iconvdata/ibm930.c

   1 /* Conversion from and to IBM930.
   2    Copyright (C) 2000-2015 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, see
  18    <http://www.gnu.org/licenses/>.  */
  19
  20 #include <dlfcn.h>
  21 #include <stdint.h>
  22 #include <wchar.h>
  23 #include <byteswap.h>
  24 #include "ibm930.h"
  25
  26 /* The shift sequences for this charset (it does not use ESC). */
  27 #define SI              0x0F  /* Shift In, host code to turn DBCS off.  */
  28 #define SO              0x0E  /* Shift Out, host code to turn DBCS on.  */
  29
  30 /* Definitions used in the body of the `gconv' function.  */
  31 #define CHARSET_NAME    "IBM930//"
  32 #define FROM_LOOP       from_ibm930
  33 #define TO_LOOP         to_ibm930
  34 #define ONE_DIRECTION                   0
  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 (__glibc_unlikely (outbuf >= outend))                            \
  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
  95 /* First, define the conversion function from IBM-930 to UCS4.  */
  96 #define MIN_NEEDED_INPUT        FROM_LOOP_MIN_NEEDED_FROM
  97 #define MAX_NEEDED_INPUT        FROM_LOOP_MAX_NEEDED_FROM
  98 #define MIN_NEEDED_OUTPUT       FROM_LOOP_MIN_NEEDED_TO
  99 #define MAX_NEEDED_OUTPUT       FROM_LOOP_MAX_NEEDED_TO
 100 #define LOOPFCT                 FROM_LOOP
 101 #define BODY \
 102   {                                                                           \
 103     uint32_t ch = *inptr;                                                     \
 104     uint32_t res;                                                             \
 105                                                                               \
 106     if (__builtin_expect (ch, 0) == SO)                                       \
 107       {                                                                       \
 108         /* Shift OUT, change to DBCS converter.  */                           \
 109         if (curcs == db)                                                      \
 110           {                                                                   \
 111             result = __GCONV_ILLEGAL_INPUT;                                   \
 112             break;                                                            \
 113           }                                                                   \
 114         curcs = db;                                                           \
 115         ++inptr;                                                              \
 116         continue;                                                             \
 117       }                                                                       \
 118     else if (__builtin_expect (ch, 0) == SI)                                  \
 119       {                                                                       \
 120         /* Shift IN, change to SBCS converter */                              \
 121         if (curcs == sb)                                                      \
 122           {                                                                   \
 123             result = __GCONV_ILLEGAL_INPUT;                                   \
 124             break;                                                            \
 125           }                                                                   \
 126         curcs = sb;                                                           \
 127         ++inptr;                                                              \
 128         continue;                                                             \
 129       }                                                                       \
 130                                                                               \
 131     if (curcs == sb)                                                          \
 132       {                                                                       \
 133         /* Use the IBM930 table for single byte.  */                          \
 134         res = __ibm930sb_to_ucs4[ch];                                         \
 135         if (__builtin_expect (res, L'\1') == L'\0' && ch != '\0')             \
 136           {                                                                   \
 137             /* This is an illegal character.  */                              \
 138             STANDARD_FROM_LOOP_ERR_HANDLER (1);                               \
 139           }                                                                   \
 140         else                                                                  \
 141           {                                                                   \
 142             put32 (outptr, res);                                              \
 143             outptr += 4;                                                      \
 144           }                                                                   \
 145         ++inptr;                                                              \
 146       }                                                                       \
 147     else                                                                      \
 148       {                                                                       \
 149         /* Use the IBM930 table for double byte. */                           \
 150         const struct gap *rp2 = __ibm930db_to_ucs4_idx;                       \
 151                                                                               \
 152         assert (curcs == db);                                                 \
 153                                                                               \
 154         if (__glibc_unlikely (inptr + 1 >= inend))                            \
 155           {                                                                   \
 156             /* The second character is not available.  Store the              \
 157                intermediate result. */                                        \
 158             result = __GCONV_INCOMPLETE_INPUT;                                \
 159             break;                                                            \
 160           }                                                                   \
 161                                                                               \
 162         ch = (ch * 0x100) + inptr[1];                                         \
 163         while (ch > rp2->end)                                                 \
 164           ++rp2;                                                              \
 165                                                                               \
 166         if (__builtin_expect (rp2->start == 0xffff, 0)                        \
 167             || __builtin_expect (ch < rp2->start, 0)                          \
 168             || (res = __ibm930db_to_ucs4[ch + rp2->idx],                      \
 169                 __builtin_expect (res, L'\1') == L'\0' && ch != '\0'))        \
 170           {                                                                   \
 171             /* This is an illegal character.  */                              \
 172             STANDARD_FROM_LOOP_ERR_HANDLER (2);                               \
 173           }                                                                   \
 174         else                                                                  \
 175           {                                                                   \
 176             put32 (outptr, res);                                              \
 177             outptr += 4;                                                      \
 178           }                                                                   \
 179         inptr += 2;                                                           \
