This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add code for debugging locale initialization
[perl5.git] / locale.c
CommitLineData
98994639
HS
1/* locale.c
2 *
1129b882
NC
3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
4 * 2002, 2003, 2005, 2006, 2007, 2008 by Larry Wall and others
98994639
HS
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
9 */
10
11/*
4ac71550 12 * A Elbereth Gilthoniel,
cdad3b53 13 * silivren penna míriel
4ac71550 14 * o menel aglar elenath!
cdad3b53 15 * Na-chaered palan-díriel
4ac71550
TC
16 * o galadhremmin ennorath,
17 * Fanuilos, le linnathon
18 * nef aear, si nef aearon!
19 *
20 * [p.238 of _The Lord of the Rings_, II/i: "Many Meetings"]
98994639
HS
21 */
22
166f8a29
DM
23/* utility functions for handling locale-specific stuff like what
24 * character represents the decimal point.
0d071d52
KW
25 *
26 * All C programs have an underlying locale. Perl generally doesn't pay any
27 * attention to it except within the scope of a 'use locale'. For most
28 * categories, it accomplishes this by just using different operations if it is
29 * in such scope than if not. However, various libc functions called by Perl
30 * are affected by the LC_NUMERIC category, so there are macros in perl.h that
31 * are used to toggle between the current locale and the C locale depending on
32 * the desired behavior of those functions at the moment.
166f8a29
DM
33 */
34
98994639
HS
35#include "EXTERN.h"
36#define PERL_IN_LOCALE_C
37#include "perl.h"
38
b310b053
JH
39#ifdef I_LANGINFO
40# include <langinfo.h>
41#endif
42
a4af207c
JH
43#include "reentr.h"
44
8ef6e574
KW
45#ifdef USE_LOCALE
46
98994639 47/*
0d071d52
KW
48 * Standardize the locale name from a string returned by 'setlocale', possibly
49 * modifying that string.
98994639 50 *
0ef2a2b2 51 * The typical return value of setlocale() is either
98994639
HS
52 * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL
53 * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL
54 * (the space-separated values represent the various sublocales,
0ef2a2b2 55 * in some unspecified order). This is not handled by this function.
98994639
HS
56 *
57 * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n",
0ef2a2b2
KW
58 * which is harmful for further use of the string in setlocale(). This
59 * function removes the trailing new line and everything up through the '='
98994639
HS
60 *
61 */
62STATIC char *
63S_stdize_locale(pTHX_ char *locs)
64{
7452cf6a 65 const char * const s = strchr(locs, '=');
98994639
HS
66 bool okay = TRUE;
67
7918f24d
NC
68 PERL_ARGS_ASSERT_STDIZE_LOCALE;
69
8772537c
AL
70 if (s) {
71 const char * const t = strchr(s, '.');
98994639 72 okay = FALSE;
8772537c
AL
73 if (t) {
74 const char * const u = strchr(t, '\n');
75 if (u && (u[1] == 0)) {
76 const STRLEN len = u - s;
77 Move(s + 1, locs, len, char);
78 locs[len] = 0;
79 okay = TRUE;
98994639
HS
80 }
81 }
82 }
83
84 if (!okay)
85 Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs);
86
87 return locs;
88}
89
8ef6e574
KW
90#endif
91
98994639
HS
92void
93Perl_set_numeric_radix(pTHX)
94{
95#ifdef USE_LOCALE_NUMERIC
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
KW
571 result = setlocale(category, locale);
572
481465ea 573 if (! override_LC_ALL) {
89f7b9aa
KW
574 return result;
575 }
576
dfd77d7a 577 /* Here the input category was LC_ALL, and we have set it to what is in the
481465ea
KW
578 * LANG variable or the system default if there is no LANG. But these have
579 * lower priority than the other LC_foo variables, so override it for each
580 * one that is set. (If they are set to "", it means to use the same thing
581 * we just set LC_ALL to, so can skip) */
89f7b9aa
KW
582# ifdef USE_LOCALE_TIME
583 result = PerlEnv_getenv("LC_TIME");
730252b2 584 if (result && strNE(result, "")) {
89f7b9aa
KW
585 setlocale(LC_TIME, result);
586 }
587# endif
588# ifdef USE_LOCALE_CTYPE
589 result = PerlEnv_getenv("LC_CTYPE");
730252b2 590 if (result && strNE(result, "")) {
89f7b9aa
KW
591 setlocale(LC_CTYPE, result);
592 }
593# endif
594# ifdef USE_LOCALE_COLLATE
595 result = PerlEnv_getenv("LC_COLLATE");
730252b2 596 if (result && strNE(result, "")) {
89f7b9aa
KW
597 setlocale(LC_COLLATE, result);
598 }
599# endif
600# ifdef USE_LOCALE_MONETARY
601 result = PerlEnv_getenv("LC_MONETARY");
730252b2 602 if (result && strNE(result, "")) {
89f7b9aa
KW
603 setlocale(LC_MONETARY, result);
604 }
605# endif
606# ifdef USE_LOCALE_NUMERIC
607 result = PerlEnv_getenv("LC_NUMERIC");
730252b2 608 if (result && strNE(result, "")) {
89f7b9aa
KW
609 setlocale(LC_NUMERIC, result);
610 }
611# endif
612# ifdef USE_LOCALE_MESSAGES
613 result = PerlEnv_getenv("LC_MESSAGES");
730252b2 614 if (result && strNE(result, "")) {
89f7b9aa
KW
615 setlocale(LC_MESSAGES, result);
616 }
617# endif
618
619 return setlocale(LC_ALL, NULL);
620
b385bb4d
KW
621}
622
623#endif
624
625
98994639
HS
626/*
627 * Initialize locale awareness.
628 */
629int
630Perl_init_i18nl10n(pTHX_ int printwarn)
631{
0e92a118
KW
632 /* printwarn is
633 *
634 * 0 if not to output warning when setup locale is bad
635 * 1 if to output warning based on value of PERL_BADLANG
636 * >1 if to output regardless of PERL_BADLANG
637 *
638 * returns
98994639 639 * 1 = set ok or not applicable,
0e92a118
KW
640 * 0 = fallback to a locale of lower priority
641 * -1 = fallback to all locales failed, not even to the C locale
6b058d42
KW
642 *
643 * Under -DDEBUGGING, if the environment variable PERL_DEBUG_LOCALE_INIT is
644 * set, debugging information is output.
645 *
646 * This looks more complicated than it is, mainly due to the #ifdefs.
647 *
648 * We try to set LC_ALL to the value determined by the environment. If
649 * there is no LC_ALL on this platform, we try the individual categories we
650 * know about. If this works, we are done.
651 *
652 * But if it doesn't work, we have to do something else. We search the
653 * environment variables ourselves instead of relying on the system to do
654 * it. We look at, in order, LC_ALL, LANG, a system default locale (if we
655 * think there is one), and the ultimate fallback "C". This is all done in
656 * the same loop as above to avoid duplicating code, but it makes things
657 * more complex. After the original failure, we add the fallback
658 * possibilities to the list of locales to try, and iterate the loop
659 * through them all until one succeeds.
660 *
661 * On Ultrix, the locale MUST come from the environment, so there is
662 * preliminary code to set it. I (khw) am not sure that it is necessary,
663 * and that this couldn't be folded into the loop, but barring any real
664 * platforms to test on, it's staying as-is
665 *
666 * A slight complication is that in embedded Perls, the locale may already
667 * be set-up, and we don't want to get it from the normal environment
668 * variables. This is handled by having a special environment variable
669 * indicate we're in this situation. We simply set setlocale's 2nd
670 * parameter to be a NULL instead of "". That indicates to setlocale that
671 * it is not to change anything, but to return the current value,
672 * effectively initializing perl's db to what the locale already is.
