This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
numeric.c: White-space only
[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)
c1284011 112 && _is_cur_LC_category_utf8(LC_NUMERIC))
28acfe03
KW
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
c1284011 257 PL_in_utf8_CTYPE_locale = _is_cur_LC_category_utf8(LC_CTYPE);
31f05a37
KW
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
481465ea 351 bool override_LC_ALL = 0;
89f7b9aa
KW
352 char * result;
353
b385bb4d
KW
354 if (locale && strEQ(locale, "")) {
355# ifdef LC_ALL
356 locale = PerlEnv_getenv("LC_ALL");
357 if (! locale) {
358#endif
359 switch (category) {
360# ifdef LC_ALL
361 case LC_ALL:
481465ea 362 override_LC_ALL = TRUE;
b385bb4d
KW
363 break; /* We already know its variable isn't set */
364# endif
365# ifdef USE_LOCALE_TIME
366 case LC_TIME:
367 locale = PerlEnv_getenv("LC_TIME");
368 break;
369# endif
370# ifdef USE_LOCALE_CTYPE
371 case LC_CTYPE:
372 locale = PerlEnv_getenv("LC_CTYPE");
373 break;
374# endif
375# ifdef USE_LOCALE_COLLATE
376 case LC_COLLATE:
377 locale = PerlEnv_getenv("LC_COLLATE");
378 break;
379# endif
380# ifdef USE_LOCALE_MONETARY
381 case LC_MONETARY:
382 locale = PerlEnv_getenv("LC_MONETARY");
383 break;
384# endif
385# ifdef USE_LOCALE_NUMERIC
386 case LC_NUMERIC:
387 locale = PerlEnv_getenv("LC_NUMERIC");
388 break;
389# endif
390# ifdef USE_LOCALE_MESSAGES
391 case LC_MESSAGES:
392 locale = PerlEnv_getenv("LC_MESSAGES");
393 break;
394# endif
395 default:
396 /* This is a category, like PAPER_SIZE that we don't
397 * know about; and so can't provide a wrapper. */
398 break;
399 }
400 if (! locale) {
401 locale = PerlEnv_getenv("LANG");
481465ea 402 if (! locale) {
b385bb4d
KW
403 locale = "";
404 }
405 }
406# ifdef LC_ALL
407 }
408# endif
409 }
410
89f7b9aa
KW
411 result = setlocale(category, locale);
412
481465ea 413 if (! override_LC_ALL) {
89f7b9aa
KW
414 return result;
415 }
416
417 /* Here the input locale was LC_ALL, and we have set it to what is in the
481465ea
KW
418 * LANG variable or the system default if there is no LANG. But these have
419 * lower priority than the other LC_foo variables, so override it for each
420 * one that is set. (If they are set to "", it means to use the same thing
421 * we just set LC_ALL to, so can skip) */
89f7b9aa
KW
422# ifdef USE_LOCALE_TIME
423 result = PerlEnv_getenv("LC_TIME");
730252b2 424 if (result && strNE(result, "")) {
89f7b9aa
KW
425 setlocale(LC_TIME, result);
426 }
427# endif
428# ifdef USE_LOCALE_CTYPE
429 result = PerlEnv_getenv("LC_CTYPE");
730252b2 430 if (result && strNE(result, "")) {
89f7b9aa
KW
431 setlocale(LC_CTYPE, result);
432 }
433# endif
434# ifdef USE_LOCALE_COLLATE
435 result = PerlEnv_getenv("LC_COLLATE");
730252b2 436 if (result && strNE(result, "")) {
89f7b9aa
KW
437 setlocale(LC_COLLATE, result);
438 }
439# endif
440# ifdef USE_LOCALE_MONETARY
441 result = PerlEnv_getenv("LC_MONETARY");
730252b2 442 if (result && strNE(result, "")) {
89f7b9aa
KW
443 setlocale(LC_MONETARY, result);
444 }
445# endif
446# ifdef USE_LOCALE_NUMERIC
447 result = PerlEnv_getenv("LC_NUMERIC");
730252b2 448 if (result && strNE(result, "")) {
89f7b9aa
KW
449 setlocale(LC_NUMERIC, result);
450 }
451# endif
452# ifdef USE_LOCALE_MESSAGES
453 result = PerlEnv_getenv("LC_MESSAGES");
730252b2 454 if (result && strNE(result, "")) {
89f7b9aa
KW
455 setlocale(LC_MESSAGES, result);
456 }
457# endif
458
459 return setlocale(LC_ALL, NULL);
460
b385bb4d
KW
461}
462
463#endif
464
465
98994639
HS
466/*
467 * Initialize locale awareness.
