This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Improve fallback during locale initialization
[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");
65ebb059
KW
447#else
448 const char * const language = NULL;
98994639 449#endif
65ebb059 450
ccd65d51
KW
451 /* NULL uses the existing already set up locale */
452 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
453 ? NULL
454 : "";
65ebb059
KW
455 const char* trial_locales[5] = { setlocale_init }; /* 5 = 1 each for "",
456 LC_ALL, LANG, "", C
457 */
458 unsigned int trial_locales_count = 1;
7452cf6a
AL
459 char * const lc_all = PerlEnv_getenv("LC_ALL");
460 char * const lang = PerlEnv_getenv("LANG");
98994639 461 bool setlocale_failure = FALSE;
65ebb059
KW
462 unsigned int i;
463 char *p;
464 const bool locwarn = (printwarn > 1 ||
465 (printwarn &&
466 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
0e92a118 467 bool done = FALSE;
65ebb059
KW
468 const char *description;
469 const char *system_default_locale = NULL;
0e92a118 470
98994639 471
0e92a118
KW
472#ifndef LOCALE_ENVIRON_REQUIRED
473 PERL_UNUSED_VAR(done);
474#else
98994639
HS
475
476 /*
477 * Ultrix setlocale(..., "") fails if there are no environment
478 * variables from which to get a locale name.
479 */
480
b3e384bf 481# ifdef LC_ALL
98994639 482 if (lang) {
b385bb4d 483 if (my_setlocale(LC_ALL, setlocale_init))
98994639
HS
484 done = TRUE;
485 else
486 setlocale_failure = TRUE;
487 }
488 if (!setlocale_failure) {
b3e384bf 489# ifdef USE_LOCALE_CTYPE
56279a21 490 Safefree(curctype);
98994639 491 if (! (curctype =
b385bb4d 492 my_setlocale(LC_CTYPE,
98994639 493 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
ccd65d51 494 ? setlocale_init : NULL)))
98994639
HS
495 setlocale_failure = TRUE;
496 else
497 curctype = savepv(curctype);
b3e384bf
KW
498# endif /* USE_LOCALE_CTYPE */
499# ifdef USE_LOCALE_COLLATE
56279a21 500 Safefree(curcoll);
98994639 501 if (! (curcoll =
b385bb4d 502 my_setlocale(LC_COLLATE,
98994639 503 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
ccd65d51 504 ? setlocale_init : NULL)))
98994639
HS
505 setlocale_failure = TRUE;
506 else
507 curcoll = savepv(curcoll);
b3e384bf
KW
508# endif /* USE_LOCALE_COLLATE */
509# ifdef USE_LOCALE_NUMERIC
56279a21 510 Safefree(curnum);
98994639 511 if (! (curnum =
b385bb4d 512 my_setlocale(LC_NUMERIC,
98994639 513 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
ccd65d51 514 ? setlocale_init : NULL)))
98994639
HS
515 setlocale_failure = TRUE;
516 else
517 curnum = savepv(curnum);
b3e384bf 518# endif /* USE_LOCALE_NUMERIC */
98994639
HS
519 }
520
b3e384bf 521# endif /* LC_ALL */
98994639
HS
522
523#endif /* !LOCALE_ENVIRON_REQUIRED */
524
65ebb059
KW
525 /* We try each locale in the list until we get one that works, or exhaust
526 * the list */
527 for (i= 0; i < trial_locales_count; i++) {
528 const char * trial_locale = trial_locales[i];
529
530 if (i > 0) {
531
532 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
533 * when i==0, but I (khw) don't think that behavior makes much
534 * sense */
535 setlocale_failure = FALSE;
536
537#ifdef WIN32
538
539 /* On Windows machines, an entry of "" after the 0th means to use
540 * the system default locale, which we now proceed to get. */
541 if (strEQ(trial_locale, "")) {
542 unsigned int j;
543
544 /* Note that this may change the locale, but we are going to do
545 * that anyway just below */
546 system_default_locale = setlocale(LC_ALL, "");
547
548 /* Skip if invalid or it's already on the list of locales to
549 * try */