673 *
674 * We play the same trick with NULL if a LC_ALL succeeds. We call
675 * setlocale() on the individual categores with NULL to get their existing
676 * values for our db, instead of trying to change them.
677 * */
98994639 678
0e92a118
KW
679 int ok = 1;
680
98994639 681#if defined(USE_LOCALE)
98994639
HS
682#ifdef USE_LOCALE_CTYPE
683 char *curctype = NULL;
684#endif /* USE_LOCALE_CTYPE */
685#ifdef USE_LOCALE_COLLATE
686 char *curcoll = NULL;
687#endif /* USE_LOCALE_COLLATE */
688#ifdef USE_LOCALE_NUMERIC
689 char *curnum = NULL;
690#endif /* USE_LOCALE_NUMERIC */
691#ifdef __GLIBC__
175c4cf9 692 const char * const language = savepv(PerlEnv_getenv("LANGUAGE"));
98994639 693#endif
65ebb059 694
ccd65d51
KW
695 /* NULL uses the existing already set up locale */
696 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
697 ? NULL
698 : "";
5d1187d1
KW
699#ifdef DEBUGGING
700 const bool debug = (PerlEnv_getenv("PERL_DEBUG_LOCALE_INIT"))
701 ? TRUE
702 : FALSE;
703# define DEBUG_LOCALE_INIT(category, locale, result) \
704 STMT_START { \
705 if (debug) { \
706 PerlIO_printf(Perl_debug_log, \
707 "%s:%d: %s\n", \
708 __FILE__, __LINE__, \
709 _setlocale_debug_string(category, \
710 locale, \
711 result)); \
712 } \
713 } STMT_END
714#else
715# define DEBUG_LOCALE_INIT(a,b,c)
716#endif
c3fcd832
KW
717 const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
718 unsigned int trial_locales_count;
175c4cf9
KW
719 const char * const lc_all = savepv(PerlEnv_getenv("LC_ALL"));
720 const char * const lang = savepv(PerlEnv_getenv("LANG"));
98994639 721 bool setlocale_failure = FALSE;
65ebb059
KW
722 unsigned int i;
723 char *p;
175c4cf9
KW
724
725 /* A later getenv() could zap this, so only use here */
726 const char * const bad_lang_use_once = PerlEnv_getenv("PERL_BADLANG");
727
728 const bool locwarn = (printwarn > 1
729 || (printwarn
730 && (! bad_lang_use_once
22ff3130
HS
731 || (
732 /* disallow with "" or "0" */
733 *bad_lang_use_once
734 && strNE("0", bad_lang_use_once)))));
0e92a118 735 bool done = FALSE;
5d1187d1
KW
736 char * sl_result; /* return from setlocale() */
737 char * locale_param;
6bce99ee
JH
738#ifdef WIN32
739 /* In some systems you can find out the system default locale
740 * and use that as the fallback locale. */
741# define SYSTEM_DEFAULT_LOCALE
742#endif
743#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059 744 const char *system_default_locale = NULL;
6bce99ee 745#endif
98994639 746
0e92a118
KW
747#ifndef LOCALE_ENVIRON_REQUIRED
748 PERL_UNUSED_VAR(done);
5d1187d1 749 PERL_UNUSED_VAR(locale_param);
0e92a118 750#else
98994639
HS
751
752 /*
753 * Ultrix setlocale(..., "") fails if there are no environment
754 * variables from which to get a locale name.
755 */
756
b3e384bf 757# ifdef LC_ALL
98994639 758 if (lang) {
5d1187d1
KW
759 sl_result = my_setlocale(LC_ALL, setlocale_init);
760 DEBUG_LOCALE_INIT(LC_ALL, setlocale_init, sl_result);
761 if (sl_result)
98994639
HS
762 done = TRUE;
763 else
764 setlocale_failure = TRUE;
765 }
5d1187d1 766 if (! setlocale_failure) {
b3e384bf 767# ifdef USE_LOCALE_CTYPE
5d1187d1
KW
768 locale_param = (! done && (lang || PerlEnv_getenv("LC_CTYPE")))
769 ? setlocale_init
770 : NULL;
771 curctype = my_setlocale(LC_CTYPE, locale_param);
772 DEBUG_LOCALE_INIT(LC_CTYPE, locale_param, sl_result);
773 if (! curctype)
98994639
HS
774 setlocale_failure = TRUE;
775 else
776 curctype = savepv(curctype);
b3e384bf
KW
777# endif /* USE_LOCALE_CTYPE */
778# ifdef USE_LOCALE_COLLATE
5d1187d1
KW
779 locale_param = (! done && (lang || PerlEnv_getenv("LC_COLLATE")))
780 ? setlocale_init
781 : NULL;
782 curcoll = my_setlocale(LC_COLLATE, locale_param);
783 DEBUG_LOCALE_INIT(LC_COLLATE, locale_param, sl_result);
784 if (! curcoll)
98994639
HS
785 setlocale_failure = TRUE;
786 else
787 curcoll = savepv(curcoll);
b3e384bf
KW
788# endif /* USE_LOCALE_COLLATE */
789# ifdef USE_LOCALE_NUMERIC
5d1187d1
KW
790 locale_param = (! done && (lang || PerlEnv_getenv("LC_NUMERIC")))
791 ? setlocale_init
792 : NULL;
793 curnum = my_setlocale(LC_NUMERIC, locale_param);
794 DEBUG_LOCALE_INIT(LC_NUMERIC, locale_param, sl_result);
795 if (! curnum)
98994639
HS
796 setlocale_failure = TRUE;
797 else
798 curnum = savepv(curnum);
b3e384bf 799# endif /* USE_LOCALE_NUMERIC */
a782673d 800# ifdef USE_LOCALE_MESSAGES
5d1187d1
KW
801 locale_param = (! done && (lang || PerlEnv_getenv("LC_MESSAGES")))
802 ? setlocale_init
803 : NULL;
804 sl_result = my_setlocale(LC_MESSAGES, locale_param);
805 DEBUG_LOCALE_INIT(LC_MESSAGES, locale_param, sl_result);
806 if (! sl_result)
a782673d
KW
807 setlocale_failure = TRUE;
808 }
809# endif /* USE_LOCALE_MESSAGES */
c835d6be 810# ifdef USE_LOCALE_MONETARY
5d1187d1
KW
811 locale_param = (! done && (lang || PerlEnv_getenv("LC_MONETARY")))
812 ? setlocale_init
813 : NULL;
814 sl_result = my_setlocale(LC_MONETARY, locale_param);
815 DEBUG_LOCALE_INIT(LC_MONETARY, locale_param, sl_result);
816 if (! sl_result) {
c835d6be
KW
817 setlocale_failure = TRUE;
818 }
819# endif /* USE_LOCALE_MONETARY */
98994639
HS
820 }
821
b3e384bf 822# endif /* LC_ALL */
98994639
HS
823
824#endif /* !LOCALE_ENVIRON_REQUIRED */
825
65ebb059 826 /* We try each locale in the list until we get one that works, or exhaust
20a240df
KW
827 * the list. Normally the loop is executed just once. But if setting the
828 * locale fails, inside the loop we add fallback trials to the array and so
829 * will execute the loop multiple times */
c3fcd832
KW
830 trial_locales[0] = setlocale_init;
831 trial_locales_count = 1;
65ebb059
KW
832 for (i= 0; i < trial_locales_count; i++) {
833 const char * trial_locale = trial_locales[i];
834
835 if (i > 0) {
836
837 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
838 * when i==0, but I (khw) don't think that behavior makes much
839 * sense */
840 setlocale_failure = FALSE;
841
6bce99ee
JH
842#ifdef SYSTEM_DEFAULT_LOCALE
843# ifdef WIN32
65ebb059
KW
844 /* On Windows machines, an entry of "" after the 0th means to use
845 * the system default locale, which we now proceed to get. */
846 if (strEQ(trial_locale, "")) {
847 unsigned int j;
848
849 /* Note that this may change the locale, but we are going to do
850 * that anyway just below */
851 system_default_locale = setlocale(LC_ALL, "");
5d1187d1 852 DEBUG_LOCALE_INIT(LC_ALL, "", system_default_locale);
65ebb059
KW
853
854 /* Skip if invalid or it's already on the list of locales to
855 * try */
856 if (! system_default_locale) {
857 goto next_iteration;
858 }
859 for (j = 0; j < trial_locales_count; j++) {
860 if (strEQ(system_default_locale, trial_locales[j])) {
861 goto next_iteration;
862 }
863 }
864
865 trial_locale = system_default_locale;
866 }
6bce99ee
JH
867# endif /* WIN32 */
868#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
869 }
870
98994639 871#ifdef LC_ALL
5d1187d1
KW
872 sl_result = my_setlocale(LC_ALL, trial_locale);
873 DEBUG_LOCALE_INIT(LC_ALL, trial_locale, sl_result);
874 if (! sl_result) {
49c85077 875 setlocale_failure = TRUE;
7cd8b568
KW
876 }
877 else {
878 /* Since LC_ALL succeeded, it should have changed all the other
879 * categories it can to its value; so we massage things so that the
880 * setlocales below just return their category's current values.
