This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Rename an internal flag
[perl5.git] / locale.c
CommitLineData
98994639
HS
1/* locale.c
2 *
1129b882
NC
3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
4 * 2002, 2003, 2005, 2006, 2007, 2008 by Larry Wall and others
98994639
HS
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
9 */
10
11/*
4ac71550 12 * A Elbereth Gilthoniel,
cdad3b53 13 * silivren penna míriel
4ac71550 14 * o menel aglar elenath!
cdad3b53 15 * Na-chaered palan-díriel
4ac71550
TC
16 * o galadhremmin ennorath,
17 * Fanuilos, le linnathon
18 * nef aear, si nef aearon!
19 *
20 * [p.238 of _The Lord of the Rings_, II/i: "Many Meetings"]
98994639
HS
21 */
22
166f8a29
DM
23/* utility functions for handling locale-specific stuff like what
24 * character represents the decimal point.
0d071d52
KW
25 *
26 * All C programs have an underlying locale. Perl generally doesn't pay any
27 * attention to it except within the scope of a 'use locale'. For most
28 * categories, it accomplishes this by just using different operations if it is
29 * in such scope than if not. However, various libc functions called by Perl
30 * are affected by the LC_NUMERIC category, so there are macros in perl.h that
31 * are used to toggle between the current locale and the C locale depending on
32 * the desired behavior of those functions at the moment.
166f8a29
DM
33 */
34
98994639
HS
35#include "EXTERN.h"
36#define PERL_IN_LOCALE_C
37#include "perl.h"
38
b310b053
JH
39#ifdef I_LANGINFO
40# include <langinfo.h>
41#endif
42
a4af207c
JH
43#include "reentr.h"
44
8ef6e574
KW
45#ifdef USE_LOCALE
46
98994639 47/*
0d071d52
KW
48 * Standardize the locale name from a string returned by 'setlocale', possibly
49 * modifying that string.
98994639 50 *
0ef2a2b2 51 * The typical return value of setlocale() is either
98994639
HS
52 * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL
53 * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL
54 * (the space-separated values represent the various sublocales,
0ef2a2b2 55 * in some unspecified order). This is not handled by this function.
98994639
HS
56 *
57 * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n",
0ef2a2b2
KW
58 * which is harmful for further use of the string in setlocale(). This
59 * function removes the trailing new line and everything up through the '='
98994639
HS
60 *
61 */
62STATIC char *
63S_stdize_locale(pTHX_ char *locs)
64{
7452cf6a 65 const char * const s = strchr(locs, '=');
98994639
HS
66 bool okay = TRUE;
67
7918f24d
NC
68 PERL_ARGS_ASSERT_STDIZE_LOCALE;
69
8772537c
AL
70 if (s) {
71 const char * const t = strchr(s, '.');
98994639 72 okay = FALSE;
8772537c
AL
73 if (t) {
74 const char * const u = strchr(t, '\n');
75 if (u && (u[1] == 0)) {
76 const STRLEN len = u - s;
77 Move(s + 1, locs, len, char);
78 locs[len] = 0;
79 okay = TRUE;
98994639
HS
80 }
81 }
82 }
83
84 if (!okay)
85 Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs);
86
87 return locs;
88}
89
8ef6e574
KW
90#endif
91
98994639
HS
92void
93Perl_set_numeric_radix(pTHX)
94{
95#ifdef USE_LOCALE_NUMERIC
97aff369 96 dVAR;
98994639 97# ifdef HAS_LOCALECONV
7452cf6a 98 const struct lconv* const lc = localeconv();
98994639 99
98994639
HS
100 if (lc && lc->decimal_point) {
101 if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) {
102 SvREFCNT_dec(PL_numeric_radix_sv);
a0714e2c 103 PL_numeric_radix_sv = NULL;
98994639
HS
104 }
105 else {
106 if (PL_numeric_radix_sv)
107 sv_setpv(PL_numeric_radix_sv, lc->decimal_point);
108 else
109 PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0);
28acfe03
KW
110 if (! is_ascii_string((U8 *) lc->decimal_point, 0)
111 && is_utf8_string((U8 *) lc->decimal_point, 0)
112 && is_cur_LC_category_utf8(LC_NUMERIC))
113 {
114 SvUTF8_on(PL_numeric_radix_sv);
115 }
98994639
HS
116 }
117 }
118 else
a0714e2c 119 PL_numeric_radix_sv = NULL;
98994639
HS
120# endif /* HAS_LOCALECONV */
121#endif /* USE_LOCALE_NUMERIC */
122}
123
98994639 124void
8772537c 125Perl_new_numeric(pTHX_ const char *newnum)
98994639
HS
126{
127#ifdef USE_LOCALE_NUMERIC
0d071d52
KW
128
129 /* Called after all libc setlocale() calls affecting LC_NUMERIC, to tell
130 * core Perl this and that 'newnum' is the name of the new locale.
