iconvdata/euc-jp.c

   1 /* Mapping tables for EUC-JP 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 <stdint.h>
  22 #include <gconv.h>
  23 #include <jis0201.h>
  24 #include <jis0208.h>
  25 #include <jis0212.h>
  26
  27 /* Definitions used in the body of the `gconv' function.  */
  28 #define CHARSET_NAME            "EUC-JP//"
  29 #define FROM_LOOP               from_euc_jp
  30 #define TO_LOOP                 to_euc_jp
  31 #define DEFINE_INIT             1
  32 #define DEFINE_FINI             1
  33 #define MIN_NEEDED_FROM         1
  34 #define MAX_NEEDED_FROM         3
  35 #define MIN_NEEDED_TO           4
  36
  37
  38 /* First define the conversion function from EUC-JP 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       ++inptr;                                                                \
  49     else if ((__builtin_expect (ch, 0xa1) <= 0xa0 && ch != 0x8e && ch != 0x8f)\
  50              || __builtin_expect (ch, 0xfe) > 0xfe)                           \
  51       {                                                                       \
  52         /* This is illegal.  */                                               \
  53         if (! ignore_errors_p ())                                             \
  54           {                                                                   \
  55             result = __GCONV_ILLEGAL_INPUT;                                   \
  56             break;                                                            \
  57           }                                                                   \
  58                                                                               \
  59         ++inptr;                                                              \
  60         ++*irreversible;                                                      \
  61         continue;                                                             \
  62       }                                                                       \
  63     else                                                                      \
  64       {                                                                       \
  65         /* Two or more byte character.  First test whether the next           \
  66            character is also available.  */                                   \
  67         int ch2;                                                              \
  68                                                                               \
  69         if (NEED_LENGTH_TEST && __builtin_expect (inptr + 1 >= inend, 0))     \
  70           {                                                                   \
  71             /* The second character is not available.  Store the              \
  72                intermediate result.  */                                       \
  73             result = __GCONV_INCOMPLETE_INPUT;                                \
  74             break;                                                            \
  75           }                                                                   \
  76                                                                               \
  77         ch2 = inptr[1];                                                       \
  78                                                                               \
  79         /* All second bytes of a multibyte character must be >= 0xa1. */      \
  80         if (__builtin_expect (ch2, 0xa1) < 0xa1)                              \
  81           {                                                                   \
  82             /* This is an illegal character.  */                              \
  83             if (! ignore_errors_p ())                                         \
  84               {                                                               \
  85                 result = __GCONV_ILLEGAL_INPUT;                               \
  86                 break;                                                        \
  87               }                                                               \
  88                                                                               \
  89             ++inptr;                                                          \
  90             ++*irreversible;                                                  \
  91             continue;                                                         \
  92           }                                                                   \
  93                                                                               \
  94         if (ch == 0x8e)                                                       \
  95           {                                                                   \
  96             /* This is code set 2: half-width katakana.  */                   \
  97             ch = jisx0201_to_ucs4 (ch2);                                      \
  98             inptr += 2;                                                       \
  99           }                                                                   \
 100         else                                                                  \
 101           {                                                                   \
 102             const unsigned char *endp;                                        \
 103                                                                               \
 104             if (ch == 0x8f)                                                   \
 105               {                                                               \
 106                 /* This is code set 3: JIS X 0212-1990.  */                   \
 107                 endp = inptr + 1;                                             \
 108                                                                               \
 109                 ch = jisx0212_to_ucs4 (&endp,                                 \
 110                                        NEED_LENGTH_TEST ? inend - endp : 2,   \
 111                                        0x80);                                 \
 112               }                                                               \
 113             else                                                              \
 114               {                                                               \
 115                 /* This is code set 1: JIS X 0208.  */                        \
 116                 endp = inptr;                                                 \
 117                                                                               \
 118                 ch = jisx0208_to_ucs4 (&endp,                                 \
 119                                        NEED_LENGTH_TEST ? inend - inptr : 2,  \
 120                                        0x80);                                 \
 121               }                                                               \
 122                                                                               \
 123             if (NEED_LENGTH_TEST && __builtin_expect (ch, 1) == 0)            \
 124               {                                                               \
 125                 /* Not enough input available.  */                            \
 126                 result = __GCONV_INCOMPLETE_INPUT;                            \
 127                 break;                                                        \
 128               }                                                               \
 129             if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR)             \
 130               {                                                               \
 131                 /* Illegal character.  */                                     \
 132                 if (! ignore_errors_p ())                                     \
 133                   {                                                           \
 134                     /* This is an illegal character.  */                      \
 135                     result = __GCONV_ILLEGAL_INPUT;                           \
 136                     break;                                                    \
