This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Emulate POSIX locale setting on Windows
[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{
419 int ok = 1;
420 /* returns
421 * 1 = set ok or not applicable,
422 * 0 = fallback to C locale,
423 * -1 = fallback to C locale failed
424 */
425
426#if defined(USE_LOCALE)
97aff369 427 dVAR;
98994639
HS
428
429#ifdef USE_LOCALE_CTYPE
430 char *curctype = NULL;
431#endif /* USE_LOCALE_CTYPE */
432#ifdef USE_LOCALE_COLLATE
433 char *curcoll = NULL;
434#endif /* USE_LOCALE_COLLATE */
435#ifdef USE_LOCALE_NUMERIC
436 char *curnum = NULL;
437#endif /* USE_LOCALE_NUMERIC */
438#ifdef __GLIBC__
7452cf6a 439 char * const language = PerlEnv_getenv("LANGUAGE");
98994639 440#endif
ccd65d51
KW
441 /* NULL uses the existing already set up locale */
442 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
443 ? NULL
444 : "";
7452cf6a
AL
445 char * const lc_all = PerlEnv_getenv("LC_ALL");
446 char * const lang = PerlEnv_getenv("LANG");
98994639
HS
447 bool setlocale_failure = FALSE;
448
449#ifdef LOCALE_ENVIRON_REQUIRED
450
451 /*
452 * Ultrix setlocale(..., "") fails if there are no environment
453 * variables from which to get a locale name.
454 */
455
456 bool done = FALSE;
457
b3e384bf 458# ifdef LC_ALL
98994639 459 if (lang) {
b385bb4d 460 if (my_setlocale(LC_ALL, setlocale_init))
98994639
HS
461 done = TRUE;
462 else
463 setlocale_failure = TRUE;
464 }
465 if (!setlocale_failure) {
b3e384bf 466# ifdef USE_LOCALE_CTYPE
56279a21 467 Safefree(curctype);
98994639 468 if (! (curctype =
b385bb4d 469 my_setlocale(LC_CTYPE,
98994639 470 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
ccd65d51 471 ? setlocale_init : NULL)))
98994639
HS
472 setlocale_failure = TRUE;
473 else
474 curctype = savepv(curctype);
b3e384bf
KW
475# endif /* USE_LOCALE_CTYPE */
476# ifdef USE_LOCALE_COLLATE
56279a21 477 Safefree(curcoll);
98994639 478 if (! (curcoll =
b385bb4d 479 my_setlocale(LC_COLLATE,
98994639 480 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
ccd65d51 481 ? setlocale_init : NULL)))
98994639
HS
482 setlocale_failure = TRUE;
483 else
484 curcoll = savepv(curcoll);
b3e384bf
KW
485# endif /* USE_LOCALE_COLLATE */
486# ifdef USE_LOCALE_NUMERIC
56279a21 487 Safefree(curnum);
98994639 488 if (! (curnum =
b385bb4d 489 my_setlocale(LC_NUMERIC,
98994639 490 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
ccd65d51 491 ? setlocale_init : NULL)))
98994639
HS
492 setlocale_failure = TRUE;
493 else
494 curnum = savepv(curnum);
b3e384bf 495# endif /* USE_LOCALE_NUMERIC */
98994639
HS
496 }
497
b3e384bf 498# endif /* LC_ALL */
98994639
HS
499
500#endif /* !LOCALE_ENVIRON_REQUIRED */
501
502#ifdef LC_ALL
b385bb4d 503 if (! my_setlocale(LC_ALL, setlocale_init))
98994639
HS
504 setlocale_failure = TRUE;
505#endif /* LC_ALL */
506
507 if (!setlocale_failure) {
508#ifdef USE_LOCALE_CTYPE
6cb43dbf 509 Safefree(curctype);
b385bb4d 510 if (! (curctype = my_setlocale(LC_CTYPE, setlocale_init)))
98994639
HS
511 setlocale_failure = TRUE;
512 else