550 if (! system_default_locale) {
551 goto next_iteration;
552 }
553 for (j = 0; j < trial_locales_count; j++) {
554 if (strEQ(system_default_locale, trial_locales[j])) {
555 goto next_iteration;
556 }
557 }
558
559 trial_locale = system_default_locale;
560 }
561#endif
562 }
563
98994639 564#ifdef LC_ALL
65ebb059 565 if (! my_setlocale(LC_ALL, trial_locale))
98994639
HS
566 setlocale_failure = TRUE;
567#endif /* LC_ALL */
568
569 if (!setlocale_failure) {
570#ifdef USE_LOCALE_CTYPE
6cb43dbf 571 Safefree(curctype);
65ebb059 572 if (! (curctype = my_setlocale(LC_CTYPE, trial_locale)))
98994639
HS
573 setlocale_failure = TRUE;
574 else
575 curctype = savepv(curctype);
576#endif /* USE_LOCALE_CTYPE */
577#ifdef USE_LOCALE_COLLATE
6cb43dbf 578 Safefree(curcoll);
65ebb059 579 if (! (curcoll = my_setlocale(LC_COLLATE, trial_locale)))
98994639
HS
580 setlocale_failure = TRUE;
581 else
582 curcoll = savepv(curcoll);
583#endif /* USE_LOCALE_COLLATE */
584#ifdef USE_LOCALE_NUMERIC
6cb43dbf 585 Safefree(curnum);
65ebb059 586 if (! (curnum = my_setlocale(LC_NUMERIC, trial_locale)))
98994639
HS
587 setlocale_failure = TRUE;
588 else
589 curnum = savepv(curnum);
590#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
591
592 if (! setlocale_failure) { /* Success */
593 break;
594 }
98994639
HS
595 }
596
65ebb059
KW
597 /* Here, something failed; will need to try a fallback. */
598 ok = 0;
599
600 if (i == 0) {
601 unsigned int j;
98994639 602
65ebb059 603 if (locwarn) { /* Output failure info only on the first one */
98994639
HS
604#ifdef LC_ALL
605
606 PerlIO_printf(Perl_error_log,
607 "perl: warning: Setting locale failed.\n");
608
609#else /* !LC_ALL */
610
611 PerlIO_printf(Perl_error_log,
612 "perl: warning: Setting locale failed for the categories:\n\t");
613#ifdef USE_LOCALE_CTYPE
614 if (! curctype)
615 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
616#endif /* USE_LOCALE_CTYPE */
617#ifdef USE_LOCALE_COLLATE
618 if (! curcoll)
619 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
620#endif /* USE_LOCALE_COLLATE */
621#ifdef USE_LOCALE_NUMERIC
622 if (! curnum)
623 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
624#endif /* USE_LOCALE_NUMERIC */
625 PerlIO_printf(Perl_error_log, "\n");
626
627#endif /* LC_ALL */
628
629 PerlIO_printf(Perl_error_log,
630 "perl: warning: Please check that your locale settings:\n");
631
632#ifdef __GLIBC__
633 PerlIO_printf(Perl_error_log,
634 "\tLANGUAGE = %c%s%c,\n",
635 language ? '"' : '(',
636 language ? language : "unset",
637 language ? '"' : ')');
638#endif
639
640 PerlIO_printf(Perl_error_log,
641 "\tLC_ALL = %c%s%c,\n",
642 lc_all ? '"' : '(',
643 lc_all ? lc_all : "unset",
644 lc_all ? '"' : ')');
645
646#if defined(USE_ENVIRON_ARRAY)
647 {
648 char **e;
649 for (e = environ; *e; e++) {
650 if (strnEQ(*e, "LC_", 3)
651 && strnNE(*e, "LC_ALL=", 7)
652 && (p = strchr(*e, '=')))
653 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
654 (int)(p - *e), *e, p + 1);
655 }
656 }
657#else
658 PerlIO_printf(Perl_error_log,
659 "\t(possibly more locale environment variables)\n");
660#endif
661
662 PerlIO_printf(Perl_error_log,
663 "\tLANG = %c%s%c\n",
664 lang ? '"' : '(',
665 lang ? lang : "unset",
666 lang ? '"' : ')');
667
668 PerlIO_printf(Perl_error_log,
669 " are supported and installed on your system.\n");
670 }
671
65ebb059
KW
672 /* Calculate what fallback locales to try. We have avoided this
673 * until we have to, becuase failure is quite unlikely. This will
674 * usually change the upper bound of the loop we are in.
