Home | History | Annotate | Download | only in intl
      1 /* Determine a canonical name for the current locale's character encoding.
      2 
      3    Copyright (C) 2000-2003 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 (at) 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 __EMX__ || defined __DJGPP__
     77   /* Win32, 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 = relocate (LIBDIR);
    120       const char *base = "charset.alias";
    121       char *file_name;
    122 
    123       /* Concatenate dir and base into freshly allocated file_name.  */
    124       {
    125 	size_t dir_len = strlen (dir);
    126 	size_t base_len = strlen (base);
    127 	int add_slash = (dir_len > 0 && !ISSLASH (dir[dir_len - 1]));
    128 	file_name = (char *) malloc (dir_len + add_slash + base_len + 1);
    129 	if (file_name != NULL)
    130 	  {
    131 	    memcpy (file_name, dir, dir_len);
    132 	    if (add_slash)
    133 	      file_name[dir_len] = DIRECTORY_SEPARATOR;
    134 	    memcpy (file_name + dir_len + add_slash, base, base_len + 1);
    135 	  }
    136       }
    137 
    138       if (file_name == NULL || (fp = fopen (file_name, "r")) == NULL)
    139 	/* Out of memory or file not found, treat it as empty.  */
    140 	cp = "";
    141       else
    142 	{
    143 	  /* Parse the file's contents.  */
    144 	  int c;
    145 	  char buf1[50+1];
    146 	  char buf2[50+1];
    147 	  char *res_ptr = NULL;
    148 	  size_t res_size = 0;
    149 	  size_t l1, l2;
    150 
    151 	  for (;;)
    152 	    {
    153 	      c = getc (fp);
    154 	      if (c == EOF)
    155 		break;
    156 	      if (c == '\n' || c == ' ' || c == '\t')
    157 		continue;
    158 	      if (c == '#')
    159 		{
    160 		  /* Skip comment, to end of line.  */
    161 		  do
    162 		    c = getc (fp);
    163 		  while (!(c == EOF || c == '\n'));
    164 		  if (c == EOF)
    165 		    break;
    166 		  continue;
    167 		}
    168 	      ungetc (c, fp);
    169 	      if (fscanf (fp, "%50s %50s", buf1, buf2) < 2)
    170 		break;
    171 	      l1 = strlen (buf1);
    172 	      l2 = strlen (buf2);
    173 	      if (res_size == 0)
    174 		{
    175 		  res_size = l1 + 1 + l2 + 1;
    176 		  res_ptr = (char *) malloc (res_size + 1);
    177 		}
    178 	      else
    179 		{
    180 		  res_size += l1 + 1 + l2 + 1;
    181 		  res_ptr = (char *) realloc (res_ptr, res_size + 1);
    182 		}
    183 	      if (res_ptr == NULL)
    184 		{
    185 		  /* Out of memory. */
    186 		  res_size = 0;
    187 		  break;
    188 		}
    189 	      strcpy (res_ptr + res_size - (l2 + 1) - (l1 + 1), buf1);
    190 	      strcpy (res_ptr + res_size - (l2 + 1), buf2);
    191 	    }
    192 	  fclose (fp);
    193 	  if (res_size == 0)
    194 	    cp = "";
    195 	  else
    196 	    {
    197 	      *(res_ptr + res_size) = '\0';
    198 	      cp = res_ptr;
    199 	    }
    200 	}
    201 
    202       free (file_name);
    203 
    204 #else
    205 
    206 # if defined VMS
    207       /* To avoid the troubles of an extra file charset.alias_vms in the
    208 	 sources of many GNU packages, simply inline the aliases here.  */
    209       /* The list of encodings is taken from the OpenVMS 7.3-1 documentation
    210 	 "Compaq C Run-Time Library Reference Manual for OpenVMS systems"
    211 	 section 10.7 "Handling Different Character Sets".  */
    212       cp = "ISO8859-1" "\0" "ISO-8859-1" "\0"
    213 	   "ISO8859-2" "\0" "ISO-8859-2" "\0"
    214 	   "ISO8859-5" "\0" "ISO-8859-5" "\0"
    215 	   "ISO8859-7" "\0" "ISO-8859-7" "\0"
    216 	   "ISO8859-8" "\0" "ISO-8859-8" "\0"
    217 	   "ISO8859-9" "\0" "ISO-8859-9" "\0"
    218 	   /* Japanese */
    219 	   "eucJP" "\0" "EUC-JP" "\0"
    220 	   "SJIS" "\0" "SHIFT_JIS" "\0"
    221 	   "DECKANJI" "\0" "DEC-KANJI" "\0"
    222 	   "SDECKANJI" "\0" "EUC-JP" "\0"
    223 	   /* Chinese */
    224 	   "eucTW" "\0" "EUC-TW" "\0"
    225 	   "DECHANYU" "\0" "DEC-HANYU" "\0"
    226 	   "DECHANZI" "\0" "GB2312" "\0"
    227 	   /* Korean */
    228 	   "DECKOREAN" "\0" "EUC-KR" "\0";
    229 # endif
    230 
    231 # if defined WIN32
    232       /* To avoid the troubles of installing a separate file in the same
    233 	 directory as the DLL and of retrieving the DLL's directory at
    234 	 runtime, simply inline the aliases here.  */
    235 
    236       cp = "CP936" "\0" "GBK" "\0"
    237 	   "CP1361" "\0" "JOHAB" "\0"
    238 	   "CP20127" "\0" "ASCII" "\0"
    239 	   "CP20866" "\0" "KOI8-R" "\0"
    240 	   "CP21866" "\0" "KOI8-RU" "\0"
    241 	   "CP28591" "\0" "ISO-8859-1" "\0"
    242 	   "CP28592" "\0" "ISO-8859-2" "\0"
    243 	   "CP28593" "\0" "ISO-8859-3" "\0"
    244 	   "CP28594" "\0" "ISO-8859-4" "\0"
    245 	   "CP28595" "\0" "ISO-8859-5" "\0"
    246 	   "CP28596" "\0" "ISO-8859-6" "\0"
    247 	   "CP28597" "\0" "ISO-8859-7" "\0"
    248 	   "CP28598" "\0" "ISO-8859-8" "\0"
    249 	   "CP28599" "\0" "ISO-8859-9" "\0"
    250 	   "CP28605" "\0" "ISO-8859-15" "\0";
    251 # endif
    252 #endif
    253 
    254       charset_aliases = cp;
    255     }
    256 
    257   return cp;
    258 }
    259 
    260 /* Determine the current locale's character encoding, and canonicalize it
    261    into one of the canonical names listed in config.charset.