513 curctype = savepv(curctype);
514#endif /* USE_LOCALE_CTYPE */
515#ifdef USE_LOCALE_COLLATE
6cb43dbf 516 Safefree(curcoll);
b385bb4d 517 if (! (curcoll = my_setlocale(LC_COLLATE, setlocale_init)))
98994639
HS
518 setlocale_failure = TRUE;
519 else
520 curcoll = savepv(curcoll);
521#endif /* USE_LOCALE_COLLATE */
522#ifdef USE_LOCALE_NUMERIC
6cb43dbf 523 Safefree(curnum);
b385bb4d 524 if (! (curnum = my_setlocale(LC_NUMERIC, setlocale_init)))
98994639
HS
525 setlocale_failure = TRUE;
526 else
527 curnum = savepv(curnum);
528#endif /* USE_LOCALE_NUMERIC */
529 }
530
531 if (setlocale_failure) {
532 char *p;
0bd48802 533 const bool locwarn = (printwarn > 1 ||
98994639
HS
534 (printwarn &&
535 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
536
537 if (locwarn) {
538#ifdef LC_ALL
539
540 PerlIO_printf(Perl_error_log,
541 "perl: warning: Setting locale failed.\n");
542
543#else /* !LC_ALL */
544
545 PerlIO_printf(Perl_error_log,
546 "perl: warning: Setting locale failed for the categories:\n\t");
547#ifdef USE_LOCALE_CTYPE
548 if (! curctype)
549 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
550#endif /* USE_LOCALE_CTYPE */
551#ifdef USE_LOCALE_COLLATE
552 if (! curcoll)
553 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
554#endif /* USE_LOCALE_COLLATE */
555#ifdef USE_LOCALE_NUMERIC
556 if (! curnum)
557 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
558#endif /* USE_LOCALE_NUMERIC */
559 PerlIO_printf(Perl_error_log, "\n");
560
561#endif /* LC_ALL */
562
563 PerlIO_printf(Perl_error_log,
564 "perl: warning: Please check that your locale settings:\n");
565
566#ifdef __GLIBC__
567 PerlIO_printf(Perl_error_log,
568 "\tLANGUAGE = %c%s%c,\n",
569 language ? '"' : '(',
570 language ? language : "unset",
571 language ? '"' : ')');
572#endif
573
574 PerlIO_printf(Perl_error_log,
575 "\tLC_ALL = %c%s%c,\n",
576 lc_all ? '"' : '(',
577 lc_all ? lc_all : "unset",
578 lc_all ? '"' : ')');
579
580#if defined(USE_ENVIRON_ARRAY)
581 {
582 char **e;
583 for (e = environ; *e; e++) {
584 if (strnEQ(*e, "LC_", 3)
585 && strnNE(*e, "LC_ALL=", 7)
586 && (p = strchr(*e, '=')))
587 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
588 (int)(p - *e), *e, p + 1);
589 }
590 }
591#else
592 PerlIO_printf(Perl_error_log,
593 "\t(possibly more locale environment variables)\n");
594#endif
595
596 PerlIO_printf(Perl_error_log,
597 "\tLANG = %c%s%c\n",
598 lang ? '"' : '(',
599 lang ? lang : "unset",
600 lang ? '"' : ')');
601
602 PerlIO_printf(Perl_error_log,
603 " are supported and installed on your system.\n");
604 }
605
606#ifdef LC_ALL
607
608 if (setlocale(LC_ALL, "C")) {
609 if (locwarn)
610 PerlIO_printf(Perl_error_log,
611 "perl: warning: Falling back to the standard locale (\"C\").\n");
612 ok = 0;
613 }
614 else {
615 if (locwarn)
616 PerlIO_printf(Perl_error_log,
617 "perl: warning: Failed to fall back to the standard locale (\"C\").\n");
618 ok = -1;
619 }
620
621#else /* ! LC_ALL */
622
623 if (0
624#ifdef USE_LOCALE_CTYPE
625 || !(curctype || setlocale(LC_CTYPE, "C"))
