Commit | Line | Data |
---|---|---|
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 | /* | |
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 | */ | |
53 | STATIC char * | |
54 | S_stdize_locale(pTHX_ char *locs) | |
55 | { | |
7452cf6a | 56 | const char * const s = strchr(locs, '='); |
98994639 HS |
57 | bool okay = TRUE; |
58 | ||
7918f24d NC |
59 | PERL_ARGS_ASSERT_STDIZE_LOCALE; |
60 | ||
8772537c AL |
61 | if (s) { |
62 | const char * const t = strchr(s, '.'); | |
98994639 | 63 | okay = FALSE; |
8772537c AL |
64 | if (t) { |
65 | const char * const u = strchr(t, '\n'); | |
66 | if (u && (u[1] == 0)) { | |
67 | const STRLEN len = u - s; | |
68 | Move(s + 1, locs, len, char); | |
69 | locs[len] = 0; | |
70 | okay = TRUE; | |
98994639 HS |
71 | } |
72 | } | |
73 | } | |
74 | ||
75 | if (!okay) | |
76 | Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs); | |
77 | ||
78 | return locs; | |
79 | } | |
27da23d5 | 80 | #endif |
98994639 HS |
81 | |
82 | void | |
83 | Perl_set_numeric_radix(pTHX) | |
84 | { | |
85 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 86 | dVAR; |
98994639 | 87 | # ifdef HAS_LOCALECONV |
7452cf6a | 88 | const struct lconv* const lc = localeconv(); |
98994639 | 89 | |
98994639 HS |
90 | if (lc && lc->decimal_point) { |
91 | if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) { | |
92 | SvREFCNT_dec(PL_numeric_radix_sv); | |
a0714e2c | 93 | PL_numeric_radix_sv = NULL; |
98994639 HS |
94 | } |
95 | else { | |
96 | if (PL_numeric_radix_sv) | |
97 | sv_setpv(PL_numeric_radix_sv, lc->decimal_point); | |
98 | else | |
99 | PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0); | |
100 | } | |
101 | } | |
102 | else | |
a0714e2c | 103 | PL_numeric_radix_sv = NULL; |
98994639 HS |
104 | # endif /* HAS_LOCALECONV */ |
105 | #endif /* USE_LOCALE_NUMERIC */ | |
106 | } | |
107 | ||
108 | /* | |
109 | * Set up for a new numeric locale. | |
110 | */ | |
111 | void | |
8772537c | 112 | Perl_new_numeric(pTHX_ const char *newnum) |
98994639 HS |
113 | { |
114 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 115 | dVAR; |
98994639 HS |
116 | |
117 | if (! newnum) { | |
43c5f42d NC |
118 | Safefree(PL_numeric_name); |
119 | PL_numeric_name = NULL; | |
98994639 HS |
120 | PL_numeric_standard = TRUE; |
121 | PL_numeric_local = TRUE; | |
122 | return; | |
123 | } | |
124 | ||
125 | if (! PL_numeric_name || strNE(PL_numeric_name, newnum)) { | |
126 | Safefree(PL_numeric_name); | |
127 | PL_numeric_name = stdize_locale(savepv(newnum)); | |
770526c1 NC |
128 | PL_numeric_standard = ((*newnum == 'C' && newnum[1] == '\0') |
129 | || strEQ(newnum, "POSIX")); | |
98994639 HS |
130 | PL_numeric_local = TRUE; |
131 | set_numeric_radix(); | |
132 | } | |
133 | ||
134 | #endif /* USE_LOCALE_NUMERIC */ | |
135 | } | |
136 | ||
137 | void | |
138 | Perl_set_numeric_standard(pTHX) | |
139 | { | |