131 * It installs this locale as the current underlying default.
132 *
133 * The default locale and the C locale can be toggled between by use of the
134 * set_numeric_local() and set_numeric_standard() functions, which should
135 * probably not be called directly, but only via macros like
136 * SET_NUMERIC_STANDARD() in perl.h.
137 *
138 * The toggling is necessary mainly so that a non-dot radix decimal point
139 * character can be output, while allowing internal calculations to use a
140 * dot.
141 *
142 * This sets several interpreter-level variables:
143 * PL_numeric_name The default locale's name: a copy of 'newnum'
144 * PL_numeric_local A boolean indicating if the toggled state is such
145 * that the current locale is the default locale
146 * PL_numeric_standard A boolean indicating if the toggled state is such
147 * that the current locale is the C locale
148 * Note that both of the last two variables can be true at the same time,
149 * if the underlying locale is C. (Toggling is a no-op under these
150 * circumstances.)
151 *
152 * Any code changing the locale (outside this file) should use
153 * POSIX::setlocale, which calls this function. Therefore this function
154 * should be called directly only from this file and from
155 * POSIX::setlocale() */
156
b03f34cf 157 char *save_newnum;
97aff369 158 dVAR;
98994639
HS
159
160 if (! newnum) {
43c5f42d
NC
161 Safefree(PL_numeric_name);
162 PL_numeric_name = NULL;
98994639
HS
163 PL_numeric_standard = TRUE;
164 PL_numeric_local = TRUE;
165 return;
166 }
167
b03f34cf
KW
168 save_newnum = stdize_locale(savepv(newnum));
169 if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) {
98994639 170 Safefree(PL_numeric_name);
b03f34cf 171 PL_numeric_name = save_newnum;
b03f34cf 172 }
98994639 173
e19f01cb
KW
174 PL_numeric_standard = ((*save_newnum == 'C' && save_newnum[1] == '\0')
175 || strEQ(save_newnum, "POSIX"));
176 PL_numeric_local = TRUE;
177 set_numeric_radix();
6959d69d 178
98994639
HS
179#endif /* USE_LOCALE_NUMERIC */
180}
181
182void
183Perl_set_numeric_standard(pTHX)
184{
185#ifdef USE_LOCALE_NUMERIC
97aff369 186 dVAR;
98994639 187
0d071d52
KW
188 /* Toggle the LC_NUMERIC locale to C, if not already there. Probably
189 * should use the macros like SET_NUMERIC_STANDARD() in perl.h instead of
190 * calling this directly. */
191
98994639
HS
192 if (! PL_numeric_standard) {
193 setlocale(LC_NUMERIC, "C");
194 PL_numeric_standard = TRUE;
195 PL_numeric_local = FALSE;
196 set_numeric_radix();
197 }
198
199#endif /* USE_LOCALE_NUMERIC */
200}
201
202void
203Perl_set_numeric_local(pTHX)
204{
205#ifdef USE_LOCALE_NUMERIC
97aff369 206 dVAR;
98994639 207
0d071d52
KW
208 /* Toggle the LC_NUMERIC locale to the current underlying default, if not
209 * already there. Probably should use the macros like SET_NUMERIC_LOCAL()
210 * in perl.h instead of calling this directly. */
211
98994639
HS
212 if (! PL_numeric_local) {
213 setlocale(LC_NUMERIC, PL_numeric_name);
214 PL_numeric_standard = FALSE;
215 PL_numeric_local = TRUE;
216 set_numeric_radix();
217 }
218
219#endif /* USE_LOCALE_NUMERIC */
220}
221
222/*
223 * Set up for a new ctype locale.
224 */
225void
8772537c 226Perl_new_ctype(pTHX_ const char *newctype)
98994639
HS
227{
228#ifdef USE_LOCALE_CTYPE
0d071d52
KW
229
230 /* Called after all libc setlocale() calls affecting LC_CTYPE, to tell
231 * core Perl this and that 'newctype' is the name of the new locale.
