This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Make behavior of $Carp::MaxArgNums match docs
[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
96# ifdef HAS_LOCALECONV
7452cf6a 97 const struct lconv* const lc = localeconv();
98994639 98
98994639
HS
99 if (lc && lc->decimal_point) {
100 if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) {
101 SvREFCNT_dec(PL_numeric_radix_sv);
a0714e2c 102 PL_numeric_radix_sv = NULL;
98994639
HS
103 }
104 else {
105 if (PL_numeric_radix_sv)
106 sv_setpv(PL_numeric_radix_sv, lc->decimal_point);
107 else
108 PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0);
9f10db87 109 if (! is_invariant_string((U8 *) lc->decimal_point, 0)
28acfe03 110 && is_utf8_string((U8 *) lc->decimal_point, 0)
c1284011 111 && _is_cur_LC_category_utf8(LC_NUMERIC))
28acfe03
KW
112 {
113 SvUTF8_on(PL_numeric_radix_sv);
114 }
98994639
HS
115 }
116 }
117 else
a0714e2c 118 PL_numeric_radix_sv = NULL;
69014004
KW
119
120 DEBUG_L(PerlIO_printf(Perl_debug_log, "Locale radix is %s\n",
121 (PL_numeric_radix_sv)
122 ? lc->decimal_point
123 : "NULL"));
124
98994639
HS
125# endif /* HAS_LOCALECONV */
126#endif /* USE_LOCALE_NUMERIC */
127}
128
a39edc4c
KW
129/* Is the C string input 'name' "C" or "POSIX"? If so, and 'name' is the
130 * return of setlocale(), then this is extremely likely to be the C or POSIX
131 * locale. However, the output of setlocale() is documented to be opaque, but
132 * the odds are extremely small that it would return these two strings for some
133 * other locale. Note that VMS in these two locales includes many non-ASCII
134 * characters as controls and punctuation (below are hex bytes):
135 * cntrl: 00-1F 7F 84-97 9B-9F
136 * punct: 21-2F 3A-40 5B-60 7B-7E A1-A3 A5 A7-AB B0-B3 B5-B7 B9-BD BF-CF D1-DD DF-EF F1-FD
137 * Oddly, none there are listed as alphas, though some represent alphabetics
138 * http://www.nntp.perl.org/group/perl.perl5.porters/2013/02/msg198753.html */
98b630b3
KW
139#define isNAME_C_OR_POSIX(name) ((name) != NULL \
140 && ((*(name) == 'C' && (*(name + 1)) == '\0') \
a39edc4c
KW
141 || strEQ((name), "POSIX")))
142
98994639 143void
8772537c 144Perl_new_numeric(pTHX_ const char *newnum)
98994639
HS
145{
146#ifdef USE_LOCALE_NUMERIC
0d071d52
KW
147
148 /* Called after all libc setlocale() calls affecting LC_NUMERIC, to tell
149 * core Perl this and that 'newnum' is the name of the new locale.
150 * It installs this locale as the current underlying default.
151 *
152 * The default locale and the C locale can be toggled between by use of the
153 * set_numeric_local() and set_numeric_standard() functions, which should
154 * probably not be called directly, but only via macros like
155 * SET_NUMERIC_STANDARD() in perl.h.
156 *
157 * The toggling is necessary mainly so that a non-dot radix decimal point
158 * character can be output, while allowing internal calculations to use a
159 * dot.
160 *
161 * This sets several interpreter-level variables:
bb304765 162 * PL_numeric_name The underlying locale's name: a copy of 'newnum'
0d071d52 163 * PL_numeric_local A boolean indicating if the toggled state is such
7738054c
KW
164 * that the current locale is the program's underlying
165 * locale
166 * PL_numeric_standard An int indicating if the toggled state is such
167 * that the current locale is the C locale. If non-zero,
168 * it is in C; if > 1, it means it may not be toggled away
169 * from C.
0d071d52
KW
170 * Note that both of the last two variables can be true at the same time,
171 * if the underlying locale is C. (Toggling is a no-op under these
172 * circumstances.)
173 *
174 * Any code changing the locale (outside this file) should use
175 * POSIX::setlocale, which calls this function. Therefore this function
176 * should be called directly only from this file and from
177 * POSIX::setlocale() */
178
b03f34cf 179 char *save_newnum;
98994639
HS
180
181 if (! newnum) {
43c5f42d
NC
182 Safefree(PL_numeric_name);
183 PL_numeric_name = NULL;
98994639
HS
184 PL_numeric_standard = TRUE;
185 PL_numeric_local = TRUE;
186 return;
187 }
188
b03f34cf
KW
189 save_newnum = stdize_locale(savepv(newnum));
190 if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) {
98994639 191 Safefree(PL_numeric_name);
b03f34cf 192 PL_numeric_name = save_newnum;
b03f34cf 193 }
98994639 194
a39edc4c 195 PL_numeric_standard = isNAME_C_OR_POSIX(save_newnum);
e19f01cb 196 PL_numeric_local = TRUE;
4c28b29c
KW
197
198 /* Keep LC_NUMERIC in the C locale. This is for XS modules, so they don't
199 * have to worry about the radix being a non-dot. (Core operations that
200 * need the underlying locale change to it temporarily). */
201 set_numeric_standard();
202
e19f01cb 203 set_numeric_radix();
6959d69d 204
f2ce9e1c
JH
205#else
206 PERL_UNUSED_ARG(newnum);
98994639
HS
207#endif /* USE_LOCALE_NUMERIC */
208}
209
210void
211Perl_set_numeric_standard(pTHX)
212{
213#ifdef USE_LOCALE_NUMERIC
28c1bf33
KW
214 /* Toggle the LC_NUMERIC locale to C. Most code should use the macros like
215 * SET_NUMERIC_STANDARD() in perl.h instead of calling this directly. The
216 * macro avoids calling this routine if toggling isn't necessary according
217 * to our records (which could be wrong if some XS code has changed the
218 * locale behind our back) */
0d071d52 219
a9b8c0d8
KW
220 setlocale(LC_NUMERIC, "C");
221 PL_numeric_standard = TRUE;
222 PL_numeric_local = isNAME_C_OR_POSIX(PL_numeric_name);
223 set_numeric_radix();
69014004
KW
224 DEBUG_L(PerlIO_printf(Perl_debug_log,
225 "Underlying LC_NUMERIC locale now is C\n"));
98994639
HS
226
227#endif /* USE_LOCALE_NUMERIC */
228}
229
230void
231Perl_set_numeric_local(pTHX)
232{
233#ifdef USE_LOCALE_NUMERIC
28c1bf33
KW
234 /* Toggle the LC_NUMERIC locale to the current underlying default. Most
235 * code should use the macros like SET_NUMERIC_LOCAL() in perl.h instead of
236 * calling this directly. The macro avoids calling this routine if
237 * toggling isn't necessary according to our records (which could be wrong
238 * if some XS code has changed the locale behind our back) */
a9b8c0d8
KW
239
240 setlocale(LC_NUMERIC, PL_numeric_name);
241 PL_numeric_standard = isNAME_C_OR_POSIX(PL_numeric_name);
242 PL_numeric_local = TRUE;
243 set_numeric_radix();
69014004
KW
244 DEBUG_L(PerlIO_printf(Perl_debug_log,
245 "Underlying LC_NUMERIC locale now is %s\n",
246 PL_numeric_name));
98994639
HS
247
248#endif /* USE_LOCALE_NUMERIC */
249}
250
251/*
252 * Set up for a new ctype locale.
253 */
254void
8772537c 255Perl_new_ctype(pTHX_ const char *newctype)
98994639
HS
256{
257#ifdef USE_LOCALE_CTYPE
0d071d52
KW
258
259 /* Called after all libc setlocale() calls affecting LC_CTYPE, to tell
260 * core Perl this and that 'newctype' is the name of the new locale.
