| 1 | n/a | /* |
|---|
| 2 | n/a | * _codecs_jp.c: Codecs collection for Japanese encodings |
|---|
| 3 | n/a | * |
|---|
| 4 | n/a | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
|---|
| 5 | n/a | */ |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | #define USING_BINARY_PAIR_SEARCH |
|---|
| 8 | n/a | #define EMPBASE 0x20000 |
|---|
| 9 | n/a | |
|---|
| 10 | n/a | #include "cjkcodecs.h" |
|---|
| 11 | n/a | #include "mappings_jp.h" |
|---|
| 12 | n/a | #include "mappings_jisx0213_pair.h" |
|---|
| 13 | n/a | #include "alg_jisx0201.h" |
|---|
| 14 | n/a | #include "emu_jisx0213_2000.h" |
|---|
| 15 | n/a | |
|---|
| 16 | n/a | /* |
|---|
| 17 | n/a | * CP932 codec |
|---|
| 18 | n/a | */ |
|---|
| 19 | n/a | |
|---|
| 20 | n/a | ENCODER(cp932) |
|---|
| 21 | n/a | { |
|---|
| 22 | n/a | while (*inpos < inlen) { |
|---|
| 23 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 24 | n/a | DBCHAR code; |
|---|
| 25 | n/a | unsigned char c1, c2; |
|---|
| 26 | n/a | |
|---|
| 27 | n/a | if (c <= 0x80) { |
|---|
| 28 | n/a | WRITEBYTE1((unsigned char)c); |
|---|
| 29 | n/a | NEXT(1, 1); |
|---|
| 30 | n/a | continue; |
|---|
| 31 | n/a | } |
|---|
| 32 | n/a | else if (c >= 0xff61 && c <= 0xff9f) { |
|---|
| 33 | n/a | WRITEBYTE1(c - 0xfec0); |
|---|
| 34 | n/a | NEXT(1, 1); |
|---|
| 35 | n/a | continue; |
|---|
| 36 | n/a | } |
|---|
| 37 | n/a | else if (c >= 0xf8f0 && c <= 0xf8f3) { |
|---|
| 38 | n/a | /* Windows compatibility */ |
|---|
| 39 | n/a | REQUIRE_OUTBUF(1); |
|---|
| 40 | n/a | if (c == 0xf8f0) |
|---|
| 41 | n/a | OUTBYTE1(0xa0); |
|---|
| 42 | n/a | else |
|---|
| 43 | n/a | OUTBYTE1(c - 0xfef1 + 0xfd); |
|---|
| 44 | n/a | NEXT(1, 1); |
|---|
| 45 | n/a | continue; |
|---|
| 46 | n/a | } |
|---|
| 47 | n/a | |
|---|
| 48 | n/a | if (c > 0xFFFF) |
|---|
| 49 | n/a | return 1; |
|---|
| 50 | n/a | REQUIRE_OUTBUF(2); |
|---|
| 51 | n/a | |
|---|
| 52 | n/a | if (TRYMAP_ENC(cp932ext, code, c)) { |
|---|
| 53 | n/a | OUTBYTE1(code >> 8); |
|---|
| 54 | n/a | OUTBYTE2(code & 0xff); |
|---|
| 55 | n/a | } |
|---|
| 56 | n/a | else if (TRYMAP_ENC(jisxcommon, code, c)) { |
|---|
| 57 | n/a | if (code & 0x8000) /* MSB set: JIS X 0212 */ |
|---|
| 58 | n/a | return 1; |
|---|
| 59 | n/a | |
|---|
| 60 | n/a | /* JIS X 0208 */ |
|---|
| 61 | n/a | c1 = code >> 8; |
|---|
| 62 | n/a | c2 = code & 0xff; |
|---|
| 63 | n/a | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); |
|---|
| 64 | n/a | c1 = (c1 - 0x21) >> 1; |
|---|
| 65 | n/a | OUTBYTE1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1); |
|---|
| 66 | n/a | OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41); |
|---|
| 67 | n/a | } |
|---|
| 68 | n/a | else if (c >= 0xe000 && c < 0xe758) { |
|---|
| 69 | n/a | /* User-defined area */ |
|---|
| 70 | n/a | c1 = (Py_UCS4)(c - 0xe000) / 188; |
|---|
| 71 | n/a | c2 = (Py_UCS4)(c - 0xe000) % 188; |
|---|
| 72 | n/a | OUTBYTE1(c1 + 0xf0); |
|---|
| 73 | n/a | OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41); |
|---|
| 74 | n/a | } |
|---|
| 75 | n/a | else |
|---|
| 76 | n/a | return 1; |
|---|
| 77 | n/a | |
|---|
| 78 | n/a | NEXT(1, 2); |
|---|
| 79 | n/a | } |
|---|
| 80 | n/a | |
|---|
| 81 | n/a | return 0; |
|---|
| 82 | n/a | } |
|---|
| 83 | n/a | |
|---|
| 84 | n/a | DECODER(cp932) |
|---|
| 85 | n/a | { |
|---|
| 86 | n/a | while (inleft > 0) { |
|---|
| 87 | n/a | unsigned char c = INBYTE1, c2; |
|---|
| 88 | n/a | Py_UCS4 decoded; |
|---|
| 89 | n/a | |
|---|
| 90 | n/a | if (c <= 0x80) { |
|---|
| 91 | n/a | OUTCHAR(c); |
|---|
| 92 | n/a | NEXT_IN(1); |
|---|
| 93 | n/a | continue; |
|---|
| 94 | n/a | } |
|---|
| 95 | n/a | else if (c >= 0xa0 && c <= 0xdf) { |
|---|
