This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
locale.c: Remove conditionals.
[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);
28acfe03
KW
109 if (! is_ascii_string((U8 *) lc->decimal_point, 0)
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 */
139#define isNAME_C_OR_POSIX(name) ((name) != NULL \
140 && ((*(name) == 'C' && (*(name) + 1) == '\0') \
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:
162 * PL_numeric_name The default locale's name: a copy of 'newnum'
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
98994639
HS
220 setlocale(LC_NUMERIC, "C");
221 PL_numeric_standard = TRUE;
126aedc4 222 PL_numeric_local = isNAME_C_OR_POSIX(PL_numeric_name);
98994639 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) */
98994639 239 setlocale(LC_NUMERIC, PL_numeric_name);
126aedc4 240 PL_numeric_standard = isNAME_C_OR_POSIX(PL_numeric_name);
98994639
HS
241 PL_numeric_local = TRUE;
242 set_numeric_radix();
69014004
KW
243 DEBUG_L(PerlIO_printf(Perl_debug_log,
244 "Underlying LC_NUMERIC locale now is %s\n",
245 PL_numeric_name));
98994639
HS
246
247#endif /* USE_LOCALE_NUMERIC */
248}
249
250/*
251 * Set up for a new ctype locale.
252 */
253void
8772537c 254Perl_new_ctype(pTHX_ const char *newctype)
98994639
HS
255{
256#ifdef USE_LOCALE_CTYPE
0d071d52
KW
257
258 /* Called after all libc setlocale() calls affecting LC_CTYPE, to tell
259 * core Perl this and that 'newctype' is the name of the new locale.
260 *
261 * This function sets up the folding arrays for all 256 bytes, assuming
262 * that tofold() is tolc() since fold case is not a concept in POSIX,
263 *
264 * Any code changing the locale (outside this file) should use
265 * POSIX::setlocale, which calls this function. Therefore this function
266 * should be called directly only from this file and from
267 * POSIX::setlocale() */
268
27da23d5 269 dVAR;
68067e4e 270 UV i;
98994639 271
7918f24d
NC
272 PERL_ARGS_ASSERT_NEW_CTYPE;
273
c1284011 274 PL_in_utf8_CTYPE_locale = _is_cur_LC_category_utf8(LC_CTYPE);
31f05a37
KW
275
276 /* A UTF-8 locale gets standard rules. But note that code still has to
277 * handle this specially because of the three problematic code points */
278 if (PL_in_utf8_CTYPE_locale) {
279 Copy(PL_fold_latin1, PL_fold_locale, 256, U8);
280 }
281 else {
baa60164
KW
282 for (i = 0; i < 256; i++) {
283 if (isUPPER_LC((U8) i))
284 PL_fold_locale[i] = (U8) toLOWER_LC((U8) i);
285 else if (isLOWER_LC((U8) i))
286 PL_fold_locale[i] = (U8) toUPPER_LC((U8) i);
287 else
288 PL_fold_locale[i] = (U8) i;
289 }
31f05a37 290 }
98994639
HS
291
292#endif /* USE_LOCALE_CTYPE */
7918f24d 293 PERL_ARGS_ASSERT_NEW_CTYPE;
8772537c 294 PERL_UNUSED_ARG(newctype);
96a5add6 295 PERL_UNUSED_CONTEXT;
98994639
HS
296}
297
98994639 298void
8772537c 299Perl_new_collate(pTHX_ const char *newcoll)
98994639
HS
300{
301#ifdef USE_LOCALE_COLLATE
0d071d52
KW
302
303 /* Called after all libc setlocale() calls affecting LC_COLLATE, to tell
304 * core Perl this and that 'newcoll' is the name of the new locale.
305 *
306 * Any code changing the locale (outside this file) should use
307 * POSIX::setlocale, which calls this function. Therefore this function
308 * should be called directly only from this file and from
309 * POSIX::setlocale() */
310
98994639
HS
311 if (! newcoll) {
312 if (PL_collation_name) {
313 ++PL_collation_ix;
314 Safefree(PL_collation_name);
315 PL_collation_name = NULL;
316 }
317 PL_collation_standard = TRUE;
318 PL_collxfrm_base = 0;
319 PL_collxfrm_mult = 2;
320 return;
321 }
322
323 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
324 ++PL_collation_ix;
325 Safefree(PL_collation_name);
326 PL_collation_name = stdize_locale(savepv(newcoll));
a39edc4c 327 PL_collation_standard = isNAME_C_OR_POSIX(newcoll);
98994639
HS
328
329 {
330 /* 2: at most so many chars ('a', 'b'). */
331 /* 50: surely no system expands a char more. */
332#define XFRMBUFSIZE (2 * 50)
333 char xbuf[XFRMBUFSIZE];
8772537c
AL
334 const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE);
335 const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE);
336 const SSize_t mult = fb - fa;
e0601d3c 337 if (mult < 1 && !(fa == 0 && fb == 0))
ad49ad39
NC
338 Perl_croak(aTHX_ "panic: strxfrm() gets absurd - a => %"UVuf", ab => %"UVuf,
339 (UV) fa, (UV) fb);
eb160463 340 PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0;
98994639
HS
341 PL_collxfrm_mult = mult;
342 }
343 }
344
f2ce9e1c
JH
345#else
346 PERL_UNUSED_ARG(newcoll);
98994639
HS
347#endif /* USE_LOCALE_COLLATE */
348}
349
b385bb4d
KW
350#ifdef WIN32
351
352char *
353Perl_my_setlocale(pTHX_ int category, const char* locale)
354{
355 /* This, for Windows, emulates POSIX setlocale() behavior. There is no
356 * difference unless the input locale is "", which means on Windows to get
357 * the machine default, which is set via the computer's "Regional and
358 * Language Options" (or its current equivalent). In POSIX, it instead
359 * means to find the locale from the user's environment. This routine
360 * looks in the environment, and, if anything is found, uses that instead
361 * of going to the machine default. If there is no environment override,
362 * the machine default is used, as normal, by calling the real setlocale()
363 * with "". The POSIX behavior is to use the LC_ALL variable if set;
