1 /* locale.c
2 *
3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2005, by Larry Wall and others
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 /*
12 * A Elbereth Gilthoniel,
13 * silivren penna m�riel
14 * o menel aglar elenath!
15 * Na-chaered palan-d�riel
16 * o galadhremmin ennorath,
17 * Fanuilos, le linnathon
18 * nef aear, si nef aearon!
19 */
20
21 /* utility functions for handling locale-specific stuff like what
22 * character represents the decimal point.
23 */
24
25 #include "EXTERN.h"
26 #define PERL_IN_LOCALE_C
27 #include "perl.h"
28
29 #ifdef I_LOCALE
30 # include <locale.h>
31 #endif
32
33 #ifdef I_LANGINFO
34 # include <langinfo.h>
35 #endif
36
37 #include "reentr.h"
38
39 #if defined(USE_LOCALE_NUMERIC) || defined(USE_LOCALE_COLLATE)
40 /*
41 * Standardize the locale name from a string returned by 'setlocale'.
42 *
43 * The standard return value of setlocale() is either
44 * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL
45 * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL
46 * (the space-separated values represent the various sublocales,
47 * in some unspecificed order)
48 *
49 * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n",
50 * which is harmful for further use of the string in setlocale().
51 *
52 */
53 STATIC char *
S_stdize_locale(pTHX_ char * locs)54 S_stdize_locale(pTHX_ char *locs)
55 {
56 const char *s = strchr(locs, '=');
57 bool okay = TRUE;
58
59 if (s) {
60 const char * const t = strchr(s, '.');
61 okay = FALSE;
62 if (t) {
63 const char * const u = strchr(t, '\n');
64 if (u && (u[1] == 0)) {
65 const STRLEN len = u - s;
66 Move(s + 1, locs, len, char);
67 locs[len] = 0;
68 okay = TRUE;
69 }
70 }
71 }
72
73 if (!okay)
74 Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs);
75
76 return locs;
77 }
78 #endif
79
80 void
Perl_set_numeric_radix(pTHX)81 Perl_set_numeric_radix(pTHX)
82 {
83 #ifdef USE_LOCALE_NUMERIC
84 # ifdef HAS_LOCALECONV
85 struct lconv* lc;
86
87 lc = localeconv();
88 if (lc && lc->decimal_point) {
89 if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) {
90 SvREFCNT_dec(PL_numeric_radix_sv);
91 PL_numeric_radix_sv = Nullsv;
92 }
93 else {
94 if (PL_numeric_radix_sv)
95 sv_setpv(PL_numeric_radix_sv, lc->decimal_point);
96 else
97 PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0);
98 }
99 }
100 else
101 PL_numeric_radix_sv = Nullsv;
102 # endif /* HAS_LOCALECONV */
103 #endif /* USE_LOCALE_NUMERIC */
104 }
105
106 /*
107 * Set up for a new numeric locale.
108 */
109 void
Perl_new_numeric(pTHX_ char * newnum)110 Perl_new_numeric(pTHX_ char *newnum)
111 {
112 #ifdef USE_LOCALE_NUMERIC
113
114 if (! newnum) {
115 Safefree(PL_numeric_name);
116 PL_numeric_name = NULL;
117 PL_numeric_standard = TRUE;
118 PL_numeric_local = TRUE;
119 return;
120 }
121
122 if (! PL_numeric_name || strNE(PL_numeric_name, newnum)) {
123 Safefree(PL_numeric_name);
124 PL_numeric_name = stdize_locale(savepv(newnum));
125 PL_numeric_standard = ((*newnum == 'C' && newnum[1] == '\0')
126 || strEQ(newnum, "POSIX"));
127 PL_numeric_local = TRUE;
128 set_numeric_radix();
129 }
130
131 #endif /* USE_LOCALE_NUMERIC */
132 }
133
134 void
Perl_set_numeric_standard(pTHX)135 Perl_set_numeric_standard(pTHX)
136 {
137 #ifdef USE_LOCALE_NUMERIC
138
139 if (! PL_numeric_standard) {
140 setlocale(LC_NUMERIC, "C");
141 PL_numeric_standard = TRUE;
142 PL_numeric_local = FALSE;
143 set_numeric_radix();
144 }
145
146 #endif /* USE_LOCALE_NUMERIC */
147 }
148
149 void
Perl_set_numeric_local(pTHX)150 Perl_set_numeric_local(pTHX)
151 {
152 #ifdef USE_LOCALE_NUMERIC
153
154 if (! PL_numeric_local) {
155 setlocale(LC_NUMERIC, PL_numeric_name);
156 PL_numeric_standard = FALSE;
157 PL_numeric_local = TRUE;
158 set_numeric_radix();
159 }
160
161 #endif /* USE_LOCALE_NUMERIC */
162 }
163
164 /*
165 * Set up for a new ctype locale.
