lib/gb18030_2022.h

   1 /*
   2  * Copyright (C) 1999-2001, 2005, 2012, 2016, 2023 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  * GB18030:2022
  22  */
  23
  24 /*
  25  * GB18030, in the version from 2022, is like the version from 2005,
  26  * except that a few mappings to the Unicode PUA have been replaced with
  27  * mappings to assigned Unicode characters.
  28  */
  29
  30 static int
  31 gb18030_2022_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n)
  32 {
  33   int ret;
  34
  35   /* Code set 0 (ASCII) */
  36   if (*s < 0x80)
  37     return ascii_mbtowc(conv,pwc,s,n);
  38
  39   /* Code set 1 (GBK extended) */
  40   ret = gbk_mbtowc(conv,pwc,s,n);
  41   if (ret != RET_ILSEQ)
  42     return ret;
  43
  44   ret = gb18030_2022_ext_mbtowc(conv,pwc,s,n);
  45   if (ret != RET_ILSEQ)
  46     return ret;
  47
  48   /* Code set 2 (remainder of Unicode U+0000..U+FFFF), including
  49      User-defined characters, two-byte part of range U+E766..U+E864 */
  50   ret = gb18030_2022_uni_mbtowc(conv,pwc,s,n);
  51   if (ret != RET_ILSEQ)
  52     return ret;
  53   /* User-defined characters range U+E000..U+E765 */
  54   {
  55     unsigned char c1 = s[0];
  56     if ((c1 >= 0xaa && c1 <= 0xaf) || (c1 >= 0xf8 && c1 <= 0xfe)) {
  57       if (n >= 2) {
  58         unsigned char c2 = s[1];
  59         if (c2 >= 0xa1 && c2 <= 0xfe) {
  60           *pwc = 0xe000 + 94 * (c1 >= 0xf8 ? c1 - 0xf2 : c1 - 0xaa) + (c2 - 0xa1);
  61           return 2;
  62         }
  63       } else
  64         return RET_TOOFEW(0);
  65     } else if (c1 >= 0xa1 && c1 <= 0xa7) {
  66       if (n >= 2) {
  67         unsigned char c2 = s[1];
  68         if (c2 >= 0x40 && c2 <= 0xa1 && c2 != 0x7f) {
  69           *pwc = 0xe4c6 + 96 * (c1 - 0xa1) + c2 - (c2 >= 0x80 ? 0x41 : 0x40);
  70           return 2;
  71         }
  72       } else
  73         return RET_TOOFEW(0);
  74     }
  75   }
  76
  77   /* Code set 3 (Unicode U+10000..U+10FFFF) */
  78   {
  79     unsigned char c1 = s[0];
  80     if (c1 >= 0x90 && c1 <= 0xe3) {
  81       if (n >= 2) {
  82         unsigned char c2 = s[1];
  83         if (c2 >= 0x30 && c2 <= 0x39) {
  84           if (n >= 3) {
  85             unsigned char c3 = s[2];
  86             if (c3 >= 0x81 && c3 <= 0xfe) {
  87               if (n >= 4) {
  88                 unsigned char c4 = s[3];
  89                 if (c4 >= 0x30 && c4 <= 0x39) {
  90                   unsigned int i = (((c1 - 0x90) * 10 + (c2 - 0x30)) * 126 + (c3 - 0x81)) * 10 + (c4 - 0x30);
  91                   if (i >= 0 && i < 0x100000) {
  92                     *pwc = (ucs4_t) (0x10000 + i);
  93                     return 4;
  94                   }
  95                 }
  96                 return RET_ILSEQ;
  97               }
  98               return RET_TOOFEW(0);
  99             }
 100             return RET_ILSEQ;
 101           }
 102           return RET_TOOFEW(0);
 103         }
 104         return RET_ILSEQ;
 105       }
 106       return RET_TOOFEW(0);
 107     }
 108     return RET_ILSEQ;
 109   }
 110 }
 111
 112 static const unsigned short gb18030_2022_pua2charset[23*3] = {
 113 /* Unicode range   GB18030 range */
 114   0xe766, 0xe76b,  0xa2ab, /*.. 0xa2b0, */
 115   0xe76d, 0xe76d,  0xa2e4,
 116   0xe76e, 0xe76f,  0xa2ef, /*.. 0xa2f0, */
 117   0xe770, 0xe771,  0xa2fd, /*.. 0xa2fe, */
 118   0xe772, 0xe77c,  0xa4f4, /*.. 0xa4fe, */