675 *
676 * Since the system's default way of setting the locale has not
677 * found one that works, We use Perl's defined ordering: LC_ALL,
678 * LANG, and the C locale. We don't try the same locale twice, so
679 * don't add to the list if already there. (On POSIX systems, the
680 * LC_ALL element will likely be a repeat of the 0th element "",
681 * but there's no harm done by doing it explicitly */
682 if (lc_all) {
683 for (j = 0; j < trial_locales_count; j++) {
684 if (strEQ(lc_all, trial_locales[j])) {
685 goto done_lc_all;
686 }
687 }
688 trial_locales[trial_locales_count++] = lc_all;
689 }
690 done_lc_all:
98994639 691
65ebb059
KW
692 if (lang) {
693 for (j = 0; j < trial_locales_count; j++) {
694 if (strEQ(lang, trial_locales[j])) {
695 goto done_lang;
696 }
697 }
698 trial_locales[trial_locales_count++] = lang;
699 }
700 done_lang:
701
702#if defined(WIN32) && defined(LC_ALL)
703 /* For Windows, we also try the system default locale before "C".
704 * (If there exists a Windows without LC_ALL we skip this because
705 * it gets too complicated. For those, the "C" is the next
706 * fallback possibility). The "" is the same as the 0th element of
707 * the array, but the code at the loop above knows to treat it
708 * differently when not the 0th */
709 trial_locales[trial_locales_count++] = "";
710#endif
711
712 for (j = 0; j < trial_locales_count; j++) {
713 if (strEQ("C", trial_locales[j])) {
714 goto done_C;
715 }
716 }
717 trial_locales[trial_locales_count++] = "C";
98994639 718
65ebb059
KW
719 done_C: ;
720 } /* end of first time through the loop */
98994639 721
65ebb059
KW
722#ifdef WIN32
723 next_iteration: ;
724#endif
725
726 } /* end of looping through the trial locales */
727
728 if (ok < 1) { /* If we tried to fallback */
729 const char* msg;
730 if (! setlocale_failure) { /* fallback succeeded */
731 msg = "Falling back to";
732 }
733 else { /* fallback failed */
98994639 734
65ebb059
KW
735 /* We dropped off the end of the loop, so have to decrement i to
736 * get back to the value the last time through */
737 i--;
98994639 738
65ebb059
KW
739 ok = -1;
740 msg = "Failed to fall back to";
741
742 /* To continue, we should use whatever values we've got */
98994639 743#ifdef USE_LOCALE_CTYPE
56279a21 744 Safefree(curctype);
bd61b366 745 curctype = savepv(setlocale(LC_CTYPE, NULL));
98994639
HS
746#endif /* USE_LOCALE_CTYPE */
747#ifdef USE_LOCALE_COLLATE
56279a21 748 Safefree(curcoll);
bd61b366 749 curcoll = savepv(setlocale(LC_COLLATE, NULL));
98994639
HS
750#endif /* USE_LOCALE_COLLATE */
751#ifdef USE_LOCALE_NUMERIC
56279a21 752 Safefree(curnum);
bd61b366 753 curnum = savepv(setlocale(LC_NUMERIC, NULL));
98994639 754#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
755 }
756