| 96 | n/a | if (c == 0xa0) |
|---|
| 97 | n/a | OUTCHAR(0xf8f0); /* half-width katakana */ |
|---|
| 98 | n/a | else |
|---|
| 99 | n/a | OUTCHAR(0xfec0 + c); |
|---|
| 100 | n/a | NEXT_IN(1); |
|---|
| 101 | n/a | continue; |
|---|
| 102 | n/a | } |
|---|
| 103 | n/a | else if (c >= 0xfd/* && c <= 0xff*/) { |
|---|
| 104 | n/a | /* Windows compatibility */ |
|---|
| 105 | n/a | OUTCHAR(0xf8f1 - 0xfd + c); |
|---|
| 106 | n/a | NEXT_IN(1); |
|---|
| 107 | n/a | continue; |
|---|
| 108 | n/a | } |
|---|
| 109 | n/a | |
|---|
| 110 | n/a | REQUIRE_INBUF(2); |
|---|
| 111 | n/a | c2 = INBYTE2; |
|---|
| 112 | n/a | |
|---|
| 113 | n/a | if (TRYMAP_DEC(cp932ext, decoded, c, c2)) |
|---|
| 114 | n/a | OUTCHAR(decoded); |
|---|
| 115 | n/a | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){ |
|---|
| 116 | n/a | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
|---|
| 117 | n/a | return 1; |
|---|
| 118 | n/a | |
|---|
| 119 | n/a | c = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
|---|
| 120 | n/a | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
|---|
| 121 | n/a | c = (2 * c + (c2 < 0x5e ? 0 : 1) + 0x21); |
|---|
| 122 | n/a | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
|---|
| 123 | n/a | |
|---|
| 124 | n/a | if (TRYMAP_DEC(jisx0208, decoded, c, c2)) |
|---|
| 125 | n/a | OUTCHAR(decoded); |
|---|
| 126 | n/a | else |
|---|
| 127 | n/a | return 1; |
|---|
| 128 | n/a | } |
|---|
| 129 | n/a | else if (c >= 0xf0 && c <= 0xf9) { |
|---|
| 130 | n/a | if ((c2 >= 0x40 && c2 <= 0x7e) || |
|---|
| 131 | n/a | (c2 >= 0x80 && c2 <= 0xfc)) |
|---|
| 132 | n/a | OUTCHAR(0xe000 + 188 * (c - 0xf0) + |
|---|
| 133 | n/a | (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41)); |
|---|
| 134 | n/a | else |
|---|
| 135 | n/a | return 1; |
|---|
| 136 | n/a | } |
|---|
| 137 | n/a | else |
|---|
| 138 | n/a | return 1; |
|---|
| 139 | n/a | |
|---|
| 140 | n/a | NEXT_IN(2); |
|---|
| 141 | n/a | } |
|---|
| 142 | n/a | |
|---|
| 143 | n/a | return 0; |
|---|
| 144 | n/a | } |
|---|
| 145 | n/a | |
|---|
| 146 | n/a | |
|---|
| 147 | n/a | /* |
|---|
| 148 | n/a | * EUC-JIS-2004 codec |
|---|
| 149 | n/a | */ |
|---|
| 150 | n/a | |
|---|
| 151 | n/a | ENCODER(euc_jis_2004) |
|---|
| 152 | n/a | { |
|---|
| 153 | n/a | while (*inpos < inlen) { |
|---|
| 154 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 155 | n/a | DBCHAR code; |
|---|
| 156 | n/a | Py_ssize_t insize; |
|---|
| 157 | n/a | |
|---|
| 158 | n/a | if (c < 0x80) { |
|---|
| 159 | n/a | WRITEBYTE1(c); |
|---|
| 160 | n/a | NEXT(1, 1); |
|---|
| 161 | n/a | continue; |
|---|
| 162 | n/a | } |
|---|
| 163 | n/a | |
|---|
| 164 | n/a | insize = 1; |
|---|
| 165 | n/a | |
|---|
| 166 | n/a | if (c <= 0xFFFF) { |
|---|
| 167 | n/a | EMULATE_JISX0213_2000_ENCODE_BMP(code, c) |
|---|
| 168 | n/a | else if (TRYMAP_ENC(jisx0213_bmp, code, c)) { |
|---|
| 169 | n/a | if (code == MULTIC) { |
|---|
| 170 | n/a | if (inlen - *inpos < 2) { |
|---|
| 171 | n/a | if (flags & MBENC_FLUSH) { |
|---|
| 172 | n/a | code = find_pairencmap( |
|---|
| 173 | n/a | (ucs2_t)c, 0, |
|---|
| 174 | n/a | jisx0213_pair_encmap, |
|---|
| 175 | n/a | JISX0213_ENCPAIRS); |
|---|
| 176 | n/a | if (code == DBCINV) |
|---|
| 177 | n/a | return 1; |
|---|
| 178 | n/a | } |
|---|
| 179 | n/a | else |
|---|
| 180 | n/a | return MBERR_TOOFEW; |
|---|
| 181 | n/a | } |
|---|
| 182 | n/a | else { |
|---|
| 183 | n/a | Py_UCS4 c2 = INCHAR2; |
|---|
| 184 | n/a | code = find_pairencmap( |
|---|
| 185 | n/a | (ucs2_t)c, c2, |
|---|
| 186 | n/a | jisx0213_pair_encmap, |
|---|
| 187 | n/a | JISX0213_ENCPAIRS); |
|---|
| 188 | n/a | if (code == DBCINV) { |
|---|
| 189 | n/a | code = find_pairencmap( |