881 * This adequately handles the case in NetBSD where LC_COLLATE may
882 * not be defined for a locale, and setting it individually will
883 * fail, whereas setting LC_ALL suceeds, leaving LC_COLLATE set to
884 * the POSIX locale. */
885 trial_locale = NULL;
886 }
98994639
HS
887#endif /* LC_ALL */
888
49c85077 889 if (!setlocale_failure) {
98994639 890#ifdef USE_LOCALE_CTYPE
49c85077 891 Safefree(curctype);
5d1187d1
KW
892 curctype = my_setlocale(LC_CTYPE, trial_locale);
893 DEBUG_LOCALE_INIT(LC_CTYPE, trial_locale, curctype);
894 if (! curctype)
49c85077
KW
895 setlocale_failure = TRUE;
896 else
897 curctype = savepv(curctype);
98994639
HS
898#endif /* USE_LOCALE_CTYPE */
899#ifdef USE_LOCALE_COLLATE
49c85077 900 Safefree(curcoll);
5d1187d1
KW
901 curcoll = my_setlocale(LC_COLLATE, trial_locale);
902 DEBUG_LOCALE_INIT(LC_COLLATE, trial_locale, curcoll);
903 if (! curcoll)
49c85077
KW
904 setlocale_failure = TRUE;
905 else
906 curcoll = savepv(curcoll);
98994639
HS
907#endif /* USE_LOCALE_COLLATE */
908#ifdef USE_LOCALE_NUMERIC
49c85077 909 Safefree(curnum);
5d1187d1
KW
910 curnum = my_setlocale(LC_NUMERIC, trial_locale);
911 DEBUG_LOCALE_INIT(LC_NUMERIC, trial_locale, curnum);
912 if (! curnum)
49c85077
KW
913 setlocale_failure = TRUE;
914 else
915 curnum = savepv(curnum);
98994639 916#endif /* USE_LOCALE_NUMERIC */
a782673d 917#ifdef USE_LOCALE_MESSAGES
5d1187d1
KW
918 sl_result = my_setlocale(LC_MESSAGES, trial_locale);
919 DEBUG_LOCALE_INIT(LC_MESSAGES, trial_locale, sl_result);
920 if (! (sl_result))
a782673d
KW
921 setlocale_failure = TRUE;
922#endif /* USE_LOCALE_MESSAGES */
c835d6be 923#ifdef USE_LOCALE_MONETARY
5d1187d1
KW
924 sl_result = my_setlocale(LC_MONETARY, trial_locale);
925 DEBUG_LOCALE_INIT(LC_MONETARY, trial_locale, sl_result);
926 if (! (sl_result))
c835d6be
KW
927 setlocale_failure = TRUE;
928#endif /* USE_LOCALE_MONETARY */
929
49c85077
KW
930 if (! setlocale_failure) { /* Success */
931 break;
932 }
65ebb059 933 }
98994639 934
49c85077
KW
935 /* Here, something failed; will need to try a fallback. */
936 ok = 0;
65ebb059 937
49c85077
KW
938 if (i == 0) {
939 unsigned int j;
98994639 940
65ebb059 941 if (locwarn) { /* Output failure info only on the first one */
98994639
HS
942#ifdef LC_ALL
943
49c85077
KW
944 PerlIO_printf(Perl_error_log,
945 "perl: warning: Setting locale failed.\n");
98994639
HS
946
947#else /* !LC_ALL */
948
49c85077
KW
949 PerlIO_printf(Perl_error_log,
950 "perl: warning: Setting locale failed for the categories:\n\t");
20a240df 951# ifdef USE_LOCALE_CTYPE
49c85077
KW
952 if (! curctype)
953 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
20a240df
KW
954# endif /* USE_LOCALE_CTYPE */
955# ifdef USE_LOCALE_COLLATE
49c85077
KW
956 if (! curcoll)
957 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
20a240df
KW
958# endif /* USE_LOCALE_COLLATE */
959# ifdef USE_LOCALE_NUMERIC
49c85077
KW
960 if (! curnum)
961 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
20a240df 962# endif /* USE_LOCALE_NUMERIC */
a782673d 963 PerlIO_printf(Perl_error_log, "and possibly others\n");
98994639
HS
964
965#endif /* LC_ALL */
966
49c85077
KW
967 PerlIO_printf(Perl_error_log,
968 "perl: warning: Please check that your locale settings:\n");
98994639
HS
969
970#ifdef __GLIBC__
49c85077
KW
971 PerlIO_printf(Perl_error_log,
972 "\tLANGUAGE = %c%s%c,\n",
973 language ? '"' : '(',
974 language ? language : "unset",
975 language ? '"' : ')');
98994639
HS
976#endif
977
49c85077
KW
978 PerlIO_printf(Perl_error_log,
979 "\tLC_ALL = %c%s%c,\n",
980 lc_all ? '"' : '(',
981 lc_all ? lc_all : "unset",
982 lc_all ? '"' : ')');
98994639
HS
983
984#if defined(USE_ENVIRON_ARRAY)
49c85077
KW
985 {
986 char **e;
987 for (e = environ; *e; e++) {
988 if (strnEQ(*e, "LC_", 3)
989 && strnNE(*e, "LC_ALL=", 7)
990 && (p = strchr(*e, '=')))
991 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
992 (int)(p - *e), *e, p + 1);
993 }
994 }
98994639 995#else
49c85077
KW
996 PerlIO_printf(Perl_error_log,
997 "\t(possibly more locale environment variables)\n");
98994639
HS
998#endif
999
49c85077
KW
1000 PerlIO_printf(Perl_error_log,
1001 "\tLANG = %c%s%c\n",
1002 lang ? '"' : '(',
1003 lang ? lang : "unset",
1004 lang ? '"' : ')');
98994639 1005
49c85077
KW
1006 PerlIO_printf(Perl_error_log,
1007 " are supported and installed on your system.\n");
1008 }
98994639 1009
65ebb059 1010 /* Calculate what fallback locales to try. We have avoided this
f6bab5f6 1011 * until we have to, because failure is quite unlikely. This will
65ebb059
KW
1012 * usually change the upper bound of the loop we are in.