468 */
469int
470Perl_init_i18nl10n(pTHX_ int printwarn)
471{
0e92a118
KW
472 /* printwarn is
473 *
474 * 0 if not to output warning when setup locale is bad
475 * 1 if to output warning based on value of PERL_BADLANG
476 * >1 if to output regardless of PERL_BADLANG
477 *
478 * returns
98994639 479 * 1 = set ok or not applicable,
0e92a118
KW
480 * 0 = fallback to a locale of lower priority
481 * -1 = fallback to all locales failed, not even to the C locale
98994639
HS
482 */
483
0e92a118
KW
484 int ok = 1;
485
98994639 486#if defined(USE_LOCALE)
97aff369 487 dVAR;
98994639
HS
488
489#ifdef USE_LOCALE_CTYPE
490 char *curctype = NULL;
491#endif /* USE_LOCALE_CTYPE */
492#ifdef USE_LOCALE_COLLATE
493 char *curcoll = NULL;
494#endif /* USE_LOCALE_COLLATE */
495#ifdef USE_LOCALE_NUMERIC
496 char *curnum = NULL;
497#endif /* USE_LOCALE_NUMERIC */
498#ifdef __GLIBC__
7452cf6a 499 char * const language = PerlEnv_getenv("LANGUAGE");
98994639 500#endif
65ebb059 501
ccd65d51
KW
502 /* NULL uses the existing already set up locale */
503 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
504 ? NULL
505 : "";
c3fcd832
KW
506 const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
507 unsigned int trial_locales_count;
7452cf6a
AL
508 char * const lc_all = PerlEnv_getenv("LC_ALL");
509 char * const lang = PerlEnv_getenv("LANG");
98994639 510 bool setlocale_failure = FALSE;
65ebb059
KW
511 unsigned int i;
512 char *p;
513 const bool locwarn = (printwarn > 1 ||
514 (printwarn &&
515 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
0e92a118 516 bool done = FALSE;
6bce99ee
JH
517#ifdef WIN32
518 /* In some systems you can find out the system default locale
519 * and use that as the fallback locale. */
520# define SYSTEM_DEFAULT_LOCALE
521#endif
522#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059 523 const char *system_default_locale = NULL;
6bce99ee 524#endif
98994639 525
0e92a118
KW
526#ifndef LOCALE_ENVIRON_REQUIRED
527 PERL_UNUSED_VAR(done);
528#else
98994639
HS
529
530 /*
531 * Ultrix setlocale(..., "") fails if there are no environment
532 * variables from which to get a locale name.
533 */
534
b3e384bf 535# ifdef LC_ALL
98994639 536 if (lang) {
b385bb4d 537 if (my_setlocale(LC_ALL, setlocale_init))
98994639
HS
538 done = TRUE;
539 else
540 setlocale_failure = TRUE;
541 }
542 if (!setlocale_failure) {
b3e384bf 543# ifdef USE_LOCALE_CTYPE
56279a21 544 Safefree(curctype);
98994639 545 if (! (curctype =
b385bb4d 546 my_setlocale(LC_CTYPE,
98994639 547 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
ccd65d51 548 ? setlocale_init : NULL)))
98994639
HS
549 setlocale_failure = TRUE;
550 else
551 curctype = savepv(curctype);
b3e384bf
KW
552# endif /* USE_LOCALE_CTYPE */
553# ifdef USE_LOCALE_COLLATE
56279a21 554 Safefree(curcoll);
98994639 555 if (! (curcoll =
b385bb4d 556 my_setlocale(LC_COLLATE,
98994639 557 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
ccd65d51 558 ? setlocale_init : NULL)))
98994639
HS
559 setlocale_failure = TRUE;
560 else
561 curcoll = savepv(curcoll);
b3e384bf
KW
562# endif /* USE_LOCALE_COLLATE */
563# ifdef USE_LOCALE_NUMERIC
56279a21 564 Safefree(curnum);
98994639 565 if (! (curnum =
b385bb4d 566 my_setlocale(LC_NUMERIC,
98994639 567 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
ccd65d51 568 ? setlocale_init : NULL)))
98994639
HS
569 setlocale_failure = TRUE;
570 else
571 curnum = savepv(curnum);
b3e384bf 572# endif /* USE_LOCALE_NUMERIC */
a782673d
KW
573# ifdef USE_LOCALE_MESSAGES
574 if (! my_setlocale(LC_MESSAGES,
575 (!done && (lang || PerlEnv_getenv("LC_MESSAGES")))
576 ? setlocale_init : NULL))
577 {
578 setlocale_failure = TRUE;
579 }
580# endif /* USE_LOCALE_MESSAGES */
c835d6be
KW
581# ifdef USE_LOCALE_MONETARY
582 if (! my_setlocale(LC_MONETARY,
583 (!done && (lang || PerlEnv_getenv("LC_MONETARY")))
584 ? setlocale_init : NULL))
585 {
586 setlocale_failure = TRUE;
587 }
588# endif /* USE_LOCALE_MONETARY */
98994639
HS
589 }
590
b3e384bf 591# endif /* LC_ALL */
98994639
HS
592
593#endif /* !LOCALE_ENVIRON_REQUIRED */
594
65ebb059
KW
595 /* We try each locale in the list until we get one that works, or exhaust
596 * the list */
c3fcd832
KW