140 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 141 | dVAR; |
98994639 HS |
142 | |
143 | if (! PL_numeric_standard) { | |
144 | setlocale(LC_NUMERIC, "C"); | |
145 | PL_numeric_standard = TRUE; | |
146 | PL_numeric_local = FALSE; | |
147 | set_numeric_radix(); | |
148 | } | |
149 | ||
150 | #endif /* USE_LOCALE_NUMERIC */ | |
151 | } | |
152 | ||
153 | void | |
154 | Perl_set_numeric_local(pTHX) | |
155 | { | |
156 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 157 | dVAR; |
98994639 HS |
158 | |
159 | if (! PL_numeric_local) { | |
160 | setlocale(LC_NUMERIC, PL_numeric_name); | |
161 | PL_numeric_standard = FALSE; | |
162 | PL_numeric_local = TRUE; | |
163 | set_numeric_radix(); | |
164 | } | |
165 | ||
166 | #endif /* USE_LOCALE_NUMERIC */ | |
167 | } | |
168 | ||
169 | /* | |
170 | * Set up for a new ctype locale. | |
171 | */ | |
172 | void | |
8772537c | 173 | Perl_new_ctype(pTHX_ const char *newctype) |
98994639 HS |
174 | { |
175 | #ifdef USE_LOCALE_CTYPE | |
27da23d5 | 176 | dVAR; |
98994639 HS |
177 | int i; |
178 | ||
7918f24d NC |
179 | PERL_ARGS_ASSERT_NEW_CTYPE; |
180 | ||
98994639 HS |
181 | for (i = 0; i < 256; i++) { |
182 | if (isUPPER_LC(i)) | |
183 | PL_fold_locale[i] = toLOWER_LC(i); | |
184 | else if (isLOWER_LC(i)) | |
185 | PL_fold_locale[i] = toUPPER_LC(i); | |
186 | else | |
187 | PL_fold_locale[i] = i; | |
188 | } | |
189 | ||
190 | #endif /* USE_LOCALE_CTYPE */ | |
7918f24d | 191 | PERL_ARGS_ASSERT_NEW_CTYPE; |
8772537c | 192 | PERL_UNUSED_ARG(newctype); |
96a5add6 | 193 | PERL_UNUSED_CONTEXT; |
98994639 HS |
194 | } |
195 | ||
196 | /* | |
197 | * Set up for a new collation locale. | |
198 | */ | |
199 | void | |
8772537c | 200 | Perl_new_collate(pTHX_ const char *newcoll) |
98994639 HS |
201 | { |
202 | #ifdef USE_LOCALE_COLLATE | |
97aff369 | 203 | dVAR; |
98994639 HS |
204 | |
205 | if (! newcoll) { | |
206 | if (PL_collation_name) { | |
207 | ++PL_collation_ix; | |
208 | Safefree(PL_collation_name); | |
209 | PL_collation_name = NULL; | |
210 | } | |
211 | PL_collation_standard = TRUE; | |
212 | PL_collxfrm_base = 0; | |
213 | PL_collxfrm_mult = 2; | |
214 | return; | |
215 | } | |
216 | ||
217 | if (! PL_collation_name || strNE(PL_collation_name, newcoll)) { | |
218 | ++PL_collation_ix; | |
219 | Safefree(PL_collation_name); | |
220 | PL_collation_name = stdize_locale(savepv(newcoll)); | |
770526c1 NC |
221 | PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0') |
222 | || strEQ(newcoll, "POSIX")); | |
98994639 HS |
223 | |
224 | { | |
225 | /* 2: at most so many chars ('a', 'b'). */ | |
226 | /* 50: surely no system expands a char more. */ | |
227 | #define XFRMBUFSIZE (2 * 50) | |
228 | char xbuf[XFRMBUFSIZE]; | |
8772537c AL |
229 | const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE); |
230 | const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE); | |
231 | const SSize_t mult = fb - fa; | |