232 *
233 * This function sets up the folding arrays for all 256 bytes, assuming
234 * that tofold() is tolc() since fold case is not a concept in POSIX,
235 *
236 * Any code changing the locale (outside this file) should use
237 * POSIX::setlocale, which calls this function. Therefore this function
238 * should be called directly only from this file and from
239 * POSIX::setlocale() */
240
27da23d5 241 dVAR;
68067e4e 242 UV i;
98994639 243
7918f24d
NC
244 PERL_ARGS_ASSERT_NEW_CTYPE;
245
77806dea
KW
246 for (i = 0; i < 256; i++) {
247 if (isUPPER_LC((U8) i))
248 PL_fold_locale[i] = toLOWER_LC((U8) i);
249 else if (isLOWER_LC((U8) i))
250 PL_fold_locale[i] = toUPPER_LC((U8) i);
98994639 251 else
e7e76d1b 252 PL_fold_locale[i] = (U8) i;
98994639
HS
253 }
254
255#endif /* USE_LOCALE_CTYPE */
7918f24d 256 PERL_ARGS_ASSERT_NEW_CTYPE;
8772537c 257 PERL_UNUSED_ARG(newctype);
96a5add6 258 PERL_UNUSED_CONTEXT;
98994639
HS
259}
260
98994639 261void
8772537c 262Perl_new_collate(pTHX_ const char *newcoll)
98994639
HS
263{
264#ifdef USE_LOCALE_COLLATE
0d071d52
KW
265
266 /* Called after all libc setlocale() calls affecting LC_COLLATE, to tell
267 * core Perl this and that 'newcoll' is the name of the new locale.
268 *
269 * Any code changing the locale (outside this file) should use
270 * POSIX::setlocale, which calls this function. Therefore this function
271 * should be called directly only from this file and from
272 * POSIX::setlocale() */
273
97aff369 274 dVAR;
98994639
HS
275
276 if (! newcoll) {
277 if (PL_collation_name) {
278 ++PL_collation_ix;
279 Safefree(PL_collation_name);
280 PL_collation_name = NULL;
281 }
282 PL_collation_standard = TRUE;
283 PL_collxfrm_base = 0;
284 PL_collxfrm_mult = 2;
285 return;
286 }
287
288 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
289 ++PL_collation_ix;
290 Safefree(PL_collation_name);
291 PL_collation_name = stdize_locale(savepv(newcoll));
770526c1
NC
292 PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0')
293 || strEQ(newcoll, "POSIX"));
98994639
HS
294
295 {
296 /* 2: at most so many chars ('a', 'b'). */
297 /* 50: surely no system expands a char more. */
298#define XFRMBUFSIZE (2 * 50)
299 char xbuf[XFRMBUFSIZE];
8772537c
AL
300 const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE);
301 const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE);
302 const SSize_t mult = fb - fa;
e0601d3c 303 if (mult < 1 && !(fa == 0 && fb == 0))
ad49ad39
NC
304 Perl_croak(aTHX_ "panic: strxfrm() gets absurd - a => %"UVuf", ab => %"UVuf,
305 (UV) fa, (UV) fb);
eb160463 306 PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0;
98994639
HS
307 PL_collxfrm_mult = mult;
308 }
309 }
310
311#endif /* USE_LOCALE_COLLATE */
312}
313
314/*
315 * Initialize locale awareness.
316 */
317int
318Perl_init_i18nl10n(pTHX_ int printwarn)
319{
320 int ok = 1;
321 /* returns
322 * 1 = set ok or not applicable,
323 * 0 = fallback to C locale,
324 * -1 = fallback to C locale failed
325 */
326
327#if defined(USE_LOCALE)
97aff369 328 dVAR;
98994639
HS
329
330#ifdef USE_LOCALE_CTYPE
331 char *curctype = NULL;
332#endif /* USE_LOCALE_CTYPE */
333#ifdef USE_LOCALE_COLLATE
334 char *curcoll = NULL;
335#endif /* USE_LOCALE_COLLATE */
336#ifdef USE_LOCALE_NUMERIC
337 char *curnum = NULL;
338#endif /* USE_LOCALE_NUMERIC */
339#ifdef __GLIBC__
7452cf6a 340 char * const language = PerlEnv_getenv("LANGUAGE");
98994639 341#endif
ccd65d51
KW
342 /* NULL uses the existing already set up locale */
343 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
344 ? NULL
345 : "";
7452cf6a
AL
346 char * const lc_all = PerlEnv_getenv("LC_ALL");
347 char * const lang = PerlEnv_getenv("LANG");
98994639
HS
348 bool setlocale_failure = FALSE;
349
350#ifdef LOCALE_ENVIRON_REQUIRED
351
352 /*
353 * Ultrix setlocale(..., "") fails if there are no environment
354 * variables from which to get a locale name.
