iconvdata/ibm933.c

   1 /* Conversion from and to IBM933.
   2    Copyright (C) 2000-2002 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Masahide Washizawa <washi@yamato.ibm.co.jp>, 2000.
   5
   6    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Lesser General Public
   8    License as published by the Free Software Foundation; either
   9    version 2.1 of the 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    Lesser General Public License for more details.
  15
  16    You should have received a copy of the GNU Lesser General Public
  17    License along with the GNU C Library; if not, write to the Free
  18    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  19    02111-1307 USA.  */
  20
  21 #include <dlfcn.h>
  22 #include <stdint.h>
  23 #include <wchar.h>
  24 #include <byteswap.h>
  25 #include "ibm933.h"
  26
  27 /* The shift sequences for this charset (it does not use ESC).  */
  28 #define SI              0x0F  /* Shift In, host code to turn DBCS off.  */
  29 #define SO              0x0E  /* Shift Out, host code to turn DBCS on.  */
  30
  31 /* Definitions used in the body of the `gconv' function.  */
  32 #define CHARSET_NAME    "IBM933//"
  33 #define FROM_LOOP       from_ibm933
  34 #define TO_LOOP         to_ibm933
  35 #define FROM_LOOP_MIN_NEEDED_FROM       1
  36 #define FROM_LOOP_MAX_NEEDED_FROM       2
  37 #define FROM_LOOP_MIN_NEEDED_TO         4
  38 #define FROM_LOOP_MAX_NEEDED_TO         4
  39 #define TO_LOOP_MIN_NEEDED_FROM         4
  40 #define TO_LOOP_MAX_NEEDED_FROM         4
  41 #define TO_LOOP_MIN_NEEDED_TO           1
  42 #define TO_LOOP_MAX_NEEDED_TO           3
  43 #define PREPARE_LOOP \
  44   int save_curcs;                                                             \
  45   int *curcsp = &data->__statep->__count;
  46 #define EXTRA_LOOP_ARGS         , curcsp
  47
  48 /* Definitions of initialization and destructor function.  */
  49 #define DEFINE_INIT     1
  50 #define DEFINE_FINI     1
  51
  52
  53 /* Since this is a stateful encoding we have to provide code which resets
  54    the output state to the initial state.  This has to be done during the
  55    flushing.  */
  56 #define EMIT_SHIFT_TO_INIT \
  57   if ((data->__statep->__count & ~7) != sb)                                   \
  58     {                                                                         \
  59       if (FROM_DIRECTION)                                                     \
  60         data->__statep->__count &= 7;                                         \
  61       else                                                                    \
  62         {                                                                     \
  63           /* We are not in the initial state.  To switch back we have         \
  64              to emit `SI'.  */                                                \
  65           if (__builtin_expect (outbuf >= outend, 0))                         \
  66             /* We don't have enough room in the output buffer.  */            \
  67             status = __GCONV_FULL_OUTPUT;                                     \
  68           else                                                                \
  69             {                                                                 \
  70               /* Write out the shift sequence.  */                            \
  71               *outbuf++ = SI;                                                 \
  72               data->__statep->__count &= 7;                                   \
  73             }                                                                 \
  74         }                                                                     \
  75     }
  76
  77
  78 /* Since we might have to reset input pointer we must be able to save
  79    and retore the state.  */
  80 #define SAVE_RESET_STATE(Save) \
  81   if (Save)                                                                   \
  82     save_curcs = *curcsp;                                                     \
  83   else                                                                        \
  84     *curcsp = save_curcs
  85
  86
  87 /* Current codeset type.  */
  88 enum
  89 {
  90   sb = 0,
  91   db = 64
  92 };
  93
  94 /* First, define the conversion function from IBM-933 to UCS4.  */
  95 #define MIN_NEEDED_INPUT        FROM_LOOP_MIN_NEEDED_FROM
  96 #define MAX_NEEDED_INPUT        FROM_LOOP_MAX_NEEDED_FROM
  97 #define MIN_NEEDED_OUTPUT       FROM_LOOP_MIN_NEEDED_TO
  98 #define MAX_NEEDED_OUTPUT       FROM_LOOP_MAX_NEEDED_TO
  99 #define LOOPFCT                 FROM_LOOP
 100 #define BODY \
 101   {                                                                           \
 102     uint32_t ch = *inptr;                                                     \
 103     uint32_t res;                                                             \
 104                                                                               \
 105     if (__builtin_expect (ch, 0) == SO)                                       \
 106       {                                                                       \
 107         /* Shift OUT, change to DBCS converter.  */                           \
 108         if (curcs == db)                                                      \
 109           {                                                                   \
 110             result = __GCONV_ILLEGAL_INPUT;                                   \
 111             break;                                                            \
 112           }                                                                   \
 113         curcs = db;                                                           \
 114         ++inptr;                                                              \
 115         continue;                                                             \
 116       }                                                                       \
 117     else if (__builtin_expect (ch, 0) == SI)                                  \
 118       {                                                                       \
 119         /* Shift IN, change to SBCS converter.  */                            \
 120         if (curcs == sb)                                                      \
 121           {                                                                   \
 122             result = __GCONV_ILLEGAL_INPUT;                                   \
 123             break;                                                            \
 124           }                                                                   \
 125         curcs = sb;                                                           \
 126         ++inptr;                                                              \
 127         continue;                                                             \