626#endif /* USE_LOCALE_CTYPE */
627#ifdef USE_LOCALE_COLLATE
628 || !(curcoll || setlocale(LC_COLLATE, "C"))
629#endif /* USE_LOCALE_COLLATE */
630#ifdef USE_LOCALE_NUMERIC
631 || !(curnum || setlocale(LC_NUMERIC, "C"))
632#endif /* USE_LOCALE_NUMERIC */
633 )
634 {
635 if (locwarn)
636 PerlIO_printf(Perl_error_log,
637 "perl: warning: Cannot fall back to the standard locale (\"C\").\n");
638 ok = -1;
639 }
640
641#endif /* ! LC_ALL */
642
643#ifdef USE_LOCALE_CTYPE
56279a21 644 Safefree(curctype);
bd61b366 645 curctype = savepv(setlocale(LC_CTYPE, NULL));
98994639
HS
646#endif /* USE_LOCALE_CTYPE */
647#ifdef USE_LOCALE_COLLATE
56279a21 648 Safefree(curcoll);
bd61b366 649 curcoll = savepv(setlocale(LC_COLLATE, NULL));
98994639
HS
650#endif /* USE_LOCALE_COLLATE */
651#ifdef USE_LOCALE_NUMERIC
56279a21 652 Safefree(curnum);
bd61b366 653 curnum = savepv(setlocale(LC_NUMERIC, NULL));
98994639
HS
654#endif /* USE_LOCALE_NUMERIC */
655 }
656 else {
657
658#ifdef USE_LOCALE_CTYPE
659 new_ctype(curctype);
660#endif /* USE_LOCALE_CTYPE */
661
662#ifdef USE_LOCALE_COLLATE
663 new_collate(curcoll);
664#endif /* USE_LOCALE_COLLATE */
665
666#ifdef USE_LOCALE_NUMERIC
667 new_numeric(curnum);
668#endif /* USE_LOCALE_NUMERIC */
b310b053 669
98994639
HS
670 }
671
8ef6e574 672#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
b310b053 673 {
fde18df1 674 /* Set PL_utf8locale to TRUE if using PerlIO _and_
7d74bb61 675 the current LC_CTYPE locale is UTF-8.
8aa8f774 676 If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
a05d7ebb 677 are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
fde18df1 678 on STDIN, STDOUT, STDERR, _and_ the default open discipline.
085a54d9 679 */
7d74bb61 680 PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE);
fde18df1 681 }
a05d7ebb 682 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
683 This is an alternative to using the -C command line switch
684 (the -C if present will override this). */
685 {
dd374669 686 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 687 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
688 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
689 PL_utf8cache = -1;
b310b053 690 }
ec71e770 691#endif
b310b053 692
98994639 693#ifdef USE_LOCALE_CTYPE
43c5f42d 694 Safefree(curctype);
98994639
HS
695#endif /* USE_LOCALE_CTYPE */
696#ifdef USE_LOCALE_COLLATE
43c5f42d 697 Safefree(curcoll);
98994639
HS
698#endif /* USE_LOCALE_COLLATE */
699#ifdef USE_LOCALE_NUMERIC
43c5f42d 700 Safefree(curnum);
98994639 701#endif /* USE_LOCALE_NUMERIC */
8ef6e574
KW
702
703#endif /* USE_LOCALE */
704
98994639
HS
705 return ok;
706}
707
98994639
HS
708#ifdef USE_LOCALE_COLLATE
709
710/*
711 * mem_collxfrm() is a bit like strxfrm() but with two important
712 * differences. First, it handles embedded NULs. Second, it allocates
713 * a bit more memory than needed for the transformed data itself.
714 * The real transformed data begins at offset sizeof(collationix).
715 * Please see sv_collxfrm() to see how this is used.