166 */
167 void
Perl_new_ctype(pTHX_ char * newctype)168 Perl_new_ctype(pTHX_ char *newctype)
169 {
170 #ifdef USE_LOCALE_CTYPE
171 int i;
172
173 for (i = 0; i < 256; i++) {
174 if (isUPPER_LC(i))
175 PL_fold_locale[i] = toLOWER_LC(i);
176 else if (isLOWER_LC(i))
177 PL_fold_locale[i] = toUPPER_LC(i);
178 else
179 PL_fold_locale[i] = i;
180 }
181
182 #endif /* USE_LOCALE_CTYPE */
183 PERL_UNUSED_ARG(newctype);
184 }
185
186 /*
187 * Set up for a new collation locale.
188 */
189 void
Perl_new_collate(pTHX_ char * newcoll)190 Perl_new_collate(pTHX_ char *newcoll)
191 {
192 #ifdef USE_LOCALE_COLLATE
193
194 if (! newcoll) {
195 if (PL_collation_name) {
196 ++PL_collation_ix;
197 Safefree(PL_collation_name);
198 PL_collation_name = NULL;
199 }
200 PL_collation_standard = TRUE;
201 PL_collxfrm_base = 0;
202 PL_collxfrm_mult = 2;
203 return;
204 }
205
206 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
207 ++PL_collation_ix;
208 Safefree(PL_collation_name);
209 PL_collation_name = stdize_locale(savepv(newcoll));
210 PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0')
211 || strEQ(newcoll, "POSIX"));
212
213 {
214 /* 2: at most so many chars ('a', 'b'). */
215 /* 50: surely no system expands a char more. */
216 #define XFRMBUFSIZE (2 * 50)
217 char xbuf[XFRMBUFSIZE];
218 const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE);
219 const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE);
220 const SSize_t mult = fb - fa;
221 if (mult < 1)
222 Perl_croak(aTHX_ "strxfrm() gets absurd");
223 PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0;
224 PL_collxfrm_mult = mult;
225 }
226 }
227
228 #endif /* USE_LOCALE_COLLATE */
229 }
230
231 /*
232 * Initialize locale awareness.
233 */
234 int
Perl_init_i18nl10n(pTHX_ int printwarn)235 Perl_init_i18nl10n(pTHX_ int printwarn)
236 {
237 int ok = 1;
238 /* returns
239 * 1 = set ok or not applicable,
240 * 0 = fallback to C locale,
241 * -1 = fallback to C locale failed
242 */
243
244 #if defined(USE_LOCALE)
245
246 #ifdef USE_LOCALE_CTYPE
247 char *curctype = NULL;
248 #endif /* USE_LOCALE_CTYPE */
249 #ifdef USE_LOCALE_COLLATE
250 char *curcoll = NULL;
251 #endif /* USE_LOCALE_COLLATE */
252 #ifdef USE_LOCALE_NUMERIC
253 char *curnum = NULL;
254 #endif /* USE_LOCALE_NUMERIC */
255 #ifdef __GLIBC__
256 char *language = PerlEnv_getenv("LANGUAGE");
257 #endif
258 char *lc_all = PerlEnv_getenv("LC_ALL");
259 char *lang = PerlEnv_getenv("LANG");
260 bool setlocale_failure = FALSE;
261
262 #ifdef LOCALE_ENVIRON_REQUIRED
263
264 /*
265 * Ultrix setlocale(..., "") fails if there are no environment
266 * variables from which to get a locale name.