364 * otherwise to use the particular category's variable if set; otherwise to
365 * use the LANG variable. */
366
481465ea 367 bool override_LC_ALL = 0;
89f7b9aa
KW
368 char * result;
369
b385bb4d
KW
370 if (locale && strEQ(locale, "")) {
371# ifdef LC_ALL
372 locale = PerlEnv_getenv("LC_ALL");
373 if (! locale) {
374#endif
375 switch (category) {
376# ifdef LC_ALL
377 case LC_ALL:
481465ea 378 override_LC_ALL = TRUE;
b385bb4d
KW
379 break; /* We already know its variable isn't set */
380# endif
381# ifdef USE_LOCALE_TIME
382 case LC_TIME:
383 locale = PerlEnv_getenv("LC_TIME");
384 break;
385# endif
386# ifdef USE_LOCALE_CTYPE
387 case LC_CTYPE:
388 locale = PerlEnv_getenv("LC_CTYPE");
389 break;
390# endif
391# ifdef USE_LOCALE_COLLATE
392 case LC_COLLATE:
393 locale = PerlEnv_getenv("LC_COLLATE");
394 break;
395# endif
396# ifdef USE_LOCALE_MONETARY
397 case LC_MONETARY:
398 locale = PerlEnv_getenv("LC_MONETARY");
399 break;
400# endif
401# ifdef USE_LOCALE_NUMERIC
402 case LC_NUMERIC:
403 locale = PerlEnv_getenv("LC_NUMERIC");
404 break;
405# endif
406# ifdef USE_LOCALE_MESSAGES
407 case LC_MESSAGES:
408 locale = PerlEnv_getenv("LC_MESSAGES");
409 break;
410# endif
411 default:
412 /* This is a category, like PAPER_SIZE that we don't
413 * know about; and so can't provide a wrapper. */
414 break;
415 }
416 if (! locale) {
417 locale = PerlEnv_getenv("LANG");
481465ea 418 if (! locale) {
b385bb4d
KW
419 locale = "";
420 }
421 }
422# ifdef LC_ALL
423 }
424# endif
425 }
426
89f7b9aa
KW
427 result = setlocale(category, locale);
428
481465ea 429 if (! override_LC_ALL) {
89f7b9aa
KW
430 return result;
431 }
432
433 /* Here the input locale was LC_ALL, and we have set it to what is in the
481465ea
KW
434 * LANG variable or the system default if there is no LANG. But these have
435 * lower priority than the other LC_foo variables, so override it for each
436 * one that is set. (If they are set to "", it means to use the same thing
437 * we just set LC_ALL to, so can skip) */
89f7b9aa
KW
438# ifdef USE_LOCALE_TIME
439 result = PerlEnv_getenv("LC_TIME");
730252b2 440 if (result && strNE(result, "")) {
89f7b9aa
KW
441 setlocale(LC_TIME, result);
442 }
443# endif
444# ifdef USE_LOCALE_CTYPE
445 result = PerlEnv_getenv("LC_CTYPE");
730252b2 446 if (result && strNE(result, "")) {
89f7b9aa
KW
447 setlocale(LC_CTYPE, result);
448 }
449# endif
450# ifdef USE_LOCALE_COLLATE
451 result = PerlEnv_getenv("LC_COLLATE");
730252b2 452 if (result && strNE(result, "")) {
89f7b9aa
KW
453 setlocale(LC_COLLATE, result);
454 }
455# endif
456# ifdef USE_LOCALE_MONETARY
457 result = PerlEnv_getenv("LC_MONETARY");
730252b2 458 if (result && strNE(result, "")) {
89f7b9aa
KW
459 setlocale(LC_MONETARY, result);
460 }
461# endif
462# ifdef USE_LOCALE_NUMERIC
463 result = PerlEnv_getenv("LC_NUMERIC");
730252b2 464 if (result && strNE(result, "")) {
89f7b9aa
KW
465 setlocale(LC_NUMERIC, result);
466 }
467# endif
468# ifdef USE_LOCALE_MESSAGES
469 result = PerlEnv_getenv("LC_MESSAGES");
730252b2 470 if (result && strNE(result, "")) {
89f7b9aa
KW
471 setlocale(LC_MESSAGES, result);
472 }
473# endif
474
475 return setlocale(LC_ALL, NULL);
476
b385bb4d
KW
477}
478
479#endif
480
481
98994639
HS
482/*
483 * Initialize locale awareness.
484 */
485int
486Perl_init_i18nl10n(pTHX_ int printwarn)
487{
0e92a118
KW
488 /* printwarn is
489 *
490 * 0 if not to output warning when setup locale is bad
491 * 1 if to output warning based on value of PERL_BADLANG
492 * >1 if to output regardless of PERL_BADLANG
493 *
494 * returns
98994639 495 * 1 = set ok or not applicable,
0e92a118
KW
496 * 0 = fallback to a locale of lower priority
497 * -1 = fallback to all locales failed, not even to the C locale
98994639
HS
498 */
499
0e92a118
KW
500 int ok = 1;
501
98994639 502#if defined(USE_LOCALE)
98994639
HS
503#ifdef USE_LOCALE_CTYPE
504 char *curctype = NULL;
505#endif /* USE_LOCALE_CTYPE */
506#ifdef USE_LOCALE_COLLATE
507 char *curcoll = NULL;
508#endif /* USE_LOCALE_COLLATE */
509#ifdef USE_LOCALE_NUMERIC
510 char *curnum = NULL;
511#endif /* USE_LOCALE_NUMERIC */
512#ifdef __GLIBC__
7452cf6a 513 char * const language = PerlEnv_getenv("LANGUAGE");
98994639 514#endif
65ebb059 515
ccd65d51
KW
516 /* NULL uses the existing already set up locale */
517 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
518 ? NULL
519 : "";
c3fcd832
KW
520 const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
521 unsigned int trial_locales_count;
7452cf6a
AL
522 char * const lc_all = PerlEnv_getenv("LC_ALL");
523 char * const lang = PerlEnv_getenv("LANG");
98994639 524 bool setlocale_failure = FALSE;
65ebb059
KW
525 unsigned int i;
526 char *p;
527 const bool locwarn = (printwarn > 1 ||
528 (printwarn &&
529 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
0e92a118 530 bool done = FALSE;
6bce99ee
JH
531#ifdef WIN32
532 /* In some systems you can find out the system default locale
533 * and use that as the fallback locale. */
534# define SYSTEM_DEFAULT_LOCALE
535#endif
536#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059 537 const char *system_default_locale = NULL;
6bce99ee 538#endif
98994639 539
0e92a118
KW
540#ifndef LOCALE_ENVIRON_REQUIRED
541 PERL_UNUSED_VAR(done);
542#else
98994639
HS
543
544 /*
545 * Ultrix setlocale(..., "") fails if there are no environment
546 * variables from which to get a locale name.