716 */
166f8a29 717
98994639
HS
718char *
719Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
720{
97aff369 721 dVAR;
98994639
HS
722 char *xbuf;
723 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
724
7918f24d
NC
725 PERL_ARGS_ASSERT_MEM_COLLXFRM;
726
98994639
HS
727 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
728 /* the +1 is for the terminating NUL. */
729
730 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 731 Newx(xbuf, xAlloc, char);
98994639
HS
732 if (! xbuf)
733 goto bad;
734
735 *(U32*)xbuf = PL_collation_ix;
736 xout = sizeof(PL_collation_ix);
737 for (xin = 0; xin < len; ) {
224e8ef5 738 Size_t xused;
98994639
HS
739
740 for (;;) {
741 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 742 if (xused >= PERL_INT_MAX)
98994639 743 goto bad;
eb160463 744 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
745 break;
746 xAlloc = (2 * xAlloc) + 1;
747 Renew(xbuf, xAlloc, char);
748 if (! xbuf)
749 goto bad;
750 }
751
752 xin += strlen(s + xin) + 1;
753 xout += xused;
754
755 /* Embedded NULs are understood but silently skipped
756 * because they make no sense in locale collation. */
757 }
758
759 xbuf[xout] = '\0';
760 *xlen = xout - sizeof(PL_collation_ix);
761 return xbuf;
762
763 bad:
764 Safefree(xbuf);
765 *xlen = 0;
766 return NULL;
767}
768
769#endif /* USE_LOCALE_COLLATE */
770
8ef6e574
KW
771#ifdef USE_LOCALE
772
637917bb 773STATIC bool
7d74bb61
KW
774S_is_cur_LC_category_utf8(pTHX_ int category)
775{
776 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
777 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b
KW
778 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
779 * could give the wrong result. It errs on the side of not being a UTF-8
780 * locale. */
7d74bb61
KW
781
782 char *save_input_locale = NULL;
7d74bb61
KW
783 STRLEN final_pos;
784
8ef6e574 785#ifdef LC_ALL
7d74bb61 786 assert(category != LC_ALL);
8ef6e574 787#endif
7d74bb61
KW
788
789 /* First dispose of the trivial cases */
b07fffd1 790 save_input_locale = setlocale(category, NULL);
7d74bb61 791 if (! save_input_locale) {
69014004
KW
792 DEBUG_L(PerlIO_printf(Perl_debug_log,
793 "Could not find current locale for category %d\n",
794 category));
7d74bb61
KW
795 return FALSE; /* XXX maybe should croak */
796 }
b07fffd1 797 save_input_locale = stdize_locale(savepv(save_input_locale));
7d74bb61
KW
798 if ((*save_input_locale == 'C' && save_input_locale[1] == '\0')
799 || strEQ(save_input_locale, "POSIX"))
800 {
69014004
KW
801 DEBUG_L(PerlIO_printf(Perl_debug_log,
802 "Current locale for category %d is %s\n",
803 category, save_input_locale));
b07fffd1 804 Safefree(save_input_locale);
7d74bb61
KW
805 return FALSE;
806 }
807
1d958db2
KW
808#if defined(USE_LOCALE_CTYPE) \
809 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 810
1d958db2 811 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
812
813 char *save_ctype_locale = NULL;
119ee68b 814 bool is_utf8;
7d74bb61 815
119ee68b 816 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
817
818 /* Get the current LC_CTYPE locale */
819 save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL)));
820 if (! save_ctype_locale) {
69014004
KW
821 DEBUG_L(PerlIO_printf(Perl_debug_log,
822 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
823 goto cant_use_nllanginfo;
824 }
825
826 /* If LC_CTYPE and the desired category use the same locale, this
827 * means that finding the value for LC_CTYPE is the same as finding
828 * the value for the desired category. Otherwise, switch LC_CTYPE
829 * to the desired category's locale */
830 if (strEQ(save_ctype_locale, save_input_locale)) {
831 Safefree(save_ctype_locale);
832 save_ctype_locale = NULL;
833 }
834 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
