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 | /* | |
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. | |
25 | */ | |
26 | ||
98994639 HS |
27 | #include "EXTERN.h" |
28 | #define PERL_IN_LOCALE_C | |
29 | #include "perl.h" | |
30 | ||
b310b053 JH |
31 | #ifdef I_LANGINFO |
32 | # include <langinfo.h> | |
33 | #endif | |
34 | ||
a4af207c JH |
35 | #include "reentr.h" |
36 | ||
27da23d5 | 37 | #if defined(USE_LOCALE_NUMERIC) || defined(USE_LOCALE_COLLATE) |
98994639 HS |
38 | /* |
39 | * Standardize the locale name from a string returned by 'setlocale'. | |
40 | * | |
0ef2a2b2 | 41 | * The typical return value of setlocale() is either |
98994639 HS |
42 | * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL |
43 | * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL | |
44 | * (the space-separated values represent the various sublocales, | |
0ef2a2b2 | 45 | * in some unspecified order). This is not handled by this function. |
98994639 HS |
46 | * |
47 | * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n", | |
0ef2a2b2 KW |
48 | * which is harmful for further use of the string in setlocale(). This |
49 | * function removes the trailing new line and everything up through the '=' | |
98994639 HS |
50 | * |
51 | */ | |
52 | STATIC char * | |
53 | S_stdize_locale(pTHX_ char *locs) | |
54 | { | |
7452cf6a | 55 | const char * const s = strchr(locs, '='); |
98994639 HS |
56 | bool okay = TRUE; |
57 | ||
7918f24d NC |
58 | PERL_ARGS_ASSERT_STDIZE_LOCALE; |
59 | ||
8772537c AL |
60 | if (s) { |
61 | const char * const t = strchr(s, '.'); | |
98994639 | 62 | okay = FALSE; |
8772537c AL |
63 | if (t) { |
64 | const char * const u = strchr(t, '\n'); | |
65 | if (u && (u[1] == 0)) { | |
66 | const STRLEN len = u - s; | |
67 | Move(s + 1, locs, len, char); | |
68 | locs[len] = 0; | |
69 | okay = TRUE; | |
98994639 HS |
70 | } |
71 | } | |
72 | } | |
73 | ||
74 | if (!okay) | |
75 | Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs); | |
76 | ||
77 | return locs; | |
78 | } | |
27da23d5 | 79 | #endif |
98994639 HS |
80 | |
81 | void | |
82 | Perl_set_numeric_radix(pTHX) | |
83 | { | |
84 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 85 | dVAR; |
98994639 | 86 | # ifdef HAS_LOCALECONV |
7452cf6a | 87 | const struct lconv* const lc = localeconv(); |
98994639 | 88 | |
98994639 HS |
89 | if (lc && lc->decimal_point) { |
90 | if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) { | |
91 | SvREFCNT_dec(PL_numeric_radix_sv); | |
a0714e2c | 92 | PL_numeric_radix_sv = NULL; |
98994639 HS |
93 | } |
94 | else { | |
95 | if (PL_numeric_radix_sv) | |
96 | sv_setpv(PL_numeric_radix_sv, lc->decimal_point); | |
97 | else | |
98 | PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0); | |
99 | } | |
100 | } | |
101 | else | |
a0714e2c | 102 | PL_numeric_radix_sv = NULL; |
98994639 HS |
103 | # endif /* HAS_LOCALECONV */ |
104 | #endif /* USE_LOCALE_NUMERIC */ | |
105 | } | |
106 | ||
107 | /* | |
