iconvdata/euc-tw.c

   1 /* Mapping tables for EUC-TW handling.
   2    Copyright (C) 1998, 1999, 2000 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 <dlfcn.h>
  22 #include <stdint.h>
  23 #include <cns11643l1.h>
  24 #include <cns11643.h>
  25
  26 /* Definitions used in the body of the `gconv' function.  */
  27 #define CHARSET_NAME            "EUC-TW//"
  28 #define FROM_LOOP               from_euc_tw
  29 #define TO_LOOP                 to_euc_tw
  30 #define DEFINE_INIT             1
  31 #define DEFINE_FINI             1
  32 #define MIN_NEEDED_FROM         1
  33 #define MAX_NEEDED_FROM         4
  34 #define MIN_NEEDED_TO           4
  35
  36
  37 /* First define the conversion function from EUC-TW to UCS4.  */
  38 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
  39 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
  40 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
  41 #define LOOPFCT                 FROM_LOOP
  42 #define BODY \
  43   {                                                                           \
  44     uint32_t ch = *inptr;                                                     \
  45                                                                               \
  46     if (ch <= 0x7f)                                                           \
  47       /* Plain ASCII.  */                                                     \
  48       ++inptr;                                                                \
  49     else if ((ch <= 0xa0 || ch > 0xfe) && ch != 0x8e)                         \
  50       {                                                                       \
  51         /* This is illegal.  */                                               \
  52         if (! ignore_errors_p ())                                             \
  53           {                                                                   \
  54             result = __GCONV_ILLEGAL_INPUT;                                   \
  55             break;                                                            \
  56           }                                                                   \
  57                                                                               \
  58         ++inptr;                                                              \
  59         ++*irreversible;                                                      \
  60         continue;                                                             \
  61       }                                                                       \
  62     else                                                                      \
  63       {                                                                       \
  64         /* Two or more byte character.  First test whether the next           \
  65            character is also available.  */                                   \
  66         uint32_t ch2;                                                         \
  67                                                                               \
  68         if (inptr + (ch == 0x8e ? 3 : 1) >= inend)                            \
  69           {                                                                   \
  70             /* The second character is not available.  Store the              \
  71                intermediate result.  */                                       \
  72             result = __GCONV_INCOMPLETE_INPUT;                                \
  73             break;                                                            \
  74           }                                                                   \
  75                                                                               \
  76         ch2 = *inptr;                                                         \
  77                                                                               \
  78         /* All second bytes of a multibyte character must be >= 0xa1. */      \
  79         if (ch2 < 0xa1 || ch2 == 0xff)                                        \
  80           {                                                                   \
  81             /* This is illegal.  */                                           \
  82             if (! ignore_errors_p ())                                         \
  83               {                                                               \
  84                 result = __GCONV_ILLEGAL_INPUT;                               \
  85                 break;                                                        \
  86               }                                                               \
  87                                                                               \
  88             ++inptr;                                                          \
  89             ++*irreversible;                                                  \
  90             continue;                                                         \
  91           }                                                                   \
  92                                                                               \
  93         if (ch == 0x8e)                                                       \
  94           {                                                                   \
  95             /* This is code set 2: CNS 11643, planes 1 to 16.  */             \
  96             const char *endp = inptr + 1;                                     \
  97                                                                               \
  98             ch = cns11643_to_ucs4 (&endp, inend - inptr - 1, 0x80);           \
  99             /* Please note that we need not test for the missing input        \
 100                characters here anymore.  */                                   \
 101             if (ch == __UNKNOWN_10646_CHAR)                                   \
 102               {                                                               \
 103                 /* Illegal input.  */                                         \
 104                 if (! ignore_errors_p ())                                     \
 105                   {                                                           \
 106                     /* This is an illegal character.  */                      \
 107                     result = __GCONV_ILLEGAL_INPUT;                           \
 108                     break;                                                    \
 109                   }                                                           \
 110                                                                               \
 111                 ++inptr;                                                      \
 112                 ++*irreversible;                                              \
 113                 continue;                                                     \
 114               }                                                               \
 115                                                                               \
 116             inptr += 4;                                                       \
 117           }                                                                   \
 118         else                                                                  \
 119           {                                                                   \
 120             /* This is code set 1: CNS 11643, plane 1.  */                    \
 121             const unsigned char *endp = inptr;                                \
 122                                                                               \
 123             ch = cns11643l1_to_ucs4 (&endp, inend - inptr, 0x80);             \
 124             /* Please note that we need not test for the missing input        \
 125                characters here anymore.  */                                   \
 126             if (ch == __UNKNOWN_10646_CHAR)                                   \
