lib/iconv.c

   1 /*
   2  * Copyright (C) 1999-2001 Free Software Foundation, Inc.
   3  * This file is part of the GNU LIBICONV Library.
   4  *
   5  * The GNU LIBICONV Library is free software; you can redistribute it
   6  * and/or modify it under the terms of the GNU Library General Public
   7  * License as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * The GNU LIBICONV Library is distributed in the hope that it will be
  11  * useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * Library General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Library General Public
  16  * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
  17  * If not, write to the Free Software Foundation, Inc., 59 Temple Place -
  18  * Suite 330, Boston, MA 02111-1307, USA.
  19  */
  20
  21 #include <iconv.h>
  22
  23 #include <stdlib.h>
  24 #include <string.h>
  25 #include "config.h"
  26 #include "libcharset.h"
  27
  28 /*
  29  * Consider those system dependent encodings that are needed for the
  30  * current system.
  31  */
  32 #ifdef _AIX
  33 #define USE_AIX
  34 #endif
  35 #ifdef __osf__
  36 #define USE_OSF1
  37 #endif
  38 #ifdef __DJGPP__
  39 #define USE_DOS
  40 #endif
  41
  42 /*
  43  * Data type for general conversion loop.
  44  */
  45 struct loop_funcs {
  46   size_t (*loop_convert) (iconv_t icd,
  47                           const char* * inbuf, size_t *inbytesleft,
  48                           char* * outbuf, size_t *outbytesleft);
  49   size_t (*loop_reset) (iconv_t icd,
  50                         char* * outbuf, size_t *outbytesleft);
  51 };
  52
  53 /*
  54  * Converters.
  55  */
  56 #include "converters.h"
  57
  58 /*
  59  * Transliteration tables.
  60  */
  61 #include "cjk_variants.h"
  62 #include "translit.h"
  63
  64 /*
  65  * Table of all supported encodings.
  66  */
  67 struct encoding {
  68   struct mbtowc_funcs ifuncs; /* conversion multibyte -> unicode */
  69   struct wctomb_funcs ofuncs; /* conversion unicode -> multibyte */
  70   int oflags;                 /* flags for unicode -> multibyte conversion */
  71 };
  72 enum {
  73 #define DEFENCODING(xxx_names,xxx,xxx_ifuncs,xxx_ofuncs1,xxx_ofuncs2) \
  74   ei_##xxx ,
  75 #include "encodings.def"
  76 #ifdef USE_AIX
  77 #include "encodings_aix.def"
  78 #endif
  79 #ifdef USE_OSF1
  80 #include "encodings_osf1.def"
  81 #endif
  82 #ifdef USE_DOS
  83 #include "encodings_dos.def"
  84 #endif
  85 #include "encodings_local.def"
  86 #undef DEFENCODING
  87 ei_for_broken_compilers_that_dont_like_trailing_commas
  88 };
  89 #include "flags.h"
  90 static struct encoding const all_encodings[] = {
  91 #define DEFENCODING(xxx_names,xxx,xxx_ifuncs,xxx_ofuncs1,xxx_ofuncs2) \
  92   { xxx_ifuncs, xxx_ofuncs1,xxx_ofuncs2, ei_##xxx##_oflags },
  93 #include "encodings.def"
  94 #ifdef USE_AIX
  95 #include "encodings_aix.def"
  96 #endif
  97 #ifdef USE_OSF1
  98 #include "encodings_osf1.def"
  99 #endif
 100 #ifdef USE_DOS
 101 #include "encodings_dos.def"
 102 #endif
 103 #undef DEFENCODING
 104 #define DEFENCODING(xxx_names,xxx,xxx_ifuncs,xxx_ofuncs1,xxx_ofuncs2) \
 105   { xxx_ifuncs, xxx_ofuncs1,xxx_ofuncs2, 0 },
 106 #include "encodings_local.def"
 107 #undef DEFENCODING
 108 };
 109
 110 /*
 111  * Conversion loops.
 112  */
 113 #include "loops.h"
 114
 115 /*
 116  * Alias lookup function.
 117  * Defines
 118  *   struct alias { const char* name; unsigned int encoding_index; };
 119  *   const struct alias * aliases_lookup (const char *str, unsigned int len);
 120  *   #define MAX_WORD_LENGTH ...
 121  */
 122 #include "aliases.h"
 123
 124 /*
 125  * System dependent alias lookup function.