|---|
| 190 | n/a | (ucs2_t)c, 0, |
|---|
| 191 | n/a | jisx0213_pair_encmap, |
|---|
| 192 | n/a | JISX0213_ENCPAIRS); |
|---|
| 193 | n/a | if (code == DBCINV) |
|---|
| 194 | n/a | return 1; |
|---|
| 195 | n/a | } else |
|---|
| 196 | n/a | insize = 2; |
|---|
| 197 | n/a | } |
|---|
| 198 | n/a | } |
|---|
| 199 | n/a | } |
|---|
| 200 | n/a | else if (TRYMAP_ENC(jisxcommon, code, c)) |
|---|
| 201 | n/a | ; |
|---|
| 202 | n/a | else if (c >= 0xff61 && c <= 0xff9f) { |
|---|
| 203 | n/a | /* JIS X 0201 half-width katakana */ |
|---|
| 204 | n/a | WRITEBYTE2(0x8e, c - 0xfec0); |
|---|
| 205 | n/a | NEXT(1, 2); |
|---|
| 206 | n/a | continue; |
|---|
| 207 | n/a | } |
|---|
| 208 | n/a | else if (c == 0xff3c) |
|---|
| 209 | n/a | /* F/W REVERSE SOLIDUS (see NOTES) */ |
|---|
| 210 | n/a | code = 0x2140; |
|---|
| 211 | n/a | else if (c == 0xff5e) |
|---|
| 212 | n/a | /* F/W TILDE (see NOTES) */ |
|---|
| 213 | n/a | code = 0x2232; |
|---|
| 214 | n/a | else |
|---|
| 215 | n/a | return 1; |
|---|
| 216 | n/a | } |
|---|
| 217 | n/a | else if (c >> 16 == EMPBASE >> 16) { |
|---|
| 218 | n/a | EMULATE_JISX0213_2000_ENCODE_EMP(code, c) |
|---|
| 219 | n/a | else if (TRYMAP_ENC(jisx0213_emp, code, c & 0xffff)) |
|---|
| 220 | n/a | ; |
|---|
| 221 | n/a | else |
|---|
| 222 | n/a | return insize; |
|---|
| 223 | n/a | } |
|---|
| 224 | n/a | else |
|---|
| 225 | n/a | return insize; |
|---|
| 226 | n/a | |
|---|
| 227 | n/a | if (code & 0x8000) { |
|---|
| 228 | n/a | /* Codeset 2 */ |
|---|
| 229 | n/a | WRITEBYTE3(0x8f, code >> 8, (code & 0xFF) | 0x80); |
|---|
| 230 | n/a | NEXT(insize, 3); |
|---|
| 231 | n/a | } else { |
|---|
| 232 | n/a | /* Codeset 1 */ |
|---|
| 233 | n/a | WRITEBYTE2((code >> 8) | 0x80, (code & 0xFF) | 0x80); |
|---|
| 234 | n/a | NEXT(insize, 2); |
|---|
| 235 | n/a | } |
|---|
| 236 | n/a | } |
|---|
| 237 | n/a | |
|---|
| 238 | n/a | return 0; |
|---|
| 239 | n/a | } |
|---|
| 240 | n/a | |
|---|
| 241 | n/a | DECODER(euc_jis_2004) |
|---|
| 242 | n/a | { |
|---|
| 243 | n/a | while (inleft > 0) { |
|---|
| 244 | n/a | unsigned char c = INBYTE1; |
|---|
| 245 | n/a | Py_UCS4 code, decoded; |
|---|
| 246 | n/a | |
|---|
| 247 | n/a | if (c < 0x80) { |
|---|
| 248 | n/a | OUTCHAR(c); |
|---|
| 249 | n/a | NEXT_IN(1); |
|---|
| 250 | n/a | continue; |
|---|
| 251 | n/a | } |
|---|
| 252 | n/a | |
|---|
| 253 | n/a | if (c == 0x8e) { |
|---|
| 254 | n/a | /* JIS X 0201 half-width katakana */ |
|---|
| 255 | n/a | unsigned char c2; |
|---|
| 256 | n/a | |
|---|
| 257 | n/a | REQUIRE_INBUF(2); |
|---|
| 258 | n/a | c2 = INBYTE2; |
|---|
| 259 | n/a | if (c2 >= 0xa1 && c2 <= 0xdf) { |
|---|
| 260 | n/a | OUTCHAR(0xfec0 + c2); |
|---|
| 261 | n/a | NEXT_IN(2); |
|---|
| 262 | n/a | } |
|---|
| 263 | n/a | else |
|---|
| 264 | n/a | return 1; |
|---|
| 265 | n/a | } |
|---|
| 266 | n/a | else if (c == 0x8f) { |
|---|
| 267 | n/a | unsigned char c2, c3; |
|---|
| 268 | n/a | |
|---|
| 269 | n/a | REQUIRE_INBUF(3); |
|---|
| 270 | n/a | c2 = INBYTE2 ^ 0x80; |
|---|
| 271 | n/a | c3 = INBYTE3 ^ 0x80; |
|---|
| 272 | n/a | |
|---|
| 273 | n/a | /* JIS X 0213 Plane 2 or JIS X 0212 (see NOTES) */ |
|---|
| 274 | n/a | EMULATE_JISX0213_2000_DECODE_PLANE2(writer, c2, c3) |
|---|
| 275 | n/a | else if (TRYMAP_DEC(jisx0213_2_bmp, decoded, c2, c3)) |
|---|
| 276 | n/a | OUTCHAR(decoded); |
|---|
| 277 | n/a | else if (TRYMAP_DEC(jisx0213_2_emp, code, c2, c3)) { |
|---|
| 278 | n/a | OUTCHAR(EMPBASE | code); |
|---|
| 279 | n/a | NEXT_IN(3); |
|---|
| 280 | n/a | continue; |
|---|
| 281 | n/a | } |
|---|
| 282 | n/a | else if (TRYMAP_DEC(jisx0212, decoded, c2, c3)) |
|---|
| 283 | n/a | OUTCHAR(decoded); |
|---|