597 trial_locales[0] = setlocale_init;
598 trial_locales_count = 1;
65ebb059
KW
599 for (i= 0; i < trial_locales_count; i++) {
600 const char * trial_locale = trial_locales[i];
601
602 if (i > 0) {
603
604 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
605 * when i==0, but I (khw) don't think that behavior makes much
606 * sense */
607 setlocale_failure = FALSE;
608
6bce99ee
JH
609#ifdef SYSTEM_DEFAULT_LOCALE
610# ifdef WIN32
65ebb059
KW
611 /* On Windows machines, an entry of "" after the 0th means to use
612 * the system default locale, which we now proceed to get. */
613 if (strEQ(trial_locale, "")) {
614 unsigned int j;
615
616 /* Note that this may change the locale, but we are going to do
617 * that anyway just below */
618 system_default_locale = setlocale(LC_ALL, "");
619
620 /* Skip if invalid or it's already on the list of locales to
621 * try */
622 if (! system_default_locale) {
623 goto next_iteration;
624 }
625 for (j = 0; j < trial_locales_count; j++) {
626 if (strEQ(system_default_locale, trial_locales[j])) {
627 goto next_iteration;
628 }
629 }
630
631 trial_locale = system_default_locale;
632 }
6bce99ee
JH
633# endif /* WIN32 */
634#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
635 }
636
98994639 637#ifdef LC_ALL
7cd8b568 638 if (! my_setlocale(LC_ALL, trial_locale)) {
49c85077 639 setlocale_failure = TRUE;
7cd8b568
KW
640 }
641 else {
642 /* Since LC_ALL succeeded, it should have changed all the other
643 * categories it can to its value; so we massage things so that the
644 * setlocales below just return their category's current values.
645 * This adequately handles the case in NetBSD where LC_COLLATE may
646 * not be defined for a locale, and setting it individually will
647 * fail, whereas setting LC_ALL suceeds, leaving LC_COLLATE set to
648 * the POSIX locale. */
649 trial_locale = NULL;
650 }
98994639
HS
651#endif /* LC_ALL */
652
49c85077 653 if (!setlocale_failure) {
98994639 654#ifdef USE_LOCALE_CTYPE
49c85077
KW
655 Safefree(curctype);
656 if (! (curctype = my_setlocale(LC_CTYPE, trial_locale)))
657 setlocale_failure = TRUE;
658 else
659 curctype = savepv(curctype);
98994639
HS
660#endif /* USE_LOCALE_CTYPE */
661#ifdef USE_LOCALE_COLLATE
49c85077
KW
662 Safefree(curcoll);
663 if (! (curcoll = my_setlocale(LC_COLLATE, trial_locale)))
664 setlocale_failure = TRUE;
665 else
666 curcoll = savepv(curcoll);
98994639
HS
667#endif /* USE_LOCALE_COLLATE */
668#ifdef USE_LOCALE_NUMERIC
49c85077
KW
669 Safefree(curnum);
670 if (! (curnum = my_setlocale(LC_NUMERIC, trial_locale)))
671 setlocale_failure = TRUE;
672 else
673 curnum = savepv(curnum);
98994639 674#endif /* USE_LOCALE_NUMERIC */
a782673d
KW
675#ifdef USE_LOCALE_MESSAGES
676 if (! (my_setlocale(LC_MESSAGES, trial_locale)))
677 setlocale_failure = TRUE;
678#endif /* USE_LOCALE_MESSAGES */
c835d6be
KW
679#ifdef USE_LOCALE_MONETARY
680 if (! (my_setlocale(LC_MONETARY, trial_locale)))
681 setlocale_failure = TRUE;
682#endif /* USE_LOCALE_MONETARY */
683
49c85077
KW
684 if (! setlocale_failure) { /* Success */
685 break;
686 }
65ebb059 687 }
98994639 688
49c85077
KW
689 /* Here, something failed; will need to try a fallback. */
690 ok = 0;
65ebb059 691
49c85077
KW
692 if (i == 0) {
693 unsigned int j;
98994639 694
65ebb059 695 if (locwarn) { /* Output failure info only on the first one */
98994639
HS
696#ifdef LC_ALL
697
49c85077
KW
698 PerlIO_printf(Perl_error_log,
699 "perl: warning: Setting locale failed.\n");
98994639
HS
700
701#else /* !LC_ALL */
702
49c85077
KW
703 PerlIO_printf(Perl_error_log,
704 "perl: warning: Setting locale failed for the categories:\n\t");
98994639 705#ifdef USE_LOCALE_CTYPE
49c85077
KW
706 if (! curctype)
707 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
98994639
HS
708#endif /* USE_LOCALE_CTYPE */
709#ifdef USE_LOCALE_COLLATE
49c85077
KW
710 if (! curcoll)
711 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
98994639
HS
712#endif /* USE_LOCALE_COLLATE */
713#ifdef USE_LOCALE_NUMERIC
49c85077
KW
714 if (! curnum)
715 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
98994639 716#endif /* USE_LOCALE_NUMERIC */
a782673d 717 PerlIO_printf(Perl_error_log, "and possibly others\n");