267 */
268
269 bool done = FALSE;
270
271 #ifdef LC_ALL
272 if (lang) {
273 if (setlocale(LC_ALL, ""))
274 done = TRUE;
275 else
276 setlocale_failure = TRUE;
277 }
278 if (!setlocale_failure) {
279 #ifdef USE_LOCALE_CTYPE
280 if (! (curctype =
281 setlocale(LC_CTYPE,
282 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
283 ? "" : Nullch)))
284 setlocale_failure = TRUE;
285 else
286 curctype = savepv(curctype);
287 #endif /* USE_LOCALE_CTYPE */
288 #ifdef USE_LOCALE_COLLATE
289 if (! (curcoll =
290 setlocale(LC_COLLATE,
291 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
292 ? "" : Nullch)))
293 setlocale_failure = TRUE;
294 else
295 curcoll = savepv(curcoll);
296 #endif /* USE_LOCALE_COLLATE */
297 #ifdef USE_LOCALE_NUMERIC
298 if (! (curnum =
299 setlocale(LC_NUMERIC,
300 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
301 ? "" : Nullch)))
302 setlocale_failure = TRUE;
303 else
304 curnum = savepv(curnum);
305 #endif /* USE_LOCALE_NUMERIC */
306 }
307
308 #endif /* LC_ALL */
309
310 #endif /* !LOCALE_ENVIRON_REQUIRED */
311
312 #ifdef LC_ALL
313 if (! setlocale(LC_ALL, ""))
314 setlocale_failure = TRUE;
315 #endif /* LC_ALL */
316
317 if (!setlocale_failure) {
318 #ifdef USE_LOCALE_CTYPE
319 if (! (curctype = setlocale(LC_CTYPE, "")))
320 setlocale_failure = TRUE;
321 else
322 curctype = savepv(curctype);
323 #endif /* USE_LOCALE_CTYPE */
324 #ifdef USE_LOCALE_COLLATE
325 if (! (curcoll = setlocale(LC_COLLATE, "")))
326 setlocale_failure = TRUE;
327 else
328 curcoll = savepv(curcoll);
329 #endif /* USE_LOCALE_COLLATE */
330 #ifdef USE_LOCALE_NUMERIC
331 if (! (curnum = setlocale(LC_NUMERIC, "")))
332 setlocale_failure = TRUE;
333 else
334 curnum = savepv(curnum);
335 #endif /* USE_LOCALE_NUMERIC */
336 }
337
338 if (setlocale_failure) {
339 char *p;
340 bool locwarn = (printwarn > 1 ||
341 (printwarn &&
342 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
343
344 if (locwarn) {
345 #ifdef LC_ALL
346
347 PerlIO_printf(Perl_error_log,
348 "perl: warning: Setting locale failed.\n");
349
350 #else /* !LC_ALL */
351
352 PerlIO_printf(Perl_error_log,
353 "perl: warning: Setting locale failed for the categories:\n\t");
354 #ifdef USE_LOCALE_CTYPE
355 if (! curctype)
356 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
357 #endif /* USE_LOCALE_CTYPE */
358 #ifdef USE_LOCALE_COLLATE
359 if (! curcoll)
360 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
361 #endif /* USE_LOCALE_COLLATE */
362 #ifdef USE_LOCALE_NUMERIC
363 if (! curnum)
364 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
365 #endif /* USE_LOCALE_NUMERIC */
366 PerlIO_printf(Perl_error_log, "\n");
367
368 #endif /* LC_ALL */
369
370 PerlIO_printf(Perl_error_log,
371 "perl: warning: Please check that your locale settings:\n");
372
373 #ifdef __GLIBC__
374 PerlIO_printf(Perl_error_log,
375 "\tLANGUAGE = %c%s%c,\n",
376 language ? '"' : '(',
377 language ? language : "unset",
378 language ? '"' : ')');
379 #endif
380
381 PerlIO_printf(Perl_error_log,
382 "\tLC_ALL = %c%s%c,\n",
383 lc_all ? '"' : '(',
384 lc_all ? lc_all : "unset",
