Commit | Line | Data |
---|---|---|
98994639 HS |
1 | /* locale.c |
2 | * | |
1129b882 NC |
3 | * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, |
4 | * 2002, 2003, 2005, 2006, 2007, 2008 by Larry Wall and others | |
98994639 HS |
5 | * |
6 | * You may distribute under the terms of either the GNU General Public | |
7 | * License or the Artistic License, as specified in the README file. | |
8 | * | |
9 | */ | |
10 | ||
11 | /* | |
4ac71550 | 12 | * A Elbereth Gilthoniel, |
cdad3b53 | 13 | * silivren penna míriel |
4ac71550 | 14 | * o menel aglar elenath! |
cdad3b53 | 15 | * Na-chaered palan-díriel |
4ac71550 TC |
16 | * o galadhremmin ennorath, |
17 | * Fanuilos, le linnathon | |
18 | * nef aear, si nef aearon! | |
19 | * | |
20 | * [p.238 of _The Lord of the Rings_, II/i: "Many Meetings"] | |
98994639 HS |
21 | */ |
22 | ||
166f8a29 DM |
23 | /* utility functions for handling locale-specific stuff like what |
24 | * character represents the decimal point. | |
0d071d52 KW |
25 | * |
26 | * All C programs have an underlying locale. Perl generally doesn't pay any | |
27 | * attention to it except within the scope of a 'use locale'. For most | |
28 | * categories, it accomplishes this by just using different operations if it is | |
29 | * in such scope than if not. However, various libc functions called by Perl | |
30 | * are affected by the LC_NUMERIC category, so there are macros in perl.h that | |
31 | * are used to toggle between the current locale and the C locale depending on | |
32 | * the desired behavior of those functions at the moment. | |
166f8a29 DM |
33 | */ |
34 | ||
98994639 HS |
35 | #include "EXTERN.h" |
36 | #define PERL_IN_LOCALE_C | |
37 | #include "perl.h" | |
38 | ||
b310b053 JH |
39 | #ifdef I_LANGINFO |
40 | # include <langinfo.h> | |
41 | #endif | |
42 | ||
a4af207c JH |
43 | #include "reentr.h" |
44 | ||
8ef6e574 KW |
45 | #ifdef USE_LOCALE |
46 | ||
98994639 | 47 | /* |
0d071d52 KW |
48 | * Standardize the locale name from a string returned by 'setlocale', possibly |
49 | * modifying that string. | |
98994639 | 50 | * |
0ef2a2b2 | 51 | * The typical return value of setlocale() is either |
98994639 HS |
52 | * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL |
53 | * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL | |
54 | * (the space-separated values represent the various sublocales, | |
0ef2a2b2 | 55 | * in some unspecified order). This is not handled by this function. |
98994639 HS |
56 | * |
57 | * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n", | |
0ef2a2b2 KW |
58 | * which is harmful for further use of the string in setlocale(). This |
59 | * function removes the trailing new line and everything up through the '=' | |
98994639 HS |
60 | * |
61 | */ | |
62 | STATIC char * | |
63 | S_stdize_locale(pTHX_ char *locs) | |
64 | { | |
7452cf6a | 65 | const char * const s = strchr(locs, '='); |
98994639 HS |
66 | bool okay = TRUE; |
67 | ||
7918f24d NC |
68 | PERL_ARGS_ASSERT_STDIZE_LOCALE; |
69 | ||
8772537c AL |
70 | if (s) { |
71 | const char * const t = strchr(s, '.'); | |
98994639 | 72 | okay = FALSE; |
8772537c AL |
73 | if (t) { |
74 | const char * const u = strchr(t, '\n'); | |
75 | if (u && (u[1] == 0)) { | |
76 | const STRLEN len = u - s; | |
77 | Move(s + 1, locs, len, char); | |
78 | locs[len] = 0; | |
79 | okay = TRUE; | |
98994639 HS |
80 | } |
81 | } | |
82 | } | |
83 | ||
84 | if (!okay) | |
85 | Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs); | |
86 | ||
87 | return locs; | |
88 | } | |
89 | ||
8ef6e574 KW |
90 | #endif |
91 | ||
98994639 HS |
92 | void |
93 | Perl_set_numeric_radix(pTHX) | |
94 | { | |
95 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 96 | dVAR; |
98994639 | 97 | # ifdef HAS_LOCALECONV |
7452cf6a | 98 | const struct lconv* const lc = localeconv(); |
98994639 | 99 | |
98994639 HS |
100 | if (lc && lc->decimal_point) { |
101 | if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) { | |
102 | SvREFCNT_dec(PL_numeric_radix_sv); | |
a0714e2c | 103 | PL_numeric_radix_sv = NULL; |
98994639 HS |
104 | } |
105 | else { | |
106 | if (PL_numeric_radix_sv) | |
107 | sv_setpv(PL_numeric_radix_sv, lc->decimal_point); | |
108 | else | |
109 | PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0); | |
28acfe03 KW |
110 | if (! is_ascii_string((U8 *) lc->decimal_point, 0) |
111 | && is_utf8_string((U8 *) lc->decimal_point, 0) | |
112 | && is_cur_LC_category_utf8(LC_NUMERIC)) | |
113 | { | |
114 | SvUTF8_on(PL_numeric_radix_sv); | |
115 | } | |
98994639 HS |
116 | } |
117 | } | |
118 | else | |
a0714e2c | 119 | PL_numeric_radix_sv = NULL; |
69014004 KW |
120 | |
121 | DEBUG_L(PerlIO_printf(Perl_debug_log, "Locale radix is %s\n", | |
122 | (PL_numeric_radix_sv) | |
123 | ? lc->decimal_point | |
124 | : "NULL")); | |
125 | ||
98994639 HS |
126 | # endif /* HAS_LOCALECONV */ |
127 | #endif /* USE_LOCALE_NUMERIC */ | |
128 | } | |
129 | ||
98994639 | 130 | void |
8772537c | 131 | Perl_new_numeric(pTHX_ const char *newnum) |
98994639 HS |
132 | { |
133 | #ifdef USE_LOCALE_NUMERIC | |
0d071d52 KW |
134 | |
135 | /* Called after all libc setlocale() calls affecting LC_NUMERIC, to tell | |
136 | * core Perl this and that 'newnum' is the name of the new locale. | |
137 | * It installs this locale as the current underlying default. | |
138 | * | |
139 | * The default locale and the C locale can be toggled between by use of the | |
140 | * set_numeric_local() and set_numeric_standard() functions, which should | |
141 | * probably not be called directly, but only via macros like | |
142 | * SET_NUMERIC_STANDARD() in perl.h. | |
143 | * | |
144 | * The toggling is necessary mainly so that a non-dot radix decimal point | |
