iconvdata/ibm939.c

   1 /* Conversion to and from IBM939.
   2    Copyright (C) 2000-2002, 2005, 2008 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 "ibm939.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    "IBM939//"
  33 #define FROM_LOOP       from_ibm939
  34 #define TO_LOOP         to_ibm939
  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-939 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 IBM939 table for single byte.  */                          \
 133         res = __ibm939sb_to_ucs4[ch];                                         \
 134         if (__builtin_expect (res == L'\0', 0) && ch != '\0')                 \
 135           {                                                                   \
 136             /* This is an illegal character.  */                              \
 137             STANDARD_FROM_LOOP_ERR_HANDLER (1);                               \
 138           }                                                                   \
 139         else                                                                  \
 140           {                                                                   \
 141             put32 (outptr, res);                                              \
 142             outptr += 4;                                                      \
 143           }                                                                   \
 144         ++inptr;                                                              \
 145       }                                                                       \
 146     else                                                                      \
 147       {                                                                       \
 148         /* Use the IBM939 table for double byte.  */                          \
 149         const struct gap *rp2 = __ibm939db_to_ucs4_idx;                       \
 150                                                                               \
 151         assert (curcs == db);                                                 \
 152                                                                               \
 153         if (__builtin_expect (inptr + 1 >= inend, 0))                         \
 154           {                                                                   \
 155             /* The second character is not available.  Store the              \
 156                intermediate result. */                                        \
 157             result = __GCONV_INCOMPLETE_INPUT;                                \
 158             break;                                                            \
 159           }                                                                   \
 160                                                                               \
 161         ch = (ch * 0x100) + inptr[1];                                         \
 162         while (ch > rp2->end)                                                 \
 163           ++rp2;                                                              \
 164                                                                               \
 165         if (__builtin_expect (rp2 == NULL, 0)                                 \
 166             || __builtin_expect (ch < rp2->start, 0)                          \
 167             || (res = __ibm939db_to_ucs4[ch + rp2->idx],                      \
 168                 __builtin_expect (res, L'\1') == L'\0' && ch != '\0'))        \
 169           {                                                                   \
 170             /* This is an illegal character.  */                              \
 171             STANDARD_FROM_LOOP_ERR_HANDLER (2);                               \
 172           }                                                                   \
 173         else                                                                  \
 174           {                                                                   \
 175             put32 (outptr, res);                                              \
 176             outptr += 4;                                                      \
 177           }                                                                   \
 178         inptr += 2;                                                           \
 179       }                                                                       \
 180   }
 181 #define LOOP_NEED_FLAGS
 182 #define EXTRA_LOOP_DECLS        , int *curcsp
 183 #define INIT_PARAMS             int curcs = *curcsp & ~7
 184 #define UPDATE_PARAMS           *curcsp = curcs
 185 #include <iconv/loop.c>
 186
 187 /* Next, define the other direction */
 188 #define MIN_NEEDED_INPUT        TO_LOOP_MIN_NEEDED_FROM
 189 #define MAX_NEEDED_INPUT        TO_LOOP_MAX_NEEDED_FROM
 190 #define MIN_NEEDED_OUTPUT       TO_LOOP_MIN_NEEDED_TO
 191 #define MAX_NEEDED_OUTPUT       TO_LOOP_MAX_NEEDED_TO
 192 #define LOOPFCT                 TO_LOOP
 193 #define BODY \
 194   {                                                                           \
 195     uint32_t ch = get32 (inptr);                                              \
 196     const struct gap *rp1 = __ucs4_to_ibm939sb_idx;                           \
 197     const struct gap *rp2 = __ucs4_to_ibm939db_idx;                           \
 198     const char *cp;                                                           \
 199                                                                               \
 200     if (__builtin_expect (ch >= 0xffff, 0))                                   \
 201       {                                                                       \
 202         UNICODE_TAG_HANDLER (ch, 4);                                          \
 203         goto ibm939_invalid_char;                                             \
 204       }                                                                       \
 205                                                                               \
 206     while (ch > rp1->end)                                                     \
 207       ++rp1;                                                                  \
 208                                                                               \
 209     /* Use the UCS4 table for single byte.  */                                \
 210     if (__builtin_expect (ch < rp1->start, 0)                                 \
 211         || (cp = __ucs4_to_ibm939sb[ch + rp1->idx],                           \
 212             __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0'))          \
 213       {                                                                       \
 214         /* Use the UCS4 table for double byte.  */                            \
 215         while (ch > rp2->end)                                                 \
 216           ++rp2;                                                              \
 217                                                                               \
 218         if (__builtin_expect (ch < rp2->start, 0)                             \
 219             || (cp = __ucs4_to_ibm939db[ch + rp2->idx],                       \
 220                 __builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0'))        \
 221           {                                                                   \
 222             /* This is an illegal character.  */                              \
 223           ibm939_invalid_char:                                                \
 224             STANDARD_TO_LOOP_ERR_HANDLER (4);                                 \
 225           }                                                                   \
 226         else                                                                  \
 227           {                                                                   \
 228             if (curcs == sb)                                                  \
 229               {                                                               \
 230                 if (__builtin_expect (outptr + 1 > outend, 0))                \
 231                   {                                                           \
 232                     result = __GCONV_FULL_OUTPUT;                             \
 233                     break;                                                    \
 234                   }                                                           \
 235                 *outptr++ = SO;                                               \
 236                 curcs = db;                                                   \
 237               }                                                               \
 238                                                                               \
 239             if (__builtin_expect (outptr + 2 > outend, 0))                    \
 240               {                                                               \
 241                 result = __GCONV_FULL_OUTPUT;                                 \
 242                 break;                                                        \
 243               }                                                               \
 244             *outptr++ = cp[0];                                                \
 245             *outptr++ = cp[1];                                                \
 246           }                                                                   \
 247       }                                                                       \
 248     else                                                                      \
 249       {                                                                       \
 250         if (curcs == db)                                                      \
 251           {                                                                   \
 252             if (__builtin_expect (outptr + 1 > outend, 0))                    \
 253               {                                                               \
 254                 result = __GCONV_FULL_OUTPUT;                                 \
 255                 break;                                                        \
 256               }                                                               \
 257             *outptr++ = SI;                                                   \
 258           }                                                                   \
 259                                                                               \
 260         if (__builtin_expect (outptr + 1 > outend, 0))                        \
 261           {                                                                   \
 262             result = __GCONV_FULL_OUTPUT;                                     \
 263             break;                                                            \
 264           }                                                                   \
 265         if (ch == 0x7e)                                                       \
 266           *outptr++ = 0xa0;                                                   \
 267         else if (ch == 0x5c)                                                  \
 268           *outptr++ = 0xb2;                                                   \
 269         else                                                                  \
 270           *outptr++ = cp[0];                                                  \
 271         curcs = sb;                                                           \
 272       }                                                                       \
 273                                                                               \
 274     /* Now that we wrote the output increment the input pointer.  */          \
 275     inptr += 4;                                                               \
 276   }
 277 #define LOOP_NEED_FLAGS
 278 #define EXTRA_LOOP_DECLS        , int *curcsp
 279 #define INIT_PARAMS             int curcs = *curcsp & ~7
 280 #define REINIT_PARAMS           curcs = *curcsp & ~7
 281 #define UPDATE_PARAMS           *curcsp = curcs
 282 #include <iconv/loop.c>
 283
 284 /* Now define the toplevel functions. */
 285 #include <iconv/skeleton.c>