355 */
356
357 bool done = FALSE;
358
b3e384bf 359# ifdef LC_ALL
98994639 360 if (lang) {
ccd65d51 361 if (setlocale(LC_ALL, setlocale_init))
98994639
HS
362 done = TRUE;
363 else
364 setlocale_failure = TRUE;
365 }
366 if (!setlocale_failure) {
b3e384bf 367# ifdef USE_LOCALE_CTYPE
56279a21 368 Safefree(curctype);
98994639
HS
369 if (! (curctype =
370 setlocale(LC_CTYPE,
371 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
ccd65d51 372 ? setlocale_init : NULL)))
98994639
HS
373 setlocale_failure = TRUE;
374 else
375 curctype = savepv(curctype);
b3e384bf
KW
376# endif /* USE_LOCALE_CTYPE */
377# ifdef USE_LOCALE_COLLATE
56279a21 378 Safefree(curcoll);
98994639
HS
379 if (! (curcoll =
380 setlocale(LC_COLLATE,
381 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
ccd65d51 382 ? setlocale_init : NULL)))
98994639
HS
383 setlocale_failure = TRUE;
384 else
385 curcoll = savepv(curcoll);
b3e384bf
KW
386# endif /* USE_LOCALE_COLLATE */
387# ifdef USE_LOCALE_NUMERIC
56279a21 388 Safefree(curnum);
98994639
HS
389 if (! (curnum =
390 setlocale(LC_NUMERIC,
391 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
ccd65d51 392 ? setlocale_init : NULL)))
98994639
HS
393 setlocale_failure = TRUE;
394 else
395 curnum = savepv(curnum);
b3e384bf 396# endif /* USE_LOCALE_NUMERIC */
98994639
HS
397 }
398
b3e384bf 399# endif /* LC_ALL */
98994639
HS
400
401#endif /* !LOCALE_ENVIRON_REQUIRED */
402
403#ifdef LC_ALL
ccd65d51 404 if (! setlocale(LC_ALL, setlocale_init))
98994639
HS
405 setlocale_failure = TRUE;
406#endif /* LC_ALL */
407
408 if (!setlocale_failure) {
409#ifdef USE_LOCALE_CTYPE
6cb43dbf 410 Safefree(curctype);
ccd65d51 411 if (! (curctype = setlocale(LC_CTYPE, setlocale_init)))
98994639
HS
412 setlocale_failure = TRUE;
413 else
414 curctype = savepv(curctype);
415#endif /* USE_LOCALE_CTYPE */
416#ifdef USE_LOCALE_COLLATE
6cb43dbf 417 Safefree(curcoll);
ccd65d51 418 if (! (curcoll = setlocale(LC_COLLATE, setlocale_init)))
98994639
HS
419 setlocale_failure = TRUE;
420 else
421 curcoll = savepv(curcoll);
422#endif /* USE_LOCALE_COLLATE */
423#ifdef USE_LOCALE_NUMERIC
6cb43dbf 424 Safefree(curnum);
ccd65d51 425 if (! (curnum = setlocale(LC_NUMERIC, setlocale_init)))
98994639
HS
426 setlocale_failure = TRUE;
427 else
428 curnum = savepv(curnum);
429#endif /* USE_LOCALE_NUMERIC */
430 }
431
432 if (setlocale_failure) {
433 char *p;
0bd48802 434 const bool locwarn = (printwarn > 1 ||
98994639
HS
435 (printwarn &&
436 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
437
438 if (locwarn) {
439#ifdef LC_ALL
440
441 PerlIO_printf(Perl_error_log,
442 "perl: warning: Setting locale failed.\n");
443
444#else /* !LC_ALL */
445
446 PerlIO_printf(Perl_error_log,
447 "perl: warning: Setting locale failed for the categories:\n\t");
448#ifdef USE_LOCALE_CTYPE
449 if (! curctype)
450 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
451#endif /* USE_LOCALE_CTYPE */
452#ifdef USE_LOCALE_COLLATE
453 if (! curcoll)
454 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
455#endif /* USE_LOCALE_COLLATE */
456#ifdef USE_LOCALE_NUMERIC
457 if (! curnum)
458 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
459#endif /* USE_LOCALE_NUMERIC */
460 PerlIO_printf(Perl_error_log, "\n");
461
462#endif /* LC_ALL */
463
464 PerlIO_printf(Perl_error_log,
465 "perl: warning: Please check that your locale settings:\n");
466
467#ifdef __GLIBC__
468 PerlIO_printf(Perl_error_log,
469 "\tLANGUAGE = %c%s%c,\n",
470 language ? '"' : '(',
471 language ? language : "unset",
472 language ? '"' : ')');
473#endif
474
475 PerlIO_printf(Perl_error_log,
476 "\tLC_ALL = %c%s%c,\n",
477 lc_all ? '"' : '(',
478 lc_all ? lc_all : "unset",
479 lc_all ? '"' : ')');
480
481#if defined(USE_ENVIRON_ARRAY)
482 {
483 char **e;
484 for (e = environ; *e; e++) {
485 if (strnEQ(*e, "LC_", 3)
486 && strnNE(*e, "LC_ALL=", 7)