757 if (locwarn) {
758 const char * description;
759 const char * name = "";
760 if (strEQ(trial_locales[i], "C")) {
761 description = "the standard locale";
762 name = "C";
763 }
764 else if (strEQ(trial_locales[i], "")) {
765 description = "the system default locale";
766 if (system_default_locale) {
767 name = system_default_locale;
768 }
769 }
770 else {
771 description = "a fallback locale";
772 name = trial_locales[i];
773 }
774 if (name && strNE(name, "")) {
775 PerlIO_printf(Perl_error_log,
776 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
777 }
778 else {
779 PerlIO_printf(Perl_error_log,
780 "perl: warning: %s %s.\n", msg, description);
781 }
782 }
783 } /* End of tried to fallback */
98994639
HS
784
785#ifdef USE_LOCALE_CTYPE
786 new_ctype(curctype);
787#endif /* USE_LOCALE_CTYPE */
788
789#ifdef USE_LOCALE_COLLATE
790 new_collate(curcoll);
791#endif /* USE_LOCALE_COLLATE */
792
793#ifdef USE_LOCALE_NUMERIC
794 new_numeric(curnum);
795#endif /* USE_LOCALE_NUMERIC */
b310b053 796
8ef6e574 797#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
b310b053 798 {
fde18df1 799 /* Set PL_utf8locale to TRUE if using PerlIO _and_
7d74bb61 800 the current LC_CTYPE locale is UTF-8.
8aa8f774 801 If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
a05d7ebb 802 are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
fde18df1 803 on STDIN, STDOUT, STDERR, _and_ the default open discipline.
085a54d9 804 */
7d74bb61 805 PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE);
fde18df1 806 }
a05d7ebb 807 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
808 This is an alternative to using the -C command line switch
809 (the -C if present will override this). */
810 {
dd374669 811 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 812 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
813 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
814 PL_utf8cache = -1;
b310b053 815 }
ec71e770 816#endif
b310b053 817
98994639 818#ifdef USE_LOCALE_CTYPE
43c5f42d 819 Safefree(curctype);
98994639
HS
820#endif /* USE_LOCALE_CTYPE */
821#ifdef USE_LOCALE_COLLATE
43c5f42d 822 Safefree(curcoll);
98994639
HS
823#endif /* USE_LOCALE_COLLATE */
824#ifdef USE_LOCALE_NUMERIC
43c5f42d 825 Safefree(curnum);
98994639 826#endif /* USE_LOCALE_NUMERIC */
8ef6e574
KW
827
828#endif /* USE_LOCALE */
829
98994639
HS
830 return ok;
831}
832
98994639
HS
833#ifdef USE_LOCALE_COLLATE
834
835/*
836 * mem_collxfrm() is a bit like strxfrm() but with two important
837 * differences. First, it handles embedded NULs. Second, it allocates
838 * a bit more memory than needed for the transformed data itself.
839 * The real transformed data begins at offset sizeof(collationix).
840 * Please see sv_collxfrm() to see how this is used.