547 */
548
b3e384bf 549# ifdef LC_ALL
98994639 550 if (lang) {
b385bb4d 551 if (my_setlocale(LC_ALL, setlocale_init))
98994639
HS
552 done = TRUE;
553 else
554 setlocale_failure = TRUE;
555 }
556 if (!setlocale_failure) {
b3e384bf 557# ifdef USE_LOCALE_CTYPE
56279a21 558 Safefree(curctype);
98994639 559 if (! (curctype =
b385bb4d 560 my_setlocale(LC_CTYPE,
98994639 561 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
ccd65d51 562 ? setlocale_init : NULL)))
98994639
HS
563 setlocale_failure = TRUE;
564 else
565 curctype = savepv(curctype);
b3e384bf
KW
566# endif /* USE_LOCALE_CTYPE */
567# ifdef USE_LOCALE_COLLATE
56279a21 568 Safefree(curcoll);
98994639 569 if (! (curcoll =
b385bb4d 570 my_setlocale(LC_COLLATE,
98994639 571 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
ccd65d51 572 ? setlocale_init : NULL)))
98994639
HS
573 setlocale_failure = TRUE;
574 else
575 curcoll = savepv(curcoll);
b3e384bf
KW
576# endif /* USE_LOCALE_COLLATE */
577# ifdef USE_LOCALE_NUMERIC
56279a21 578 Safefree(curnum);
98994639 579 if (! (curnum =
b385bb4d 580 my_setlocale(LC_NUMERIC,
98994639 581 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
ccd65d51 582 ? setlocale_init : NULL)))
98994639
HS
583 setlocale_failure = TRUE;
584 else
585 curnum = savepv(curnum);
b3e384bf 586# endif /* USE_LOCALE_NUMERIC */
a782673d
KW
587# ifdef USE_LOCALE_MESSAGES
588 if (! my_setlocale(LC_MESSAGES,
589 (!done && (lang || PerlEnv_getenv("LC_MESSAGES")))
590 ? setlocale_init : NULL))
591 {
592 setlocale_failure = TRUE;
593 }
594# endif /* USE_LOCALE_MESSAGES */
c835d6be
KW
595# ifdef USE_LOCALE_MONETARY
596 if (! my_setlocale(LC_MONETARY,
597 (!done && (lang || PerlEnv_getenv("LC_MONETARY")))
598 ? setlocale_init : NULL))
599 {
600 setlocale_failure = TRUE;
601 }
602# endif /* USE_LOCALE_MONETARY */
98994639
HS
603 }
604
b3e384bf 605# endif /* LC_ALL */
98994639
HS
606
607#endif /* !LOCALE_ENVIRON_REQUIRED */
608
65ebb059
KW
609 /* We try each locale in the list until we get one that works, or exhaust
610 * the list */
c3fcd832
KW
611 trial_locales[0] = setlocale_init;
612 trial_locales_count = 1;
65ebb059
KW
613 for (i= 0; i < trial_locales_count; i++) {
614 const char * trial_locale = trial_locales[i];
615
616 if (i > 0) {
617
618 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
619 * when i==0, but I (khw) don't think that behavior makes much
620 * sense */
621 setlocale_failure = FALSE;
622
6bce99ee
JH
623#ifdef SYSTEM_DEFAULT_LOCALE
624# ifdef WIN32
65ebb059
KW
625 /* On Windows machines, an entry of "" after the 0th means to use
626 * the system default locale, which we now proceed to get. */
627 if (strEQ(trial_locale, "")) {
628 unsigned int j;
629
630 /* Note that this may change the locale, but we are going to do
631 * that anyway just below */
632 system_default_locale = setlocale(LC_ALL, "");
633
634 /* Skip if invalid or it's already on the list of locales to
635 * try */
636 if (! system_default_locale) {
637 goto next_iteration;
638 }
639 for (j = 0; j < trial_locales_count; j++) {
640 if (strEQ(system_default_locale, trial_locales[j])) {
641 goto next_iteration;
642 }
643 }
644
645 trial_locale = system_default_locale;
646 }
6bce99ee
JH
647# endif /* WIN32 */
648#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
649 }
650
98994639 651#ifdef LC_ALL
7cd8b568 652 if (! my_setlocale(LC_ALL, trial_locale)) {
49c85077 653 setlocale_failure = TRUE;
7cd8b568
KW
654 }
655 else {
656 /* Since LC_ALL succeeded, it should have changed all the other
657 * categories it can to its value; so we massage things so that the
658 * setlocales below just return their category's current values.