108 | * Set up for a new numeric locale. | |
109 | */ | |
110 | void | |
8772537c | 111 | Perl_new_numeric(pTHX_ const char *newnum) |
98994639 HS |
112 | { |
113 | #ifdef USE_LOCALE_NUMERIC | |
b03f34cf | 114 | char *save_newnum; |
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 | ||
b03f34cf KW |
125 | save_newnum = stdize_locale(savepv(newnum)); |
126 | if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) { | |
98994639 | 127 | Safefree(PL_numeric_name); |
b03f34cf KW |
128 | PL_numeric_name = save_newnum; |
129 | PL_numeric_standard = ((*save_newnum == 'C' && save_newnum[1] == '\0') | |
130 | || strEQ(save_newnum, "POSIX")); | |
98994639 HS |
131 | PL_numeric_local = TRUE; |
132 | set_numeric_radix(); | |
133 | } | |
b03f34cf KW |
134 | else { |
135 | Safefree(save_newnum); | |
136 | } | |
98994639 HS |
137 | |
138 | #endif /* USE_LOCALE_NUMERIC */ | |
139 | } | |
140 | ||
141 | void | |
142 | Perl_set_numeric_standard(pTHX) | |
143 | { | |
144 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 145 | dVAR; |
98994639 HS |
146 | |
147 | if (! PL_numeric_standard) { | |
148 | setlocale(LC_NUMERIC, "C"); | |
149 | PL_numeric_standard = TRUE; | |
150 | PL_numeric_local = FALSE; | |
151 | set_numeric_radix(); | |
152 | } | |
153 | ||
154 | #endif /* USE_LOCALE_NUMERIC */ | |
155 | } | |
156 | ||
157 | void | |
158 | Perl_set_numeric_local(pTHX) | |
159 | { | |
160 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 161 | dVAR; |
98994639 HS |
162 | |
163 | if (! PL_numeric_local) { | |
164 | setlocale(LC_NUMERIC, PL_numeric_name); | |
165 | PL_numeric_standard = FALSE; | |
166 | PL_numeric_local = TRUE; | |
167 | set_numeric_radix(); | |
168 | } | |
169 | ||
170 | #endif /* USE_LOCALE_NUMERIC */ | |
171 | } | |
172 | ||
173 | /* | |
174 | * Set up for a new ctype locale. | |
175 | */ | |
176 | void | |
8772537c | 177 | Perl_new_ctype(pTHX_ const char *newctype) |
98994639 HS |
178 | { |
179 | #ifdef USE_LOCALE_CTYPE | |
27da23d5 | 180 | dVAR; |
98994639 HS |
181 | int i; |
182 | ||
7918f24d NC |
183 | PERL_ARGS_ASSERT_NEW_CTYPE; |
184 | ||
98994639 HS |
185 | for (i = 0; i < 256; i++) { |
186 | if (isUPPER_LC(i)) | |
187 | PL_fold_locale[i] = toLOWER_LC(i); | |
188 | else if (isLOWER_LC(i)) | |
189 | PL_fold_locale[i] = toUPPER_LC(i); | |
190 | else | |
191 | PL_fold_locale[i] = i; | |
192 | } | |
193 | ||
194 | #endif /* USE_LOCALE_CTYPE */ | |
7918f24d | 195 | PERL_ARGS_ASSERT_NEW_CTYPE; |
8772537c | 196 | PERL_UNUSED_ARG(newctype); |
96a5add6 | 197 | PERL_UNUSED_CONTEXT; |
98994639 HS |
198 | } |
199 | ||
200 | /* | |
201 | * Set up for a new collation locale. | |
202 | */ | |
203 | void | |
8772537c | 204 | Perl_new_collate(pTHX_ const char *newcoll) |
98994639 HS |
205 | { |
206 | #ifdef USE_LOCALE_COLLATE | |
97aff369 | 207 | dVAR; |
98994639 HS |
208 | |
209 | if (! newcoll) { | |
210 | if (PL_collation_name) { | |
211 | ++PL_collation_ix; | |
212 | Safefree(PL_collation_name); | |