841 */
166f8a29 842
98994639
HS
843char *
844Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
845{
97aff369 846 dVAR;
98994639
HS
847 char *xbuf;
848 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
849
7918f24d
NC
850 PERL_ARGS_ASSERT_MEM_COLLXFRM;
851
98994639
HS
852 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
853 /* the +1 is for the terminating NUL. */
854
855 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 856 Newx(xbuf, xAlloc, char);
98994639
HS
857 if (! xbuf)
858 goto bad;
859
860 *(U32*)xbuf = PL_collation_ix;
861 xout = sizeof(PL_collation_ix);
862 for (xin = 0; xin < len; ) {
224e8ef5 863 Size_t xused;
98994639
HS
864
865 for (;;) {
866 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 867 if (xused >= PERL_INT_MAX)
98994639 868 goto bad;
eb160463 869 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
870 break;
871 xAlloc = (2 * xAlloc) + 1;
872 Renew(xbuf, xAlloc, char);
873 if (! xbuf)
874 goto bad;
875 }
876
877 xin += strlen(s + xin) + 1;
878 xout += xused;
879
880 /* Embedded NULs are understood but silently skipped
881 * because they make no sense in locale collation. */
882 }
883
884 xbuf[xout] = '\0';
885 *xlen = xout - sizeof(PL_collation_ix);
886 return xbuf;
887
888 bad:
889 Safefree(xbuf);
890 *xlen = 0;
891 return NULL;
892}
893
894#endif /* USE_LOCALE_COLLATE */
895
8ef6e574
KW
896#ifdef USE_LOCALE
897
637917bb 898STATIC bool
7d74bb61
KW
899S_is_cur_LC_category_utf8(pTHX_ int category)
900{
901 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
902 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b
KW
903 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
904 * could give the wrong result. It errs on the side of not being a UTF-8
905 * locale. */
7d74bb61
KW
906
907 char *save_input_locale = NULL;
7d74bb61
KW
908 STRLEN final_pos;
909
8ef6e574 910#ifdef LC_ALL
7d74bb61 911 assert(category != LC_ALL);
8ef6e574 912#endif
7d74bb61
KW
913
914 /* First dispose of the trivial cases */
b07fffd1 915 save_input_locale = setlocale(category, NULL);
7d74bb61 916 if (! save_input_locale) {
69014004
KW
917 DEBUG_L(PerlIO_printf(Perl_debug_log,
918 "Could not find current locale for category %d\n",
919 category));
7d74bb61
KW
920 return FALSE; /* XXX maybe should croak */
921 }
b07fffd1 922 save_input_locale = stdize_locale(savepv(save_input_locale));
7d74bb61
KW
923 if ((*save_input_locale == 'C' && save_input_locale[1] == '\0')
924 || strEQ(save_input_locale, "POSIX"))
925 {
69014004
KW
926 DEBUG_L(PerlIO_printf(Perl_debug_log,
927 "Current locale for category %d is %s\n",
928 category, save_input_locale));
b07fffd1 929 Safefree(save_input_locale);
7d74bb61
KW
930 return FALSE;
931 }
932
1d958db2
KW
933#if defined(USE_LOCALE_CTYPE) \
934 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 935
1d958db2 936 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
937
938 char *save_ctype_locale = NULL;
119ee68b 939 bool is_utf8;
7d74bb61 940
119ee68b 941 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
942
943 /* Get the current LC_CTYPE locale */
944 save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL)));
945 if (! save_ctype_locale) {
69014004
KW
946 DEBUG_L(PerlIO_printf(Perl_debug_log,
947 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
948 goto cant_use_nllanginfo;
949 }
950
951 /* If LC_CTYPE and the desired category use the same locale, this
952 * means that finding the value for LC_CTYPE is the same as finding
953 * the value for the desired category. Otherwise, switch LC_CTYPE
954 * to the desired category's locale */
955 if (strEQ(save_ctype_locale, save_input_locale)) {
956 Safefree(save_ctype_locale);
957 save_ctype_locale = NULL;
958 }
959 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
960 DEBUG_L(PerlIO_printf(Perl_debug_log,