98994639
HS
718
719#endif /* LC_ALL */
720
49c85077
KW
721 PerlIO_printf(Perl_error_log,
722 "perl: warning: Please check that your locale settings:\n");
98994639
HS
723
724#ifdef __GLIBC__
49c85077
KW
725 PerlIO_printf(Perl_error_log,
726 "\tLANGUAGE = %c%s%c,\n",
727 language ? '"' : '(',
728 language ? language : "unset",
729 language ? '"' : ')');
98994639
HS
730#endif
731
49c85077
KW
732 PerlIO_printf(Perl_error_log,
733 "\tLC_ALL = %c%s%c,\n",
734 lc_all ? '"' : '(',
735 lc_all ? lc_all : "unset",
736 lc_all ? '"' : ')');
98994639
HS
737
738#if defined(USE_ENVIRON_ARRAY)
49c85077
KW
739 {
740 char **e;
741 for (e = environ; *e; e++) {
742 if (strnEQ(*e, "LC_", 3)
743 && strnNE(*e, "LC_ALL=", 7)
744 && (p = strchr(*e, '=')))
745 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
746 (int)(p - *e), *e, p + 1);
747 }
748 }
98994639 749#else
49c85077
KW
750 PerlIO_printf(Perl_error_log,
751 "\t(possibly more locale environment variables)\n");
98994639
HS
752#endif
753
49c85077
KW
754 PerlIO_printf(Perl_error_log,
755 "\tLANG = %c%s%c\n",
756 lang ? '"' : '(',
757 lang ? lang : "unset",
758 lang ? '"' : ')');
98994639 759
49c85077
KW
760 PerlIO_printf(Perl_error_log,
761 " are supported and installed on your system.\n");
762 }
98994639 763
65ebb059
KW
764 /* Calculate what fallback locales to try. We have avoided this
765 * until we have to, becuase failure is quite unlikely. This will
766 * usually change the upper bound of the loop we are in.
767 *
768 * Since the system's default way of setting the locale has not
769 * found one that works, We use Perl's defined ordering: LC_ALL,
770 * LANG, and the C locale. We don't try the same locale twice, so
771 * don't add to the list if already there. (On POSIX systems, the
772 * LC_ALL element will likely be a repeat of the 0th element "",
773 * but there's no harm done by doing it explicitly */
774 if (lc_all) {
775 for (j = 0; j < trial_locales_count; j++) {
776 if (strEQ(lc_all, trial_locales[j])) {
777 goto done_lc_all;
778 }
779 }
780 trial_locales[trial_locales_count++] = lc_all;
781 }
782 done_lc_all:
98994639 783
65ebb059
KW
784 if (lang) {
785 for (j = 0; j < trial_locales_count; j++) {
786 if (strEQ(lang, trial_locales[j])) {
787 goto done_lang;
788 }
789 }
790 trial_locales[trial_locales_count++] = lang;
791 }
792 done_lang:
793
794#if defined(WIN32) && defined(LC_ALL)
795 /* For Windows, we also try the system default locale before "C".
796 * (If there exists a Windows without LC_ALL we skip this because
797 * it gets too complicated. For those, the "C" is the next
798 * fallback possibility). The "" is the same as the 0th element of
799 * the array, but the code at the loop above knows to treat it
800 * differently when not the 0th */
801 trial_locales[trial_locales_count++] = "";
802#endif
803
804 for (j = 0; j < trial_locales_count; j++) {
805 if (strEQ("C", trial_locales[j])) {
806 goto done_C;
807 }
808 }
809 trial_locales[trial_locales_count++] = "C";
98994639 810
65ebb059
KW
811 done_C: ;
812 } /* end of first time through the loop */
98994639 813
65ebb059
KW
814#ifdef WIN32
815 next_iteration: ;
816#endif
817
818 } /* end of looping through the trial locales */
819
820 if (ok < 1) { /* If we tried to fallback */
821 const char* msg;
822 if (! setlocale_failure) { /* fallback succeeded */
823 msg = "Falling back to";
824 }
825 else { /* fallback failed */
98994639 826
65ebb059
KW
827 /* We dropped off the end of the loop, so have to decrement i to
828 * get back to the value the last time through */
829 i--;
98994639 830
65ebb059
KW
831 ok = -1;
832 msg = "Failed to fall back to";
833
834 /* To continue, we should use whatever values we've got */
98994639 835#ifdef USE_LOCALE_CTYPE
49c85077
KW
836 Safefree(curctype);
837 curctype = savepv(setlocale(LC_CTYPE, NULL));
98994639
HS
838#endif /* USE_LOCALE_CTYPE */
839#ifdef USE_LOCALE_COLLATE
49c85077
KW
840 Safefree(curcoll);
841 curcoll = savepv(setlocale(LC_COLLATE, NULL));
98994639
HS
842#endif /* USE_LOCALE_COLLATE */
843#ifdef USE_LOCALE_NUMERIC
49c85077
KW
844 Safefree(curnum);
845 curnum = savepv(setlocale(LC_NUMERIC, NULL));
98994639 846#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
