iconvdata/gbgbk.c

   1 /* Mapping tables from GBK to GB2312 and vice versa.
   2    Copyright (C) 1999, 2000 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1999.
   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 <gconv.h>
  22 #include <stdint.h>
  23
  24
  25 /* Definitions used in the body of the `gconv' function.  */
  26 #define CHARSET_NAME            "GBK//"
  27 #define FROM_LOOP               from_gbk_to_gb
  28 #define TO_LOOP                 from_gb_to_gbk
  29 #define DEFINE_INIT             1
  30 #define DEFINE_FINI             1
  31 #define MIN_NEEDED_FROM         1
  32 #define MAX_NEEDED_FROM         2
  33 #define MIN_NEEDED_TO           1
  34 #define MAX_NEEDED_TO           2
  35
  36
  37 /* First define the conversion function from GBK to GB2312.  */
  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 MAX_NEEDED_OUTPUT       MAX_NEEDED_TO
  42 #define LOOPFCT                 FROM_LOOP
  43 #define BODY \
  44   {                                                                           \
  45     uint32_t ch = *inptr;                                                     \
  46                                                                               \
  47     if (ch <= 0x7f)                                                           \
  48       *outptr++ = *inptr++;                                                   \
  49     else                                                                      \
  50       {                                                                       \
  51         /* It's a two-byte sequence.  We have to mask out all the sequences   \
  52            which are not in GB2312.  Besides all of them in the range         \
  53            0x8140 to 0xA0FE this also includes in the remaining range the     \
  54            sequences which the second byte being in the range from 0x40 to    \
  55            0xA0 and the following exceptions:                                 \
  56                                                                               \
  57              0xA2A1 to 0xA2A9,                                                \
  58              0xA2AA,                                                          \
  59              0xA6E0 to 0xA6EB,                                                \
  60              0xA6EE to 0xA6F2,                                                \
  61              0xA6F4, 0xA6F5,                                                  \
  62              0xA8BB to 0xA8C0                                                 \
  63                                                                               \
  64            All these characters are not defined in GB2312.  Besides this      \
  65            there is an incomatibility in the mapping.  The Unicode tables     \
  66            say that 0xA1A4 maps in GB2312 to U30FB while in GBK it maps to    \
  67            U00B7.  Since we are free to do whatever we want if a mapping      \
  68            is not available we will not flag this as an error but instead     \
  69            map the two positions.  But this means that the mapping            \
  70                                                                               \
  71                 UCS4 -> GB2312 -> GBK -> UCS4                                 \
  72                                                                               \
  73            might not produce identical text.  */                              \
  74         if (NEED_LENGTH_TEST && __builtin_expect (inptr + 1 >= inend, 0))     \
  75           {                                                                   \
  76             /* The second character is not available.  Store                  \
  77                the intermediate result.  */                                   \
  78             result = __GCONV_INCOMPLETE_INPUT;                                \
  79             break;                                                            \
  80           }                                                                   \
  81                                                                               \
  82         if (NEED_LENGTH_TEST && __builtin_expect (outend - outptr < 2, 0))    \
  83           {                                                                   \
  84             /* We ran out of space.  */                                       \
  85             result = __GCONV_FULL_OUTPUT;                                     \
  86             break;                                                            \
  87           }                                                                   \
  88                                                                               \
  89         ch = (ch << 8) | inptr[1];                                            \
  90                                                                               \
  91         /* Now determine whether the character is valid.  */                  \
  92         if (__builtin_expect (ch, 0xa1a1) < 0xa1a1                            \
  93             || __builtin_expect (ch, 0xa1a1) > 0xf7fe                         \
  94             || __builtin_expect (inptr[1], 0xa1) < 0xa1                       \
  95             /* Now test the exceptions.  */                                   \
  96             || (__builtin_expect (ch, 0xa1a1) >= 0xa2a1                       \
  97                 && __builtin_expect (ch, 0xa1a1) <= 0xa2aa)                   \
  98             || (__builtin_expect (ch, 0xa1a1) >= 0xa6e0                       \
  99                 && __builtin_expect (ch, 0xa1a1) <= 0xa6f5)                   \
 100             || (__builtin_expect (ch, 0xa1a1) >= 0xa8bb                       \
 101                 && __builtin_expect (ch, 0xa1a1) <= 0xa8c0))                  \
 102           {                                                                   \
 103             /* One of the characters we cannot map.  */                       \
 104             if (! ignore_errors_p ())                                         \
 105               {                                                               \
 106                 result = __GCONV_ILLEGAL_INPUT;                               \
 107                 break;                                                        \
 108               }                                                               \
 109                                                                               \
 110             inptr += 2;                                                       \
 111             ++*irreversible;                                                  \
 112           }                                                                   \
 113         else                                                                  \
 114           {                                                                   \
 115             /* Copy the two bytes.  */                                        \
 116             *outptr++ = *inptr++;                                             \
 117             *outptr++ = *inptr++;                                             \
 118           }                                                                   \
 119       }                                                                       \
 120   }
 121 #include <iconv/loop.c>
 122
 123
 124 /* Next, define the other direction.  */
 125 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 126 #define MAX_NEEDED_INPUT        MAX_NEEDED_TO
 127 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 128 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 129 #define LOOPFCT                 TO_LOOP
 130 #define BODY \
 131   {                                                                           \
 132     /* We don't have to care about characters we cannot map.  The only        \
 133        problem is the mapping of 0xA1A4 but as explained above we do not      \
 134        do anything special here.  */                                          \
 135     unsigned char ch = *inptr++;                                              \
 136                                                                               \
 137     if (ch > 0x7f)                                                            \
 138       {                                                                       \
 139         if (NEED_LENGTH_TEST && __builtin_expect (inptr + 1 >= inend, 0))     \
 140           {                                                                   \
 141             /* The second character is not available.  Store                  \
 142                  the intermediate result.  */                                 \
 143             result = __GCONV_INCOMPLETE_INPUT;                                \
 144             break;                                                            \
 145           }                                                                   \
 146                                                                               \
 147         if (NEED_LENGTH_TEST && __builtin_expect (outend - outptr < 2, 0))    \
 148           {                                                                   \
 149             /* We ran out of space.  */                                       \
 150             result = __GCONV_FULL_OUTPUT;                                     \
 151             break;                                                            \
 152           }                                                                   \
 153                                                                               \
 154         *outptr++ = ch;                                                       \
 155         ch = *inptr++;                                                        \
 156       }                                                                       \
 157     *outptr++ = ch;                                                           \
 158   }
 159 #include <iconv/loop.c>
 160
 161
 162 /* Now define the toplevel functions.  */
 163 #include <iconv/skeleton.c>