iconvdata/euc-tw.c

   1 /* Mapping tables for EUC-TW handling.
   2    Copyright (C) 1998, 1999, 2000-2002, 2003, 2007
   3    Free Software Foundation, Inc.
   4    This file is part of the GNU C Library.
   5    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
   6
   7    The GNU C Library is free software; you can redistribute it and/or
   8    modify it under the terms of the GNU Lesser General Public
   9    License as published by the Free Software Foundation; either
  10    version 2.1 of the License, or (at your option) any later version.
  11
  12    The GNU C Library is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15    Lesser General Public License for more details.
  16
  17    You should have received a copy of the GNU Lesser General Public
  18    License along with the GNU C Library; if not, write to the Free
  19    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  20    02111-1307 USA.  */
  21
  22 #include <dlfcn.h>
  23 #include <stdint.h>
  24 #include <cns11643l1.h>
  25 #include <cns11643.h>
  26
  27 /* Definitions used in the body of the `gconv' function.  */
  28 #define CHARSET_NAME            "EUC-TW//"
  29 #define FROM_LOOP               from_euc_tw
  30 #define TO_LOOP                 to_euc_tw
  31 #define DEFINE_INIT             1
  32 #define DEFINE_FINI             1
  33 #define MIN_NEEDED_FROM         1
  34 #define MAX_NEEDED_FROM         4
  35 #define MIN_NEEDED_TO           4
  36
  37
  38 /* First define the conversion function from EUC-TW to UCS4.  */
  39 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
  40 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
  41 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
  42 #define LOOPFCT                 FROM_LOOP
  43 #define BODY \
  44   {                                                                           \
  45     uint32_t ch = *inptr;                                                     \
  46                                                                               \
  47     if (ch <= 0x7f)                                                           \
  48       /* Plain ASCII.  */                                                     \
  49       ++inptr;                                                                \
  50     else if ((ch <= 0xa0 || ch > 0xfe) && ch != 0x8e)                         \
  51       {                                                                       \
  52         /* This is illegal.  */                                               \
  53         STANDARD_FROM_LOOP_ERR_HANDLER (1);                                   \
  54       }                                                                       \
  55     else                                                                      \
  56       {                                                                       \
  57         /* Two or more byte character.  First test whether the next byte      \
  58            is also available.  */                                             \
  59         uint32_t ch2;                                                         \
  60                                                                               \
  61         if (inptr + 1 >= inend)                                               \
  62           {                                                                   \
  63             /* The second byte is not available.  Store the intermediate      \
  64                result.  */                                                    \
  65             result = __GCONV_INCOMPLETE_INPUT;                                \
  66             break;                                                            \
  67           }                                                                   \
  68                                                                               \
  69         ch2 = *(inptr + 1);                                                   \
  70                                                                               \
  71         /* All second bytes of a multibyte character must be >= 0xa1. */      \
  72         if (ch2 < 0xa1 || ch2 == 0xff)                                        \
  73           STANDARD_FROM_LOOP_ERR_HANDLER (1);                                 \
  74                                                                               \
  75         if (ch == 0x8e)                                                       \
  76           {                                                                   \
  77             /* This is code set 2: CNS 11643, planes 1 to 16.  */             \
  78             const unsigned char *endp = inptr + 1;                            \
  79                                                                               \
  80             ch = cns11643_to_ucs4 (&endp, inend - inptr - 1, 0x80);           \
  81                                                                               \
  82             if (ch == 0)                                                      \
  83               {                                                               \
  84                 /* The third or fourth byte is not available.  Store          \
  85                    the intermediate result.  */                               \
  86                 result = __GCONV_INCOMPLETE_INPUT;                            \
  87                 break;                                                        \
  88               }                                                               \
  89                                                                               \
  90             if (ch == __UNKNOWN_10646_CHAR)                                   \
  91               /* Illegal input.  */                                           \
  92               STANDARD_FROM_LOOP_ERR_HANDLER (1);                             \
  93                                                                               \
  94             inptr += 4;                                                       \
  95           }                                                                   \
  96         else                                                                  \
  97           {                                                                   \
  98             /* This is code set 1: CNS 11643, plane 1.  */                    \
  99             const unsigned char *endp = inptr;                                \