145 | * character can be output, while allowing internal calculations to use a | |
146 | * dot. | |
147 | * | |
148 | * This sets several interpreter-level variables: | |
149 | * PL_numeric_name The default locale's name: a copy of 'newnum' | |
150 | * PL_numeric_local A boolean indicating if the toggled state is such | |
151 | * that the current locale is the default locale | |
152 | * PL_numeric_standard A boolean indicating if the toggled state is such | |
153 | * that the current locale is the C locale | |
154 | * Note that both of the last two variables can be true at the same time, | |
155 | * if the underlying locale is C. (Toggling is a no-op under these | |
156 | * circumstances.) | |
157 | * | |
158 | * Any code changing the locale (outside this file) should use | |
159 | * POSIX::setlocale, which calls this function. Therefore this function | |
160 | * should be called directly only from this file and from | |
161 | * POSIX::setlocale() */ | |
162 | ||
b03f34cf | 163 | char *save_newnum; |
97aff369 | 164 | dVAR; |
98994639 HS |
165 | |
166 | if (! newnum) { | |
43c5f42d NC |
167 | Safefree(PL_numeric_name); |
168 | PL_numeric_name = NULL; | |
98994639 HS |
169 | PL_numeric_standard = TRUE; |
170 | PL_numeric_local = TRUE; | |
171 | return; | |
172 | } | |
173 | ||
b03f34cf KW |
174 | save_newnum = stdize_locale(savepv(newnum)); |
175 | if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) { | |
98994639 | 176 | Safefree(PL_numeric_name); |
b03f34cf | 177 | PL_numeric_name = save_newnum; |
b03f34cf | 178 | } |
98994639 | 179 | |
e19f01cb KW |
180 | PL_numeric_standard = ((*save_newnum == 'C' && save_newnum[1] == '\0') |
181 | || strEQ(save_newnum, "POSIX")); | |
182 | PL_numeric_local = TRUE; | |
183 | set_numeric_radix(); | |
6959d69d | 184 | |
98994639 HS |
185 | #endif /* USE_LOCALE_NUMERIC */ |
186 | } | |
187 | ||
188 | void | |
189 | Perl_set_numeric_standard(pTHX) | |
190 | { | |
191 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 192 | dVAR; |
98994639 | 193 | |
0d071d52 KW |
194 | /* Toggle the LC_NUMERIC locale to C, if not already there. Probably |
195 | * should use the macros like SET_NUMERIC_STANDARD() in perl.h instead of | |
196 | * calling this directly. */ | |
197 | ||
98994639 HS |
198 | if (! PL_numeric_standard) { |
199 | setlocale(LC_NUMERIC, "C"); | |
200 | PL_numeric_standard = TRUE; | |
201 | PL_numeric_local = FALSE; | |
202 | set_numeric_radix(); | |
203 | } | |
69014004 KW |
204 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
205 | "Underlying LC_NUMERIC locale now is C\n")); | |
98994639 HS |
206 | |
207 | #endif /* USE_LOCALE_NUMERIC */ | |
208 | } | |
209 | ||
210 | void | |
211 | Perl_set_numeric_local(pTHX) | |
212 | { | |
213 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 214 | dVAR; |
98994639 | 215 | |
0d071d52 KW |
216 | /* Toggle the LC_NUMERIC locale to the current underlying default, if not |
217 | * already there. Probably should use the macros like SET_NUMERIC_LOCAL() | |
218 | * in perl.h instead of calling this directly. */ | |
219 | ||
98994639 HS |
220 | if (! PL_numeric_local) { |
221 | setlocale(LC_NUMERIC, PL_numeric_name); | |
222 | PL_numeric_standard = FALSE; | |
223 | PL_numeric_local = TRUE; | |
224 | set_numeric_radix(); | |
225 | } | |
69014004 KW |
226 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
227 | "Underlying LC_NUMERIC locale now is %s\n", | |
228 | PL_numeric_name)); | |
98994639 HS |
229 | |
230 | #endif /* USE_LOCALE_NUMERIC */ | |
231 | } | |
232 | ||
233 | /* | |
234 | * Set up for a new ctype locale. | |
235 | */ | |
236 | void | |
8772537c | 237 | Perl_new_ctype(pTHX_ const char *newctype) |
98994639 HS |
238 | { |
239 | #ifdef USE_LOCALE_CTYPE | |
0d071d52 KW |
240 | |
241 | /* Called after all libc setlocale() calls affecting LC_CTYPE, to tell | |
242 | * core Perl this and that 'newctype' is the name of the new locale. | |
243 | * | |
244 | * This function sets up the folding arrays for all 256 bytes, assuming | |
245 | * that tofold() is tolc() since fold case is not a concept in POSIX, | |
246 | * | |
247 | * Any code changing the locale (outside this file) should use | |
248 | * POSIX::setlocale, which calls this function. Therefore this function | |
249 | * should be called directly only from this file and from | |
250 | * POSIX::setlocale() */ | |
251 | ||
27da23d5 | 252 | dVAR; |
68067e4e | 253 | UV i; |
98994639 | 254 | |
7918f24d NC |
255 | PERL_ARGS_ASSERT_NEW_CTYPE; |
256 | ||
31f05a37 KW |
257 | PL_in_utf8_CTYPE_locale = is_cur_LC_category_utf8(LC_CTYPE); |
258 | ||
259 | /* A UTF-8 locale gets standard rules. But note that code still has to | |
260 | * handle this specially because of the three problematic code points */ | |
261 | if (PL_in_utf8_CTYPE_locale) { | |
262 | Copy(PL_fold_latin1, PL_fold_locale, 256, U8); | |
263 | } | |
264 | else { | |
baa60164 KW |
265 | for (i = 0; i < 256; i++) { |
266 | if (isUPPER_LC((U8) i)) | |
267 | PL_fold_locale[i] = (U8) toLOWER_LC((U8) i); | |
268 | else if (isLOWER_LC((U8) i)) | |
269 | PL_fold_locale[i] = (U8) toUPPER_LC((U8) i); | |
270 | else | |
271 | PL_fold_locale[i] = (U8) i; | |
272 | } | |
31f05a37 | 273 | } |
98994639 HS |
274 | |
275 | #endif /* USE_LOCALE_CTYPE */ | |
7918f24d | 276 | PERL_ARGS_ASSERT_NEW_CTYPE; |
8772537c | 277 | PERL_UNUSED_ARG(newctype); |
96a5add6 | 278 | PERL_UNUSED_CONTEXT; |
98994639 HS |
279 | } |
280 | ||
98994639 | 281 | void |
8772537c | 282 | Perl_new_collate(pTHX_ const char *newcoll) |
98994639 HS |
283 | { |
284 | #ifdef USE_LOCALE_COLLATE | |
0d071d52 KW |
285 | |
286 | /* Called after all libc setlocale() calls affecting LC_COLLATE, to tell | |