659 * This adequately handles the case in NetBSD where LC_COLLATE may
660 * not be defined for a locale, and setting it individually will
661 * fail, whereas setting LC_ALL suceeds, leaving LC_COLLATE set to
662 * the POSIX locale. */
663 trial_locale = NULL;
664 }
98994639
HS
665#endif /* LC_ALL */
666
49c85077 667 if (!setlocale_failure) {
98994639 668#ifdef USE_LOCALE_CTYPE
49c85077
KW
669 Safefree(curctype);
670 if (! (curctype = my_setlocale(LC_CTYPE, trial_locale)))
671 setlocale_failure = TRUE;
672 else
673 curctype = savepv(curctype);
98994639
HS
674#endif /* USE_LOCALE_CTYPE */
675#ifdef USE_LOCALE_COLLATE
49c85077
KW
676 Safefree(curcoll);
677 if (! (curcoll = my_setlocale(LC_COLLATE, trial_locale)))
678 setlocale_failure = TRUE;
679 else
680 curcoll = savepv(curcoll);
98994639
HS
681#endif /* USE_LOCALE_COLLATE */
682#ifdef USE_LOCALE_NUMERIC
49c85077
KW
683 Safefree(curnum);
684 if (! (curnum = my_setlocale(LC_NUMERIC, trial_locale)))
685 setlocale_failure = TRUE;
686 else
687 curnum = savepv(curnum);
98994639 688#endif /* USE_LOCALE_NUMERIC */
a782673d
KW
689#ifdef USE_LOCALE_MESSAGES
690 if (! (my_setlocale(LC_MESSAGES, trial_locale)))
691 setlocale_failure = TRUE;
692#endif /* USE_LOCALE_MESSAGES */
c835d6be
KW
693#ifdef USE_LOCALE_MONETARY
694 if (! (my_setlocale(LC_MONETARY, trial_locale)))
695 setlocale_failure = TRUE;
696#endif /* USE_LOCALE_MONETARY */
697
49c85077
KW
698 if (! setlocale_failure) { /* Success */
699 break;
700 }
65ebb059 701 }
98994639 702
49c85077
KW
703 /* Here, something failed; will need to try a fallback. */
704 ok = 0;
65ebb059 705
49c85077
KW
706 if (i == 0) {
707 unsigned int j;
98994639 708
65ebb059 709 if (locwarn) { /* Output failure info only on the first one */
98994639
HS
710#ifdef LC_ALL
711
49c85077
KW
712 PerlIO_printf(Perl_error_log,
713 "perl: warning: Setting locale failed.\n");
98994639
HS
714
715#else /* !LC_ALL */
716
49c85077
KW
717 PerlIO_printf(Perl_error_log,
718 "perl: warning: Setting locale failed for the categories:\n\t");
98994639 719#ifdef USE_LOCALE_CTYPE
49c85077
KW
720 if (! curctype)
721 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
98994639
HS
722#endif /* USE_LOCALE_CTYPE */
723#ifdef USE_LOCALE_COLLATE
49c85077
KW
724 if (! curcoll)
725 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
98994639
HS
726#endif /* USE_LOCALE_COLLATE */
727#ifdef USE_LOCALE_NUMERIC
49c85077
KW
728 if (! curnum)
729 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
98994639 730#endif /* USE_LOCALE_NUMERIC */
a782673d 731 PerlIO_printf(Perl_error_log, "and possibly others\n");
98994639
HS
732
733#endif /* LC_ALL */
734
49c85077
KW
735 PerlIO_printf(Perl_error_log,
736 "perl: warning: Please check that your locale settings:\n");
98994639
HS
737
738#ifdef __GLIBC__
49c85077
KW
739 PerlIO_printf(Perl_error_log,
740 "\tLANGUAGE = %c%s%c,\n",
741 language ? '"' : '(',
742 language ? language : "unset",
743 language ? '"' : ')');
98994639
HS
744#endif
745
49c85077
KW
746 PerlIO_printf(Perl_error_log,
747 "\tLC_ALL = %c%s%c,\n",
748 lc_all ? '"' : '(',
749 lc_all ? lc_all : "unset",
750 lc_all ? '"' : ')');
98994639
HS
751
752#if defined(USE_ENVIRON_ARRAY)
49c85077
KW
753 {
754 char **e;
755 for (e = environ; *e; e++) {
756 if (strnEQ(*e, "LC_", 3)
757 && strnNE(*e, "LC_ALL=", 7)
758 && (p = strchr(*e, '=')))
759 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
760 (int)(p - *e), *e, p + 1);
761 }
762 }
98994639 763#else
49c85077
KW
764 PerlIO_printf(Perl_error_log,
765 "\t(possibly more locale environment variables)\n");
98994639
HS
766#endif
767
49c85077
KW
768 PerlIO_printf(Perl_error_log,
769 "\tLANG = %c%s%c\n",
770 lang ? '"' : '(',
771 lang ? lang : "unset",
772 lang ? '"' : ')');
98994639 773
49c85077
KW
774 PerlIO_printf(Perl_error_log,
775 " are supported and installed on your system.\n");
776 }
98994639 777
65ebb059
KW
778 /* Calculate what fallback locales to try. We have avoided this
779 * until we have to, becuase failure is quite unlikely. This will
780 * usually change the upper bound of the loop we are in.
781 *
782 * Since the system's default way of setting the locale has not
783 * found one that works, We use Perl's defined ordering: LC_ALL,
784 * LANG, and the C locale. We don't try the same locale twice, so
785 * don't add to the list if already there. (On POSIX systems, the
786 * LC_ALL element will likely be a repeat of the 0th element "",
787 * but there's no harm done by doing it explicitly */
788 if (lc_all) {
789 for (j = 0; j < trial_locales_count; j++) {
790 if (strEQ(lc_all, trial_locales[j])) {
791 goto done_lc_all;
792 }
793 }
794 trial_locales[trial_locales_count++] = lc_all;
795 }
796 done_lc_all:
98994639 797
65ebb059
KW
798 if (lang) {
799 for (j = 0; j < trial_locales_count; j++) {
800 if (strEQ(lang, trial_locales[j])) {
801 goto done_lang;
802 }
803 }
804 trial_locales[trial_locales_count++] = lang;
805 }
806 done_lang:
807
808#if defined(WIN32) && defined(LC_ALL)
809 /* For Windows, we also try the system default locale before "C".
