lib/euc_jisx0213.h

   1 /*
   2  * Copyright (C) 1999-2002, 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  * EUC-JISX0213
  22  */
  23
  24 /* The structure of EUC-JISX0213 is as follows:
  25
  26    0x00..0x7F: ASCII
  27
  28    0x8E{A1..FE}: JISX0201 Katakana, with prefix 0x8E, offset by +0x80.
  29
  30    0x8F{A1..FE}{A1..FE}: JISX0213 plane 2, with prefix 0x8F, offset by +0x8080.
  31
  32    0x{A1..FE}{A1..FE}: JISX0213 plane 1, offset by +0x8080.
  33
  34    Note that some JISX0213 characters are not contained in Unicode 3.2
  35    and are therefore best represented as sequences of Unicode characters.
  36 */
  37
  38 #include "jisx0213.h"
  39 #include "flushwc.h"
  40
  41 static int
  42 euc_jisx0213_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n)
  43 {
  44   ucs4_t last_wc = conv->istate;
  45   if (last_wc) {
  46     /* Output the buffered character. */
  47     conv->istate = 0;
  48     *pwc = last_wc;
  49     return 0; /* Don't advance the input pointer. */
  50   } else {
  51     unsigned char c = *s;
  52     if (c < 0x80) {
  53       /* Plain ASCII character. */
  54       *pwc = (ucs4_t) c;
  55       return 1;
  56     } else {
  57       if ((c >= 0xa1 && c <= 0xfe) || c == 0x8e || c == 0x8f) {
  58         /* Two or three byte character. */
  59         if (n >= 2) {
  60           unsigned char c2 = s[1];
  61           if (c2 >= 0xa1 && c2 <= 0xfe) {
  62             if (c == 0x8e) {
  63               /* Half-width katakana. */
  64               if (c2 <= 0xdf) {
  65                 *pwc = c2 + 0xfec0;
  66                 return 2;
  67               }
  68             } else {
  69               ucs4_t wc;
  70               if (c == 0x8f) {
  71                 /* JISX 0213 plane 2. */
  72                 if (n >= 3) {
  73                   unsigned char c3 = s[2];
  74                   wc = jisx0213_to_ucs4(0x200-0x80+c2,c3^0x80);
  75                 } else
  76                   return RET_TOOFEW(0);
  77               } else {
  78                 /* JISX 0213 plane 1. */
  79                 wc = jisx0213_to_ucs4(0x100-0x80+c,c2^0x80);
  80               }
  81               if (wc) {
  82                 if (wc < 0x80) {
  83                   /* It's a combining character. */
  84                   ucs4_t wc1 = jisx0213_to_ucs_combining[wc - 1][0];
  85                   ucs4_t wc2 = jisx0213_to_ucs_combining[wc - 1][1];
  86                   /* We cannot output two Unicode characters at once. So,
  87                      output the first character and buffer the second one. */
  88                   *pwc = wc1;
  89                   conv->istate = wc2;
  90                 } else
  91                   *pwc = wc;
  92                 return (c == 0x8f ? 3 : 2);
  93               }
  94             }
  95           }
  96         } else
  97           return RET_TOOFEW(0);
  98       }
  99       return RET_ILSEQ;
 100     }
 101   }
 102 }
 103
 104 #define euc_jisx0213_flushwc normal_flushwc
 105