98994639 HS |
232 | if (mult < 1) |
233 | Perl_croak(aTHX_ "strxfrm() gets absurd"); | |
eb160463 | 234 | PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0; |
98994639 HS |
235 | PL_collxfrm_mult = mult; |
236 | } | |
237 | } | |
238 | ||
239 | #endif /* USE_LOCALE_COLLATE */ | |
240 | } | |
241 | ||
242 | /* | |
243 | * Initialize locale awareness. | |
244 | */ | |
245 | int | |
246 | Perl_init_i18nl10n(pTHX_ int printwarn) | |
247 | { | |
248 | int ok = 1; | |
249 | /* returns | |
250 | * 1 = set ok or not applicable, | |
251 | * 0 = fallback to C locale, | |
252 | * -1 = fallback to C locale failed | |
253 | */ | |
254 | ||
255 | #if defined(USE_LOCALE) | |
97aff369 | 256 | dVAR; |
98994639 HS |
257 | |
258 | #ifdef USE_LOCALE_CTYPE | |
259 | char *curctype = NULL; | |
260 | #endif /* USE_LOCALE_CTYPE */ | |
261 | #ifdef USE_LOCALE_COLLATE | |
262 | char *curcoll = NULL; | |
263 | #endif /* USE_LOCALE_COLLATE */ | |
264 | #ifdef USE_LOCALE_NUMERIC | |
265 | char *curnum = NULL; | |
266 | #endif /* USE_LOCALE_NUMERIC */ | |
267 | #ifdef __GLIBC__ | |
7452cf6a | 268 | char * const language = PerlEnv_getenv("LANGUAGE"); |
98994639 | 269 | #endif |
7452cf6a AL |
270 | char * const lc_all = PerlEnv_getenv("LC_ALL"); |
271 | char * const lang = PerlEnv_getenv("LANG"); | |
98994639 HS |
272 | bool setlocale_failure = FALSE; |
273 | ||
274 | #ifdef LOCALE_ENVIRON_REQUIRED | |
275 | ||
276 | /* | |
277 | * Ultrix setlocale(..., "") fails if there are no environment | |
278 | * variables from which to get a locale name. | |
279 | */ | |
280 | ||
281 | bool done = FALSE; | |
282 | ||
283 | #ifdef LC_ALL | |
284 | if (lang) { | |
285 | if (setlocale(LC_ALL, "")) | |
286 | done = TRUE; | |
287 | else | |
288 | setlocale_failure = TRUE; | |
289 | } | |
290 | if (!setlocale_failure) { | |
291 | #ifdef USE_LOCALE_CTYPE | |
56279a21 | 292 | Safefree(curctype); |
98994639 HS |
293 | if (! (curctype = |
294 | setlocale(LC_CTYPE, | |
295 | (!done && (lang || PerlEnv_getenv("LC_CTYPE"))) | |
bd61b366 | 296 | ? "" : NULL))) |
98994639 HS |
297 | setlocale_failure = TRUE; |
298 | else | |
299 | curctype = savepv(curctype); | |
300 | #endif /* USE_LOCALE_CTYPE */ | |
301 | #ifdef USE_LOCALE_COLLATE | |
56279a21 | 302 | Safefree(curcoll); |
98994639 HS |
303 | if (! (curcoll = |
304 | setlocale(LC_COLLATE, | |
305 | (!done && (lang || PerlEnv_getenv("LC_COLLATE"))) | |
bd61b366 | 306 | ? "" : NULL))) |
98994639 HS |
307 | setlocale_failure = TRUE; |
308 | else | |
309 | curcoll = savepv(curcoll); | |
310 | #endif /* USE_LOCALE_COLLATE */ | |
311 | #ifdef USE_LOCALE_NUMERIC | |
56279a21 | 312 | Safefree(curnum); |
98994639 HS |
313 | if (! (curnum = |
314 | setlocale(LC_NUMERIC, | |
315 | (!done && (lang || PerlEnv_getenv("LC_NUMERIC"))) | |
bd61b366 | 316 | ? "" : NULL))) |
98994639 HS |
317 | setlocale_failure = TRUE; |
318 | else | |