287 | * core Perl this and that 'newcoll' is the name of the new locale. | |
288 | * | |
289 | * Any code changing the locale (outside this file) should use | |
290 | * POSIX::setlocale, which calls this function. Therefore this function | |
291 | * should be called directly only from this file and from | |
292 | * POSIX::setlocale() */ | |
293 | ||
97aff369 | 294 | dVAR; |
98994639 HS |
295 | |
296 | if (! newcoll) { | |
297 | if (PL_collation_name) { | |
298 | ++PL_collation_ix; | |
299 | Safefree(PL_collation_name); | |
300 | PL_collation_name = NULL; | |
301 | } | |
302 | PL_collation_standard = TRUE; | |
303 | PL_collxfrm_base = 0; | |
304 | PL_collxfrm_mult = 2; | |
305 | return; | |
306 | } | |
307 | ||
308 | if (! PL_collation_name || strNE(PL_collation_name, newcoll)) { | |
309 | ++PL_collation_ix; | |
310 | Safefree(PL_collation_name); | |
311 | PL_collation_name = stdize_locale(savepv(newcoll)); | |
770526c1 NC |
312 | PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0') |
313 | || strEQ(newcoll, "POSIX")); | |
98994639 HS |
314 | |
315 | { | |
316 | /* 2: at most so many chars ('a', 'b'). */ | |
317 | /* 50: surely no system expands a char more. */ | |
318 | #define XFRMBUFSIZE (2 * 50) | |
319 | char xbuf[XFRMBUFSIZE]; | |
8772537c AL |
320 | const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE); |
321 | const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE); | |
322 | const SSize_t mult = fb - fa; | |
e0601d3c | 323 | if (mult < 1 && !(fa == 0 && fb == 0)) |
ad49ad39 NC |
324 | Perl_croak(aTHX_ "panic: strxfrm() gets absurd - a => %"UVuf", ab => %"UVuf, |
325 | (UV) fa, (UV) fb); | |
eb160463 | 326 | PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0; |
98994639 HS |
327 | PL_collxfrm_mult = mult; |
328 | } | |
329 | } | |
330 | ||
331 | #endif /* USE_LOCALE_COLLATE */ | |
332 | } | |
333 | ||
b385bb4d KW |
334 | #ifdef WIN32 |
335 | ||
336 | char * | |
337 | Perl_my_setlocale(pTHX_ int category, const char* locale) | |
338 | { | |
339 | /* This, for Windows, emulates POSIX setlocale() behavior. There is no | |
340 | * difference unless the input locale is "", which means on Windows to get | |
341 | * the machine default, which is set via the computer's "Regional and | |
342 | * Language Options" (or its current equivalent). In POSIX, it instead | |
343 | * means to find the locale from the user's environment. This routine | |
344 | * looks in the environment, and, if anything is found, uses that instead | |
345 | * of going to the machine default. If there is no environment override, | |
346 | * the machine default is used, as normal, by calling the real setlocale() | |
347 | * with "". The POSIX behavior is to use the LC_ALL variable if set; | |
348 | * otherwise to use the particular category's variable if set; otherwise to | |
349 | * use the LANG variable. */ | |
350 | ||
351 | if (locale && strEQ(locale, "")) { | |
352 | # ifdef LC_ALL | |
353 | locale = PerlEnv_getenv("LC_ALL"); | |
354 | if (! locale) { | |
355 | #endif | |
356 | switch (category) { | |
357 | # ifdef LC_ALL | |
358 | case LC_ALL: | |
359 | break; /* We already know its variable isn't set */ | |
360 | # endif | |
361 | # ifdef USE_LOCALE_TIME | |
362 | case LC_TIME: | |
363 | locale = PerlEnv_getenv("LC_TIME"); | |
364 | break; | |
365 | # endif | |
366 | # ifdef USE_LOCALE_CTYPE | |
367 | case LC_CTYPE: | |
368 | locale = PerlEnv_getenv("LC_CTYPE"); | |
369 | break; | |
370 | # endif | |
371 | # ifdef USE_LOCALE_COLLATE | |
372 | case LC_COLLATE: | |
373 | locale = PerlEnv_getenv("LC_COLLATE"); | |
374 | break; | |
375 | # endif | |
376 | # ifdef USE_LOCALE_MONETARY | |
377 | case LC_MONETARY: | |
378 | locale = PerlEnv_getenv("LC_MONETARY"); | |
379 | break; | |
380 | # endif | |
381 | # ifdef USE_LOCALE_NUMERIC | |
382 | case LC_NUMERIC: | |
383 | locale = PerlEnv_getenv("LC_NUMERIC"); | |
384 | break; | |
385 | # endif | |
386 | # ifdef USE_LOCALE_MESSAGES | |
387 | case LC_MESSAGES: | |
388 | locale = PerlEnv_getenv("LC_MESSAGES"); | |
389 | break; | |
390 | # endif | |
391 | default: | |
392 | /* This is a category, like PAPER_SIZE that we don't | |
393 | * know about; and so can't provide a wrapper. */ | |
394 | break; | |
395 | } | |
396 | if (! locale) { | |
397 | locale = PerlEnv_getenv("LANG"); | |
398 | if (! locale) { | |
399 | locale = ""; | |
400 | } | |
401 | } | |
402 | # ifdef LC_ALL | |
403 | } | |
404 | # endif | |
405 | } | |
406 | ||
407 | return setlocale(category, locale); | |
408 | } | |
409 | ||
410 | #endif | |
411 | ||
412 | ||
98994639 HS |
413 | /* |
414 | * Initialize locale awareness. | |
415 | */ | |
416 | int | |
417 | Perl_init_i18nl10n(pTHX_ int printwarn) | |
418 | { | |
0e92a118 KW |
419 | /* printwarn is |
420 | * | |
421 | * 0 if not to output warning when setup locale is bad | |
422 | * 1 if to output warning based on value of PERL_BADLANG | |
423 | * >1 if to output regardless of PERL_BADLANG | |
424 | * | |
425 | * returns | |
98994639 | 426 | * 1 = set ok or not applicable, |
0e92a118 KW |
427 | * 0 = fallback to a locale of lower priority |
428 | * -1 = fallback to all locales failed, not even to the C locale | |
98994639 HS |
429 | */ |
430 | ||
0e92a118 KW |
431 | int ok = 1; |
432 | ||
98994639 | 433 | #if defined(USE_LOCALE) |
97aff369 | 434 | dVAR; |
98994639 HS |
435 | |
436 | #ifdef USE_LOCALE_CTYPE | |
437 | char *curctype = NULL; | |
438 | #endif /* USE_LOCALE_CTYPE */ | |
439 | #ifdef USE_LOCALE_COLLATE | |
440 | char *curcoll = NULL; | |
441 | #endif /* USE_LOCALE_COLLATE */ | |
442 | #ifdef USE_LOCALE_NUMERIC | |
443 | char *curnum = NULL; | |
444 | #endif /* USE_LOCALE_NUMERIC */ | |
445 | #ifdef __GLIBC__ | |
