iconvdata/unicode.c

   1 /* Conversion module for Unicode
   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 <byteswap.h>
  22 #include <dlfcn.h>
  23 #include <gconv.h>
  24 #include <stddef.h>
  25 #include <stdint.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28
  29 /* This is the Byte Order Mark character (BOM).  */
  30 #define BOM     0xfeff
  31 /* And in the other endian format.  */
  32 #define BOM_OE  0xfffe
  33
  34
  35 /* Definitions used in the body of the `gconv' function.  */
  36 #define FROM_LOOP               from_unicode_loop
  37 #define TO_LOOP                 to_unicode_loop
  38 #define DEFINE_INIT             0
  39 #define DEFINE_FINI             0
  40 #define MIN_NEEDED_FROM         2
  41 #define MIN_NEEDED_TO           4
  42 #define FROM_DIRECTION          (dir == from_unicode)
  43 #define PREPARE_LOOP \
  44   enum direction dir = ((struct unicode_data *) step->__data)->dir;           \
  45   int swap;                                                                   \
  46   if (FROM_DIRECTION)                                                         \
  47     {                                                                         \
  48       if (data->__invocation_counter == 0)                                    \
  49         {                                                                     \
  50           /* We have to find out which byte order the file is encoded in.  */ \
  51           if (inptr + 2 > inend)                                              \
  52             return __GCONV_EMPTY_INPUT;                                       \
  53                                                                               \
  54           if (get16u (inptr) == BOM)                                          \
  55             /* Simply ignore the BOM character.  */                           \
  56             inptr += 2;                                                       \
  57           else if (get16u (inptr) == BOM_OE)                                  \
  58             {                                                                 \
  59               ((struct unicode_data *) step->__data)->swap = 1;               \
  60               inptr += 2;                                                     \
  61             }                                                                 \
  62         }                                                                     \
  63     }                                                                         \
  64   else if (!data->__internal_use && data->__invocation_counter == 0)          \
  65     {                                                                         \
  66       /* Emit the Byte Order Mark.  */                                        \
  67       if (__builtin_expect (outbuf + 2 > outend, 0))                          \
  68         return __GCONV_FULL_OUTPUT;                                           \
  69                                                                               \
  70       put16u (outbuf, BOM);                                                   \
  71       outbuf += 2;                                                            \
  72     }                                                                         \
  73   swap = ((struct unicode_data *) step->__data)->swap;
  74 #define EXTRA_LOOP_ARGS         , swap
  75
  76
  77 /* Direction of the transformation.  */
  78 enum direction
  79 {
  80   illegal_dir,
  81   to_unicode,
  82   from_unicode
  83 };
  84
  85 struct unicode_data
  86 {
  87   enum direction dir;
  88   int swap;
  89 };
  90
  91
  92 int
  93 gconv_init (struct __gconv_step *step)
  94 {
  95   /* Determine which direction.  */
  96   struct unicode_data *new_data;
  97   enum direction dir = illegal_dir;
  98   int result;
  99
 100   if (__strcasecmp (step->__from_name, "UNICODE") == 0)
 101     dir = from_unicode;
 102   else
 103     dir = to_unicode;
 104
 105   new_data = (struct unicode_data *) malloc (sizeof (struct unicode_data));
 106
 107   result = __GCONV_NOMEM;
 108   if (new_data != NULL)
 109     {
 110       new_data->dir = dir;
 111       new_data->swap = 0;
 112       step->__data = new_data;
 113
 114       if (dir == from_unicode)
 115         {
 116           step->__min_needed_from = MIN_NEEDED_FROM;
 117           step->__max_needed_from = MIN_NEEDED_FROM;
 118           step->__min_needed_to = MIN_NEEDED_TO;
 119           step->__max_needed_to = MIN_NEEDED_TO;
 120         }
 121       else
 122         {
 123           step->__min_needed_from = MIN_NEEDED_TO;
 124           step->__max_needed_from = MIN_NEEDED_TO;
 125           step->__min_needed_to = MIN_NEEDED_FROM;
 126           step->__max_needed_to = MIN_NEEDED_FROM;
 127         }
 128
 129       step->__stateful = 0;
 130
 131       result = __GCONV_OK;
 132     }
 133
 134   return result;
 135 }
 136
 137
 138 void
 139 gconv_end (struct __gconv_step *data)
 140 {
 141   free (data->__data);
 142 }
 143
 144
 145 /* Convert from the internal (UCS4-like) format to UCS2.  */
 146 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 147 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 148 #define LOOPFCT                 TO_LOOP
 149 #define BODY \
 150   {                                                                           \
 151     uint32_t c = get32 (inptr);                                               \
 152                                                                               \
 153     if (__builtin_expect (c, 0) >= 0x10000)                                   \
 154       {                                                                       \
 155         if (step_data->__trans.__trans_fct != NULL)                           \
 156           {                                                                   \
 157             result = DL_CALL_FCT (step_data->__trans.__trans_fct,             \
 158                                   (step, step_data, *inptrp, &inptr, inend,   \
 159                                    *outptrp, &outptr, outend, irreversible)); \
 160             if (result != __GCONV_OK)                                         \
 161               break;                                                          \
 162           }                                                                   \
 163         else if (! ignore_errors_p ())                                        \
 164           {                                                                   \
 165             /* This is an illegal character.  */                              \
 166             result = __GCONV_ILLEGAL_INPUT;                                   \
 167             break;                                                            \
 168           }                                                                   \
 169         else                                                                  \
 170           {                                                                   \
 171             ++*irreversible;                                                  \
 172             inptr += 4;                                                       \
 173           }                                                                   \
 174         continue;                                                             \
 175       }                                                                       \
 176     else                                                                      \
 177       {                                                                       \
 178         put16 (outptr, c);                                                    \
 179         outptr += 2;                                                          \
 180       }                                                                       \
 181                                                                               \
 182     inptr += 4;                                                               \
 183   }
 184 #define LOOP_NEED_FLAGS
 185 #define EXTRA_LOOP_DECLS \
 186         , int swap
 187 #include <iconv/loop.c>
 188
 189
 190 /* Convert from UCS2 to the internal (UCS4-like) format.  */
 191 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 192 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 193 #define LOOPFCT                 FROM_LOOP
 194 #define BODY \
 195   {                                                                           \
 196     uint16_t u1 = get16 (inptr);                                              \
 197                                                                               \
 198     if (swap)                                                                 \
 199       u1 = bswap_16 (u1);                                                     \
 200                                                                               \
 201     put32 (outptr, u1);                                                       \
 202                                                                               \
 203     inptr += 2;                                                               \
 204     outptr += 4;                                                              \
 205   }
 206 #define EXTRA_LOOP_DECLS \
 207         , int swap
 208 #include <iconv/loop.c>
 209
 210
 211 /* Now define the toplevel functions.  */
 212 #include <iconv/skeleton.c>