487 && (p = strchr(*e, '=')))
488 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
489 (int)(p - *e), *e, p + 1);
490 }
491 }
492#else
493 PerlIO_printf(Perl_error_log,
494 "\t(possibly more locale environment variables)\n");
495#endif
496
497 PerlIO_printf(Perl_error_log,
498 "\tLANG = %c%s%c\n",
499 lang ? '"' : '(',
500 lang ? lang : "unset",
501 lang ? '"' : ')');
502
503 PerlIO_printf(Perl_error_log,
504 " are supported and installed on your system.\n");
505 }
506
507#ifdef LC_ALL
508
509 if (setlocale(LC_ALL, "C")) {
510 if (locwarn)
511 PerlIO_printf(Perl_error_log,
512 "perl: warning: Falling back to the standard locale (\"C\").\n");
513 ok = 0;
514 }
515 else {
516 if (locwarn)
517 PerlIO_printf(Perl_error_log,
518 "perl: warning: Failed to fall back to the standard locale (\"C\").\n");
519 ok = -1;
520 }
521
522#else /* ! LC_ALL */
523
524 if (0
525#ifdef USE_LOCALE_CTYPE
526 || !(curctype || setlocale(LC_CTYPE, "C"))
527#endif /* USE_LOCALE_CTYPE */
528#ifdef USE_LOCALE_COLLATE
529 || !(curcoll || setlocale(LC_COLLATE, "C"))
530#endif /* USE_LOCALE_COLLATE */
531#ifdef USE_LOCALE_NUMERIC
532 || !(curnum || setlocale(LC_NUMERIC, "C"))
533#endif /* USE_LOCALE_NUMERIC */
534 )
535 {
536 if (locwarn)
537 PerlIO_printf(Perl_error_log,
538 "perl: warning: Cannot fall back to the standard locale (\"C\").\n");
539 ok = -1;
540 }
541
542#endif /* ! LC_ALL */
543
544#ifdef USE_LOCALE_CTYPE
56279a21 545 Safefree(curctype);
bd61b366 546 curctype = savepv(setlocale(LC_CTYPE, NULL));
98994639
HS
547#endif /* USE_LOCALE_CTYPE */
548#ifdef USE_LOCALE_COLLATE
56279a21 549 Safefree(curcoll);
bd61b366 550 curcoll = savepv(setlocale(LC_COLLATE, NULL));
98994639
HS
551#endif /* USE_LOCALE_COLLATE */
552#ifdef USE_LOCALE_NUMERIC
56279a21 553 Safefree(curnum);
bd61b366 554 curnum = savepv(setlocale(LC_NUMERIC, NULL));
98994639
HS
555#endif /* USE_LOCALE_NUMERIC */
556 }
557 else {
558
559#ifdef USE_LOCALE_CTYPE
560 new_ctype(curctype);
561#endif /* USE_LOCALE_CTYPE */
562
563#ifdef USE_LOCALE_COLLATE
564 new_collate(curcoll);
565#endif /* USE_LOCALE_COLLATE */
566
567#ifdef USE_LOCALE_NUMERIC
568 new_numeric(curnum);
569#endif /* USE_LOCALE_NUMERIC */
b310b053 570
98994639
HS
571 }
572
8ef6e574 573#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
b310b053 574 {
fde18df1 575 /* Set PL_utf8locale to TRUE if using PerlIO _and_
7d74bb61 576 the current LC_CTYPE locale is UTF-8.
8aa8f774 577 If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
a05d7ebb 578 are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
fde18df1 579 on STDIN, STDOUT, STDERR, _and_ the default open discipline.
085a54d9 580 */
7d74bb61 581 PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE);
fde18df1 582 }
a05d7ebb 583 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
584 This is an alternative to using the -C command line switch
585 (the -C if present will override this). */
586 {
dd374669 587 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 588 PL_unicode = p ? parse_unicode_opts(&p) : 0;
5a22a2bb
NC
589 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
590 PL_utf8cache = -1;
b310b053 591 }
ec71e770 592#endif
b310b053 593
98994639 594#ifdef USE_LOCALE_CTYPE
43c5f42d 595 Safefree(curctype);
98994639
HS
596#endif /* USE_LOCALE_CTYPE */
597#ifdef USE_LOCALE_COLLATE
43c5f42d 598 Safefree(curcoll);
98994639
HS
599#endif /* USE_LOCALE_COLLATE */
600#ifdef USE_LOCALE_NUMERIC
43c5f42d 601 Safefree(curnum);
98994639 602#endif /* USE_LOCALE_NUMERIC */
8ef6e574
KW
603
604#endif /* USE_LOCALE */
605
98994639
HS
606 return ok;
607}
608
98994639
HS
609#ifdef USE_LOCALE_COLLATE
610
611/*
612 * mem_collxfrm() is a bit like strxfrm() but with two important
613 * differences. First, it handles embedded NULs. Second, it allocates
614 * a bit more memory than needed for the transformed data itself.