7452cf6a | 446 | char * const language = PerlEnv_getenv("LANGUAGE"); |
98994639 | 447 | #endif |
ccd65d51 KW |
448 | /* NULL uses the existing already set up locale */ |
449 | const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT")) | |
450 | ? NULL | |
451 | : ""; | |
7452cf6a AL |
452 | char * const lc_all = PerlEnv_getenv("LC_ALL"); |
453 | char * const lang = PerlEnv_getenv("LANG"); | |
98994639 | 454 | bool setlocale_failure = FALSE; |
0e92a118 KW |
455 | bool done = FALSE; |
456 | ||
98994639 | 457 | |
0e92a118 KW |
458 | #ifndef LOCALE_ENVIRON_REQUIRED |
459 | PERL_UNUSED_VAR(done); | |
460 | #else | |
98994639 HS |
461 | |
462 | /* | |
463 | * Ultrix setlocale(..., "") fails if there are no environment | |
464 | * variables from which to get a locale name. | |
465 | */ | |
466 | ||
b3e384bf | 467 | # ifdef LC_ALL |
98994639 | 468 | if (lang) { |
b385bb4d | 469 | if (my_setlocale(LC_ALL, setlocale_init)) |
98994639 HS |
470 | done = TRUE; |
471 | else | |
472 | setlocale_failure = TRUE; | |
473 | } | |
474 | if (!setlocale_failure) { | |
b3e384bf | 475 | # ifdef USE_LOCALE_CTYPE |
56279a21 | 476 | Safefree(curctype); |
98994639 | 477 | if (! (curctype = |
b385bb4d | 478 | my_setlocale(LC_CTYPE, |
98994639 | 479 | (!done && (lang || PerlEnv_getenv("LC_CTYPE"))) |
ccd65d51 | 480 | ? setlocale_init : NULL))) |
98994639 HS |
481 | setlocale_failure = TRUE; |
482 | else | |
483 | curctype = savepv(curctype); | |
b3e384bf KW |
484 | # endif /* USE_LOCALE_CTYPE */ |
485 | # ifdef USE_LOCALE_COLLATE | |
56279a21 | 486 | Safefree(curcoll); |
98994639 | 487 | if (! (curcoll = |
b385bb4d | 488 | my_setlocale(LC_COLLATE, |
98994639 | 489 | (!done && (lang || PerlEnv_getenv("LC_COLLATE"))) |
ccd65d51 | 490 | ? setlocale_init : NULL))) |
98994639 HS |
491 | setlocale_failure = TRUE; |
492 | else | |
493 | curcoll = savepv(curcoll); | |
b3e384bf KW |
494 | # endif /* USE_LOCALE_COLLATE */ |
495 | # ifdef USE_LOCALE_NUMERIC | |
56279a21 | 496 | Safefree(curnum); |
98994639 | 497 | if (! (curnum = |
b385bb4d | 498 | my_setlocale(LC_NUMERIC, |
98994639 | 499 | (!done && (lang || PerlEnv_getenv("LC_NUMERIC"))) |
ccd65d51 | 500 | ? setlocale_init : NULL))) |
98994639 HS |
501 | setlocale_failure = TRUE; |
502 | else | |
503 | curnum = savepv(curnum); | |
b3e384bf | 504 | # endif /* USE_LOCALE_NUMERIC */ |
98994639 HS |
505 | } |
506 | ||
b3e384bf | 507 | # endif /* LC_ALL */ |
98994639 HS |
508 | |
509 | #endif /* !LOCALE_ENVIRON_REQUIRED */ | |
510 | ||
511 | #ifdef LC_ALL | |
b385bb4d | 512 | if (! my_setlocale(LC_ALL, setlocale_init)) |
98994639 HS |
513 | setlocale_failure = TRUE; |
514 | #endif /* LC_ALL */ | |
515 | ||
516 | if (!setlocale_failure) { | |
517 | #ifdef USE_LOCALE_CTYPE | |
6cb43dbf | 518 | Safefree(curctype); |
b385bb4d | 519 | if (! (curctype = my_setlocale(LC_CTYPE, setlocale_init))) |
98994639 HS |
520 | setlocale_failure = TRUE; |
521 | else | |
522 | curctype = savepv(curctype); | |
523 | #endif /* USE_LOCALE_CTYPE */ | |
524 | #ifdef USE_LOCALE_COLLATE | |
6cb43dbf | 525 | Safefree(curcoll); |
b385bb4d | 526 | if (! (curcoll = my_setlocale(LC_COLLATE, setlocale_init))) |
98994639 HS |
527 | setlocale_failure = TRUE; |
528 | else | |
529 | curcoll = savepv(curcoll); | |
530 | #endif /* USE_LOCALE_COLLATE */ | |
531 | #ifdef USE_LOCALE_NUMERIC | |
6cb43dbf | 532 | Safefree(curnum); |
b385bb4d | 533 | if (! (curnum = my_setlocale(LC_NUMERIC, setlocale_init))) |
98994639 HS |
534 | setlocale_failure = TRUE; |
535 | else | |
536 | curnum = savepv(curnum); | |
537 | #endif /* USE_LOCALE_NUMERIC */ | |
538 | } | |
539 | ||
540 | if (setlocale_failure) { | |
541 | char *p; | |
0bd48802 | 542 | const bool locwarn = (printwarn > 1 || |
98994639 HS |
543 | (printwarn && |
544 | (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p)))); | |
545 | ||
546 | if (locwarn) { | |
547 | #ifdef LC_ALL | |
548 | ||
549 | PerlIO_printf(Perl_error_log, | |
550 | "perl: warning: Setting locale failed.\n"); | |
551 | ||
552 | #else /* !LC_ALL */ | |
553 | ||
554 | PerlIO_printf(Perl_error_log, | |
555 | "perl: warning: Setting locale failed for the categories:\n\t"); | |
556 | #ifdef USE_LOCALE_CTYPE | |
557 | if (! curctype) | |
558 | PerlIO_printf(Perl_error_log, "LC_CTYPE "); | |
559 | #endif /* USE_LOCALE_CTYPE */ | |
560 | #ifdef USE_LOCALE_COLLATE | |
561 | if (! curcoll) | |
562 | PerlIO_printf(Perl_error_log, "LC_COLLATE "); | |
563 | #endif /* USE_LOCALE_COLLATE */ | |
564 | #ifdef USE_LOCALE_NUMERIC | |
565 | if (! curnum) | |
566 | PerlIO_printf(Perl_error_log, "LC_NUMERIC "); | |
567 | #endif /* USE_LOCALE_NUMERIC */ | |
568 | PerlIO_printf(Perl_error_log, "\n"); | |
569 | ||
570 | #endif /* LC_ALL */ | |
571 | ||
572 | PerlIO_printf(Perl_error_log, | |
573 | "perl: warning: Please check that your locale settings:\n"); | |
574 | ||
575 | #ifdef __GLIBC__ | |
576 | PerlIO_printf(Perl_error_log, | |
577 | "\tLANGUAGE = %c%s%c,\n", | |
578 | language ? '"' : '(', | |
579 | language ? language : "unset", | |
580 | language ? '"' : ')'); | |
581 | #endif | |
582 | ||
583 | PerlIO_printf(Perl_error_log, | |
584 | "\tLC_ALL = %c%s%c,\n", | |
585 | lc_all ? '"' : '(', | |
586 | lc_all ? lc_all : "unset", | |
587 | lc_all ? '"' : ')'); | |
588 | ||
589 | #if defined(USE_ENVIRON_ARRAY) | |
590 | { | |
591 | char **e; | |
592 | for (e = environ; *e; e++) { | |
593 | if (strnEQ(*e, "LC_", 3) | |
594 | && strnNE(*e, "LC_ALL=", 7) | |
595 | && (p = strchr(*e, '='))) | |
596 | PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n", | |
597 | (int)(p - *e), *e, p + 1); | |
598 | } | |
599 | } | |
600 | #else | |
601 | PerlIO_printf(Perl_error_log, | |
