This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Make taint checking regex compile time instead of runtime
[perl5.git] / locale.c
CommitLineData
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 */
62STATIC char *
63S_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
92void
93Perl_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 130void
8772537c 131Perl_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
188void
189Perl_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
210void
211Perl_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 */
236void
8772537c 237Perl_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 281void
8772537c 282Perl_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
336char *
337Perl_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 */
416int
417Perl_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
65ebb059 448
ccd65d51
KW
449 /* NULL uses the existing already set up locale */
450 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
451 ? NULL
452 : "";
c3fcd832
KW
453 const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
454 unsigned int trial_locales_count;
7452cf6a
AL
455 char * const lc_all = PerlEnv_getenv("LC_ALL");
456 char * const lang = PerlEnv_getenv("LANG");
98994639 457 bool setlocale_failure = FALSE;
65ebb059
KW
458 unsigned int i;
459 char *p;
460 const bool locwarn = (printwarn > 1 ||
461 (printwarn &&
462 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
0e92a118 463 bool done = FALSE;
65ebb059 464 const char *system_default_locale = NULL;
0e92a118 465
98994639 466
0e92a118
KW
467#ifndef LOCALE_ENVIRON_REQUIRED
468 PERL_UNUSED_VAR(done);
469#else
98994639
HS
470
471 /*
472 * Ultrix setlocale(..., "") fails if there are no environment
473 * variables from which to get a locale name.
474 */
475
b3e384bf 476# ifdef LC_ALL
98994639 477 if (lang) {
b385bb4d 478 if (my_setlocale(LC_ALL, setlocale_init))
98994639
HS
479 done = TRUE;
480 else
481 setlocale_failure = TRUE;
482 }
483 if (!setlocale_failure) {
b3e384bf 484# ifdef USE_LOCALE_CTYPE
56279a21 485 Safefree(curctype);
98994639 486 if (! (curctype =
b385bb4d 487 my_setlocale(LC_CTYPE,
98994639 488 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
ccd65d51 489 ? setlocale_init : NULL)))
98994639
HS
490 setlocale_failure = TRUE;
491 else
492 curctype = savepv(curctype);
b3e384bf
KW
493# endif /* USE_LOCALE_CTYPE */
494# ifdef USE_LOCALE_COLLATE
56279a21 495 Safefree(curcoll);
98994639 496 if (! (curcoll =
b385bb4d 497 my_setlocale(LC_COLLATE,
98994639 498 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
ccd65d51 499 ? setlocale_init : NULL)))
98994639
HS
500 setlocale_failure = TRUE;
501 else
502 curcoll = savepv(curcoll);
b3e384bf
KW
503# endif /* USE_LOCALE_COLLATE */
504# ifdef USE_LOCALE_NUMERIC
56279a21 505 Safefree(curnum);
98994639 506 if (! (curnum =
b385bb4d 507 my_setlocale(LC_NUMERIC,
98994639 508 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
ccd65d51 509 ? setlocale_init : NULL)))
98994639
HS
510 setlocale_failure = TRUE;
511 else
512 curnum = savepv(curnum);
b3e384bf 513# endif /* USE_LOCALE_NUMERIC */
a782673d
KW
514# ifdef USE_LOCALE_MESSAGES
515 if (! my_setlocale(LC_MESSAGES,
516 (!done && (lang || PerlEnv_getenv("LC_MESSAGES")))
517 ? setlocale_init : NULL))
518 {
519 setlocale_failure = TRUE;
520 }
521# endif /* USE_LOCALE_MESSAGES */
c835d6be
KW
522# ifdef USE_LOCALE_MONETARY
523 if (! my_setlocale(LC_MONETARY,