1013 *
1014 * Since the system's default way of setting the locale has not
1015 * found one that works, We use Perl's defined ordering: LC_ALL,
1016 * LANG, and the C locale. We don't try the same locale twice, so
1017 * don't add to the list if already there. (On POSIX systems, the
1018 * LC_ALL element will likely be a repeat of the 0th element "",
6b058d42
KW
1019 * but there's no harm done by doing it explicitly.
1020 *
1021 * Note that this tries the LC_ALL environment variable even on
1022 * systems which have no LC_ALL locale setting. This may or may
1023 * not have been originally intentional, but there's no real need
1024 * to change the behavior. */
65ebb059
KW
1025 if (lc_all) {
1026 for (j = 0; j < trial_locales_count; j++) {
1027 if (strEQ(lc_all, trial_locales[j])) {
1028 goto done_lc_all;
1029 }
1030 }
1031 trial_locales[trial_locales_count++] = lc_all;
1032 }
1033 done_lc_all:
98994639 1034
65ebb059
KW
1035 if (lang) {
1036 for (j = 0; j < trial_locales_count; j++) {
1037 if (strEQ(lang, trial_locales[j])) {
1038 goto done_lang;
1039 }
1040 }
1041 trial_locales[trial_locales_count++] = lang;
1042 }
1043 done_lang:
1044
1045#if defined(WIN32) && defined(LC_ALL)
1046 /* For Windows, we also try the system default locale before "C".
1047 * (If there exists a Windows without LC_ALL we skip this because
1048 * it gets too complicated. For those, the "C" is the next
1049 * fallback possibility). The "" is the same as the 0th element of
1050 * the array, but the code at the loop above knows to treat it
1051 * differently when not the 0th */
1052 trial_locales[trial_locales_count++] = "";
1053#endif
1054
1055 for (j = 0; j < trial_locales_count; j++) {
1056 if (strEQ("C", trial_locales[j])) {
1057 goto done_C;
1058 }
1059 }
1060 trial_locales[trial_locales_count++] = "C";
98994639 1061
65ebb059
KW
1062 done_C: ;
1063 } /* end of first time through the loop */
98994639 1064
65ebb059
KW
1065#ifdef WIN32
1066 next_iteration: ;
1067#endif
1068
1069 } /* end of looping through the trial locales */
1070
1071 if (ok < 1) { /* If we tried to fallback */
1072 const char* msg;
1073 if (! setlocale_failure) { /* fallback succeeded */
1074 msg = "Falling back to";
1075 }
1076 else { /* fallback failed */
98994639 1077
65ebb059
KW
1078 /* We dropped off the end of the loop, so have to decrement i to
1079 * get back to the value the last time through */
1080 i--;
98994639 1081
65ebb059
KW
1082 ok = -1;
1083 msg = "Failed to fall back to";
1084
1085 /* To continue, we should use whatever values we've got */
98994639 1086#ifdef USE_LOCALE_CTYPE
49c85077
KW
1087 Safefree(curctype);
1088 curctype = savepv(setlocale(LC_CTYPE, NULL));
5d1187d1 1089 DEBUG_LOCALE_INIT(LC_CTYPE, NULL, curctype);
98994639
HS
1090#endif /* USE_LOCALE_CTYPE */
1091#ifdef USE_LOCALE_COLLATE
49c85077
KW
1092 Safefree(curcoll);
1093 curcoll = savepv(setlocale(LC_COLLATE, NULL));
5d1187d1 1094 DEBUG_LOCALE_INIT(LC_COLLATE, NULL, curcoll);
98994639
HS
1095#endif /* USE_LOCALE_COLLATE */
1096#ifdef USE_LOCALE_NUMERIC
49c85077
KW
1097 Safefree(curnum);
1098 curnum = savepv(setlocale(LC_NUMERIC, NULL));
5d1187d1 1099 DEBUG_LOCALE_INIT(LC_NUMERIC, NULL, curnum);
98994639 1100#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
1101 }
1102
1103 if (locwarn) {
1104 const char * description;
1105 const char * name = "";
1106 if (strEQ(trial_locales[i], "C")) {
1107 description = "the standard locale";
1108 name = "C";
1109 }
6bce99ee 1110#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059
KW
1111 else if (strEQ(trial_locales[i], "")) {
1112 description = "the system default locale";
1113 if (system_default_locale) {
1114 name = system_default_locale;
1115 }
1116 }
6bce99ee 1117#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
1118 else {
1119 description = "a fallback locale";
1120 name = trial_locales[i];
1121 }
1122 if (name && strNE(name, "")) {
1123 PerlIO_printf(Perl_error_log,
1124 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
1125 }
1126 else {
1127 PerlIO_printf(Perl_error_log,
1128 "perl: warning: %s %s.\n", msg, description);
1129 }
1130 }
1131 } /* End of tried to fallback */
98994639
HS
1132
1133#ifdef USE_LOCALE_CTYPE
1134 new_ctype(curctype);
1135#endif /* USE_LOCALE_CTYPE */
1136
1137#ifdef USE_LOCALE_COLLATE
1138 new_collate(curcoll);
1139#endif /* USE_LOCALE_COLLATE */
1140
1141#ifdef USE_LOCALE_NUMERIC
1142 new_numeric(curnum);
1143#endif /* USE_LOCALE_NUMERIC */
b310b053 1144
8ef6e574 1145#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
49c85077
KW
1146 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
1147 * locale is UTF-8. If PL_utf8locale and PL_unicode (set by -C or by
1148 * $ENV{PERL_UNICODE}) are true, perl.c:S_parse_body() will turn on the
1149 * PerlIO :utf8 layer on STDIN, STDOUT, STDERR, _and_ the default open
1150 * discipline. */
c1284011 1151 PL_utf8locale = _is_cur_LC_category_utf8(LC_CTYPE);
49c85077 1152
a05d7ebb 1153 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
1154 This is an alternative to using the -C command line switch
1155 (the -C if present will override this). */
1156 {
dd374669 1157 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 1158 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
1159 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
1160 PL_utf8cache = -1;
b310b053 1161 }
ec71e770 1162#endif
b310b053 1163
98994639 1164#ifdef USE_LOCALE_CTYPE
43c5f42d 1165 Safefree(curctype);
98994639
HS
1166#endif /* USE_LOCALE_CTYPE */
1167#ifdef USE_LOCALE_COLLATE
43c5f42d 1168 Safefree(curcoll);
98994639
HS
1169#endif /* USE_LOCALE_COLLATE */
1170#ifdef USE_LOCALE_NUMERIC
43c5f42d 1171 Safefree(curnum);
98994639 1172#endif /* USE_LOCALE_NUMERIC */
8ef6e574 1173
175c4cf9
KW
1174#ifdef __GLIBC__
1175 Safefree(language);
1176#endif
1177
1178 Safefree(lc_all);
1179 Safefree(lang);
1180
e3305790
KW
1181#else /* !USE_LOCALE */
1182 PERL_UNUSED_ARG(printwarn);
1183#endif /* USE_LOCALE */
1184
98994639
HS
1185 return ok;
1186}
1187
49c85077 1188
98994639
HS
1189#ifdef USE_LOCALE_COLLATE
1190
1191/*
1192 * mem_collxfrm() is a bit like strxfrm() but with two important
1193 * differences. First, it handles embedded NULs. Second, it allocates
1194 * a bit more memory than needed for the transformed data itself.