385 lc_all ? '"' : ')');
386
387 #if defined(USE_ENVIRON_ARRAY)
388 {
389 char **e;
390 for (e = environ; *e; e++) {
391 if (strnEQ(*e, "LC_", 3)
392 && strnNE(*e, "LC_ALL=", 7)
393 && (p = strchr(*e, '=')))
394 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
395 (int)(p - *e), *e, p + 1);
396 }
397 }
398 #else
399 PerlIO_printf(Perl_error_log,
400 "\t(possibly more locale environment variables)\n");
401 #endif
402
403 PerlIO_printf(Perl_error_log,
404 "\tLANG = %c%s%c\n",
405 lang ? '"' : '(',
406 lang ? lang : "unset",
407 lang ? '"' : ')');
408
409 PerlIO_printf(Perl_error_log,
410 " are supported and installed on your system.\n");
411 }
412
413 #ifdef LC_ALL
414
415 if (setlocale(LC_ALL, "C")) {
416 if (locwarn)
417 PerlIO_printf(Perl_error_log,
418 "perl: warning: Falling back to the standard locale (\"C\").\n");
419 ok = 0;
420 }
421 else {
422 if (locwarn)
423 PerlIO_printf(Perl_error_log,
424 "perl: warning: Failed to fall back to the standard locale (\"C\").\n");
425 ok = -1;
426 }
427
428 #else /* ! LC_ALL */
429
430 if (0
431 #ifdef USE_LOCALE_CTYPE
432 || !(curctype || setlocale(LC_CTYPE, "C"))
433 #endif /* USE_LOCALE_CTYPE */
434 #ifdef USE_LOCALE_COLLATE
435 || !(curcoll || setlocale(LC_COLLATE, "C"))
436 #endif /* USE_LOCALE_COLLATE */
437 #ifdef USE_LOCALE_NUMERIC
438 || !(curnum || setlocale(LC_NUMERIC, "C"))
439 #endif /* USE_LOCALE_NUMERIC */
440 )
441 {
442 if (locwarn)
443 PerlIO_printf(Perl_error_log,
444 "perl: warning: Cannot fall back to the standard locale (\"C\").\n");
445 ok = -1;
446 }
447
448 #endif /* ! LC_ALL */
449
450 #ifdef USE_LOCALE_CTYPE
451 curctype = savepv(setlocale(LC_CTYPE, Nullch));
452 #endif /* USE_LOCALE_CTYPE */
453 #ifdef USE_LOCALE_COLLATE
454 curcoll = savepv(setlocale(LC_COLLATE, Nullch));
455 #endif /* USE_LOCALE_COLLATE */
456 #ifdef USE_LOCALE_NUMERIC
457 curnum = savepv(setlocale(LC_NUMERIC, Nullch));
458 #endif /* USE_LOCALE_NUMERIC */
459 }
460 else {
461
462 #ifdef USE_LOCALE_CTYPE
463 new_ctype(curctype);
464 #endif /* USE_LOCALE_CTYPE */
465
466 #ifdef USE_LOCALE_COLLATE
467 new_collate(curcoll);
468 #endif /* USE_LOCALE_COLLATE */
469
470 #ifdef USE_LOCALE_NUMERIC
471 new_numeric(curnum);
472 #endif /* USE_LOCALE_NUMERIC */
473
474 }
475
476 #endif /* USE_LOCALE */
477
478 #ifdef USE_PERLIO
479 {
480 /* Set PL_utf8locale to TRUE if using PerlIO _and_
481 any of the following are true:
482 - nl_langinfo(CODESET) contains /^utf-?8/i
483 - $ENV{LC_ALL} contains /^utf-?8/i
484 - $ENV{LC_CTYPE} contains /^utf-?8/i
485 - $ENV{LANG} contains /^utf-?8/i
486 The LC_ALL, LC_CTYPE, LANG obey the usual override
487 hierarchy of locale environment variables. (LANGUAGE
488 affects only LC_MESSAGES only under glibc.) (If present,
489 it overrides LC_MESSAGES for GNU gettext, and it also
490 can have more than one locale, separated by spaces,
491 in case you need to know.)
492 If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
493 are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
494 on STDIN, STDOUT, STDERR, _and_ the default open discipline.