 119   0xe77d, 0xe784,  0xa5f7, /*.. 0xa5fe, */
 120   0xe785, 0xe78c,  0xa6b9, /*.. 0xa6c0, */
 121   0xe797, 0xe79f,  0xa6f6, /*.. 0xa6fe, */
 122   0xe7a0, 0xe7ae,  0xa7c2, /*.. 0xa7d0, */
 123   0xe7af, 0xe7bb,  0xa7f2, /*.. 0xa7fe, */
 124   0xe7bc, 0xe7c6,  0xa896, /*.. 0xa8a0, */
 125   0xe7c9, 0xe7cc,  0xa8c1, /*.. 0xa8c4, */
 126   0xe7cd, 0xe7e1,  0xa8ea, /*.. 0xa8fe, */
 127   0xe7e2, 0xe7e2,  0xa958,
 128   0xe7e3, 0xe7e3,  0xa95b,
 129   0xe7e4, 0xe7e6,  0xa95d, /*.. 0xa95f, */
 130   0xe7f4, 0xe800,  0xa997, /*.. 0xa9a3, */
 131   0xe801, 0xe80f,  0xa9f0, /*.. 0xa9fe, */
 132   0xe810, 0xe814,  0xd7fa, /*.. 0xd7fe, */
 133   0xe816, 0xe818,  0xfe51, /*.. 0xfe53, */
 134   0xe831, 0xe831,  0xfe6c,
 135   0xe83b, 0xe83b,  0xfe76,
 136   0xe855, 0xe855,  0xfe91,
 137 };
 138
 139 static int
 140 gb18030_2022_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, size_t n)
 141 {
 142   int ret;
 143
 144   /* Code set 0 (ASCII) */
 145   ret = ascii_wctomb(conv,r,wc,n);
 146   if (ret != RET_ILUNI)
 147     return ret;
 148
 149   /* Code set 1 (GBK extended) */
 150   ret = gbk_wctomb(conv,r,wc,n);
 151   if (ret != RET_ILUNI)
 152     return ret;
 153
 154   ret = gb18030ext_wctomb(conv,r,wc,n);
 155   if (ret != RET_ILUNI)
 156     return ret;
 157
 158   /* Code set 2 (remainder of Unicode U+0000..U+FFFF) */
 159   if (wc >= 0xe000 && wc <= 0xe864) {
 160     if (n >= 2) {
 161       if (wc < 0xe766) {
 162         /* User-defined characters range U+E000..U+E765 */
 163         if (wc < 0xe4c6) {
 164           unsigned int i = wc - 0xe000;
 165           r[1] = (i % 94) + 0xa1; i = i / 94;
 166           r[0] = (i < 6 ? i + 0xaa : i + 0xf2);
 167           return 2;
 168         } else {
 169           unsigned int i = wc - 0xe4c6;
 170           r[0] = (i / 96) + 0xa1; i = i % 96;
 171           r[1] = i + (i >= 0x3f ? 0x41 : 0x40);
 172           return 2;
 173         }
 174       } else {
 175         /* User-defined characters, two-byte part of range U+E766..U+E864 */
 176         unsigned int k1 = 0;
 177         unsigned int k2 = 23;
 178         /* Invariant: We know that if wc occurs in Unicode interval in
 179            gb18030_2022_pua2charset, it does so at a k with  k1 <= k < k2. */
 180         while (k1 < k2) {
 181           unsigned int k = (k1 + k2) / 2;
 182           if (wc < gb18030_2022_pua2charset[k*3+0])
 183             k2 = k;
 184           else if (wc > gb18030_2022_pua2charset[k*3+1])
 185             k1 = k + 1;
 186           else {
 187             unsigned short c =
 188               gb18030_2022_pua2charset[k*3+2] + (wc - gb18030_2022_pua2charset[k*3+0]);
 189             r[0] = (c >> 8);
 190             r[1] = (c & 0xff);
 191             return 2;
 192           }
 193         }
 194       }
 195     } else
 196       return RET_TOOSMALL;
 197   }
 198   ret = gb18030_2022_uni_wctomb(conv,r,wc,n);
 199   if (ret != RET_ILUNI)
 200     return ret;
 201
 202   /* Code set 3 (Unicode U+10000..U+10FFFF) */
 203   if (n >= 4) {
 204     if (wc >= 0x10000 && wc < 0x110000) {
 205       unsigned int i = wc - 0x10000;
 206       r[3] = (i % 10) + 0x30; i = i / 10;
 207       r[2] = (i % 126) + 0x81; i = i / 126;
 208       r[1] = (i % 10) + 0x30; i = i / 10;
 209       r[0] = i + 0x90;
 210       return 4;
 211     }
 212     return RET_ILUNI;
 213   }
 214   return RET_TOOSMALL;
 215 }