013-10-28 23:18:39 +0100 | [diff] [blame] | 687 | else if (TRYMAP_DEC(jisx0213_2_bmp, decoded, c1, c2)) |
| 688 | OUTCHAR(decoded); |
| 689 | else if (TRYMAP_DEC(jisx0213_2_emp, code, c1, c2)) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 690 | OUTCHAR(EMPBASE | code); |
| 691 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 692 | continue; |
| 693 | } |
| 694 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 695 | return 1; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 696 | NEXT_IN(2); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 697 | } |
| 698 | continue; |
| 699 | } |
| 700 | else |
Victor Stinner | 2cded9c | 2011-07-08 01:45:13 +0200 | [diff] [blame] | 701 | return 1; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 702 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 703 | NEXT_IN(1); /* JIS X 0201 */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 704 | } |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 705 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 706 | return 0; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 707 | } |
| 708 | |
| 709 | |
| 710 | BEGIN_MAPPINGS_LIST |
| 711 | MAPPING_DECONLY(jisx0208) |
| 712 | MAPPING_DECONLY(jisx0212) |
| 713 | MAPPING_ENCONLY(jisxcommon) |
| 714 | MAPPING_DECONLY(jisx0213_1_bmp) |
| 715 | MAPPING_DECONLY(jisx0213_2_bmp) |
| 716 | MAPPING_ENCONLY(jisx0213_bmp) |
| 717 | MAPPING_DECONLY(jisx0213_1_emp) |
| 718 | MAPPING_DECONLY(jisx0213_2_emp) |
| 719 | MAPPING_ENCONLY(jisx0213_emp) |
| 720 | MAPPING_ENCDEC(jisx0213_pair) |
| 721 | MAPPING_ENCDEC(cp932ext) |
| 722 | END_MAPPINGS_LIST |
| 723 | |
| 724 | BEGIN_CODECS_LIST |
| 725 | CODEC_STATELESS(shift_jis) |
| 726 | CODEC_STATELESS(cp932) |
| 727 | CODEC_STATELESS(euc_jp) |
| 728 | CODEC_STATELESS(shift_jis_2004) |
| 729 | CODEC_STATELESS(euc_jis_2004) |
| 730 | { "euc_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(euc_jis_2004) }, |
| 731 | { "shift_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(shift_jis_2004) }, |
| 732 | END_CODECS_LIST |
| 733 | |
| 734 | I_AM_A_MODULE_FOR(jp) |