lib/iso2022_jpms.h

   1 /*
   2  * Copyright (C) 1999-2001, 2008, 2011-2012, 2016, 2018 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 Lesser General Public
   7  * License as published by the Free Software Foundation; either version 2.1
   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  * Lesser General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General Public
  16  * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
  17  * If not, see <https://www.gnu.org/licenses/>.
  18  */
  19
  20 /*
  21  * ISO-2022-JP-MS
  22  * alias CP50221
  23  *
  24  * This is an extension of ISO-2022-JP-1 with larger character sets.
  25  * It uses ESC $ B and ESC $ ( D to denote *extensions* of JIS X 0208 and
  26  * JIS X 0212, respectively.  This violates the principles of ISO 2022,
  27  * where
  28  *   1. character sets to be used by ISO 2022 have to be registered at the
  29  *      ISO IR registry <https://www.itscj.ipsj.or.jp/ISO-IR/>,
  30  *   2. different character sets are designated by different escape
  31  *      sequences.
  32  * It's a typical instance of the "embrace and extend" strategy by Microsoft
  33  * <https://en.wikipedia.org/wiki/Embrace,_extend_and_extinguish>.
  34  */
  35
  36 /*
  37  * Windows has three encodings CP50220, CP50221, CP50222.
  38  * The common parts are:
  39  *   - US-ASCII (0x00..0x7F)
  40  *   - JIS X 0208 extended by
  41  *       - one row (0x2D),
  42  *       - a private use area (rows 0x75..0x7E = U+E000..U+E3AB),
  43  *     enabled with ESC $ B, disabled with ESC ( B.
  44  *   - JIS X 0212 extended by
  45  *       - two rows (0x73..0x74),
  46  *       - a private use area (rows 0x75..0x7E = U+E3AC..U+E757),
  47  *     enabled with ESC $ ( D, disabled with ESC ( B.
  48  * They differ in the handling of JIS X 0201 characters (halfwidth Katakana)
  49  * in the conversion direction Unicode -> CP5022x:
  50  *   * CP50220 maps the halfwidth Katakana to fullwidth Katakana characters.
  51  *   * CP50221 contains the JIS X 0201 halfwidth Katakana characters,
  52  *     enabled with ESC ( I, disabled with ESC ( B.
  53  *   * CP50222 contains the JIS X 0201 halfwidth Katakana characters,
  54  *     enabled with ESC ( J 0x0E, disabled with ESC ( B.
  55  * In the conversion direction CP5022x -> Unicode, all three operate the same:
  56  *   - ESC ( I is supported and understood.
  57  *   - ESC ( J 0x0E is not accepted.  (Tested on Windows XP SP3.)
  58  * Conclusion:
  59  *   - CP50222 should not be used, because the multibyte sequence that it
  60  *     produces cannot be parsed by either of the three encodings.
  61  *   - CP50221 is preferrable to CP50220, because it can faithfully represent
  62  *     the halfwidth Katakana characters.
  63  * We therefore implement CP50221.  As an extension, in the mbtowc conversion
  64  * direction, we support also ESC ( J 0x0E, just in case.
  65  */
  66
  67 #include "cp50221_0208_ext.h"
  68 #include "cp50221_0212_ext.h"
  69
  70 #define ESC 0x1b
  71 #define SO  0x0e
  72 #define SI  0x0f
  73
  74 /*
  75  * The state can be one of the following values.
