1 /* Determine a canonical name for the current locale's character encoding.
2 
3    Copyright (C) 2000-2001 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 <haible@clisp.cons.org>.  */
21 
22 #ifdef HAVE_CONFIG_H
23 # include <autoconf.h>
24 #endif
25 
26 #if HAVE_STDDEF_H
27 # include <stddef.h>
28 #endif
29 
30 #include <stdio.h>
31 #if HAVE_STRING_H
32 # include <string.h>
33 #else
34 # include <strings.h>
35 #endif
36 #if HAVE_STDLIB_H
37 # include <stdlib.h>
38 #endif
39 
40 #if defined _WIN32 || defined __WIN32__
41 # undef WIN32   /* avoid warning on mingw32 */
42 # define WIN32
43 #endif
44 
45 #ifndef WIN32
46 # if HAVE_LANGINFO_CODESET
47 #  include <langinfo.h>
48 # else
49 #  if HAVE_SETLOCALE
50 #   include <locale.h>
51 #  endif
52 # endif
53 #else /* WIN32 */
54 # define WIN32_LEAN_AND_MEAN
55 # include <windows.h>
56 #endif
57 
58 #ifndef DIRECTORY_SEPARATOR
59 # define DIRECTORY_SEPARATOR '/'
60 #endif
61 
62 #ifndef ISSLASH
63 # define ISSLASH(C) ((C) == DIRECTORY_SEPARATOR)
64 #endif
65 
66 /* The following static variable is declared 'volatile' to avoid a
67    possible multithread problem in the function get_charset_aliases. If we
68    are running in a threaded environment, and if two threads initialize
69    'charset_aliases' simultaneously, both will produce the same value,
70    and everything will be ok if the two assignments to 'charset_aliases'
71    are atomic. But I don't know what will happen if the two assignments mix.  */
72 #if __STDC__ != 1
73 # define volatile /* empty */
74 #endif
75 /* Pointer to the contents of the charset.alias file, if it has already been
76    read, else NULL.  Its format is:
77    ALIAS_1 '\0' CANONICAL_1 '\0' ... ALIAS_n '\0' CANONICAL_n '\0' '\0'  */
78 static const char * volatile charset_aliases;
79 
80 /* Return a pointer to the contents of the charset.alias file.  */
81 static const char *
get_charset_aliases(void)82 get_charset_aliases (void)
83 {
84   const char *cp;
85 
86   cp = charset_aliases;
87   if (cp == NULL)
88     {
89 #ifndef WIN32
90       FILE *fp;
91       const char *dir = LIBDIR;
92       const char *base = "charset.alias";
93       char *file_name;
94 
95       /* Concatenate dir and base into freshly allocated file_name.  */
96       {
97 	size_t dir_len = strlen (dir);
98 	size_t base_len = strlen (base);
99 	int add_slash = (dir_len > 0 && !ISSLASH (dir[dir_len - 1]));
100 	file_name = (char *) malloc (dir_len + add_slash + base_len + 1);
101 	if (file_name != NULL)
102 	  {
103 	    memcpy (file_name, dir, dir_len);
104 	    if (add_slash)
105 	      file_name[dir_len] = DIRECTORY_SEPARATOR;
106 	    memcpy (file_name + dir_len + add_slash, base, base_len + 1);
107 	  }
108       }
109 
110       if (file_name == NULL || (fp = fopen (file_name, "r")) == NULL)
111 	/* Out of memory or file not found, treat it as empty.  */
112 	cp = "";
113       else
114 	{
115 	  /* Parse the file's contents.  */
116 	  int c;
117 	  char buf1[50+1];
118 	  char buf2[50+1];
119 	  char *res_ptr = NULL;
120 	  size_t res_size = 0;
121 	  size_t l1, l2;
122 
123 	  for (;;)
124 	    {
125 	      c = getc (fp);
126 	      if (c == EOF)
127 		break;
128 	      if (c == '\n' || c == ' ' || c == '\t')
129 		continue;
130 	      if (c == '#')
131 		{
132 		  /* Skip comment, to end of line.  */
133 		  do
134 		    c = getc (fp);
135 		  while (!(c == EOF || c == '\n'));
136 		  if (c == EOF)
137 		    break;
138 		  continue;
139 		}
140 	      ungetc (c, fp);
141 	      if (fscanf(fp, "%50s %50s", buf1, buf2) < 2)
142 		break;
