iconvdata/eucjp.c

   1 /* Mapping tables for EUC-JP handling.
   2    Copyright (C) 1998 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
   5
   6    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Library General Public License as
   8    published by the Free Software Foundation; either version 2 of the
   9    License, or (at your option) any later version.
  10
  11    The GNU C Library is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14    Library General Public License for more details.
  15
  16    You should have received a copy of the GNU Library General Public
  17    License along with the GNU C Library; see the file COPYING.LIB.  If not,
  18    write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  19    Boston, MA 02111-1307, USA.  */
  20
  21 #include <gconv.h>
  22 #include <stdint.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25 #include <wchar.h>
  26 #include <jis0201.h>
  27 #include <jis0208.h>
  28 #include <jis0212.h>
  29
  30 /* Direction of the transformation.  */
  31 enum direction
  32 {
  33   illegal,
  34   to_eucjp,
  35   from_eucjp
  36 };
  37
  38 struct eucjp_data
  39 {
  40   enum direction dir;
  41 };
  42
  43
  44 int
  45 gconv_init (struct gconv_step *step)
  46 {
  47   /* Determine which direction.  */
  48   struct eucjp_data *new_data;
  49   enum direction dir;
  50   int result;
  51
  52   if (strcasestr (step->from_name, "EUC-JP") != NULL)
  53     dir = from_eucjp;
  54   else if (strcasestr (step->to_name, "EUC-JP") != NULL)
  55     dir = to_eucjp;
  56   else
  57     dir = illegal;
  58
  59   result = GCONV_NOCONV;
  60   if (dir != illegal
  61       && ((new_data
  62            = (struct eucjp_data *) malloc (sizeof (struct eucjp_data)))
  63           != NULL))
  64     {
  65       new_data->dir = dir;
  66       step->data = new_data;
  67       result = GCONV_OK;
  68     }
  69
  70   return result;
  71 }
  72
  73
  74 void
  75 gconv_end (struct gconv_step *data)
  76 {
  77   free (data->data);
  78 }
  79
  80
  81 int
  82 gconv (struct gconv_step *step, struct gconv_step_data *data,
  83        const char *inbuf, size_t *inbufsize, size_t *written, int do_flush)
  84 {
  85   struct gconv_step *next_step = step + 1;
  86   struct gconv_step_data *next_data = data + 1;
  87   gconv_fct fct = next_step->fct;
  88   size_t do_write;
  89   int result;
  90
  91   /* If the function is called with no input this means we have to reset
  92      to the initial state.  The possibly partly converted input is
  93      dropped.  */
  94   if (do_flush)
  95     {
  96       do_write = 0;
  97
  98       /* Call the steps down the chain if there are any.  */
  99       if (data->is_last)
 100         result = GCONV_OK;
 101       else
 102         {
 103           struct gconv_step *next_step = step + 1;
 104           struct gconv_step_data *next_data = data + 1;
 105
 106           result = (*fct) (next_step, next_data, NULL, 0, written, 1);
 107
 108           /* Clear output buffer.  */
 109           data->outbufavail = 0;
 110         }
 111     }
 112   else
 113     {
 114       enum direction dir = ((struct eucjp_data *) step->data)->dir;
 115
 116       do_write = 0;
 117
 118       do
 119         {
 120           result = GCONV_OK;
 121
 122           if (dir == from_eucjp)
 123             {
 124               size_t inchars = *inbufsize;
 125               size_t outwchars = data->outbufavail;
 126               char *outbuf = data->outbuf;
 127               size_t cnt = 0;
 128
 129               while (cnt < inchars
 130                      && (outwchars + sizeof (wchar_t) <= data->outbufsize))
 131                 {
 132                   int inchar = (unsigned char) inbuf[cnt];
 133                   wchar_t ch;
 134
 135                   if (inchar <= 0x7f)
 136                     ch = (wchar_t) inchar;
 137                   else if ((inchar <= 0xa0 || inchar > 0xfe)
 138                            && inchar != 0x8e && inchar != 0x8f)
 139                       /* This is illegal.  */
 140                       ch = L'\0';
 141                   else
 142                     {
 143                       /* Two or more byte character.  First test whether the
 144                          next character is also available.  */
 145                       int inchar2;
 146
 147                       if (cnt + 1 >= inchars)
 148                         {
 149                           /* The second character is not available.  Store
 150                              the intermediate result.  */
 151                           result = GCONV_INCOMPLETE_INPUT;
 152                           break;
 153                         }
 154
 155                       inchar2 = (unsigned char) inbuf[++cnt];
 156
 157                       /* All second bytes of a multibyte character must be
 158                          >= 0xa1. */
 159                       if (inchar2 < 0xa1)
 160                         {
 161                           /* This is an illegal character.  */
 162                           --cnt;
 163                           result = GCONV_ILLEGAL_INPUT;
 164                           break;
 165                         }
 166
 167                       if (inchar == '\x8e')
 168                         /* This is code set 2: half-width katakana.  */
 169                         ch = jisx0201_to_ucs4 (inchar2);
 170                       else if (inchar == '\x8f')
 171                         {
 172                           /* This is code set 3: JIS X 0212-1990.  */
 173                           const char *endp = &inbuf[cnt];
 174
 175                           ch = jisx0212_to_ucs4 (&endp, 1 + inchars - cnt,
 176                                                  0x80);
 177                           cnt = endp - inbuf;
 178                         }
 179                       else
 180                         {
 181                           /* This is code set 1: JIS X 0208.  */
 182                           const char *endp = &inbuf[cnt - 1];
 183
 184                           ch = jisx0208_to_ucs4 (&endp, 2 + inchars - cnt,
 185                                                  0x80);
 186                           if (ch != L'\0')
 187                             ++cnt;
 188                         }
 189
 190                       if (ch == UNKNOWN_10646_CHAR)
 191                          ch = L'\0';
 192