  76  */
  77 #define STATE_ASCII             0  /* Esc ( B */
  78 #define STATE_JISX0201ROMAN     1  /* Esc ( J */ /* only in mbtowc direction */
  79 #define STATE_JISX0201KATAKANA  2  /* Esc ( I */
  80 #define STATE_JISX0208MS        3  /* Esc $ @ or Esc $ B */
  81 #define STATE_JISX0212MS        4  /* Esc $ ( D */
  82
  83 static int
  84 iso2022_jpms_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n)
  85 {
  86   state_t state = conv->istate;
  87   int count = 0;
  88   unsigned char c;
  89   for (;;) {
  90     c = *s;
  91     if (c == ESC) {
  92       if (n < count+3)
  93         goto none;
  94       if (s[1] == '(') {
  95         if (s[2] == 'B') {
  96           state = STATE_ASCII;
  97           s += 3; count += 3;
  98           if (n < count+1)
  99             goto none;
 100           continue;
 101         }
 102         if (s[2] == 'I') {
 103           state = STATE_JISX0201KATAKANA;
 104           s += 3; count += 3;
 105           if (n < count+1)
 106             goto none;
 107           continue;
 108         }
 109         if (s[2] == 'J') {
 110           state = STATE_JISX0201ROMAN;
 111           s += 3; count += 3;
 112           if (n < count+1)
 113             goto none;
 114           continue;
 115         }
 116         goto ilseq;
 117       }
 118       if (s[1] == '$') {
 119         if (s[2] == '@' || s[2] == 'B') {
 120           /* We don't distinguish JIS X 0208-1978 and JIS X 0208-1983. */
 121           state = STATE_JISX0208MS;
 122           s += 3; count += 3;
 123           if (n < count+1)
 124             goto none;
 125           continue;
 126         }
 127         if (s[2] == '(') {
 128           if (n < count+4)
 129             goto none;
 130           if (s[3] == 'D') {
 131             state = STATE_JISX0212MS;
 132             s += 4; count += 4;
 133             if (n < count+1)
 134               goto none;
 135             continue;
 136           }
 137         }
 138         goto ilseq;
 139       }
 140       goto ilseq;
 141     }
 142     if (c == SO) {
 143       if (state == STATE_JISX0201ROMAN)
 144         state = STATE_JISX0201KATAKANA;
 145       s += 1; count += 1;
 146       if (n < count+1)
 147         goto none;
 148       continue;
 149     }
 150     if (c == SI) {
 151       if (state == STATE_JISX0201KATAKANA)
 152         state = STATE_JISX0201ROMAN;
 153       s += 1; count += 1;
 154       if (n < count+1)
 155         goto none;
 156       continue;
 157     }
 158     break;
 159   }
 160   switch (state) {
 161     case STATE_ASCII:
 162       if (c < 0x80) {
 163         int ret = ascii_mbtowc(conv,pwc,s,1);
 164         if (ret == RET_ILSEQ)
 165           goto ilseq;
 166         if (ret != 1) abort();
 167         conv->istate = state;
 168         return count+1;
 169       } else
 170         goto ilseq;
 171     case STATE_JISX0201ROMAN:
 172       if (c < 0x80) {
 173         int ret = jisx0201_mbtowc(conv,pwc,s,1);
 174         if (ret == RET_ILSEQ)
 175           goto ilseq;
 176         if (ret != 1) abort();
 177         conv->istate = state;
 178         return count+1;
 179       } else
 180         goto ilseq;
 181     case STATE_JISX0201KATAKANA:
 182       if (c < 0x80) {
 183         unsigned char buf = c+0x80;
 184         int ret = jisx0201_mbtowc(conv,pwc,&buf,1);
 185         if (ret == RET_ILSEQ)
 186           goto ilseq;
 187         if (ret != 1) abort();
 188         conv->istate = state;
 189         return count+1;
 190       } else
 191         goto ilseq;
 192     case STATE_JISX0208MS:
 193       if (n < count+2)
 194         goto none;
 195       if (s[0] < 0x80 && s[1] < 0x80) {
 196         int ret;
 197         if (s[0] < 0x75) {
 198           if (s[0] == 0x2d) {
 199             /* Extension of JIS X 0208.  */
 200             if (s[1] >= 0x21 && s[1] <= 0x79) {
 201               unsigned char i = (s[1] - 0x21) + 1;
 202               ret = cp50221_0208_ext_mbtowc(conv,pwc,&i,1);
 203               if (ret == 1)
 204                 ret = 2;
 205             } else
 206               ret = RET_ILSEQ;
 207           } else {
 208             /* JIS X 0208.  */
 209             ret = jisx0208_mbtowc(conv,pwc,s,2);
 210           }
 211         } else {
 212           /* Extension of JIS X 0208.
 213              0x{75..7E}{21..8E} maps to U+E000..U+E3AB.
