3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2005, by Larry Wall and others
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.
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!
21 /* utility functions for handling locale-specific stuff like what
22 * character represents the decimal point.
26 #define PERL_IN_LOCALE_C
34 # include <langinfo.h>
40 * Standardize the locale name from a string returned by 'setlocale'.
42 * The standard return value of setlocale() is either
43 * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL
44 * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL
45 * (the space-separated values represent the various sublocales,
46 * in some unspecificed order)
48 * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n",
49 * which is harmful for further use of the string in setlocale().
53 S_stdize_locale(pTHX_ char *locs)
58 if ((s = strchr(locs, '='))) {
62 if ((t = strchr(s, '.'))) {
65 if ((u = strchr(t, '\n'))) {
69 Move(s + 1, locs, len, char);
78 Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs);
84 Perl_set_numeric_radix(pTHX)
86 #ifdef USE_LOCALE_NUMERIC
87 # ifdef HAS_LOCALECONV
91 if (lc && lc->decimal_point) {
92 if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) {
93 SvREFCNT_dec(PL_numeric_radix_sv);
94 PL_numeric_radix_sv = Nullsv;
97 if (PL_numeric_radix_sv)
98 sv_setpv(PL_numeric_radix_sv, lc->decimal_point);
100 PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0);
104 PL_numeric_radix_sv = Nullsv;
105 # endif /* HAS_LOCALECONV */
106 #endif /* USE_LOCALE_NUMERIC */
110 * Set up for a new numeric locale.
113 Perl_new_numeric(pTHX_ char *newnum)
115 #ifdef USE_LOCALE_NUMERIC
118 if (PL_numeric_name) {
119 Safefree(PL_numeric_name);
120 PL_numeric_name = NULL;
122 PL_numeric_standard = TRUE;
123 PL_numeric_local = TRUE;
127 if (! PL_numeric_name || strNE(PL_numeric_name, newnum)) {
128 Safefree(PL_numeric_name);
129 PL_numeric_name = stdize_locale(savepv(newnum));
130 PL_numeric_standard = ((*newnum == 'C' && newnum[1] == '\0')
131 || strEQ(newnum, "POSIX"));
132 PL_numeric_local = TRUE;
136 #endif /* USE_LOCALE_NUMERIC */
140 Perl_set_numeric_standard(pTHX)
142 #ifdef USE_LOCALE_NUMERIC
144 if (! PL_numeric_standard) {
145 setlocale(LC_NUMERIC, "C");
146 PL_numeric_standard = TRUE;
147 PL_numeric_local = FALSE;
151 #endif /* USE_LOCALE_NUMERIC */
155 Perl_set_numeric_local(pTHX)
157 #ifdef USE_LOCALE_NUMERIC
159 if (! PL_numeric_local) {
160 setlocale(LC_NUMERIC, PL_numeric_name);
161 PL_numeric_standard = FALSE;
162 PL_numeric_local = TRUE;
166 #endif /* USE_LOCALE_NUMERIC */
170 * Set up for a new ctype locale.
173 Perl_new_ctype(pTHX_ char *newctype)
175 #ifdef USE_LOCALE_CTYPE
179 for (i = 0; i < 256; i++) {
181 PL_fold_locale[i] = toLOWER_LC(i);
182 else if (isLOWER_LC(i))
183 PL_fold_locale[i] = toUPPER_LC(i);
185 PL_fold_locale[i] = i;
188 #endif /* USE_LOCALE_CTYPE */
192 * Set up for a new collation locale.