261 *
262 * This function sets up the folding arrays for all 256 bytes, assuming
263 * that tofold() is tolc() since fold case is not a concept in POSIX,
264 *
265 * Any code changing the locale (outside this file) should use
266 * POSIX::setlocale, which calls this function. Therefore this function
267 * should be called directly only from this file and from
268 * POSIX::setlocale() */
269
27da23d5 270 dVAR;
68067e4e 271 UV i;
98994639 272
7918f24d
NC
273 PERL_ARGS_ASSERT_NEW_CTYPE;
274
215c5139
KW
275 /* We will replace any bad locale warning with 1) nothing if the new one is
276 * ok; or 2) a new warning for the bad new locale */
277 if (PL_warn_locale) {
278 SvREFCNT_dec_NN(PL_warn_locale);
279 PL_warn_locale = NULL;
280 }
281
c1284011 282 PL_in_utf8_CTYPE_locale = _is_cur_LC_category_utf8(LC_CTYPE);
31f05a37
KW
283
284 /* A UTF-8 locale gets standard rules. But note that code still has to
285 * handle this specially because of the three problematic code points */
286 if (PL_in_utf8_CTYPE_locale) {
287 Copy(PL_fold_latin1, PL_fold_locale, 256, U8);
288 }
289 else {
8c6180a9
KW
290 /* Assume enough space for every character being bad. 4 spaces each
291 * for the 94 printable characters that are output like "'x' "; and 5
292 * spaces each for "'\\' ", "'\t' ", and "'\n' "; plus a terminating
293 * NUL */
294 char bad_chars_list[ (94 * 4) + (3 * 5) + 1 ];
295
296 bool check_for_problems = ckWARN_d(WARN_LOCALE); /* No warnings means
297 no check */
298 bool multi_byte_locale = FALSE; /* Assume is a single-byte locale
299 to start */
300 unsigned int bad_count = 0; /* Count of bad characters */
301
baa60164
KW
302 for (i = 0; i < 256; i++) {
303 if (isUPPER_LC((U8) i))
304 PL_fold_locale[i] = (U8) toLOWER_LC((U8) i);
305 else if (isLOWER_LC((U8) i))
306 PL_fold_locale[i] = (U8) toUPPER_LC((U8) i);
307 else
308 PL_fold_locale[i] = (U8) i;
8c6180a9
KW
309
310 /* If checking for locale problems, see if the native ASCII-range
311 * printables plus \n and \t are in their expected categories in
312 * the new locale. If not, this could mean big trouble, upending
313 * Perl's and most programs' assumptions, like having a
314 * metacharacter with special meaning become a \w. Fortunately,
315 * it's very rare to find locales that aren't supersets of ASCII
316 * nowadays. It isn't a problem for most controls to be changed
317 * into something else; we check only \n and \t, though perhaps \r
318 * could be an issue as well. */
319 if (check_for_problems
320 && (isGRAPH_A(i) || isBLANK_A(i) || i == '\n'))
321 {
322 if ((isALPHANUMERIC_A(i) && ! isALPHANUMERIC_LC(i))
323 || (isPUNCT_A(i) && ! isPUNCT_LC(i))
324 || (isBLANK_A(i) && ! isBLANK_LC(i))
325 || (i == '\n' && ! isCNTRL_LC(i)))
326 {
327 if (bad_count) { /* Separate multiple entries with a
328 blank */
329 bad_chars_list[bad_count++] = ' ';
330 }
331 bad_chars_list[bad_count++] = '\'';
332 if (isPRINT_A(i)) {
333 bad_chars_list[bad_count++] = (char) i;
334 }
335 else {
336 bad_chars_list[bad_count++] = '\\';
337 if (i == '\n') {
338 bad_chars_list[bad_count++] = 'n';
339 }
340 else {
341 assert(i == '\t');
342 bad_chars_list[bad_count++] = 't';
343 }
344 }
345 bad_chars_list[bad_count++] = '\'';
346 bad_chars_list[bad_count] = '\0';
347 }
348 }
349 }
350
351#ifdef MB_CUR_MAX
352 /* We only handle single-byte locales (outside of UTF-8 ones; so if
353 * this locale requires than one byte, there are going to be
354 * problems. */
ba1a4362
KW
355 if (check_for_problems && MB_CUR_MAX > 1
356
357 /* Some platforms return MB_CUR_MAX > 1 for even the "C"
358 * locale. Just assume that the implementation for them (plus
359 * for POSIX) is correct and the > 1 value is spurious. (Since
360 * these are specially handled to never be considered UTF-8
361 * locales, as long as this is the only problem, everything
362 * should work fine */
363 && strNE(newctype, "C") && strNE(newctype, "POSIX"))
364 {
8c6180a9
KW
365 multi_byte_locale = TRUE;
366 }
367#endif
368
369 if (bad_count || multi_byte_locale) {
780fcc9f 370 PL_warn_locale = Perl_newSVpvf(aTHX_
8c6180a9 371 "Locale '%s' may not work well.%s%s%s\n",
780fcc9f 372 newctype,
8c6180a9
KW
373 (multi_byte_locale)
374 ? " Some characters in it are not recognized by"
375 " Perl."
376 : "",
377 (bad_count)
378 ? "\nThe following characters (and maybe others)"
379 " may not have the same meaning as the Perl"
380 " program expects:\n"
381 : "",
382 (bad_count)
383 ? bad_chars_list
384 : ""
385 );
780fcc9f
KW
386 /* If we are actually in the scope of the locale, output the
387 * message now. Otherwise we save it to be output at the first
388 * operation using this locale, if that actually happens. Most
389 * programs don't use locales, so they are immune to bad ones */
390 if (IN_LC(LC_CTYPE)) {
391
392 /* We have to save 'newctype' because the setlocale() just
393 * below may destroy it. The next setlocale() further down
394 * should restore it properly so that the intermediate change
395 * here is transparent to this function's caller */
396 const char * const badlocale = savepv(newctype);
397
398 setlocale(LC_CTYPE, "C");
399
400 /* The '0' below suppresses a bogus gcc compiler warning */
401 Perl_warner(aTHX_ packWARN(WARN_LOCALE), SvPVX(PL_warn_locale), 0);
402 setlocale(LC_CTYPE, badlocale);
c0f3a893 403 Safefree(badlocale);
780fcc9f
KW
404 SvREFCNT_dec_NN(PL_warn_locale);
405 PL_warn_locale = NULL;
406 }
baa60164 407 }
31f05a37 408 }
98994639
HS
409
410#endif /* USE_LOCALE_CTYPE */
7918f24d 411 PERL_ARGS_ASSERT_NEW_CTYPE;
8772537c 412 PERL_UNUSED_ARG(newctype);
96a5add6 413 PERL_UNUSED_CONTEXT;
98994639
HS
414}
415
98994639 416void
2726666d
KW
417Perl__warn_problematic_locale()
418{
2726666d
KW
419
420#ifdef USE_LOCALE_CTYPE
421
5f04a188
KW
422 dTHX;
423
424 /* Internal-to-core function that outputs the message in PL_warn_locale,
425 * and then NULLS it. Should be called only through the macro
426 * _CHECK_AND_WARN_PROBLEMATIC_LOCALE */
427
2726666d
KW
428 if (PL_warn_locale) {
429 /*GCC_DIAG_IGNORE(-Wformat-security); Didn't work */
430 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
431 SvPVX(PL_warn_locale),
432 0 /* dummy to avoid compiler warning */ );
433 /* GCC_DIAG_RESTORE; */
434 SvREFCNT_dec_NN(PL_warn_locale);
435 PL_warn_locale = NULL;
436 }
437
438#endif
439
440}
441
442void
8772537c 443Perl_new_collate(pTHX_ const char *newcoll)
98994639
HS
444{
445#ifdef USE_LOCALE_COLLATE
0d071d52
KW
446
447 /* Called after all libc setlocale() calls affecting LC_COLLATE, to tell
448 * core Perl this and that 'newcoll' is the name of the new locale.
449 *
450 * Any code changing the locale (outside this file) should use
451 * POSIX::setlocale, which calls this function. Therefore this function
452 * should be called directly only from this file and from
453 * POSIX::setlocale() */
454
98994639
HS
455 if (! newcoll) {
456 if (PL_collation_name) {
457 ++PL_collation_ix;
458 Safefree(PL_collation_name);
459 PL_collation_name = NULL;
460 }
461 PL_collation_standard = TRUE;
462 PL_collxfrm_base = 0;
463 PL_collxfrm_mult = 2;
464 return;
465 }
466
467 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
468 ++PL_collation_ix;
469 Safefree(PL_collation_name);
470 PL_collation_name = stdize_locale(savepv(newcoll));
a39edc4c 471 PL_collation_standard = isNAME_C_OR_POSIX(newcoll);
98994639
HS
472
473 {
474 /* 2: at most so many chars ('a', 'b'). */
475 /* 50: surely no system expands a char more. */
476#define XFRMBUFSIZE (2 * 50)
477 char xbuf[XFRMBUFSIZE];
8772537c
AL
478 const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE);
479 const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE);
480 const SSize_t mult = fb - fa;
e0601d3c 481 if (mult < 1 && !(fa == 0 && fb == 0))
ad49ad39
NC
482 Perl_croak(aTHX_ "panic: strxfrm() gets absurd - a => %"UVuf", ab => %"UVuf,
483 (UV) fa, (UV) fb);
eb160463 484 PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0;
98994639
HS
485 PL_collxfrm_mult = mult;
486 }
487 }
488
f2ce9e1c
JH
489#else
490 PERL_UNUSED_ARG(newcoll);
98994639
HS
491#endif /* USE_LOCALE_COLLATE */
492}
493
b385bb4d
KW
494#ifdef WIN32
495
496char *
497Perl_my_setlocale(pTHX_ int category, const char* locale)
498{
499 /* This, for Windows, emulates POSIX setlocale() behavior. There is no
500 * difference unless the input locale is "", which means on Windows to get
501 * the machine default, which is set via the computer's "Regional and
502 * Language Options" (or its current equivalent). In POSIX, it instead
503 * means to find the locale from the user's environment. This routine
504 * looks in the environment, and, if anything is found, uses that instead
505 * of going to the machine default. If there is no environment override,
506 * the machine default is used, as normal, by calling the real setlocale()
507 * with "". The POSIX behavior is to use the LC_ALL variable if set;
508 * otherwise to use the particular category's variable if set; otherwise to
509 * use the LANG variable. */
510
175c4cf9 511 bool override_LC_ALL = FALSE;
89f7b9aa
KW
512 char * result;
513
b385bb4d
KW
514 if (locale && strEQ(locale, "")) {
515# ifdef LC_ALL
516 locale = PerlEnv_getenv("LC_ALL");
517 if (! locale) {
518#endif
519 switch (category) {
520# ifdef LC_ALL
521 case LC_ALL:
481465ea 522 override_LC_ALL = TRUE;
b385bb4d
KW
523 break; /* We already know its variable isn't set */
524# endif
525# ifdef USE_LOCALE_TIME
526 case LC_TIME:
527 locale = PerlEnv_getenv("LC_TIME");
528 break;
529# endif
530# ifdef USE_LOCALE_CTYPE
531 case LC_CTYPE:
532 locale = PerlEnv_getenv("LC_CTYPE");
533 break;
534# endif
535# ifdef USE_LOCALE_COLLATE
536 case LC_COLLATE:
537 locale = PerlEnv_getenv("LC_COLLATE");
538 break;
539# endif
540# ifdef USE_LOCALE_MONETARY
541 case LC_MONETARY:
542 locale = PerlEnv_getenv("LC_MONETARY");
543 break;
544# endif
545# ifdef USE_LOCALE_NUMERIC
546 case LC_NUMERIC:
547 locale = PerlEnv_getenv("LC_NUMERIC");
548 break;
549# endif
550# ifdef USE_LOCALE_MESSAGES
551 case LC_MESSAGES:
552 locale = PerlEnv_getenv("LC_MESSAGES");
553 break;
554# endif
555 default:
556 /* This is a category, like PAPER_SIZE that we don't
557 * know about; and so can't provide a wrapper. */
558 break;
559 }
560 if (! locale) {
561 locale = PerlEnv_getenv("LANG");
481465ea 562 if (! locale) {
b385bb4d
KW
563 locale = "";
564 }
565 }
566# ifdef LC_ALL
567 }
568# endif
569 }
570
89f7b9aa 571 result = setlocale(category, locale);
bbc98134
KW
572 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n", __FILE__, __LINE__,
573 _setlocale_debug_string(category, locale, result)));
89f7b9aa 574
481465ea 575 if (! override_LC_ALL) {
89f7b9aa
KW
576 return result;
577 }
578
dfd77d7a 579 /* Here the input category was LC_ALL, and we have set it to what is in the
481465ea
KW
580 * LANG variable or the system default if there is no LANG. But these have
581 * lower priority than the other LC_foo variables, so override it for each
582 * one that is set. (If they are set to "", it means to use the same thing
583 * we just set LC_ALL to, so can skip) */
89f7b9aa
KW
584# ifdef USE_LOCALE_TIME
585 result = PerlEnv_getenv("LC_TIME");
730252b2 586 if (result && strNE(result, "")) {
89f7b9aa 587 setlocale(LC_TIME, result);
bbc98134
KW
588 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
589 __FILE__, __LINE__,
590 _setlocale_debug_string(LC_TIME, result, "not captured")));
89f7b9aa
KW
591 }
592# endif
593# ifdef USE_LOCALE_CTYPE
594 result = PerlEnv_getenv("LC_CTYPE");
730252b2 595 if (result && strNE(result, "")) {
89f7b9aa 596 setlocale(LC_CTYPE, result);
bbc98134
KW
597 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
598 __FILE__, __LINE__,
599 _setlocale_debug_string(LC_CTYPE, result, "not captured")));
89f7b9aa
KW
600 }
601# endif
602# ifdef USE_LOCALE_COLLATE
603 result = PerlEnv_getenv("LC_COLLATE");
730252b2 604 if (result && strNE(result, "")) {
89f7b9aa 605 setlocale(LC_COLLATE, result);
bbc98134
KW
606 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
607 __FILE__, __LINE__,
608 _setlocale_debug_string(LC_COLLATE, result, "not captured")));
89f7b9aa
KW
609 }
610# endif
611# ifdef USE_LOCALE_MONETARY
612 result = PerlEnv_getenv("LC_MONETARY");
730252b2 613 if (result && strNE(result, "")) {
89f7b9aa 614 setlocale(LC_MONETARY, result);
bbc98134
KW
615 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
616 __FILE__, __LINE__,
617 _setlocale_debug_string(LC_MONETARY, result, "not captured")));
89f7b9aa
KW
618 }
619# endif
620# ifdef USE_LOCALE_NUMERIC
621 result = PerlEnv_getenv("LC_NUMERIC");
730252b2 622 if (result && strNE(result, "")) {
89f7b9aa 623 setlocale(LC_NUMERIC, result);
bbc98134
KW
624 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
625 __FILE__, __LINE__,
626 _setlocale_debug_string(LC_NUMERIC, result, "not captured")));
89f7b9aa
KW
627 }
628# endif
629# ifdef USE_LOCALE_MESSAGES
630 result = PerlEnv_getenv("LC_MESSAGES");
730252b2 631 if (result && strNE(result, "")) {
89f7b9aa 632 setlocale(LC_MESSAGES, result);
bbc98134
KW
633 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
634 __FILE__, __LINE__,
635 _setlocale_debug_string(LC_MESSAGES, result, "not captured")));
89f7b9aa
KW
636 }
637# endif
638
bbc98134
KW
639 result = setlocale(LC_ALL, NULL);
640 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
641 __FILE__, __LINE__,
642 _setlocale_debug_string(LC_ALL, NULL, result)));
89f7b9aa 643
bbc98134 644 return result;
b385bb4d
KW
645}
646
647#endif
648
649
98994639
HS
650/*
651 * Initialize locale awareness.
652 */
653int
654Perl_init_i18nl10n(pTHX_ int printwarn)
655{
0e92a118
KW
656 /* printwarn is
657 *
658 * 0 if not to output warning when setup locale is bad
659 * 1 if to output warning based on value of PERL_BADLANG
660 * >1 if to output regardless of PERL_BADLANG
661 *
662 * returns
98994639 663 * 1 = set ok or not applicable,
0e92a118
KW
664 * 0 = fallback to a locale of lower priority
665 * -1 = fallback to all locales failed, not even to the C locale
6b058d42
KW
666 *
667 * Under -DDEBUGGING, if the environment variable PERL_DEBUG_LOCALE_INIT is
668 * set, debugging information is output.
669 *
670 * This looks more complicated than it is, mainly due to the #ifdefs.
671 *
672 * We try to set LC_ALL to the value determined by the environment. If
673 * there is no LC_ALL on this platform, we try the individual categories we
674 * know about. If this works, we are done.
675 *
676 * But if it doesn't work, we have to do something else. We search the
677 * environment variables ourselves instead of relying on the system to do
678 * it. We look at, in order, LC_ALL, LANG, a system default locale (if we
679 * think there is one), and the ultimate fallback "C". This is all done in
680 * the same loop as above to avoid duplicating code, but it makes things
681 * more complex. After the original failure, we add the fallback
682 * possibilities to the list of locales to try, and iterate the loop
683 * through them all until one succeeds.
684 *
685 * On Ultrix, the locale MUST come from the environment, so there is
686 * preliminary code to set it. I (khw) am not sure that it is necessary,
687 * and that this couldn't be folded into the loop, but barring any real
688 * platforms to test on, it's staying as-is
689 *
690 * A slight complication is that in embedded Perls, the locale may already
691 * be set-up, and we don't want to get it from the normal environment
692 * variables. This is handled by having a special environment variable
693 * indicate we're in this situation. We simply set setlocale's 2nd
694 * parameter to be a NULL instead of "". That indicates to setlocale that
695 * it is not to change anything, but to return the current value,
696 * effectively initializing perl's db to what the locale already is.
697 *
698 * We play the same trick with NULL if a LC_ALL succeeds. We call
699 * setlocale() on the individual categores with NULL to get their existing
700 * values for our db, instead of trying to change them.
701 * */
98994639 702
0e92a118
KW
703 int ok = 1;
704
98994639 705#if defined(USE_LOCALE)
98994639
HS
706#ifdef USE_LOCALE_CTYPE
707 char *curctype = NULL;
708#endif /* USE_LOCALE_CTYPE */
709#ifdef USE_LOCALE_COLLATE
710 char *curcoll = NULL;
711#endif /* USE_LOCALE_COLLATE */
712#ifdef USE_LOCALE_NUMERIC
713 char *curnum = NULL;
714#endif /* USE_LOCALE_NUMERIC */
715#ifdef __GLIBC__
175c4cf9 716 const char * const language = savepv(PerlEnv_getenv("LANGUAGE"));
98994639 717#endif
65ebb059 718
ccd65d51
KW
719 /* NULL uses the existing already set up locale */
720 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
721 ? NULL
722 : "";
5d1187d1
KW
723#ifdef DEBUGGING
724 const bool debug = (PerlEnv_getenv("PERL_DEBUG_LOCALE_INIT"))
725 ? TRUE
726 : FALSE;
727# define DEBUG_LOCALE_INIT(category, locale, result) \
728 STMT_START { \
729 if (debug) { \
730 PerlIO_printf(Perl_debug_log, \
731 "%s:%d: %s\n", \
732 __FILE__, __LINE__, \
733 _setlocale_debug_string(category, \
734 locale, \
735 result)); \
736 } \
737 } STMT_END
738#else
739# define DEBUG_LOCALE_INIT(a,b,c)
740#endif
c3fcd832
KW
741 const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
742 unsigned int trial_locales_count;
175c4cf9
KW
743 const char * const lc_all = savepv(PerlEnv_getenv("LC_ALL"));
744 const char * const lang = savepv(PerlEnv_getenv("LANG"));
98994639 745 bool setlocale_failure = FALSE;
65ebb059
KW
746 unsigned int i;
747 char *p;
175c4cf9
KW
748
749 /* A later getenv() could zap this, so only use here */
750 const char * const bad_lang_use_once = PerlEnv_getenv("PERL_BADLANG");
751
752 const bool locwarn = (printwarn > 1
753 || (printwarn
754 && (! bad_lang_use_once
22ff3130
HS
755 || (
756 /* disallow with "" or "0" */
757 *bad_lang_use_once
758 && strNE("0", bad_lang_use_once)))));
0e92a118 759 bool done = FALSE;
5d1187d1
KW
760 char * sl_result; /* return from setlocale() */
761 char * locale_param;
6bce99ee
JH
762#ifdef WIN32
763 /* In some systems you can find out the system default locale
764 * and use that as the fallback locale. */
765# define SYSTEM_DEFAULT_LOCALE
766#endif
767#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059 768 const char *system_default_locale = NULL;
6bce99ee 769#endif
98994639 770
0e92a118
KW
771#ifndef LOCALE_ENVIRON_REQUIRED
772 PERL_UNUSED_VAR(done);
5d1187d1 773 PERL_UNUSED_VAR(locale_param);
0e92a118 774#else
98994639
HS
775
776 /*
777 * Ultrix setlocale(..., "") fails if there are no environment
778 * variables from which to get a locale name.