835 DEBUG_L(PerlIO_printf(Perl_debug_log,
836 "Could not change LC_CTYPE locale to %s\n",
837 save_input_locale));
7d74bb61
KW
838 Safefree(save_ctype_locale);
839 goto cant_use_nllanginfo;
840 }
841 }
842
69014004
KW
843 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
844 save_input_locale));
845
7d74bb61 846 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
847 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
848 * should give the correct results */
119ee68b 849
1d958db2
KW
850# if defined(HAS_NL_LANGINFO) && defined(CODESET)
851 {
852 char *codeset = savepv(nl_langinfo(CODESET));
853 if (codeset && strNE(codeset, "")) {
119ee68b 854
1d958db2
KW
855 /* If we switched LC_CTYPE, switch back */
856 if (save_ctype_locale) {
857 setlocale(LC_CTYPE, save_ctype_locale);
858 Safefree(save_ctype_locale);
859 }
860
861 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
862 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
863
69014004
KW
864 DEBUG_L(PerlIO_printf(Perl_debug_log,
865 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
866 codeset, is_utf8));
1d958db2
KW
867 Safefree(codeset);
868 Safefree(save_input_locale);
869 return is_utf8;
870 }
119ee68b
KW
871 }
872
1d958db2
KW
873# endif
874# ifdef MB_CUR_MAX
875
876 /* Here, either we don't have nl_langinfo, or it didn't return a
877 * codeset. Try MB_CUR_MAX */
878
119ee68b
KW
879 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
880 * Unicode code point. Since UTF-8 is the only non-single byte
881 * encoding we handle, we just say any such encoding is UTF-8, and if
882 * turns out to be wrong, other things will fail */
883 is_utf8 = MB_CUR_MAX >= 4;
884
69014004
KW
885 DEBUG_L(PerlIO_printf(Perl_debug_log,
886 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
887 (int) MB_CUR_MAX, is_utf8));
888
119ee68b
KW
889 Safefree(save_input_locale);
890
891# ifdef HAS_MBTOWC
892
893 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
894 * since they are both in the C99 standard. We can feed a known byte
895 * string to the latter function, and check that it gives the expected
896 * result */
897 if (is_utf8) {
898 wchar_t wc;
a6715020 899 GCC_DIAG_IGNORE(-Wunused-result);
1d958db2 900 (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */
a6715020 901 GCC_DIAG_RESTORE;
69014004 902 errno = 0;
119ee68b
KW
903 if (mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
904 != strlen(HYPHEN_UTF8)
905 || wc != (wchar_t) 0x2010)
906 {
907 is_utf8 = FALSE;
69014004
KW
908 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", wc));
909 DEBUG_L(PerlIO_printf(Perl_debug_log,
910 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
911 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
912 }
913 }
119ee68b
KW
914# endif
915
1d958db2
KW
916 /* If we switched LC_CTYPE, switch back */
917 if (save_ctype_locale) {
918 setlocale(LC_CTYPE, save_ctype_locale);
919 Safefree(save_ctype_locale);
119ee68b 920 }
7d74bb61 921
1d958db2 922 return is_utf8;
119ee68b 923# endif
7d74bb61 924 }
119ee68b 925
7d74bb61
KW
926 cant_use_nllanginfo:
927
928#endif /* HAS_NL_LANGINFO etc */
929
930 /* nl_langinfo not available or failed somehow. Look at the locale name to
751e9426 931 * see if it matches qr/UTF -? 8 /ix */
7d74bb61
KW
932
933 final_pos = strlen(save_input_locale) - 1;
751e9426
KW
934 if (final_pos >= 3) {
935 char *name = save_input_locale;
936
937 /* Find next 'U' or 'u' and look from there */
938 while ((name += strcspn(name, "Uu") + 1)
939 <= save_input_locale + final_pos - 2)
7d74bb61 940 {
751e9426
KW
941 if (toFOLD(*(name)) != 't'
942 || toFOLD(*(name + 1)) != 'f')
943 {
944 continue;
945 }
946 name += 2;
947 if (*(name) == '-') {
948 if ((name > save_input_locale + final_pos - 1)) {