195 Perl_new_collate(pTHX_ char *newcoll)
197 #ifdef USE_LOCALE_COLLATE
200 if (PL_collation_name) {
202 Safefree(PL_collation_name);
203 PL_collation_name = NULL;
205 PL_collation_standard = TRUE;
206 PL_collxfrm_base = 0;
207 PL_collxfrm_mult = 2;
211 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
213 Safefree(PL_collation_name);
214 PL_collation_name = stdize_locale(savepv(newcoll));
215 PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0')
216 || strEQ(newcoll, "POSIX"));
219 /* 2: at most so many chars ('a', 'b'). */
220 /* 50: surely no system expands a char more. */
221 #define XFRMBUFSIZE (2 * 50)
222 char xbuf[XFRMBUFSIZE];
223 Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE);
224 Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE);
225 SSize_t mult = fb - fa;
227 Perl_croak(aTHX_ "strxfrm() gets absurd");
228 PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0;
229 PL_collxfrm_mult = mult;
233 #endif /* USE_LOCALE_COLLATE */
237 * Initialize locale awareness.
240 Perl_init_i18nl10n(pTHX_ int printwarn)
244 * 1 = set ok or not applicable,
245 * 0 = fallback to C locale,
246 * -1 = fallback to C locale failed
249 #if defined(USE_LOCALE)
251 #ifdef USE_LOCALE_CTYPE
252 char *curctype = NULL;
253 #endif /* USE_LOCALE_CTYPE */
254 #ifdef USE_LOCALE_COLLATE
255 char *curcoll = NULL;
256 #endif /* USE_LOCALE_COLLATE */
257 #ifdef USE_LOCALE_NUMERIC
259 #endif /* USE_LOCALE_NUMERIC */
261 char *language = PerlEnv_getenv("LANGUAGE");
263 char *lc_all = PerlEnv_getenv("LC_ALL");
264 char *lang = PerlEnv_getenv("LANG");
265 bool setlocale_failure = FALSE;
267 #ifdef LOCALE_ENVIRON_REQUIRED
270 * Ultrix setlocale(..., "") fails if there are no environment
271 * variables from which to get a locale name.
278 if (setlocale(LC_ALL, ""))
281 setlocale_failure = TRUE;
283 if (!setlocale_failure) {
284 #ifdef USE_LOCALE_CTYPE
287 (!done && (lang || PerlEnv_getenv("LC_CTYPE")))
289 setlocale_failure = TRUE;
291 curctype = savepv(curctype);
292 #endif /* USE_LOCALE_CTYPE */
293 #ifdef USE_LOCALE_COLLATE
295 setlocale(LC_COLLATE,
296 (!done && (lang || PerlEnv_getenv("LC_COLLATE")))
298 setlocale_failure = TRUE;
300 curcoll = savepv(curcoll);
301 #endif /* USE_LOCALE_COLLATE */
302 #ifdef USE_LOCALE_NUMERIC
304 setlocale(LC_NUMERIC,
305 (!done && (lang || PerlEnv_getenv("LC_NUMERIC")))
307 setlocale_failure = TRUE;
309 curnum = savepv(curnum);
310 #endif /* USE_LOCALE_NUMERIC */
315 #endif /* !LOCALE_ENVIRON_REQUIRED */
318 if (! setlocale(LC_ALL, ""))
319 setlocale_failure = TRUE;
322 if (!setlocale_failure) {
323 #ifdef USE_LOCALE_CTYPE
324 if (! (curctype = setlocale(LC_CTYPE, "")))
325 setlocale_failure = TRUE;
327 curctype = savepv(curctype);
328 #endif /* USE_LOCALE_CTYPE */
329 #ifdef USE_LOCALE_COLLATE
330 if (! (curcoll = setlocale(LC_COLLATE, "")))
331 setlocale_failure = TRUE;
333 curcoll = savepv(curcoll);
334 #endif /* USE_LOCALE_COLLATE */
335 #ifdef USE_LOCALE_NUMERIC
336 if (! (curnum = setlocale(LC_NUMERIC, "")))
337 setlocale_failure = TRUE;
339 curnum = savepv(curnum);
340 #endif /* USE_LOCALE_NUMERIC */
343 if (setlocale_failure) {
345 bool locwarn = (printwarn > 1 ||
347 (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
352 PerlIO_printf(Perl_error_log,