 127               {                                                               \
 128                 /* Illegal input.  */                                         \
 129                 if (! ignore_errors_p ())                                     \
 130                   {                                                           \
 131                     /* This is an illegal character.  */                      \
 132                     result = __GCONV_ILLEGAL_INPUT;                           \
 133                     break;                                                    \
 134                   }                                                           \
 135                                                                               \
 136                 inptr += 2;                                                   \
 137                 ++*irreversible;                                              \
 138                 continue;                                                     \
 139               }                                                               \
 140                                                                               \
 141             inptr += 2;                                                       \
 142           }                                                                   \
 143       }                                                                       \
 144                                                                               \
 145     put32 (outptr, ch);                                                       \
 146     outptr += 4;                                                              \
 147   }
 148 #define LOOP_NEED_FLAGS
 149 #include <iconv/loop.c>
 150
 151
 152 /* Next, define the other direction.  */
 153 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 154 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 155 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 156 #define LOOPFCT                 TO_LOOP
 157 #define BODY \
 158   {                                                                           \
 159     uint32_t ch = get32 (inptr);                                              \
 160                                                                               \
 161     if (ch <= 0x7f)                                                           \
 162       /* It's plain ASCII.  */                                                \
 163       *outptr++ = ch;                                                         \
 164     else                                                                      \
 165       {                                                                       \
 166         /* Try the JIS character sets.  */                                    \
 167         size_t found;                                                         \
 168                                                                               \
 169         found = ucs4_to_cns11643l1 (ch, outptr, outend - outptr);             \
 170         if (__builtin_expect (found, 1) == 0)                                 \
 171           {                                                                   \
 172             /* We ran out of space.  */                                       \
 173             result = __GCONV_INCOMPLETE_INPUT;                                \
 174             break;                                                            \
 175           }                                                                   \
 176         if (__builtin_expect (found, 1) != __UNKNOWN_10646_CHAR)              \
 177           {                                                                   \
 178             /* It's a CNS 11643, plane 1 character, adjust it for EUC-TW.  */ \
 179             *outptr++ += 0x80;                                                \
 180             *outptr++ += 0x80;                                                \
 181           }                                                                   \
 182         else                                                                  \
 183           {                                                                   \
 184             /* No CNS 11643, plane 1 character.  */                           \
 185                                                                               \
 186             found = ucs4_to_cns11643 (ch, outptr + 1, outend - outptr - 1);   \
 187             if (__builtin_expect (found, 1) == 0)                             \
 188               {                                                               \
 189                 /* We ran out of space.  */                                   \
 190                 result = __GCONV_INCOMPLETE_INPUT;                            \
 191                 break;                                                        \
 192               }                                                               \
 193             if (__builtin_expect (found, 0) == __UNKNOWN_10646_CHAR)          \
 194               {                                                               \
 195                 /* Illegal character.  */                                     \
 196                 if (step_data->__trans.__trans_fct != NULL)                   \
 197                   {                                                           \
 198                     result = DL_CALL_FCT (step_data->__trans.__trans_fct,     \
 199                                           (step, step_data, *inptrp, &inptr,  \
 200                                            inend, *outptrp, &outptr, outend,  \
 201                                            irreversible));                    \
 202                     if (result != __GCONV_OK)                                 \
 203                       break;                                                  \
 204                   }                                                           \
 205                 else if (! ignore_errors_p ())                                \
 206                   {                                                           \
 207                     result = __GCONV_ILLEGAL_INPUT;                           \
 208                     break;                                                    \
 209                   }                                                           \
 210                 else                                                          \
 211                   {                                                           \
 212                     inptr += 4;                                               \
 213                     ++*irreversible;                                          \
 214                   }                                                           \
 215                 continue;                                                     \
 216               }                                                               \
 217                                                                               \
 218             /* It's a CNS 11643 character, adjust it for EUC-TW.  */          \
 219             *outptr++ = '\x8e';                                               \
 220             *outptr++ += 0xa0;                                                \
 221             *outptr++ += 0x80;                                                \
 222             *outptr++ += 0x80;                                                \
 223           }                                                                   \
 224       }                                                                       \
 225                                                                               \
 226     inptr += 4;                                                               \
 227   }
 228 #define LOOP_NEED_FLAGS
 229 #include <iconv/loop.c>
 230
 231
 232 /* Now define the toplevel functions.  */
 233 #include <iconv/skeleton.c>