961 "Could not change LC_CTYPE locale to %s\n",
962 save_input_locale));
7d74bb61
KW
963 Safefree(save_ctype_locale);
964 goto cant_use_nllanginfo;
965 }
966 }
967
69014004
KW
968 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
969 save_input_locale));
970
7d74bb61 971 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
972 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
973 * should give the correct results */
119ee68b 974
1d958db2
KW
975# if defined(HAS_NL_LANGINFO) && defined(CODESET)
976 {
977 char *codeset = savepv(nl_langinfo(CODESET));
978 if (codeset && strNE(codeset, "")) {
119ee68b 979
1d958db2
KW
980 /* If we switched LC_CTYPE, switch back */
981 if (save_ctype_locale) {
982 setlocale(LC_CTYPE, save_ctype_locale);
983 Safefree(save_ctype_locale);
984 }
985
986 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
987 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
988
69014004
KW
989 DEBUG_L(PerlIO_printf(Perl_debug_log,
990 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
991 codeset, is_utf8));
1d958db2
KW
992 Safefree(codeset);
993 Safefree(save_input_locale);
994 return is_utf8;
995 }
119ee68b
KW
996 }
997
1d958db2
KW
998# endif
999# ifdef MB_CUR_MAX
1000
1001 /* Here, either we don't have nl_langinfo, or it didn't return a
1002 * codeset. Try MB_CUR_MAX */
1003
119ee68b
KW
1004 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
1005 * Unicode code point. Since UTF-8 is the only non-single byte
1006 * encoding we handle, we just say any such encoding is UTF-8, and if
1007 * turns out to be wrong, other things will fail */
1008 is_utf8 = MB_CUR_MAX >= 4;
1009
69014004
KW
1010 DEBUG_L(PerlIO_printf(Perl_debug_log,
1011 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
1012 (int) MB_CUR_MAX, is_utf8));
1013
119ee68b
KW
1014 Safefree(save_input_locale);
1015
1016# ifdef HAS_MBTOWC
1017
1018 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
1019 * since they are both in the C99 standard. We can feed a known byte
1020 * string to the latter function, and check that it gives the expected
1021 * result */
1022 if (is_utf8) {
1023 wchar_t wc;
a6715020 1024 GCC_DIAG_IGNORE(-Wunused-result);
1d958db2 1025 (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */
a6715020 1026 GCC_DIAG_RESTORE;
69014004 1027 errno = 0;
119ee68b
KW
1028 if (mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
1029 != strlen(HYPHEN_UTF8)
1030 || wc != (wchar_t) 0x2010)
1031 {
1032 is_utf8 = FALSE;
69014004
KW
1033 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", wc));
1034 DEBUG_L(PerlIO_printf(Perl_debug_log,
1035 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
1036 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
1037 }
1038 }
119ee68b
KW
1039# endif
1040
1d958db2
KW
1041 /* If we switched LC_CTYPE, switch back */
1042 if (save_ctype_locale) {
1043 setlocale(LC_CTYPE, save_ctype_locale);
1044 Safefree(save_ctype_locale);
119ee68b 1045 }
7d74bb61 1046
1d958db2 1047 return is_utf8;
119ee68b 1048# endif
7d74bb61 1049 }
119ee68b 1050
7d74bb61
KW
1051 cant_use_nllanginfo:
1052
1053#endif /* HAS_NL_LANGINFO etc */
1054
1055 /* nl_langinfo not available or failed somehow. Look at the locale name to
751e9426 1056 * see if it matches qr/UTF -? 8 /ix */
7d74bb61
KW
1057
1058 final_pos = strlen(save_input_locale) - 1;
751e9426
KW
1059 if (final_pos >= 3) {
1060 char *name = save_input_locale;
1061
1062 /* Find next 'U' or 'u' and look from there */
1063 while ((name += strcspn(name, "Uu") + 1)
1064 <= save_input_locale + final_pos - 2)
7d74bb61 1065 {
751e9426
KW
1066 if (toFOLD(*(name)) != 't'
1067 || toFOLD(*(name + 1)) != 'f')
1068 {
1069 continue;
1070 }
1071 name += 2;
1072 if (*(name) == '-') {
1073 if ((name > save_input_locale + final_pos - 1)) {