949 break;
950 }
951 name++;
952 }
953 if (*(name) == '8') {
954 Safefree(save_input_locale);
69014004
KW
955 DEBUG_L(PerlIO_printf(Perl_debug_log,
956 "Locale %s ends with UTF-8 in name\n",
957 save_input_locale));
751e9426
KW
958 return TRUE;
959 }
7d74bb61 960 }
69014004
KW
961 DEBUG_L(PerlIO_printf(Perl_debug_log,
962 "Locale %s doesn't end with UTF-8 in name\n",
963 save_input_locale));
7d74bb61
KW
964 }
965
966#ifdef WIN32
967 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
968 if (final_pos >= 4
969 && *(save_input_locale + final_pos - 0) == '1'
970 && *(save_input_locale + final_pos - 1) == '0'
971 && *(save_input_locale + final_pos - 2) == '0'
972 && *(save_input_locale + final_pos - 3) == '5'
973 && *(save_input_locale + final_pos - 4) == '6')
974 {
975 Safefree(save_input_locale);
69014004
KW
976 DEBUG_L(PerlIO_printf(Perl_debug_log,
977 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
978 save_input_locale));
7d74bb61
KW
979 return TRUE;
980 }
981#endif
982
fa9b773e
KW
983 /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
984 if (instr(save_input_locale, "8859")) {
985 Safefree(save_input_locale);
69014004
KW
986 DEBUG_L(PerlIO_printf(Perl_debug_log,
987 "Locale %s has 8859 in name, not UTF-8 locale\n",
988 save_input_locale));
fa9b773e
KW
989 return FALSE;
990 }
991
992#ifdef HAS_LOCALECONV
993
994# ifdef USE_LOCALE_MONETARY
995
996 /* Here, there is nothing in the locale name to indicate whether the locale
997 * is UTF-8 or not. This "name", the return of setlocale(), is actually
998 * defined to be opaque, so we can't really rely on the absence of various
999 * substrings in the name to indicate its UTF-8ness. Look at the locale's
1000 * currency symbol. Often that will be in the native script, and if the
1001 * symbol isn't in UTF-8, we know that the locale isn't. If it is
1002 * non-ASCII UTF-8, we infer that the locale is too.
1003 * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the
1004 * locale of the desired category, if it isn't that locale already */
1005
1006 {
1007 char *save_monetary_locale = NULL;
1008 bool illegal_utf8 = FALSE;
1009 bool only_ascii = FALSE;
1010 const struct lconv* const lc = localeconv();
1011
1012 if (category != LC_MONETARY) {
1013
1014 save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY,
1015 NULL)));
1016 if (! save_monetary_locale) {
69014004
KW
1017 DEBUG_L(PerlIO_printf(Perl_debug_log,
1018 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1019 goto cant_use_monetary;
1020 }
1021
1022 if (strNE(save_monetary_locale, save_input_locale)) {
1023 if (! setlocale(LC_MONETARY, save_input_locale)) {
69014004
KW
1024 DEBUG_L(PerlIO_printf(Perl_debug_log,
1025 "Could not change LC_MONETARY locale to %s\n",
1026 save_input_locale));
fa9b773e
KW
1027 Safefree(save_monetary_locale);
1028 goto cant_use_monetary;
1029 }
1030 }
1031 }
1032
1033 /* Here the current LC_MONETARY is set to the locale of the category
1034 * whose information is desired. */
1035
1036 if (lc && lc->currency_symbol) {
1037 if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) {
69014004
KW
1038 DEBUG_L(PerlIO_printf(Perl_debug_log,
1039 "Currency symbol for %s is not legal UTF-8\n",
1040 save_input_locale));
fa9b773e
KW
1041 illegal_utf8 = TRUE;
1042 }
1043 else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) {
69014004 1044 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
1045 only_ascii = TRUE;
1046 }
1047 }
1048
1049 /* If we changed it, restore LC_MONETARY to its original locale */
1050 if (save_monetary_locale) {
1051 setlocale(LC_MONETARY, save_monetary_locale);
1052 Safefree(save_monetary_locale);
1053 }
1054
1055 Safefree(save_input_locale);
1056
1057 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise
