iconvdata/ibm933.c

   1 /* Conversion from and to IBM933.
   2    Copyright (C) 2000 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 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 <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
  36 /* Definitions of initialization and destructor function.  */
  37 #define DEFINE_INIT     1
  38 #define DEFINE_FINI     1
  39
  40 #define MIN_NEEDED_FROM 1
  41 #define MIN_NEEDED_TO   4
  42
  43 /* Current codeset type.  */
  44 enum
  45 {
  46   init = 0,
  47   sb,
  48   db
  49 };
  50
  51 /* First, define the conversion function from IBM-933 to UCS4.  */
  52 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
  53 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
  54 #define INIT_PARAMS             int curcs = init;
  55 #define LOOPFCT                 FROM_LOOP
  56 #define BODY \
  57   {                                                                           \
  58     uint32_t ch = *inptr;                                                     \
  59     uint32_t res;                                                             \
  60     const struct gap *rp1 = __ibm933sb_to_ucs4_idx;                           \
  61     const struct gap *rp2 = __ibm933db_to_ucs4_idx;                           \
  62                                                                               \
  63     if (__builtin_expect (ch, 0) == SO)                                       \
  64       {                                                                       \
  65         if (__builtin_expect (inptr + 1 >= inend, 0))                         \
  66           {                                                                   \
  67             result = __GCONV_INCOMPLETE_INPUT;                                \
  68             break;                                                            \
  69           }                                                                   \
  70                                                                               \
  71         /* Shift OUT, change to DBCS converter.  */                           \
  72         if (curcs == db)                                                      \
  73           {                                                                   \
  74             result = __GCONV_ILLEGAL_INPUT;                                   \
  75             break;                                                            \
  76           }                                                                   \
  77         curcs = db;                                                           \
  78         ++inptr;                                                              \
  79         ch = *inptr;                                                          \
  80       }                                                                       \
  81     else if (__builtin_expect (ch, 0) == SI)                                  \
  82       {                                                                       \
  83         if (__builtin_expect (inptr + 1 >= inend, 0))                         \
  84           {                                                                   \
  85             result = __GCONV_INCOMPLETE_INPUT;                                \
  86             break;                                                            \
  87           }                                                                   \
  88                                                                               \
  89         /* Shift IN, change to SBCS converter.  */                            \
  90         if (curcs == sb)                                                      \
  91           {                                                                   \
  92             result = __GCONV_ILLEGAL_INPUT;                                   \
  93             break;                                                            \
  94           }                                                                   \
  95         curcs = sb;                                                           \
  96         ++inptr;                                                              \
  97         ch = *inptr;                                                          \
  98       }                                                                       \
  99                                                                               \
 100     if (curcs == sb || curcs == init)                                         \
 101       {                                                                       \
 102         /* Use the IBM933 table for single byte.  */                          \
 103         while (ch > rp1->end)                                                 \
 104           ++rp1;                                                              \
 105                                                                               \
 106         if (__builtin_expect (rp1 == NULL, 0)                                 \
 107             || __builtin_expect (ch < rp1->start, 0)                          \
 108             || (res = __ibm933sb_to_ucs4[ch + rp1->idx],                      \
 109                 __builtin_expect (res, L'\1') == L'\0' && ch != '\0'))        \
 110           {                                                                   \
 111             /* This is an illegal character.  */                              \
 112             if (! ignore_errors_p ())                                         \
 113               {                                                               \
 114                 result = __GCONV_ILLEGAL_INPUT;                               \
 115                 break;                                                        \
 116               }                                                               \
 117             ++*irreversible;                                                  \
 118             ++inptr;                                                          \
 119             continue;                                                         \
 120           }                                                                   \
 121         else                                                                  \
 122           {                                                                   \
 123             put32 (outptr, res);                                              \
 124             outptr += 4;                                                      \
 125             ++inptr;                                                          \
 126           }                                                                   \
 127       }                                                                       \
 128     else if (curcs == db)                                                     \
 129       {                                                                       \
 130         /* Use the IBM933 table for double byte.  */                          \
 131         if (__builtin_expect (inptr + 1 >= inend, 0))                         \
 132           {                                                                   \
 133             /* The second character is not available.  Store the              \
 134                intermediate result. */                                        \
 135             result = __GCONV_INCOMPLETE_INPUT;                                \
 136             break;                                                            \
 137           }                                                                   \
 138                                                                               \
 139         ch = (ch * 0x100) + inptr[1];                                         \
 140         while (ch > rp2->end)                                                 \
 141           ++rp2;                                                              \
 142                                                                               \
 143         if (__builtin_expect (rp2 == NULL, 0)                                 \
 144             || __builtin_expect (ch < rp2->start, 0)                          \
 145             || (res = __ibm933db_to_ucs4[ch + rp2->idx],                      \
 146                 __builtin_expect (res, L'\1') == L'\0' && ch != '\0'))        \
 147           {                                                                   \
 148             /* This is an illegal character.  */                              \
 149             if (! ignore_errors_p ())                                         \
