gnu/dist/gettext/gettext-tools/lib/localcharset.c

   1 /* Determine a canonical name for the current locale's character encoding.
   2
   3    Copyright (C) 2000-2004 Free Software Foundation, Inc.
   4
   5    This program is free software; you can redistribute it and/or modify it
   6    under the terms of the GNU Library General Public License as published
   7    by the Free Software Foundation; either version 2, or (at your option)
   8    any later version.
   9
  10    This program 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    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 this program; if not, write to the Free Software
  17    Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
  18    USA.  */
  19
  20 /* Written by Bruno Haible <bruno@clisp.org>.  */
  21
  22 #ifdef HAVE_CONFIG_H
  23 # include <config.h>
  24 #endif
  25
  26 /* Specification.  */
  27 #include "localcharset.h"
  28
  29 #if HAVE_STDDEF_H
  30 # include <stddef.h>
  31 #endif
  32
  33 #include <stdio.h>
  34 #if HAVE_STRING_H
  35 # include <string.h>
  36 #else
  37 # include <strings.h>
  38 #endif
  39 #if HAVE_STDLIB_H
  40 # include <stdlib.h>
  41 #endif
  42
  43 #if defined _WIN32 || defined __WIN32__
  44 # undef WIN32   /* avoid warning on mingw32 */
  45 # define WIN32
  46 #endif
  47
  48 #if defined __EMX__
  49 /* Assume EMX program runs on OS/2, even if compiled under DOS.  */
  50 # define OS2
  51 #endif
  52
  53 #if !defined WIN32
  54 # if HAVE_LANGINFO_CODESET
  55 #  include <langinfo.h>
  56 # else
  57 #  if HAVE_SETLOCALE
  58 #   include <locale.h>
  59 #  endif
  60 # endif
  61 #elif defined WIN32
  62 # define WIN32_LEAN_AND_MEAN
  63 # include <windows.h>
  64 #endif
  65 #if defined OS2
  66 # define INCL_DOS
  67 # include <os2.h>
  68 #endif
  69
  70 #if ENABLE_RELOCATABLE
  71 # include "relocatable.h"
  72 #else
  73 # define relocate(pathname) (pathname)
  74 #endif
  75
  76 #if defined _WIN32 || defined __WIN32__ || defined __CYGWIN__ || defined __EMX__ || defined __DJGPP__
  77   /* Win32, Cygwin, OS/2, DOS */
  78 # define ISSLASH(C) ((C) == '/' || (C) == '\\')
  79 #endif
  80
  81 #ifndef DIRECTORY_SEPARATOR
  82 # define DIRECTORY_SEPARATOR '/'
  83 #endif
  84
  85 #ifndef ISSLASH
  86 # define ISSLASH(C) ((C) == DIRECTORY_SEPARATOR)
  87 #endif
  88
  89 #if HAVE_DECL_GETC_UNLOCKED
  90 # undef getc
  91 # define getc getc_unlocked
  92 #endif
  93
  94 /* The following static variable is declared 'volatile' to avoid a
  95    possible multithread problem in the function get_charset_aliases. If we
  96    are running in a threaded environment, and if two threads initialize
  97    'charset_aliases' simultaneously, both will produce the same value,
  98    and everything will be ok if the two assignments to 'charset_aliases'
  99    are atomic. But I don't know what will happen if the two assignments mix.  */
 100 #if __STDC__ != 1
 101 # define volatile /* empty */
 102 #endif
 103 /* Pointer to the contents of the charset.alias file, if it has already been
 104    read, else NULL.  Its format is:
 105    ALIAS_1 '\0' CANONICAL_1 '\0' ... ALIAS_n '\0' CANONICAL_n '\0' '\0'  */
 106 static const char * volatile charset_aliases;
 107
 108 /* Return a pointer to the contents of the charset.alias file.  */
 109 static const char *
 110 get_charset_aliases ()
 111 {
 112   const char *cp;
 113
 114   cp = charset_aliases;
 115   if (cp == NULL)
 116     {
 117 #if !(defined VMS || defined WIN32)
 118       FILE *fp;
 119       const char *dir;
 120       const char *base = "charset.alias";
 121       char *file_name;
 122
 123       /* Make it possible to override the charset.alias location.  This is
 124          necessary for running the testsuite before "make install".  */
 125       dir = getenv ("CHARSETALIASDIR");
 126       if (dir == NULL || dir[0] == '\0')
 127         dir = relocate (LIBDIR);
 128
 129       /* Concatenate dir and base into freshly allocated file_name.  */
 130       {
 131         size_t dir_len = strlen (dir);
 132         size_t base_len = strlen (base);
 133         int add_slash = (dir_len > 0 && !ISSLASH (dir[dir_len - 1]));
 134         file_name = (char *) malloc (dir_len + add_slash + base_len + 1);
 135         if (file_name != NULL)
 136           {
 137             memcpy (file_name, dir, dir_len);
 138             if (add_slash)
 139               file_name[dir_len] = DIRECTORY_SEPARATOR;
 140             memcpy (file_name + dir_len + add_slash, base, base_len + 1);
 141           }
 142       }
 143
 144       if (file_name == NULL || (fp = fopen (file_name, "r")) == NULL)
 145         /* Out of memory or file not found, treat it as empty.  */
 146         cp = "";
