Commit | Line | Data |
---|---|---|
7f8e40b7 NJ |
1 | /* Determine a canonical name for the current locale's character encoding. |
2 | ||
3 | Copyright (C) 2000-2006, 2008 Free Software Foundation, Inc. | |
4 | ||
5 | This program is free software; you can redistribute it and/or modify | |
6 | it under the terms of the GNU Lesser General Public License as published by | |
7 | 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 | |
13 | GNU Lesser General Public License for more details. | |
14 | ||
15 | You should have received a copy of the GNU Lesser General Public License along | |
16 | with this program; if not, write to the Free Software Foundation, | |
17 | Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ | |
18 | ||
19 | /* Written by Bruno Haible <bruno@clisp.org>. */ | |
20 | ||
21 | #include <config.h> | |
22 | ||
23 | /* Specification. */ | |
24 | #include "localcharset.h" | |
25 | ||
26 | #include <stddef.h> | |
27 | #include <stdio.h> | |
28 | #include <string.h> | |
29 | #include <stdlib.h> | |
30 | ||
31 | #if defined _WIN32 || defined __WIN32__ | |
32 | # define WIN32_NATIVE | |
33 | #endif | |
34 | ||
35 | #if defined __EMX__ | |
36 | /* Assume EMX program runs on OS/2, even if compiled under DOS. */ | |
37 | # ifndef OS2 | |
38 | # define OS2 | |
39 | # endif | |
40 | #endif | |
41 | ||
42 | #if !defined WIN32_NATIVE | |
43 | # if HAVE_LANGINFO_CODESET | |
44 | # include <langinfo.h> | |
45 | # else | |
46 | # if 0 /* see comment below */ | |
47 | # include <locale.h> | |
48 | # endif | |
49 | # endif | |
50 | # ifdef __CYGWIN__ | |
51 | # define WIN32_LEAN_AND_MEAN | |
52 | # include <windows.h> | |
53 | # endif | |
54 | #elif defined WIN32_NATIVE | |
55 | # define WIN32_LEAN_AND_MEAN | |
56 | # include <windows.h> | |
57 | #endif | |
58 | #if defined OS2 | |
59 | # define INCL_DOS | |
60 | # include <os2.h> | |
61 | #endif | |
62 | ||
63 | #if ENABLE_RELOCATABLE | |
64 | # include "relocatable.h" | |
65 | #else | |
66 | # define relocate(pathname) (pathname) | |
67 | #endif | |
68 | ||
69 | /* Get LIBDIR. */ | |
70 | #ifndef LIBDIR | |
71 | # include "configmake.h" | |
72 | #endif | |
73 | ||
74 | #if defined _WIN32 || defined __WIN32__ || defined __CYGWIN__ || defined __EMX__ || defined __DJGPP__ | |
75 | /* Win32, Cygwin, OS/2, DOS */ | |
76 | # define ISSLASH(C) ((C) == '/' || (C) == '\\') | |
77 | #endif | |
78 | ||
79 | #ifndef DIRECTORY_SEPARATOR | |
80 | # define DIRECTORY_SEPARATOR '/' | |
81 | #endif | |
82 | ||
83 | #ifndef ISSLASH | |
84 | # define ISSLASH(C) ((C) == DIRECTORY_SEPARATOR) | |
85 | #endif | |
86 | ||
87 | #if HAVE_DECL_GETC_UNLOCKED | |
88 | # undef getc | |
89 | # define getc getc_unlocked | |
90 | #endif | |
91 | ||
92 | /* The following static variable is declared 'volatile' to avoid a | |
93 | possible multithread problem in the function get_charset_aliases. If we | |
94 | are running in a threaded environment, and if two threads initialize | |
95 | 'charset_aliases' simultaneously, both will produce the same value, | |
96 | and everything will be ok if the two assignments to 'charset_aliases' | |