 106 /* Composition tables for each of the relevant combining characters.  */
 107 static const struct { unsigned short base; unsigned short composed; } euc_jisx0213_comp_table_data[] = {
 108 #define euc_jisx0213_comp_table02e5_idx 0
 109 #define euc_jisx0213_comp_table02e5_len 1
 110   { 0xabe4, 0xabe5 }, /* 0x12B65 = 0x12B64 U+02E5 */
 111 #define euc_jisx0213_comp_table02e9_idx (euc_jisx0213_comp_table02e5_idx+euc_jisx0213_comp_table02e5_len)
 112 #define euc_jisx0213_comp_table02e9_len 1
 113   { 0xabe0, 0xabe6 }, /* 0x12B66 = 0x12B60 U+02E9 */
 114 #define euc_jisx0213_comp_table0300_idx (euc_jisx0213_comp_table02e9_idx+euc_jisx0213_comp_table02e9_len)
 115 #define euc_jisx0213_comp_table0300_len 5
 116   { 0xa9dc, 0xabc4 }, /* 0x12B44 = 0x1295C U+0300 */
 117   { 0xabb8, 0xabc8 }, /* 0x12B48 = 0x12B38 U+0300 */
 118   { 0xabb7, 0xabca }, /* 0x12B4A = 0x12B37 U+0300 */
 119   { 0xabb0, 0xabcc }, /* 0x12B4C = 0x12B30 U+0300 */
 120   { 0xabc3, 0xabce }, /* 0x12B4E = 0x12B43 U+0300 */
 121 #define euc_jisx0213_comp_table0301_idx (euc_jisx0213_comp_table0300_idx+euc_jisx0213_comp_table0300_len)
 122 #define euc_jisx0213_comp_table0301_len 4
 123   { 0xabb8, 0xabc9 }, /* 0x12B49 = 0x12B38 U+0301 */
 124   { 0xabb7, 0xabcb }, /* 0x12B4B = 0x12B37 U+0301 */
 125   { 0xabb0, 0xabcd }, /* 0x12B4D = 0x12B30 U+0301 */
 126   { 0xabc3, 0xabcf }, /* 0x12B4F = 0x12B43 U+0301 */
 127 #define euc_jisx0213_comp_table309a_idx (euc_jisx0213_comp_table0301_idx+euc_jisx0213_comp_table0301_len)
 128 #define euc_jisx0213_comp_table309a_len 14
 129   { 0xa4ab, 0xa4f7 }, /* 0x12477 = 0x1242B U+309A */
 130   { 0xa4ad, 0xa4f8 }, /* 0x12478 = 0x1242D U+309A */
 131   { 0xa4af, 0xa4f9 }, /* 0x12479 = 0x1242F U+309A */
 132   { 0xa4b1, 0xa4fa }, /* 0x1247A = 0x12431 U+309A */
 133   { 0xa4b3, 0xa4fb }, /* 0x1247B = 0x12433 U+309A */
 134   { 0xa5ab, 0xa5f7 }, /* 0x12577 = 0x1252B U+309A */
 135   { 0xa5ad, 0xa5f8 }, /* 0x12578 = 0x1252D U+309A */
 136   { 0xa5af, 0xa5f9 }, /* 0x12579 = 0x1252F U+309A */
 137   { 0xa5b1, 0xa5fa }, /* 0x1257A = 0x12531 U+309A */
 138   { 0xa5b3, 0xa5fb }, /* 0x1257B = 0x12533 U+309A */
 139   { 0xa5bb, 0xa5fc }, /* 0x1257C = 0x1253B U+309A */
 140   { 0xa5c4, 0xa5fd }, /* 0x1257D = 0x12544 U+309A */
 141   { 0xa5c8, 0xa5fe }, /* 0x1257E = 0x12548 U+309A */
 142   { 0xa6f5, 0xa6f8 }, /* 0x12678 = 0x12675 U+309A */
 143 };
 144
 145 static int
 146 euc_jisx0213_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, size_t n)
 147 {
 148   int count = 0;
 149   unsigned short lasttwo = conv->ostate;
 150
 151   if (lasttwo) {
 152     /* Attempt to combine the last character with this one. */
 153     unsigned int idx;
 154     unsigned int len;
 155
 156     if (wc == 0x02e5)
 157       idx = euc_jisx0213_comp_table02e5_idx,
 158       len = euc_jisx0213_comp_table02e5_len;
 159     else if (wc == 0x02e9)
 160       idx = euc_jisx0213_comp_table02e9_idx,
 161       len = euc_jisx0213_comp_table02e9_len;
 162     else if (wc == 0x0300)
 163       idx = euc_jisx0213_comp_table0300_idx,
 164       len = euc_jisx0213_comp_table0300_len;