353 "perl: warning: Setting locale failed.\n");
357 PerlIO_printf(Perl_error_log,
358 "perl: warning: Setting locale failed for the categories:\n\t");
359 #ifdef USE_LOCALE_CTYPE
361 PerlIO_printf(Perl_error_log, "LC_CTYPE ");
362 #endif /* USE_LOCALE_CTYPE */
363 #ifdef USE_LOCALE_COLLATE
365 PerlIO_printf(Perl_error_log, "LC_COLLATE ");
366 #endif /* USE_LOCALE_COLLATE */
367 #ifdef USE_LOCALE_NUMERIC
369 PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
370 #endif /* USE_LOCALE_NUMERIC */
371 PerlIO_printf(Perl_error_log, "\n");
375 PerlIO_printf(Perl_error_log,
376 "perl: warning: Please check that your locale settings:\n");
379 PerlIO_printf(Perl_error_log,
380 "\tLANGUAGE = %c%s%c,\n",
381 language ? '"' : '(',
382 language ? language : "unset",
383 language ? '"' : ')');
386 PerlIO_printf(Perl_error_log,
387 "\tLC_ALL = %c%s%c,\n",
389 lc_all ? lc_all : "unset",
392 #if defined(USE_ENVIRON_ARRAY)
395 for (e = environ; *e; e++) {
396 if (strnEQ(*e, "LC_", 3)
397 && strnNE(*e, "LC_ALL=", 7)
398 && (p = strchr(*e, '=')))
399 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
400 (int)(p - *e), *e, p + 1);
404 PerlIO_printf(Perl_error_log,
405 "\t(possibly more locale environment variables)\n");
408 PerlIO_printf(Perl_error_log,
411 lang ? lang : "unset",
414 PerlIO_printf(Perl_error_log,
415 " are supported and installed on your system.\n");
420 if (setlocale(LC_ALL, "C")) {
422 PerlIO_printf(Perl_error_log,
423 "perl: warning: Falling back to the standard locale (\"C\").\n");
428 PerlIO_printf(Perl_error_log,
429 "perl: warning: Failed to fall back to the standard locale (\"C\").\n");
436 #ifdef USE_LOCALE_CTYPE
437 || !(curctype || setlocale(LC_CTYPE, "C"))
438 #endif /* USE_LOCALE_CTYPE */
439 #ifdef USE_LOCALE_COLLATE
440 || !(curcoll || setlocale(LC_COLLATE, "C"))
441 #endif /* USE_LOCALE_COLLATE */
442 #ifdef USE_LOCALE_NUMERIC
443 || !(curnum || setlocale(LC_NUMERIC, "C"))
444 #endif /* USE_LOCALE_NUMERIC */
448 PerlIO_printf(Perl_error_log,
449 "perl: warning: Cannot fall back to the standard locale (\"C\").\n");
453 #endif /* ! LC_ALL */
455 #ifdef USE_LOCALE_CTYPE
456 curctype = savepv(setlocale(LC_CTYPE, Nullch));
457 #endif /* USE_LOCALE_CTYPE */
458 #ifdef USE_LOCALE_COLLATE
459 curcoll = savepv(setlocale(LC_COLLATE, Nullch));
460 #endif /* USE_LOCALE_COLLATE */
461 #ifdef USE_LOCALE_NUMERIC
462 curnum = savepv(setlocale(LC_NUMERIC, Nullch));
463 #endif /* USE_LOCALE_NUMERIC */
467 #ifdef USE_LOCALE_CTYPE
469 #endif /* USE_LOCALE_CTYPE */
471 #ifdef USE_LOCALE_COLLATE
472 new_collate(curcoll);
473 #endif /* USE_LOCALE_COLLATE */
475 #ifdef USE_LOCALE_NUMERIC
477 #endif /* USE_LOCALE_NUMERIC */
481 #endif /* USE_LOCALE */
485 /* Set PL_utf8locale to TRUE if using PerlIO _and_
486 any of the following are true:
487 - nl_langinfo(CODESET) contains /^utf-?8/i
488 - $ENV{LC_ALL} contains /^utf-?8/i
489 - $ENV{LC_CTYPE} contains /^utf-?8/i
490 - $ENV{LANG} contains /^utf-?8/i
491 The LC_ALL, LC_CTYPE, LANG obey the usual override
492 hierarchy of locale environment variables. (LANGUAGE
493 affects only LC_MESSAGES only under glibc.) (If present,
494 it overrides LC_MESSAGES for GNU gettext, and it also
495 can have more than one locale, separated by spaces,
496 in case you need to know.)