97 | are atomic. But I don't know what will happen if the two assignments mix. */ | |
98 | #if __STDC__ != 1 | |
99 | # define volatile /* empty */ | |
100 | #endif | |
101 | /* Pointer to the contents of the charset.alias file, if it has already been | |
102 | read, else NULL. Its format is: | |
103 | ALIAS_1 '\0' CANONICAL_1 '\0' ... ALIAS_n '\0' CANONICAL_n '\0' '\0' */ | |
104 | static const char * volatile charset_aliases; | |
105 | ||
106 | /* Return a pointer to the contents of the charset.alias file. */ | |
107 | static const char * | |
108 | get_charset_aliases (void) | |
109 | { | |
110 | const char *cp; | |
111 | ||
112 | cp = charset_aliases; | |
113 | if (cp == NULL) | |
114 | { | |
115 | #if !(defined VMS || defined WIN32_NATIVE || defined __CYGWIN__) | |
116 | FILE *fp; | |
117 | const char *dir; | |
118 | const char *base = "charset.alias"; | |
119 | char *file_name; | |
120 | ||
121 | /* Make it possible to override the charset.alias location. This is | |
122 | necessary for running the testsuite before "make install". */ | |
123 | dir = getenv ("CHARSETALIASDIR"); | |
124 | if (dir == NULL || dir[0] == '\0') | |
125 | dir = relocate (LIBDIR); | |
126 | ||
127 | /* Concatenate dir and base into freshly allocated file_name. */ | |
128 | { | |
129 | size_t dir_len = strlen (dir); | |
130 | size_t base_len = strlen (base); | |
131 | int add_slash = (dir_len > 0 && !ISSLASH (dir[dir_len - 1])); | |
132 | file_name = (char *) malloc (dir_len + add_slash + base_len + 1); | |
133 | if (file_name != NULL) | |
134 | { | |
135 | memcpy (file_name, dir, dir_len); | |
136 | if (add_slash) | |
137 | file_name[dir_len] = DIRECTORY_SEPARATOR; | |
138 | memcpy (file_name + dir_len + add_slash, base, base_len + 1); | |
139 | } | |
140 | } | |
141 | ||
142 | if (file_name == NULL || (fp = fopen (file_name, "r")) == NULL) | |
143 | /* Out of memory or file not found, treat it as empty. */ | |
144 | cp = ""; | |
145 | else | |
146 | { | |
147 | /* Parse the file's contents. */ | |
148 | char *res_ptr = NULL; | |
149 | size_t res_size = 0; | |
150 | ||
151 | for (;;) | |
152 | { | |
153 | int c; | |
154 | char buf1[50+1]; | |
155 | char buf2[50+1]; | |
156 | size_t l1, l2; | |
157 | char *old_res_ptr; | |
158 | ||
159 | c = getc (fp); | |
160 | if (c == EOF) | |
161 | break; | |
162 | if (c == '\n' || c == ' ' || c == '\t') | |
163 | continue; | |
164 | if (c == '#') | |
165 | { | |
166 | /* Skip comment, to end of line. */ | |
167 | do | |
168 | c = getc (fp); | |
169 | while (!(c == EOF || c == '\n')); | |
170 | if (c == EOF) | |
171 | break; | |
172 | continue; | |
173 | } | |
174 | ungetc (c, fp); | |
175 | if (fscanf (fp, "%50s %50s", buf1, buf2) < 2) | |
176 | break; | |
177 | l1 = strlen (buf1); | |
178 | l2 = strlen (buf2); | |
179 | old_res_ptr = res_ptr; | |
180 | if (res_size == 0) | |
181 | { | |
182 | res_size = l1 + 1 + l2 + 1; | |
183 | res_ptr = (char *) malloc (res_size + 1); | |
184 | } | |
185 | else | |
186 | { | |
187 | res_size += l1 + 1 + l2 + 1; | |
188 | res_ptr = (char *) realloc (res_ptr, res_size + 1); | |
