lib/cp932.h

   1 /*
   2  * Copyright (C) 1999-2002, 2005, 2016 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  * CP932
  22  */
  23
  24 /*
  25  * Microsoft CP932 is a slightly extended version of SHIFT_JIS.
  26  * The differences between the EASTASIA/JIS/SHIFTJIS.TXT and the
  27  * VENDORS/MICSFT/WINDOWS/CP932.TXT tables found on ftp.unicode.org are
  28  * as follows:
  29  *
  30  * 1. CP932 uses ASCII, not JISX0201 Roman.
  31  *
  32  * 2. Some characters in the JISX0208 range are defined differently:
  33  *
  34  *     code   SHIFTJIS.TXT                   CP932.TXT
  35  *    0x815F  0x005C # REVERSE SOLIDUS       0xFF3C # FULLWIDTH REVERSE SOLIDUS
  36  *    0x8160  0x301C # WAVE DASH             0xFF5E # FULLWIDTH TILDE
  37  *    0x8161  0x2016 # DOUBLE VERTICAL LINE  0x2225 # PARALLEL TO
  38  *    0x817C  0x2212 # MINUS SIGN            0xFF0D # FULLWIDTH HYPHEN-MINUS
  39  *    0x8191  0x00A2 # CENT SIGN             0xFFE0 # FULLWIDTH CENT SIGN
  40  *    0x8192  0x00A3 # POUND SIGN            0xFFE1 # FULLWIDTH POUND SIGN
  41  *    0x81CA  0x00AC # NOT SIGN              0xFFE2 # FULLWIDTH NOT SIGN
  42  *
  43  *    We don't implement the latter 6 of these changes, only the first one.
  44  *    SHIFTJIS.TXT makes more sense. However, as a compromise with user
  45  *    expectation, we implement the middle 5 of these changes in the
  46  *    Unicode to CP932 direction. We don't implement the last one at all,
  47  *    because it would collide with the mapping of 0xFA54.
  48  *
  49  * 3. A few new rows. See cp932ext.h.
  50  *
  51  * Many variants of CP932 (in GNU libc, JDK, OSF/1, Windows-2000, ICU) also
  52  * add:
  53  *
  54  * 4. Private area mappings:
  55  *
  56  *              code                 Unicode
  57  *    0x{F0..F9}{40..7E,80..FC}  U+E000..U+E757
  58  *
  59  * We add them too because, although there are backward compatibility problems
  60  * when a character from a private area is moved to an official Unicode code
  61  * point, they are useful for some people in practice.
  62  */
  63
  64 #include "cp932ext.h"
  65
  66 /*
  67    Conversion between SJIS codes (s1,s2) and JISX0208 codes (c1,c2):
  68    Example. (s1,s2) = 0x8140, (c1,c2) = 0x2121.
  69    0x81 <= s1 <= 0x9F || 0xE0 <= s1 <= 0xEA,
  70    0x40 <= s2 <= 0x7E || 0x80 <= s2 <= 0xFC,
  71    0x21 <= c1 <= 0x74, 0x21 <= c2 <= 0x7E.
  72    Invariant:
  73      94*2*(s1 < 0xE0 ? s1-0x81 : s1-0xC1) + (s2 < 0x80 ? s2-0x40 : s2-0x41)
  74      = 94*(c1-0x21)+(c2-0x21)
  75    Conversion (s1,s2) -> (c1,c2):
  76      t1 := (s1 < 0xE0 ? s1-0x81 : s1-0xC1)
  77      t2 := (s2 < 0x80 ? s2-0x40 : s2-0x41)
  78      c1 := 2*t1 + (t2 < 0x5E ? 0 : 1) + 0x21
  79      c2 := (t2 < 0x5E ? t2 : t2-0x5E) + 0x21
  80    Conversion (c1,c2) -> (s1,s2):
  81      t1 := (c1 - 0x21) >> 1
  82      t2 := ((c1 - 0x21) & 1) * 0x5E + (c2 - 0x21)
  83      s1 := (t1 < 0x1F ? t1+0x81 : t1+0xC1)
  84      s2 := (t2 < 0x3F ? t2+0x40 : t2+0x41)
  85  */
  86
  87 static int
  88 cp932_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n)
  89 {
  90   unsigned char c = *s;
  91   if (c < 0x80)
  92     return ascii_mbtowc(conv,pwc,s,n);
  93   else if (c >= 0xa1 && c <= 0xdf)
  94     return jisx0201_mbtowc(conv,pwc,s,n);
  95   else {
  96     unsigned char s1, s2;
  97     s1 = c;
  98     if ((s1 >= 0x81 && s1 <= 0x9f && s1 != 0x87) || (s1 >= 0xe0 && s1 <= 0xea)) {
  99       if (n < 2)
 100         return RET_TOOFEW(0);
 101       s2 = s[1];
 102       if ((s2 >= 0x40 && s2 <= 0x7e) || (s2 >= 0x80 && s2 <= 0xfc)) {
 103         unsigned char t1 = (s1 < 0xe0 ? s1-0x81 : s1-0xc1);
 104         unsigned char t2 = (s2 < 0x80 ? s2-0x40 : s2-0x41);
 105         unsigned char buf[2];