497 If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
498 are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
499 on STDIN, STDOUT, STDERR, _and_ the default open discipline.
501 bool utf8locale = FALSE;
502 char *codeset = NULL;
503 #if defined(HAS_NL_LANGINFO) && defined(CODESET)
504 codeset = nl_langinfo(CODESET);
507 utf8locale = (ibcmp(codeset, "UTF-8", 5) == 0 ||
508 ibcmp(codeset, "UTF8", 4) == 0);
509 #if defined(USE_LOCALE)
510 else { /* nl_langinfo(CODESET) is supposed to correctly
511 * interpret the locale environment variables,
512 * but just in case it fails, let's do this manually. */
514 utf8locale = (ibcmp(lang, "UTF-8", 5) == 0 ||
515 ibcmp(lang, "UTF8", 4) == 0);
516 #ifdef USE_LOCALE_CTYPE
518 utf8locale = (ibcmp(curctype, "UTF-8", 5) == 0 ||
519 ibcmp(curctype, "UTF8", 4) == 0);
522 utf8locale = (ibcmp(lc_all, "UTF-8", 5) == 0 ||
523 ibcmp(lc_all, "UTF8", 4) == 0);
525 #endif /* USE_LOCALE */
527 PL_utf8locale = TRUE;
529 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
530 This is an alternative to using the -C command line switch
531 (the -C if present will override this). */
533 char *p = PerlEnv_getenv("PERL_UNICODE");
534 PL_unicode = p ? parse_unicode_opts(&p) : 0;
538 #ifdef USE_LOCALE_CTYPE
539 if (curctype != NULL)
541 #endif /* USE_LOCALE_CTYPE */
542 #ifdef USE_LOCALE_COLLATE
545 #endif /* USE_LOCALE_COLLATE */
546 #ifdef USE_LOCALE_NUMERIC
549 #endif /* USE_LOCALE_NUMERIC */
553 /* Backwards compatibility. */
555 Perl_init_i18nl14n(pTHX_ int printwarn)
557 return init_i18nl10n(printwarn);
560 #ifdef USE_LOCALE_COLLATE
563 * mem_collxfrm() is a bit like strxfrm() but with two important
564 * differences. First, it handles embedded NULs. Second, it allocates
565 * a bit more memory than needed for the transformed data itself.
566 * The real transformed data begins at offset sizeof(collationix).
567 * Please see sv_collxfrm() to see how this is used.
571 Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen)
574 STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */
576 /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */
577 /* the +1 is for the terminating NUL. */
579 xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1;
580 New(171, xbuf, xAlloc, char);
584 *(U32*)xbuf = PL_collation_ix;
585 xout = sizeof(PL_collation_ix);
586 for (xin = 0; xin < len; ) {
590 xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout);
593 if ((STRLEN)xused < xAlloc - xout)
595 xAlloc = (2 * xAlloc) + 1;
596 Renew(xbuf, xAlloc, char);
601 xin += strlen(s + xin) + 1;
604 /* Embedded NULs are understood but silently skipped
605 * because they make no sense in locale collation. */
609 *xlen = xout - sizeof(PL_collation_ix);
618 #endif /* USE_LOCALE_COLLATE */