495 */
496 bool utf8locale = FALSE;
497 char *codeset = NULL;
498 #if defined(HAS_NL_LANGINFO) && defined(CODESET)
499 codeset = nl_langinfo(CODESET);
500 #endif
501 if (codeset)
502 utf8locale = (ibcmp(codeset, "UTF-8", 5) == 0 ||
503 ibcmp(codeset, "UTF8", 4) == 0);
504 #if defined(USE_LOCALE)
505 else { /* nl_langinfo(CODESET) is supposed to correctly
506 * interpret the locale environment variables,
507 * but just in case it fails, let's do this manually. */
508 if (lang)
509 utf8locale = (ibcmp(lang, "UTF-8", 5) == 0 ||
510 ibcmp(lang, "UTF8", 4) == 0);
511 #ifdef USE_LOCALE_CTYPE
512 if (curctype)
513 utf8locale = (ibcmp(curctype, "UTF-8", 5) == 0 ||
514 ibcmp(curctype, "UTF8", 4) == 0);
515 #endif
516 if (lc_all)
517 utf8locale = (ibcmp(lc_all, "UTF-8", 5) == 0 ||
518 ibcmp(lc_all, "UTF8", 4) == 0);
519 }
520 #endif /* USE_LOCALE */
521 if (utf8locale)
522 PL_utf8locale = TRUE;
523 }
524 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
525 This is an alternative to using the -C command line switch
526 (the -C if present will override this). */
527 {
528 char *p = PerlEnv_getenv("PERL_UNICODE");
529 PL_unicode = p ? parse_unicode_opts(&p) : 0;
530 }
531 #endif
532
533 #ifdef USE_LOCALE_CTYPE
534 Safefree(curctype);
535 #endif /* USE_LOCALE_CTYPE */
536 #ifdef USE_LOCALE_COLLATE
537 Safefree(curcoll);
538 #endif /* USE_LOCALE_COLLATE */
539 #ifdef USE_LOCALE_NUMERIC
540 Safefree(curnum);
541 #endif /* USE_LOCALE_NUMERIC */
542 return ok;
543 }
544
545 /* Backwards compatibility. */
546 int
Perl_init_i18nl14n(pTHX_ int printwarn)547 Perl_init_i18nl14n(pTHX_ int printwarn)
548 {
549 return init_i18nl10n(printwarn);
550 }
551
552 #ifdef USE_LOCALE_COLLATE
553
554 /*
555 * mem_collxfrm() is a bit like strxfrm() but with two important
556 * differences. First, it handles embedded NULs. Second, it allocates
557 * a bit more memory than needed for the transformed data itself.
558 * The real transformed data begins at offset sizeof(collationix).
559 * Please see sv_collxfrm() to see how this is used.
560 */
561
562 char *
Perl_mem_collxfrm(pTHX_ const char * s,STRLEN len,STRLEN * xlen)563 Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
564 {
565 char *xbuf;
566 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
567
568 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
569 /* the +1 is for the terminating NUL. */
570
571 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
572 Newx(xbuf, xAlloc, char);
573 if (! xbuf)
574 goto bad;
575
576 *(U32*)xbuf = PL_collation_ix;
577 xout = sizeof(PL_collation_ix);
578 for (xin = 0; xin < len; ) {
579 SSize_t xused;
580
581 for (;;) {
582 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
583 if (xused == -1)
584 goto bad;
585 if ((STRLEN)xused < xAlloc - xout)
586 break;
587 xAlloc = (2 * xAlloc) + 1;
588 Renew(xbuf, xAlloc, char);
589 if (! xbuf)
590 goto bad;
591 }
592
593 xin += strlen(s + xin) + 1;
594 xout += xused;
595
596 /* Embedded NULs are understood but silently skipped
597 * because they make no sense in locale collation. */
598 }
599
600 xbuf[xout] = '\0';
601 *xlen = xout - sizeof(PL_collation_ix);
602 return xbuf;
603
604 bad:
605 Safefree(xbuf);
606 *xlen = 0;
607 return NULL;
608 }
609
610 #endif /* USE_LOCALE_COLLATE */
611
612 /*
613 * Local variables:
614 * c-indentation-style: bsd
615 * c-basic-offset: 4
616 * indent-tabs-mode: t
617 * End:
618 *
619 * ex: set ts=8 sts=4 sw=4 noet:
620 */
621