847 }
848
849 if (locwarn) {
850 const char * description;
851 const char * name = "";
852 if (strEQ(trial_locales[i], "C")) {
853 description = "the standard locale";
854 name = "C";
855 }
6bce99ee 856#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059
KW
857 else if (strEQ(trial_locales[i], "")) {
858 description = "the system default locale";
859 if (system_default_locale) {
860 name = system_default_locale;
861 }
862 }
6bce99ee 863#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
864 else {
865 description = "a fallback locale";
866 name = trial_locales[i];
867 }
868 if (name && strNE(name, "")) {
869 PerlIO_printf(Perl_error_log,
870 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
871 }
872 else {
873 PerlIO_printf(Perl_error_log,
874 "perl: warning: %s %s.\n", msg, description);
875 }
876 }
877 } /* End of tried to fallback */
98994639
HS
878
879#ifdef USE_LOCALE_CTYPE
880 new_ctype(curctype);
881#endif /* USE_LOCALE_CTYPE */
882
883#ifdef USE_LOCALE_COLLATE
884 new_collate(curcoll);
885#endif /* USE_LOCALE_COLLATE */
886
887#ifdef USE_LOCALE_NUMERIC
888 new_numeric(curnum);
889#endif /* USE_LOCALE_NUMERIC */
b310b053 890
8ef6e574 891#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
49c85077
KW
892 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
893 * locale is UTF-8. If PL_utf8locale and PL_unicode (set by -C or by
894 * $ENV{PERL_UNICODE}) are true, perl.c:S_parse_body() will turn on the
895 * PerlIO :utf8 layer on STDIN, STDOUT, STDERR, _and_ the default open
896 * discipline. */
c1284011 897 PL_utf8locale = _is_cur_LC_category_utf8(LC_CTYPE);
49c85077 898
a05d7ebb 899 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
900 This is an alternative to using the -C command line switch
901 (the -C if present will override this). */
902 {
dd374669 903 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 904 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
905 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
906 PL_utf8cache = -1;
b310b053 907 }
ec71e770 908#endif
b310b053 909
98994639 910#ifdef USE_LOCALE_CTYPE
43c5f42d 911 Safefree(curctype);
98994639
HS
912#endif /* USE_LOCALE_CTYPE */
913#ifdef USE_LOCALE_COLLATE
43c5f42d 914 Safefree(curcoll);
98994639
HS
915#endif /* USE_LOCALE_COLLATE */
916#ifdef USE_LOCALE_NUMERIC
43c5f42d 917 Safefree(curnum);
98994639 918#endif /* USE_LOCALE_NUMERIC */
8ef6e574
KW
919
920#endif /* USE_LOCALE */
921
98994639
HS
922 return ok;
923}
924
49c85077 925
98994639
HS
926#ifdef USE_LOCALE_COLLATE
927
928/*
929 * mem_collxfrm() is a bit like strxfrm() but with two important
930 * differences. First, it handles embedded NULs. Second, it allocates
931 * a bit more memory than needed for the transformed data itself.
932 * The real transformed data begins at offset sizeof(collationix).
933 * Please see sv_collxfrm() to see how this is used.
934 */
166f8a29 935
98994639
HS
936char *
937Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
938{
97aff369 939 dVAR;
98994639
HS
940 char *xbuf;
941 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
942
7918f24d
NC
943 PERL_ARGS_ASSERT_MEM_COLLXFRM;
944
98994639
HS
945 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
946 /* the +1 is for the terminating NUL. */
947
948 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 949 Newx(xbuf, xAlloc, char);
98994639
HS
950 if (! xbuf)
951 goto bad;
952
953 *(U32*)xbuf = PL_collation_ix;
954 xout = sizeof(PL_collation_ix);
955 for (xin = 0; xin < len; ) {
224e8ef5 956 Size_t xused;
98994639
HS
957
958 for (;;) {
959 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 960 if (xused >= PERL_INT_MAX)
98994639 961 goto bad;
eb160463 962 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
963 break;
964 xAlloc = (2 * xAlloc) + 1;
965 Renew(xbuf, xAlloc, char);
966 if (! xbuf)
967 goto bad;
968 }
969
970 xin += strlen(s + xin) + 1;
971 xout += xused;
972
973 /* Embedded NULs are understood but silently skipped
974 * because they make no sense in locale collation. */
975 }
976
977 xbuf[xout] = '\0';
978 *xlen = xout - sizeof(PL_collation_ix);
979 return xbuf;
980
981 bad:
982 Safefree(xbuf);
983 *xlen = 0;
984 return NULL;