 214              But some rows maps to characters present in CP932.  */
 215           if (s[0] <= 0x7e && (s[1] >= 0x21 && s[1] <= 0x7e)) {
 216             unsigned short wc = 0xfffd;
 217             if (s[0] >= 0x79 && s[0] <= 0x7c)
 218               wc = cp932ext_2uni_pageed[(s[0] - 0x79) * 94 + (s[1] - 0x21)];
 219             if (wc == 0xfffd)
 220               wc = (s[0] - 0x75) * 94 + (s[1] - 0x21) + 0xe000;
 221             *pwc = wc;
 222             ret = 2;
 223           } else
 224             ret = RET_ILSEQ;
 225         }
 226         if (ret == RET_ILSEQ)
 227           goto ilseq;
 228         if (ret != 2) abort();
 229         conv->istate = state;
 230         return count+2;
 231       } else
 232         goto ilseq;
 233     case STATE_JISX0212MS:
 234       if (n < count+2)
 235         goto none;
 236       if (s[0] < 0x80 && s[1] < 0x80) {
 237         int ret;
 238         if (s[0] < 0x73) {
 239           /* JIS X 0212.  */
 240           ret = jisx0212_mbtowc(conv,pwc,s,2);
 241         } else {
 242           if (s[0] < 0x75) {
 243             /* Extension of JIS X 0212.  */
 244             if (s[1] >= 0x21 && s[1] <= 0x7e) {
 245               unsigned char i = (s[0] - 0x73) * 94 + (s[1] - 0x21) + 1;
 246               ret = cp50221_0212_ext_mbtowc(conv,pwc,&i,1);
 247               if (ret == 1)
 248                 ret = 2;
 249             } else
 250               ret = RET_ILSEQ;
 251           } else {
 252             /* Extension of JIS X 0208.
 253                0x{75..7E}{21..8E} maps to U+E3AC..U+E757.  */
 254             if (s[0] <= 0x7e && (s[1] >= 0x21 && s[1] <= 0x7e)) {
 255               *pwc = (s[0] - 0x75) * 94 + (s[1] - 0x21) + 0xe3ac;
 256               ret = 2;
 257             } else
 258               ret = RET_ILSEQ;
 259           }
 260         }
 261         if (ret == RET_ILSEQ)
 262           goto ilseq;
 263         if (ret != 2) abort();
 264         conv->istate = state;
 265         return count+2;
 266       } else
 267         goto ilseq;
 268     default: abort();
 269   }
 270
 271 none:
 272   conv->istate = state;
 273   return RET_TOOFEW(count);
 274
 275 ilseq:
 276   conv->istate = state;
 277   return RET_SHIFT_ILSEQ(count);
 278 }
 279
 280 static int
 281 iso2022_jpms_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, size_t n)
 282 {
 283   state_t state = conv->ostate;
 284   unsigned char buf[2];
 285   int ret;
 286
 287   /* Try ASCII. */
 288   ret = ascii_wctomb(conv,buf,wc,1);
 289   if (ret != RET_ILUNI) {
 290     if (ret != 1) abort();
 291     if (buf[0] < 0x80) {
 292       int count = (state == STATE_ASCII ? 1 : 4);
 293       if (n < count)
 294         return RET_TOOSMALL;
 295       if (state != STATE_ASCII) {
 296         r[0] = ESC;
 297         r[1] = '(';
 298         r[2] = 'B';
 299         r += 3;
 300         state = STATE_ASCII;
 301       }
 302       r[0] = buf[0];
 303       conv->ostate = state;
 304       return count;
 305     }
 306   }
 307
 308   /* Try JIS X 0201-1976 Katakana. */
 309   ret = jisx0201_wctomb(conv,buf,wc,1);
 310   if (ret != RET_ILUNI) {
 311     if (ret != 1) abort();
 312     if (buf[0] >= 0x80) {
 313       int count = (state == STATE_JISX0201KATAKANA ? 1 : 4);
 314       if (n < count)
 315         return RET_TOOSMALL;
 316       if (state != STATE_JISX0201KATAKANA) {
 317         r[0] = ESC;
 318         r[1] = '(';
 319         r[2] = 'I';
 320         r += 3;
 321         state = STATE_JISX0201KATAKANA;
