iconvdata/utf-32.c

   1 /* Conversion module for UTF-32.
   2    Copyright (C) 1999, 2000-2002 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4
   5    The GNU C Library is free software; you can redistribute it and/or
   6    modify it under the terms of the GNU Lesser General Public
   7    License as published by the Free Software Foundation; either
   8    version 2.1 of the License, or (at your option) any later version.
   9
  10    The GNU C Library is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13    Lesser General Public License for more details.
  14
  15    You should have received a copy of the GNU Lesser General Public
  16    License along with the GNU C Library; if not, write to the Free
  17    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  18    02111-1307 USA.  */
  19
  20 #include <byteswap.h>
  21 #include <dlfcn.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     0x0000feffu
  30 /* And in the other byte order.  */
  31 #define BOM_OE  0xfffe0000u
  32
  33
  34 /* Definitions used in the body of the `gconv' function.  */
  35 #define FROM_LOOP               from_utf32_loop
  36 #define TO_LOOP                 to_utf32_loop
  37 #define DEFINE_INIT             0
  38 #define DEFINE_FINI             0
  39 #define MIN_NEEDED_FROM         4
  40 #define MIN_NEEDED_TO           4
  41 #define FROM_DIRECTION          (dir == from_utf32)
  42 #define PREPARE_LOOP \
  43   enum direction dir = ((struct utf32_data *) step->__data)->dir;             \
  44   enum variant var = ((struct utf32_data *) step->__data)->var;               \
  45   int swap;                                                                   \
  46   if (FROM_DIRECTION && var == UTF_32)                                        \
  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 + 4 > inend)                                              \
  52             return (inptr == inend                                            \
  53                     ? __GCONV_EMPTY_INPUT : __GCONV_INCOMPLETE_INPUT);        \
  54                                                                               \
  55           if (get32u (inptr) == BOM)                                          \
  56             /* Simply ignore the BOM character.  */                           \
  57             *inptrp = inptr += 4;                                             \
  58           else if (get32u (inptr) == BOM_OE)                                  \
  59             {                                                                 \
  60               ((struct utf32_data *) step->__data)->swap = 1;                 \
  61               *inptrp = inptr += 4;                                           \
  62             }                                                                 \
  63         }                                                                     \
  64     }                                                                         \
  65   else if (!FROM_DIRECTION && var == UTF_32 && !data->__internal_use          \
  66            && data->__invocation_counter == 0)                                \
  67     {                                                                         \
  68       /* Emit the Byte Order Mark.  */                                        \
  69       if (__builtin_expect (outbuf + 4 > outend, 0))                          \
  70         return __GCONV_FULL_OUTPUT;                                           \
  71                                                                               \
  72       put32u (outbuf, BOM);                                                   \
  73       outbuf += 4;                                                            \
  74     }                                                                         \
  75   swap = ((struct utf32_data *) step->__data)->swap;
  76 #define EXTRA_LOOP_ARGS         , var, swap
  77
  78
  79 /* Direction of the transformation.  */
  80 enum direction
  81 {
  82   illegal_dir,
  83   to_utf32,
  84   from_utf32
  85 };
  86
  87 enum variant
  88 {
  89   illegal_var,
  90   UTF_32,
  91   UTF_32LE,
  92   UTF_32BE
  93 };
  94
  95 struct utf32_data
  96 {
  97   enum direction dir;
  98   enum variant var;
  99   int swap;
 100 };
 101
 102
 103 extern int gconv_init (struct __gconv_step *step);
 104 int
 105 gconv_init (struct __gconv_step *step)
 106 {
 107   /* Determine which direction.  */
 108   struct utf32_data *new_data;
 109   enum direction dir = illegal_dir;
 110   enum variant var = illegal_var;
 111   int result;
 112
 113   if (__strcasecmp (step->__from_name, "UTF-32//") == 0)
 114     {
 115       dir = from_utf32;
 116       var = UTF_32;
 117     }
 118   else if (__strcasecmp (step->__to_name, "UTF-32//") == 0)
 119     {
 120       dir = to_utf32;
 121       var = UTF_32;
 122     }
 123   else if (__strcasecmp (step->__from_name, "UTF-32BE//") == 0)
 124     {
 125       dir = from_utf32;
 126       var = UTF_32BE;
 127     }
 128   else if (__strcasecmp (step->__to_name, "UTF-32BE//") == 0)
 129     {
 130       dir = to_utf32;
 131       var = UTF_32BE;
 132     }
 133   else if (__strcasecmp (step->__from_name, "UTF-32LE//") == 0)
 134     {
 135       dir = from_utf32;
 136       var = UTF_32LE;
 137     }
 138   else if (__strcasecmp (step->__to_name, "UTF-32LE//") == 0)
 139     {
 140       dir = to_utf32;
 141       var = UTF_32LE;
 142     }
 143
 144   result = __GCONV_NOCONV;
 145   if (__builtin_expect (dir, to_utf32) != illegal_dir)
 146     {
 147       new_data = (struct utf32_data *) malloc (sizeof (struct utf32_data));
 148
 149       result = __GCONV_NOMEM;
 150       if (new_data != NULL)
 151         {