524 (!done && (lang || PerlEnv_getenv("LC_MONETARY")))
525 ? setlocale_init : NULL))
526 {
527 setlocale_failure = TRUE;
528 }
529# endif /* USE_LOCALE_MONETARY */
98994639
HS
530 }
531
b3e384bf 532# endif /* LC_ALL */
98994639
HS
533
534#endif /* !LOCALE_ENVIRON_REQUIRED */
535
65ebb059
KW
536 /* We try each locale in the list until we get one that works, or exhaust
537 * the list */
c3fcd832
KW
538 trial_locales[0] = setlocale_init;
539 trial_locales_count = 1;
65ebb059
KW
540 for (i= 0; i < trial_locales_count; i++) {
541 const char * trial_locale = trial_locales[i];
542
543 if (i > 0) {
544
545 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
546 * when i==0, but I (khw) don't think that behavior makes much
547 * sense */
548 setlocale_failure = FALSE;
549
550#ifdef WIN32
551
552 /* On Windows machines, an entry of "" after the 0th means to use
553 * the system default locale, which we now proceed to get. */
554 if (strEQ(trial_locale, "")) {
555 unsigned int j;
556
557 /* Note that this may change the locale, but we are going to do
558 * that anyway just below */
559 system_default_locale = setlocale(LC_ALL, "");
560
561 /* Skip if invalid or it's already on the list of locales to
562 * try */
563 if (! system_default_locale) {
564 goto next_iteration;
565 }
566 for (j = 0; j < trial_locales_count; j++) {
567 if (strEQ(system_default_locale, trial_locales[j])) {
568 goto next_iteration;
569 }
570 }
571
572 trial_locale = system_default_locale;
573 }
574#endif
575 }
576
98994639 577#ifdef LC_ALL
7cd8b568 578 if (! my_setlocale(LC_ALL, trial_locale)) {
49c85077 579 setlocale_failure = TRUE;
7cd8b568
KW
580 }
581 else {
582 /* Since LC_ALL succeeded, it should have changed all the other
583 * categories it can to its value; so we massage things so that the
584 * setlocales below just return their category's current values.
585 * This adequately handles the case in NetBSD where LC_COLLATE may
586 * not be defined for a locale, and setting it individually will
587 * fail, whereas setting LC_ALL suceeds, leaving LC_COLLATE set to
588 * the POSIX locale. */
589 trial_locale = NULL;
590 }
98994639
HS
591#endif /* LC_ALL */
592
49c85077 593 if (!setlocale_failure) {
98994639 594#ifdef USE_LOCALE_CTYPE
49c85077
KW
595 Safefree(curctype);
596 if (! (curctype = my_setlocale(LC_CTYPE, trial_locale)))
597 setlocale_failure = TRUE;
598 else
599 curctype = savepv(curctype);
98994639
HS
600#endif /* USE_LOCALE_CTYPE */
601#ifdef USE_LOCALE_COLLATE
49c85077
KW
602 Safefree(curcoll);
603 if (! (curcoll = my_setlocale(LC_COLLATE, trial_locale)))
604 setlocale_failure = TRUE;
605 else
606 curcoll = savepv(curcoll);
98994639
HS
607#endif /* USE_LOCALE_COLLATE */
608#ifdef USE_LOCALE_NUMERIC
49c85077
KW
609 Safefree(curnum);
610 if (! (curnum = my_setlocale(LC_NUMERIC, trial_locale)))
611 setlocale_failure = TRUE;
612 else
613 curnum = savepv(curnum);
98994639 614#endif /* USE_LOCALE_NUMERIC */
a782673d
KW
615#ifdef USE_LOCALE_MESSAGES
616 if (! (my_setlocale(LC_MESSAGES, trial_locale)))
617 setlocale_failure = TRUE;
618#endif /* USE_LOCALE_MESSAGES */
c835d6be
KW
619#ifdef USE_LOCALE_MONETARY
620 if (! (my_setlocale(LC_MONETARY, trial_locale)))
621 setlocale_failure = TRUE;
622#endif /* USE_LOCALE_MONETARY */
623
49c85077
KW
624 if (! setlocale_failure) { /* Success */
625 break;
626 }
65ebb059 627 }
98994639 628
49c85077
KW
629 /* Here, something failed; will need to try a fallback. */
630 ok = 0;