985}
986
987#endif /* USE_LOCALE_COLLATE */
988
8ef6e574
KW
989#ifdef USE_LOCALE
990
c1284011
KW
991bool
992Perl__is_cur_LC_category_utf8(pTHX_ int category)
7d74bb61
KW
993{
994 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
995 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b
KW
996 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
997 * could give the wrong result. It errs on the side of not being a UTF-8
998 * locale. */
7d74bb61
KW
999
1000 char *save_input_locale = NULL;
7d74bb61
KW
1001 STRLEN final_pos;
1002
8ef6e574 1003#ifdef LC_ALL
7d74bb61 1004 assert(category != LC_ALL);
8ef6e574 1005#endif
7d74bb61
KW
1006
1007 /* First dispose of the trivial cases */
b07fffd1 1008 save_input_locale = setlocale(category, NULL);
7d74bb61 1009 if (! save_input_locale) {
69014004
KW
1010 DEBUG_L(PerlIO_printf(Perl_debug_log,
1011 "Could not find current locale for category %d\n",
1012 category));
7d74bb61
KW
1013 return FALSE; /* XXX maybe should croak */
1014 }
b07fffd1 1015 save_input_locale = stdize_locale(savepv(save_input_locale));
7d74bb61
KW
1016 if ((*save_input_locale == 'C' && save_input_locale[1] == '\0')
1017 || strEQ(save_input_locale, "POSIX"))
1018 {
69014004
KW
1019 DEBUG_L(PerlIO_printf(Perl_debug_log,
1020 "Current locale for category %d is %s\n",
1021 category, save_input_locale));
b07fffd1 1022 Safefree(save_input_locale);
7d74bb61
KW
1023 return FALSE;
1024 }
1025
1d958db2
KW
1026#if defined(USE_LOCALE_CTYPE) \
1027 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 1028
1d958db2 1029 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
1030
1031 char *save_ctype_locale = NULL;
119ee68b 1032 bool is_utf8;
7d74bb61 1033
119ee68b 1034 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
1035
1036 /* Get the current LC_CTYPE locale */
1037 save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL)));
1038 if (! save_ctype_locale) {
69014004
KW
1039 DEBUG_L(PerlIO_printf(Perl_debug_log,
1040 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
1041 goto cant_use_nllanginfo;
1042 }
1043
1044 /* If LC_CTYPE and the desired category use the same locale, this
1045 * means that finding the value for LC_CTYPE is the same as finding
1046 * the value for the desired category. Otherwise, switch LC_CTYPE
1047 * to the desired category's locale */
1048 if (strEQ(save_ctype_locale, save_input_locale)) {
1049 Safefree(save_ctype_locale);
1050 save_ctype_locale = NULL;
1051 }
1052 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
1053 DEBUG_L(PerlIO_printf(Perl_debug_log,
1054 "Could not change LC_CTYPE locale to %s\n",
1055 save_input_locale));
7d74bb61
KW
1056 Safefree(save_ctype_locale);
1057 goto cant_use_nllanginfo;
1058 }
1059 }
1060
69014004
KW
1061 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
1062 save_input_locale));
1063
7d74bb61 1064 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
1065 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
1066 * should give the correct results */
119ee68b 1067
1d958db2
KW
1068# if defined(HAS_NL_LANGINFO) && defined(CODESET)
1069 {
1070 char *codeset = savepv(nl_langinfo(CODESET));
1071 if (codeset && strNE(codeset, "")) {
119ee68b 1072
1d958db2
KW
1073 /* If we switched LC_CTYPE, switch back */
1074 if (save_ctype_locale) {
1075 setlocale(LC_CTYPE, save_ctype_locale);
1076 Safefree(save_ctype_locale);
1077 }
1078
1079 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
1080 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
1081
69014004
KW
1082 DEBUG_L(PerlIO_printf(Perl_debug_log,
1083 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
1084 codeset, is_utf8));
1d958db2
KW
1085 Safefree(codeset);
1086 Safefree(save_input_locale);
1087 return is_utf8;
1088 }
0ac78434 1089 Safefree(codeset);
119ee68b
KW
1090 }
1091
1d958db2
KW
1092# endif
1093# ifdef MB_CUR_MAX
1094
1095 /* Here, either we don't have nl_langinfo, or it didn't return a
1096 * codeset. Try MB_CUR_MAX */
1097
119ee68b
KW
1098 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
1099 * Unicode code point. Since UTF-8 is the only non-single byte
1100 * encoding we handle, we just say any such encoding is UTF-8, and if