602 | "\t(possibly more locale environment variables)\n"); | |
603 | #endif | |
604 | ||
605 | PerlIO_printf(Perl_error_log, | |
606 | "\tLANG = %c%s%c\n", | |
607 | lang ? '"' : '(', | |
608 | lang ? lang : "unset", | |
609 | lang ? '"' : ')'); | |
610 | ||
611 | PerlIO_printf(Perl_error_log, | |
612 | " are supported and installed on your system.\n"); | |
613 | } | |
614 | ||
615 | #ifdef LC_ALL | |
616 | ||
617 | if (setlocale(LC_ALL, "C")) { | |
618 | if (locwarn) | |
619 | PerlIO_printf(Perl_error_log, | |
620 | "perl: warning: Falling back to the standard locale (\"C\").\n"); | |
621 | ok = 0; | |
622 | } | |
623 | else { | |
624 | if (locwarn) | |
625 | PerlIO_printf(Perl_error_log, | |
626 | "perl: warning: Failed to fall back to the standard locale (\"C\").\n"); | |
627 | ok = -1; | |
628 | } | |
629 | ||
630 | #else /* ! LC_ALL */ | |
631 | ||
632 | if (0 | |
633 | #ifdef USE_LOCALE_CTYPE | |
634 | || !(curctype || setlocale(LC_CTYPE, "C")) | |
635 | #endif /* USE_LOCALE_CTYPE */ | |
636 | #ifdef USE_LOCALE_COLLATE | |
637 | || !(curcoll || setlocale(LC_COLLATE, "C")) | |
638 | #endif /* USE_LOCALE_COLLATE */ | |
639 | #ifdef USE_LOCALE_NUMERIC | |
640 | || !(curnum || setlocale(LC_NUMERIC, "C")) | |
641 | #endif /* USE_LOCALE_NUMERIC */ | |
642 | ) | |
643 | { | |
644 | if (locwarn) | |
645 | PerlIO_printf(Perl_error_log, | |
646 | "perl: warning: Cannot fall back to the standard locale (\"C\").\n"); | |
647 | ok = -1; | |
648 | } | |
649 | ||
650 | #endif /* ! LC_ALL */ | |
651 | ||
652 | #ifdef USE_LOCALE_CTYPE | |
56279a21 | 653 | Safefree(curctype); |
bd61b366 | 654 | curctype = savepv(setlocale(LC_CTYPE, NULL)); |
98994639 HS |
655 | #endif /* USE_LOCALE_CTYPE */ |
656 | #ifdef USE_LOCALE_COLLATE | |
56279a21 | 657 | Safefree(curcoll); |
bd61b366 | 658 | curcoll = savepv(setlocale(LC_COLLATE, NULL)); |
98994639 HS |
659 | #endif /* USE_LOCALE_COLLATE */ |
660 | #ifdef USE_LOCALE_NUMERIC | |
56279a21 | 661 | Safefree(curnum); |
bd61b366 | 662 | curnum = savepv(setlocale(LC_NUMERIC, NULL)); |
98994639 HS |
663 | #endif /* USE_LOCALE_NUMERIC */ |
664 | } | |
665 | else { | |
666 | ||
667 | #ifdef USE_LOCALE_CTYPE | |
668 | new_ctype(curctype); | |
669 | #endif /* USE_LOCALE_CTYPE */ | |
670 | ||
671 | #ifdef USE_LOCALE_COLLATE | |
672 | new_collate(curcoll); | |
673 | #endif /* USE_LOCALE_COLLATE */ | |
674 | ||
675 | #ifdef USE_LOCALE_NUMERIC | |
676 | new_numeric(curnum); | |
677 | #endif /* USE_LOCALE_NUMERIC */ | |
b310b053 | 678 | |
98994639 HS |
679 | } |
680 | ||
8ef6e574 | 681 | #if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE) |
b310b053 | 682 | { |
fde18df1 | 683 | /* Set PL_utf8locale to TRUE if using PerlIO _and_ |
7d74bb61 | 684 | the current LC_CTYPE locale is UTF-8. |
8aa8f774 | 685 | If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE}) |
a05d7ebb | 686 | are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer |
fde18df1 | 687 | on STDIN, STDOUT, STDERR, _and_ the default open discipline. |
085a54d9 | 688 | */ |
7d74bb61 | 689 | PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE); |
fde18df1 | 690 | } |
a05d7ebb | 691 | /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO. |
fde18df1 JH |
692 | This is an alternative to using the -C command line switch |
693 | (the -C if present will override this). */ | |
694 | { | |
dd374669 | 695 | const char *p = PerlEnv_getenv("PERL_UNICODE"); |
a05d7ebb | 696 | PL_unicode = p ? parse_unicode_opts(&p) : 0; |
5a22a2bb NC |
697 | if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG) |
698 | PL_utf8cache = -1; | |
b310b053 | 699 | } |
ec71e770 | 700 | #endif |
b310b053 | 701 | |
98994639 | 702 | #ifdef USE_LOCALE_CTYPE |
43c5f42d | 703 | Safefree(curctype); |
98994639 HS |
704 | #endif /* USE_LOCALE_CTYPE */ |
705 | #ifdef USE_LOCALE_COLLATE | |
43c5f42d | 706 | Safefree(curcoll); |
98994639 HS |
707 | #endif /* USE_LOCALE_COLLATE */ |
708 | #ifdef USE_LOCALE_NUMERIC | |
43c5f42d | 709 | Safefree(curnum); |
98994639 | 710 | #endif /* USE_LOCALE_NUMERIC */ |
8ef6e574 KW |
711 | |
712 | #endif /* USE_LOCALE */ | |
713 | ||
98994639 HS |
714 | return ok; |
715 | } | |
716 | ||
98994639 HS |
717 | #ifdef USE_LOCALE_COLLATE |
718 | ||
719 | /* | |
720 | * mem_collxfrm() is a bit like strxfrm() but with two important | |
721 | * differences. First, it handles embedded NULs. Second, it allocates | |
722 | * a bit more memory than needed for the transformed data itself. | |
723 | * The real transformed data begins at offset sizeof(collationix). | |
724 | * Please see sv_collxfrm() to see how this is used. | |
725 | */ | |
166f8a29 | 726 | |
98994639 HS |
727 | char * |
728 | Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen) | |
729 | { | |
97aff369 | 730 | dVAR; |
98994639 HS |
731 | char *xbuf; |
732 | STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */ | |
733 | ||
7918f24d NC |
734 | PERL_ARGS_ASSERT_MEM_COLLXFRM; |
735 | ||
98994639 HS |
736 | /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */ |
737 | /* the +1 is for the terminating NUL. */ | |
738 | ||
739 | xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1; | |
a02a5408 | 740 | Newx(xbuf, xAlloc, char); |
98994639 HS |
741 | if (! xbuf) |
742 | goto bad; | |
743 | ||
744 | *(U32*)xbuf = PL_collation_ix; | |
745 | xout = sizeof(PL_collation_ix); | |
746 | for (xin = 0; xin < len; ) { | |
224e8ef5 | 747 | Size_t xused; |
98994639 HS |
748 | |
749 | for (;;) { | |
750 | xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout); | |
224e8ef5 | 751 | if (xused >= PERL_INT_MAX) |
98994639 | 752 | goto bad; |
eb160463 | 753 | if ((STRLEN)xused < xAlloc - xout) |
98994639 HS |
754 | break; |
755 | xAlloc = (2 * xAlloc) + 1; | |
