lib/iso2022_jp2.h

   1 /*
   2  * Copyright (C) 1999-2000 Free Software Foundation, Inc.
   3  * This file is part of the GNU LIBICONV Library.
   4  *
   5  * The GNU LIBICONV Library is free software; you can redistribute it
   6  * and/or modify it under the terms of the GNU Library General Public
   7  * License as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * The GNU LIBICONV Library is distributed in the hope that it will be
  11  * useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * Library General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Library General Public
  16  * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
  17  * If not, write to the Free Software Foundation, Inc., 59 Temple Place -
  18  * Suite 330, Boston, MA 02111-1307, USA.
  19  */
  20
  21 /*
  22  * ISO-2022-JP-2
  23  */
  24
  25 /* Specification: RFC 1554 */
  26 /* ESC '(' 'I' for JISX0201 Katakana is an extension not found in RFC 1554 or
  27    CJK.INF, but implemented in glibc-2.1 and qt-2.0. */
  28
  29 #define ESC 0x1b
  30
  31 /*
  32  * The state is composed of one of the following values
  33  */
  34 #define STATE_ASCII             0
  35 #define STATE_JISX0201ROMAN     1
  36 #define STATE_JISX0201KATAKANA  2
  37 #define STATE_JISX0208          3
  38 #define STATE_JISX0212          4
  39 #define STATE_GB2312            5
  40 #define STATE_KSC5601           6
  41 /*
  42  * and one of the following values, << 8
  43  */
  44 #define STATE_G2_NONE           0
  45 #define STATE_G2_ISO8859_1      1
  46 #define STATE_G2_ISO8859_7      2
  47
  48 #define SPLIT_STATE \
  49   unsigned int state1 = state & 0xff, state2 = state >> 8
  50 #define COMBINE_STATE \
  51   state = (state2 << 8) | state1
  52
  53 static int
  54 iso2022_jp2_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, int n)
  55 {
  56   state_t state = conv->istate;
  57   SPLIT_STATE;
  58   int count = 0;
  59   unsigned char c;
  60   for (;;) {
  61     c = *s;
  62     if (c == ESC) {
  63       if (n < count+3)
  64         goto none;
  65       if (s[1] == '(') {
  66         if (s[2] == 'B') {
  67           state1 = STATE_ASCII;
  68           s += 3; count += 3;
  69           if (n < count+1)
  70             goto none;
  71           continue;
  72         }
  73         if (s[2] == 'J') {
  74           state1 = STATE_JISX0201ROMAN;
  75           s += 3; count += 3;
  76           if (n < count+1)
  77             goto none;
  78           continue;
  79         }
  80         if (s[2] == 'I') {
  81           state1 = STATE_JISX0201KATAKANA;
  82           s += 3; count += 3;
  83           if (n < count+1)
  84             goto none;
  85           continue;
  86         }
  87         return RET_ILSEQ;
  88       }
  89       if (s[1] == '$') {
  90         if (s[2] == '@' || s[2] == 'B') {
  91           /* We don't distinguish JIS X 0208-1978 and JIS X 0208-1983. */
  92           state1 = STATE_JISX0208;
  93           s += 3; count += 3;
  94           if (n < count+1)
  95             goto none;
  96           continue;
  97         }
  98         if (s[2] == 'A') {
  99           state1 = STATE_GB2312;
 100           s += 3; count += 3;
 101           if (n < count+1)
 102             goto none;
 103           continue;
 104         }
 105         if (s[2] == '(') {
 106           if (n < count+4)
 107             goto none;
 108           if (s[3] == 'D') {
 109             state1 = STATE_JISX0212;
 110             s += 4; count += 4;
 111             if (n < count+1)
 112               goto none;
 113             continue;
 114           }
 115           if (s[3] == 'C') {
 116             state1 = STATE_KSC5601;
 117             s += 4; count += 4;
 118             if (n < count+1)
 119               goto none;
 120             continue;
 121           }
 122           return RET_ILSEQ;
 123         }
 124         return RET_ILSEQ;
 125       }
 126       if (s[1] == '.') {