 128       }                                                                       \
 129                                                                               \
 130     if (curcs == sb)                                                          \
 131       {                                                                       \
 132         /* Use the IBM933 table for single byte.  */                          \
 133         res = __ibm933sb_to_ucs4[ch];                                         \
 134         if (__builtin_expect (res, L'\1') == L'\0' && ch != '\0')             \
 135           {                                                                   \
 136             /* This is an illegal character.  */                              \
 137             if (! ignore_errors_p ())                                         \
 138               {                                                               \
 139                 result = __GCONV_ILLEGAL_INPUT;                               \
 140                 break;                                                        \
 141               }                                                               \
 142             ++*irreversible;                                                  \
 143           }                                                                   \
 144         else                                                                  \
 145           {                                                                   \
 146             put32 (outptr, res);                                              \
 147             outptr += 4;                                                      \
 148           }                                                                   \
 149         ++inptr;                                                              \
 150       }                                                                       \
 151     else                                                                      \
 152       {                                                                       \
 153         const struct gap *rp2 = __ibm933db_to_ucs4_idx;                       \
 154                                                                               \
 155         assert (curcs == db);                                                 \
 156                                                                               \
 157         /* Use the IBM933 table for double byte.  */                          \
 158         if (__builtin_expect (inptr + 1 >= inend, 0))                         \
 159           {                                                                   \
 160             /* The second character is not available.  Store the              \
 161                intermediate result. */                                        \
 162             result = __GCONV_INCOMPLETE_INPUT;                                \
 163             break;                                                            \
 164           }                                                                   \
 165                                                                               \
 166         ch = (ch * 0x100) + inptr[1];                                         \
 167         while (ch > rp2->end)                                                 \
 168           ++rp2;                                                              \
 169                                                                               \
 170         if (__builtin_expect (rp2 == NULL, 0)                                 \
 171             || __builtin_expect (ch < rp2->start, 0)                          \
 172             || (res = __ibm933db_to_ucs4[ch + rp2->idx],                      \
 173                 __builtin_expect (res, L'\1') == L'\0' && ch != '\0'))        \
 174           {                                                                   \
 175             /* This is an illegal character.  */                              \
 176             if (! ignore_errors_p ())                                         \
 177               {                                                               \
 178                 result = __GCONV_ILLEGAL_INPUT;                               \
 179                 break;                                                        \
 180               }                                                               \
 181             ++*irreversible;                                                  \
 182             inptr += 2;                                                       \
 183             continue;                                                         \
 184           }                                                                   \
 185         else                                                                  \
 186           {                                                                   \
 187             put32 (outptr, res);                                              \
 188             outptr += 4;                                                      \
 189             inptr += 2;                                                       \
 190           }                                                                   \
 191       }                                                                       \
 192   }
 193 #define LOOP_NEED_FLAGS
 194 #define EXTRA_LOOP_DECLS        , int *curcsp
 195 #define INIT_PARAMS             int curcs = *curcsp & ~7
 196 #define UPDATE_PARAMS           *curcsp = curcs
 197 #include <iconv/loop.c>
 198
 199 /* Next, define the other direction.  */
 200 #define MIN_NEEDED_INPUT        TO_LOOP_MIN_NEEDED_FROM
 201 #define MAX_NEEDED_INPUT        TO_LOOP_MAX_NEEDED_FROM
 202 #define MIN_NEEDED_OUTPUT       TO_LOOP_MIN_NEEDED_TO
 203 #define MAX_NEEDED_OUTPUT       TO_LOOP_MAX_NEEDED_TO
 204 #define LOOPFCT                 TO_LOOP
 205 #define BODY \
 206   {                                                                           \
 207     uint32_t ch = get32 (inptr);                                              \
 208     const struct gap *rp1 = __ucs4_to_ibm933sb_idx;                           \
 209     const struct gap *rp2 = __ucs4_to_ibm933db_idx;                           \
 210     const char *cp;                                                           \
 211                                                                               \
 212     if (__builtin_expect (ch >= 0xffff, 0))                                   \
 213       {                                                                       \
 214         UNICODE_TAG_HANDLER (ch, 4);                                          \
 215                                                                               \
 216         if (! ignore_errors_p ())                                             \