 126  * Defines
 127  *   const struct alias * aliases2_lookup (const char *str);
 128  */
 129 #if defined(USE_AIX) || defined(USE_OSF1) || defined(USE_DOS) /* || ... */
 130 static struct alias sysdep_aliases[] = {
 131 #ifdef USE_AIX
 132 #include "aliases_aix.h"
 133 #endif
 134 #ifdef USE_OSF1
 135 #include "aliases_osf1.h"
 136 #endif
 137 #ifdef USE_DOS
 138 #include "aliases_dos.h"
 139 #endif
 140 };
 141 #ifdef __GNUC__
 142 __inline
 143 #endif
 144 const struct alias *
 145 aliases2_lookup (register const char *str)
 146 {
 147   struct alias * ptr;
 148   unsigned int count;
 149   for (ptr = sysdep_aliases, count = sizeof(sysdep_aliases)/sizeof(sysdep_aliases[0]); count > 0; ptr++, count--)
 150     if (!strcmp(str,ptr->name))
 151       return ptr;
 152   return NULL;
 153 }
 154 #else
 155 #define aliases2_lookup(str)  NULL
 156 #endif
 157
 158 #if 0
 159 /* Like !strcasecmp, except that the both strings can be assumed to be ASCII
 160    and the first string can be assumed to be in uppercase. */
 161 static int strequal (const char* str1, const char* str2)
 162 {
 163   unsigned char c1;
 164   unsigned char c2;
 165   for (;;) {
 166     c1 = * (unsigned char *) str1++;
 167     c2 = * (unsigned char *) str2++;
 168     if (c1 == 0)
 169       break;
 170     if (c2 >= 'a' && c2 <= 'z')
 171       c2 -= 'a'-'A';
 172     if (c1 != c2)
 173       break;
 174   }
 175   return (c1 == c2);
 176 }
 177 #endif
 178
 179 iconv_t iconv_open (const char* tocode, const char* fromcode)
 180 {
 181   struct conv_struct * cd;
 182   char buf[MAX_WORD_LENGTH+10+1];
 183   const char* cp;
 184   char* bp;
 185   const struct alias * ap;
 186   unsigned int count;
 187   unsigned int from_index;
 188   int from_wchar;
 189   unsigned int to_index;
 190   int to_wchar;
 191   int transliterate = 0;
 192
 193   /* Before calling aliases_lookup, convert the input string to upper case,
 194    * and check whether it's entirely ASCII (we call gperf with option "-7"
 195    * to achieve a smaller table) and non-empty. If it's not entirely ASCII,
 196    * or if it's too long, it is not a valid encoding name.
 197    */
 198   for (to_wchar = 0;;) {
 199     /* Search tocode in the table. */
 200     for (cp = tocode, bp = buf, count = MAX_WORD_LENGTH+10+1; ; cp++, bp++) {
 201       unsigned char c = * (unsigned char *) cp;
 202       if (c >= 0x80)
 203         goto invalid;
 204       if (c >= 'a' && c <= 'z')
 205         c -= 'a'-'A';
 206       *bp = c;
 207       if (c == '\0')
 208         break;
 209       if (--count == 0)
 210         goto invalid;
 211     }
 212     if (bp-buf > 10 && memcmp(bp-10,"//TRANSLIT",10)==0) {
 213       bp -= 10;
 214       *bp = '\0';
 215       transliterate = 1;
 216     }
 217     ap = aliases_lookup(buf,bp-buf);
 218     if (ap == NULL) {
 219       ap = aliases2_lookup(buf);
 220       if (ap == NULL)
 221         goto invalid;
 222     }
 223     if (ap->encoding_index == ei_local_char) {
 224       tocode = locale_charset();
 225       continue;
 226     }
 227     if (ap->encoding_index == ei_local_wchar_t) {
 228 #if __STDC_ISO_10646__
 229       if (sizeof(wchar_t) == 4) {
 230         to_index = ei_ucs4internal;
 231         break;
 232       }
 233       if (sizeof(wchar_t) == 2) {
 234         to_index = ei_ucs2internal;
 235         break;
 236       }
 237       if (sizeof(wchar_t) == 1) {
 238         to_index = ei_iso8859_1;
 239         break;
 240       }
 241 #endif
 242 #if HAVE_MBRTOWC
 243       to_wchar = 1;
 244       tocode = locale_charset();
 245       continue;
 246 #endif
 247       goto invalid;
 248     }
 249     to_index = ap->encoding_index;
 250     break;
 251   }
 252   for (from_wchar = 0;;) {
 253     /* Search fromcode in the table. */
 254     for (cp = fromcode, bp = buf, count = MAX_WORD_LENGTH+10+1; ; cp++, bp++) {
 255       unsigned char c = * (unsigned char *) cp;
 256       if (c >= 0x80)
 257         goto invalid;
 258       if (c >= 'a' && c <= 'z')
 259         c -= 'a'-'A';
 260       *bp = c;
 261       if (c == '\0')
 262         break;
 263       if (--count == 0)
 264         goto invalid;
 265     }
 266     if (bp-buf > 10 && memcmp(bp-10,"//TRANSLIT",10)==0) {
 267       bp -= 10;
 268       *bp = '\0';
 269     }