319 | curnum = savepv(curnum); | |
320 | #endif /* USE_LOCALE_NUMERIC */ | |
321 | } | |
322 | ||
323 | #endif /* LC_ALL */ | |
324 | ||
325 | #endif /* !LOCALE_ENVIRON_REQUIRED */ | |
326 | ||
327 | #ifdef LC_ALL | |
328 | if (! setlocale(LC_ALL, "")) | |
329 | setlocale_failure = TRUE; | |
330 | #endif /* LC_ALL */ | |
331 | ||
332 | if (!setlocale_failure) { | |
333 | #ifdef USE_LOCALE_CTYPE | |
6cb43dbf | 334 | Safefree(curctype); |
98994639 HS |
335 | if (! (curctype = setlocale(LC_CTYPE, ""))) |
336 | setlocale_failure = TRUE; | |
337 | else | |
338 | curctype = savepv(curctype); | |
339 | #endif /* USE_LOCALE_CTYPE */ | |
340 | #ifdef USE_LOCALE_COLLATE | |
6cb43dbf | 341 | Safefree(curcoll); |
98994639 HS |
342 | if (! (curcoll = setlocale(LC_COLLATE, ""))) |
343 | setlocale_failure = TRUE; | |
344 | else | |
345 | curcoll = savepv(curcoll); | |
346 | #endif /* USE_LOCALE_COLLATE */ | |
347 | #ifdef USE_LOCALE_NUMERIC | |
6cb43dbf | 348 | Safefree(curnum); |
98994639 HS |
349 | if (! (curnum = setlocale(LC_NUMERIC, ""))) |
350 | setlocale_failure = TRUE; | |
351 | else | |
352 | curnum = savepv(curnum); | |
353 | #endif /* USE_LOCALE_NUMERIC */ | |
354 | } | |
355 | ||
356 | if (setlocale_failure) { | |
357 | char *p; | |
0bd48802 | 358 | const bool locwarn = (printwarn > 1 || |
98994639 HS |
359 | (printwarn && |
360 | (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p)))); | |
361 | ||
362 | if (locwarn) { | |
363 | #ifdef LC_ALL | |
364 | ||
365 | PerlIO_printf(Perl_error_log, | |
366 | "perl: warning: Setting locale failed.\n"); | |
367 | ||
368 | #else /* !LC_ALL */ | |
369 | ||
370 | PerlIO_printf(Perl_error_log, | |
371 | "perl: warning: Setting locale failed for the categories:\n\t"); | |
372 | #ifdef USE_LOCALE_CTYPE | |
373 | if (! curctype) | |
374 | PerlIO_printf(Perl_error_log, "LC_CTYPE "); | |
375 | #endif /* USE_LOCALE_CTYPE */ | |
376 | #ifdef USE_LOCALE_COLLATE | |
377 | if (! curcoll) | |
378 | PerlIO_printf(Perl_error_log, "LC_COLLATE "); | |
379 | #endif /* USE_LOCALE_COLLATE */ | |
380 | #ifdef USE_LOCALE_NUMERIC | |
381 | if (! curnum) | |
382 | PerlIO_printf(Perl_error_log, "LC_NUMERIC "); | |
383 | #endif /* USE_LOCALE_NUMERIC */ | |
384 | PerlIO_printf(Perl_error_log, "\n"); | |
385 | ||
386 | #endif /* LC_ALL */ | |
387 | ||
388 | PerlIO_printf(Perl_error_log, | |
389 | "perl: warning: Please check that your locale settings:\n"); | |
390 | ||
391 | #ifdef __GLIBC__ | |
392 | PerlIO_printf(Perl_error_log, | |
393 | "\tLANGUAGE = %c%s%c,\n", | |
394 | language ? '"' : '(', | |
395 | language ? language : "unset", | |
396 | language ? '"' : ')'); | |
397 | #endif | |
398 | ||
399 | PerlIO_printf(Perl_error_log, | |
400 | "\tLC_ALL = %c%s%c,\n", | |
401 | lc_all ? '"' : '(', | |
402 | lc_all ? lc_all : "unset", | |
403 | lc_all ? '"' : ')'); | |
404 | ||
405 | #if defined(USE_ENVIRON_ARRAY) | |
406 | { | |
407 | char **e; | |