810 * (If there exists a Windows without LC_ALL we skip this because
811 * it gets too complicated. For those, the "C" is the next
812 * fallback possibility). The "" is the same as the 0th element of
813 * the array, but the code at the loop above knows to treat it
814 * differently when not the 0th */
815 trial_locales[trial_locales_count++] = "";
816#endif
817
818 for (j = 0; j < trial_locales_count; j++) {
819 if (strEQ("C", trial_locales[j])) {
820 goto done_C;
821 }
822 }
823 trial_locales[trial_locales_count++] = "C";
98994639 824
65ebb059
KW
825 done_C: ;
826 } /* end of first time through the loop */
98994639 827
65ebb059
KW
828#ifdef WIN32
829 next_iteration: ;
830#endif
831
832 } /* end of looping through the trial locales */
833
834 if (ok < 1) { /* If we tried to fallback */
835 const char* msg;
836 if (! setlocale_failure) { /* fallback succeeded */
837 msg = "Falling back to";
838 }
839 else { /* fallback failed */
98994639 840
65ebb059
KW
841 /* We dropped off the end of the loop, so have to decrement i to
842 * get back to the value the last time through */
843 i--;
98994639 844
65ebb059
KW
845 ok = -1;
846 msg = "Failed to fall back to";
847
848 /* To continue, we should use whatever values we've got */
98994639 849#ifdef USE_LOCALE_CTYPE
49c85077
KW
850 Safefree(curctype);
851 curctype = savepv(setlocale(LC_CTYPE, NULL));
98994639
HS
852#endif /* USE_LOCALE_CTYPE */
853#ifdef USE_LOCALE_COLLATE
49c85077
KW
854 Safefree(curcoll);
855 curcoll = savepv(setlocale(LC_COLLATE, NULL));
98994639
HS
856#endif /* USE_LOCALE_COLLATE */
857#ifdef USE_LOCALE_NUMERIC
49c85077
KW
858 Safefree(curnum);
859 curnum = savepv(setlocale(LC_NUMERIC, NULL));
98994639 860#endif /* USE_LOCALE_NUMERIC */
65ebb059
KW
861 }
862
863 if (locwarn) {
864 const char * description;
865 const char * name = "";
866 if (strEQ(trial_locales[i], "C")) {
867 description = "the standard locale";
868 name = "C";
869 }
6bce99ee 870#ifdef SYSTEM_DEFAULT_LOCALE
65ebb059
KW
871 else if (strEQ(trial_locales[i], "")) {
872 description = "the system default locale";
873 if (system_default_locale) {
874 name = system_default_locale;
875 }
876 }
6bce99ee 877#endif /* SYSTEM_DEFAULT_LOCALE */
65ebb059
KW
878 else {
879 description = "a fallback locale";
880 name = trial_locales[i];
881 }
882 if (name && strNE(name, "")) {
883 PerlIO_printf(Perl_error_log,
884 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
885 }
886 else {
887 PerlIO_printf(Perl_error_log,
888 "perl: warning: %s %s.\n", msg, description);
889 }
890 }
891 } /* End of tried to fallback */
98994639
HS
892
893#ifdef USE_LOCALE_CTYPE
894 new_ctype(curctype);
895#endif /* USE_LOCALE_CTYPE */
896
897#ifdef USE_LOCALE_COLLATE
898 new_collate(curcoll);
899#endif /* USE_LOCALE_COLLATE */
900
901#ifdef USE_LOCALE_NUMERIC
902 new_numeric(curnum);
903#endif /* USE_LOCALE_NUMERIC */
b310b053 904
8ef6e574 905#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
49c85077
KW
906 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
907 * locale is UTF-8. If PL_utf8locale and PL_unicode (set by -C or by
908 * $ENV{PERL_UNICODE}) are true, perl.c:S_parse_body() will turn on the
909 * PerlIO :utf8 layer on STDIN, STDOUT, STDERR, _and_ the default open
910 * discipline. */
c1284011 911 PL_utf8locale = _is_cur_LC_category_utf8(LC_CTYPE);
49c85077 912
a05d7ebb 913 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
914 This is an alternative to using the -C command line switch
915 (the -C if present will override this). */
916 {
dd374669 917 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 918 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
919 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
920 PL_utf8cache = -1;
b310b053 921 }
ec71e770 922#endif
b310b053 923
98994639 924#ifdef USE_LOCALE_CTYPE
43c5f42d 925 Safefree(curctype);
98994639
HS
926#endif /* USE_LOCALE_CTYPE */
927#ifdef USE_LOCALE_COLLATE
43c5f42d 928 Safefree(curcoll);
98994639
HS
929#endif /* USE_LOCALE_COLLATE */
930#ifdef USE_LOCALE_NUMERIC
43c5f42d 931 Safefree(curnum);
98994639 932#endif /* USE_LOCALE_NUMERIC */
8ef6e574 933
f2ce9e1c
JH
934#else /* !USE_LOCALE */
935 PERL_UNUSED_ARG(printwarn);
8ef6e574
KW
936#endif /* USE_LOCALE */
937
98994639
HS
938 return ok;
939}
940
49c85077 941
98994639
HS
942#ifdef USE_LOCALE_COLLATE
943
944/*
945 * mem_collxfrm() is a bit like strxfrm() but with two important
946 * differences. First, it handles embedded NULs. Second, it allocates
947 * a bit more memory than needed for the transformed data itself.
948 * The real transformed data begins at offset sizeof(collationix).
949 * Please see sv_collxfrm() to see how this is used.