213 | PL_collation_name = NULL; | |
214 | } | |
215 | PL_collation_standard = TRUE; | |
216 | PL_collxfrm_base = 0; | |
217 | PL_collxfrm_mult = 2; | |
218 | return; | |
219 | } | |
220 | ||
221 | if (! PL_collation_name || strNE(PL_collation_name, newcoll)) { | |
222 | ++PL_collation_ix; | |
223 | Safefree(PL_collation_name); | |
224 | PL_collation_name = stdize_locale(savepv(newcoll)); | |
770526c1 NC |
225 | PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0') |
226 | || strEQ(newcoll, "POSIX")); | |
98994639 HS |
227 | |
228 | { | |
229 | /* 2: at most so many chars ('a', 'b'). */ | |
230 | /* 50: surely no system expands a char more. */ | |
231 | #define XFRMBUFSIZE (2 * 50) | |
232 | char xbuf[XFRMBUFSIZE]; | |
8772537c AL |
233 | const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE); |
234 | const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE); | |
235 | const SSize_t mult = fb - fa; | |
e0601d3c | 236 | if (mult < 1 && !(fa == 0 && fb == 0)) |
ad49ad39 NC |
237 | Perl_croak(aTHX_ "panic: strxfrm() gets absurd - a => %"UVuf", ab => %"UVuf, |
238 | (UV) fa, (UV) fb); | |
eb160463 | 239 | PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0; |
98994639 HS |
240 | PL_collxfrm_mult = mult; |
241 | } | |
242 | } | |
243 | ||
244 | #endif /* USE_LOCALE_COLLATE */ | |
245 | } | |
246 | ||
247 | /* | |
248 | * Initialize locale awareness. | |
249 | */ | |
250 | int | |
251 | Perl_init_i18nl10n(pTHX_ int printwarn) | |
252 | { | |
253 | int ok = 1; | |
254 | /* returns | |
255 | * 1 = set ok or not applicable, | |
256 | * 0 = fallback to C locale, | |
257 | * -1 = fallback to C locale failed | |
258 | */ | |
259 | ||
260 | #if defined(USE_LOCALE) | |
97aff369 | 261 | dVAR; |
98994639 HS |
262 | |
263 | #ifdef USE_LOCALE_CTYPE | |
264 | char *curctype = NULL; | |
265 | #endif /* USE_LOCALE_CTYPE */ | |
266 | #ifdef USE_LOCALE_COLLATE | |
267 | char *curcoll = NULL; | |
268 | #endif /* USE_LOCALE_COLLATE */ | |
269 | #ifdef USE_LOCALE_NUMERIC | |
270 | char *curnum = NULL; | |
271 | #endif /* USE_LOCALE_NUMERIC */ | |
272 | #ifdef __GLIBC__ | |
7452cf6a | 273 | char * const language = PerlEnv_getenv("LANGUAGE"); |
98994639 | 274 | #endif |
7452cf6a AL |
275 | char * const lc_all = PerlEnv_getenv("LC_ALL"); |
276 | char * const lang = PerlEnv_getenv("LANG"); | |
98994639 HS |
277 | bool setlocale_failure = FALSE; |
278 | ||
279 | #ifdef LOCALE_ENVIRON_REQUIRED | |
280 | ||
281 | /* | |
282 | * Ultrix setlocale(..., "") fails if there are no environment | |
283 | * variables from which to get a locale name. | |
284 | */ | |
285 | ||
286 | bool done = FALSE; | |
287 | ||
b3e384bf | 288 | # ifdef LC_ALL |
98994639 HS |
289 | if (lang) { |
290 | if (setlocale(LC_ALL, "")) | |
291 | done = TRUE; | |
292 | else | |
293 | setlocale_failure = TRUE; | |
294 | } | |
295 | if (!setlocale_failure) { | |
b3e384bf | 296 | # ifdef USE_LOCALE_CTYPE |
56279a21 | 297 | Safefree(curctype); |
98994639 HS |
298 | if (! (curctype = |