65ebb059 631
49c85077
KW
632 if (i == 0) {
633 unsigned int j;
98994639 634
65ebb059 635 if (locwarn) { /* Output failure info only on the first one */
98994639
HS
636#ifdef LC_ALL
637
49c85077
KW
638 PerlIO_printf(Perl_error_log,
639 "perl: warning: Setting locale failed.\n");
98994639
HS
640
641#else /* !LC_ALL */
642
49c85077
KW
643 PerlIO_printf(Perl_error_log,
644 "perl: warning: Setting locale failed for the categories:\n\t");
98994639 645#ifdef USE_LOCALE_CTYPE
49c85077
KW
646 if (! curctype)
647 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
98994639
HS
648#endif /* USE_LOCALE_CTYPE */
649#ifdef USE_LOCALE_COLLATE
49c85077
KW
650 if (! curcoll)
651 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
98994639
HS
652#endif /* USE_LOCALE_COLLATE */
653#ifdef USE_LOCALE_NUMERIC
49c85077
KW
654 if (! curnum)
655 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
98994639 656#endif /* USE_LOCALE_NUMERIC */
a782673d 657 PerlIO_printf(Perl_error_log, "and possibly others\n");
98994639
HS
658
659#endif /* LC_ALL */
660
49c85077
KW
661 PerlIO_printf(Perl_error_log,
662 "perl: warning: Please check that your locale settings:\n");
98994639
HS
663
664#ifdef __GLIBC__
49c85077
KW
665 PerlIO_printf(Perl_error_log,
666 "\tLANGUAGE = %c%s%c,\n",
667 language ? '"' : '(',
668 language ? language : "unset",
669 language ? '"' : ')');
98994639
HS
670#endif
671
49c85077
KW
672 PerlIO_printf(Perl_error_log,
673 "\tLC_ALL = %c%s%c,\n",
674 lc_all ? '"' : '(',
675 lc_all ? lc_all : "unset",
676 lc_all ? '"' : ')');
98994639
HS
677
678#if defined(USE_ENVIRON_ARRAY)
49c85077
KW
679 {
680 char **e;
681 for (e = environ; *e; e++) {
682 if (strnEQ(*e, "LC_", 3)
683 && strnNE(*e, "LC_ALL=", 7)
684 && (p = strchr(*e, '=')))
685 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
686 (int)(p - *e), *e, p + 1);
687 }
688 }
98994639 689#else
49c85077
KW
690 PerlIO_printf(Perl_error_log,
691 "\t(possibly more locale environment variables)\n");
98994639
HS
692#endif
693
49c85077
KW
694 PerlIO_printf(Perl_error_log,
695 "\tLANG = %c%s%c\n",
696 lang ? '"' : '(',
697 lang ? lang : "unset",
698 lang ? '"' : ')');
98994639 699
49c85077
KW
700 PerlIO_printf(Perl_error_log,
701 " are supported and installed on your system.\n");
702 }
98994639 703
65ebb059
KW
704 /* Calculate what fallback locales to try. We have avoided this
705 * until we have to, becuase failure is quite unlikely. This will
706 * usually change the upper bound of the loop we are in.
707 *
708 * Since the system's default way of setting the locale has not
709 * found one that works, We use Perl's defined ordering: LC_ALL,
710 * LANG, and the C locale. We don't try the same locale twice, so
711 * don't add to the list if already there. (On POSIX systems, the
712 * LC_ALL element will likely be a repeat of the 0th element "",
713 * but there's no harm done by doing it explicitly */
714 if (lc_all) {
715 for (j = 0; j < trial_locales_count; j++) {
716 if (strEQ(lc_all, trial_locales[j])) {
717 goto done_lc_all;
718 }
719 }
720 trial_locales[trial_locales_count++] = lc_all;
721 }
722 done_lc_all:
98994639 723
65ebb059
KW
724 if (lang) {
725 for (j = 0; j < trial_locales_count; j++) {
726 if (strEQ(lang, trial_locales[j])) {
727 goto done_lang;
728 }
729 }
730 trial_locales[trial_locales_count++] = lang;
731 }
732 done_lang:
733
734#if defined(WIN32) && defined(LC_ALL)
735 /* For Windows, we also try the system default locale before "C".