1101 * turns out to be wrong, other things will fail */
1102 is_utf8 = MB_CUR_MAX >= 4;
1103
69014004
KW
1104 DEBUG_L(PerlIO_printf(Perl_debug_log,
1105 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
1106 (int) MB_CUR_MAX, is_utf8));
1107
119ee68b
KW
1108 Safefree(save_input_locale);
1109
1110# ifdef HAS_MBTOWC
1111
1112 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
1113 * since they are both in the C99 standard. We can feed a known byte
1114 * string to the latter function, and check that it gives the expected
1115 * result */
1116 if (is_utf8) {
1117 wchar_t wc;
a6715020 1118 GCC_DIAG_IGNORE(-Wunused-result);
1d958db2 1119 (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */
a6715020 1120 GCC_DIAG_RESTORE;
69014004 1121 errno = 0;
f019f68f 1122 if ((size_t)mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
119ee68b
KW
1123 != strlen(HYPHEN_UTF8)
1124 || wc != (wchar_t) 0x2010)
1125 {
1126 is_utf8 = FALSE;
69014004
KW
1127 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", wc));
1128 DEBUG_L(PerlIO_printf(Perl_debug_log,
1129 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
1130 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
1131 }
1132 }
119ee68b
KW
1133# endif
1134
1d958db2
KW
1135 /* If we switched LC_CTYPE, switch back */
1136 if (save_ctype_locale) {
1137 setlocale(LC_CTYPE, save_ctype_locale);
1138 Safefree(save_ctype_locale);
119ee68b 1139 }
7d74bb61 1140
1d958db2 1141 return is_utf8;
119ee68b 1142# endif
7d74bb61 1143 }
119ee68b 1144
7d74bb61
KW
1145 cant_use_nllanginfo:
1146
1147#endif /* HAS_NL_LANGINFO etc */
1148
1149 /* nl_langinfo not available or failed somehow. Look at the locale name to
751e9426 1150 * see if it matches qr/UTF -? 8 /ix */
7d74bb61
KW
1151
1152 final_pos = strlen(save_input_locale) - 1;
751e9426
KW
1153 if (final_pos >= 3) {
1154 char *name = save_input_locale;
1155
1156 /* Find next 'U' or 'u' and look from there */
1157 while ((name += strcspn(name, "Uu") + 1)
1158 <= save_input_locale + final_pos - 2)
7d74bb61 1159 {
751e9426
KW
1160 if (toFOLD(*(name)) != 't'
1161 || toFOLD(*(name + 1)) != 'f')
1162 {
1163 continue;
1164 }
1165 name += 2;
1166 if (*(name) == '-') {
1167 if ((name > save_input_locale + final_pos - 1)) {
1168 break;
1169 }
1170 name++;
1171 }
1172 if (*(name) == '8') {
1173 Safefree(save_input_locale);
69014004
KW
1174 DEBUG_L(PerlIO_printf(Perl_debug_log,
1175 "Locale %s ends with UTF-8 in name\n",
1176 save_input_locale));
751e9426
KW
1177 return TRUE;
1178 }
7d74bb61 1179 }
69014004
KW
1180 DEBUG_L(PerlIO_printf(Perl_debug_log,
1181 "Locale %s doesn't end with UTF-8 in name\n",
1182 save_input_locale));
7d74bb61
KW
1183 }
1184
1185#ifdef WIN32
1186 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
1187 if (final_pos >= 4
1188 && *(save_input_locale + final_pos - 0) == '1'
1189 && *(save_input_locale + final_pos - 1) == '0'
1190 && *(save_input_locale + final_pos - 2) == '0'
1191 && *(save_input_locale + final_pos - 3) == '5'
1192 && *(save_input_locale + final_pos - 4) == '6')
1193 {
69014004
KW
1194 DEBUG_L(PerlIO_printf(Perl_debug_log,
1195 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
1196 save_input_locale));
d37662c0 1197 Safefree(save_input_locale);
7d74bb61
KW
1198 return TRUE;
1199 }
1200#endif
1201
fa9b773e
KW
1202 /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
1203 if (instr(save_input_locale, "8859")) {
69014004
KW
1204 DEBUG_L(PerlIO_printf(Perl_debug_log,
1205 "Locale %s has 8859 in name, not UTF-8 locale\n",
1206 save_input_locale));
d37662c0 1207 Safefree(save_input_locale);
fa9b773e
KW
1208 return FALSE;
1209 }
1210
1211#ifdef HAS_LOCALECONV
1212
1213# ifdef USE_LOCALE_MONETARY
1214
1215 /* Here, there is nothing in the locale name to indicate whether the locale
1216 * is UTF-8 or not. This "name", the return of setlocale(), is actually
1217 * defined to be opaque, so we can't really rely on the absence of various
1218 * substrings in the name to indicate its UTF-8ness. Look at the locale's
1219 * currency symbol. Often that will be in the native script, and if the
1220 * symbol isn't in UTF-8, we know that the locale isn't. If it is
1221 * non-ASCII UTF-8, we infer that the locale is too.