299 | setlocale(LC_CTYPE, | |
300 | (!done && (lang || PerlEnv_getenv("LC_CTYPE"))) | |
bd61b366 | 301 | ? "" : NULL))) |
98994639 HS |
302 | setlocale_failure = TRUE; |
303 | else | |
304 | curctype = savepv(curctype); | |
b3e384bf KW |
305 | # endif /* USE_LOCALE_CTYPE */ |
306 | # ifdef USE_LOCALE_COLLATE | |
56279a21 | 307 | Safefree(curcoll); |
98994639 HS |
308 | if (! (curcoll = |
309 | setlocale(LC_COLLATE, | |
310 | (!done && (lang || PerlEnv_getenv("LC_COLLATE"))) | |
bd61b366 | 311 | ? "" : NULL))) |
98994639 HS |
312 | setlocale_failure = TRUE; |
313 | else | |
314 | curcoll = savepv(curcoll); | |
b3e384bf KW |
315 | # endif /* USE_LOCALE_COLLATE */ |
316 | # ifdef USE_LOCALE_NUMERIC | |
56279a21 | 317 | Safefree(curnum); |
98994639 HS |
318 | if (! (curnum = |
319 | setlocale(LC_NUMERIC, | |
320 | (!done && (lang || PerlEnv_getenv("LC_NUMERIC"))) | |
bd61b366 | 321 | ? "" : NULL))) |
98994639 HS |
322 | setlocale_failure = TRUE; |
323 | else | |
324 | curnum = savepv(curnum); | |
b3e384bf | 325 | # endif /* USE_LOCALE_NUMERIC */ |
98994639 HS |
326 | } |
327 | ||
b3e384bf | 328 | # endif /* LC_ALL */ |
98994639 HS |
329 | |
330 | #endif /* !LOCALE_ENVIRON_REQUIRED */ | |
331 | ||
332 | #ifdef LC_ALL | |
333 | if (! setlocale(LC_ALL, "")) | |
334 | setlocale_failure = TRUE; | |
335 | #endif /* LC_ALL */ | |
336 | ||
337 | if (!setlocale_failure) { | |
338 | #ifdef USE_LOCALE_CTYPE | |
6cb43dbf | 339 | Safefree(curctype); |
98994639 HS |
340 | if (! (curctype = setlocale(LC_CTYPE, ""))) |
341 | setlocale_failure = TRUE; | |
342 | else | |
343 | curctype = savepv(curctype); | |
344 | #endif /* USE_LOCALE_CTYPE */ | |
345 | #ifdef USE_LOCALE_COLLATE | |
6cb43dbf | 346 | Safefree(curcoll); |
98994639 HS |
347 | if (! (curcoll = setlocale(LC_COLLATE, ""))) |
348 | setlocale_failure = TRUE; | |
349 | else | |
350 | curcoll = savepv(curcoll); | |
351 | #endif /* USE_LOCALE_COLLATE */ | |
352 | #ifdef USE_LOCALE_NUMERIC | |
6cb43dbf | 353 | Safefree(curnum); |
98994639 HS |
354 | if (! (curnum = setlocale(LC_NUMERIC, ""))) |
355 | setlocale_failure = TRUE; | |
356 | else | |
357 | curnum = savepv(curnum); | |
358 | #endif /* USE_LOCALE_NUMERIC */ | |
359 | } | |
360 | ||
361 | if (setlocale_failure) { | |
362 | char *p; | |
0bd48802 | 363 | const bool locwarn = (printwarn > 1 || |
98994639 HS |
364 | (printwarn && |
365 | (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p)))); | |
366 | ||
367 | if (locwarn) { | |
368 | #ifdef LC_ALL | |
369 | ||
370 | PerlIO_printf(Perl_error_log, | |
371 | "perl: warning: Setting locale failed.\n"); | |
372 | ||
373 | #else /* !LC_ALL */ | |
374 | ||
375 | PerlIO_printf(Perl_error_log, | |
376 | "perl: warning: Setting locale failed for the categories:\n\t"); | |
377 | #ifdef USE_LOCALE_CTYPE | |
378 | if (! curctype) | |
379 | PerlIO_printf(Perl_error_log, "LC_CTYPE "); | |
380 | #endif /* USE_LOCALE_CTYPE */ | |
381 | #ifdef USE_LOCALE_COLLATE | |