| 284 | n/a | else |
|---|
| 285 | n/a | return 1; |
|---|
| 286 | n/a | NEXT_IN(3); |
|---|
| 287 | n/a | } |
|---|
| 288 | n/a | else { |
|---|
| 289 | n/a | unsigned char c2; |
|---|
| 290 | n/a | |
|---|
| 291 | n/a | REQUIRE_INBUF(2); |
|---|
| 292 | n/a | c ^= 0x80; |
|---|
| 293 | n/a | c2 = INBYTE2 ^ 0x80; |
|---|
| 294 | n/a | |
|---|
| 295 | n/a | /* JIS X 0213 Plane 1 */ |
|---|
| 296 | n/a | EMULATE_JISX0213_2000_DECODE_PLANE1(writer, c, c2) |
|---|
| 297 | n/a | else if (c == 0x21 && c2 == 0x40) |
|---|
| 298 | n/a | OUTCHAR(0xff3c); |
|---|
| 299 | n/a | else if (c == 0x22 && c2 == 0x32) |
|---|
| 300 | n/a | OUTCHAR(0xff5e); |
|---|
| 301 | n/a | else if (TRYMAP_DEC(jisx0208, decoded, c, c2)) |
|---|
| 302 | n/a | OUTCHAR(decoded); |
|---|
| 303 | n/a | else if (TRYMAP_DEC(jisx0213_1_bmp, decoded, c, c2)) |
|---|
| 304 | n/a | OUTCHAR(decoded); |
|---|
| 305 | n/a | else if (TRYMAP_DEC(jisx0213_1_emp, code, c, c2)) { |
|---|
| 306 | n/a | OUTCHAR(EMPBASE | code); |
|---|
| 307 | n/a | NEXT_IN(2); |
|---|
| 308 | n/a | continue; |
|---|
| 309 | n/a | } |
|---|
| 310 | n/a | else if (TRYMAP_DEC(jisx0213_pair, code, c, c2)) { |
|---|
| 311 | n/a | OUTCHAR2(code >> 16, code & 0xffff); |
|---|
| 312 | n/a | NEXT_IN(2); |
|---|
| 313 | n/a | continue; |
|---|
| 314 | n/a | } |
|---|
| 315 | n/a | else |
|---|
| 316 | n/a | return 1; |
|---|
| 317 | n/a | NEXT_IN(2); |
|---|
| 318 | n/a | } |
|---|
| 319 | n/a | } |
|---|
| 320 | n/a | |
|---|
| 321 | n/a | return 0; |
|---|
| 322 | n/a | } |
|---|
| 323 | n/a | |
|---|
| 324 | n/a | |
|---|
| 325 | n/a | /* |
|---|
| 326 | n/a | * EUC-JP codec |
|---|
| 327 | n/a | */ |
|---|
| 328 | n/a | |
|---|
| 329 | n/a | ENCODER(euc_jp) |
|---|
| 330 | n/a | { |
|---|
| 331 | n/a | while (*inpos < inlen) { |
|---|
| 332 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 333 | n/a | DBCHAR code; |
|---|
| 334 | n/a | |
|---|
| 335 | n/a | if (c < 0x80) { |
|---|
| 336 | n/a | WRITEBYTE1((unsigned char)c); |
|---|
| 337 | n/a | NEXT(1, 1); |
|---|
| 338 | n/a | continue; |
|---|
| 339 | n/a | } |
|---|
| 340 | n/a | |
|---|
| 341 | n/a | if (c > 0xFFFF) |
|---|
| 342 | n/a | return 1; |
|---|
| 343 | n/a | |
|---|
| 344 | n/a | if (TRYMAP_ENC(jisxcommon, code, c)) |
|---|
| 345 | n/a | ; |
|---|
| 346 | n/a | else if (c >= 0xff61 && c <= 0xff9f) { |
|---|
| 347 | n/a | /* JIS X 0201 half-width katakana */ |
|---|
| 348 | n/a | WRITEBYTE2(0x8e, c - 0xfec0); |
|---|
| 349 | n/a | NEXT(1, 2); |
|---|
| 350 | n/a | continue; |
|---|
| 351 | n/a | } |
|---|
| 352 | n/a | #ifndef STRICT_BUILD |
|---|
| 353 | n/a | else if (c == 0xff3c) /* FULL-WIDTH REVERSE SOLIDUS */ |
|---|
| 354 | n/a | code = 0x2140; |
|---|
| 355 | n/a | else if (c == 0xa5) { /* YEN SIGN */ |
|---|
| 356 | n/a | WRITEBYTE1(0x5c); |
|---|
| 357 | n/a | NEXT(1, 1); |
|---|
| 358 | n/a | continue; |
|---|
| 359 | n/a | } else if (c == 0x203e) { /* OVERLINE */ |
|---|
| 360 | n/a | WRITEBYTE1(0x7e); |
|---|
| 361 | n/a | NEXT(1, 1); |
|---|
| 362 | n/a | continue; |
|---|
| 363 | n/a | } |
|---|
| 364 | n/a | #endif |
|---|
| 365 | n/a | else |
|---|
| 366 | n/a | return 1; |
|---|
| 367 | n/a | |
|---|
| 368 | n/a | if (code & 0x8000) { |
|---|
| 369 | n/a | /* JIS X 0212 */ |
|---|
| 370 | n/a | WRITEBYTE3(0x8f, code >> 8, (code & 0xFF) | 0x80); |
|---|
| 371 | n/a | NEXT(1, 3); |
|---|
| 372 | n/a | } else { |
|---|
| 373 | n/a | /* JIS X 0208 */ |
|---|
| 374 | n/a | WRITEBYTE2((code >> 8) | 0x80, (code & 0xFF) | 0x80); |
|---|
| 375 | n/a | NEXT(1, 2); |
|---|
| 376 | n/a | } |
|---|
| 377 | n/a | } |
|---|
| 378 | n/a | |
|---|
| 379 | n/a | return 0; |
|---|
| 380 | n/a | } |
|---|
| 381 | n/a | |
|---|
| 382 | n/a | DECODER(euc_jp) |