 193                       if (ch == L'\0')
 194                         --cnt;
 195                     }
 196
 197                   if (ch == L'\0' && inbuf[cnt] != '\0')
 198                     {
 199                       /* This is an illegal character.  */
 200                       result = GCONV_ILLEGAL_INPUT;
 201                       break;
 202                     }
 203
 204                   *((wchar_t *) (outbuf + outwchars)) = ch;
 205                   ++do_write;
 206                   outwchars += sizeof (wchar_t);
 207                   ++cnt;
 208                 }
 209               *inbufsize -= cnt;
 210               data->outbufavail = outwchars;
 211             }
 212           else
 213             {
 214               size_t inwchars = *inbufsize;
 215               size_t outchars = data->outbufavail;
 216               char *outbuf = data->outbuf;
 217               size_t cnt = 0;
 218               int extra = 0;
 219
 220               while (inwchars >= cnt + sizeof (wchar_t)
 221                      && outchars < data->outbufsize)
 222                 {
 223                   wchar_t ch = *((wchar_t *) (inbuf + cnt));
 224
 225                   if (ch <= L'\x7f')
 226                     /* It's plain ASCII.  */
 227                     outbuf[outchars] = ch;
 228                   else
 229                     {
 230                       /* Try the JIS character sets.  */
 231                       size_t found;
 232
 233                       found = ucs4_to_jisx0201 (ch, &outbuf[outchars]);
 234
 235                       if (found == UNKNOWN_10646_CHAR)
 236                         {
 237                           /* No JIS 0201 character.  */
 238                           found = ucs4_to_jisx0208 (ch, &outbuf[outchars],
 239                                                     (data->outbufsize
 240                                                      - outchars));
 241                           if (found == 0)
 242                             {
 243                               /* We ran out of space.  */
 244                               extra = 2;
 245                               break;
 246                             }
 247                           else if (found != UNKNOWN_10646_CHAR)
 248                             {
 249                               /* It's a JIS 0208 character, adjust it for
 250                                  EUC-JP.  */
 251                               outbuf[outchars++] += 0x80;
 252                               outbuf[outchars] += 0x80;
 253                             }
 254                           else
 255                             {
 256                               /* No JIS 0208 character.  */
 257                               found = ucs4_to_jisx0212 (ch, &outbuf[outchars],
 258                                                         (data->outbufsize
 259                                                          - outchars));
 260
 261                               if (found == 0)
 262                                 {
 263                                   /* We ran out of space.  */
 264                                   extra = 2;
 265                                   break;
 266                                 }
 267                               else if (found != UNKNOWN_10646_CHAR)
 268                                 {
 269                                   /* It's a JIS 0212 character, adjust it for
 270                                      EUC-JP.  */
 271                                   outbuf[outchars++] += 0x80;
 272                                   outbuf[outchars] += 0x80;
 273                                 }
 274                               else
 275                                 /* Illegal character.  */
 276                                 break;
 277                             }
 278                         }
 279                     }
 280
 281                   ++do_write;
 282                   ++outchars;
 283                   cnt += sizeof (wchar_t);
 284                 }
 285               *inbufsize -= cnt;
 286               data->outbufavail = outchars;
 287
 288               if (outchars + extra < data->outbufsize)
 289                 {
 290                   /* If there is still room in the output buffer something
 291                      is wrong with the input.  */
 292                   if (inwchars >= cnt + sizeof (wchar_t))
 293                     {
 294                       /* An error occurred.  */
 295                       result = GCONV_ILLEGAL_INPUT;
 296                       break;
 297                     }
 298                   if (inwchars != cnt)
 299                     {
 300                       /* There are some unprocessed bytes at the end of the
 301                          input buffer.  */
 302                       result = GCONV_INCOMPLETE_INPUT;
 303                       break;
 304                     }
 305                 }
 306             }
 307
 308           if (result != GCONV_OK)
 309             break;
 310
 311           if (data->is_last)
 312             {
 313               /* This is the last step.  */
 314               result = (*inbufsize > (dir == from_eucjp
 315                                       ? 0 : sizeof (wchar_t) - 1)
 316                         ? GCONV_FULL_OUTPUT : GCONV_EMPTY_INPUT);
 317               break;
 318             }
 319
 320           /* Status so far.  */
 321           result = GCONV_EMPTY_INPUT;
 322
 323           if (data->outbufavail > 0)
 324             {
 325               /* Call the functions below in the chain.  */
 326               size_t newavail = data->outbufavail;
 327
 328               result = (*fct) (next_step, next_data, data->outbuf, &newavail,
 329                                written, 0);
 330
 331               /* Correct the output buffer.  */
 332               if (newavail != data->outbufavail && newavail > 0)
 333                 {
 334                   memmove (data->outbuf,
 335                            &data->outbuf[data->outbufavail - newavail],
 336                            newavail);
 337                   data->outbufavail = newavail;
 338                 }
 339             }
 340         }
 341       while (*inbufsize > 0 && result == GCONV_EMPTY_INPUT);
 342     }
 343
 344   if (written != NULL && data->is_last)
 345     *written = do_write;
 346
 347   return result;
 348 }