This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
performance tweaking op.c
[perl5.git] / locale.c
CommitLineData
98994639
HS
1/* locale.c
2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999,
acde74e1 4 * 2000, 2001, 2002, 2003, 2005, 2006, 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/*
12 * A Elbereth Gilthoniel,
13 * silivren penna míriel
14 * o menel aglar elenath!
15 * Na-chaered palan-díriel
16 * o galadhremmin ennorath,
17 * Fanuilos, le linnathon
18 * nef aear, si nef aearon!
19 */
20
166f8a29
DM
21/* utility functions for handling locale-specific stuff like what
22 * character represents the decimal point.
23 */
24
98994639
HS
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
b310b053
JH
33#ifdef I_LANGINFO
34# include <langinfo.h>
35#endif
36
a4af207c
JH
37#include "reentr.h"
38
27da23d5 39#if defined(USE_LOCALE_NUMERIC) || defined(USE_LOCALE_COLLATE)
98994639
HS
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 */
53STATIC char *
54S_stdize_locale(pTHX_ char *locs)
55{
7452cf6a 56 const char * const s = strchr(locs, '=');
98994639
HS
57 bool okay = TRUE;
58
8772537c
AL
59 if (s) {
60 const char * const t = strchr(s, '.');
98994639 61 okay = FALSE;
8772537c
AL
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;
98994639
HS
69 }
70 }
71 }
72
73 if (!okay)
74 Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs);
75
76 return locs;
77}
27da23d5 78#endif
98994639
HS
79
80void
81Perl_set_numeric_radix(pTHX)
82{
83#ifdef USE_LOCALE_NUMERIC
84# ifdef HAS_LOCALECONV
7452cf6a 85 const struct lconv* const lc = localeconv();
98994639 86
98994639
HS
87 if (lc && lc->decimal_point) {
88 if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) {
89 SvREFCNT_dec(PL_numeric_radix_sv);
90 PL_numeric_radix_sv = Nullsv;
91 }
92 else {
93 if (PL_numeric_radix_sv)
94 sv_setpv(PL_numeric_radix_sv, lc->decimal_point);
95 else
96 PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0);
97 }
98 }
99 else
100 PL_numeric_radix_sv = Nullsv;
101# endif /* HAS_LOCALECONV */
102#endif /* USE_LOCALE_NUMERIC */
103}
104
105/*
106 * Set up for a new numeric locale.
107 */
108void
8772537c 109Perl_new_numeric(pTHX_ const char *newnum)
98994639
HS
110{
111#ifdef USE_LOCALE_NUMERIC
112
113 if (! newnum) {
43c5f42d
NC
114 Safefree(PL_numeric_name);
115 PL_numeric_name = NULL;
98994639
HS
116 PL_numeric_standard = TRUE;
117 PL_numeric_local = TRUE;
118 return;
119 }
120
121 if (! PL_numeric_name || strNE(PL_numeric_name, newnum)) {
122 Safefree(PL_numeric_name);
123 PL_numeric_name = stdize_locale(savepv(newnum));
770526c1
NC
124 PL_numeric_standard = ((*newnum == 'C' && newnum[1] == '\0')
125 || strEQ(newnum, "POSIX"));
98994639
HS
126 PL_numeric_local = TRUE;
127 set_numeric_radix();
128 }
129
130#endif /* USE_LOCALE_NUMERIC */
131}
132
133void
134Perl_set_numeric_standard(pTHX)
135{
136#ifdef USE_LOCALE_NUMERIC
137
138 if (! PL_numeric_standard) {
139 setlocale(LC_NUMERIC, "C");
140 PL_numeric_standard = TRUE;
141 PL_numeric_local = FALSE;
142 set_numeric_radix();
143 }
144
145#endif /* USE_LOCALE_NUMERIC */
146}
147
148void
149Perl_set_numeric_local(pTHX)
150{
151#ifdef USE_LOCALE_NUMERIC
152
153 if (! PL_numeric_local) {
154 setlocale(LC_NUMERIC, PL_numeric_name);
155 PL_numeric_standard = FALSE;
156 PL_numeric_local = TRUE;
157 set_numeric_radix();
158 }
159
160#endif /* USE_LOCALE_NUMERIC */
161}
162
163/*
164 * Set up for a new ctype locale.
165 */
166void
8772537c 167Perl_new_ctype(pTHX_ const char *newctype)
98994639
HS
168{
169#ifdef USE_LOCALE_CTYPE
27da23d5 170 dVAR;
98994639
HS
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 */
8772537c 183 PERL_UNUSED_ARG(newctype);
98994639
HS
184}
185
186/*
187 * Set up for a new collation locale.
188 */
189void
8772537c 190Perl_new_collate(pTHX_ const char *newcoll)
98994639
HS
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));
770526c1
NC
210 PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0')
211 || strEQ(newcoll, "POSIX"));
98994639
HS
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];
8772537c
AL
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;
98994639
HS
221 if (mult < 1)
222 Perl_croak(aTHX_ "strxfrm() gets absurd");
eb160463 223 PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0;
98994639
HS
224 PL_collxfrm_mult = mult;
225 }
226 }
227
228#endif /* USE_LOCALE_COLLATE */
229}
230
231/*
232 * Initialize locale awareness.