736 * (If there exists a Windows without LC_ALL we skip this because
737 * it gets too complicated. For those, the "C" is the next
738 * fallback possibility). The "" is the same as the 0th element of
739 * the array, but the code at the loop above knows to treat it
740 * differently when not the 0th */
741 trial_locales[trial_locales_count++] = "";
742#endif
743
744 for (j = 0; j < trial_locales_count; j++) {
745 if (strEQ("C", trial_locales[j])) {
746 goto done_C;
747 }
748 }
749 trial_locales[trial_locales_count++] = "C";
98994639 750
65ebb059
KW
751 done_C: ;
752 } /* end of first time through the loop */
98994639 753
65ebb059
KW
754#ifdef WIN32
755 next_iteration: ;
756#endif
757
758 } /* end of looping through the trial locales */
759
760 if (ok < 1) { /* If we tried to fallback */
761 const char* msg;
762 if (! setlocale_failure) { /* fallback succeeded */
763 msg = "Falling back to";
764 }
765 else { /* fallback failed */
98994639 766
65ebb059
KW
767 /* We dropped off the end of the loop, so have to decrement i to
768 * get back to the value the last time through */
769 i--;
98994639 770
65ebb059
KW
771 ok = -1;
772 msg = "Failed to fall back to";
773
774 /* To continue, we should use whatever values we've got */
98994639 775#ifdef USE_LOCALE_CTYPE
49c85077
KW
776 Safefree(curctype);
777 curctype = savepv(setlocale(LC_CTYPE, NULL));
98994639
HS
778#endif /* USE_LOCALE_CTYPE */
779#ifdef USE_LOCALE_COLLATE
49c85077
KW
780 Safefree(curcoll);
781 curcoll = savepv(setlocale(LC_COLLATE, NULL));
98994639
HS
782#endif /* USE_LOCALE_COLLATE */
783#ifdef USE_LOCALE_NUMERIC
49c85077
KW
784 Safefree(curnum);
785 curnum = savepv(setlocale(LC_NUMERIC, NULL));
98994639 786#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
787 }
788
789 if (locwarn) {
790 const char * description;
791 const char * name = "";
792 if (strEQ(trial_locales[i], "C")) {
793 description = "the standard locale";
794 name = "C";
795 }
796 else if (strEQ(trial_locales[i], "")) {
797 description = "the system default locale";
798 if (system_default_locale) {
799 name = system_default_locale;
800 }
801 }
802 else {
803 description = "a fallback locale";
804 name = trial_locales[i];
805 }
806 if (name && strNE(name, "")) {
807 PerlIO_printf(Perl_error_log,
808 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
809 }
810 else {
811 PerlIO_printf(Perl_error_log,
812 "perl: warning: %s %s.\n", msg, description);
813 }
814 }
815 } /* End of tried to fallback */
98994639
HS
816
817#ifdef USE_LOCALE_CTYPE
818 new_ctype(curctype);
819#endif /* USE_LOCALE_CTYPE */
820
821#ifdef USE_LOCALE_COLLATE
822 new_collate(curcoll);
823#endif /* USE_LOCALE_COLLATE */
824
825#ifdef USE_LOCALE_NUMERIC
826 new_numeric(curnum);
827#endif /* USE_LOCALE_NUMERIC */
b310b053 828
8ef6e574 829#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
49c85077
KW
830 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
831 * locale is UTF-8. If PL_utf8locale and PL_unicode (set by -C or by
832 * $ENV{PERL_UNICODE}) are true, perl.c:S_parse_body() will turn on the
833 * PerlIO :utf8 layer on STDIN, STDOUT, STDERR, _and_ the default open
834 * discipline. */
835 PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE);
836
a05d7ebb 837 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
838 This is an alternative to using the -C command line switch
839 (the -C if present will override this). */
840 {
dd374669 841 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 842 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
843 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
844 PL_utf8cache = -1;
b310b053 845 }
ec71e770 846#endif
b310b053 847
98994639 848#ifdef USE_LOCALE_CTYPE
43c5f42d 849 Safefree(curctype);
98994639
HS
850#endif /* USE_LOCALE_CTYPE */
851#ifdef USE_LOCALE_COLLATE
43c5f42d 852 Safefree(curcoll);
98994639
HS
853#endif /* USE_LOCALE_COLLATE */
854#ifdef USE_LOCALE_NUMERIC
43c5f42d 855 Safefree(curnum);
98994639 856#endif /* USE_LOCALE_NUMERIC */
8ef6e574
KW
857
858#endif /* USE_LOCALE */
859
98994639
HS
860 return ok;
861}
862
49c85077 863
98994639
HS
864#ifdef USE_LOCALE_COLLATE
865
866/*
867 * mem_collxfrm() is a bit like strxfrm() but with two important
868 * differences. First, it handles embedded NULs. Second, it allocates
869 * a bit more memory than needed for the transformed data itself.