756 | Renew(xbuf, xAlloc, char); | |
757 | if (! xbuf) | |
758 | goto bad; | |
759 | } | |
760 | ||
761 | xin += strlen(s + xin) + 1; | |
762 | xout += xused; | |
763 | ||
764 | /* Embedded NULs are understood but silently skipped | |
765 | * because they make no sense in locale collation. */ | |
766 | } | |
767 | ||
768 | xbuf[xout] = '\0'; | |
769 | *xlen = xout - sizeof(PL_collation_ix); | |
770 | return xbuf; | |
771 | ||
772 | bad: | |
773 | Safefree(xbuf); | |
774 | *xlen = 0; | |
775 | return NULL; | |
776 | } | |
777 | ||
778 | #endif /* USE_LOCALE_COLLATE */ | |
779 | ||
8ef6e574 KW |
780 | #ifdef USE_LOCALE |
781 | ||
637917bb | 782 | STATIC bool |
7d74bb61 KW |
783 | S_is_cur_LC_category_utf8(pTHX_ int category) |
784 | { | |
785 | /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE | |
786 | * otherwise. 'category' may not be LC_ALL. If the platform doesn't have | |
119ee68b KW |
787 | * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence |
788 | * could give the wrong result. It errs on the side of not being a UTF-8 | |
789 | * locale. */ | |
7d74bb61 KW |
790 | |
791 | char *save_input_locale = NULL; | |
7d74bb61 KW |
792 | STRLEN final_pos; |
793 | ||
8ef6e574 | 794 | #ifdef LC_ALL |
7d74bb61 | 795 | assert(category != LC_ALL); |
8ef6e574 | 796 | #endif |
7d74bb61 KW |
797 | |
798 | /* First dispose of the trivial cases */ | |
b07fffd1 | 799 | save_input_locale = setlocale(category, NULL); |
7d74bb61 | 800 | if (! save_input_locale) { |
69014004 KW |
801 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
802 | "Could not find current locale for category %d\n", | |
803 | category)); | |
7d74bb61 KW |
804 | return FALSE; /* XXX maybe should croak */ |
805 | } | |
b07fffd1 | 806 | save_input_locale = stdize_locale(savepv(save_input_locale)); |
7d74bb61 KW |
807 | if ((*save_input_locale == 'C' && save_input_locale[1] == '\0') |
808 | || strEQ(save_input_locale, "POSIX")) | |
809 | { | |
69014004 KW |
810 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
811 | "Current locale for category %d is %s\n", | |
812 | category, save_input_locale)); | |
b07fffd1 | 813 | Safefree(save_input_locale); |
7d74bb61 KW |
814 | return FALSE; |
815 | } | |
816 | ||
1d958db2 KW |
817 | #if defined(USE_LOCALE_CTYPE) \ |
818 | && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET))) | |
7d74bb61 | 819 | |
1d958db2 | 820 | { /* Next try nl_langinfo or MB_CUR_MAX if available */ |
7d74bb61 KW |
821 | |
822 | char *save_ctype_locale = NULL; | |
119ee68b | 823 | bool is_utf8; |
7d74bb61 | 824 | |
119ee68b | 825 | if (category != LC_CTYPE) { /* These work only on LC_CTYPE */ |
7d74bb61 KW |
826 | |
827 | /* Get the current LC_CTYPE locale */ | |
828 | save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL))); | |
829 | if (! save_ctype_locale) { | |
69014004 KW |
830 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
831 | "Could not find current locale for LC_CTYPE\n")); | |
7d74bb61 KW |
832 | goto cant_use_nllanginfo; |
833 | } | |
834 | ||
835 | /* If LC_CTYPE and the desired category use the same locale, this | |
836 | * means that finding the value for LC_CTYPE is the same as finding | |
837 | * the value for the desired category. Otherwise, switch LC_CTYPE | |
838 | * to the desired category's locale */ | |
839 | if (strEQ(save_ctype_locale, save_input_locale)) { | |
840 | Safefree(save_ctype_locale); | |
841 | save_ctype_locale = NULL; | |
842 | } | |
843 | else if (! setlocale(LC_CTYPE, save_input_locale)) { | |
69014004 KW |
844 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
845 | "Could not change LC_CTYPE locale to %s\n", | |
846 | save_input_locale)); | |
7d74bb61 KW |
847 | Safefree(save_ctype_locale); |
848 | goto cant_use_nllanginfo; | |
849 | } | |
850 | } | |
851 | ||
69014004 KW |
852 | DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n", |
853 | save_input_locale)); | |
854 | ||
7d74bb61 | 855 | /* Here the current LC_CTYPE is set to the locale of the category whose |
1d958db2 KW |
856 | * information is desired. This means that nl_langinfo() and MB_CUR_MAX |
857 | * should give the correct results */ | |
119ee68b | 858 | |
1d958db2 KW |
859 | # if defined(HAS_NL_LANGINFO) && defined(CODESET) |
860 | { | |
861 | char *codeset = savepv(nl_langinfo(CODESET)); | |
862 | if (codeset && strNE(codeset, "")) { | |
119ee68b | 863 | |
1d958db2 KW |
864 | /* If we switched LC_CTYPE, switch back */ |
865 | if (save_ctype_locale) { | |
866 | setlocale(LC_CTYPE, save_ctype_locale); | |
867 | Safefree(save_ctype_locale); | |
868 | } | |
869 | ||
870 | is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8")) | |
871 | || foldEQ(codeset, STR_WITH_LEN("UTF8")); | |
872 | ||
69014004 KW |
873 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
874 | "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n", | |
875 | codeset, is_utf8)); | |
1d958db2 KW |
876 | Safefree(codeset); |
877 | Safefree(save_input_locale); | |
878 | return is_utf8; | |
879 | } | |
119ee68b KW |
880 | } |
881 | ||
1d958db2 KW |
882 | # endif |
883 | # ifdef MB_CUR_MAX | |
884 | ||
885 | /* Here, either we don't have nl_langinfo, or it didn't return a | |
886 | * codeset. Try MB_CUR_MAX */ | |
887 | ||
119ee68b KW |
888 | /* Standard UTF-8 needs at least 4 bytes to represent the maximum |
889 | * Unicode code point. Since UTF-8 is the only non-single byte | |
890 | * encoding we handle, we just say any such encoding is UTF-8, and if | |
891 | * turns out to be wrong, other things will fail */ | |
892 | is_utf8 = MB_CUR_MAX >= 4; | |
893 | ||
69014004 KW |
894 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
895 | "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n", | |