382 | if (! curcoll) | |
383 | PerlIO_printf(Perl_error_log, "LC_COLLATE "); | |
384 | #endif /* USE_LOCALE_COLLATE */ | |
385 | #ifdef USE_LOCALE_NUMERIC | |
386 | if (! curnum) | |
387 | PerlIO_printf(Perl_error_log, "LC_NUMERIC "); | |
388 | #endif /* USE_LOCALE_NUMERIC */ | |
389 | PerlIO_printf(Perl_error_log, "\n"); | |
390 | ||
391 | #endif /* LC_ALL */ | |
392 | ||
393 | PerlIO_printf(Perl_error_log, | |
394 | "perl: warning: Please check that your locale settings:\n"); | |
395 | ||
396 | #ifdef __GLIBC__ | |
397 | PerlIO_printf(Perl_error_log, | |
398 | "\tLANGUAGE = %c%s%c,\n", | |
399 | language ? '"' : '(', | |
400 | language ? language : "unset", | |
401 | language ? '"' : ')'); | |
402 | #endif | |
403 | ||
404 | PerlIO_printf(Perl_error_log, | |
405 | "\tLC_ALL = %c%s%c,\n", | |
406 | lc_all ? '"' : '(', | |
407 | lc_all ? lc_all : "unset", | |
408 | lc_all ? '"' : ')'); | |
409 | ||
410 | #if defined(USE_ENVIRON_ARRAY) | |
411 | { | |
412 | char **e; | |
413 | for (e = environ; *e; e++) { | |
414 | if (strnEQ(*e, "LC_", 3) | |
415 | && strnNE(*e, "LC_ALL=", 7) | |
416 | && (p = strchr(*e, '='))) | |
417 | PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n", | |
418 | (int)(p - *e), *e, p + 1); | |
419 | } | |
420 | } | |
421 | #else | |
422 | PerlIO_printf(Perl_error_log, | |
423 | "\t(possibly more locale environment variables)\n"); | |
424 | #endif | |
425 | ||
426 | PerlIO_printf(Perl_error_log, | |
427 | "\tLANG = %c%s%c\n", | |
428 | lang ? '"' : '(', | |
429 | lang ? lang : "unset", | |
430 | lang ? '"' : ')'); | |
431 | ||
432 | PerlIO_printf(Perl_error_log, | |
433 | " are supported and installed on your system.\n"); | |
434 | } | |
435 | ||
436 | #ifdef LC_ALL | |
437 | ||
438 | if (setlocale(LC_ALL, "C")) { | |
439 | if (locwarn) | |
440 | PerlIO_printf(Perl_error_log, | |
441 | "perl: warning: Falling back to the standard locale (\"C\").\n"); | |
442 | ok = 0; | |
443 | } | |
444 | else { | |
445 | if (locwarn) | |
446 | PerlIO_printf(Perl_error_log, | |
447 | "perl: warning: Failed to fall back to the standard locale (\"C\").\n"); | |
448 | ok = -1; | |
449 | } | |
450 | ||
451 | #else /* ! LC_ALL */ | |
452 | ||
453 | if (0 | |
454 | #ifdef USE_LOCALE_CTYPE | |
455 | || !(curctype || setlocale(LC_CTYPE, "C")) | |
456 | #endif /* USE_LOCALE_CTYPE */ | |
457 | #ifdef USE_LOCALE_COLLATE | |
458 | || !(curcoll || setlocale(LC_COLLATE, "C")) | |
459 | #endif /* USE_LOCALE_COLLATE */ | |
460 | #ifdef USE_LOCALE_NUMERIC | |
461 | || !(curnum || setlocale(LC_NUMERIC, "C")) | |
462 | #endif /* USE_LOCALE_NUMERIC */ | |
463 | ) | |
464 | { | |
465 | if (locwarn) | |
466 | PerlIO_printf(Perl_error_log, | |
467 | "perl: warning: Cannot fall back to the standard locale (\"C\").\n"); | |
468 | ok = -1; | |
469 | } | |
470 | ||
471 | #endif /* ! LC_ALL */ | |
472 | ||
473 | #ifdef USE_LOCALE_CTYPE | |
56279a21 | 474 | Safefree(curctype); |
bd61b366 | 475 | curctype = savepv(setlocale(LC_CTYPE, NULL)); |