233 */
234int
235Perl_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__
7452cf6a 256 char * const language = PerlEnv_getenv("LANGUAGE");
98994639 257#endif
7452cf6a
AL
258 char * const lc_all = PerlEnv_getenv("LC_ALL");
259 char * const lang = PerlEnv_getenv("LANG");
98994639
HS
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;
0bd48802 340 const bool locwarn = (printwarn > 1 ||
98994639
HS
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 */
b310b053 473
98994639
HS
474 }
475
476#endif /* USE_LOCALE */
477
ec71e770 478#ifdef USE_PERLIO
b310b053 479 {
fde18df1 480 /* Set PL_utf8locale to TRUE if using PerlIO _and_
ec71e770 481 any of the following are true:
085a54d9 482 - nl_langinfo(CODESET) contains /^utf-?8/i
61de9fb5 483 - $ENV{LC_ALL} contains /^utf-?8/i
085a54d9 484 - $ENV{LC_CTYPE} contains /^utf-?8/i
61de9fb5
JH
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.)
8aa8f774 492 If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
a05d7ebb 493 are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
fde18df1 494 on STDIN, STDOUT, STDERR, _and_ the default open discipline.
085a54d9 495 */
fde18df1 496 bool utf8locale = FALSE;
b310b053
JH
497 char *codeset = NULL;
498#if defined(HAS_NL_LANGINFO) && defined(CODESET)
499 codeset = nl_langinfo(CODESET);
500#endif
61de9fb5 501 if (codeset)
89529cee
GA
502 utf8locale = (Perl_ibcmp(aTHX_ codeset, STR_WITH_LEN("UTF-8")) == 0 ||
503 Perl_ibcmp(aTHX_ codeset, STR_WITH_LEN("UTF8") ) == 0);
077a72a9 504#if defined(USE_LOCALE)
61de9fb5
JH
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)
89529cee
GA
509 utf8locale = (Perl_ibcmp(aTHX_ lang, STR_WITH_LEN("UTF-8")) == 0 ||
510 Perl_ibcmp(aTHX_ lang, STR_WITH_LEN("UTF8") ) == 0);
b310b053 511#ifdef USE_LOCALE_CTYPE
61de9fb5 512 if (curctype)
89529cee
GA
513 utf8locale = (Perl_ibcmp(aTHX_ curctype, STR_WITH_LEN("UTF-8")) == 0 ||
514 Perl_ibcmp(aTHX_ curctype, STR_WITH_LEN("UTF8") ) == 0);
b310b053 515#endif
61de9fb5 516 if (lc_all)
89529cee
GA
517 utf8locale = (Perl_ibcmp(aTHX_ lc_all, STR_WITH_LEN("UTF-8")) == 0 ||
518 Perl_ibcmp(aTHX_ lc_all, STR_WITH_LEN("UTF8") ) == 0);
61de9fb5 519 }
fde18df1
JH
520#endif /* USE_LOCALE */
521 if (utf8locale)
522 PL_utf8locale = TRUE;
523 }
a05d7ebb 524 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
fde18df1
JH
525 This is an alternative to using the -C command line switch
526 (the -C if present will override this). */
527 {
dd374669 528 const char *p = PerlEnv_getenv("PERL_UNICODE");
a05d7ebb 529 PL_unicode = p ? parse_unicode_opts(&p) : 0;
b310b053 530 }
ec71e770 531#endif
b310b053 532
98994639 533#ifdef USE_LOCALE_CTYPE
43c5f42d 534 Safefree(curctype);
98994639
HS
535#endif /* USE_LOCALE_CTYPE */
536#ifdef USE_LOCALE_COLLATE
43c5f42d 537 Safefree(curcoll);
98994639
HS
538#endif /* USE_LOCALE_COLLATE */
539#ifdef USE_LOCALE_NUMERIC
43c5f42d 540 Safefree(curnum);
98994639
HS
541#endif /* USE_LOCALE_NUMERIC */
542 return ok;
543}
544
98994639
HS
545#ifdef USE_LOCALE_COLLATE
546
547/*
548 * mem_collxfrm() is a bit like strxfrm() but with two important
549 * differences. First, it handles embedded NULs. Second, it allocates
550 * a bit more memory than needed for the transformed data itself.
551 * The real transformed data begins at offset sizeof(collationix).
552 * Please see sv_collxfrm() to see how this is used.
553 */
166f8a29 554
98994639
HS
555char *
556Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
557{
558 char *xbuf;
559 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
560
561 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
562 /* the +1 is for the terminating NUL. */
563
564 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
a02a5408 565 Newx(xbuf, xAlloc, char);
98994639
HS
566 if (! xbuf)
567 goto bad;
568
569 *(U32*)xbuf = PL_collation_ix;
570 xout = sizeof(PL_collation_ix);
571 for (xin = 0; xin < len; ) {
572 SSize_t xused;
573
574 for (;;) {
575 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
576 if (xused == -1)
577 goto bad;
eb160463 578 if ((STRLEN)xused < xAlloc - xout)
98994639
HS
579 break;
580 xAlloc = (2 * xAlloc) + 1;
581 Renew(xbuf, xAlloc, char);
582 if (! xbuf)
583 goto bad;
584 }
585
586 xin += strlen(s + xin) + 1;
587 xout += xused;
588
589 /* Embedded NULs are understood but silently skipped
590 * because they make no sense in locale collation. */
591 }
592
593 xbuf[xout] = '\0';
594 *xlen = xout - sizeof(PL_collation_ix);
595 return xbuf;
596
597 bad:
598 Safefree(xbuf);
599 *xlen = 0;
600 return NULL;
601}
602
603#endif /* USE_LOCALE_COLLATE */
604
66610fdd
RGS
605/*
606 * Local variables:
607 * c-indentation-style: bsd
608 * c-basic-offset: 4
609 * indent-tabs-mode: t
610 * End:
611 *
37442d52
RGS
612 * ex: set ts=8 sts=4 sw=4 noet:
613 */