1195 * The real transformed data begins at offset sizeof(collationix).
1196 * Please see sv_collxfrm() to see how this is used.
1197 */
166f8a29 1198
98994639
HS
1199char *
1200Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
1201{
1202 char *xbuf;
1203 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
1204
7918f24d
NC
1205 PERL_ARGS_ASSERT_MEM_COLLXFRM;
1206
98994639
HS
1207 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
1208 /* the +1 is for the terminating NUL. */
1209
1210 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 1211 Newx(xbuf, xAlloc, char);
98994639
HS
1212 if (! xbuf)
1213 goto bad;
1214
1215 *(U32*)xbuf = PL_collation_ix;
1216 xout = sizeof(PL_collation_ix);
1217 for (xin = 0; xin < len; ) {
224e8ef5 1218 Size_t xused;
98994639
HS
1219
1220 for (;;) {
1221 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 1222 if (xused >= PERL_INT_MAX)
98994639 1223 goto bad;
eb160463 1224 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
1225 break;
1226 xAlloc = (2 * xAlloc) + 1;
1227 Renew(xbuf, xAlloc, char);
1228 if (! xbuf)
1229 goto bad;
1230 }
1231
1232 xin += strlen(s + xin) + 1;
1233 xout += xused;
1234
1235 /* Embedded NULs are understood but silently skipped
1236 * because they make no sense in locale collation. */
1237 }
1238
1239 xbuf[xout] = '\0';
1240 *xlen = xout - sizeof(PL_collation_ix);
1241 return xbuf;
1242
1243 bad:
1244 Safefree(xbuf);
1245 *xlen = 0;
1246 return NULL;
1247}
1248
1249#endif /* USE_LOCALE_COLLATE */
1250
8ef6e574
KW
1251#ifdef USE_LOCALE
1252
c1284011
KW
1253bool
1254Perl__is_cur_LC_category_utf8(pTHX_ int category)
7d74bb61
KW
1255{
1256 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
1257 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b 1258 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
609548d2
KW
1259 * could give the wrong result. The result will very likely be correct for
1260 * languages that have commonly used non-ASCII characters, but for notably
1261 * English, it comes down to if the locale's name ends in something like
1262 * "UTF-8". It errs on the side of not being a UTF-8 locale. */
7d74bb61
KW
1263
1264 char *save_input_locale = NULL;
7d74bb61
KW
1265 STRLEN final_pos;
1266
8ef6e574 1267#ifdef LC_ALL
7d74bb61 1268 assert(category != LC_ALL);
8ef6e574 1269#endif
7d74bb61
KW
1270
1271 /* First dispose of the trivial cases */
b07fffd1 1272 save_input_locale = setlocale(category, NULL);
7d74bb61 1273 if (! save_input_locale) {
69014004
KW
1274 DEBUG_L(PerlIO_printf(Perl_debug_log,
1275 "Could not find current locale for category %d\n",
1276 category));
7d74bb61
KW
1277 return FALSE; /* XXX maybe should croak */
1278 }
b07fffd1 1279 save_input_locale = stdize_locale(savepv(save_input_locale));
a39edc4c 1280 if (isNAME_C_OR_POSIX(save_input_locale)) {
69014004
KW
1281 DEBUG_L(PerlIO_printf(Perl_debug_log,
1282 "Current locale for category %d is %s\n",
1283 category, save_input_locale));
b07fffd1 1284 Safefree(save_input_locale);
7d74bb61
KW
1285 return FALSE;
1286 }
1287
1d958db2
KW
1288#if defined(USE_LOCALE_CTYPE) \
1289 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 1290
1d958db2 1291 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
1292
1293 char *save_ctype_locale = NULL;
119ee68b 1294 bool is_utf8;
7d74bb61 1295
119ee68b 1296 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
1297
1298 /* Get the current LC_CTYPE locale */
4f72bb37 1299 save_ctype_locale = setlocale(LC_CTYPE, NULL);
7d74bb61 1300 if (! save_ctype_locale) {
69014004
KW
1301 DEBUG_L(PerlIO_printf(Perl_debug_log,
1302 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
1303 goto cant_use_nllanginfo;
1304 }
4f72bb37 1305 save_ctype_locale = stdize_locale(savepv(save_ctype_locale));
7d74bb61
KW
1306
1307 /* If LC_CTYPE and the desired category use the same locale, this
1308 * means that finding the value for LC_CTYPE is the same as finding
1309 * the value for the desired category. Otherwise, switch LC_CTYPE
1310 * to the desired category's locale */
1311 if (strEQ(save_ctype_locale, save_input_locale)) {
1312 Safefree(save_ctype_locale);
1313 save_ctype_locale = NULL;
1314 }
1315 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
1316 DEBUG_L(PerlIO_printf(Perl_debug_log,
1317 "Could not change LC_CTYPE locale to %s\n",
1318 save_input_locale));
7d74bb61
KW
1319 Safefree(save_ctype_locale);
1320 goto cant_use_nllanginfo;
1321 }
1322 }
1323
69014004
KW
1324 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
1325 save_input_locale));
1326
7d74bb61 1327 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
1328 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
1329 * should give the correct results */
119ee68b 1330
1d958db2
KW
1331# if defined(HAS_NL_LANGINFO) && defined(CODESET)
1332 {
4f72bb37 1333 char *codeset = nl_langinfo(CODESET);
1d958db2 1334 if (codeset && strNE(codeset, "")) {
4f72bb37 1335 codeset = savepv(codeset);
119ee68b 1336
1d958db2
KW
1337 /* If we switched LC_CTYPE, switch back */
1338 if (save_ctype_locale) {
1339 setlocale(LC_CTYPE, save_ctype_locale);
1340 Safefree(save_ctype_locale);
1341 }
1342
1343 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
1344 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
1345
69014004
KW
1346 DEBUG_L(PerlIO_printf(Perl_debug_log,
1347 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
1348 codeset, is_utf8));
1d958db2
KW
1349 Safefree(codeset);
1350 Safefree(save_input_locale);
1351 return is_utf8;
1352 }
119ee68b
KW
1353 }
1354
1d958db2
KW
1355# endif
1356# ifdef MB_CUR_MAX
1357
1358 /* Here, either we don't have nl_langinfo, or it didn't return a
1359 * codeset. Try MB_CUR_MAX */
1360
119ee68b
KW
1361 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
1362 * Unicode code point. Since UTF-8 is the only non-single byte
1363 * encoding we handle, we just say any such encoding is UTF-8, and if
1364 * turns out to be wrong, other things will fail */
1365 is_utf8 = MB_CUR_MAX >= 4;
1366
69014004
KW
1367 DEBUG_L(PerlIO_printf(Perl_debug_log,
1368 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
1369 (int) MB_CUR_MAX, is_utf8));
1370
119ee68b
KW
1371 Safefree(save_input_locale);
1372
1373# ifdef HAS_MBTOWC
1374
1375 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