|---|
| 383 | n/a | { |
|---|
| 384 | n/a | while (inleft > 0) { |
|---|
| 385 | n/a | unsigned char c = INBYTE1; |
|---|
| 386 | n/a | Py_UCS4 decoded; |
|---|
| 387 | n/a | |
|---|
| 388 | n/a | if (c < 0x80) { |
|---|
| 389 | n/a | OUTCHAR(c); |
|---|
| 390 | n/a | NEXT_IN(1); |
|---|
| 391 | n/a | continue; |
|---|
| 392 | n/a | } |
|---|
| 393 | n/a | |
|---|
| 394 | n/a | if (c == 0x8e) { |
|---|
| 395 | n/a | /* JIS X 0201 half-width katakana */ |
|---|
| 396 | n/a | unsigned char c2; |
|---|
| 397 | n/a | |
|---|
| 398 | n/a | REQUIRE_INBUF(2); |
|---|
| 399 | n/a | c2 = INBYTE2; |
|---|
| 400 | n/a | if (c2 >= 0xa1 && c2 <= 0xdf) { |
|---|
| 401 | n/a | OUTCHAR(0xfec0 + c2); |
|---|
| 402 | n/a | NEXT_IN(2); |
|---|
| 403 | n/a | } |
|---|
| 404 | n/a | else |
|---|
| 405 | n/a | return 1; |
|---|
| 406 | n/a | } |
|---|
| 407 | n/a | else if (c == 0x8f) { |
|---|
| 408 | n/a | unsigned char c2, c3; |
|---|
| 409 | n/a | |
|---|
| 410 | n/a | REQUIRE_INBUF(3); |
|---|
| 411 | n/a | c2 = INBYTE2; |
|---|
| 412 | n/a | c3 = INBYTE3; |
|---|
| 413 | n/a | /* JIS X 0212 */ |
|---|
| 414 | n/a | if (TRYMAP_DEC(jisx0212, decoded, c2 ^ 0x80, c3 ^ 0x80)) { |
|---|
| 415 | n/a | OUTCHAR(decoded); |
|---|
| 416 | n/a | NEXT_IN(3); |
|---|
| 417 | n/a | } |
|---|
| 418 | n/a | else |
|---|
| 419 | n/a | return 1; |
|---|
| 420 | n/a | } |
|---|
| 421 | n/a | else { |
|---|
| 422 | n/a | unsigned char c2; |
|---|
| 423 | n/a | |
|---|
| 424 | n/a | REQUIRE_INBUF(2); |
|---|
| 425 | n/a | c2 = INBYTE2; |
|---|
| 426 | n/a | /* JIS X 0208 */ |
|---|
| 427 | n/a | #ifndef STRICT_BUILD |
|---|
| 428 | n/a | if (c == 0xa1 && c2 == 0xc0) |
|---|
| 429 | n/a | /* FULL-WIDTH REVERSE SOLIDUS */ |
|---|
| 430 | n/a | OUTCHAR(0xff3c); |
|---|
| 431 | n/a | else |
|---|
| 432 | n/a | #endif |
|---|
| 433 | n/a | if (TRYMAP_DEC(jisx0208, decoded, c ^ 0x80, c2 ^ 0x80)) |
|---|
| 434 | n/a | OUTCHAR(decoded); |
|---|
| 435 | n/a | else |
|---|
| 436 | n/a | return 1; |
|---|
| 437 | n/a | NEXT_IN(2); |
|---|
| 438 | n/a | } |
|---|
| 439 | n/a | } |
|---|
| 440 | n/a | |
|---|
| 441 | n/a | return 0; |
|---|
| 442 | n/a | } |
|---|
| 443 | n/a | |
|---|
| 444 | n/a | |
|---|
| 445 | n/a | /* |
|---|
| 446 | n/a | * SHIFT_JIS codec |
|---|
| 447 | n/a | */ |
|---|
| 448 | n/a | |
|---|
| 449 | n/a | ENCODER(shift_jis) |
|---|
| 450 | n/a | { |
|---|
| 451 | n/a | while (*inpos < inlen) { |
|---|
| 452 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 453 | n/a | DBCHAR code; |
|---|
| 454 | n/a | unsigned char c1, c2; |
|---|
| 455 | n/a | |
|---|
| 456 | n/a | #ifdef STRICT_BUILD |
|---|
| 457 | n/a | JISX0201_R_ENCODE(c, code) |
|---|
| 458 | n/a | #else |
|---|
| 459 | n/a | if (c < 0x80) |
|---|
| 460 | n/a | code = c; |
|---|
| 461 | n/a | else if (c == 0x00a5) |
|---|
| 462 | n/a | code = 0x5c; /* YEN SIGN */ |
|---|
| 463 | n/a | else if (c == 0x203e) |
|---|
| 464 | n/a | code = 0x7e; /* OVERLINE */ |
|---|
| 465 | n/a | #endif |
|---|
| 466 | n/a | else JISX0201_K_ENCODE(c, code) |
|---|
| 467 | n/a | else if (c > 0xFFFF) |
|---|
| 468 | n/a | return 1; |
|---|
| 469 | n/a | else |
|---|
| 470 | n/a | code = NOCHAR; |
|---|
| 471 | n/a | |
|---|
| 472 | n/a | if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) { |
|---|
| 473 | n/a | REQUIRE_OUTBUF(1); |
|---|
| 474 | n/a | |
|---|
| 475 | n/a | OUTBYTE1((unsigned char)code); |
|---|
| 476 | n/a | NEXT(1, 1); |
|---|
| 477 | n/a | continue; |
|---|
| 478 | n/a | } |
|---|
| 479 | n/a | |
|---|
| 480 | n/a | REQUIRE_OUTBUF(2); |
|---|
| 481 | n/a | |
|---|
| 482 | n/a | if (code == NOCHAR) { |
|---|
| 483 | n/a | if (TRYMAP_ENC(jisxcommon, code, c)) |
|---|
| 484 | n/a | ; |