 127         if (n < count+3)
 128           goto none;
 129         if (s[2] == 'A') {
 130           state2 = STATE_G2_ISO8859_1;
 131           s += 3; count += 3;
 132           if (n < count+1)
 133             goto none;
 134           continue;
 135         }
 136         if (s[2] == 'F') {
 137           state2 = STATE_G2_ISO8859_7;
 138           s += 3; count += 3;
 139           if (n < count+1)
 140             goto none;
 141           continue;
 142         }
 143         return RET_ILSEQ;
 144       }
 145       if (s[1] == 'N') {
 146         switch (state2) {
 147           case STATE_G2_NONE:
 148             return RET_ILSEQ;
 149           case STATE_G2_ISO8859_1:
 150             if (s[2] < 0x80) {
 151               unsigned char buf = s[2]+0x80;
 152               int ret = iso8859_1_mbtowc(conv,pwc,&buf,1);
 153               if (ret == RET_ILSEQ)
 154                 return RET_ILSEQ;
 155               if (ret != 1) abort();
 156               COMBINE_STATE;
 157               conv->istate = state;
 158               return count+3;
 159             } else
 160               return RET_ILSEQ;
 161           case STATE_G2_ISO8859_7:
 162             if (s[2] < 0x80) {
 163               unsigned char buf = s[2]+0x80;
 164               int ret = iso8859_7_mbtowc(conv,pwc,&buf,1);
 165               if (ret == RET_ILSEQ)
 166                 return RET_ILSEQ;
 167               if (ret != 1) abort();
 168               COMBINE_STATE;
 169               conv->istate = state;
 170               return count+3;
 171             } else
 172               return RET_ILSEQ;
 173           default: abort();
 174         }
 175       }
 176       return RET_ILSEQ;
 177     }
 178     break;
 179   }
 180   switch (state1) {
 181     case STATE_ASCII:
 182       if (c < 0x80) {
 183         int ret = ascii_mbtowc(conv,pwc,s,1);
 184         if (ret == RET_ILSEQ)
 185           return RET_ILSEQ;
 186         if (ret != 1) abort();
 187         if (*pwc == 0x000a || *pwc == 0x000d)
 188           state2 = STATE_G2_NONE;
 189         COMBINE_STATE;
 190         conv->istate = state;
 191         return count+1;
 192       } else
 193         return RET_ILSEQ;
 194     case STATE_JISX0201ROMAN:
 195       if (c < 0x80) {
 196         int ret = jisx0201_mbtowc(conv,pwc,s,1);
 197         if (ret == RET_ILSEQ)
 198           return RET_ILSEQ;
 199         if (ret != 1) abort();
 200         if (*pwc == 0x000a || *pwc == 0x000d)
 201           state2 = STATE_G2_NONE;
 202         COMBINE_STATE;
 203         conv->istate = state;
 204         return count+1;
 205       } else
 206         return RET_ILSEQ;
 207     case STATE_JISX0201KATAKANA:
 208       if (c < 0x80) {
 209         unsigned char buf = c+0x80;
 210         int ret = jisx0201_mbtowc(conv,pwc,&buf,1);
 211         if (ret == RET_ILSEQ)
 212           return RET_ILSEQ;
 213         if (ret != 1) abort();
 214         COMBINE_STATE;
 215         conv->istate = state;
 216         return count+1;
 217       } else
 218         return RET_ILSEQ;
 219     case STATE_JISX0208:
 220       if (n < count+2)
 221         goto none;
 222       if (s[0] < 0x80 && s[1] < 0x80) {
 223         int ret = jisx0208_mbtowc(conv,pwc,s,2);
 224         if (ret == RET_ILSEQ)
 225           return RET_ILSEQ;
 226         if (ret != 2) abort();
 227         COMBINE_STATE;
 228         conv->istate = state;
 229         return count+2;
 230       } else
 231         return RET_ILSEQ;
 232     case STATE_JISX0212:
 233       if (n < count+2)
 234         goto none;
 235       if (s[0] < 0x80 && s[1] < 0x80) {
 236         int ret = jisx0212_mbtowc(conv,pwc,s,2);
 237         if (ret == RET_ILSEQ)
 238           return RET_ILSEQ;
 239         if (ret != 2) abort();
 240         COMBINE_STATE;
 241         conv->istate = state;
 242         return count+2;
 243       } else
 244         return RET_ILSEQ;
 245     case STATE_GB2312:
 246       if (n < count+2)
 247         goto none;
 248       if (s[0] < 0x80 && s[1] < 0x80) {
 249         int ret = gb2312_mbtowc(conv,pwc,s,2);
 250         if (ret == RET_ILSEQ)
 251           return RET_ILSEQ;
 252         if (ret != 2) abort();
 253         COMBINE_STATE;
 254         conv->istate = state;
 255         return count+2;
 256       } else
 257         return RET_ILSEQ;
 258     case STATE_KSC5601:
 259       if (n < count+2)
 260         goto none;
 261       if (s[0] < 0x80 && s[1] < 0x80) {
 262         int ret = ksc5601_mbtowc(conv,pwc,s,2);
 263         if (ret == RET_ILSEQ)
 264           return RET_ILSEQ;
 265         if (ret != 2) abort();
 266         COMBINE_STATE;
 267         conv->istate = state;
 268         return count+2;
 269       } else
 270         return RET_ILSEQ;
 271     default: abort();
 272   }
 273
 274 none:
 275   COMBINE_STATE;
 276   conv->istate = state;
 277   return RET_TOOFEW(count);
 278 }
 279
 280 static int
 281 iso2022_jp2_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n)
 282 {
 283   state_t state = conv->ostate;
 284   SPLIT_STATE;
 285   unsigned char buf[2];
 286   int ret;
 287
 288   /* Try ASCII. */
 289   ret = ascii_wctomb(conv,buf,wc,1);
 290   if (ret != RET_ILSEQ) {
 291     if (ret != 1) abort();
 292     if (buf[0] < 0x80) {
 293       int count = (state1 == STATE_ASCII ? 1 : 4);
 294       if (n < count)
 295         return RET_TOOSMALL;
 296       if (state1 != STATE_ASCII) {
 297         r[0] = ESC;
 298         r[1] = '(';
 299         r[2] = 'B';
 300         r += 3;
 301         state1 = STATE_ASCII;
 302       }
 303       r[0] = buf[0];
 304       if (wc == 0x000a || wc == 0x000d)
 305         state2 = STATE_G2_NONE;
 306       COMBINE_STATE;
 307       conv->ostate = state;
 308       return count;
 309     }
 310   }
 311
 312   /* Try ISO-8859-1. */
 313   ret = iso8859_1_wctomb(conv,buf,wc,1);
 314   if (ret != RET_ILSEQ) {
 315     if (ret != 1) abort();
 316     if (buf[0] >= 0x80) {
 317       int count = (state2 == STATE_G2_ISO8859_1 ? 3 : 6);
 318       if (n < count)
 319         return RET_TOOSMALL;
 320       if (state2 != STATE_G2_ISO8859_1) {
 321         r[0] = ESC;
 322         r[1] = '.';
 323         r[2] = 'A';
 324         r += 3;
 325         state2 = STATE_G2_ISO8859_1;
 326       }
 327       r[0] = ESC;
 328       r[1] = 'N';
 329       r[2] = buf[0]-0x80;
 330       COMBINE_STATE;
 331       conv->ostate = state;
 332       return count;
 333     }
 334   }
 335
 336   /* Try ISO-8859-7. */
 337   ret = iso8859_7_wctomb(conv,buf,wc,1);
 338   if (ret != RET_ILSEQ) {
 339     if (ret != 1) abort();
 340     if (buf[0] >= 0x80) {
 341       int count = (state2 == STATE_G2_ISO8859_7 ? 3 : 6);
 342       if (n < count)
 343         return RET_TOOSMALL;
 344       if (state2 != STATE_G2_ISO8859_7) {
 345         r[0] = ESC;
 346         r[1] = '.';
 347         r[2] = 'F';
 348         r += 3;
 349         state2 = STATE_G2_ISO8859_7;
 350       }
 351       r[0] = ESC;
 352       r[1] = 'N';
 353       r[2] = buf[0]-0x80;
 354       COMBINE_STATE;
 355       conv->ostate = state;
 356       return count;
 357     }
 358   }
 359
 360   /* Try JIS X 0201-1976 Roman and Kana. */
 361   ret = jisx0201_wctomb(conv,buf,wc,1);
 362   if (ret != RET_ILSEQ) {
 363     if (ret != 1) abort();
 364     if (buf[0] < 0x80) {
 365       int count = (state1 == STATE_JISX0201ROMAN ? 1 : 4);
 366       if (n < count)
 367         return RET_TOOSMALL;
 368       if (state1 != STATE_JISX0201ROMAN) {
 369         r[0] = ESC;
 370         r[1] = '(';
 371         r[2] = 'J';
 372         r += 3;
 373         state1 = STATE_JISX0201ROMAN;
 374       }
 375       r[0] = buf[0];
 376       if (wc == 0x000a || wc == 0x000d)
 377         state2 = STATE_G2_NONE;
 378       COMBINE_STATE;
 379       conv->ostate = state;
 380       return count;
 381     } else {
 382       int count = (state1 == STATE_JISX0201KATAKANA ? 1 : 4);