779 */
780
b3e384bf 781# ifdef LC_ALL
98994639 782 if (lang) {
5d1187d1
KW
783 sl_result = my_setlocale(LC_ALL, setlocale_init);
784 DEBUG_LOCALE_INIT(LC_ALL, setlocale_init, sl_result);
785 if (sl_result)
98994639
HS
786 done = TRUE;
787 else
788 setlocale_failure = TRUE;
789 }
5d1187d1 790 if (! setlocale_failure) {
b3e384bf 791# ifdef USE_LOCALE_CTYPE
5d1187d1
KW
792 locale_param = (! done && (lang || PerlEnv_getenv("LC_CTYPE")))
793 ? setlocale_init
794 : NULL;
795 curctype = my_setlocale(LC_CTYPE, locale_param);
796 DEBUG_LOCALE_INIT(LC_CTYPE, locale_param, sl_result);
797 if (! curctype)
98994639
HS
798 setlocale_failure = TRUE;
799 else
800 curctype = savepv(curctype);
b3e384bf
KW
801# endif /* USE_LOCALE_CTYPE */
802# ifdef USE_LOCALE_COLLATE
5d1187d1
KW
803 locale_param = (! done && (lang || PerlEnv_getenv("LC_COLLATE")))
804 ? setlocale_init
805 : NULL;
806 curcoll = my_setlocale(LC_COLLATE, locale_param);
807 DEBUG_LOCALE_INIT(LC_COLLATE, locale_param, sl_result);
808 if (! curcoll)
98994639
HS
809 setlocale_failure = TRUE;
810 else
811 curcoll = savepv(curcoll);
b3e384bf
KW
812# endif /* USE_LOCALE_COLLATE */
813# ifdef USE_LOCALE_NUMERIC
5d1187d1
KW
814 locale_param = (! done && (lang || PerlEnv_getenv("LC_NUMERIC")))
815 ? setlocale_init
816 : NULL;
817 curnum = my_setlocale(LC_NUMERIC, locale_param);
818 DEBUG_LOCALE_INIT(LC_NUMERIC, locale_param, sl_result);
819 if (! curnum)
98994639
HS
820 setlocale_failure = TRUE;
821 else
822 curnum = savepv(curnum);
b3e384bf 823# endif /* USE_LOCALE_NUMERIC */
a782673d 824# ifdef USE_LOCALE_MESSAGES
5d1187d1
KW
825 locale_param = (! done && (lang || PerlEnv_getenv("LC_MESSAGES")))
826 ? setlocale_init
827 : NULL;
828 sl_result = my_setlocale(LC_MESSAGES, locale_param);
829 DEBUG_LOCALE_INIT(LC_MESSAGES, locale_param, sl_result);
830 if (! sl_result)
a782673d
KW
831 setlocale_failure = TRUE;
832 }
833# endif /* USE_LOCALE_MESSAGES */
c835d6be 834# ifdef USE_LOCALE_MONETARY
5d1187d1
KW
835 locale_param = (! done && (lang || PerlEnv_getenv("LC_MONETARY")))
836 ? setlocale_init
837 : NULL;
838 sl_result = my_setlocale(LC_MONETARY, locale_param);
839 DEBUG_LOCALE_INIT(LC_MONETARY, locale_param, sl_result);
840 if (! sl_result) {
c835d6be
KW
841 setlocale_failure = TRUE;
842 }
843# endif /* USE_LOCALE_MONETARY */
98994639
HS
844 }
845
b3e384bf 846# endif /* LC_ALL */
98994639
HS
847
848#endif /* !LOCALE_ENVIRON_REQUIRED */
849
65ebb059 850 /* We try each locale in the list until we get one that works, or exhaust
20a240df
KW
851 * the list. Normally the loop is executed just once. But if setting the
852 * locale fails, inside the loop we add fallback trials to the array and so
853 * will execute the loop multiple times */
c3fcd832
KW
854 trial_locales[0] = setlocale_init;
855 trial_locales_count = 1;
65ebb059
KW
856 for (i= 0; i < trial_locales_count; i++) {
857 const char * trial_locale = trial_locales[i];
858
859 if (i > 0) {
860
861 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
862 * when i==0, but I (khw) don't think that behavior makes much
863 * sense */
864 setlocale_failure = FALSE;
865
6bce99ee
JH
866#ifdef SYSTEM_DEFAULT_LOCALE
867# ifdef WIN32
65ebb059
KW
868 /* On Windows machines, an entry of "" after the 0th means to use
869 * the system default locale, which we now proceed to get. */
870 if (strEQ(trial_locale, "")) {
871 unsigned int j;
872
873 /* Note that this may change the locale, but we are going to do
874 * that anyway just below */
875 system_default_locale = setlocale(LC_ALL, "");
5d1187d1 876 DEBUG_LOCALE_INIT(LC_ALL, "", system_default_locale);
65ebb059
KW
877
878 /* Skip if invalid or it's already on the list of locales to
879 * try */
880 if (! system_default_locale) {
881 goto next_iteration;
882 }
883 for (j = 0; j < trial_locales_count; j++) {
884 if (strEQ(system_default_locale, trial_locales[j])) {
885 goto next_iteration;
886 }
887 }
888
889 trial_locale = system_default_locale;
890 }
6bce99ee
JH
891# endif /* WIN32 */
892#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
893 }
894
98994639 895#ifdef LC_ALL
5d1187d1
KW
896 sl_result = my_setlocale(LC_ALL, trial_locale);
897 DEBUG_LOCALE_INIT(LC_ALL, trial_locale, sl_result);
898 if (! sl_result) {
49c85077 899 setlocale_failure = TRUE;
7cd8b568
KW
900 }
901 else {
902 /* Since LC_ALL succeeded, it should have changed all the other
903 * categories it can to its value; so we massage things so that the
904 * setlocales below just return their category's current values.
905 * This adequately handles the case in NetBSD where LC_COLLATE may
906 * not be defined for a locale, and setting it individually will
907 * fail, whereas setting LC_ALL suceeds, leaving LC_COLLATE set to
908 * the POSIX locale. */
909 trial_locale = NULL;
910 }
98994639
HS
911#endif /* LC_ALL */
912
49c85077 913 if (!setlocale_failure) {
98994639 914#ifdef USE_LOCALE_CTYPE
49c85077 915 Safefree(curctype);
5d1187d1
KW
916 curctype = my_setlocale(LC_CTYPE, trial_locale);
917 DEBUG_LOCALE_INIT(LC_CTYPE, trial_locale, curctype);
918 if (! curctype)
49c85077
KW
919 setlocale_failure = TRUE;
920 else
921 curctype = savepv(curctype);
98994639
HS
922#endif /* USE_LOCALE_CTYPE */
923#ifdef USE_LOCALE_COLLATE
49c85077 924 Safefree(curcoll);
5d1187d1
KW
925 curcoll = my_setlocale(LC_COLLATE, trial_locale);
926 DEBUG_LOCALE_INIT(LC_COLLATE, trial_locale, curcoll);
927 if (! curcoll)
49c85077
KW
928 setlocale_failure = TRUE;
929 else
930 curcoll = savepv(curcoll);
98994639
HS
931#endif /* USE_LOCALE_COLLATE */
932#ifdef USE_LOCALE_NUMERIC
49c85077 933 Safefree(curnum);
5d1187d1
KW
934 curnum = my_setlocale(LC_NUMERIC, trial_locale);
935 DEBUG_LOCALE_INIT(LC_NUMERIC, trial_locale, curnum);
936 if (! curnum)
49c85077
KW
937 setlocale_failure = TRUE;
938 else
939 curnum = savepv(curnum);
98994639 940#endif /* USE_LOCALE_NUMERIC */
a782673d 941#ifdef USE_LOCALE_MESSAGES
5d1187d1
KW
942 sl_result = my_setlocale(LC_MESSAGES, trial_locale);
943 DEBUG_LOCALE_INIT(LC_MESSAGES, trial_locale, sl_result);
944 if (! (sl_result))
a782673d
KW
945 setlocale_failure = TRUE;
946#endif /* USE_LOCALE_MESSAGES */
c835d6be 947#ifdef USE_LOCALE_MONETARY
5d1187d1
KW
948 sl_result = my_setlocale(LC_MONETARY, trial_locale);
949 DEBUG_LOCALE_INIT(LC_MONETARY, trial_locale, sl_result);
950 if (! (sl_result))
c835d6be
KW
951 setlocale_failure = TRUE;
952#endif /* USE_LOCALE_MONETARY */
953
49c85077
KW
954 if (! setlocale_failure) { /* Success */
955 break;
956 }
65ebb059 957 }
98994639 958
49c85077
KW
959 /* Here, something failed; will need to try a fallback. */
960 ok = 0;
65ebb059 961
49c85077
KW
962 if (i == 0) {
963 unsigned int j;
98994639 964
65ebb059 965 if (locwarn) { /* Output failure info only on the first one */
98994639
HS
966#ifdef LC_ALL
967
49c85077
KW
968 PerlIO_printf(Perl_error_log,
969 "perl: warning: Setting locale failed.\n");
98994639
HS
970
971#else /* !LC_ALL */
972
49c85077
KW
973 PerlIO_printf(Perl_error_log,
974 "perl: warning: Setting locale failed for the categories:\n\t");
20a240df 975# ifdef USE_LOCALE_CTYPE
49c85077
KW
976 if (! curctype)
977 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
20a240df
KW
978# endif /* USE_LOCALE_CTYPE */
979# ifdef USE_LOCALE_COLLATE
49c85077
KW
980 if (! curcoll)
981 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
20a240df
KW
982# endif /* USE_LOCALE_COLLATE */
983# ifdef USE_LOCALE_NUMERIC
49c85077
KW
984 if (! curnum)
985 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
20a240df 986# endif /* USE_LOCALE_NUMERIC */
a782673d 987 PerlIO_printf(Perl_error_log, "and possibly others\n");
98994639
HS
988
989#endif /* LC_ALL */
990
49c85077
KW
991 PerlIO_printf(Perl_error_log,
992 "perl: warning: Please check that your locale settings:\n");
98994639
HS
993
994#ifdef __GLIBC__
49c85077
KW
995 PerlIO_printf(Perl_error_log,
996 "\tLANGUAGE = %c%s%c,\n",
997 language ? '"' : '(',
998 language ? language : "unset",
999 language ? '"' : ')');
98994639
HS
1000#endif
1001
49c85077
KW
1002 PerlIO_printf(Perl_error_log,
1003 "\tLC_ALL = %c%s%c,\n",
1004 lc_all ? '"' : '(',
1005 lc_all ? lc_all : "unset",
1006 lc_all ? '"' : ')');
98994639
HS
1007
1008#if defined(USE_ENVIRON_ARRAY)
49c85077
KW
1009 {
1010 char **e;
1011 for (e = environ; *e; e++) {
1012 if (strnEQ(*e, "LC_", 3)
1013 && strnNE(*e, "LC_ALL=", 7)
1014 && (p = strchr(*e, '=')))
1015 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
1016 (int)(p - *e), *e, p + 1);
1017 }
1018 }
98994639 1019#else
49c85077
KW
1020 PerlIO_printf(Perl_error_log,
1021 "\t(possibly more locale environment variables)\n");
98994639
HS
1022#endif
1023
49c85077
KW
1024 PerlIO_printf(Perl_error_log,
1025 "\tLANG = %c%s%c\n",
1026 lang ? '"' : '(',
1027 lang ? lang : "unset",
1028 lang ? '"' : ')');
98994639 1029
49c85077
KW
1030 PerlIO_printf(Perl_error_log,
1031 " are supported and installed on your system.\n");
1032 }
98994639 1033
65ebb059 1034 /* Calculate what fallback locales to try. We have avoided this
f6bab5f6 1035 * until we have to, because failure is quite unlikely. This will
65ebb059
KW
1036 * usually change the upper bound of the loop we are in.