|---|
| 485 | n/a | #ifndef STRICT_BUILD |
|---|
| 486 | n/a | else if (c == 0xff3c) |
|---|
| 487 | n/a | code = 0x2140; /* FULL-WIDTH REVERSE SOLIDUS */ |
|---|
| 488 | n/a | #endif |
|---|
| 489 | n/a | else |
|---|
| 490 | n/a | return 1; |
|---|
| 491 | n/a | |
|---|
| 492 | n/a | if (code & 0x8000) /* MSB set: JIS X 0212 */ |
|---|
| 493 | n/a | return 1; |
|---|
| 494 | n/a | } |
|---|
| 495 | n/a | |
|---|
| 496 | n/a | c1 = code >> 8; |
|---|
| 497 | n/a | c2 = code & 0xff; |
|---|
| 498 | n/a | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); |
|---|
| 499 | n/a | c1 = (c1 - 0x21) >> 1; |
|---|
| 500 | n/a | OUTBYTE1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1); |
|---|
| 501 | n/a | OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41); |
|---|
| 502 | n/a | NEXT(1, 2); |
|---|
| 503 | n/a | } |
|---|
| 504 | n/a | |
|---|
| 505 | n/a | return 0; |
|---|
| 506 | n/a | } |
|---|
| 507 | n/a | |
|---|
| 508 | n/a | DECODER(shift_jis) |
|---|
| 509 | n/a | { |
|---|
| 510 | n/a | while (inleft > 0) { |
|---|
| 511 | n/a | unsigned char c = INBYTE1; |
|---|
| 512 | n/a | Py_UCS4 decoded; |
|---|
| 513 | n/a | |
|---|
| 514 | n/a | #ifdef STRICT_BUILD |
|---|
| 515 | n/a | JISX0201_R_DECODE(c, writer) |
|---|
| 516 | n/a | #else |
|---|
| 517 | n/a | if (c < 0x80) |
|---|
| 518 | n/a | OUTCHAR(c); |
|---|
| 519 | n/a | #endif |
|---|
| 520 | n/a | else JISX0201_K_DECODE(c, writer) |
|---|
| 521 | n/a | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){ |
|---|
| 522 | n/a | unsigned char c1, c2; |
|---|
| 523 | n/a | |
|---|
| 524 | n/a | REQUIRE_INBUF(2); |
|---|
| 525 | n/a | c2 = INBYTE2; |
|---|
| 526 | n/a | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
|---|
| 527 | n/a | return 1; |
|---|
| 528 | n/a | |
|---|
| 529 | n/a | c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
|---|
| 530 | n/a | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
|---|
| 531 | n/a | c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1) + 0x21); |
|---|
| 532 | n/a | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
|---|
| 533 | n/a | |
|---|
| 534 | n/a | #ifndef STRICT_BUILD |
|---|
| 535 | n/a | if (c1 == 0x21 && c2 == 0x40) { |
|---|
| 536 | n/a | /* FULL-WIDTH REVERSE SOLIDUS */ |
|---|
| 537 | n/a | OUTCHAR(0xff3c); |
|---|
| 538 | n/a | NEXT_IN(2); |
|---|
| 539 | n/a | continue; |
|---|
| 540 | n/a | } |
|---|
| 541 | n/a | #endif |
|---|
| 542 | n/a | if (TRYMAP_DEC(jisx0208, decoded, c1, c2)) { |
|---|
| 543 | n/a | OUTCHAR(decoded); |
|---|
| 544 | n/a | NEXT_IN(2); |
|---|
| 545 | n/a | continue; |
|---|
| 546 | n/a | } |
|---|
| 547 | n/a | else |
|---|
| 548 | n/a | return 1; |
|---|
| 549 | n/a | } |
|---|
| 550 | n/a | else |
|---|
| 551 | n/a | return 1; |
|---|
| 552 | n/a | |
|---|
| 553 | n/a | NEXT_IN(1); /* JIS X 0201 */ |
|---|
| 554 | n/a | } |
|---|
| 555 | n/a | |
|---|
| 556 | n/a | return 0; |
|---|
| 557 | n/a | } |
|---|
| 558 | n/a | |
|---|
| 559 | n/a | |
|---|
| 560 | n/a | /* |
|---|
| 561 | n/a | * SHIFT_JIS-2004 codec |
|---|
| 562 | n/a | */ |
|---|
| 563 | n/a | |
|---|
| 564 | n/a | ENCODER(shift_jis_2004) |
|---|
| 565 | n/a | { |
|---|
| 566 | n/a | while (*inpos < inlen) { |
|---|
| 567 | n/a | Py_UCS4 c = INCHAR1; |
|---|
| 568 | n/a | DBCHAR code = NOCHAR; |
|---|
| 569 | n/a | int c1, c2; |
|---|
| 570 | n/a | Py_ssize_t insize; |
|---|
| 571 | n/a | |
|---|
| 572 | n/a | JISX0201_ENCODE(c, code) |
|---|
| 573 | n/a | |
|---|
| 574 | n/a | if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) { |
|---|
| 575 | n/a | WRITEBYTE1((unsigned char)code); |
|---|
| 576 | n/a | NEXT(1, 1); |
|---|
| 577 | n/a | continue; |
|---|
| 578 | n/a | } |
|---|