 383       if (n < count)
 384         return RET_TOOSMALL;
 385       if (state1 != STATE_JISX0201KATAKANA) {
 386         r[0] = ESC;
 387         r[1] = '(';
 388         r[2] = 'I';
 389         r += 3;
 390         state1 = STATE_JISX0201KATAKANA;
 391       }
 392       r[0] = buf[0]-0x80;
 393       COMBINE_STATE;
 394       conv->ostate = state;
 395       return count;
 396     }
 397   }
 398
 399   /* Try JIS X 0208-1990 in place of JIS X 0208-1978 and JIS X 0208-1983. */
 400   ret = jisx0208_wctomb(conv,buf,wc,2);
 401   if (ret != RET_ILSEQ) {
 402     if (ret != 2) abort();
 403     if (buf[0] < 0x80 && buf[1] < 0x80) {
 404       int count = (state1 == STATE_JISX0208 ? 2 : 5);
 405       if (n < count)
 406         return RET_TOOSMALL;
 407       if (state1 != STATE_JISX0208) {
 408         r[0] = ESC;
 409         r[1] = '$';
 410         r[2] = 'B';
 411         r += 3;
 412         state1 = STATE_JISX0208;
 413       }
 414       r[0] = buf[0];
 415       r[1] = buf[1];
 416       COMBINE_STATE;
 417       conv->ostate = state;
 418       return count;
 419     }
 420   }
 421
 422   /* Try JIS X 0212-1990. */
 423   ret = jisx0212_wctomb(conv,buf,wc,2);
 424   if (ret != RET_ILSEQ) {
 425     if (ret != 2) abort();
 426     if (buf[0] < 0x80 && buf[1] < 0x80) {
 427       int count = (state1 == STATE_JISX0212 ? 2 : 6);
 428       if (n < count)
 429         return RET_TOOSMALL;
 430       if (state1 != STATE_JISX0212) {
 431         r[0] = ESC;
 432         r[1] = '$';
 433         r[2] = '(';
 434         r[3] = 'D';
 435         r += 4;
 436         state1 = STATE_JISX0212;
 437       }
 438       r[0] = buf[0];
 439       r[1] = buf[1];
 440       COMBINE_STATE;
 441       conv->ostate = state;
 442       return count;
 443     }
 444   }
 445
 446   /* Try GB 2312-1980. */
 447   ret = gb2312_wctomb(conv,buf,wc,2);
 448   if (ret != RET_ILSEQ) {
 449     if (ret != 2) abort();
 450     if (buf[0] < 0x80 && buf[1] < 0x80) {
 451       int count = (state1 == STATE_GB2312 ? 2 : 5);
 452       if (n < count)
 453         return RET_TOOSMALL;
 454       if (state1 != STATE_GB2312) {
 455         r[0] = ESC;
 456         r[1] = '$';
 457         r[2] = 'A';
 458         r += 3;
 459         state1 = STATE_GB2312;
 460       }
 461       r[0] = buf[0];
 462       r[1] = buf[1];
 463       COMBINE_STATE;
 464       conv->ostate = state;
 465       return count;
 466     }
 467   }
 468
 469   /* Try KS C 5601-1992. */
 470   ret = ksc5601_wctomb(conv,buf,wc,2);
 471   if (ret != RET_ILSEQ) {
 472     if (ret != 2) abort();
 473     if (buf[0] < 0x80 && buf[1] < 0x80) {
 474       int count = (state1 == STATE_KSC5601 ? 2 : 6);
 475       if (n < count)
 476         return RET_TOOSMALL;
 477       if (state1 != STATE_KSC5601) {
 478         r[0] = ESC;
 479         r[1] = '$';
 480         r[2] = '(';
 481         r[3] = 'C';
 482         r += 4;
 483         state1 = STATE_KSC5601;
 484       }
 485       r[0] = buf[0];
 486       r[1] = buf[1];
 487       COMBINE_STATE;
 488       conv->ostate = state;
 489       return count;
 490     }
 491   }
 492
 493   return RET_ILSEQ;
 494 }
 495
 496 static int
 497 iso2022_jp2_reset (conv_t conv, unsigned char *r, int n)
 498 {
 499   state_t state = conv->ostate;
 500   SPLIT_STATE;
 501   (void)state2;
 502   if (state1 != STATE_ASCII) {
 503     if (n < 3)
 504       return RET_TOOSMALL;
 505     r[0] = ESC;
 506     r[1] = '(';
 507     r[2] = 'B';
 508     /* conv->ostate = 0; will be done by the caller */
 509     return 3;
 510   } else
 511     return 0;
 512 }
 513
 514 #undef COMBINE_STATE
 515 #undef SPLIT_STATE
 516 #undef STATE_G2_ISO8859_7
 517 #undef STATE_G2_ISO8859_1
 518 #undef STATE_G2_NONE
 519 #undef STATE_KSC5601
 520 #undef STATE_GB2312
 521 #undef STATE_JISX0212
 522 #undef STATE_JISX0208
 523 #undef STATE_JISX0201KATAKANA
 524 #undef STATE_JISX0201ROMAN
 525 #undef STATE_ASCII