615 * The real transformed data begins at offset sizeof(collationix).
616 * Please see sv_collxfrm() to see how this is used.
617 */
166f8a29 618
98994639
HS
619char *
620Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
621{
97aff369 622 dVAR;
98994639
HS
623 char *xbuf;
624 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
625
7918f24d
NC
626 PERL_ARGS_ASSERT_MEM_COLLXFRM;
627
98994639
HS
628 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
629 /* the +1 is for the terminating NUL. */
630
631 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 632 Newx(xbuf, xAlloc, char);
98994639
HS
633 if (! xbuf)
634 goto bad;
635
636 *(U32*)xbuf = PL_collation_ix;
637 xout = sizeof(PL_collation_ix);
638 for (xin = 0; xin < len; ) {
224e8ef5 639 Size_t xused;
98994639
HS
640
641 for (;;) {
642 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
224e8ef5 643 if (xused >= PERL_INT_MAX)
98994639 644 goto bad;
eb160463 645 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
646 break;
647 xAlloc = (2 * xAlloc) + 1;
648 Renew(xbuf, xAlloc, char);
649 if (! xbuf)
650 goto bad;
651 }
652
653 xin += strlen(s + xin) + 1;
654 xout += xused;
655
656 /* Embedded NULs are understood but silently skipped
657 * because they make no sense in locale collation. */
658 }
659
660 xbuf[xout] = '\0';
661 *xlen = xout - sizeof(PL_collation_ix);
662 return xbuf;
663
664 bad:
665 Safefree(xbuf);
666 *xlen = 0;
667 return NULL;
668}
669
670#endif /* USE_LOCALE_COLLATE */
671
8ef6e574
KW
672#ifdef USE_LOCALE
673
637917bb 674STATIC bool
7d74bb61
KW
675S_is_cur_LC_category_utf8(pTHX_ int category)
676{
677 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
678 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
119ee68b
KW
679 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
680 * could give the wrong result. It errs on the side of not being a UTF-8
681 * locale. */
7d74bb61
KW
682
683 char *save_input_locale = NULL;
7d74bb61
KW
684 STRLEN final_pos;
685
8ef6e574 686#ifdef LC_ALL
7d74bb61 687 assert(category != LC_ALL);
8ef6e574 688#endif
7d74bb61
KW
689
690 /* First dispose of the trivial cases */
b07fffd1 691 save_input_locale = setlocale(category, NULL);
7d74bb61
KW
692 if (! save_input_locale) {
693 return FALSE; /* XXX maybe should croak */
694 }
b07fffd1 695 save_input_locale = stdize_locale(savepv(save_input_locale));
7d74bb61
KW
696 if ((*save_input_locale == 'C' && save_input_locale[1] == '\0')
697 || strEQ(save_input_locale, "POSIX"))
698 {
b07fffd1 699 Safefree(save_input_locale);
7d74bb61
KW
700 return FALSE;
701 }
702
119ee68b 703#if defined(USE_LOCALE_CTYPE) && (defined(MB_CUR_MAX) || defined(HAS_NL_LANGINFO) && defined(CODESET))
7d74bb61 704
119ee68b 705 { /* Next try MB_CUR_MAX or nl_langinfo if available */
7d74bb61
KW
706
707 char *save_ctype_locale = NULL;
119ee68b 708 bool is_utf8;
7d74bb61 709
119ee68b 710 if (category != LC_CTYPE) { /* These work only on LC_CTYPE */
7d74bb61
KW
711
712 /* Get the current LC_CTYPE locale */
713 save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL)));
714 if (! save_ctype_locale) {
715 goto cant_use_nllanginfo;
716 }
717
718 /* If LC_CTYPE and the desired category use the same locale, this
719 * means that finding the value for LC_CTYPE is the same as finding
720 * the value for the desired category. Otherwise, switch LC_CTYPE
721 * to the desired category's locale */
722 if (strEQ(save_ctype_locale, save_input_locale)) {
723 Safefree(save_ctype_locale);
724 save_ctype_locale = NULL;
725 }
726 else if (! setlocale(LC_CTYPE, save_input_locale)) {
727 Safefree(save_ctype_locale);
728 goto cant_use_nllanginfo;
729 }
730 }
731
732 /* Here the current LC_CTYPE is set to the locale of the category whose
119ee68b
KW
733 * information is desired. This means that MB_CUR_MAX and