1037 *
1038 * Since the system's default way of setting the locale has not
1039 * found one that works, We use Perl's defined ordering: LC_ALL,
1040 * LANG, and the C locale. We don't try the same locale twice, so
1041 * don't add to the list if already there. (On POSIX systems, the
1042 * LC_ALL element will likely be a repeat of the 0th element "",
6b058d42
KW
1043 * but there's no harm done by doing it explicitly.
1044 *
1045 * Note that this tries the LC_ALL environment variable even on
1046 * systems which have no LC_ALL locale setting. This may or may
1047 * not have been originally intentional, but there's no real need
1048 * to change the behavior. */
65ebb059
KW
1049 if (lc_all) {
1050 for (j = 0; j < trial_locales_count; j++) {
1051 if (strEQ(lc_all, trial_locales[j])) {
1052 goto done_lc_all;
1053 }
1054 }
1055 trial_locales[trial_locales_count++] = lc_all;
1056 }
1057 done_lc_all:
98994639 1058
65ebb059
KW
1059 if (lang) {
1060 for (j = 0; j < trial_locales_count; j++) {
1061 if (strEQ(lang, trial_locales[j])) {
1062 goto done_lang;
1063 }
1064 }
1065 trial_locales[trial_locales_count++] = lang;
1066 }
1067 done_lang:
1068
1069#if defined(WIN32) && defined(LC_ALL)
1070 /* For Windows, we also try the system default locale before "C".
1071 * (If there exists a Windows without LC_ALL we skip this because
1072 * it gets too complicated. For those, the "C" is the next
1073 * fallback possibility). The "" is the same as the 0th element of
1074 * the array, but the code at the loop above knows to treat it
1075 * differently when not the 0th */
1076 trial_locales[trial_locales_count++] = "";
1077#endif
1078
1079 for (j = 0; j < trial_locales_count; j++) {
1080 if (strEQ("C", trial_locales[j])) {
1081 goto done_C;
1082 }
1083 }
1084 trial_locales[trial_locales_count++] = "C";
98994639 1085
65ebb059
KW
1086 done_C: ;
1087 } /* end of first time through the loop */
98994639 1088
65ebb059
KW
1089#ifdef WIN32
1090 next_iteration: ;
1091#endif
1092
1093 } /* end of looping through the trial locales */
1094
1095 if (ok < 1) { /* If we tried to fallback */
1096 const char* msg;
1097 if (! setlocale_failure) { /* fallback succeeded */
1098 msg = "Falling back to";
1099 }
1100 else { /* fallback failed */
98994639 1101
65ebb059
KW
1102 /* We dropped off the end of the loop, so have to decrement i to
1103 * get back to the value the last time through */
1104 i--;
98994639 1105
65ebb059
KW
1106 ok = -1;
1107 msg = "Failed to fall back to";
1108
1109 /* To continue, we should use whatever values we've got */
98994639 1110#ifdef USE_LOCALE_CTYPE
49c85077
KW
1111 Safefree(curctype);
1112 curctype = savepv(setlocale(LC_CTYPE, NULL));
5d1187d1 1113 DEBUG_LOCALE_INIT(LC_CTYPE, NULL, curctype);
98994639
HS
1114#endif /* USE_LOCALE_CTYPE */
1115#ifdef USE_LOCALE_COLLATE
49c85077
KW
1116 Safefree(curcoll);
1117 curcoll = savepv(setlocale(LC_COLLATE, NULL));
5d1187d1 1118 DEBUG_LOCALE_INIT(LC_COLLATE, NULL, curcoll);
98994639
HS
1119#endif /* USE_LOCALE_COLLATE */
1120#ifdef USE_LOCALE_NUMERIC
49c85077
KW
1121 Safefree(curnum);
1122 curnum = savepv(setlocale(LC_NUMERIC, NULL));
5d1187d1 1123 DEBUG_LOCALE_INIT(LC_NUMERIC, NULL, curnum);
98994639 1124#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
1125 }
1126
1127 if (locwarn) {
1128 const char * description;
1129 const char * name = "";
1130 if (strEQ(trial_locales[i], "C")) {
1131 description = "the standard locale";
1132 name = "C";
1133 }
6bce99ee 1134#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059
KW
1135 else if (strEQ(trial_locales[i], "")) {
1136 description = "the system default locale";
1137 if (system_default_locale) {
1138 name = system_default_locale;
1139 }
1140 }
6bce99ee 1141#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
1142 else {
1143 description = "a fallback locale";
1144 name = trial_locales[i];
1145 }
1146 if (name && strNE(name, "")) {
1147 PerlIO_printf(Perl_error_log,
1148 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
1149 }
1150 else {
1151 PerlIO_printf(Perl_error_log,
1152 "perl: warning: %s %s.\n", msg, description);
1153 }
1154 }
1155 } /* End of tried to fallback */
98994639
HS
1156
1157#ifdef USE_LOCALE_CTYPE
1158 new_ctype(curctype);
1159#endif /* USE_LOCALE_CTYPE */
1160
1161#ifdef USE_LOCALE_COLLATE
1162 new_collate(curcoll);
1163#endif /* USE_LOCALE_COLLATE */
1164
1165#ifdef USE_LOCALE_NUMERIC
1166 new_numeric(curnum);
1167#endif /* USE_LOCALE_NUMERIC */
b310b053 1168
8ef6e574 1169#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
49c85077
KW
1170 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
1171 * locale is UTF-8. If PL_utf8locale and PL_unicode (set by -C or by
1172 * $ENV{PERL_UNICODE}) are true, perl.c:S_parse_body() will turn on the
1173 * PerlIO :utf8 layer on STDIN, STDOUT, STDERR, _and_ the default open
1174 * discipline. */
c1284011 1175 PL_utf8locale = _is_cur_LC_category_utf8(LC_CTYPE);
49c85077 1176
a05d7ebb 1177 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
1178 This is an alternative to using the -C command line switch
1179 (the -C if present will override this). */
1180 {
dd374669 1181 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 1182 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
1183 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
1184 PL_utf8cache = -1;
b310b053 1185 }
ec71e770 1186#endif
b310b053 1187
98994639 1188#ifdef USE_LOCALE_CTYPE
43c5f42d 1189 Safefree(curctype);
98994639
HS
1190#endif /* USE_LOCALE_CTYPE */
1191#ifdef USE_LOCALE_COLLATE
43c5f42d 1192 Safefree(curcoll);
98994639
HS
1193#endif /* USE_LOCALE_COLLATE */
1194#ifdef USE_LOCALE_NUMERIC
43c5f42d 1195 Safefree(curnum);
98994639 1196#endif /* USE_LOCALE_NUMERIC */
8ef6e574 1197
175c4cf9
KW
1198#ifdef __GLIBC__
1199 Safefree(language);
1200#endif
1201
1202 Safefree(lc_all);
1203 Safefree(lang);
1204
e3305790
KW
1205#else /* !USE_LOCALE */
1206 PERL_UNUSED_ARG(printwarn);
1207#endif /* USE_LOCALE */
1208
98994639
HS
1209 return ok;
1210}
1211
49c85077 1212
98994639
HS
1213#ifdef USE_LOCALE_COLLATE
1214
1215/*
1216 * mem_collxfrm() is a bit like strxfrm() but with two important
1217 * differences. First, it handles embedded NULs. Second, it allocates
1218 * a bit more memory than needed for the transformed data itself.
1219 * The real transformed data begins at offset sizeof(collationix).
1220 * Please see sv_collxfrm() to see how this is used.