189 | } | |
190 | if (res_ptr == NULL) | |
191 | { | |
192 | /* Out of memory. */ | |
193 | res_size = 0; | |
194 | if (old_res_ptr != NULL) | |
195 | free (old_res_ptr); | |
196 | break; | |
197 | } | |
198 | strcpy (res_ptr + res_size - (l2 + 1) - (l1 + 1), buf1); | |
199 | strcpy (res_ptr + res_size - (l2 + 1), buf2); | |
200 | } | |
201 | fclose (fp); | |
202 | if (res_size == 0) | |
203 | cp = ""; | |
204 | else | |
205 | { | |
206 | *(res_ptr + res_size) = '\0'; | |
207 | cp = res_ptr; | |
208 | } | |
209 | } | |
210 | ||
211 | if (file_name != NULL) | |
212 | free (file_name); | |
213 | ||
214 | #else | |
215 | ||
216 | # if defined VMS | |
217 | /* To avoid the troubles of an extra file charset.alias_vms in the | |
218 | sources of many GNU packages, simply inline the aliases here. */ | |
219 | /* The list of encodings is taken from the OpenVMS 7.3-1 documentation | |
220 | "Compaq C Run-Time Library Reference Manual for OpenVMS systems" | |
221 | section 10.7 "Handling Different Character Sets". */ | |
222 | cp = "ISO8859-1" "\0" "ISO-8859-1" "\0" | |
223 | "ISO8859-2" "\0" "ISO-8859-2" "\0" | |
224 | "ISO8859-5" "\0" "ISO-8859-5" "\0" | |
225 | "ISO8859-7" "\0" "ISO-8859-7" "\0" | |
226 | "ISO8859-8" "\0" "ISO-8859-8" "\0" | |
227 | "ISO8859-9" "\0" "ISO-8859-9" "\0" | |
228 | /* Japanese */ | |
229 | "eucJP" "\0" "EUC-JP" "\0" | |
230 | "SJIS" "\0" "SHIFT_JIS" "\0" | |
231 | "DECKANJI" "\0" "DEC-KANJI" "\0" | |
232 | "SDECKANJI" "\0" "EUC-JP" "\0" | |
233 | /* Chinese */ | |
234 | "eucTW" "\0" "EUC-TW" "\0" | |
235 | "DECHANYU" "\0" "DEC-HANYU" "\0" | |
236 | "DECHANZI" "\0" "GB2312" "\0" | |
237 | /* Korean */ | |
238 | "DECKOREAN" "\0" "EUC-KR" "\0"; | |
239 | # endif | |
240 | ||
241 | # if defined WIN32_NATIVE || defined __CYGWIN__ | |
242 | /* To avoid the troubles of installing a separate file in the same | |
243 | directory as the DLL and of retrieving the DLL's directory at | |
244 | runtime, simply inline the aliases here. */ | |
245 | ||
246 | cp = "CP936" "\0" "GBK" "\0" | |
247 | "CP1361" "\0" "JOHAB" "\0" | |
248 | "CP20127" "\0" "ASCII" "\0" | |
249 | "CP20866" "\0" "KOI8-R" "\0" | |
250 | "CP20936" "\0" "GB2312" "\0" | |
251 | "CP21866" "\0" "KOI8-RU" "\0" | |
252 | "CP28591" "\0" "ISO-8859-1" "\0" | |
253 | "CP28592" "\0" "ISO-8859-2" "\0" | |
254 | "CP28593" "\0" "ISO-8859-3" "\0" | |
255 | "CP28594" "\0" "ISO-8859-4" "\0" | |
256 | "CP28595" "\0" "ISO-8859-5" "\0" | |
257 | "CP28596" "\0" "ISO-8859-6" "\0" | |
258 | "CP28597" "\0" "ISO-8859-7" "\0" | |
259 | "CP28598" "\0" "ISO-8859-8" "\0" | |
260 | "CP28599" "\0" "ISO-8859-9" "\0" | |
261 | "CP28605" "\0" "ISO-8859-15" "\0" | |
262 | "CP38598" "\0" "ISO-8859-8" "\0" | |
263 | "CP51932" "\0" "EUC-JP" "\0" | |
264 | "CP51936" "\0" "GB2312" "\0" | |
265 | "CP51949" "\0" "EUC-KR" "\0" | |
266 | "CP51950" "\0" "EUC-TW" "\0" | |
267 | "CP54936" "\0" "GB18030" "\0" | |
268 | "CP65001" "\0" "UTF-8" "\0"; | |
269 | # endif | |
270 | #endif | |
271 | ||
272 | charset_aliases = cp; | |
273 | } | |
274 | ||
275 | return cp; | |
276 | } | |