734 * nl_langinfo() should give the correct results */
735
736# ifdef MB_CUR_MAX
737
738 /* If we switched LC_CTYPE, switch back */
739 if (save_ctype_locale) {
740 setlocale(LC_CTYPE, save_ctype_locale);
741 Safefree(save_ctype_locale);
742 }
743
744 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
745 * Unicode code point. Since UTF-8 is the only non-single byte
746 * encoding we handle, we just say any such encoding is UTF-8, and if
747 * turns out to be wrong, other things will fail */
748 is_utf8 = MB_CUR_MAX >= 4;
749
750 Safefree(save_input_locale);
751
752# ifdef HAS_MBTOWC
753
754 /* ... But, most system that have MB_CUR_MAX will also have mbtowc(),
755 * since they are both in the C99 standard. We can feed a known byte
756 * string to the latter function, and check that it gives the expected
757 * result */
758 if (is_utf8) {
759 wchar_t wc;
760 if (mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
761 != strlen(HYPHEN_UTF8)
762 || wc != (wchar_t) 0x2010)
763 {
764 is_utf8 = FALSE;
765 }
766 }
767
768# endif
769
770 return is_utf8;
771#else
772# if defined(HAS_NL_LANGINFO) && defined(CODESET)
773 {
648f6ebd
KW
774 char *codeset = savepv(nl_langinfo(CODESET));
775 if (codeset) {
7d74bb61 776
648f6ebd
KW
777 /* If we switched LC_CTYPE, switch back */
778 if (save_ctype_locale) {
779 setlocale(LC_CTYPE, save_ctype_locale);
780 Safefree(save_ctype_locale);
781 }
7d74bb61 782
648f6ebd
KW
783 is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8"))
784 || foldEQ(codeset, STR_WITH_LEN("UTF8"));
7d74bb61 785
648f6ebd
KW
786 Safefree(codeset);
787 Safefree(save_input_locale);
788 return is_utf8;
789 }
119ee68b 790 }
7d74bb61 791
119ee68b
KW
792# endif
793# endif
7d74bb61 794 }
119ee68b 795
7d74bb61
KW
796 cant_use_nllanginfo:
797
798#endif /* HAS_NL_LANGINFO etc */
799
800 /* nl_langinfo not available or failed somehow. Look at the locale name to
751e9426 801 * see if it matches qr/UTF -? 8 /ix */
7d74bb61
KW
802
803 final_pos = strlen(save_input_locale) - 1;
751e9426
KW
804 if (final_pos >= 3) {
805 char *name = save_input_locale;
806
807 /* Find next 'U' or 'u' and look from there */
808 while ((name += strcspn(name, "Uu") + 1)
809 <= save_input_locale + final_pos - 2)
7d74bb61 810 {
751e9426
KW
811 if (toFOLD(*(name)) != 't'
812 || toFOLD(*(name + 1)) != 'f')
813 {
814 continue;
815 }
816 name += 2;
817 if (*(name) == '-') {
818 if ((name > save_input_locale + final_pos - 1)) {
819 break;
820 }
821 name++;
822 }
823 if (*(name) == '8') {
824 Safefree(save_input_locale);
825 return TRUE;
826 }
7d74bb61
KW
827 }
828 }
829
830#ifdef WIN32
831 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
832 if (final_pos >= 4
833 && *(save_input_locale + final_pos - 0) == '1'
834 && *(save_input_locale + final_pos - 1) == '0'
835 && *(save_input_locale + final_pos - 2) == '0'
836 && *(save_input_locale + final_pos - 3) == '5'
837 && *(save_input_locale + final_pos - 4) == '6')
838 {
839 Safefree(save_input_locale);
840 return TRUE;
841 }
842#endif
843
fa9b773e
KW
844 /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
845 if (instr(save_input_locale, "8859")) {
846 Safefree(save_input_locale);
847 return FALSE;
848 }
849
850#ifdef HAS_LOCALECONV
851
852# ifdef USE_LOCALE_MONETARY
853
854 /* Here, there is nothing in the locale name to indicate whether the locale
855 * is UTF-8 or not. This "name", the return of setlocale(), is actually
856 * defined to be opaque, so we can't really rely on the absence of various
857 * substrings in the name to indicate its UTF-8ness. Look at the locale's
858 * currency symbol. Often that will be in the native script, and if the
859 * symbol isn't in UTF-8, we know that the locale isn't. If it is
860 * non-ASCII UTF-8, we infer that the locale is too.