1221 */
166f8a29 1222
98994639
HS
1223char *
1224Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
1225{
1226 char *xbuf;
1227 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
1228
7918f24d
NC
1229 PERL_ARGS_ASSERT_MEM_COLLXFRM;
1230
98994639
HS
1231 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
1232 /* the +1 is for the terminating NUL. */
1233
1234 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 1235 Newx(xbuf, xAlloc, char);
98994639
HS
1236 if (! xbuf)
1237 goto bad;
1238
1239 *(U32*)xbuf = PL_collation_ix;
1240 xout = sizeof(PL_collation_ix);
1241 for (xin = 0; xin < len; ) {
224e8ef5 1242 Size_t xused;
98994639
HS
1243
1244 for (;;) {
1245 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 1246 if (xused >= PERL_INT_MAX)
98994639 1247 goto bad;
eb160463 1248 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
1249 break;
1250 xAlloc = (2 * xAlloc) + 1;
1251 Renew(xbuf, xAlloc, char);
1252 if (! xbuf)
1253 goto bad;
1254 }
1255
1256 xin += strlen(s + xin) + 1;
1257 xout += xused;
1258
1259 /* Embedded NULs are understood but silently skipped
1260 * because they make no sense in locale collation. */
1261 }
1262
1263 xbuf[xout] = '\0';
1264 *xlen = xout - sizeof(PL_collation_ix);
1265 return xbuf;
1266
1267 bad:
1268 Safefree(xbuf);
1269 *xlen = 0;
1270 return NULL;
1271}
1272
1273#endif /* USE_LOCALE_COLLATE */
1274
8ef6e574
KW
1275#ifdef USE_LOCALE
1276
c1284011
KW
1277bool
1278Perl__is_cur_LC_category_utf8(pTHX_ int category)
7d74bb61
KW
1279{
1280 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
1281 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b 1282 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
609548d2
KW
1283 * could give the wrong result. The result will very likely be correct for
1284 * languages that have commonly used non-ASCII characters, but for notably
1285 * English, it comes down to if the locale's name ends in something like
1286 * "UTF-8". It errs on the side of not being a UTF-8 locale. */
7d74bb61
KW
1287
1288 char *save_input_locale = NULL;
7d74bb61
KW
1289 STRLEN final_pos;
1290
8ef6e574 1291#ifdef LC_ALL
7d74bb61 1292 assert(category != LC_ALL);
8ef6e574 1293#endif
7d74bb61
KW
1294
1295 /* First dispose of the trivial cases */
b07fffd1 1296 save_input_locale = setlocale(category, NULL);
7d74bb61 1297 if (! save_input_locale) {
69014004
KW
1298 DEBUG_L(PerlIO_printf(Perl_debug_log,
1299 "Could not find current locale for category %d\n",
1300 category));
7d74bb61
KW
1301 return FALSE; /* XXX maybe should croak */
1302 }
b07fffd1 1303 save_input_locale = stdize_locale(savepv(save_input_locale));
a39edc4c 1304 if (isNAME_C_OR_POSIX(save_input_locale)) {
69014004
KW
1305 DEBUG_L(PerlIO_printf(Perl_debug_log,
1306 "Current locale for category %d is %s\n",
1307 category, save_input_locale));
b07fffd1 1308 Safefree(save_input_locale);
7d74bb61
KW
1309 return FALSE;
1310 }
1311
1d958db2
KW
1312#if defined(USE_LOCALE_CTYPE) \
1313 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 1314
1d958db2 1315 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
1316
1317 char *save_ctype_locale = NULL;
119ee68b 1318 bool is_utf8;
7d74bb61 1319
119ee68b 1320 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
1321
1322 /* Get the current LC_CTYPE locale */
4f72bb37 1323 save_ctype_locale = setlocale(LC_CTYPE, NULL);
7d74bb61 1324 if (! save_ctype_locale) {
69014004
KW
1325 DEBUG_L(PerlIO_printf(Perl_debug_log,
1326 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
1327 goto cant_use_nllanginfo;
1328 }
4f72bb37 1329 save_ctype_locale = stdize_locale(savepv(save_ctype_locale));
7d74bb61
KW
1330
1331 /* If LC_CTYPE and the desired category use the same locale, this
1332 * means that finding the value for LC_CTYPE is the same as finding
1333 * the value for the desired category. Otherwise, switch LC_CTYPE
1334 * to the desired category's locale */
1335 if (strEQ(save_ctype_locale, save_input_locale)) {
1336 Safefree(save_ctype_locale);
1337 save_ctype_locale = NULL;
1338 }
1339 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
1340 DEBUG_L(PerlIO_printf(Perl_debug_log,
1341 "Could not change LC_CTYPE locale to %s\n",
1342 save_input_locale));
7d74bb61
KW
1343 Safefree(save_ctype_locale);
1344 goto cant_use_nllanginfo;
1345 }
1346 }
1347
69014004
KW
1348 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
1349 save_input_locale));
1350
7d74bb61 1351 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
1352 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
1353 * should give the correct results */
119ee68b 1354
1d958db2
KW
1355# if defined(HAS_NL_LANGINFO) && defined(CODESET)
1356 {
4f72bb37 1357 char *codeset = nl_langinfo(CODESET);
1d958db2 1358 if (codeset && strNE(codeset, "")) {
4f72bb37 1359 codeset = savepv(codeset);
119ee68b 1360
1d958db2
KW
1361 /* If we switched LC_CTYPE, switch back */
1362 if (save_ctype_locale) {
1363 setlocale(LC_CTYPE, save_ctype_locale);
1364 Safefree(save_ctype_locale);
1365 }
1366
1367 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
1368 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
1369
69014004
KW
1370 DEBUG_L(PerlIO_printf(Perl_debug_log,
1371 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
1372 codeset, is_utf8));
1d958db2
KW
1373 Safefree(codeset);
1374 Safefree(save_input_locale);
1375 return is_utf8;
1376 }
119ee68b
KW
1377 }
1378
1d958db2
KW
1379# endif
1380# ifdef MB_CUR_MAX
1381
1382 /* Here, either we don't have nl_langinfo, or it didn't return a
1383 * codeset. Try MB_CUR_MAX */
1384
119ee68b
KW
1385 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
1386 * Unicode code point. Since UTF-8 is the only non-single byte
1387 * encoding we handle, we just say any such encoding is UTF-8, and if
1388 * turns out to be wrong, other things will fail */
1389 is_utf8 = MB_CUR_MAX >= 4;
1390
69014004
KW
1391 DEBUG_L(PerlIO_printf(Perl_debug_log,
1392 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
1393 (int) MB_CUR_MAX, is_utf8));
1394
119ee68b
KW
1395 Safefree(save_input_locale);
1396
1397# ifdef HAS_MBTOWC
1398
1399 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
1400 * since they are both in the C99 standard. We can feed a known byte
1401 * string to the latter function, and check that it gives the expected
1402 * result */
1403 if (is_utf8) {
1404 wchar_t wc;
856b881c 1405 PERL_UNUSED_RESULT(mbtowc(&wc, NULL, 0));/* Reset any shift state */
69014004 1406 errno = 0;
f019f68f 1407 if ((size_t)mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
119ee68b
KW
1408 != strlen(HYPHEN_UTF8)
1409 || wc != (wchar_t) 0x2010)
1410 {
1411 is_utf8 = FALSE;
abdcbdb8 1412 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", (unsigned int)wc));
69014004
KW
1413 DEBUG_L(PerlIO_printf(Perl_debug_log,
1414 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
1415 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
1416 }
1417 }
119ee68b
KW
1418# endif
1419
1d958db2
KW
1420 /* If we switched LC_CTYPE, switch back */
1421 if (save_ctype_locale) {
1422 setlocale(LC_CTYPE, save_ctype_locale);
1423 Safefree(save_ctype_locale);
119ee68b 1424 }
7d74bb61 1425
1d958db2 1426 return is_utf8;
119ee68b 1427# endif
7d74bb61 1428 }
119ee68b 1429
7d74bb61
KW
1430 cant_use_nllanginfo:
1431
0080c90a
KW
1432#else /* nl_langinfo should work if available, so don't bother compiling this
1433 fallback code. The final fallback of looking at the name is
1434 compiled, and will be executed if nl_langinfo fails */
7d74bb61 1435
97f4de96
KW
1436 /* nl_langinfo not available or failed somehow. Next try looking at the
1437 * currency symbol to see if it disambiguates things. Often that will be
1438 * in the native script, and if the symbol isn't in UTF-8, we know that the
1439 * locale isn't. If it is non-ASCII UTF-8, we infer that the locale is
609548d2
KW
1440 * too, as the odds of a non-UTF8 string being valid UTF-8 are quite small
1441 * */
fa9b773e
KW
1442
1443#ifdef HAS_LOCALECONV
fa9b773e 1444# ifdef USE_LOCALE_MONETARY
fa9b773e
KW
1445 {
1446 char *save_monetary_locale = NULL;
fa9b773e 1447 bool only_ascii = FALSE;
13542a67
KW
1448 bool is_utf8 = FALSE;
1449 struct lconv* lc;
fa9b773e 1450
97f4de96
KW
1451 /* Like above for LC_CTYPE, we first set LC_MONETARY to the locale of
1452 * the desired category, if it isn't that locale already */
1453
fa9b773e
KW
1454 if (category != LC_MONETARY) {
1455