277 | ||
278 | /* Determine the current locale's character encoding, and canonicalize it | |
279 | into one of the canonical names listed in config.charset. | |
280 | The result must not be freed; it is statically allocated. | |
281 | If the canonical name cannot be determined, the result is a non-canonical | |
282 | name. */ | |
283 | ||
284 | #ifdef STATIC | |
285 | STATIC | |
286 | #endif | |
287 | const char * | |
288 | locale_charset (void) | |
289 | { | |
290 | const char *codeset; | |
291 | const char *aliases; | |
292 | ||
293 | #if !(defined WIN32_NATIVE || defined OS2) | |
294 | ||
295 | # if HAVE_LANGINFO_CODESET | |
296 | ||
297 | /* Most systems support nl_langinfo (CODESET) nowadays. */ | |
298 | codeset = nl_langinfo (CODESET); | |
299 | ||
300 | # ifdef __CYGWIN__ | |
301 | /* Cygwin 2006 does not have locales. nl_langinfo (CODESET) always | |
302 | returns "US-ASCII". As long as this is not fixed, return the suffix | |
303 | of the locale name from the environment variables (if present) or | |
304 | the codepage as a number. */ | |
305 | if (codeset != NULL && strcmp (codeset, "US-ASCII") == 0) | |
306 | { | |
307 | const char *locale; | |
308 | static char buf[2 + 10 + 1]; | |
309 | ||
310 | locale = getenv ("LC_ALL"); | |
311 | if (locale == NULL || locale[0] == '\0') | |
312 | { | |
313 | locale = getenv ("LC_CTYPE"); | |
314 | if (locale == NULL || locale[0] == '\0') | |
315 | locale = getenv ("LANG"); | |
316 | } | |
317 | if (locale != NULL && locale[0] != '\0') | |
318 | { | |
319 | /* If the locale name contains an encoding after the dot, return | |
320 | it. */ | |
321 | const char *dot = strchr (locale, '.'); | |
322 | ||
323 | if (dot != NULL) | |
324 | { | |
325 | const char *modifier; | |
326 | ||
327 | dot++; | |
328 | /* Look for the possible @... trailer and remove it, if any. */ | |
329 | modifier = strchr (dot, '@'); | |
330 | if (modifier == NULL) | |
331 | return dot; | |
332 | if (modifier - dot < sizeof (buf)) | |
333 | { | |
334 | memcpy (buf, dot, modifier - dot); | |
335 | buf [modifier - dot] = '\0'; | |
336 | return buf; | |
337 | } | |
338 | } | |
339 | } | |
340 | ||
341 | /* Woe32 has a function returning the locale's codepage as a number. */ | |
342 | sprintf (buf, "CP%u", GetACP ()); | |
343 | codeset = buf; | |
344 | } | |
345 | # endif | |
346 | ||
347 | # else | |
348 | ||
349 | /* On old systems which lack it, use setlocale or getenv. */ | |
350 | const char *locale = NULL; | |
351 | ||
352 | /* But most old systems don't have a complete set of locales. Some | |
353 | (like SunOS 4 or DJGPP) have only the C locale. Therefore we don't | |
354 | use setlocale here; it would return "C" when it doesn't support the | |
355 | locale name the user has set. */ | |
356 | # if 0 | |
357 | locale = setlocale (LC_CTYPE, NULL); | |
358 | # endif | |
359 | if (locale == NULL || locale[0] == '\0') | |
360 | { | |
361 | locale = getenv ("LC_ALL"); | |
362 | if (locale == NULL || locale[0] == '\0') | |
363 | { | |
364 | locale = getenv ("LC_CTYPE"); | |
365 | if (locale == NULL || locale[0] == '\0') | |