950 */
166f8a29 951
98994639
HS
952char *
953Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
954{
955 char *xbuf;
956 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
957
7918f24d
NC
958 PERL_ARGS_ASSERT_MEM_COLLXFRM;
959
98994639
HS
960 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
961 /* the +1 is for the terminating NUL. */
962
963 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 964 Newx(xbuf, xAlloc, char);
98994639
HS
965 if (! xbuf)
966 goto bad;
967
968 *(U32*)xbuf = PL_collation_ix;
969 xout = sizeof(PL_collation_ix);
970 for (xin = 0; xin < len; ) {
224e8ef5 971 Size_t xused;
98994639
HS
972
973 for (;;) {
974 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 975 if (xused >= PERL_INT_MAX)
98994639 976 goto bad;
eb160463 977 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
978 break;
979 xAlloc = (2 * xAlloc) + 1;
980 Renew(xbuf, xAlloc, char);
981 if (! xbuf)
982 goto bad;
983 }
984
985 xin += strlen(s + xin) + 1;
986 xout += xused;
987
988 /* Embedded NULs are understood but silently skipped
989 * because they make no sense in locale collation. */
990 }
991
992 xbuf[xout] = '\0';
993 *xlen = xout - sizeof(PL_collation_ix);
994 return xbuf;
995
996 bad:
997 Safefree(xbuf);
998 *xlen = 0;
999 return NULL;
1000}
1001
1002#endif /* USE_LOCALE_COLLATE */
1003
8ef6e574
KW
1004#ifdef USE_LOCALE
1005
c1284011
KW
1006bool
1007Perl__is_cur_LC_category_utf8(pTHX_ int category)
7d74bb61
KW
1008{
1009 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
1010 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b
KW
1011 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
1012 * could give the wrong result. It errs on the side of not being a UTF-8
1013 * locale. */
7d74bb61
KW
1014
1015 char *save_input_locale = NULL;
7d74bb61
KW
1016 STRLEN final_pos;
1017
8ef6e574 1018#ifdef LC_ALL
7d74bb61 1019 assert(category != LC_ALL);
8ef6e574 1020#endif
7d74bb61
KW
1021
1022 /* First dispose of the trivial cases */
b07fffd1 1023 save_input_locale = setlocale(category, NULL);
7d74bb61 1024 if (! save_input_locale) {
69014004
KW
1025 DEBUG_L(PerlIO_printf(Perl_debug_log,
1026 "Could not find current locale for category %d\n",
1027 category));
7d74bb61
KW
1028 return FALSE; /* XXX maybe should croak */
1029 }
b07fffd1 1030 save_input_locale = stdize_locale(savepv(save_input_locale));
a39edc4c 1031 if (isNAME_C_OR_POSIX(save_input_locale)) {
69014004
KW
1032 DEBUG_L(PerlIO_printf(Perl_debug_log,
1033 "Current locale for category %d is %s\n",
1034 category, save_input_locale));
b07fffd1 1035 Safefree(save_input_locale);
7d74bb61
KW
1036 return FALSE;
1037 }
1038
1d958db2
KW
1039#if defined(USE_LOCALE_CTYPE) \
1040 && (defined(MB_CUR_MAX) || (defined(HAS_NL_LANGINFO) && defined(CODESET)))
7d74bb61 1041
1d958db2 1042 { /* Next try nl_langinfo or MB_CUR_MAX if available */
7d74bb61
KW
1043
1044 char *save_ctype_locale = NULL;
119ee68b 1045 bool is_utf8;
7d74bb61 1046
119ee68b 1047 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
1048
1049 /* Get the current LC_CTYPE locale */
1050 save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL)));
1051 if (! save_ctype_locale) {
69014004
KW
1052 DEBUG_L(PerlIO_printf(Perl_debug_log,
1053 "Could not find current locale for LC_CTYPE\n"));
7d74bb61
KW
1054 goto cant_use_nllanginfo;
1055 }
1056
1057 /* If LC_CTYPE and the desired category use the same locale, this
1058 * means that finding the value for LC_CTYPE is the same as finding
1059 * the value for the desired category. Otherwise, switch LC_CTYPE
1060 * to the desired category's locale */
1061 if (strEQ(save_ctype_locale, save_input_locale)) {
1062 Safefree(save_ctype_locale);
1063 save_ctype_locale = NULL;
1064 }
1065 else if (! setlocale(LC_CTYPE, save_input_locale)) {
69014004
KW
1066 DEBUG_L(PerlIO_printf(Perl_debug_log,
1067 "Could not change LC_CTYPE locale to %s\n",
1068 save_input_locale));
7d74bb61
KW
1069 Safefree(save_ctype_locale);
1070 goto cant_use_nllanginfo;
1071 }
1072 }
1073
69014004
KW
1074 DEBUG_L(PerlIO_printf(Perl_debug_log, "Current LC_CTYPE locale=%s\n",
1075 save_input_locale));
1076
7d74bb61 1077 /* Here the current LC_CTYPE is set to the locale of the category whose
1d958db2
KW
1078 * information is desired. This means that nl_langinfo() and MB_CUR_MAX
1079 * should give the correct results */
119ee68b 1080
1d958db2
KW
1081# if defined(HAS_NL_LANGINFO) && defined(CODESET)
1082 {
1083 char *codeset = savepv(nl_langinfo(CODESET));
1084 if (codeset && strNE(codeset, "")) {
119ee68b 1085
1d958db2
KW
1086 /* If we switched LC_CTYPE, switch back */
1087 if (save_ctype_locale) {
1088 setlocale(LC_CTYPE, save_ctype_locale);
1089 Safefree(save_ctype_locale);
1090 }
1091
1092 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
1093 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
1094
69014004
KW
1095 DEBUG_L(PerlIO_printf(Perl_debug_log,
1096 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
1097 codeset, is_utf8));
1d958db2
KW
1098 Safefree(codeset);
1099 Safefree(save_input_locale);
1100 return is_utf8;
1101 }
0ac78434 1102 Safefree(codeset);
119ee68b
KW
1103 }
1104
1d958db2
KW
1105# endif
1106# ifdef MB_CUR_MAX
1107
1108 /* Here, either we don't have nl_langinfo, or it didn't return a
1109 * codeset. Try MB_CUR_MAX */
1110
119ee68b
KW
1111 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
1112 * Unicode code point. Since UTF-8 is the only non-single byte
1113 * encoding we handle, we just say any such encoding is UTF-8, and if
1114 * turns out to be wrong, other things will fail */
1115 is_utf8 = MB_CUR_MAX >= 4;
1116
69014004
KW
1117 DEBUG_L(PerlIO_printf(Perl_debug_log,
1118 "\tMB_CUR_MAX=%d; ?UTF8 locale=%d\n",