| 579 | n/a | |
|---|
| 580 | n/a | REQUIRE_OUTBUF(2); |
|---|
| 581 | n/a | insize = 1; |
|---|
| 582 | n/a | |
|---|
| 583 | n/a | if (code == NOCHAR) { |
|---|
| 584 | n/a | if (c <= 0xffff) { |
|---|
| 585 | n/a | EMULATE_JISX0213_2000_ENCODE_BMP(code, c) |
|---|
| 586 | n/a | else if (TRYMAP_ENC(jisx0213_bmp, code, c)) { |
|---|
| 587 | n/a | if (code == MULTIC) { |
|---|
| 588 | n/a | if (inlen - *inpos < 2) { |
|---|
| 589 | n/a | if (flags & MBENC_FLUSH) { |
|---|
| 590 | n/a | code = find_pairencmap |
|---|
| 591 | n/a | ((ucs2_t)c, 0, |
|---|
| 592 | n/a | jisx0213_pair_encmap, |
|---|
| 593 | n/a | JISX0213_ENCPAIRS); |
|---|
| 594 | n/a | if (code == DBCINV) |
|---|
| 595 | n/a | return 1; |
|---|
| 596 | n/a | } |
|---|
| 597 | n/a | else |
|---|
| 598 | n/a | return MBERR_TOOFEW; |
|---|
| 599 | n/a | } |
|---|
| 600 | n/a | else { |
|---|
| 601 | n/a | Py_UCS4 ch2 = INCHAR2; |
|---|
| 602 | n/a | code = find_pairencmap( |
|---|
| 603 | n/a | (ucs2_t)c, ch2, |
|---|
| 604 | n/a | jisx0213_pair_encmap, |
|---|
| 605 | n/a | JISX0213_ENCPAIRS); |
|---|
| 606 | n/a | if (code == DBCINV) { |
|---|
| 607 | n/a | code = find_pairencmap( |
|---|
| 608 | n/a | (ucs2_t)c, 0, |
|---|
| 609 | n/a | jisx0213_pair_encmap, |
|---|
| 610 | n/a | JISX0213_ENCPAIRS); |
|---|
| 611 | n/a | if (code == DBCINV) |
|---|
| 612 | n/a | return 1; |
|---|
| 613 | n/a | } |
|---|
| 614 | n/a | else |
|---|
| 615 | n/a | insize = 2; |
|---|
| 616 | n/a | } |
|---|
| 617 | n/a | } |
|---|
| 618 | n/a | } |
|---|
| 619 | n/a | else if (TRYMAP_ENC(jisxcommon, code, c)) { |
|---|
| 620 | n/a | /* abandon JIS X 0212 codes */ |
|---|
| 621 | n/a | if (code & 0x8000) |
|---|
| 622 | n/a | return 1; |
|---|
| 623 | n/a | } |
|---|
| 624 | n/a | else |
|---|
| 625 | n/a | return 1; |
|---|
| 626 | n/a | } |
|---|
| 627 | n/a | else if (c >> 16 == EMPBASE >> 16) { |
|---|
| 628 | n/a | EMULATE_JISX0213_2000_ENCODE_EMP(code, c) |
|---|
| 629 | n/a | else if (TRYMAP_ENC(jisx0213_emp, code, c&0xffff)) |
|---|
| 630 | n/a | ; |
|---|
| 631 | n/a | else |
|---|
| 632 | n/a | return insize; |
|---|
| 633 | n/a | } |
|---|
| 634 | n/a | else |
|---|
| 635 | n/a | return insize; |
|---|
| 636 | n/a | } |
|---|
| 637 | n/a | |
|---|
| 638 | n/a | c1 = code >> 8; |
|---|
| 639 | n/a | c2 = (code & 0xff) - 0x21; |
|---|
| 640 | n/a | |
|---|
| 641 | n/a | if (c1 & 0x80) { |
|---|
| 642 | n/a | /* Plane 2 */ |
|---|
| 643 | n/a | if (c1 >= 0xee) |
|---|
| 644 | n/a | c1 -= 0x87; |
|---|
| 645 | n/a | else if (c1 >= 0xac || c1 == 0xa8) |
|---|
| 646 | n/a | c1 -= 0x49; |
|---|
| 647 | n/a | else |
|---|
| 648 | n/a | c1 -= 0x43; |
|---|
| 649 | n/a | } |
|---|
| 650 | n/a | else { |
|---|
| 651 | n/a | /* Plane 1 */ |
|---|
| 652 | n/a | c1 -= 0x21; |
|---|
| 653 | n/a | } |
|---|
| 654 | n/a | |
|---|
| 655 | n/a | if (c1 & 1) |
|---|
| 656 | n/a | c2 += 0x5e; |
|---|
| 657 | n/a | c1 >>= 1; |
|---|
| 658 | n/a | OUTBYTE1(c1 + (c1 < 0x1f ? 0x81 : 0xc1)); |
|---|
| 659 | n/a | OUTBYTE2(c2 + (c2 < 0x3f ? 0x40 : 0x41)); |
|---|
| 660 | n/a | |
|---|
| 661 | n/a | NEXT(insize, 2); |
|---|
| 662 | n/a | } |
|---|
| 663 | n/a | |
|---|
| 664 | n/a | return 0; |
|---|
| 665 | n/a | } |
|---|
| 666 | n/a | |
|---|
| 667 | n/a | DECODER(shift_jis_2004) |
|---|
| 668 | n/a | { |
|---|
| 669 | n/a | while (inleft > 0) { |
|---|
| 670 | n/a | unsigned char c = INBYTE1; |
|---|
| 671 | n/a | |
|---|
| 672 | n/a | JISX0201_DECODE(c, writer) |
|---|
| 673 | n/a | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc)){ |
|---|
| 674 | n/a | unsigned char c1, c2; |
|---|
| 675 | n/a | Py_UCS4 code, decoded; |
|---|
| 676 | n/a | |
|---|
| 677 | n/a | REQUIRE_INBUF(2); |
|---|