1222 * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the
1223 * locale of the desired category, if it isn't that locale already */
1224
1225 {
1226 char *save_monetary_locale = NULL;
1227 bool illegal_utf8 = FALSE;
1228 bool only_ascii = FALSE;
1229 const struct lconv* const lc = localeconv();
1230
1231 if (category != LC_MONETARY) {
1232
1233 save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY,
1234 NULL)));
1235 if (! save_monetary_locale) {
69014004
KW
1236 DEBUG_L(PerlIO_printf(Perl_debug_log,
1237 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1238 goto cant_use_monetary;
1239 }
1240
1241 if (strNE(save_monetary_locale, save_input_locale)) {
1242 if (! setlocale(LC_MONETARY, save_input_locale)) {
69014004
KW
1243 DEBUG_L(PerlIO_printf(Perl_debug_log,
1244 "Could not change LC_MONETARY locale to %s\n",
1245 save_input_locale));
fa9b773e
KW
1246 Safefree(save_monetary_locale);
1247 goto cant_use_monetary;
1248 }
1249 }
1250 }
1251
1252 /* Here the current LC_MONETARY is set to the locale of the category
1253 * whose information is desired. */
1254
1255 if (lc && lc->currency_symbol) {
1256 if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) {
69014004
KW
1257 DEBUG_L(PerlIO_printf(Perl_debug_log,
1258 "Currency symbol for %s is not legal UTF-8\n",
1259 save_input_locale));
fa9b773e
KW
1260 illegal_utf8 = TRUE;
1261 }
1262 else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) {
69014004 1263 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
1264 only_ascii = TRUE;
1265 }
1266 }
1267
1268 /* If we changed it, restore LC_MONETARY to its original locale */
1269 if (save_monetary_locale) {
1270 setlocale(LC_MONETARY, save_monetary_locale);
1271 Safefree(save_monetary_locale);
1272 }
1273
1274 Safefree(save_input_locale);
1275
1276 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise
1277 * assume the locale is UTF-8 if and only if the symbol is non-ascii
1278 * UTF-8. (We can't really tell if the locale is UTF-8 or not if the
1279 * symbol is just a '$', so we err on the side of it not being UTF-8)
1280 * */
69014004
KW
1281 DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", (illegal_utf8)
1282 ? FALSE
1283 : ! only_ascii));
fa9b773e
KW
1284 return (illegal_utf8)
1285 ? FALSE
1286 : ! only_ascii;
1287
1288 }
1289 cant_use_monetary:
1290
1291# endif /* USE_LOCALE_MONETARY */
1292#endif /* HAS_LOCALECONV */
1293
1294#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES)
1295
1296/* This code is ifdefd out because it was found to not be necessary in testing
1297 * on our dromedary test machine, which has over 700 locales. There, looking
1298 * at just the currency symbol gave essentially the same results as doing this
1299 * extra work. Executing this also caused segfaults in miniperl. I left it in
1300 * so as to avoid rewriting it if real-world experience indicates that
1301 * dromedary is an outlier. Essentially, instead of returning abpve if we
1302 * haven't found illegal utf8, we continue on and examine all the strerror()
1303 * messages on the platform for utf8ness. If all are ASCII, we still don't
1304 * know the answer; but otherwise we have a pretty good indication of the
1305 * utf8ness. The reason this doesn't necessarily help much is that the
1306 * messages may not have been translated into the locale. The currency symbol
1307 * is much more likely to have been translated. The code below would need to
1308 * be altered somewhat to just be a continuation of testing the currency
1309 * symbol. */
1310 int e;
1311 unsigned int failures = 0, non_ascii = 0;
1312 char *save_messages_locale = NULL;
1313
1314 /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the
1315 * desired category, if it isn't that locale already */
1316
1317 if (category != LC_MESSAGES) {
1318
1319 save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES,
1320 NULL)));
1321 if (! save_messages_locale) {
1322 goto cant_use_messages;
1323 }
1324
1325 if (strEQ(save_messages_locale, save_input_locale)) {
1326 Safefree(save_input_locale);
1327 }
1328 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
1329 Safefree(save_messages_locale);
1330 goto cant_use_messages;
1331 }
1332 }
1333
1334 /* Here the current LC_MESSAGES is set to the locale of the category
1335 * whose information is desired. Look through all the messages */
1336
1337 for (e = 0;
1338#ifdef HAS_SYS_ERRLIST
1339 e <= sys_nerr
1340#endif
1341 ; e++)
1342 {
1343 const U8* const errmsg = (U8 *) Strerror(e) ;
1344 if (!errmsg)
1345 break;
1346 if (! is_utf8_string(errmsg, 0)) {
1347 failures++;
1348 break;
1349 }
1350 else if (! is_ascii_string(errmsg, 0)) {
1351 non_ascii++;
1352 }
1353 }
1354
1355 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1356 if (save_messages_locale) {
1357 setlocale(LC_MESSAGES, save_messages_locale);
1358 Safefree(save_messages_locale);
1359 }
1360
1361 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1362 * any non-ascii means it is one; otherwise we assume it isn't */
1363 return (failures) ? FALSE : non_ascii;
1364
1365 }
1366 cant_use_messages:
1367
1368#endif
1369
69014004
KW
1370 DEBUG_L(PerlIO_printf(Perl_debug_log,
1371 "Assuming locale %s is not a UTF-8 locale\n",
1372 save_input_locale));
fa9b773e 1373 Safefree(save_input_locale);
7d74bb61
KW
1374 return FALSE;
1375}
1376
8ef6e574 1377#endif
7d74bb61 1378
66610fdd
RGS
1379/*
1380 * Local variables:
1381 * c-indentation-style: bsd
1382 * c-basic-offset: 4
14d04a33 1383 * indent-tabs-mode: nil
66610fdd
RGS
1384 * End:
1385 *
14d04a33 1386 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1387 */