1119 (int) MB_CUR_MAX, is_utf8));
1120
119ee68b
KW
1121 Safefree(save_input_locale);
1122
1123# ifdef HAS_MBTOWC
1124
1125 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
1126 * since they are both in the C99 standard. We can feed a known byte
1127 * string to the latter function, and check that it gives the expected
1128 * result */
1129 if (is_utf8) {
1130 wchar_t wc;
a6715020 1131 GCC_DIAG_IGNORE(-Wunused-result);
1d958db2 1132 (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */
a6715020 1133 GCC_DIAG_RESTORE;
69014004 1134 errno = 0;
f019f68f 1135 if ((size_t)mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
119ee68b
KW
1136 != strlen(HYPHEN_UTF8)
1137 || wc != (wchar_t) 0x2010)
1138 {
1139 is_utf8 = FALSE;
69014004
KW
1140 DEBUG_L(PerlIO_printf(Perl_debug_log, "\thyphen=U+%x\n", wc));
1141 DEBUG_L(PerlIO_printf(Perl_debug_log,
1142 "\treturn from mbtowc=%d; errno=%d; ?UTF8 locale=0\n",
1143 mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)), errno));
119ee68b
KW
1144 }
1145 }
119ee68b
KW
1146# endif
1147
1d958db2
KW
1148 /* If we switched LC_CTYPE, switch back */
1149 if (save_ctype_locale) {
1150 setlocale(LC_CTYPE, save_ctype_locale);
1151 Safefree(save_ctype_locale);
119ee68b 1152 }
7d74bb61 1153
1d958db2 1154 return is_utf8;
119ee68b 1155# endif
7d74bb61 1156 }
119ee68b 1157
7d74bb61
KW
1158 cant_use_nllanginfo:
1159
1160#endif /* HAS_NL_LANGINFO etc */
1161
1162 /* nl_langinfo not available or failed somehow. Look at the locale name to
751e9426 1163 * see if it matches qr/UTF -? 8 /ix */
7d74bb61
KW
1164
1165 final_pos = strlen(save_input_locale) - 1;
751e9426
KW
1166 if (final_pos >= 3) {
1167 char *name = save_input_locale;
1168
1169 /* Find next 'U' or 'u' and look from there */
1170 while ((name += strcspn(name, "Uu") + 1)
1171 <= save_input_locale + final_pos - 2)
7d74bb61 1172 {
751e9426
KW
1173 if (toFOLD(*(name)) != 't'
1174 || toFOLD(*(name + 1)) != 'f')
1175 {
1176 continue;
1177 }
1178 name += 2;
1179 if (*(name) == '-') {
1180 if ((name > save_input_locale + final_pos - 1)) {
1181 break;
1182 }
1183 name++;
1184 }
1185 if (*(name) == '8') {
1186 Safefree(save_input_locale);
69014004
KW
1187 DEBUG_L(PerlIO_printf(Perl_debug_log,
1188 "Locale %s ends with UTF-8 in name\n",
1189 save_input_locale));
751e9426
KW
1190 return TRUE;
1191 }
7d74bb61 1192 }
69014004
KW
1193 DEBUG_L(PerlIO_printf(Perl_debug_log,
1194 "Locale %s doesn't end with UTF-8 in name\n",
1195 save_input_locale));
7d74bb61
KW
1196 }
1197
1198#ifdef WIN32
1199 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
1200 if (final_pos >= 4
1201 && *(save_input_locale + final_pos - 0) == '1'
1202 && *(save_input_locale + final_pos - 1) == '0'
1203 && *(save_input_locale + final_pos - 2) == '0'
1204 && *(save_input_locale + final_pos - 3) == '5'
1205 && *(save_input_locale + final_pos - 4) == '6')
1206 {
69014004
KW
1207 DEBUG_L(PerlIO_printf(Perl_debug_log,
1208 "Locale %s ends with 10056 in name, is UTF-8 locale\n",
1209 save_input_locale));
d37662c0 1210 Safefree(save_input_locale);
7d74bb61
KW
1211 return TRUE;
1212 }
1213#endif
1214
fa9b773e
KW
1215 /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
1216 if (instr(save_input_locale, "8859")) {
69014004
KW
1217 DEBUG_L(PerlIO_printf(Perl_debug_log,
1218 "Locale %s has 8859 in name, not UTF-8 locale\n",
1219 save_input_locale));
d37662c0 1220 Safefree(save_input_locale);
fa9b773e
KW
1221 return FALSE;
1222 }
1223
1224#ifdef HAS_LOCALECONV
1225
1226# ifdef USE_LOCALE_MONETARY
1227
1228 /* Here, there is nothing in the locale name to indicate whether the locale
1229 * is UTF-8 or not. This "name", the return of setlocale(), is actually
1230 * defined to be opaque, so we can't really rely on the absence of various
1231 * substrings in the name to indicate its UTF-8ness. Look at the locale's
1232 * currency symbol. Often that will be in the native script, and if the
1233 * symbol isn't in UTF-8, we know that the locale isn't. If it is
1234 * non-ASCII UTF-8, we infer that the locale is too.
1235 * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the
1236 * locale of the desired category, if it isn't that locale already */
1237
1238 {
1239 char *save_monetary_locale = NULL;
1240 bool illegal_utf8 = FALSE;
1241 bool only_ascii = FALSE;
1242 const struct lconv* const lc = localeconv();
1243
1244 if (category != LC_MONETARY) {
1245
1246 save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY,
1247 NULL)));
1248 if (! save_monetary_locale) {
69014004
KW
1249 DEBUG_L(PerlIO_printf(Perl_debug_log,
1250 "Could not find current locale for LC_MONETARY\n"));
fa9b773e
KW
1251 goto cant_use_monetary;
1252 }
1253
1254 if (strNE(save_monetary_locale, save_input_locale)) {
1255 if (! setlocale(LC_MONETARY, save_input_locale)) {
69014004
KW
1256 DEBUG_L(PerlIO_printf(Perl_debug_log,
1257 "Could not change LC_MONETARY locale to %s\n",
1258 save_input_locale));
fa9b773e
KW
1259 Safefree(save_monetary_locale);
1260 goto cant_use_monetary;
1261 }
1262 }
1263 }
1264
1265 /* Here the current LC_MONETARY is set to the locale of the category
1266 * whose information is desired. */
1267
1268 if (lc && lc->currency_symbol) {
1269 if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) {
69014004
KW
1270 DEBUG_L(PerlIO_printf(Perl_debug_log,
1271 "Currency symbol for %s is not legal UTF-8\n",
1272 save_input_locale));
fa9b773e
KW
1273 illegal_utf8 = TRUE;
1274 }
1275 else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) {
69014004 1276 DEBUG_L(PerlIO_printf(Perl_debug_log, "Currency symbol for %s contains only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
fa9b773e
KW
1277 only_ascii = TRUE;
1278 }
1279 }
1280