 180       }                                                                       \
 181   }
 182 #define LOOP_NEED_FLAGS
 183 #define EXTRA_LOOP_DECLS        , int *curcsp
 184 #define INIT_PARAMS             int curcs = *curcsp & ~7
 185 #define UPDATE_PARAMS           *curcsp = curcs
 186 #include <iconv/loop.c>
 187
 188 /* Next, define the other direction.  */
 189 #define MIN_NEEDED_INPUT        TO_LOOP_MIN_NEEDED_FROM
 190 #define MAX_NEEDED_INPUT        TO_LOOP_MAX_NEEDED_FROM
 191 #define MIN_NEEDED_OUTPUT       TO_LOOP_MIN_NEEDED_TO
 192 #define MAX_NEEDED_OUTPUT       TO_LOOP_MAX_NEEDED_TO
 193 #define LOOPFCT                 TO_LOOP
 194 #define BODY \
 195   {                                                                           \
 196     uint32_t ch = get32 (inptr);                                              \
 197     const struct gap *rp1 = __ucs4_to_ibm930sb_idx;                           \
 198     const struct gap *rp2 = __ucs4_to_ibm930db_idx;                           \
 199     const char *cp;                                                           \
 200                                                                               \
 201     if (__glibc_unlikely (ch >= 0xffff))                                      \
 202       {                                                                       \
 203         UNICODE_TAG_HANDLER (ch, 4);                                          \
 204                                                                               \
 205         STANDARD_TO_LOOP_ERR_HANDLER (4);                                     \
 206       }                                                                       \
 207                                                                               \
 208     while (ch > rp1->end)                                                     \
 209       ++rp1;                                                                  \
 210                                                                               \
 211     /* Use the UCS4 table for single byte.  */                                \
 212     if (__builtin_expect (ch < rp1->start, 0)                                 \
 213         || (cp = __ucs4_to_ibm930sb[ch + rp1->idx],                           \
 214             __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0'))          \
 215       {                                                                       \
 216         /* Use the UCS4 table for double byte. */                             \
 217         while (ch > rp2->end)                                                 \
 218           ++rp2;                                                              \
 219                                                                               \
 220         if (__builtin_expect (ch < rp2->start, 0)                             \
 221             || (cp = __ucs4_to_ibm930db[ch + rp2->idx],                       \
 222                 __builtin_expect (cp[0], L'\1')== L'\0' && ch != '\0'))       \
 223           {                                                                   \
 224             /* This is an illegal character.  */                              \
 225             STANDARD_TO_LOOP_ERR_HANDLER (4);                                 \
 226           }                                                                   \
 227         else                                                                  \
 228           {                                                                   \
 229             if (curcs == sb)                                                  \
 230               {                                                               \
 231                 if (__glibc_unlikely (outptr + 1 > outend))                   \
 232                   {                                                           \
 233                     result = __GCONV_FULL_OUTPUT;                             \
 234                     break;                                                    \
 235                   }                                                           \
 236                 *outptr++ = SO;                                               \
 237                 curcs = db;                                                   \
 238               }                                                               \
 239                                                                               \
 240             if (__glibc_unlikely (outptr + 2 > outend))                       \
 241               {                                                               \
 242                 result = __GCONV_FULL_OUTPUT;                                 \
 243                 break;                                                        \
 244               }                                                               \
 245             *outptr++ = cp[0];                                                \
 246             *outptr++ = cp[1];                                                \
 247           }                                                                   \
 248       }                                                                       \
 249     else                                                                      \
 250       {                                                                       \
 251         if (curcs == db)                                                      \
 252           {                                                                   \
 253             if (__glibc_unlikely (outptr + 1 > outend))                       \
 254               {                                                               \
 255                 result = __GCONV_FULL_OUTPUT;                                 \
 256                 break;                                                        \
 257               }                                                               \
 258             *outptr++ = SI;                                                   \
 259           }                                                                   \
 260                                                                               \
 261         if (__glibc_unlikely (outptr + 1 > outend))                           \
 262           {                                                                   \
 263             result = __GCONV_FULL_OUTPUT;                                     \
 264             break;                                                            \
 265           }                                                                   \
 266         if (ch == 0x7e)                                                       \
 267           *outptr++ = 0xa1;                                                   \
 268         else if (ch == 0x5c)                                                  \
 269           *outptr++ = 0x5b;                                                   \
 270         else                                                                  \
 271           *outptr++ = cp[0];                                                  \
 272         curcs = sb;                                                           \
 273       }                                                                       \
 274                                                                               \
 275     /* Now that we wrote the output increment the input pointer.  */          \
 276     inptr += 4;                                                               \
 277   }
 278 #define LOOP_NEED_FLAGS
 279 #define EXTRA_LOOP_DECLS        , int *curcsp
 280 #define INIT_PARAMS             int curcs = *curcsp & ~7
 281 #define REINIT_PARAMS           curcs = *curcsp & ~7
 282 #define UPDATE_PARAMS           *curcsp = curcs
 283 #include <iconv/loop.c>
 284
 285 /* Now define the toplevel functions.  */
 286 #include <iconv/skeleton.c>