1074 break;
1075 }
1076 name++;
1077 }
1078 if (*(name) == '8') {
1079 Safefree(save_input_locale);
69014004
KW
1080 DEBUG_L(PerlIO_printf(Perl_debug_log,
1081 "Locale %s ends with UTF-8 in name\n",
1082 save_input_locale));
751e9426
KW
1083 return TRUE;
1084 }
7d74bb61 1085 }
69014004
KW
1086 DEBUG_L(PerlIO_printf(Perl_debug_log,
1087 "Locale %s doesn't end with UTF-8 in name\n",
1088 save_input_locale));
7d74bb61
KW
1089 }
1090
1091#ifdef WIN32
1092 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
1093 if (final_pos >= 4
1094 && *(save_input_locale + final_pos - 0) == '1'
1095 && *(save_input_locale + final_pos - 1) == '0'
1096 && *(save_input_locale + final_pos - 2) == '0'
1097 && *(save_input_locale + final_pos - 3) == '5'
1098 && *(save_input_locale + final_pos - 4) == '6')
1099 {
1100 Safefree(save_input_locale);
69014004
KW
1101 DEBUG_L(PerlIO_printf(Perl_debug_log,
1102 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
1103 save_input_locale));
7d74bb61
KW
1104 return TRUE;
1105 }
1106#endif
1107
fa9b773e
KW
1108 /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
1109 if (instr(save_input_locale, "8859")) {
1110 Safefree(save_input_locale);
69014004
KW
1111 DEBUG_L(PerlIO_printf(Perl_debug_log,
1112 "Locale %s has 8859 in name, not UTF-8 locale\n",
1113 save_input_locale));
fa9b773e
KW
1114 return FALSE;
1115 }
1116
1117#ifdef HAS_LOCALECONV
1118
1119# ifdef USE_LOCALE_MONETARY
1120
1121 /* Here, there is nothing in the locale name to indicate whether the locale
1122 * is UTF-8 or not. This "name", the return of setlocale(), is actually
1123 * defined to be opaque, so we can't really rely on the absence of various
1124 * substrings in the name to indicate its UTF-8ness. Look at the locale's
1125 * currency symbol. Often that will be in the native script, and if the
1126 * symbol isn't in UTF-8, we know that the locale isn't. If it is
1127 * non-ASCII UTF-8, we infer that the locale is too.
1128 * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the
1129 * locale of the desired category, if it isn't that locale already */
1130
1131 {
1132 char *save_monetary_locale = NULL;
1133 bool illegal_utf8 = FALSE;
1134 bool only_ascii = FALSE;
1135 const struct lconv* const lc = localeconv();
1136
1137 if (category != LC_MONETARY) {
1138
1139 save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY,
1140 NULL)));
1141 if (! save_monetary_locale) {
69014004
KW
1142 DEBUG_L(PerlIO_printf(Perl_debug_log,
1143 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1144 goto cant_use_monetary;
1145 }
1146
1147 if (strNE(save_monetary_locale, save_input_locale)) {
1148 if (! setlocale(LC_MONETARY, save_input_locale)) {
69014004
KW
1149 DEBUG_L(PerlIO_printf(Perl_debug_log,
1150 "Could not change LC_MONETARY locale to %s\n",
1151 save_input_locale));
fa9b773e
KW
1152 Safefree(save_monetary_locale);
1153 goto cant_use_monetary;
1154 }
1155 }
1156 }
1157
1158 /* Here the current LC_MONETARY is set to the locale of the category
1159 * whose information is desired. */
1160
1161 if (lc && lc->currency_symbol) {
1162 if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) {
69014004
KW
1163 DEBUG_L(PerlIO_printf(Perl_debug_log,
1164 "Currency symbol for %s is not legal UTF-8\n",
1165 save_input_locale));
fa9b773e
KW
1166 illegal_utf8 = TRUE;
1167 }
1168 else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) {
69014004 1169 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
1170 only_ascii = TRUE;
1171 }
1172 }
1173
1174 /* If we changed it, restore LC_MONETARY to its original locale */
1175 if (save_monetary_locale) {
1176 setlocale(LC_MONETARY, save_monetary_locale);
1177 Safefree(save_monetary_locale);
1178 }
1179
1180 Safefree(save_input_locale);
1181