870 * The real transformed data begins at offset sizeof(collationix).
871 * Please see sv_collxfrm() to see how this is used.
872 */
166f8a29 873
98994639
HS
874char *
875Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
876{
97aff369 877 dVAR;
98994639
HS
878 char *xbuf;
879 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
880
7918f24d
NC
881 PERL_ARGS_ASSERT_MEM_COLLXFRM;
882
98994639
HS
883 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
884 /* the +1 is for the terminating NUL. */
885
886 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 887 Newx(xbuf, xAlloc, char);
98994639
HS
888 if (! xbuf)
889 goto bad;
890
891 *(U32*)xbuf = PL_collation_ix;
892 xout = sizeof(PL_collation_ix);
893 for (xin = 0; xin < len; ) {
224e8ef5 894 Size_t xused;
98994639
HS
895
896 for (;;) {
897 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 898 if (xused >= PERL_INT_MAX)
98994639 899 goto bad;
eb160463 900 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
901 break;
902 xAlloc = (2 * xAlloc) + 1;
903 Renew(xbuf, xAlloc, char);
904 if (! xbuf)
905 goto bad;
906 }
907
908 xin += strlen(s + xin) + 1;
909 xout += xused;
910
911 /* Embedded NULs are understood but silently skipped
912 * because they make no sense in locale collation. */
913 }
914
915 xbuf[xout] = '\0';
916 *xlen = xout - sizeof(PL_collation_ix);
917 return xbuf;
918
919 bad:
920 Safefree(xbuf);
921 *xlen = 0;
922 return NULL;
923}
924
925#endif /* USE_LOCALE_COLLATE */
926
8ef6e574
KW
927#ifdef USE_LOCALE
928
637917bb 929STATIC bool
7d74bb61
KW
930S_is_cur_LC_category_utf8(pTHX_ int category)
931{
932 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
933 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b
KW
934 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
935 * could give the wrong result. It errs on the side of not being a UTF-8
936 * locale. */
7d74bb61
KW
937
938 char *save_input_locale = NULL;
7d74bb61
KW
939 STRLEN final_pos;
940
8ef6e574 941#ifdef LC_ALL
7d74bb61 942 assert(category != LC_ALL);
8ef6e574 943#endif
7d74bb61
KW
944
945 /* First dispose of the trivial cases */
b07fffd1 946 save_input_locale = setlocale(category, NULL);
7d74bb61 947 if (! save_input_locale) {
69014004
KW
948 DEBUG_L(PerlIO_printf(Perl_debug_log,
949 "Could not find current locale for category %d\n",
950 category));
7d74bb61
KW
951 return FALSE; /* XXX maybe should croak */
952 }
b07fffd1 953 save_input_locale = stdize_locale(savepv(save_input_locale));
7d74bb61
KW
954 if ((*save_input_locale == 'C' && save_input_locale[1] == '\0')
955 || strEQ(save_input_locale, "POSIX"))
956 {
69014004
KW
957 DEBUG_L(PerlIO_printf(Perl_debug_log,
958 "Current locale for category %d is %s\n",
959 category, save_input_locale));
b07fffd1 960 Safefree(save_input_locale);
7d74bb61
KW
961 return FALSE;
962 }
963
1d958db2
KW
964#if defined(USE_LOCALE_CTYPE) \
965 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 966
1d958db2 967 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
968
969 char *save_ctype_locale = NULL;
119ee68b 970 bool is_utf8;
7d74bb61 971
119ee68b 972 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
973
974 /* Get the current LC_CTYPE locale */
975 save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL)));
976 if (! save_ctype_locale) {
69014004
KW
977 DEBUG_L(PerlIO_printf(Perl_debug_log,
978 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
979 goto cant_use_nllanginfo;
980 }
981
982 /* If LC_CTYPE and the desired category use the same locale, this
983 * means that finding the value for LC_CTYPE is the same as finding
984 * the value for the desired category. Otherwise, switch LC_CTYPE
985 * to the desired category's locale */
986 if (strEQ(save_ctype_locale, save_input_locale)) {
987 Safefree(save_ctype_locale);
988 save_ctype_locale = NULL;