| 678 | n/a | c2 = INBYTE2; |
|---|
| 679 | n/a | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) |
|---|
| 680 | n/a | return 1; |
|---|
| 681 | n/a | |
|---|
| 682 | n/a | c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1); |
|---|
| 683 | n/a | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); |
|---|
| 684 | n/a | c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1)); |
|---|
| 685 | n/a | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; |
|---|
| 686 | n/a | |
|---|
| 687 | n/a | if (c1 < 0x5e) { /* Plane 1 */ |
|---|
| 688 | n/a | c1 += 0x21; |
|---|
| 689 | n/a | EMULATE_JISX0213_2000_DECODE_PLANE1(writer, |
|---|
| 690 | n/a | c1, c2) |
|---|
| 691 | n/a | else if (TRYMAP_DEC(jisx0208, decoded, c1, c2)) |
|---|
| 692 | n/a | OUTCHAR(decoded); |
|---|
| 693 | n/a | else if (TRYMAP_DEC(jisx0213_1_bmp, decoded, c1, c2)) |
|---|
| 694 | n/a | OUTCHAR(decoded); |
|---|
| 695 | n/a | else if (TRYMAP_DEC(jisx0213_1_emp, code, c1, c2)) |
|---|
| 696 | n/a | OUTCHAR(EMPBASE | code); |
|---|
| 697 | n/a | else if (TRYMAP_DEC(jisx0213_pair, code, c1, c2)) |
|---|
| 698 | n/a | OUTCHAR2(code >> 16, code & 0xffff); |
|---|
| 699 | n/a | else |
|---|
| 700 | n/a | return 1; |
|---|
| 701 | n/a | NEXT_IN(2); |
|---|
| 702 | n/a | } |
|---|
| 703 | n/a | else { /* Plane 2 */ |
|---|
| 704 | n/a | if (c1 >= 0x67) |
|---|
| 705 | n/a | c1 += 0x07; |
|---|
| 706 | n/a | else if (c1 >= 0x63 || c1 == 0x5f) |
|---|
| 707 | n/a | c1 -= 0x37; |
|---|
| 708 | n/a | else |
|---|
| 709 | n/a | c1 -= 0x3d; |
|---|
| 710 | n/a | |
|---|
| 711 | n/a | EMULATE_JISX0213_2000_DECODE_PLANE2(writer, |
|---|
| 712 | n/a | c1, c2) |
|---|
| 713 | n/a | else if (TRYMAP_DEC(jisx0213_2_bmp, decoded, c1, c2)) |
|---|
| 714 | n/a | OUTCHAR(decoded); |
|---|
| 715 | n/a | else if (TRYMAP_DEC(jisx0213_2_emp, code, c1, c2)) { |
|---|
| 716 | n/a | OUTCHAR(EMPBASE | code); |
|---|
| 717 | n/a | NEXT_IN(2); |
|---|
| 718 | n/a | continue; |
|---|
| 719 | n/a | } |
|---|
| 720 | n/a | else |
|---|
| 721 | n/a | return 1; |
|---|
| 722 | n/a | NEXT_IN(2); |
|---|
| 723 | n/a | } |
|---|
| 724 | n/a | continue; |
|---|
| 725 | n/a | } |
|---|
| 726 | n/a | else |
|---|
| 727 | n/a | return 1; |
|---|
| 728 | n/a | |
|---|
| 729 | n/a | NEXT_IN(1); /* JIS X 0201 */ |
|---|
| 730 | n/a | } |
|---|
| 731 | n/a | |
|---|
| 732 | n/a | return 0; |
|---|
| 733 | n/a | } |
|---|
| 734 | n/a | |
|---|
| 735 | n/a | |
|---|
| 736 | n/a | BEGIN_MAPPINGS_LIST |
|---|
| 737 | n/a | MAPPING_DECONLY(jisx0208) |
|---|
| 738 | n/a | MAPPING_DECONLY(jisx0212) |
|---|
| 739 | n/a | MAPPING_ENCONLY(jisxcommon) |
|---|
| 740 | n/a | MAPPING_DECONLY(jisx0213_1_bmp) |
|---|
| 741 | n/a | MAPPING_DECONLY(jisx0213_2_bmp) |
|---|
| 742 | n/a | MAPPING_ENCONLY(jisx0213_bmp) |
|---|
| 743 | n/a | MAPPING_DECONLY(jisx0213_1_emp) |
|---|
| 744 | n/a | MAPPING_DECONLY(jisx0213_2_emp) |
|---|
| 745 | n/a | MAPPING_ENCONLY(jisx0213_emp) |
|---|
| 746 | n/a | MAPPING_ENCDEC(jisx0213_pair) |
|---|
| 747 | n/a | MAPPING_ENCDEC(cp932ext) |
|---|
| 748 | n/a | END_MAPPINGS_LIST |
|---|
| 749 | n/a | |
|---|
| 750 | n/a | BEGIN_CODECS_LIST |
|---|
| 751 | n/a | CODEC_STATELESS(shift_jis) |
|---|
| 752 | n/a | CODEC_STATELESS(cp932) |
|---|
| 753 | n/a | CODEC_STATELESS(euc_jp) |
|---|
| 754 | n/a | CODEC_STATELESS(shift_jis_2004) |
|---|
| 755 | n/a | CODEC_STATELESS(euc_jis_2004) |
|---|
| 756 | n/a | { "euc_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(euc_jis_2004) }, |
|---|
| 757 | n/a | { "shift_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(shift_jis_2004) }, |
|---|
| 758 | n/a | END_CODECS_LIST |
|---|
| 759 | n/a | |
|---|
| 760 | n/a | I_AM_A_MODULE_FOR(jp) |
|---|