408 | for (e = environ; *e; e++) { | |
409 | if (strnEQ(*e, "LC_", 3) | |
410 | && strnNE(*e, "LC_ALL=", 7) | |
411 | && (p = strchr(*e, '='))) | |
412 | PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n", | |
413 | (int)(p - *e), *e, p + 1); | |
414 | } | |
415 | } | |
416 | #else | |
417 | PerlIO_printf(Perl_error_log, | |
418 | "\t(possibly more locale environment variables)\n"); | |
419 | #endif | |
420 | ||
421 | PerlIO_printf(Perl_error_log, | |
422 | "\tLANG = %c%s%c\n", | |
423 | lang ? '"' : '(', | |
424 | lang ? lang : "unset", | |
425 | lang ? '"' : ')'); | |
426 | ||
427 | PerlIO_printf(Perl_error_log, | |
428 | " are supported and installed on your system.\n"); | |
429 | } | |
430 | ||
431 | #ifdef LC_ALL | |
432 | ||
433 | if (setlocale(LC_ALL, "C")) { | |
434 | if (locwarn) | |
435 | PerlIO_printf(Perl_error_log, | |
436 | "perl: warning: Falling back to the standard locale (\"C\").\n"); | |
437 | ok = 0; | |
438 | } | |
439 | else { | |
440 | if (locwarn) | |
441 | PerlIO_printf(Perl_error_log, | |
442 | "perl: warning: Failed to fall back to the standard locale (\"C\").\n"); | |
443 | ok = -1; | |
444 | } | |
445 | ||
446 | #else /* ! LC_ALL */ | |
447 | ||
448 | if (0 | |
449 | #ifdef USE_LOCALE_CTYPE | |
450 | || !(curctype || setlocale(LC_CTYPE, "C")) | |
451 | #endif /* USE_LOCALE_CTYPE */ | |
452 | #ifdef USE_LOCALE_COLLATE | |
453 | || !(curcoll || setlocale(LC_COLLATE, "C")) | |
454 | #endif /* USE_LOCALE_COLLATE */ | |
455 | #ifdef USE_LOCALE_NUMERIC | |
456 | || !(curnum || setlocale(LC_NUMERIC, "C")) | |
457 | #endif /* USE_LOCALE_NUMERIC */ | |
458 | ) | |
459 | { | |
460 | if (locwarn) | |
461 | PerlIO_printf(Perl_error_log, | |
462 | "perl: warning: Cannot fall back to the standard locale (\"C\").\n"); | |
463 | ok = -1; | |
464 | } | |
465 | ||
466 | #endif /* ! LC_ALL */ | |
467 | ||
468 | #ifdef USE_LOCALE_CTYPE | |
56279a21 | 469 | Safefree(curctype); |
bd61b366 | 470 | curctype = savepv(setlocale(LC_CTYPE, NULL)); |
98994639 HS |
471 | #endif /* USE_LOCALE_CTYPE */ |
472 | #ifdef USE_LOCALE_COLLATE | |
56279a21 | 473 | Safefree(curcoll); |
bd61b366 | 474 | curcoll = savepv(setlocale(LC_COLLATE, NULL)); |
98994639 HS |
475 | #endif /* USE_LOCALE_COLLATE */ |
476 | #ifdef USE_LOCALE_NUMERIC | |
56279a21 | 477 | Safefree(curnum); |
bd61b366 | 478 | curnum = savepv(setlocale(LC_NUMERIC, NULL)); |
98994639 HS |
479 | #endif /* USE_LOCALE_NUMERIC */ |
480 | } | |
481 | else { | |
482 | ||
483 | #ifdef USE_LOCALE_CTYPE | |
484 | new_ctype(curctype); | |
485 | #endif /* USE_LOCALE_CTYPE */ | |
486 | ||
487 | #ifdef USE_LOCALE_COLLATE | |
488 | new_collate(curcoll); | |
489 | #endif /* USE_LOCALE_COLLATE */ | |
490 | ||
491 | #ifdef USE_LOCALE_NUMERIC | |
492 | new_numeric(curnum); | |
493 | #endif /* USE_LOCALE_NUMERIC */ | |
b310b053 | 494 | |
98994639 HS |
495 | } |
496 | ||
497 | #endif /* USE_LOCALE */ | |