989 }
990 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
991 DEBUG_L(PerlIO_printf(Perl_debug_log,
992 "Could not change LC_CTYPE locale to %s\n",
993 save_input_locale));
7d74bb61
KW
994 Safefree(save_ctype_locale);
995 goto cant_use_nllanginfo;
996 }
997 }
998
69014004
KW
999 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
1000 save_input_locale));
1001
7d74bb61 1002 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
1003 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
1004 * should give the correct results */
119ee68b 1005
1d958db2
KW
1006# if defined(HAS_NL_LANGINFO) && defined(CODESET)
1007 {
1008 char *codeset = savepv(nl_langinfo(CODESET));
1009 if (codeset && strNE(codeset, "")) {
119ee68b 1010
1d958db2
KW
1011 /* If we switched LC_CTYPE, switch back */
1012 if (save_ctype_locale) {
1013 setlocale(LC_CTYPE, save_ctype_locale);
1014 Safefree(save_ctype_locale);
1015 }
1016
1017 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
1018 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
1019
69014004
KW
1020 DEBUG_L(PerlIO_printf(Perl_debug_log,
1021 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
1022 codeset, is_utf8));
1d958db2
KW
1023 Safefree(codeset);
1024 Safefree(save_input_locale);
1025 return is_utf8;
1026 }
119ee68b
KW
1027 }
1028
1d958db2
KW
1029# endif
1030# ifdef MB_CUR_MAX
1031
1032 /* Here, either we don't have nl_langinfo, or it didn't return a
1033 * codeset. Try MB_CUR_MAX */
1034
119ee68b
KW
1035 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
1036 * Unicode code point. Since UTF-8 is the only non-single byte
1037 * encoding we handle, we just say any such encoding is UTF-8, and if
1038 * turns out to be wrong, other things will fail */
1039 is_utf8 = MB_CUR_MAX >= 4;
1040
69014004
KW
1041 DEBUG_L(PerlIO_printf(Perl_debug_log,
1042 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
1043 (int) MB_CUR_MAX, is_utf8));
1044
119ee68b
KW
1045 Safefree(save_input_locale);
1046
1047# ifdef HAS_MBTOWC
1048
1049 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
1050 * since they are both in the C99 standard. We can feed a known byte
1051 * string to the latter function, and check that it gives the expected
1052 * result */
1053 if (is_utf8) {
1054 wchar_t wc;
a6715020 1055 GCC_DIAG_IGNORE(-Wunused-result);
1d958db2 1056 (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */
a6715020 1057 GCC_DIAG_RESTORE;
69014004 1058 errno = 0;
119ee68b
KW
1059 if (mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
1060 != strlen(HYPHEN_UTF8)
1061 || wc != (wchar_t) 0x2010)
1062 {
1063 is_utf8 = FALSE;
69014004
KW
1064 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", wc));
1065 DEBUG_L(PerlIO_printf(Perl_debug_log,
1066 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
1067 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
1068 }
1069 }
119ee68b
KW
1070# endif
1071
1d958db2
KW
1072 /* If we switched LC_CTYPE, switch back */
1073 if (save_ctype_locale) {
1074 setlocale(LC_CTYPE, save_ctype_locale);
1075 Safefree(save_ctype_locale);
119ee68b 1076 }
7d74bb61 1077
1d958db2 1078 return is_utf8;
119ee68b 1079# endif
7d74bb61 1080 }
119ee68b 1081
7d74bb61
KW
1082 cant_use_nllanginfo:
1083
1084#endif /* HAS_NL_LANGINFO etc */
1085
1086 /* nl_langinfo not available or failed somehow. Look at the locale name to
751e9426 1087 * see if it matches qr/UTF -? 8 /ix */
7d74bb61
KW
1088
1089 final_pos = strlen(save_input_locale) - 1;
751e9426
KW
1090 if (final_pos >= 3) {
1091 char *name = save_input_locale;
1092
1093 /* Find next 'U' or 'u' and look from there */
1094 while ((name += strcspn(name, "Uu") + 1)
1095 <= save_input_locale + final_pos - 2)
7d74bb61 1096 {
751e9426
KW
1097 if (toFOLD(*(name)) != 't'
1098 || toFOLD(*(name + 1)) != 'f')
1099 {
1100 continue;
1101 }
1102 name += 2;