896 | (int) MB_CUR_MAX, is_utf8)); | |
897 | ||
119ee68b KW |
898 | Safefree(save_input_locale); |
899 | ||
900 | # ifdef HAS_MBTOWC | |
901 | ||
902 | /* ... But, most system that have MB_CUR_MAX will also have mbtowc(), | |
903 | * since they are both in the C99 standard. We can feed a known byte | |
904 | * string to the latter function, and check that it gives the expected | |
905 | * result */ | |
906 | if (is_utf8) { | |
907 | wchar_t wc; | |
a6715020 | 908 | GCC_DIAG_IGNORE(-Wunused-result); |
1d958db2 | 909 | (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */ |
a6715020 | 910 | GCC_DIAG_RESTORE; |
69014004 | 911 | errno = 0; |
119ee68b KW |
912 | if (mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)) |
913 | != strlen(HYPHEN_UTF8) | |
914 | || wc != (wchar_t) 0x2010) | |
915 | { | |
916 | is_utf8 = FALSE; | |
69014004 KW |
917 | DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", wc)); |
918 | DEBUG_L(PerlIO_printf(Perl_debug_log, | |
919 | "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n", | |
920 | mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno)); | |
119ee68b KW |
921 | } |
922 | } | |
119ee68b KW |
923 | # endif |
924 | ||
1d958db2 KW |
925 | /* If we switched LC_CTYPE, switch back */ |
926 | if (save_ctype_locale) { | |
927 | setlocale(LC_CTYPE, save_ctype_locale); | |
928 | Safefree(save_ctype_locale); | |
119ee68b | 929 | } |
7d74bb61 | 930 | |
1d958db2 | 931 | return is_utf8; |
119ee68b | 932 | # endif |
7d74bb61 | 933 | } |
119ee68b | 934 | |
7d74bb61 KW |
935 | cant_use_nllanginfo: |
936 | ||
937 | #endif /* HAS_NL_LANGINFO etc */ | |
938 | ||
939 | /* nl_langinfo not available or failed somehow. Look at the locale name to | |
751e9426 | 940 | * see if it matches qr/UTF -? 8 /ix */ |
7d74bb61 KW |
941 | |
942 | final_pos = strlen(save_input_locale) - 1; | |
751e9426 KW |
943 | if (final_pos >= 3) { |
944 | char *name = save_input_locale; | |
945 | ||
946 | /* Find next 'U' or 'u' and look from there */ | |
947 | while ((name += strcspn(name, "Uu") + 1) | |
948 | <= save_input_locale + final_pos - 2) | |
7d74bb61 | 949 | { |
751e9426 KW |
950 | if (toFOLD(*(name)) != 't' |
951 | || toFOLD(*(name + 1)) != 'f') | |
952 | { | |
953 | continue; | |
954 | } | |
955 | name += 2; | |
956 | if (*(name) == '-') { | |
957 | if ((name > save_input_locale + final_pos - 1)) { | |
958 | break; | |
959 | } | |
960 | name++; | |
961 | } | |
962 | if (*(name) == '8') { | |
963 | Safefree(save_input_locale); | |
69014004 KW |
964 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
965 | "Locale %s ends with UTF-8 in name\n", | |
966 | save_input_locale)); | |
751e9426 KW |
967 | return TRUE; |
968 | } | |
7d74bb61 | 969 | } |
69014004 KW |
970 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
971 | "Locale %s doesn't end with UTF-8 in name\n", | |
972 | save_input_locale)); | |
7d74bb61 KW |
973 | } |
974 | ||
975 | #ifdef WIN32 | |
976 | /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */ | |
977 | if (final_pos >= 4 | |
978 | && *(save_input_locale + final_pos - 0) == '1' | |
979 | && *(save_input_locale + final_pos - 1) == '0' | |
980 | && *(save_input_locale + final_pos - 2) == '0' | |
981 | && *(save_input_locale + final_pos - 3) == '5' | |
982 | && *(save_input_locale + final_pos - 4) == '6') | |
983 | { | |
984 | Safefree(save_input_locale); | |
69014004 KW |
985 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
986 | "Locale %s ends with 10056 in name, is UTF-8 locale\n", | |
987 | save_input_locale)); | |
7d74bb61 KW |
988 | return TRUE; |
989 | } | |
990 | #endif | |
991 | ||
fa9b773e KW |
992 | /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */ |
993 | if (instr(save_input_locale, "8859")) { | |
994 | Safefree(save_input_locale); | |
69014004 KW |
995 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
996 | "Locale %s has 8859 in name, not UTF-8 locale\n", | |
997 | save_input_locale)); | |
fa9b773e KW |
998 | return FALSE; |
999 | } | |
1000 | ||
1001 | #ifdef HAS_LOCALECONV | |
1002 | ||
1003 | # ifdef USE_LOCALE_MONETARY | |
1004 | ||
1005 | /* Here, there is nothing in the locale name to indicate whether the locale | |
1006 | * is UTF-8 or not. This "name", the return of setlocale(), is actually | |
1007 | * defined to be opaque, so we can't really rely on the absence of various | |
1008 | * substrings in the name to indicate its UTF-8ness. Look at the locale's | |
1009 | * currency symbol. Often that will be in the native script, and if the | |
1010 | * symbol isn't in UTF-8, we know that the locale isn't. If it is | |
1011 | * non-ASCII UTF-8, we infer that the locale is too. | |
1012 | * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the | |
1013 | * locale of the desired category, if it isn't that locale already */ | |
1014 | ||
1015 | { | |
1016 | char *save_monetary_locale = NULL; | |
1017 | bool illegal_utf8 = FALSE; | |
1018 | bool only_ascii = FALSE; | |
1019 | const struct lconv* const lc = localeconv(); | |
1020 | ||
1021 | if (category != LC_MONETARY) { | |
1022 | ||
1023 | save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY, | |
1024 | NULL))); | |
1025 | if (! save_monetary_locale) { | |
69014004 KW |
1026 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
1027 | "Could not find current locale for LC_MONETARY\n")); | |
fa9b773e KW |
1028 | goto cant_use_monetary; |
1029 | } | |
1030 | ||
1031 | if (strNE(save_monetary_locale, save_input_locale)) { | |
1032 | if (! setlocale(LC_MONETARY, save_input_locale)) { | |
69014004 KW |
1033 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
1034 | "Could not change LC_MONETARY locale to %s\n", | |
1035 | save_input_locale)); | |
fa9b773e KW |