143 	      l1 = strlen (buf1);
144 	      l2 = strlen (buf2);
145 	      if (res_size == 0)
146 		{
147 		  res_size = l1 + 1 + l2 + 1;
148 		  res_ptr = malloc (res_size + 1);
149 		}
150 	      else
151 		{
152 		  res_size += l1 + 1 + l2 + 1;
153 		  res_ptr = realloc (res_ptr, res_size + 1);
154 		}
155 	      if (res_ptr == NULL)
156 		{
157 		  /* Out of memory. */
158 		  res_size = 0;
159 		  break;
160 		}
161 	      strcpy (res_ptr + res_size - (l2 + 1) - (l1 + 1), buf1);
162 	      strcpy (res_ptr + res_size - (l2 + 1), buf2);
163 	    }
164 	  fclose (fp);
165 	  if (res_size == 0)
166 	    cp = "";
167 	  else
168 	    {
169 	      *(res_ptr + res_size) = '\0';
170 	      cp = res_ptr;
171 	    }
172 	}
173 
174       if (file_name != NULL)
175 	free (file_name);
176 
177 #else /* WIN32 */
178 
179       /* To avoid the troubles of installing a separate file in the same
180 	 directory as the DLL and of retrieving the DLL's directory at
181 	 runtime, simply inline the aliases here.  */
182 
183       cp = "CP936" "\0" "GBK" "\0"
184 	   "CP1361" "\0" "JOHAB" "\0";
185 #endif
186 
187       charset_aliases = cp;
188     }
189 
190   return cp;
191 }
192 
193 /* Determine the current locale's character encoding, and canonicalize it
194    into one of the canonical names listed in config.charset.
195    The result must not be freed; it is statically allocated.
196    If the canonical name cannot be determined, the result is a non-canonical
197    name.  */
198 
199 #ifdef STATIC
200 STATIC
201 #endif
202 const char *
locale_charset(void)203 locale_charset (void)
204 {
205   const char *codeset;
206   const char *aliases;
207 
208 #ifndef WIN32
209 
210 # if HAVE_LANGINFO_CODESET
211 
212   /* Most systems support nl_langinfo (CODESET) nowadays.  */
213   codeset = nl_langinfo (CODESET);
214 
215 # else
216 
217   /* On old systems which lack it, use setlocale or getenv.  */
218   const char *locale = NULL;
219 
220   /* But most old systems don't have a complete set of locales.  Some
221      (like SunOS 4 or DJGPP) have only the C locale.  Therefore we don't
222      use setlocale here; it would return "C" when it doesn't support the
223      locale name the user has set.  */
224 #  if HAVE_SETLOCALE && 0
225   locale = setlocale (LC_CTYPE, NULL);
226 #  endif
227   if (locale == NULL || locale[0] == '\0')
228     {
229       locale = getenv ("LC_ALL");
230       if (locale == NULL || locale[0] == '\0')
231 	{
232 	  locale = getenv ("LC_CTYPE");
233 	  if (locale == NULL || locale[0] == '\0')
234 	    locale = getenv ("LANG");
235 	}
236     }
237 
238   /* On some old systems, one used to set locale = "iso8859_1". On others,
239      you set it to "language_COUNTRY.charset". In any case, we resolve it
240      through the charset.alias file.  */
241   codeset = locale;
242 
243 # endif
244 
245 #else /* WIN32 */
246 
247   static char buf[2 + 10 + 1];
248 
249   /* Win32 has a function returning the locale's codepage as a number.  */
250   sprintf (buf, "CP%u", GetACP ());
251   codeset = buf;
252 
253 #endif
254 
255   if (codeset == NULL)
256     /* The canonical name cannot be determined.  */
257     codeset = "";
258 
259   /* Resolve alias. */
260   for (aliases = get_charset_aliases ();
261        *aliases != '\0';
262        aliases += strlen (aliases) + 1, aliases += strlen (aliases) + 1)
263     if (strcmp (codeset, aliases) == 0
264 	|| (aliases[0] == '*' && aliases[1] == '\0'))
265       {
266 	codeset = aliases + strlen (aliases) + 1;
267 	break;
268       }
269 
270   return codeset;
271 }
272