861 * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the
862 * locale of the desired category, if it isn't that locale already */
863
864 {
865 char *save_monetary_locale = NULL;
866 bool illegal_utf8 = FALSE;
867 bool only_ascii = FALSE;
868 const struct lconv* const lc = localeconv();
869
870 if (category != LC_MONETARY) {
871
872 save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY,
873 NULL)));
874 if (! save_monetary_locale) {
875 goto cant_use_monetary;
876 }
877
878 if (strNE(save_monetary_locale, save_input_locale)) {
879 if (! setlocale(LC_MONETARY, save_input_locale)) {
880 Safefree(save_monetary_locale);
881 goto cant_use_monetary;
882 }
883 }
884 }
885
886 /* Here the current LC_MONETARY is set to the locale of the category
887 * whose information is desired. */
888
889 if (lc && lc->currency_symbol) {
890 if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) {
891 illegal_utf8 = TRUE;
892 }
893 else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) {
894 only_ascii = TRUE;
895 }
896 }
897
898 /* If we changed it, restore LC_MONETARY to its original locale */
899 if (save_monetary_locale) {
900 setlocale(LC_MONETARY, save_monetary_locale);
901 Safefree(save_monetary_locale);
902 }
903
904 Safefree(save_input_locale);
905
906 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise
907 * assume the locale is UTF-8 if and only if the symbol is non-ascii
908 * UTF-8. (We can't really tell if the locale is UTF-8 or not if the
909 * symbol is just a '$', so we err on the side of it not being UTF-8)
910 * */
911 return (illegal_utf8)
912 ? FALSE
913 : ! only_ascii;
914
915 }
916 cant_use_monetary:
917
918# endif /* USE_LOCALE_MONETARY */
919#endif /* HAS_LOCALECONV */
920
921#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES)
922
923/* This code is ifdefd out because it was found to not be necessary in testing
924 * on our dromedary test machine, which has over 700 locales. There, looking
925 * at just the currency symbol gave essentially the same results as doing this
926 * extra work. Executing this also caused segfaults in miniperl. I left it in
927 * so as to avoid rewriting it if real-world experience indicates that
928 * dromedary is an outlier. Essentially, instead of returning abpve if we
929 * haven't found illegal utf8, we continue on and examine all the strerror()
930 * messages on the platform for utf8ness. If all are ASCII, we still don't
931 * know the answer; but otherwise we have a pretty good indication of the
932 * utf8ness. The reason this doesn't necessarily help much is that the
933 * messages may not have been translated into the locale. The currency symbol
934 * is much more likely to have been translated. The code below would need to
935 * be altered somewhat to just be a continuation of testing the currency
936 * symbol. */
937 int e;
938 unsigned int failures = 0, non_ascii = 0;
939 char *save_messages_locale = NULL;
940
941 /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the
942 * desired category, if it isn't that locale already */
943
944 if (category != LC_MESSAGES) {
945
946 save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES,
947 NULL)));
948 if (! save_messages_locale) {
949 goto cant_use_messages;
950 }
951
952 if (strEQ(save_messages_locale, save_input_locale)) {
953 Safefree(save_input_locale);
954 }
955 else if (! setlocale(LC_MESSAGES, save_input_locale)) {
956 Safefree(save_messages_locale);
957 goto cant_use_messages;
958 }
959 }
960
961 /* Here the current LC_MESSAGES is set to the locale of the category
962 * whose information is desired. Look through all the messages */
963
964 for (e = 0;
965#ifdef HAS_SYS_ERRLIST
966 e <= sys_nerr
967#endif
968 ; e++)
969 {
970 const U8* const errmsg = (U8 *) Strerror(e) ;
971 if (!errmsg)
972 break;
973 if (! is_utf8_string(errmsg, 0)) {
974 failures++;
975 break;
976 }
977 else if (! is_ascii_string(errmsg, 0)) {
978 non_ascii++;
979 }
980 }
981
982 /* And, if we changed it, restore LC_MESSAGES to its original locale */
983 if (save_messages_locale) {
984 setlocale(LC_MESSAGES, save_messages_locale);
985 Safefree(save_messages_locale);
986 }
987
988 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
989 * any non-ascii means it is one; otherwise we assume it isn't */
990 return (failures) ? FALSE : non_ascii;
991
992 }
993 cant_use_messages:
994
995#endif
996
997 Safefree(save_input_locale);
7d74bb61
KW
998 return FALSE;
999}
1000
8ef6e574 1001#endif
7d74bb61 1002
66610fdd
RGS
1003/*
1004 * Local variables:
1005 * c-indentation-style: bsd
1006 * c-basic-offset: 4
14d04a33 1007 * indent-tabs-mode: nil
66610fdd
RGS
1008 * End:
1009 *
14d04a33 1010 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1011 */