1103 if (*(name) == '-') {
1104 if ((name > save_input_locale + final_pos - 1)) {
1105 break;
1106 }
1107 name++;
1108 }
1109 if (*(name) == '8') {
1110 Safefree(save_input_locale);
69014004
KW
1111 DEBUG_L(PerlIO_printf(Perl_debug_log,
1112 "Locale %s ends with UTF-8 in name\n",
1113 save_input_locale));
751e9426
KW
1114 return TRUE;
1115 }
7d74bb61 1116 }
69014004
KW
1117 DEBUG_L(PerlIO_printf(Perl_debug_log,
1118 "Locale %s doesn't end with UTF-8 in name\n",
1119 save_input_locale));
7d74bb61
KW
1120 }
1121
1122#ifdef WIN32
1123 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
1124 if (final_pos >= 4
1125 && *(save_input_locale + final_pos - 0) == '1'
1126 && *(save_input_locale + final_pos - 1) == '0'
1127 && *(save_input_locale + final_pos - 2) == '0'
1128 && *(save_input_locale + final_pos - 3) == '5'
1129 && *(save_input_locale + final_pos - 4) == '6')
1130 {
1131 Safefree(save_input_locale);
69014004
KW
1132 DEBUG_L(PerlIO_printf(Perl_debug_log,
1133 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
1134 save_input_locale));
7d74bb61
KW
1135 return TRUE;
1136 }
1137#endif
1138
fa9b773e
KW
1139 /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
1140 if (instr(save_input_locale, "8859")) {
1141 Safefree(save_input_locale);
69014004
KW
1142 DEBUG_L(PerlIO_printf(Perl_debug_log,
1143 "Locale %s has 8859 in name, not UTF-8 locale\n",
1144 save_input_locale));
fa9b773e
KW
1145 return FALSE;
1146 }
1147
1148#ifdef HAS_LOCALECONV
1149
1150# ifdef USE_LOCALE_MONETARY
1151
1152 /* Here, there is nothing in the locale name to indicate whether the locale
1153 * is UTF-8 or not. This "name", the return of setlocale(), is actually
1154 * defined to be opaque, so we can't really rely on the absence of various
1155 * substrings in the name to indicate its UTF-8ness. Look at the locale's
1156 * currency symbol. Often that will be in the native script, and if the
1157 * symbol isn't in UTF-8, we know that the locale isn't. If it is
1158 * non-ASCII UTF-8, we infer that the locale is too.
1159 * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the
1160 * locale of the desired category, if it isn't that locale already */
1161
1162 {
1163 char *save_monetary_locale = NULL;
1164 bool illegal_utf8 = FALSE;
1165 bool only_ascii = FALSE;
1166 const struct lconv* const lc = localeconv();
1167
1168 if (category != LC_MONETARY) {
1169
1170 save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY,
1171 NULL)));
1172 if (! save_monetary_locale) {
69014004
KW
1173 DEBUG_L(PerlIO_printf(Perl_debug_log,
1174 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1175 goto cant_use_monetary;
1176 }
1177
1178 if (strNE(save_monetary_locale, save_input_locale)) {
1179 if (! setlocale(LC_MONETARY, save_input_locale)) {
69014004
KW
1180 DEBUG_L(PerlIO_printf(Perl_debug_log,
1181 "Could not change LC_MONETARY locale to %s\n",
1182 save_input_locale));
fa9b773e
KW
1183 Safefree(save_monetary_locale);
1184 goto cant_use_monetary;
1185 }
1186 }
1187 }
1188
1189 /* Here the current LC_MONETARY is set to the locale of the category
1190 * whose information is desired. */
1191
1192 if (lc && lc->currency_symbol) {
1193 if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) {
69014004
KW
1194 DEBUG_L(PerlIO_printf(Perl_debug_log,
1195 "Currency symbol for %s is not legal UTF-8\n",
1196 save_input_locale));
fa9b773e
KW
1197 illegal_utf8 = TRUE;
1198 }
1199 else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) {
69014004 1200 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
1201 only_ascii = TRUE;
1202 }
1203 }
1204
1205 /* If we changed it, restore LC_MONETARY to its original locale */
1206 if (save_monetary_locale) {
1207 setlocale(LC_MONETARY, save_monetary_locale);
1208 Safefree(save_monetary_locale);
1209 }
1210
1211 Safefree(save_input_locale);