 137                   }                                                           \
 138                                                                               \
 139                 inptr += 2;                                                   \
 140                 ++*irreversible;                                              \
 141                 continue;                                                     \
 142               }                                                               \
 143             inptr = endp;                                                     \
 144           }                                                                   \
 145       }                                                                       \
 146                                                                               \
 147     put32 (outptr, ch);                                                       \
 148     outptr += 4;                                                              \
 149   }
 150 #include <iconv/loop.c>
 151
 152
 153 /* Next, define the other direction.  */
 154 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 155 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 156 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 157 #define LOOPFCT                 TO_LOOP
 158 #define BODY \
 159   {                                                                           \
 160     uint32_t ch = get32 (inptr);                                              \
 161                                                                               \
 162     if (ch <= 0x7f)                                                           \
 163       /* It's plain ASCII.  */                                                \
 164       *outptr++ = ch;                                                         \
 165     else if (ch == 0xa5)                                                      \
 166       /* YEN sign => backslash  */                                            \
 167       *outptr++ = 0x5c;                                                       \
 168     else if (ch == 0x203e)                                                    \
 169       /* overscore => asciitilde */                                           \
 170       *outptr++ = 0x7e;                                                       \
 171     else                                                                      \
 172       {                                                                       \
 173         /* Try the JIS character sets.  */                                    \
 174         size_t found;                                                         \
 175                                                                               \
 176         /* See whether we have room for at least two characters.  */          \
 177         if (NEED_LENGTH_TEST && __builtin_expect (outptr + 1 >= outend, 0))   \
 178           {                                                                   \
 179             result = __GCONV_FULL_OUTPUT;                                     \
 180             break;                                                            \
 181           }                                                                   \
 182                                                                               \
 183         found = ucs4_to_jisx0201 (ch, outptr + 1);                            \
 184         if (found != __UNKNOWN_10646_CHAR)                                    \
 185           {                                                                   \
 186             /* Yes, it's a JIS 0201 character.  Store the shift byte.  */     \
 187             *outptr = 0x8e;                                                   \
 188             outptr += 2;                                                      \
 189           }                                                                   \
 190         else                                                                  \
 191           {                                                                   \
 192             /* No JIS 0201 character.  */                                     \
 193             found = ucs4_to_jisx0208 (ch, outptr, 2);                         \
 194             /* Please note that we always have enough room for the output. */ \
 195             if (found != __UNKNOWN_10646_CHAR)                                \
 196               {                                                               \
 197                 /* It's a JIS 0208 character, adjust it for EUC-JP.  */       \
 198                 *outptr++ += 0x80;                                            \
 199                 *outptr++ += 0x80;                                            \
 200               }                                                               \
 201             else                                                              \
 202               {                                                               \
 203                 /* No JIS 0208 character.  */                                 \
 204                 found = ucs4_to_jisx0212 (ch, outptr + 1,                     \
 205                                           (NEED_LENGTH_TEST                   \
 206                                            ? outend - outptr - 1 : 2));       \
 207                                                                               \
 208                 if (__builtin_expect (found, 1) == 0)                         \
 209                   {                                                           \
 210                     /* We ran out of space.  */                               \
 211                     result = __GCONV_FULL_OUTPUT;                             \
 212                     break;                                                    \
 213                   }                                                           \
 214                 else if (__builtin_expect (found, 0) != __UNKNOWN_10646_CHAR) \
 215                   {                                                           \
 216                     /* It's a JIS 0212 character, adjust it for EUC-JP.  */   \
 217                     *outptr++ = 0x8f;                                         \
 218                     *outptr++ += 0x80;                                        \
 219                     *outptr++ += 0x80;                                        \
 220                   }                                                           \
 221                 else                                                          \
 222                   {                                                           \
 223                     /* Illegal character.  */                                 \
 224                     if (! ignore_errors_p ())                                 \
 225                       {                                                       \
 226                         result = __GCONV_ILLEGAL_INPUT;                       \
 227                         break;                                                \
 228                       }                                                       \
 229                                                                               \
 230                     inptr += 4;                                               \
 231                     ++*irreversible;                                          \
 232                     continue;                                                 \
 233                   }                                                           \
 234               }                                                               \
 235           }                                                                   \
 236       }                                                                       \
 237                                                                               \
 238     inptr += 4;                                                               \
 239   }
 240 #include <iconv/loop.c>
 241
 242
 243 /* Now define the toplevel functions.  */
 244 #include <iconv/skeleton.c>