498 | ||
ec71e770 | 499 | #ifdef USE_PERLIO |
b310b053 | 500 | { |
fde18df1 | 501 | /* Set PL_utf8locale to TRUE if using PerlIO _and_ |
ec71e770 | 502 | any of the following are true: |
085a54d9 | 503 | - nl_langinfo(CODESET) contains /^utf-?8/i |
61de9fb5 | 504 | - $ENV{LC_ALL} contains /^utf-?8/i |
085a54d9 | 505 | - $ENV{LC_CTYPE} contains /^utf-?8/i |
61de9fb5 JH |
506 | - $ENV{LANG} contains /^utf-?8/i |
507 | The LC_ALL, LC_CTYPE, LANG obey the usual override | |
508 | hierarchy of locale environment variables. (LANGUAGE | |
509 | affects only LC_MESSAGES only under glibc.) (If present, | |
510 | it overrides LC_MESSAGES for GNU gettext, and it also | |
511 | can have more than one locale, separated by spaces, | |
512 | in case you need to know.) | |
8aa8f774 | 513 | If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE}) |
a05d7ebb | 514 | are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer |
fde18df1 | 515 | on STDIN, STDOUT, STDERR, _and_ the default open discipline. |
085a54d9 | 516 | */ |
fde18df1 | 517 | bool utf8locale = FALSE; |
b310b053 JH |
518 | char *codeset = NULL; |
519 | #if defined(HAS_NL_LANGINFO) && defined(CODESET) | |
520 | codeset = nl_langinfo(CODESET); | |
521 | #endif | |
61de9fb5 | 522 | if (codeset) |
89529cee GA |
523 | utf8locale = (Perl_ibcmp(aTHX_ codeset, STR_WITH_LEN("UTF-8")) == 0 || |
524 | Perl_ibcmp(aTHX_ codeset, STR_WITH_LEN("UTF8") ) == 0); | |
077a72a9 | 525 | #if defined(USE_LOCALE) |
61de9fb5 JH |
526 | else { /* nl_langinfo(CODESET) is supposed to correctly |
527 | * interpret the locale environment variables, | |
528 | * but just in case it fails, let's do this manually. */ | |
529 | if (lang) | |
89529cee GA |
530 | utf8locale = (Perl_ibcmp(aTHX_ lang, STR_WITH_LEN("UTF-8")) == 0 || |
531 | Perl_ibcmp(aTHX_ lang, STR_WITH_LEN("UTF8") ) == 0); | |
b310b053 | 532 | #ifdef USE_LOCALE_CTYPE |
61de9fb5 | 533 | if (curctype) |
89529cee GA |
534 | utf8locale = (Perl_ibcmp(aTHX_ curctype, STR_WITH_LEN("UTF-8")) == 0 || |
535 | Perl_ibcmp(aTHX_ curctype, STR_WITH_LEN("UTF8") ) == 0); | |
b310b053 | 536 | #endif |
61de9fb5 | 537 | if (lc_all) |
89529cee GA |
538 | utf8locale = (Perl_ibcmp(aTHX_ lc_all, STR_WITH_LEN("UTF-8")) == 0 || |
539 | Perl_ibcmp(aTHX_ lc_all, STR_WITH_LEN("UTF8") ) == 0); | |
61de9fb5 | 540 | } |
fde18df1 JH |
541 | #endif /* USE_LOCALE */ |
542 | if (utf8locale) | |
543 | PL_utf8locale = TRUE; | |
544 | } | |
a05d7ebb | 545 | /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO. |
fde18df1 JH |
546 | This is an alternative to using the -C command line switch |
547 | (the -C if present will override this). */ | |
548 | { | |
dd374669 | 549 | const char *p = PerlEnv_getenv("PERL_UNICODE"); |
a05d7ebb | 550 | PL_unicode = p ? parse_unicode_opts(&p) : 0; |
5a22a2bb NC |
551 | if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG) |
552 | PL_utf8cache = -1; | |