1376 * since they are both in the C99 standard. We can feed a known byte
1377 * string to the latter function, and check that it gives the expected
1378 * result */
1379 if (is_utf8) {
1380 wchar_t wc;
856b881c 1381 PERL_UNUSED_RESULT(mbtowc(&wc, NULL, 0));/* Reset any shift state */
69014004 1382 errno = 0;
f019f68f 1383 if ((size_t)mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
119ee68b
KW
1384 != strlen(HYPHEN_UTF8)
1385 || wc != (wchar_t) 0x2010)
1386 {
1387 is_utf8 = FALSE;
abdcbdb8 1388 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", (unsigned int)wc));
69014004
KW
1389 DEBUG_L(PerlIO_printf(Perl_debug_log,
1390 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
1391 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
1392 }
1393 }
119ee68b
KW
1394# endif
1395
1d958db2
KW
1396 /* If we switched LC_CTYPE, switch back */
1397 if (save_ctype_locale) {
1398 setlocale(LC_CTYPE, save_ctype_locale);
1399 Safefree(save_ctype_locale);
119ee68b 1400 }
7d74bb61 1401
1d958db2 1402 return is_utf8;
119ee68b 1403# endif
7d74bb61 1404 }
119ee68b 1405
7d74bb61
KW
1406 cant_use_nllanginfo:
1407
0080c90a
KW
1408#else /* nl_langinfo should work if available, so don't bother compiling this
1409 fallback code. The final fallback of looking at the name is
1410 compiled, and will be executed if nl_langinfo fails */
7d74bb61 1411
97f4de96
KW
1412 /* nl_langinfo not available or failed somehow. Next try looking at the
1413 * currency symbol to see if it disambiguates things. Often that will be
1414 * in the native script, and if the symbol isn't in UTF-8, we know that the
1415 * locale isn't. If it is non-ASCII UTF-8, we infer that the locale is
609548d2
KW
1416 * too, as the odds of a non-UTF8 string being valid UTF-8 are quite small
1417 * */
fa9b773e
KW
1418
1419#ifdef HAS_LOCALECONV
fa9b773e 1420# ifdef USE_LOCALE_MONETARY
fa9b773e
KW
1421 {
1422 char *save_monetary_locale = NULL;
fa9b773e 1423 bool only_ascii = FALSE;
13542a67
KW
1424 bool is_utf8 = FALSE;
1425 struct lconv* lc;
fa9b773e 1426
97f4de96
KW
1427 /* Like above for LC_CTYPE, we first set LC_MONETARY to the locale of
1428 * the desired category, if it isn't that locale already */
1429
fa9b773e
KW
1430 if (category != LC_MONETARY) {
1431
4f72bb37 1432 save_monetary_locale = setlocale(LC_MONETARY, NULL);
fa9b773e 1433 if (! save_monetary_locale) {
69014004
KW
1434 DEBUG_L(PerlIO_printf(Perl_debug_log,
1435 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1436 goto cant_use_monetary;
1437 }
4f72bb37 1438 save_monetary_locale = stdize_locale(savepv(save_monetary_locale));
fa9b773e 1439
13542a67
KW
1440 if (strEQ(save_monetary_locale, save_input_locale)) {
1441 Safefree(save_monetary_locale);
1442 save_monetary_locale = NULL;
1443 }
1444 else if (! setlocale(LC_MONETARY, save_input_locale)) {
59c234b4
KW
1445 DEBUG_L(PerlIO_printf(Perl_debug_log,
1446 "Could not change LC_MONETARY locale to %s\n",
1447 save_input_locale));
1448 Safefree(save_monetary_locale);
1449 goto cant_use_monetary;
fa9b773e
KW
1450 }
1451 }
1452
1453 /* Here the current LC_MONETARY is set to the locale of the category
1454 * whose information is desired. */
1455
13542a67
KW
1456 lc = localeconv();
1457 if (! lc
1458 || ! lc->currency_symbol
9f10db87 1459 || is_invariant_string((U8 *) lc->currency_symbol, 0))
13542a67
KW
1460 {
1461 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));
1462 only_ascii = TRUE;
1463 }
1464 else {
1465 is_utf8 = is_utf8_string((U8 *) lc->currency_symbol, 0);
fa9b773e
KW
1466 }
1467
1468 /* If we changed it, restore LC_MONETARY to its original locale */
1469 if (save_monetary_locale) {
1470 setlocale(LC_MONETARY, save_monetary_locale);
1471 Safefree(save_monetary_locale);
1472 }
1473
13542a67 1474 if (! only_ascii) {
fa9b773e 1475
59c234b4
KW
1476 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8;
1477 * otherwise assume the locale is UTF-8 if and only if the symbol
1478 * is non-ascii UTF-8. */
1479 DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?Currency symbol for %s is UTF-8=%d\n",
1480 save_input_locale, is_utf8));
1481 Safefree(save_input_locale);
1482 return is_utf8;
13542a67 1483 }
fa9b773e
KW
1484 }
1485 cant_use_monetary:
1486
1487# endif /* USE_LOCALE_MONETARY */
1488#endif /* HAS_LOCALECONV */
1489
15f7e74e
KW
1490#if defined(HAS_STRFTIME) && defined(USE_LOCALE_TIME)
1491
1492/* Still haven't found a non-ASCII string to disambiguate UTF-8 or not. Try
1493 * the names of the months and weekdays, timezone, and am/pm indicator */
1494 {
1495 char *save_time_locale = NULL;
1496 int hour = 10;
1497 bool is_dst = FALSE;
1498 int dom = 1;
1499 int month = 0;
1500 int i;
1501 char * formatted_time;
1502
1503
1504 /* Like above for LC_MONETARY, we set LC_TIME to the locale of the
1505 * desired category, if it isn't that locale already */
1506
1507 if (category != LC_TIME) {
1508
1509 save_time_locale = setlocale(LC_TIME, NULL);
1510 if (! save_time_locale) {
1511 DEBUG_L(PerlIO_printf(Perl_debug_log,
1512 "Could not find current locale for LC_TIME\n"));
1513 goto cant_use_time;
1514 }
1515 save_time_locale = stdize_locale(savepv(save_time_locale));
1516
1517 if (strEQ(save_time_locale, save_input_locale)) {
1518 Safefree(save_time_locale);
1519 save_time_locale = NULL;
1520 }
1521 else if (! setlocale(LC_TIME, save_input_locale)) {
1522 DEBUG_L(PerlIO_printf(Perl_debug_log,
1523 "Could not change LC_TIME locale to %s\n",
1524 save_input_locale));
1525 Safefree(save_time_locale);
1526 goto cant_use_time;
1527 }
1528 }
1529
1530 /* Here the current LC_TIME is set to the locale of the category
1531 * whose information is desired. Look at all the days of the week and
9f10db87 1532 * month names, and the timezone and am/pm indicator for UTF-8 variant
15f7e74e
KW
1533 * characters. The first such a one found will tell us if the locale
1534 * is UTF-8 or not */
1535
1536 for (i = 0; i < 7 + 12; i++) { /* 7 days; 12 months */
1537 formatted_time = my_strftime("%A %B %Z %p",
1538 0, 0, hour, dom, month, 112, 0, 0, is_dst);
9f10db87 1539 if (! formatted_time || is_invariant_string((U8 *) formatted_time, 0)) {
15f7e74e
KW
1540
1541 /* Here, we didn't find a non-ASCII. Try the next time through
1542 * with the complemented dst and am/pm, and try with the next