1212
1213 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise
1214 * assume the locale is UTF-8 if and only if the symbol is non-ascii
1215 * UTF-8. (We can't really tell if the locale is UTF-8 or not if the
1216 * symbol is just a '$', so we err on the side of it not being UTF-8)
1217 * */
69014004
KW
1218 DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", (illegal_utf8)
1219 ? FALSE
1220 : ! only_ascii));
fa9b773e
KW
1221 return (illegal_utf8)
1222 ? FALSE
1223 : ! only_ascii;
1224
1225 }
1226 cant_use_monetary:
1227
1228# endif /* USE_LOCALE_MONETARY */
1229#endif /* HAS_LOCALECONV */
1230
1231#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES)
1232
1233/* This code is ifdefd out because it was found to not be necessary in testing
1234 * on our dromedary test machine, which has over 700 locales. There, looking
1235 * at just the currency symbol gave essentially the same results as doing this
1236 * extra work. Executing this also caused segfaults in miniperl. I left it in
1237 * so as to avoid rewriting it if real-world experience indicates that
1238 * dromedary is an outlier. Essentially, instead of returning abpve if we
1239 * haven't found illegal utf8, we continue on and examine all the strerror()
1240 * messages on the platform for utf8ness. If all are ASCII, we still don't
1241 * know the answer; but otherwise we have a pretty good indication of the
1242 * utf8ness. The reason this doesn't necessarily help much is that the
1243 * messages may not have been translated into the locale. The currency symbol
1244 * is much more likely to have been translated. The code below would need to
1245 * be altered somewhat to just be a continuation of testing the currency
1246 * symbol. */
1247 int e;
1248 unsigned int failures = 0, non_ascii = 0;
1249 char *save_messages_locale = NULL;
1250
1251 /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the
1252 * desired category, if it isn't that locale already */
1253
1254 if (category != LC_MESSAGES) {
1255
1256 save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES,
1257 NULL)));
1258 if (! save_messages_locale) {
1259 goto cant_use_messages;
1260 }
1261
1262 if (strEQ(save_messages_locale, save_input_locale)) {
1263 Safefree(save_input_locale);
1264 }
1265 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
1266 Safefree(save_messages_locale);
1267 goto cant_use_messages;
1268 }
1269 }
1270
1271 /* Here the current LC_MESSAGES is set to the locale of the category
1272 * whose information is desired. Look through all the messages */
1273
1274 for (e = 0;
1275#ifdef HAS_SYS_ERRLIST
1276 e <= sys_nerr
1277#endif
1278 ; e++)
1279 {
1280 const U8* const errmsg = (U8 *) Strerror(e) ;
1281 if (!errmsg)
1282 break;
1283 if (! is_utf8_string(errmsg, 0)) {
1284 failures++;
1285 break;
1286 }
1287 else if (! is_ascii_string(errmsg, 0)) {
1288 non_ascii++;
1289 }
1290 }
1291
1292 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1293 if (save_messages_locale) {
1294 setlocale(LC_MESSAGES, save_messages_locale);
1295 Safefree(save_messages_locale);
1296 }
1297
1298 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1299 * any non-ascii means it is one; otherwise we assume it isn't */
1300 return (failures) ? FALSE : non_ascii;
1301
1302 }
1303 cant_use_messages:
1304
1305#endif
1306
69014004
KW
1307 DEBUG_L(PerlIO_printf(Perl_debug_log,
1308 "Assuming locale %s is not a UTF-8 locale\n",
1309 save_input_locale));
fa9b773e 1310 Safefree(save_input_locale);
7d74bb61
KW
1311 return FALSE;
1312}
1313
8ef6e574 1314#endif
7d74bb61 1315
66610fdd
RGS
1316/*
1317 * Local variables:
1318 * c-indentation-style: bsd
1319 * c-basic-offset: 4
14d04a33 1320 * indent-tabs-mode: nil
66610fdd
RGS
1321 * End:
1322 *
14d04a33 1323 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1324 */