 150               {                                                               \
 151                 result = __GCONV_ILLEGAL_INPUT;                               \
 152                 break;                                                        \
 153               }                                                               \
 154             ++*irreversible;                                                  \
 155             inptr += 2;                                                       \
 156             continue;                                                         \
 157           }                                                                   \
 158         else                                                                  \
 159           {                                                                   \
 160             put32 (outptr, res);                                              \
 161             outptr += 4;                                                      \
 162             inptr += 2;                                                       \
 163           }                                                                   \
 164       }                                                                       \
 165   }
 166 #define LOOP_NEED_FLAGS
 167 #include <iconv/loop.c>
 168
 169 /* Next, define the other direction.  */
 170 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 171 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 172 #define INIT_PARAMS             int curcs = init;
 173 #define LOOPFCT                 TO_LOOP
 174 #define BODY \
 175   {                                                                           \
 176     uint32_t ch = get32 (inptr);                                              \
 177     const struct gap *rp1 = __ucs4_to_ibm933sb_idx;                           \
 178     const struct gap *rp2 = __ucs4_to_ibm933db_idx;                           \
 179     const char *cp;                                                           \
 180                                                                               \
 181     if (__builtin_expect (ch, 0) >= 0xffff)                                   \
 182       {                                                                       \
 183         rp1 = NULL;                                                           \
 184         rp2 = NULL;                                                           \
 185       }                                                                       \
 186     else                                                                      \
 187       while (ch > rp1->end)                                                   \
 188         ++rp1;                                                                \
 189                                                                               \
 190     /* Use the UCS4 table for single byte.  */                                \
 191     if (__builtin_expect (rp1 == NULL, 0)                                     \
 192         || __builtin_expect (ch < rp1->start, 0)                              \
 193         || (cp = __ucs4_to_ibm933sb[ch + rp1->idx],                           \
 194             __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0'))          \
 195       {                                                                       \
 196         /* Use the UCS4 table for double byte.  */                            \
 197         while (ch > rp2->end)                                                 \
 198           ++rp2;                                                              \
 199                                                                               \
 200         if (__builtin_expect (rp2 == NULL, 0)                                 \
 201             || __builtin_expect (ch < rp2->start, 0)                          \
 202             || (cp = __ucs4_to_ibm933db[ch + rp2->idx],                       \
 203                 __builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0'))        \
 204           {                                                                   \
 205             /* This is an illegal character.  */                              \
 206             if (! ignore_errors_p ())                                         \
 207               {                                                               \
 208                 result = __GCONV_ILLEGAL_INPUT;                               \
 209                 break;                                                        \
 210               }                                                               \
 211             ++*irreversible;                                                  \
 212           }                                                                   \
 213         else                                                                  \
 214           {                                                                   \
 215             if (curcs == init || curcs == sb)                                 \
 216               {                                                               \
 217                 if (__builtin_expect (outptr+1 > outend, 0))                  \
 218                   {                                                           \
 219                     result = __GCONV_FULL_OUTPUT;                             \
 220                     break;                                                    \
 221                   }                                                           \
 222                 *outptr++ = SO;                                               \
 223                 curcs = db;                                                   \
 224               }                                                               \
 225                                                                               \
 226             if (__builtin_expect (outptr + 2 > outend, 0))                    \
 227               {                                                               \
 228                 result = __GCONV_FULL_OUTPUT;                                 \
 229                 break;                                                        \
 230               }                                                               \
 231             *outptr++ = cp[0];                                                \
 232             *outptr++ = cp[1];                                                \
 233           }                                                                   \
 234       }                                                                       \
 235     else                                                                      \
 236       {                                                                       \
 237         if (curcs == db)                                                      \
 238           {                                                                   \
 239             if (__builtin_expect (outptr + 1 > outend, 0))                    \
 240               {                                                               \
 241                 result = __GCONV_FULL_OUTPUT;                                 \
 242                 break;                                                        \
 243               }                                                               \
 244             *outptr++ = SI;                                                   \
 245           }                                                                   \
 246                                                                               \
 247         if (__builtin_expect (outptr + 1 > outend, 0))                        \
 248           {                                                                   \
 249             result = __GCONV_FULL_OUTPUT;                                     \
 250             break;                                                            \
 251           }                                                                   \
 252         *outptr++ = cp[0];                                                    \
 253         curcs = sb;                                                           \
 254       }                                                                       \
 255                                                                               \
 256     /* Now that we wrote the output increment the input pointer.  */          \
 257     inptr += 4;                                                               \
 258   }
 259 #define LOOP_NEED_FLAGS
 260 #include <iconv/loop.c>
 261
 262 /* Now define the toplevel functions.  */
 263 #include <iconv/skeleton.c>