 152           new_data->dir = dir;
 153           new_data->var = var;
 154           new_data->swap = ((var == UTF_32LE && BYTE_ORDER == BIG_ENDIAN)
 155                             || (var == UTF_32BE
 156                                 && BYTE_ORDER == LITTLE_ENDIAN));
 157           step->__data = new_data;
 158
 159           if (dir == from_utf32)
 160             {
 161               step->__min_needed_from = MIN_NEEDED_FROM;
 162               step->__max_needed_from = MIN_NEEDED_FROM;
 163               step->__min_needed_to = MIN_NEEDED_TO;
 164               step->__max_needed_to = MIN_NEEDED_TO;
 165             }
 166           else
 167             {
 168               step->__min_needed_from = MIN_NEEDED_TO;
 169               step->__max_needed_from = MIN_NEEDED_TO;
 170               step->__min_needed_to = MIN_NEEDED_FROM;
 171               step->__max_needed_to = MIN_NEEDED_FROM;
 172             }
 173
 174           step->__stateful = 0;
 175
 176           result = __GCONV_OK;
 177         }
 178     }
 179
 180   return result;
 181 }
 182
 183
 184 extern void gconv_end (struct __gconv_step *data);
 185 void
 186 gconv_end (struct __gconv_step *data)
 187 {
 188   free (data->__data);
 189 }
 190
 191
 192 /* Convert from the internal (UCS4-like) format to UTF-32.  */
 193 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 194 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 195 #define LOOPFCT                 TO_LOOP
 196 #define BODY \
 197   {                                                                           \
 198     uint32_t c = get32 (inptr);                                               \
 199                                                                               \
 200     if (__builtin_expect (c >= 0x110000, 0))                                  \
 201       {                                                                       \
 202         STANDARD_TO_LOOP_ERR_HANDLER (4);                                     \
 203       }                                                                       \
 204     else if (__builtin_expect (c >= 0xd800 && c < 0xe000, 0))                 \
 205       {                                                                       \
 206         /* Surrogate characters in UCS-4 input are not valid.                 \
 207            We must catch this.  If we let surrogates pass through,            \
 208            attackers could make a security hole exploit by                    \
 209            generating "irregular UTF-32" sequences.  */                       \
 210         result = __GCONV_ILLEGAL_INPUT;                                       \
 211         if (! ignore_errors_p ())                                             \
 212           break;                                                              \
 213         inptr += 4;                                                           \
 214         ++*irreversible;                                                      \
 215         continue;                                                             \
 216       }                                                                       \
 217                                                                               \
 218     if (swap)                                                                 \
 219       put32 (outptr, bswap_32 (c));                                           \
 220     else                                                                      \
 221       put32 (outptr, c);                                                      \
 222                                                                               \
 223     outptr += 4;                                                              \
 224     inptr += 4;                                                               \
 225   }
 226 #define LOOP_NEED_FLAGS
 227 #define EXTRA_LOOP_DECLS \
 228         , enum variant var, int swap
 229 #include <iconv/loop.c>
 230
 231
 232 /* Convert from UTF-32 to the internal (UCS4-like) format.  */
 233 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 234 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 235 #define LOOPFCT                 FROM_LOOP
 236 #define BODY \
 237   {                                                                           \
 238     uint32_t u1 = get32 (inptr);                                              \
 239                                                                               \
 240     if (swap)                                                                 \
 241       u1 = bswap_32 (u1);                                                     \
 242                                                                               \
 243     if (__builtin_expect (u1 >= 0x110000, 0))                                 \
 244       {                                                                       \
 245         /* This is illegal.  */                                               \
 246         STANDARD_FROM_LOOP_ERR_HANDLER (4);                                   \
 247       }                                                                       \
 248                                                                               \
 249     put32 (outptr, u1);                                                       \
 250     inptr += 4;                                                               \
 251     outptr += 4;                                                              \
 252   }
 253 #define LOOP_NEED_FLAGS
 254 #define EXTRA_LOOP_DECLS \
 255         , enum variant var, int swap
 256 #include <iconv/loop.c>
 257
 258
 259 /* Now define the toplevel functions.  */
 260 #include <iconv/skeleton.c>