1543 * weekday. After we have gotten all weekdays, try the next
1544 * month */
1545 is_dst = ! is_dst;
1546 hour = (hour + 12) % 24;
1547 dom++;
1548 if (i > 6) {
1549 month++;
1550 }
1551 continue;
1552 }
1553
1554 /* Here, we have a non-ASCII. Return TRUE is it is valid UTF8;
1555 * false otherwise. But first, restore LC_TIME to its original
1556 * locale if we changed it */
1557 if (save_time_locale) {
1558 setlocale(LC_TIME, save_time_locale);
1559 Safefree(save_time_locale);
1560 }
1561
1562 DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?time-related strings for %s are UTF-8=%d\n",
1563 save_input_locale,
1564 is_utf8_string((U8 *) formatted_time, 0)));
1565 Safefree(save_input_locale);
1566 return is_utf8_string((U8 *) formatted_time, 0);
1567 }
1568
1569 /* Falling off the end of the loop indicates all the names were just
1570 * ASCII. Go on to the next test. If we changed it, restore LC_TIME
1571 * to its original locale */
1572 if (save_time_locale) {
1573 setlocale(LC_TIME, save_time_locale);
1574 Safefree(save_time_locale);
1575 }
1576 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));
1577 }
1578 cant_use_time:
1579
1580#endif
1581
1582#if 0 && defined(USE_LOCALE_MESSAGES) && defined(HAS_SYS_ERRLIST)
855aeb93
JH
1583
1584/* This code is ifdefd out because it was found to not be necessary in testing
5857e934
KW
1585 * on our dromedary test machine, which has over 700 locales. There, this
1586 * added no value to looking at the currency symbol and the time strings. I
1587 * left it in so as to avoid rewriting it if real-world experience indicates
1588 * that dromedary is an outlier. Essentially, instead of returning abpve if we
855aeb93
JH
1589 * haven't found illegal utf8, we continue on and examine all the strerror()
1590 * messages on the platform for utf8ness. If all are ASCII, we still don't
1591 * know the answer; but otherwise we have a pretty good indication of the
5857e934
KW
1592 * utf8ness. The reason this doesn't help much is that the messages may not
1593 * have been translated into the locale. The currency symbol and time strings
1594 * are much more likely to have been translated. */
1595 {
855aeb93 1596 int e;
5857e934
KW
1597 bool is_utf8 = FALSE;
1598 bool non_ascii = FALSE;
855aeb93 1599 char *save_messages_locale = NULL;
5857e934 1600 const char * errmsg = NULL;
855aeb93 1601
5857e934
KW
1602 /* Like above, we set LC_MESSAGES to the locale of the desired
1603 * category, if it isn't that locale already */
855aeb93
JH
1604
1605 if (category != LC_MESSAGES) {
1606
5857e934 1607 save_messages_locale = setlocale(LC_MESSAGES, NULL);
855aeb93 1608 if (! save_messages_locale) {
5857e934
KW
1609 DEBUG_L(PerlIO_printf(Perl_debug_log,
1610 "Could not find current locale for LC_MESSAGES\n"));
855aeb93
JH
1611 goto cant_use_messages;
1612 }
5857e934 1613 save_messages_locale = stdize_locale(savepv(save_messages_locale));
855aeb93
JH
1614
1615 if (strEQ(save_messages_locale, save_input_locale)) {
5857e934
KW
1616 Safefree(save_messages_locale);
1617 save_messages_locale = NULL;
855aeb93
JH
1618 }
1619 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
5857e934
KW
1620 DEBUG_L(PerlIO_printf(Perl_debug_log,
1621 "Could not change LC_MESSAGES locale to %s\n",
1622 save_input_locale));
855aeb93
JH
1623 Safefree(save_messages_locale);
1624 goto cant_use_messages;
1625 }
1626 }
1627
1628 /* Here the current LC_MESSAGES is set to the locale of the category
5857e934
KW
1629 * whose information is desired. Look through all the messages. We
1630 * can't use Strerror() here because it may expand to code that
1631 * segfaults in miniperl */
855aeb93 1632
5857e934
KW
1633 for (e = 0; e <= sys_nerr; e++) {
1634 errno = 0;
1635 errmsg = sys_errlist[e];
1636 if (errno || !errmsg) {
855aeb93
JH
1637 break;
1638 }
5857e934 1639 errmsg = savepv(errmsg);
9f10db87 1640 if (! is_invariant_string((U8 *) errmsg, 0)) {
5857e934
KW
1641 non_ascii = TRUE;
1642 is_utf8 = is_utf8_string((U8 *) errmsg, 0);
1643 break;
855aeb93
JH
1644 }
1645 }
5857e934 1646 Safefree(errmsg);
855aeb93
JH
1647
1648 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1649 if (save_messages_locale) {
1650 setlocale(LC_MESSAGES, save_messages_locale);
1651 Safefree(save_messages_locale);
1652 }
1653
5857e934
KW
1654 if (non_ascii) {
1655
1656 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1657 * any non-ascii means it is one; otherwise we assume it isn't */
1658 DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?error messages for %s are UTF-8=%d\n",
1659 save_input_locale,
1660 is_utf8));
1661 Safefree(save_input_locale);
1662 return is_utf8;
1663 }
855aeb93 1664
5857e934 1665 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
1666 }
1667 cant_use_messages:
1668
1669#endif
fa9b773e 1670
0080c90a
KW
1671#endif /* the code that is compiled when no nl_langinfo */
1672
92c0a900
KW
1673#ifndef EBCDIC /* On os390, even if the name ends with "UTF-8', it isn't a
1674 UTF-8 locale */
97f4de96
KW
1675 /* As a last resort, look at the locale name to see if it matches
1676 * qr/UTF -? * 8 /ix, or some other common locale names. This "name", the
1677 * return of setlocale(), is actually defined to be opaque, so we can't
1678 * really rely on the absence of various substrings in the name to indicate
1679 * its UTF-8ness, but if it has UTF8 in the name, it is extremely likely to
1680 * be a UTF-8 locale. Similarly for the other common names */
1681
1682 final_pos = strlen(save_input_locale) - 1;
1683 if (final_pos >= 3) {
1684 char *name = save_input_locale;
1685
1686 /* Find next 'U' or 'u' and look from there */
1687 while ((name += strcspn(name, "Uu") + 1)
1688 <= save_input_locale + final_pos - 2)
1689 {
305b8651
KW
1690 if (!isALPHA_FOLD_NE(*name, 't')
1691 || isALPHA_FOLD_NE(*(name + 1), 'f'))
97f4de96
KW
1692 {
1693 continue;
1694 }
1695 name += 2;
1696 if (*(name) == '-') {
1697 if ((name > save_input_locale + final_pos - 1)) {
1698 break;
1699 }
1700 name++;
1701 }
1702 if (*(name) == '8') {
97f4de96
KW
1703 DEBUG_L(PerlIO_printf(Perl_debug_log,
1704 "Locale %s ends with UTF-8 in name\n",
1705 save_input_locale));
00c54b9c 1706 Safefree(save_input_locale);
97f4de96
KW
1707 return TRUE;
1708 }
1709 }