1182 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise
1183 * assume the locale is UTF-8 if and only if the symbol is non-ascii
1184 * UTF-8. (We can't really tell if the locale is UTF-8 or not if the
1185 * symbol is just a '$', so we err on the side of it not being UTF-8)
1186 * */
69014004
KW
1187 DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", (illegal_utf8)
1188 ? FALSE
1189 : ! only_ascii));
fa9b773e
KW
1190 return (illegal_utf8)
1191 ? FALSE
1192 : ! only_ascii;
1193
1194 }
1195 cant_use_monetary:
1196
1197# endif /* USE_LOCALE_MONETARY */
1198#endif /* HAS_LOCALECONV */
1199
1200#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES)
1201
1202/* This code is ifdefd out because it was found to not be necessary in testing
1203 * on our dromedary test machine, which has over 700 locales. There, looking
1204 * at just the currency symbol gave essentially the same results as doing this
1205 * extra work. Executing this also caused segfaults in miniperl. I left it in
1206 * so as to avoid rewriting it if real-world experience indicates that
1207 * dromedary is an outlier. Essentially, instead of returning abpve if we
1208 * haven't found illegal utf8, we continue on and examine all the strerror()
1209 * messages on the platform for utf8ness. If all are ASCII, we still don't
1210 * know the answer; but otherwise we have a pretty good indication of the
1211 * utf8ness. The reason this doesn't necessarily help much is that the
1212 * messages may not have been translated into the locale. The currency symbol
1213 * is much more likely to have been translated. The code below would need to
1214 * be altered somewhat to just be a continuation of testing the currency
1215 * symbol. */
1216 int e;
1217 unsigned int failures = 0, non_ascii = 0;
1218 char *save_messages_locale = NULL;
1219
1220 /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the
1221 * desired category, if it isn't that locale already */
1222
1223 if (category != LC_MESSAGES) {
1224
1225 save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES,
1226 NULL)));
1227 if (! save_messages_locale) {
1228 goto cant_use_messages;
1229 }
1230
1231 if (strEQ(save_messages_locale, save_input_locale)) {
1232 Safefree(save_input_locale);
1233 }
1234 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
1235 Safefree(save_messages_locale);
1236 goto cant_use_messages;
1237 }
1238 }
1239
1240 /* Here the current LC_MESSAGES is set to the locale of the category
1241 * whose information is desired. Look through all the messages */
1242
1243 for (e = 0;
1244#ifdef HAS_SYS_ERRLIST
1245 e <= sys_nerr
1246#endif
1247 ; e++)
1248 {
1249 const U8* const errmsg = (U8 *) Strerror(e) ;
1250 if (!errmsg)
1251 break;
1252 if (! is_utf8_string(errmsg, 0)) {
1253 failures++;
1254 break;
1255 }
1256 else if (! is_ascii_string(errmsg, 0)) {
1257 non_ascii++;
1258 }
1259 }
1260
1261 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1262 if (save_messages_locale) {
1263 setlocale(LC_MESSAGES, save_messages_locale);
1264 Safefree(save_messages_locale);
1265 }
1266
1267 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1268 * any non-ascii means it is one; otherwise we assume it isn't */
1269 return (failures) ? FALSE : non_ascii;
1270
1271 }
1272 cant_use_messages:
1273
1274#endif
1275
69014004
KW
1276 DEBUG_L(PerlIO_printf(Perl_debug_log,
1277 "Assuming locale %s is not a UTF-8 locale\n",
1278 save_input_locale));
fa9b773e 1279 Safefree(save_input_locale);
7d74bb61
KW
1280 return FALSE;
1281}
1282
8ef6e574 1283#endif
7d74bb61 1284
66610fdd
RGS
1285/*
1286 * Local variables:
1287 * c-indentation-style: bsd
1288 * c-basic-offset: 4
14d04a33 1289 * indent-tabs-mode: nil
66610fdd
RGS
1290 * End:
1291 *
14d04a33 1292 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1293 */