 147       else
 148         {
 149           /* Parse the file's contents.  */
 150           char *res_ptr = NULL;
 151           size_t res_size = 0;
 152
 153           for (;;)
 154             {
 155               int c;
 156               char buf1[50+1];
 157               char buf2[50+1];
 158               size_t l1, l2;
 159               char *old_res_ptr;
 160
 161               c = getc (fp);
 162               if (c == EOF)
 163                 break;
 164               if (c == '\n' || c == ' ' || c == '\t')
 165                 continue;
 166               if (c == '#')
 167                 {
 168                   /* Skip comment, to end of line.  */
 169                   do
 170                     c = getc (fp);
 171                   while (!(c == EOF || c == '\n'));
 172                   if (c == EOF)
 173                     break;
 174                   continue;
 175                 }
 176               ungetc (c, fp);
 177               if (fscanf (fp, "%50s %50s", buf1, buf2) < 2)
 178                 break;
 179               l1 = strlen (buf1);
 180               l2 = strlen (buf2);
 181               old_res_ptr = res_ptr;
 182               if (res_size == 0)
 183                 {
 184                   res_size = l1 + 1 + l2 + 1;
 185                   res_ptr = (char *) malloc (res_size + 1);
 186                 }
 187               else
 188                 {
 189                   res_size += l1 + 1 + l2 + 1;
 190                   res_ptr = (char *) realloc (res_ptr, res_size + 1);
 191                 }
 192               if (res_ptr == NULL)
 193                 {
 194                   /* Out of memory. */
 195                   res_size = 0;
 196                   if (old_res_ptr != NULL)
 197                     free (old_res_ptr);
 198                   break;
 199                 }
 200               strcpy (res_ptr + res_size - (l2 + 1) - (l1 + 1), buf1);
 201               strcpy (res_ptr + res_size - (l2 + 1), buf2);
 202             }
 203           fclose (fp);
 204           if (res_size == 0)
 205             cp = "";
 206           else
 207             {
 208               *(res_ptr + res_size) = '\0';
 209               cp = res_ptr;
 210             }
 211         }
 212
 213       if (file_name != NULL)
 214         free (file_name);
 215
 216 #else
 217
 218 # if defined VMS
 219       /* To avoid the troubles of an extra file charset.alias_vms in the
 220          sources of many GNU packages, simply inline the aliases here.  */
 221       /* The list of encodings is taken from the OpenVMS 7.3-1 documentation
 222          "Compaq C Run-Time Library Reference Manual for OpenVMS systems"
 223          section 10.7 "Handling Different Character Sets".  */
 224       cp = "ISO8859-1" "\0" "ISO-8859-1" "\0"
 225            "ISO8859-2" "\0" "ISO-8859-2" "\0"
 226            "ISO8859-5" "\0" "ISO-8859-5" "\0"
 227            "ISO8859-7" "\0" "ISO-8859-7" "\0"
 228            "ISO8859-8" "\0" "ISO-8859-8" "\0"
 229            "ISO8859-9" "\0" "ISO-8859-9" "\0"
 230            /* Japanese */
 231            "eucJP" "\0" "EUC-JP" "\0"
 232            "SJIS" "\0" "SHIFT_JIS" "\0"
 233            "DECKANJI" "\0" "DEC-KANJI" "\0"
 234            "SDECKANJI" "\0" "EUC-JP" "\0"
 235            /* Chinese */
 236            "eucTW" "\0" "EUC-TW" "\0"
 237            "DECHANYU" "\0" "DEC-HANYU" "\0"
 238            "DECHANZI" "\0" "GB2312" "\0"
 239            /* Korean */
 240            "DECKOREAN" "\0" "EUC-KR" "\0";
 241 # endif
 242
 243 # if defined WIN32
 244       /* To avoid the troubles of installing a separate file in the same
 245          directory as the DLL and of retrieving the DLL's directory at
 246          runtime, simply inline the aliases here.  */
 247
 248       cp = "CP936" "\0" "GBK" "\0"
 249            "CP1361" "\0" "JOHAB" "\0"
 250            "CP20127" "\0" "ASCII" "\0"
 251            "CP20866" "\0" "KOI8-R" "\0"
 252            "CP21866" "\0" "KOI8-RU" "\0"
 253            "CP28591" "\0" "ISO-8859-1" "\0"
 254            "CP28592" "\0" "ISO-8859-2" "\0"
 255            "CP28593" "\0" "ISO-8859-3" "\0"
 256            "CP28594" "\0" "ISO-8859-4" "\0"
 257            "CP28595" "\0" "ISO-8859-5" "\0"
 258            "CP28596" "\0" "ISO-8859-6" "\0"
 259            "CP28597" "\0" "ISO-8859-7" "\0"
 260            "CP28598" "\0" "ISO-8859-8" "\0"
 261            "CP28599" "\0" "ISO-8859-9" "\0"
 262            "CP28605" "\0" "ISO-8859-15" "\0";
 263 # endif
 264 #endif
 265
 266       charset_aliases = cp;
 267     }
 268
 269   return cp;
 270 }
 271
 272 /* Determine the current locale's character encoding, and canonicalize it
 273    into one of the canonical names listed in config.charset.
