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