 322       }
 323       r[0] = buf[0]-0x80;
 324       conv->ostate = state;
 325       return count;
 326     }
 327   }
 328
 329   /* Try JIS X 0208-1990, in place of JIS X 0208-1978 and JIS X 0208-1983,
 330      and the extensions mentioned above.  */
 331   if (wc >= 0xe000 && wc < 0xe3ac) {
 332     unsigned short i = wc - 0xe000;
 333     buf[0] = (i / 94) + 0x75;
 334     buf[1] = (i % 94) + 0x21;
 335     ret = 2;
 336   } else {
 337     ret = jisx0208_wctomb(conv,buf,wc,2);
 338     if (ret == RET_ILUNI) {
 339       /* Extension of JIS X 0208.  */
 340       unsigned char i;
 341       ret = cp50221_0208_ext_wctomb(conv,&i,wc,1);
 342       if (ret == 1) {
 343         buf[0] = 0x2d;
 344         buf[1] = i-1 + 0x21;
 345         ret = 2;
 346       } else if (wc == 0x663B) {
 347         buf[0] = 0x7a;
 348         buf[1] = 0x36;
 349         ret = 2;
 350       } else if (wc == 0xffe2) {
 351         buf[0] = 0x7c;
 352         buf[1] = 0x7b;
 353         ret = 2;
 354       } else if (wc == 0xffe4) {
 355         buf[0] = 0x7c;
 356         buf[1] = 0x7c;
 357         ret = 2;
 358       }
 359     }
 360   }
 361   if (ret != RET_ILUNI) {
 362     if (ret != 2) abort();
 363     if (buf[0] < 0x80 && buf[1] < 0x80) {
 364       int count = (state == STATE_JISX0208MS ? 2 : 5);
 365       if (n < count)
 366         return RET_TOOSMALL;
 367       if (state != STATE_JISX0208MS) {
 368         r[0] = ESC;
 369         r[1] = '$';
 370         r[2] = 'B';
 371         r += 3;
 372         state = STATE_JISX0208MS;
 373       }
 374       r[0] = buf[0];
 375       r[1] = buf[1];
 376       conv->ostate = state;
 377       return count;
 378     }
 379   }
 380
 381   /* Try JIS X 0212-1990 and the extensions mentioned above. */
 382   if (wc >= 0xe3ac && wc < 0xe758) {
 383     unsigned short i = wc - 0xe3ac;
 384     buf[0] = (i / 94) + 0x75;
 385     buf[1] = (i % 94) + 0x21;
 386     ret = 2;
 387   } else {
 388     ret = jisx0212_wctomb(conv,buf,wc,2);
 389     if (ret == RET_ILUNI) {
 390       /* Extension of JIS X 0212.  */
 391       unsigned char i;
 392       ret = cp50221_0212_ext_wctomb(conv,&i,wc,1);
 393       if (ret == 1) {
 394         i -= 1;
 395         buf[0] = (i / 94) + 0x73;
 396         buf[1] = (i % 94) + 0x21;
 397         ret = 2;
 398       }
 399     }
 400   }
 401   if (ret != RET_ILUNI) {
 402     if (ret != 2) abort();
 403     if (buf[0] < 0x80 && buf[1] < 0x80) {
 404       int count = (state == STATE_JISX0212MS ? 2 : 6);
 405       if (n < count)
 406         return RET_TOOSMALL;
 407       if (state != STATE_JISX0212MS) {
 408         r[0] = ESC;
 409         r[1] = '$';
 410         r[2] = '(';
 411         r[3] = 'D';
 412         r += 4;
 413         state = STATE_JISX0212MS;
 414       }
 415       r[0] = buf[0];
 416       r[1] = buf[1];
 417       conv->ostate = state;
 418       return count;
 419     }
 420   }
 421
 422   return RET_ILUNI;
 423 }
 424
 425 static int
 426 iso2022_jpms_reset (conv_t conv, unsigned char *r, size_t n)
 427 {
 428   state_t state = conv->ostate;
 429   if (state != STATE_ASCII) {
 430     if (n < 3)
 431       return RET_TOOSMALL;
 432     r[0] = ESC;
 433     r[1] = '(';
 434     r[2] = 'B';
 435     /* conv->ostate = 0; will be done by the caller */
 436     return 3;
 437   } else
 438     return 0;
 439 }
 440
 441 #undef STATE_JISX0212MS
 442 #undef STATE_JISX0208MS
 443 #undef STATE_JISX0201KATAKANA
 444 #undef STATE_JISX0201ROMAN
 445 #undef STATE_ASCII