 100                                                                               \
 101             ch = cns11643l1_to_ucs4 (&endp, inend - inptr, 0x80);             \
 102             /* Please note that we need not test for the missing input        \
 103                characters here anymore.  */                                   \
 104             if (ch == __UNKNOWN_10646_CHAR)                                   \
 105               /* Illegal input.  */                                           \
 106               STANDARD_FROM_LOOP_ERR_HANDLER (2);                             \
 107                                                                               \
 108             inptr += 2;                                                       \
 109           }                                                                   \
 110       }                                                                       \
 111                                                                               \
 112     put32 (outptr, ch);                                                       \
 113     outptr += 4;                                                              \
 114   }
 115 #define LOOP_NEED_FLAGS
 116 #define ONEBYTE_BODY \
 117   {                                                                           \
 118     if (c < 0x80)                                                             \
 119       return c;                                                               \
 120     else                                                                      \
 121       return WEOF;                                                            \
 122   }
 123 #include <iconv/loop.c>
 124
 125
 126 /* Next, define the other direction.  */
 127 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 128 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 129 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 130 #define LOOPFCT                 TO_LOOP
 131 #define BODY \
 132   {                                                                           \
 133     uint32_t ch = get32 (inptr);                                              \
 134                                                                               \
 135     if (ch <= 0x7f)                                                           \
 136       /* It's plain ASCII.  */                                                \
 137       *outptr++ = ch;                                                         \
 138     else                                                                      \
 139       {                                                                       \
 140         /* Try the CNS 11643 planes.  */                                      \
 141         size_t found;                                                         \
 142                                                                               \
 143         found = ucs4_to_cns11643l1 (ch, outptr, outend - outptr);             \
 144         if (__builtin_expect (found, 1) == 0)                                 \
 145           {                                                                   \
 146             /* We ran out of space.  */                                       \
 147             result = __GCONV_FULL_OUTPUT;                                     \
 148             break;                                                            \
 149           }                                                                   \
 150         if (__builtin_expect (found, 1) != __UNKNOWN_10646_CHAR)              \
 151           {                                                                   \
 152             /* It's a CNS 11643, plane 1 character, adjust it for EUC-TW.  */ \
 153             *outptr++ += 0x80;                                                \
 154             *outptr++ += 0x80;                                                \
 155           }                                                                   \
 156         else                                                                  \
 157           {                                                                   \
 158             /* No CNS 11643, plane 1 character.  */                           \
 159                                                                               \
 160             found = ucs4_to_cns11643 (ch, outptr + 1, outend - outptr - 1);   \
 161             if (__builtin_expect (found, 1) == 0)                             \
 162               {                                                               \
 163                 /* We ran out of space.  */                                   \
 164                 result = __GCONV_FULL_OUTPUT;                                 \
 165                 break;                                                        \
 166               }                                                               \
 167             if (__builtin_expect (found, 0) == __UNKNOWN_10646_CHAR)          \
 168               {                                                               \
 169                 UNICODE_TAG_HANDLER (ch, 4);                                  \
 170                                                                               \
 171                 /* Illegal character.  */                                     \
 172                 STANDARD_TO_LOOP_ERR_HANDLER (4);                             \
 173               }                                                               \
 174                                                                               \
 175             /* It's a CNS 11643 character, adjust it for EUC-TW.  */          \
 176             *outptr++ = '\x8e';                                               \
 177             *outptr++ += 0xa0;                                                \
 178             *outptr++ += 0x80;                                                \
 179             *outptr++ += 0x80;                                                \
 180           }                                                                   \
 181       }                                                                       \
 182                                                                               \
 183     inptr += 4;                                                               \
 184   }
 185 #define LOOP_NEED_FLAGS
 186 #include <iconv/loop.c>
 187
 188
 189 /* Now define the toplevel functions.  */
 190 #include <iconv/skeleton.c>