366 | locale = getenv ("LANG"); | |
367 | } | |
368 | } | |
369 | ||
370 | /* On some old systems, one used to set locale = "iso8859_1". On others, | |
371 | you set it to "language_COUNTRY.charset". In any case, we resolve it | |
372 | through the charset.alias file. */ | |
373 | codeset = locale; | |
374 | ||
375 | # endif | |
376 | ||
377 | #elif defined WIN32_NATIVE | |
378 | ||
379 | static char buf[2 + 10 + 1]; | |
380 | ||
381 | /* Woe32 has a function returning the locale's codepage as a number. */ | |
382 | sprintf (buf, "CP%u", GetACP ()); | |
383 | codeset = buf; | |
384 | ||
385 | #elif defined OS2 | |
386 | ||
387 | const char *locale; | |
388 | static char buf[2 + 10 + 1]; | |
389 | ULONG cp[3]; | |
390 | ULONG cplen; | |
391 | ||
392 | /* Allow user to override the codeset, as set in the operating system, | |
393 | with standard language environment variables. */ | |
394 | locale = getenv ("LC_ALL"); | |
395 | if (locale == NULL || locale[0] == '\0') | |
396 | { | |
397 | locale = getenv ("LC_CTYPE"); | |
398 | if (locale == NULL || locale[0] == '\0') | |
399 | locale = getenv ("LANG"); | |
400 | } | |
401 | if (locale != NULL && locale[0] != '\0') | |
402 | { | |
403 | /* If the locale name contains an encoding after the dot, return it. */ | |
404 | const char *dot = strchr (locale, '.'); | |
405 | ||
406 | if (dot != NULL) | |
407 | { | |
408 | const char *modifier; | |
409 | ||
410 | dot++; | |
411 | /* Look for the possible @... trailer and remove it, if any. */ | |
412 | modifier = strchr (dot, '@'); | |
413 | if (modifier == NULL) | |
414 | return dot; | |
415 | if (modifier - dot < sizeof (buf)) | |
416 | { | |
417 | memcpy (buf, dot, modifier - dot); | |
418 | buf [modifier - dot] = '\0'; | |
419 | return buf; | |
420 | } | |
421 | } | |
422 | ||
423 | /* Resolve through the charset.alias file. */ | |
424 | codeset = locale; | |
425 | } | |
426 | else | |
427 | { | |
428 | /* OS/2 has a function returning the locale's codepage as a number. */ | |
429 | if (DosQueryCp (sizeof (cp), cp, &cplen)) | |
430 | codeset = ""; | |
431 | else | |
432 | { | |
433 | sprintf (buf, "CP%u", cp[0]); | |
434 | codeset = buf; | |
435 | } | |
436 | } | |
437 | ||
438 | #endif | |
439 | ||
440 | if (codeset == NULL) | |
441 | /* The canonical name cannot be determined. */ | |
442 | codeset = ""; | |
443 | ||
444 | /* Resolve alias. */ | |
445 | for (aliases = get_charset_aliases (); | |
446 | *aliases != '\0'; | |
447 | aliases += strlen (aliases) + 1, aliases += strlen (aliases) + 1) | |
448 | if (strcmp (codeset, aliases) == 0 | |
449 | || (aliases[0] == '*' && aliases[1] == '\0')) | |
450 | { | |
451 | codeset = aliases + strlen (aliases) + 1; | |
452 | break; | |
453 | } | |
454 | ||
455 | /* Don't return an empty string. GNU libc and GNU libiconv interpret | |
456 | the empty string as denoting "the locale's character encoding", | |
457 | thus GNU libiconv would call this function a second time. */ | |
458 | if (codeset[0] == '\0') | |
459 | codeset = "ASCII"; | |
460 | ||
461 | return codeset; | |
462 | } |