b310b053 | 553 | } |
ec71e770 | 554 | #endif |
b310b053 | 555 | |
98994639 | 556 | #ifdef USE_LOCALE_CTYPE |
43c5f42d | 557 | Safefree(curctype); |
98994639 HS |
558 | #endif /* USE_LOCALE_CTYPE */ |
559 | #ifdef USE_LOCALE_COLLATE | |
43c5f42d | 560 | Safefree(curcoll); |
98994639 HS |
561 | #endif /* USE_LOCALE_COLLATE */ |
562 | #ifdef USE_LOCALE_NUMERIC | |
43c5f42d | 563 | Safefree(curnum); |
98994639 HS |
564 | #endif /* USE_LOCALE_NUMERIC */ |
565 | return ok; | |
566 | } | |
567 | ||
98994639 HS |
568 | #ifdef USE_LOCALE_COLLATE |
569 | ||
570 | /* | |
571 | * mem_collxfrm() is a bit like strxfrm() but with two important | |
572 | * differences. First, it handles embedded NULs. Second, it allocates | |
573 | * a bit more memory than needed for the transformed data itself. | |
574 | * The real transformed data begins at offset sizeof(collationix). | |
575 | * Please see sv_collxfrm() to see how this is used. | |
576 | */ | |
166f8a29 | 577 | |
98994639 HS |
578 | char * |
579 | Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen) | |
580 | { | |
97aff369 | 581 | dVAR; |
98994639 HS |
582 | char *xbuf; |
583 | STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */ | |
584 | ||
7918f24d NC |
585 | PERL_ARGS_ASSERT_MEM_COLLXFRM; |
586 | ||
98994639 HS |
587 | /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */ |
588 | /* the +1 is for the terminating NUL. */ | |
589 | ||
590 | xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1; | |
a02a5408 | 591 | Newx(xbuf, xAlloc, char); |
98994639 HS |
592 | if (! xbuf) |
593 | goto bad; | |
594 | ||
595 | *(U32*)xbuf = PL_collation_ix; | |
596 | xout = sizeof(PL_collation_ix); | |
597 | for (xin = 0; xin < len; ) { | |
224e8ef5 | 598 | Size_t xused; |
98994639 HS |
599 | |
600 | for (;;) { | |
601 | xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout); | |
224e8ef5 | 602 | if (xused >= PERL_INT_MAX) |
98994639 | 603 | goto bad; |
eb160463 | 604 | if ((STRLEN)xused < xAlloc - xout) |
98994639 HS |
605 | break; |
606 | xAlloc = (2 * xAlloc) + 1; | |
607 | Renew(xbuf, xAlloc, char); | |
608 | if (! xbuf) | |
609 | goto bad; | |
610 | } | |
611 | ||
612 | xin += strlen(s + xin) + 1; | |
613 | xout += xused; | |
614 | ||
615 | /* Embedded NULs are understood but silently skipped | |
616 | * because they make no sense in locale collation. */ | |
617 | } | |
618 | ||
619 | xbuf[xout] = '\0'; | |
620 | *xlen = xout - sizeof(PL_collation_ix); | |
621 | return xbuf; | |
622 | ||
623 | bad: | |
624 | Safefree(xbuf); | |
625 | *xlen = 0; | |
626 | return NULL; | |
627 | } | |
628 | ||
629 | #endif /* USE_LOCALE_COLLATE */ | |
630 | ||
66610fdd RGS |
631 | /* |
632 | * Local variables: | |
633 | * c-indentation-style: bsd | |
634 | * c-basic-offset: 4 | |
635 | * indent-tabs-mode: t | |
636 | * End: | |
637 | * | |
37442d52 RGS |
638 | * ex: set ts=8 sts=4 sw=4 noet: |
639 | */ |