4f72bb37 1456 save_monetary_locale = setlocale(LC_MONETARY, NULL);
fa9b773e 1457 if (! save_monetary_locale) {
69014004
KW
1458 DEBUG_L(PerlIO_printf(Perl_debug_log,
1459 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1460 goto cant_use_monetary;
1461 }
4f72bb37 1462 save_monetary_locale = stdize_locale(savepv(save_monetary_locale));
fa9b773e 1463
13542a67
KW
1464 if (strEQ(save_monetary_locale, save_input_locale)) {
1465 Safefree(save_monetary_locale);
1466 save_monetary_locale = NULL;
1467 }
1468 else if (! setlocale(LC_MONETARY, save_input_locale)) {
59c234b4
KW
1469 DEBUG_L(PerlIO_printf(Perl_debug_log,
1470 "Could not change LC_MONETARY locale to %s\n",
1471 save_input_locale));
1472 Safefree(save_monetary_locale);
1473 goto cant_use_monetary;
fa9b773e
KW
1474 }
1475 }
1476
1477 /* Here the current LC_MONETARY is set to the locale of the category
1478 * whose information is desired. */
1479
13542a67
KW
1480 lc = localeconv();
1481 if (! lc
1482 || ! lc->currency_symbol
9f10db87 1483 || is_invariant_string((U8 *) lc->currency_symbol, 0))
13542a67
KW
1484 {
1485 DEBUG_L(PerlIO_printf(Perl_debug_log, "Couldn't get currency symbol for %s, or contains only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
1486 only_ascii = TRUE;
1487 }
1488 else {
1489 is_utf8 = is_utf8_string((U8 *) lc->currency_symbol, 0);
fa9b773e
KW
1490 }
1491
1492 /* If we changed it, restore LC_MONETARY to its original locale */
1493 if (save_monetary_locale) {
1494 setlocale(LC_MONETARY, save_monetary_locale);
1495 Safefree(save_monetary_locale);
1496 }
1497
13542a67 1498 if (! only_ascii) {
fa9b773e 1499
59c234b4
KW
1500 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8;
1501 * otherwise assume the locale is UTF-8 if and only if the symbol
1502 * is non-ascii UTF-8. */
1503 DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?Currency symbol for %s is UTF-8=%d\n",
1504 save_input_locale, is_utf8));
1505 Safefree(save_input_locale);
1506 return is_utf8;
13542a67 1507 }
fa9b773e
KW
1508 }
1509 cant_use_monetary:
1510
1511# endif /* USE_LOCALE_MONETARY */
1512#endif /* HAS_LOCALECONV */
1513
15f7e74e
KW
1514#if defined(HAS_STRFTIME) && defined(USE_LOCALE_TIME)
1515
1516/* Still haven't found a non-ASCII string to disambiguate UTF-8 or not. Try
1517 * the names of the months and weekdays, timezone, and am/pm indicator */
1518 {
1519 char *save_time_locale = NULL;
1520 int hour = 10;
1521 bool is_dst = FALSE;
1522 int dom = 1;
1523 int month = 0;
1524 int i;
1525 char * formatted_time;
1526
1527
1528 /* Like above for LC_MONETARY, we set LC_TIME to the locale of the
1529 * desired category, if it isn't that locale already */
1530
1531 if (category != LC_TIME) {
1532
1533 save_time_locale = setlocale(LC_TIME, NULL);
1534 if (! save_time_locale) {
1535 DEBUG_L(PerlIO_printf(Perl_debug_log,
1536 "Could not find current locale for LC_TIME\n"));
1537 goto cant_use_time;
1538 }
1539 save_time_locale = stdize_locale(savepv(save_time_locale));
1540
1541 if (strEQ(save_time_locale, save_input_locale)) {
1542 Safefree(save_time_locale);
1543 save_time_locale = NULL;
1544 }
1545 else if (! setlocale(LC_TIME, save_input_locale)) {
1546 DEBUG_L(PerlIO_printf(Perl_debug_log,
1547 "Could not change LC_TIME locale to %s\n",
1548 save_input_locale));
1549 Safefree(save_time_locale);
1550 goto cant_use_time;
1551 }
1552 }
1553
1554 /* Here the current LC_TIME is set to the locale of the category
1555 * whose information is desired. Look at all the days of the week and
9f10db87 1556 * month names, and the timezone and am/pm indicator for UTF-8 variant
15f7e74e
KW
1557 * characters. The first such a one found will tell us if the locale
1558 * is UTF-8 or not */
1559
1560 for (i = 0; i < 7 + 12; i++) { /* 7 days; 12 months */
1561 formatted_time = my_strftime("%A %B %Z %p",
1562 0, 0, hour, dom, month, 112, 0, 0, is_dst);
9f10db87 1563 if (! formatted_time || is_invariant_string((U8 *) formatted_time, 0)) {
15f7e74e
KW
1564
1565 /* Here, we didn't find a non-ASCII. Try the next time through
1566 * with the complemented dst and am/pm, and try with the next
1567 * weekday. After we have gotten all weekdays, try the next
1568 * month */
1569 is_dst = ! is_dst;
1570 hour = (hour + 12) % 24;
1571 dom++;
1572 if (i > 6) {
1573 month++;
1574 }
1575 continue;
1576 }
1577
1578 /* Here, we have a non-ASCII. Return TRUE is it is valid UTF8;
1579 * false otherwise. But first, restore LC_TIME to its original
1580 * locale if we changed it */
1581 if (save_time_locale) {
1582 setlocale(LC_TIME, save_time_locale);
1583 Safefree(save_time_locale);
1584 }
1585
1586 DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?time-related strings for %s are UTF-8=%d\n",
1587 save_input_locale,
1588 is_utf8_string((U8 *) formatted_time, 0)));
1589 Safefree(save_input_locale);
1590 return is_utf8_string((U8 *) formatted_time, 0);
1591 }
1592
1593 /* Falling off the end of the loop indicates all the names were just
1594 * ASCII. Go on to the next test. If we changed it, restore LC_TIME
1595 * to its original locale */
1596 if (save_time_locale) {
1597 setlocale(LC_TIME, save_time_locale);
1598 Safefree(save_time_locale);
1599 }
1600 DEBUG_L(PerlIO_printf(Perl_debug_log, "All time-related words for %s contain only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
1601 }
1602 cant_use_time:
1603
1604#endif
1605
1606#if 0 && defined(USE_LOCALE_MESSAGES) && defined(HAS_SYS_ERRLIST)
855aeb93
JH
1607
1608/* This code is ifdefd out because it was found to not be necessary in testing
5857e934
KW
1609 * on our dromedary test machine, which has over 700 locales. There, this
1610 * added no value to looking at the currency symbol and the time strings. I
1611 * left it in so as to avoid rewriting it if real-world experience indicates
1612 * that dromedary is an outlier. Essentially, instead of returning abpve if we
855aeb93
JH
1613 * haven't found illegal utf8, we continue on and examine all the strerror()
1614 * messages on the platform for utf8ness. If all are ASCII, we still don't
1615 * know the answer; but otherwise we have a pretty good indication of the
5857e934
KW
1616 * utf8ness. The reason this doesn't help much is that the messages may not
1617 * have been translated into the locale. The currency symbol and time strings
1618 * are much more likely to have been translated. */
1619 {
855aeb93 1620 int e;
5857e934
KW
1621 bool is_utf8 = FALSE;
1622 bool non_ascii = FALSE;
855aeb93 1623 char *save_messages_locale = NULL;
5857e934 1624 const char * errmsg = NULL;
855aeb93 1625
5857e934
KW
1626 /* Like above, we set LC_MESSAGES to the locale of the desired
1627 * category, if it isn't that locale already */
855aeb93
JH
1628
1629 if (category != LC_MESSAGES) {
1630
5857e934 1631 save_messages_locale = setlocale(LC_MESSAGES, NULL);
855aeb93 1632 if (! save_messages_locale) {
5857e934
KW
1633 DEBUG_L(PerlIO_printf(Perl_debug_log,
1634 "Could not find current locale for LC_MESSAGES\n"));
855aeb93
JH
1635 goto cant_use_messages;
1636 }
5857e934 1637 save_messages_locale = stdize_locale(savepv(save_messages_locale));
855aeb93
JH
1638
1639 if (strEQ(save_messages_locale, save_input_locale)) {
5857e934
KW
1640 Safefree(save_messages_locale);
1641 save_messages_locale = NULL;
855aeb93
JH
1642 }
1643 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
5857e934
KW
1644 DEBUG_L(PerlIO_printf(Perl_debug_log,
1645 "Could not change LC_MESSAGES locale to %s\n",
1646 save_input_locale));
855aeb93
JH
1647 Safefree(save_messages_locale);
1648 goto cant_use_messages;
1649 }
1650 }
1651
1652 /* Here the current LC_MESSAGES is set to the locale of the category
5857e934
KW
1653 * whose information is desired. Look through all the messages. We
1654 * can't use Strerror() here because it may expand to code that
1655 * segfaults in miniperl */
855aeb93 1656
5857e934
KW
1657 for (e = 0; e <= sys_nerr; e++) {
1658 errno = 0;
1659 errmsg = sys_errlist[e];
1660 if (errno || !errmsg) {
855aeb93
JH
1661 break;
1662 }
5857e934 1663 errmsg = savepv(errmsg);
9f10db87 1664 if (! is_invariant_string((U8 *) errmsg, 0)) {
5857e934
KW
1665 non_ascii = TRUE;
1666 is_utf8 = is_utf8_string((U8 *) errmsg, 0);
1667 break;
855aeb93
JH
1668 }
1669 }
5857e934 1670 Safefree(errmsg);
855aeb93
JH
1671
1672 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1673 if (save_messages_locale) {