 165     else if (wc == 0x0301)
 166       idx = euc_jisx0213_comp_table0301_idx,
 167       len = euc_jisx0213_comp_table0301_len;
 168     else if (wc == 0x309a)
 169       idx = euc_jisx0213_comp_table309a_idx,
 170       len = euc_jisx0213_comp_table309a_len;
 171     else
 172       goto not_combining;
 173
 174     do
 175       if (euc_jisx0213_comp_table_data[idx].base == lasttwo)
 176         break;
 177     while (++idx, --len > 0);
 178
 179     if (len > 0) {
 180       /* Output the combined character. */
 181       if (n >= 2) {
 182         lasttwo = euc_jisx0213_comp_table_data[idx].composed;
 183         r[0] = (lasttwo >> 8) & 0xff;
 184         r[1] = lasttwo & 0xff;
 185         conv->ostate = 0;
 186         return 2;
 187       } else
 188         return RET_TOOSMALL;
 189     }
 190
 191   not_combining:
 192     /* Output the buffered character. */
 193     if (n < 2)
 194       return RET_TOOSMALL;
 195     r[0] = (lasttwo >> 8) & 0xff;
 196     r[1] = lasttwo & 0xff;
 197     r += 2;
 198     count = 2;
 199   }
 200
 201   if (wc < 0x80) {
 202     /* Plain ASCII character. */
 203     if (n > count) {
 204       r[0] = (unsigned char) wc;
 205       conv->ostate = 0;
 206       return count+1;
 207     } else
 208       return RET_TOOSMALL;
 209   } else if (wc >= 0xff61 && wc <= 0xff9f) {
 210     /* Half-width katakana. */
 211     if (n >= count+2) {
 212       r[0] = 0x8e;
 213       r[1] = wc - 0xfec0;
 214       conv->ostate = 0;
 215       return count+2;
 216     } else
 217       return RET_TOOSMALL;
 218   } else {
 219     unsigned short jch = ucs4_to_jisx0213(wc);
 220     if (jch != 0) {
 221       if (jch & 0x0080) {
 222         /* A possible match in comp_table_data. We have to buffer it. */
 223         /* We know it's a JISX 0213 plane 1 character. */
 224         if (jch & 0x8000) abort();
 225         conv->ostate = jch | 0x8080;
 226         return count+0;
 227       }
 228       if (jch & 0x8000) {
 229         /* JISX 0213 plane 2. */
 230         if (n >= count+3) {
 231           r[0] = 0x8f;
 232           r[1] = (jch >> 8) | 0x80;
 233           r[2] = (jch & 0xff) | 0x80;
 234           conv->ostate = 0;
 235           return count+3;
 236         } else
 237           return RET_TOOSMALL;
 238       } else {
 239         /* JISX 0213 plane 1. */
 240         if (n >= count+2) {
 241           r[0] = (jch >> 8) | 0x80;
 242           r[1] = (jch & 0xff) | 0x80;
 243           conv->ostate = 0;
 244           return count+2;
 245         } else
 246           return RET_TOOSMALL;
 247       }
 248     }
 249     return RET_ILUNI;
 250   }
 251 }
 252
 253 static int
 254 euc_jisx0213_reset (conv_t conv, unsigned char *r, size_t n)
 255 {
 256   state_t lasttwo = conv->ostate;
 257
 258   if (lasttwo) {
 259     if (n < 2)
 260       return RET_TOOSMALL;
 261     r[0] = (lasttwo >> 8) & 0xff;
 262     r[1] = lasttwo & 0xff;
 263     /* conv->ostate = 0; will be done by the caller */
 264     return 2;
 265   } else
 266     return 0;
 267 }