 106         buf[0] = 2*t1 + (t2 < 0x5e ? 0 : 1) + 0x21;
 107         buf[1] = (t2 < 0x5e ? t2 : t2-0x5e) + 0x21;
 108         return jisx0208_mbtowc(conv,pwc,buf,2);
 109       }
 110     } else if ((s1 == 0x87) || (s1 >= 0xed && s1 <= 0xee) || (s1 >= 0xfa)) {
 111       if (n < 2)
 112         return RET_TOOFEW(0);
 113       return cp932ext_mbtowc(conv,pwc,s,2);
 114     } else if (s1 >= 0xf0 && s1 <= 0xf9) {
 115       /* User-defined range. See
 116        * Ken Lunde's "CJKV Information Processing", table 4-66, p. 206. */
 117       if (n < 2)
 118         return RET_TOOFEW(0);
 119       s2 = s[1];
 120       if ((s2 >= 0x40 && s2 <= 0x7e) || (s2 >= 0x80 && s2 <= 0xfc)) {
 121         *pwc = 0xe000 + 188*(s1 - 0xf0) + (s2 < 0x80 ? s2-0x40 : s2-0x41);
 122         return 2;
 123       }
 124     }
 125     return RET_ILSEQ;
 126   }
 127 }
 128
 129 static int
 130 cp932_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, size_t n)
 131 {
 132   unsigned char buf[2];
 133   int ret;
 134
 135   /* Try ASCII. */
 136   ret = ascii_wctomb(conv,buf,wc,1);
 137   if (ret != RET_ILUNI) {
 138     unsigned char c;
 139     if (ret != 1) abort();
 140     c = buf[0];
 141     if (c < 0x80) {
 142       r[0] = c;
 143       return 1;
 144     }
 145   }
 146
 147   /* Try JIS X 0201-1976 Katakana. */
 148   ret = jisx0201_wctomb(conv,buf,wc,1);
 149   if (ret != RET_ILUNI) {
 150     unsigned char c;
 151     if (ret != 1) abort();
 152     c = buf[0];
 153     if (c >= 0xa1 && c <= 0xdf) {
 154       r[0] = c;
 155       return 1;
 156     }
 157   }
 158
 159   /* Try JIS X 0208-1990. */
 160   ret = jisx0208_wctomb(conv,buf,wc,2);
 161   if (ret != RET_ILUNI) {
 162     unsigned char c1, c2;
 163     if (ret != 2) abort();
 164     if (n < 2)
 165       return RET_TOOSMALL;
 166     c1 = buf[0];
 167     c2 = buf[1];
 168     if ((c1 >= 0x21 && c1 <= 0x74) && (c2 >= 0x21 && c2 <= 0x7e)) {
 169       unsigned char t1 = (c1 - 0x21) >> 1;
 170       unsigned char t2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21);
 171       r[0] = (t1 < 0x1f ? t1+0x81 : t1+0xc1);
 172       r[1] = (t2 < 0x3f ? t2+0x40 : t2+0x41);
 173       return 2;
 174     }
 175   }
 176
 177   /* Try CP932 extensions. */
 178   ret = cp932ext_wctomb(conv,buf,wc,2);
 179   if (ret != RET_ILUNI) {
 180     if (ret != 2) abort();
 181     if (n < 2)
 182       return RET_TOOSMALL;
 183     r[0] = buf[0];
 184     r[1] = buf[1];
 185     return 2;
 186   }
 187
 188   /* User-defined range. See
 189    * Ken Lunde's "CJKV Information Processing", table 4-66, p. 206. */
 190   if (wc >= 0xe000 && wc < 0xe758) {
 191     unsigned char c1, c2;
 192     if (n < 2)
 193       return RET_TOOSMALL;
 194     c1 = (unsigned int) (wc - 0xe000) / 188;
 195     c2 = (unsigned int) (wc - 0xe000) % 188;
 196     r[0] = c1+0xf0;
 197     r[1] = (c2 < 0x3f ? c2+0x40 : c2+0x41);
 198     return 2;
 199   }
 200
 201   /* Irreversible mappings.  */
 202   if (wc == 0xff5e) {
 203     if (n < 2)
 204       return RET_TOOSMALL;
 205     r[0] = 0x81;
 206     r[1] = 0x60;
 207     return 2;
 208   }
 209   if (wc == 0x2225) {
 210     if (n < 2)
 211       return RET_TOOSMALL;
 212     r[0] = 0x81;
 213     r[1] = 0x61;
 214     return 2;
 215   }
 216   if (wc == 0xff0d) {
 217     if (n < 2)
 218       return RET_TOOSMALL;
 219     r[0] = 0x81;
 220     r[1] = 0x7c;
 221     return 2;
 222   }
 223   if (wc == 0xffe0) {
 224     if (n < 2)
 225       return RET_TOOSMALL;
 226     r[0] = 0x81;
 227     r[1] = 0x91;
 228     return 2;
 229   }
 230   if (wc == 0xffe1) {
 231     if (n < 2)
 232       return RET_TOOSMALL;
 233     r[0] = 0x81;
 234     r[1] = 0x92;
 235     return 2;
 236   }
 237
 238   return RET_ILUNI;
 239 }