1058 * assume the locale is UTF-8 if and only if the symbol is non-ascii
1059 * UTF-8. (We can't really tell if the locale is UTF-8 or not if the
1060 * symbol is just a '$', so we err on the side of it not being UTF-8)
1061 * */
69014004
KW
1062 DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", (illegal_utf8)
1063 ? FALSE
1064 : ! only_ascii));
fa9b773e
KW
1065 return (illegal_utf8)
1066 ? FALSE
1067 : ! only_ascii;
1068
1069 }
1070 cant_use_monetary:
1071
1072# endif /* USE_LOCALE_MONETARY */
1073#endif /* HAS_LOCALECONV */
1074
1075#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES)
1076
1077/* This code is ifdefd out because it was found to not be necessary in testing
1078 * on our dromedary test machine, which has over 700 locales. There, looking
1079 * at just the currency symbol gave essentially the same results as doing this
1080 * extra work. Executing this also caused segfaults in miniperl. I left it in
1081 * so as to avoid rewriting it if real-world experience indicates that
1082 * dromedary is an outlier. Essentially, instead of returning abpve if we
1083 * haven't found illegal utf8, we continue on and examine all the strerror()
1084 * messages on the platform for utf8ness. If all are ASCII, we still don't
1085 * know the answer; but otherwise we have a pretty good indication of the
1086 * utf8ness. The reason this doesn't necessarily help much is that the
1087 * messages may not have been translated into the locale. The currency symbol
1088 * is much more likely to have been translated. The code below would need to
1089 * be altered somewhat to just be a continuation of testing the currency
1090 * symbol. */
1091 int e;
1092 unsigned int failures = 0, non_ascii = 0;
1093 char *save_messages_locale = NULL;
1094
1095 /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the
1096 * desired category, if it isn't that locale already */
1097
1098 if (category != LC_MESSAGES) {
1099
1100 save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES,
1101 NULL)));
1102 if (! save_messages_locale) {
1103 goto cant_use_messages;
1104 }
1105
1106 if (strEQ(save_messages_locale, save_input_locale)) {
1107 Safefree(save_input_locale);
1108 }
1109 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
1110 Safefree(save_messages_locale);
1111 goto cant_use_messages;
1112 }
1113 }
1114
1115 /* Here the current LC_MESSAGES is set to the locale of the category
1116 * whose information is desired. Look through all the messages */
1117
1118 for (e = 0;
1119#ifdef HAS_SYS_ERRLIST
1120 e <= sys_nerr
1121#endif
1122 ; e++)
1123 {
1124 const U8* const errmsg = (U8 *) Strerror(e) ;
1125 if (!errmsg)
1126 break;
1127 if (! is_utf8_string(errmsg, 0)) {
1128 failures++;
1129 break;
1130 }
1131 else if (! is_ascii_string(errmsg, 0)) {
1132 non_ascii++;
1133 }
1134 }
1135
1136 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1137 if (save_messages_locale) {
1138 setlocale(LC_MESSAGES, save_messages_locale);
1139 Safefree(save_messages_locale);
1140 }
1141
1142 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1143 * any non-ascii means it is one; otherwise we assume it isn't */
1144 return (failures) ? FALSE : non_ascii;
1145
1146 }
1147 cant_use_messages:
1148
1149#endif
1150
69014004
KW
1151 DEBUG_L(PerlIO_printf(Perl_debug_log,
1152 "Assuming locale %s is not a UTF-8 locale\n",
1153 save_input_locale));
fa9b773e 1154 Safefree(save_input_locale);
7d74bb61
KW
1155 return FALSE;
1156}
1157
8ef6e574 1158#endif
7d74bb61 1159
66610fdd
RGS
1160/*
1161 * Local variables:
1162 * c-indentation-style: bsd
1163 * c-basic-offset: 4
14d04a33 1164 * indent-tabs-mode: nil
66610fdd
RGS
1165 * End:
1166 *
14d04a33 1167 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1168 */