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