 217           {                                                                   \
 218             result = __GCONV_ILLEGAL_INPUT;                                   \
 219             break;                                                            \
 220           }                                                                   \
 221         ++*irreversible;                                                      \
 222         inptr += 4;                                                           \
 223         continue;                                                             \
 224       }                                                                       \
 225                                                                               \
 226     while (ch > rp1->end)                                                     \
 227       ++rp1;                                                                  \
 228                                                                               \
 229     /* Use the UCS4 table for single byte.  */                                \
 230     if (__builtin_expect (ch < rp1->start, 0)                                 \
 231         || (cp = __ucs4_to_ibm933sb[ch + rp1->idx],                           \
 232             __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0'))          \
 233       {                                                                       \
 234         /* Use the UCS4 table for double byte.  */                            \
 235         while (ch > rp2->end)                                                 \
 236           ++rp2;                                                              \
 237                                                                               \
 238         if (__builtin_expect (ch < rp2->start, 0)                             \
 239             || (cp = __ucs4_to_ibm933db[ch + rp2->idx],                       \
 240                 __builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0'))        \
 241           {                                                                   \
 242             /* This is an illegal character.  */                              \
 243             if (! ignore_errors_p ())                                         \
 244               {                                                               \
 245                 result = __GCONV_ILLEGAL_INPUT;                               \
 246                 break;                                                        \
 247               }                                                               \
 248             ++*irreversible;                                                  \
 249           }                                                                   \
 250         else                                                                  \
 251           {                                                                   \
 252             if (curcs == sb)                                                  \
 253               {                                                               \
 254                 if (__builtin_expect (outptr + 1 > outend, 0))                \
 255                   {                                                           \
 256                     result = __GCONV_FULL_OUTPUT;                             \
 257                     break;                                                    \
 258                   }                                                           \
 259                 *outptr++ = SO;                                               \
 260                 curcs = db;                                                   \
 261               }                                                               \
 262                                                                               \
 263             if (__builtin_expect (outptr + 2 > outend, 0))                    \
 264               {                                                               \
 265                 result = __GCONV_FULL_OUTPUT;                                 \
 266                 break;                                                        \
 267               }                                                               \
 268             *outptr++ = cp[0];                                                \
 269             *outptr++ = cp[1];                                                \
 270           }                                                                   \
 271       }                                                                       \
 272     else                                                                      \
 273       {                                                                       \
 274         if (curcs == db)                                                      \
 275           {                                                                   \
 276             if (__builtin_expect (outptr + 1 > outend, 0))                    \
 277               {                                                               \
 278                 result = __GCONV_FULL_OUTPUT;                                 \
 279                 break;                                                        \
 280               }                                                               \
 281             *outptr++ = SI;                                                   \
 282           }                                                                   \
 283                                                                               \
 284         if (__builtin_expect (outptr + 1 > outend, 0))                        \
 285           {                                                                   \
 286             result = __GCONV_FULL_OUTPUT;                                     \
 287             break;                                                            \
 288           }                                                                   \
 289         *outptr++ = cp[0];                                                    \
 290         curcs = sb;                                                           \
 291       }                                                                       \
 292                                                                               \
 293     /* Now that we wrote the output increment the input pointer.  */          \
 294     inptr += 4;                                                               \
 295   }
 296 #define LOOP_NEED_FLAGS
 297 #define EXTRA_LOOP_DECLS        , int *curcsp
 298 #define INIT_PARAMS             int curcs = *curcsp & ~7
 299 #define UPDATE_PARAMS           *curcsp = curcs
 300 #include <iconv/loop.c>
 301
 302 /* Now define the toplevel functions.  */
 303 #include <iconv/skeleton.c>