1036 | Safefree(save_monetary_locale); |
1037 | goto cant_use_monetary; | |
1038 | } | |
1039 | } | |
1040 | } | |
1041 | ||
1042 | /* Here the current LC_MONETARY is set to the locale of the category | |
1043 | * whose information is desired. */ | |
1044 | ||
1045 | if (lc && lc->currency_symbol) { | |
1046 | if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) { | |
69014004 KW |
1047 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
1048 | "Currency symbol for %s is not legal UTF-8\n", | |
1049 | save_input_locale)); | |
fa9b773e KW |
1050 | illegal_utf8 = TRUE; |
1051 | } | |
1052 | else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) { | |
69014004 | 1053 | DEBUG_L(PerlIO_printf(Perl_debug_log, "Currency symbol for %s contains only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale)); |
fa9b773e KW |
1054 | only_ascii = TRUE; |
1055 | } | |
1056 | } | |
1057 | ||
1058 | /* If we changed it, restore LC_MONETARY to its original locale */ | |
1059 | if (save_monetary_locale) { | |
1060 | setlocale(LC_MONETARY, save_monetary_locale); | |
1061 | Safefree(save_monetary_locale); | |
1062 | } | |
1063 | ||
1064 | Safefree(save_input_locale); | |
1065 | ||
1066 | /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise | |
1067 | * assume the locale is UTF-8 if and only if the symbol is non-ascii | |
1068 | * UTF-8. (We can't really tell if the locale is UTF-8 or not if the | |
1069 | * symbol is just a '$', so we err on the side of it not being UTF-8) | |
1070 | * */ | |
69014004 KW |
1071 | DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", (illegal_utf8) |
1072 | ? FALSE | |
1073 | : ! only_ascii)); | |
fa9b773e KW |
1074 | return (illegal_utf8) |
1075 | ? FALSE | |
1076 | : ! only_ascii; | |
1077 | ||
1078 | } | |
1079 | cant_use_monetary: | |
1080 | ||
1081 | # endif /* USE_LOCALE_MONETARY */ | |
1082 | #endif /* HAS_LOCALECONV */ | |
1083 | ||
1084 | #if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES) | |
1085 | ||
1086 | /* This code is ifdefd out because it was found to not be necessary in testing | |
1087 | * on our dromedary test machine, which has over 700 locales. There, looking | |
1088 | * at just the currency symbol gave essentially the same results as doing this | |
1089 | * extra work. Executing this also caused segfaults in miniperl. I left it in | |
1090 | * so as to avoid rewriting it if real-world experience indicates that | |
1091 | * dromedary is an outlier. Essentially, instead of returning abpve if we | |
1092 | * haven't found illegal utf8, we continue on and examine all the strerror() | |
1093 | * messages on the platform for utf8ness. If all are ASCII, we still don't | |
1094 | * know the answer; but otherwise we have a pretty good indication of the | |
1095 | * utf8ness. The reason this doesn't necessarily help much is that the | |
1096 | * messages may not have been translated into the locale. The currency symbol | |
1097 | * is much more likely to have been translated. The code below would need to | |
1098 | * be altered somewhat to just be a continuation of testing the currency | |
1099 | * symbol. */ | |
1100 | int e; | |
1101 | unsigned int failures = 0, non_ascii = 0; | |
1102 | char *save_messages_locale = NULL; | |
1103 | ||
1104 | /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the | |
1105 | * desired category, if it isn't that locale already */ | |
1106 | ||
1107 | if (category != LC_MESSAGES) { | |
1108 | ||
1109 | save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES, | |
1110 | NULL))); | |
1111 | if (! save_messages_locale) { | |
1112 | goto cant_use_messages; | |
1113 | } | |
1114 | ||
1115 | if (strEQ(save_messages_locale, save_input_locale)) { | |
1116 | Safefree(save_input_locale); | |
1117 | } | |
1118 | else if (! setlocale(LC_MESSAGES, save_input_locale)) { | |
1119 | Safefree(save_messages_locale); | |
1120 | goto cant_use_messages; | |
1121 | } | |
1122 | } | |
1123 | ||
1124 | /* Here the current LC_MESSAGES is set to the locale of the category | |
1125 | * whose information is desired. Look through all the messages */ | |
1126 | ||
1127 | for (e = 0; | |
1128 | #ifdef HAS_SYS_ERRLIST | |
1129 | e <= sys_nerr | |
1130 | #endif | |
1131 | ; e++) | |
1132 | { | |
1133 | const U8* const errmsg = (U8 *) Strerror(e) ; | |
1134 | if (!errmsg) | |
1135 | break; | |
1136 | if (! is_utf8_string(errmsg, 0)) { | |
1137 | failures++; | |
1138 | break; | |
1139 | } | |
1140 | else if (! is_ascii_string(errmsg, 0)) { | |
1141 | non_ascii++; | |
1142 | } | |
1143 | } | |
1144 | ||
1145 | /* And, if we changed it, restore LC_MESSAGES to its original locale */ | |
1146 | if (save_messages_locale) { | |
1147 | setlocale(LC_MESSAGES, save_messages_locale); | |
1148 | Safefree(save_messages_locale); | |
1149 | } | |
1150 | ||
1151 | /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid, | |
1152 | * any non-ascii means it is one; otherwise we assume it isn't */ | |
1153 | return (failures) ? FALSE : non_ascii; | |
1154 | ||
1155 | } | |
1156 | cant_use_messages: | |
1157 | ||
1158 | #endif | |
1159 | ||
69014004 KW |
1160 | DEBUG_L(PerlIO_printf(Perl_debug_log, |
1161 | "Assuming locale %s is not a UTF-8 locale\n", | |
1162 | save_input_locale)); | |
fa9b773e | 1163 | Safefree(save_input_locale); |
7d74bb61 KW |
1164 | return FALSE; |
1165 | } | |
1166 | ||
8ef6e574 | 1167 | #endif |
7d74bb61 | 1168 | |
66610fdd RGS |
1169 | /* |
1170 | * Local variables: | |
1171 | * c-indentation-style: bsd | |
1172 | * c-basic-offset: 4 | |
14d04a33 | 1173 | * indent-tabs-mode: nil |
66610fdd RGS |
1174 | * End: |
1175 | * | |
14d04a33 | 1176 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 1177 | */ |