 274    The result must not be freed; it is statically allocated.
 275    If the canonical name cannot be determined, the result is a non-canonical
 276    name.  */
 277
 278 #ifdef STATIC
 279 STATIC
 280 #endif
 281 const char *
 282 locale_charset ()
 283 {
 284   const char *codeset;
 285   const char *aliases;
 286
 287 #if !(defined WIN32 || defined OS2)
 288
 289 # if HAVE_LANGINFO_CODESET
 290
 291   /* Most systems support nl_langinfo (CODESET) nowadays.  */
 292   codeset = nl_langinfo (CODESET);
 293
 294 # else
 295
 296   /* On old systems which lack it, use setlocale or getenv.  */
 297   const char *locale = NULL;
 298
 299   /* But most old systems don't have a complete set of locales.  Some
 300      (like SunOS 4 or DJGPP) have only the C locale.  Therefore we don't
 301      use setlocale here; it would return "C" when it doesn't support the
 302      locale name the user has set.  */
 303 #  if HAVE_SETLOCALE && 0
 304   locale = setlocale (LC_CTYPE, NULL);
 305 #  endif
 306   if (locale == NULL || locale[0] == '\0')
 307     {
 308       locale = getenv ("LC_ALL");
 309       if (locale == NULL || locale[0] == '\0')
 310         {
 311           locale = getenv ("LC_CTYPE");
 312           if (locale == NULL || locale[0] == '\0')
 313             locale = getenv ("LANG");
 314         }
 315     }
 316
 317   /* On some old systems, one used to set locale = "iso8859_1". On others,
 318      you set it to "language_COUNTRY.charset". In any case, we resolve it
 319      through the charset.alias file.  */
 320   codeset = locale;
 321
 322 # endif
 323
 324 #elif defined WIN32
 325
 326   static char buf[2 + 10 + 1];
 327
 328   /* Woe32 has a function returning the locale's codepage as a number.  */
 329   sprintf (buf, "CP%u", GetACP ());
 330   codeset = buf;
 331
 332 #elif defined OS2
 333
 334   const char *locale;
 335   static char buf[2 + 10 + 1];
 336   ULONG cp[3];
 337   ULONG cplen;
 338
 339   /* Allow user to override the codeset, as set in the operating system,
 340      with standard language environment variables.  */
 341   locale = getenv ("LC_ALL");
 342   if (locale == NULL || locale[0] == '\0')
 343     {
 344       locale = getenv ("LC_CTYPE");
 345       if (locale == NULL || locale[0] == '\0')
 346         locale = getenv ("LANG");
 347     }
 348   if (locale != NULL && locale[0] != '\0')
 349     {
 350       /* If the locale name contains an encoding after the dot, return it.  */
 351       const char *dot = strchr (locale, '.');
 352
 353       if (dot != NULL)
 354         {
 355           const char *modifier;
 356
 357           dot++;
 358           /* Look for the possible @... trailer and remove it, if any.  */
 359           modifier = strchr (dot, '@');
 360           if (modifier == NULL)
 361             return dot;
 362           if (modifier - dot < sizeof (buf))
 363             {
 364               memcpy (buf, dot, modifier - dot);
 365               buf [modifier - dot] = '\0';
 366               return buf;
 367             }
 368         }
 369
 370       /* Resolve through the charset.alias file.  */
 371       codeset = locale;
 372     }
 373   else
 374     {
 375       /* OS/2 has a function returning the locale's codepage as a number.  */
 376       if (DosQueryCp (sizeof (cp), cp, &cplen))
 377         codeset = "";
 378       else
 379         {
 380           sprintf (buf, "CP%u", cp[0]);
 381           codeset = buf;
 382         }
 383     }
 384
 385 #endif
 386
 387   if (codeset == NULL)
 388     /* The canonical name cannot be determined.  */
 389     codeset = "";
 390
 391   /* Resolve alias. */
 392   for (aliases = get_charset_aliases ();
 393        *aliases != '\0';
 394        aliases += strlen (aliases) + 1, aliases += strlen (aliases) + 1)
 395     if (strcmp (codeset, aliases) == 0
 396         || (aliases[0] == '*' && aliases[1] == '\0'))
 397       {
 398         codeset = aliases + strlen (aliases) + 1;
 399         break;
 400       }
 401
 402   /* Don't return an empty string.  GNU libc and GNU libiconv interpret
 403      the empty string as denoting "the locale's character encoding",
 404      thus GNU libiconv would call this function a second time.  */
 405   if (codeset[0] == '\0')
 406     codeset = "ASCII";
 407
 408   return codeset;
 409 }