1674 setlocale(LC_MESSAGES, save_messages_locale);
1675 Safefree(save_messages_locale);
1676 }
1677
5857e934
KW
1678 if (non_ascii) {
1679
1680 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1681 * any non-ascii means it is one; otherwise we assume it isn't */
1682 DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?error messages for %s are UTF-8=%d\n",
1683 save_input_locale,
1684 is_utf8));
1685 Safefree(save_input_locale);
1686 return is_utf8;
1687 }
855aeb93 1688
5857e934 1689 DEBUG_L(PerlIO_printf(Perl_debug_log, "All error messages for %s contain only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
855aeb93
JH
1690 }
1691 cant_use_messages:
1692
1693#endif
fa9b773e 1694
0080c90a
KW
1695#endif /* the code that is compiled when no nl_langinfo */
1696
92c0a900
KW
1697#ifndef EBCDIC /* On os390, even if the name ends with "UTF-8', it isn't a
1698 UTF-8 locale */
97f4de96
KW
1699 /* As a last resort, look at the locale name to see if it matches
1700 * qr/UTF -? * 8 /ix, or some other common locale names. This "name", the
1701 * return of setlocale(), is actually defined to be opaque, so we can't
1702 * really rely on the absence of various substrings in the name to indicate
1703 * its UTF-8ness, but if it has UTF8 in the name, it is extremely likely to
1704 * be a UTF-8 locale. Similarly for the other common names */
1705
1706 final_pos = strlen(save_input_locale) - 1;
1707 if (final_pos >= 3) {
1708 char *name = save_input_locale;
1709
1710 /* Find next 'U' or 'u' and look from there */
1711 while ((name += strcspn(name, "Uu") + 1)
1712 <= save_input_locale + final_pos - 2)
1713 {
305b8651
KW
1714 if (!isALPHA_FOLD_NE(*name, 't')
1715 || isALPHA_FOLD_NE(*(name + 1), 'f'))
97f4de96
KW
1716 {
1717 continue;
1718 }
1719 name += 2;
1720 if (*(name) == '-') {
1721 if ((name > save_input_locale + final_pos - 1)) {
1722 break;
1723 }
1724 name++;
1725 }
1726 if (*(name) == '8') {
97f4de96
KW
1727 DEBUG_L(PerlIO_printf(Perl_debug_log,
1728 "Locale %s ends with UTF-8 in name\n",
1729 save_input_locale));
00c54b9c 1730 Safefree(save_input_locale);
97f4de96
KW
1731 return TRUE;
1732 }
1733 }
1734 DEBUG_L(PerlIO_printf(Perl_debug_log,
1735 "Locale %s doesn't end with UTF-8 in name\n",
1736 save_input_locale));
1737 }
92c0a900 1738#endif
97f4de96
KW
1739
1740#ifdef WIN32
1741 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
1742 if (final_pos >= 4
1743 && *(save_input_locale + final_pos - 0) == '1'
1744 && *(save_input_locale + final_pos - 1) == '0'
1745 && *(save_input_locale + final_pos - 2) == '0'
1746 && *(save_input_locale + final_pos - 3) == '5'
1747 && *(save_input_locale + final_pos - 4) == '6')
1748 {
1749 DEBUG_L(PerlIO_printf(Perl_debug_log,
1750 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
1751 save_input_locale));
1752 Safefree(save_input_locale);
1753 return TRUE;
1754 }
1755#endif
1756
1757 /* Other common encodings are the ISO 8859 series, which aren't UTF-8. But
1758 * since we are about to return FALSE anyway, there is no point in doing
1759 * this extra work */
1760#if 0
1761 if (instr(save_input_locale, "8859")) {
1762 DEBUG_L(PerlIO_printf(Perl_debug_log,
1763 "Locale %s has 8859 in name, not UTF-8 locale\n",
1764 save_input_locale));
1765 Safefree(save_input_locale);
1766 return FALSE;
1767 }
1768#endif
1769
69014004
KW
1770 DEBUG_L(PerlIO_printf(Perl_debug_log,
1771 "Assuming locale %s is not a UTF-8 locale\n",
1772 save_input_locale));
fa9b773e 1773 Safefree(save_input_locale);
7d74bb61
KW
1774 return FALSE;
1775}
1776
8ef6e574 1777#endif
7d74bb61 1778
d6ded950
KW
1779
1780bool
1781Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
1782{
1a4f13e1 1783 dVAR;
d6ded950
KW
1784 /* Internal function which returns if we are in the scope of a pragma that
1785 * enables the locale category 'category'. 'compiling' should indicate if
1786 * this is during the compilation phase (TRUE) or not (FALSE). */
1787
1788 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
1789
1790 SV *categories = cop_hints_fetch_pvs(cop, "locale", 0);
1791 if (! categories || categories == &PL_sv_placeholder) {
1792 return FALSE;
1793 }
1794
1795 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
1796 * a valid unsigned */
1797 assert(category >= -1);
1798 return cBOOL(SvUV(categories) & (1U << (category + 1)));
1799}
1800
2c6ee1a7
KW
1801char *
1802Perl_my_strerror(pTHX_ const int errnum) {
1803
1804 /* Uses C locale for the error text unless within scope of 'use locale' for
1805 * LC_MESSAGES */
1806
1807#ifdef USE_LOCALE_MESSAGES
8ee4b769 1808 if (! IN_LC(LC_MESSAGES)) {
2c6ee1a7 1809 char * save_locale = setlocale(LC_MESSAGES, NULL);
a39edc4c 1810 if (! isNAME_C_OR_POSIX(save_locale)) {
2c6ee1a7
KW
1811 char *errstr;
1812
1813 /* The next setlocale likely will zap this, so create a copy */
1814 save_locale = savepv(save_locale);
1815
1816 setlocale(LC_MESSAGES, "C");
1817
1818 /* This points to the static space in Strerror, with all its
1819 * limitations */
1820 errstr = Strerror(errnum);
1821
1822 setlocale(LC_MESSAGES, save_locale);
1823 Safefree(save_locale);
1824 return errstr;
1825 }
1826 }
1827#endif
1828
1829 return Strerror(errnum);
1830}
1831
66610fdd 1832/*
747c467a
KW
1833
1834=head1 Locale-related functions and macros
1835
1836=for apidoc sync_locale
1837
1838Changing the program's locale should be avoided by XS code. Nevertheless,
1839certain non-Perl libraries called from XS, such as C<Gtk> do so. When this
1840happens, Perl needs to be told that the locale has changed. Use this function
1841to do so, before returning to Perl.
1842
1843=cut
1844*/
1845
1846void
1847Perl_sync_locale(pTHX)
1848{
1849
1850#ifdef USE_LOCALE_CTYPE
1851 new_ctype(setlocale(LC_CTYPE, NULL));
1852#endif /* USE_LOCALE_CTYPE */
1853
1854#ifdef USE_LOCALE_COLLATE
1855 new_collate(setlocale(LC_COLLATE, NULL));
1856#endif
1857
1858#ifdef USE_LOCALE_NUMERIC
1859 set_numeric_local(); /* Switch from "C" to underlying LC_NUMERIC */
1860 new_numeric(setlocale(LC_NUMERIC, NULL));
1861#endif /* USE_LOCALE_NUMERIC */
1862
1863}
1864
5d1187d1
KW
1865#if defined(DEBUGGING) && defined(USE_LOCALE)
1866
1867char *
1868Perl__setlocale_debug_string(const int category, /* category number,
1869 like LC_ALL */
1870 const char* const locale, /* locale name */
1871
1872 /* return value from setlocale() when attempting to
1873 * set 'category' to 'locale' */
1874 const char* const retval)
1875{
1876 /* Returns a pointer to a NUL-terminated string in static storage with
1877 * added text about the info passed in. This is not thread safe and will
1878 * be overwritten by the next call, so this should be used just to
1879 * formulate a string to immediately print or savepv() on.
1880 *
1881 * Buffer overflow checking is done only after the fact (via an assert),
1882 * because this is used only in DEBUGGING, and an attacker would have to
1883 * control the start up of perl with the correct environment variable or
1884 * command line option. */
1885
1886 static char ret[128] = "";
1887
1888 strcpy(ret, "setlocale(");
1889
1890 switch (category) {
1891 default:
1892 sprintf(ret, "%s? %d", ret, category);
1893 break;
1894# ifdef LC_ALL
1895 case LC_ALL:
1896 strcat(ret, "LC_ALL");
1897 break;
1898# endif
1899# ifdef LC_CTYPE
1900 case LC_CTYPE:
1901 strcat(ret, "LC_CTYPE");
1902 break;
1903# endif
1904# ifdef LC_NUMERIC
1905 case LC_NUMERIC:
1906 strcat(ret, "LC_NUMERIC");
1907 break;
1908# endif
1909# ifdef LC_COLLATE
1910 case LC_COLLATE:
1911 strcat(ret, "LC_COLLATE");
1912 break;
1913# endif
1914# ifdef LC_TIME
1915 case LC_TIME:
1916 strcat(ret, "LC_TIME");
1917 break;
1918# endif
1919# ifdef LC_MONETARY
1920 case LC_MONETARY:
1921 strcat(ret, "LC_MONETARY");
1922 break;
1923# endif
1924# ifdef LC_MESSAGES
1925 case LC_MESSAGES:
1926 strcat(ret, "LC_MESSAGES");
1927 break;
1928# endif
1929 }
1930
1931 strcat(ret, ", ");
1932
1933 if (locale) {
1934 strcat(ret, "\"");
1935 strcat(ret, locale);
1936 strcat(ret, "\"");
1937 }
1938 else {
1939 strcat(ret, "NULL");
1940 }
1941
1942 strcat(ret, ") returned ");
1943
1944 if (retval) {
1945 strcat(ret, "\"");
1946 strcat(ret, retval);
1947 strcat(ret, "\"");
1948 }
1949 else {
1950 strcat(ret, "NULL");
1951 }
1952
1953 assert(strlen(ret) < sizeof(ret));
1954
1955 return ret;
1956}
1957
1958#endif
747c467a
KW
1959
1960
1961/*
14d04a33 1962 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1963 */