1710 DEBUG_L(PerlIO_printf(Perl_debug_log,
1711 "Locale %s doesn't end with UTF-8 in name\n",
1712 save_input_locale));
1713 }
92c0a900 1714#endif
97f4de96
KW
1715
1716#ifdef WIN32
1717 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
1718 if (final_pos >= 4
1719 && *(save_input_locale + final_pos - 0) == '1'
1720 && *(save_input_locale + final_pos - 1) == '0'
1721 && *(save_input_locale + final_pos - 2) == '0'
1722 && *(save_input_locale + final_pos - 3) == '5'
1723 && *(save_input_locale + final_pos - 4) == '6')
1724 {
1725 DEBUG_L(PerlIO_printf(Perl_debug_log,
1726 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
1727 save_input_locale));
1728 Safefree(save_input_locale);
1729 return TRUE;
1730 }
1731#endif
1732
1733 /* Other common encodings are the ISO 8859 series, which aren't UTF-8. But
1734 * since we are about to return FALSE anyway, there is no point in doing
1735 * this extra work */
1736#if 0
1737 if (instr(save_input_locale, "8859")) {
1738 DEBUG_L(PerlIO_printf(Perl_debug_log,
1739 "Locale %s has 8859 in name, not UTF-8 locale\n",
1740 save_input_locale));
1741 Safefree(save_input_locale);
1742 return FALSE;
1743 }
1744#endif
1745
69014004
KW
1746 DEBUG_L(PerlIO_printf(Perl_debug_log,
1747 "Assuming locale %s is not a UTF-8 locale\n",
1748 save_input_locale));
fa9b773e 1749 Safefree(save_input_locale);
7d74bb61
KW
1750 return FALSE;
1751}
1752
8ef6e574 1753#endif
7d74bb61 1754
d6ded950
KW
1755
1756bool
1757Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
1758{
1a4f13e1 1759 dVAR;
d6ded950
KW
1760 /* Internal function which returns if we are in the scope of a pragma that
1761 * enables the locale category 'category'. 'compiling' should indicate if
1762 * this is during the compilation phase (TRUE) or not (FALSE). */
1763
1764 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
1765
1766 SV *categories = cop_hints_fetch_pvs(cop, "locale", 0);
1767 if (! categories || categories == &PL_sv_placeholder) {
1768 return FALSE;
1769 }
1770
1771 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
1772 * a valid unsigned */
1773 assert(category >= -1);
1774 return cBOOL(SvUV(categories) & (1U << (category + 1)));
1775}
1776
2c6ee1a7
KW
1777char *
1778Perl_my_strerror(pTHX_ const int errnum) {
1779
1780 /* Uses C locale for the error text unless within scope of 'use locale' for
1781 * LC_MESSAGES */
1782
1783#ifdef USE_LOCALE_MESSAGES
8ee4b769 1784 if (! IN_LC(LC_MESSAGES)) {
2c6ee1a7 1785 char * save_locale = setlocale(LC_MESSAGES, NULL);
a39edc4c 1786 if (! isNAME_C_OR_POSIX(save_locale)) {
2c6ee1a7
KW
1787 char *errstr;
1788
1789 /* The next setlocale likely will zap this, so create a copy */
1790 save_locale = savepv(save_locale);
1791
1792 setlocale(LC_MESSAGES, "C");
1793
1794 /* This points to the static space in Strerror, with all its
1795 * limitations */
1796 errstr = Strerror(errnum);
1797
1798 setlocale(LC_MESSAGES, save_locale);
1799 Safefree(save_locale);
1800 return errstr;
1801 }
1802 }
1803#endif
1804
1805 return Strerror(errnum);
1806}
1807
66610fdd 1808/*
747c467a
KW
1809
1810=head1 Locale-related functions and macros
1811
1812=for apidoc sync_locale
1813
1814Changing the program's locale should be avoided by XS code. Nevertheless,
1815certain non-Perl libraries called from XS, such as C<Gtk> do so. When this
1816happens, Perl needs to be told that the locale has changed. Use this function
1817to do so, before returning to Perl.
1818
1819=cut
1820*/
1821
1822void
1823Perl_sync_locale(pTHX)
1824{
1825
1826#ifdef USE_LOCALE_CTYPE
1827 new_ctype(setlocale(LC_CTYPE, NULL));
1828#endif /* USE_LOCALE_CTYPE */
1829
1830#ifdef USE_LOCALE_COLLATE
1831 new_collate(setlocale(LC_COLLATE, NULL));
1832#endif
1833
1834#ifdef USE_LOCALE_NUMERIC
1835 set_numeric_local(); /* Switch from "C" to underlying LC_NUMERIC */
1836 new_numeric(setlocale(LC_NUMERIC, NULL));
1837#endif /* USE_LOCALE_NUMERIC */
1838
1839}
1840
5d1187d1
KW
1841#if defined(DEBUGGING) && defined(USE_LOCALE)
1842
1843char *
1844Perl__setlocale_debug_string(const int category, /* category number,
1845 like LC_ALL */
1846 const char* const locale, /* locale name */
1847
1848 /* return value from setlocale() when attempting to
1849 * set 'category' to 'locale' */
1850 const char* const retval)
1851{
1852 /* Returns a pointer to a NUL-terminated string in static storage with
1853 * added text about the info passed in. This is not thread safe and will
1854 * be overwritten by the next call, so this should be used just to
1855 * formulate a string to immediately print or savepv() on.
1856 *
1857 * Buffer overflow checking is done only after the fact (via an assert),
1858 * because this is used only in DEBUGGING, and an attacker would have to
1859 * control the start up of perl with the correct environment variable or
1860 * command line option. */
1861
1862 static char ret[128] = "";
1863
1864 strcpy(ret, "setlocale(");
1865
1866 switch (category) {
1867 default:
1868 sprintf(ret, "%s? %d", ret, category);
1869 break;
1870# ifdef LC_ALL
1871 case LC_ALL:
1872 strcat(ret, "LC_ALL");
1873 break;
1874# endif
1875# ifdef LC_CTYPE
1876 case LC_CTYPE:
1877 strcat(ret, "LC_CTYPE");
1878 break;
1879# endif
1880# ifdef LC_NUMERIC
1881 case LC_NUMERIC:
1882 strcat(ret, "LC_NUMERIC");
1883 break;
1884# endif
1885# ifdef LC_COLLATE
1886 case LC_COLLATE:
1887 strcat(ret, "LC_COLLATE");
1888 break;
1889# endif
1890# ifdef LC_TIME
1891 case LC_TIME:
1892 strcat(ret, "LC_TIME");
1893 break;
1894# endif
1895# ifdef LC_MONETARY
1896 case LC_MONETARY:
1897 strcat(ret, "LC_MONETARY");
1898 break;
1899# endif
1900# ifdef LC_MESSAGES
1901 case LC_MESSAGES:
1902 strcat(ret, "LC_MESSAGES");
1903 break;
1904# endif
1905 }
1906
1907 strcat(ret, ", ");
1908
1909 if (locale) {
1910 strcat(ret, "\"");
1911 strcat(ret, locale);
1912 strcat(ret, "\"");
1913 }
1914 else {
1915 strcat(ret, "NULL");
1916 }
1917
1918 strcat(ret, ") returned ");
1919
1920 if (retval) {
1921 strcat(ret, "\"");
1922 strcat(ret, retval);
1923 strcat(ret, "\"");
1924 }
1925 else {
1926 strcat(ret, "NULL");
1927 }
1928
1929 assert(strlen(ret) < sizeof(ret));
1930
1931 return ret;
1932}
1933
1934#endif
747c467a
KW
1935
1936
1937/*
14d04a33 1938 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1939 */