1281 /* If we changed it, restore LC_MONETARY to its original locale */
1282 if (save_monetary_locale) {
1283 setlocale(LC_MONETARY, save_monetary_locale);
1284 Safefree(save_monetary_locale);
1285 }
1286
1287 Safefree(save_input_locale);
1288
1289 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise
1290 * assume the locale is UTF-8 if and only if the symbol is non-ascii
1291 * UTF-8. (We can't really tell if the locale is UTF-8 or not if the
1292 * symbol is just a '$', so we err on the side of it not being UTF-8)
1293 * */
69014004
KW
1294 DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", (illegal_utf8)
1295 ? FALSE
1296 : ! only_ascii));
fa9b773e
KW
1297 return (illegal_utf8)
1298 ? FALSE
1299 : ! only_ascii;
1300
1301 }
1302 cant_use_monetary:
1303
1304# endif /* USE_LOCALE_MONETARY */
1305#endif /* HAS_LOCALECONV */
1306
855aeb93
JH
1307#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES)
1308
1309/* This code is ifdefd out because it was found to not be necessary in testing
1310 * on our dromedary test machine, which has over 700 locales. There, looking
1311 * at just the currency symbol gave essentially the same results as doing this
1312 * extra work. Executing this also caused segfaults in miniperl. I left it in
1313 * so as to avoid rewriting it if real-world experience indicates that
1314 * dromedary is an outlier. Essentially, instead of returning abpve if we
1315 * haven't found illegal utf8, we continue on and examine all the strerror()
1316 * messages on the platform for utf8ness. If all are ASCII, we still don't
1317 * know the answer; but otherwise we have a pretty good indication of the
1318 * utf8ness. The reason this doesn't necessarily help much is that the
1319 * messages may not have been translated into the locale. The currency symbol
1320 * is much more likely to have been translated. The code below would need to
1321 * be altered somewhat to just be a continuation of testing the currency
1322 * symbol. */
1323 int e;
1324 unsigned int failures = 0, non_ascii = 0;
1325 char *save_messages_locale = NULL;
1326
1327 /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the
1328 * desired category, if it isn't that locale already */
1329
1330 if (category != LC_MESSAGES) {
1331
1332 save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES,
1333 NULL)));
1334 if (! save_messages_locale) {
1335 goto cant_use_messages;
1336 }
1337
1338 if (strEQ(save_messages_locale, save_input_locale)) {
1339 Safefree(save_input_locale);
1340 }
1341 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
1342 Safefree(save_messages_locale);
1343 goto cant_use_messages;
1344 }
1345 }
1346
1347 /* Here the current LC_MESSAGES is set to the locale of the category
1348 * whose information is desired. Look through all the messages */
1349
1350 for (e = 0;
1351#ifdef HAS_SYS_ERRLIST
1352 e <= sys_nerr
1353#endif
1354 ; e++)
1355 {
1356 const U8* const errmsg = (U8 *) Strerror(e) ;
1357 if (!errmsg)
1358 break;
1359 if (! is_utf8_string(errmsg, 0)) {
1360 failures++;
1361 break;
1362 }
1363 else if (! is_ascii_string(errmsg, 0)) {
1364 non_ascii++;
1365 }
1366 }
1367
1368 /* And, if we changed it, restore LC_MESSAGES to its original locale */
1369 if (save_messages_locale) {
1370 setlocale(LC_MESSAGES, save_messages_locale);
1371 Safefree(save_messages_locale);
1372 }
1373
1374 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
1375 * any non-ascii means it is one; otherwise we assume it isn't */
1376 return (failures) ? FALSE : non_ascii;
1377
1378 }
1379 cant_use_messages:
1380
1381#endif
fa9b773e 1382
69014004
KW
1383 DEBUG_L(PerlIO_printf(Perl_debug_log,
1384 "Assuming locale %s is not a UTF-8 locale\n",
1385 save_input_locale));
fa9b773e 1386 Safefree(save_input_locale);
7d74bb61
KW
1387 return FALSE;
1388}
1389
8ef6e574 1390#endif
7d74bb61 1391
d6ded950
KW
1392
1393bool
1394Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
1395{
1a4f13e1 1396 dVAR;
d6ded950
KW
1397 /* Internal function which returns if we are in the scope of a pragma that
1398 * enables the locale category 'category'. 'compiling' should indicate if
1399 * this is during the compilation phase (TRUE) or not (FALSE). */
1400
1401 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
1402
1403 SV *categories = cop_hints_fetch_pvs(cop, "locale", 0);
1404 if (! categories || categories == &PL_sv_placeholder) {
1405 return FALSE;
1406 }
1407
1408 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
1409 * a valid unsigned */
1410 assert(category >= -1);
1411 return cBOOL(SvUV(categories) & (1U << (category + 1)));
1412}
1413
2c6ee1a7
KW
1414char *
1415Perl_my_strerror(pTHX_ const int errnum) {
1416
1417 /* Uses C locale for the error text unless within scope of 'use locale' for
1418 * LC_MESSAGES */
1419
1420#ifdef USE_LOCALE_MESSAGES
8ee4b769 1421 if (! IN_LC(LC_MESSAGES)) {
2c6ee1a7 1422 char * save_locale = setlocale(LC_MESSAGES, NULL);
a39edc4c 1423 if (! isNAME_C_OR_POSIX(save_locale)) {
2c6ee1a7
KW
1424 char *errstr;
1425
1426 /* The next setlocale likely will zap this, so create a copy */
1427 save_locale = savepv(save_locale);
1428
1429 setlocale(LC_MESSAGES, "C");
1430
1431 /* This points to the static space in Strerror, with all its
1432 * limitations */
1433 errstr = Strerror(errnum);
1434
1435 setlocale(LC_MESSAGES, save_locale);
1436 Safefree(save_locale);
1437 return errstr;
1438 }
1439 }
1440#endif
1441
1442 return Strerror(errnum);
1443}
1444
66610fdd
RGS
1445/*
1446 * Local variables:
1447 * c-indentation-style: bsd
1448 * c-basic-offset: 4
14d04a33 1449 * indent-tabs-mode: nil
66610fdd
RGS
1450 * End:
1451 *
14d04a33 1452 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1453 */