    262    The result must not be freed; it is statically allocated.
    263    If the canonical name cannot be determined, the result is a non-canonical
    264    name.  */
    265 
    266 #ifdef STATIC
    267 STATIC
    268 #endif
    269 const char *
    270 locale_charset ()
    271 {
    272   const char *codeset;
    273   const char *aliases;
    274 
    275 #if !(defined WIN32 || defined OS2)
    276 
    277 # if HAVE_LANGINFO_CODESET
    278 
    279   /* Most systems support nl_langinfo (CODESET) nowadays.  */
    280   codeset = nl_langinfo (CODESET);
    281 
    282 # else
    283 
    284   /* On old systems which lack it, use setlocale or getenv.  */
    285   const char *locale = NULL;
    286 
    287   /* But most old systems don't have a complete set of locales.  Some
    288      (like SunOS 4 or DJGPP) have only the C locale.  Therefore we don't
    289      use setlocale here; it would return "C" when it doesn't support the
    290      locale name the user has set.  */
    291 #  if HAVE_SETLOCALE && 0
    292   locale = setlocale (LC_CTYPE, NULL);
    293 #  endif
    294   if (locale == NULL || locale[0] == '\0')
    295     {
    296       locale = getenv ("LC_ALL");
    297       if (locale == NULL || locale[0] == '\0')
    298 	{
    299 	  locale = getenv ("LC_CTYPE");
    300 	  if (locale == NULL || locale[0] == '\0')
    301 	    locale = getenv ("LANG");
    302 	}
    303     }
    304 
    305   /* On some old systems, one used to set locale = "iso8859_1". On others,
    306      you set it to "language_COUNTRY.charset". In any case, we resolve it
    307      through the charset.alias file.  */
    308   codeset = locale;
    309 
    310 # endif
    311 
    312 #elif defined WIN32
    313 
    314   static char buf[2 + 10 + 1];
    315 
    316   /* Woe32 has a function returning the locale's codepage as a number.  */
    317   sprintf (buf, "CP%u", GetACP ());
    318   codeset = buf;
    319 
    320 #elif defined OS2
    321 
    322   const char *locale;
    323   static char buf[2 + 10 + 1];
    324   ULONG cp[3];
    325   ULONG cplen;
    326 
    327   /* Allow user to override the codeset, as set in the operating system,
    328      with standard language environment variables.  */
    329   locale = getenv ("LC_ALL");
    330   if (locale == NULL || locale[0] == '\0')
    331     {
    332       locale = getenv ("LC_CTYPE");
    333       if (locale == NULL || locale[0] == '\0')
    334 	locale = getenv ("LANG");
    335     }
    336   if (locale != NULL && locale[0] != '\0')
    337     {
    338       /* If the locale name contains an encoding after the dot, return it.  */
    339       const char *dot = strchr (locale, '.');
    340 
    341       if (dot != NULL)
    342 	{
    343 	  const char *modifier;
    344 
    345 	  dot++;
    346 	  /* Look for the possible @... trailer and remove it, if any.  */
    347 	  modifier = strchr (dot, '@');
    348 	  if (modifier == NULL)
    349 	    return dot;
    350 	  if (modifier - dot < sizeof (buf))
    351 	    {
    352 	      memcpy (buf, dot, modifier - dot);
    353 	      buf [modifier - dot] = '\0';
    354 	      return buf;
    355 	    }
    356 	}
    357 
    358       /* Resolve through the charset.alias file.  */
    359       codeset = locale;
    360     }
    361   else
    362     {
    363       /* OS/2 has a function returning the locale's codepage as a number.  */
    364       if (DosQueryCp (sizeof (cp), cp, &cplen))
    365 	codeset = "";
    366       else
    367 	{
    368 	  sprintf (buf, "CP%u", cp[0]);
    369 	  codeset = buf;
    370 	}
    371     }
    372 
    373 #endif
    374 
    375   if (codeset == NULL)
    376     /* The canonical name cannot be determined.  */
    377     codeset = "";
    378 
    379   /* Resolve alias. */
    380   for (aliases = get_charset_aliases ();
    381        *aliases != '\0';
    382        aliases += strlen (aliases) + 1, aliases += strlen (aliases) + 1)
    383     if (strcmp (codeset, aliases) == 0
    384 	|| (aliases[0] == '*' && aliases[1] == '\0'))
    385       {
    386 	codeset = aliases + strlen (aliases) + 1;
    387 	break;
    388       }
    389 
    390   /* Don't return an empty string.  GNU libc and GNU libiconv interpret
    391      the empty string as denoting "the locale's character encoding",
    392      thus GNU libiconv would call this function a second time.  */
    393   if (codeset[0] == '\0')
    394     codeset = "ASCII";
    395 
    396   return codeset;
    397 }
    398