 270     ap = aliases_lookup(buf,bp-buf);
 271     if (ap == NULL) {
 272       ap = aliases2_lookup(buf);
 273       if (ap == NULL)
 274         goto invalid;
 275     }
 276     if (ap->encoding_index == ei_local_char) {
 277       fromcode = locale_charset();
 278       continue;
 279     }
 280     if (ap->encoding_index == ei_local_wchar_t) {
 281 #if __STDC_ISO_10646__
 282       if (sizeof(wchar_t) == 4) {
 283         from_index = ei_ucs4internal;
 284         break;
 285       }
 286       if (sizeof(wchar_t) == 2) {
 287         from_index = ei_ucs2internal;
 288         break;
 289       }
 290       if (sizeof(wchar_t) == 1) {
 291         from_index = ei_iso8859_1;
 292         break;
 293       }
 294 #endif
 295 #if HAVE_WCRTOMB
 296       from_wchar = 1;
 297       fromcode = locale_charset();
 298       continue;
 299 #endif
 300       goto invalid;
 301     }
 302     from_index = ap->encoding_index;
 303     break;
 304   }
 305   cd = (struct conv_struct *) malloc(from_wchar != to_wchar
 306                                      ? sizeof(struct wchar_conv_struct)
 307                                      : sizeof(struct conv_struct));
 308   if (cd == NULL) {
 309     errno = ENOMEM;
 310     return (iconv_t)(-1);
 311   }
 312   cd->iindex = from_index;
 313   cd->ifuncs = all_encodings[from_index].ifuncs;
 314   cd->oindex = to_index;
 315   cd->ofuncs = all_encodings[to_index].ofuncs;
 316   cd->oflags = all_encodings[to_index].oflags;
 317   /* Initialize the loop functions. */
 318 #if HAVE_MBRTOWC
 319   if (to_wchar) {
 320 #if HAVE_WCRTOMB
 321     if (from_wchar) {
 322       cd->lfuncs.loop_convert = wchar_id_loop_convert;
 323       cd->lfuncs.loop_reset = wchar_id_loop_reset;
 324     } else
 325 #endif
 326     {
 327       cd->lfuncs.loop_convert = wchar_to_loop_convert;
 328       cd->lfuncs.loop_reset = wchar_to_loop_reset;
 329     }
 330   } else
 331 #endif
 332   {
 333 #if HAVE_WCRTOMB
 334     if (from_wchar) {
 335       cd->lfuncs.loop_convert = wchar_from_loop_convert;
 336       cd->lfuncs.loop_reset = wchar_from_loop_reset;
 337     } else
 338 #endif
 339     {
 340       cd->lfuncs.loop_convert = unicode_loop_convert;
 341       cd->lfuncs.loop_reset = unicode_loop_reset;
 342     }
 343   }
 344   /* Initialize the states. */
 345   memset(&cd->istate,'\0',sizeof(state_t));
 346   memset(&cd->ostate,'\0',sizeof(state_t));
 347   /* Initialize the operation flags. */
 348   cd->transliterate = transliterate;
 349   /* Initialize additional fields. */
 350   if (from_wchar != to_wchar) {
 351     struct wchar_conv_struct * wcd = (struct wchar_conv_struct *) cd;
 352     memset(&wcd->state,'\0',sizeof(mbstate_t));
 353   }
 354   /* Done. */
 355   return (iconv_t)cd;
 356 invalid:
 357   errno = EINVAL;
 358   return (iconv_t)(-1);
 359 }
 360
 361 size_t iconv (iconv_t icd,
 362               ICONV_CONST char* * inbuf, size_t *inbytesleft,
 363               char* * outbuf, size_t *outbytesleft)
 364 {
 365   conv_t cd = (conv_t) icd;
 366   if (inbuf == NULL || *inbuf == NULL)
 367     return cd->lfuncs.loop_reset(icd,outbuf,outbytesleft);
 368   else
 369     return cd->lfuncs.loop_convert(icd,
 370                                    (const char* *)inbuf,inbytesleft,
 371                                    outbuf,outbytesleft);
 372 }
 373
 374 int iconv_close (iconv_t icd)
 375 {
 376   conv_t cd = (conv_t) icd;
 377   free(cd);
 378   return 0;
 379 }
 380
 381 #ifndef LIBICONV_PLUG
 382
 383 int iconvctl (iconv_t icd, int request, void* argument)
 384 {
 385   conv_t cd = (conv_t) icd;
 386   switch (request) {
 387     case ICONV_TRIVIALP:
 388       *(int *)argument =
 389         ((cd->lfuncs.loop_convert == unicode_loop_convert
 390           && cd->iindex == cd->oindex)
 391          || cd->lfuncs.loop_convert == wchar_id_loop_convert
 392          ? 1 : 0);
 393       return 0;
 394     case ICONV_GET_TRANSLITERATE:
 395       *(int *)argument = cd->transliterate;
 396       return 0;
 397     case ICONV_SET_TRANSLITERATE:
 398       cd->transliterate = (*(const int *)argument ? 1 : 0);
 399       return 0;
 400     default:
 401       errno = EINVAL;
 402       return -1;
 403   }
 404 }
 405
 406 int _libiconv_version = _LIBICONV_VERSION;
 407
 408 #endif