iconvdata/euc-kr.c

   1 /* Mapping tables for EUC-KR handling.
   2    Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Jungshik Shin <jshin@pantheon.yale.edu>
   5    and 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 Library General Public License as
   9    published by the Free Software Foundation; either version 2 of the
  10    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    Library General Public License for more details.
  16
  17    You should have received a copy of the GNU Library General Public
  18    License along with the GNU C Library; see the file COPYING.LIB.  If not,
  19    write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  20    Boston, MA 02111-1307, USA.  */
  21
  22 #include <dlfcn.h>
  23 #include <stdint.h>
  24 #include <ksc5601.h>
  25
  26
  27 static inline void
  28 euckr_from_ucs4 (uint32_t ch, unsigned char *cp)
  29 {
  30   if (ch > 0x7f)
  31     {
  32       if (__builtin_expect (ucs4_to_ksc5601 (ch, cp, 2), 0)
  33           != __UNKNOWN_10646_CHAR)
  34         {
  35           cp[0] |= 0x80;
  36           cp[1] |= 0x80;
  37         }
  38       else
  39         cp[0] = '\0';
  40     }
  41   /* XXX Think about 0x5c ; '\'.  */
  42   else
  43     {
  44       cp[0] = (unsigned char) ch;
  45       cp[1] = '\0';
  46     }
  47 }
  48
  49
  50 /* Definitions used in the body of the `gconv' function.  */
  51 #define CHARSET_NAME            "EUC-KR//"
  52 #define FROM_LOOP               from_euc_kr
  53 #define TO_LOOP                 to_euc_kr
  54 #define DEFINE_INIT             1
  55 #define DEFINE_FINI             1
  56 #define MIN_NEEDED_FROM         1
  57 #define MAX_NEEDED_FROM         2
  58 #define MIN_NEEDED_TO           4
  59
  60
  61 /* First define the conversion function from EUC-KR to UCS4.  */
  62 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
  63 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
  64 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
  65 #define LOOPFCT                 FROM_LOOP
  66 #define BODY \
  67   {                                                                           \
  68     uint32_t ch = *inptr;                                                     \
  69                                                                               \
  70     /* Half-width Korean Currency WON sign                                    \
  71                                                                               \
  72        if (inchar == 0x5c)                                                    \
  73          ch =  0x20a9;                                                        \
  74        else if (inchar <= 0x7f)                                               \
  75          ch = (uint32_t) inchar;                                              \
  76     */                                                                        \
  77                                                                               \
  78     if (ch <= 0x7f)                                                           \
  79       /* Plain ASCII.  */                                                     \
  80       ++inptr;                                                                \
  81     /* 0xfe(->0x7e : row 94) and 0xc9(->0x59 : row 41) are                    \
  82        user-defined areas.  */                                                \
  83     else if (__builtin_expect (ch, 0xa1) <= 0xa0                              \
  84              || __builtin_expect (ch, 0xa1) > 0xfe                            \
  85              || __builtin_expect (ch, 0xa1) == 0xc9)                          \
  86       {                                                                       \
  87         /* This is illegal.  */                                               \
  88         if (! ignore_errors_p ())                                             \
  89           {                                                                   \
  90             result = __GCONV_ILLEGAL_INPUT;                                   \
  91             break;                                                            \
  92           }                                                                   \
  93                                                                               \
  94         ++inptr;                                                              \
  95         ++*irreversible;                                                      \
  96         continue;                                                             \
  97       }                                                                       \
  98     else                                                                      \
  99       {                                                                       \
 100         /* Two-byte character.  First test whether the next character         \
 101            is also available.  */                                             \
 102         ch = ksc5601_to_ucs4 (&inptr, inptr - inend, 0x80);                   \
 103         if (__builtin_expect (ch, 1) == 0)                                    \
 104           {                                                                   \
 105             /* The second character is not available.  */                     \
 106             result = __GCONV_INCOMPLETE_INPUT;                                \
 107             break;                                                            \
 108           }                                                                   \
 109         if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR)                 \
 110           {                                                                   \
 111             /* This is an illegal character.  */                              \
 112             if (! ignore_errors_p ())                                         \
 113               {                                                               \
 114                 /* This is an illegal character.  */                          \
 115                 result = __GCONV_ILLEGAL_INPUT;                               \
 116                 break;                                                        \
 117               }                                                               \
 118                                                                               \
 119             inptr += 2;                                                       \
 120             ++*irreversible;                                                  \
 121             continue;                                                         \
 122           }                                                                   \
 123       }                                                                       \
 124                                                                               \
 125     put32 (outptr, ch);                                                       \
 126     outptr += 4;                                                              \
 127   }
 128 #define LOOP_NEED_FLAGS
 129 #include <iconv/loop.c>
 130
 131
 132 /* Next, define the other direction.  */
 133 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 134 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 135 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 136 #define LOOPFCT                 TO_LOOP
 137 #define BODY \
 138   {                                                                           \
 139     uint32_t ch = get32 (inptr);                                              \
 140     unsigned char cp[2];                                                      \
 141                                                                               \
 142     /* Decomposing Hangul syllables not available in KS C 5601 into           \
 143        Jamos should be considered either here or in euckr_from_ucs4() */      \
 144     euckr_from_ucs4 (ch, cp) ;                                                \
 145                                                                               \
 146     if (__builtin_expect (cp[0], '\1') == '\0' && ch != 0)                    \
 147       {                                                                       \
 148         /* Illegal character.  */                                             \
 149         if (step_data->__trans.__trans_fct != NULL)                           \
 150           {                                                                   \
 151             result = DL_CALL_FCT (step_data->__trans.__trans_fct,             \
 152                                   (step, step_data, *inptrp, &inptr, inend,   \
 153                                    *outptrp, &outptr, outend, irreversible)); \
 154             if (result != __GCONV_OK)                                         \
 155               break;                                                          \
 156           }                                                                   \
 157         else if (! ignore_errors_p ())                                        \
 158           {                                                                   \
 159             result = __GCONV_ILLEGAL_INPUT;                                   \
 160             break;                                                            \
 161           }                                                                   \
 162         else                                                                  \
 163           {                                                                   \
 164             inptr += 4;                                                       \
 165             ++*irreversible;                                                  \
 166           }                                                                   \
 167         continue;                                                             \
 168       }                                                                       \
 169                                                                               \
 170     *outptr++ = cp[0];                                                        \
 171     /* Now test for a possible second byte and write this if possible.  */    \
 172     if (cp[1] != '\0')                                                        \
 173       {                                                                       \
 174         if (__builtin_expect (outptr >= outend, 0))                           \
 175           {                                                                   \
 176             /* The result does not fit into the buffer.  */                   \
 177             --outptr;                                                         \
 178             result = __GCONV_FULL_OUTPUT;                                     \
 179             break;                                                            \
 180           }                                                                   \
 181         *outptr++ = cp[1];                                                    \
 182       }                                                                       \
 183                                                                               \
 184     inptr += 4;                                                               \
 185   }
 186 #define LOOP_NEED_FLAGS
 187 #include <iconv/loop.c>
 188
 189
 190 /* Now define the toplevel functions.  */
 191 #include <iconv/skeleton.c>