98994639 HS |
476 | #endif /* USE_LOCALE_CTYPE */ |
477 | #ifdef USE_LOCALE_COLLATE | |
56279a21 | 478 | Safefree(curcoll); |
bd61b366 | 479 | curcoll = savepv(setlocale(LC_COLLATE, NULL)); |
98994639 HS |
480 | #endif /* USE_LOCALE_COLLATE */ |
481 | #ifdef USE_LOCALE_NUMERIC | |
56279a21 | 482 | Safefree(curnum); |
bd61b366 | 483 | curnum = savepv(setlocale(LC_NUMERIC, NULL)); |
98994639 HS |
484 | #endif /* USE_LOCALE_NUMERIC */ |
485 | } | |
486 | else { | |
487 | ||
488 | #ifdef USE_LOCALE_CTYPE | |
489 | new_ctype(curctype); | |
490 | #endif /* USE_LOCALE_CTYPE */ | |
491 | ||
492 | #ifdef USE_LOCALE_COLLATE | |
493 | new_collate(curcoll); | |
494 | #endif /* USE_LOCALE_COLLATE */ | |
495 | ||
496 | #ifdef USE_LOCALE_NUMERIC | |
497 | new_numeric(curnum); | |
498 | #endif /* USE_LOCALE_NUMERIC */ | |
b310b053 | 499 | |
98994639 HS |
500 | } |
501 | ||
502 | #endif /* USE_LOCALE */ | |
503 | ||
ec71e770 | 504 | #ifdef USE_PERLIO |
b310b053 | 505 | { |
fde18df1 | 506 | /* Set PL_utf8locale to TRUE if using PerlIO _and_ |
ec71e770 | 507 | any of the following are true: |
085a54d9 | 508 | - nl_langinfo(CODESET) contains /^utf-?8/i |
61de9fb5 | 509 | - $ENV{LC_ALL} contains /^utf-?8/i |
085a54d9 | 510 | - $ENV{LC_CTYPE} contains /^utf-?8/i |
61de9fb5 JH |
511 | - $ENV{LANG} contains /^utf-?8/i |
512 | The LC_ALL, LC_CTYPE, LANG obey the usual override | |
513 | hierarchy of locale environment variables. (LANGUAGE | |
514 | affects only LC_MESSAGES only under glibc.) (If present, | |
515 | it overrides LC_MESSAGES for GNU gettext, and it also | |
516 | can have more than one locale, separated by spaces, | |
517 | in case you need to know.) | |
8aa8f774 | 518 | If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE}) |
a05d7ebb | 519 | are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer |
fde18df1 | 520 | on STDIN, STDOUT, STDERR, _and_ the default open discipline. |
085a54d9 | 521 | */ |
fde18df1 | 522 | bool utf8locale = FALSE; |
b310b053 JH |
523 | char *codeset = NULL; |
524 | #if defined(HAS_NL_LANGINFO) && defined(CODESET) | |
525 | codeset = nl_langinfo(CODESET); | |
526 | #endif | |
61de9fb5 | 527 | if (codeset) |
e6226b18 KW |
528 | utf8locale = (foldEQ(codeset, STR_WITH_LEN("UTF-8")) |
529 | || foldEQ(codeset, STR_WITH_LEN("UTF8") )); | |
077a72a9 | 530 | #if defined(USE_LOCALE) |
61de9fb5 JH |
531 | else { /* nl_langinfo(CODESET) is supposed to correctly |
532 | * interpret the locale environment variables, | |
533 | * but just in case it fails, let's do this manually. */ | |
534 | if (lang) | |
e6226b18 KW |
535 | utf8locale = (foldEQ(lang, STR_WITH_LEN("UTF-8")) |
536 | || foldEQ(lang, STR_WITH_LEN("UTF8") )); | |
b310b053 | 537 | #ifdef USE_LOCALE_CTYPE |
61de9fb5 | 538 | if (curctype) |
e6226b18 KW |
539 | utf8locale = (foldEQ(curctype, STR_WITH_LEN("UTF-8")) |
540 | || foldEQ(curctype, STR_WITH_LEN("UTF8") )); | |
b310b053 | 541 | #endif |
61de9fb5 | 542 | if (lc_all) |
e6226b18 KW |
543 | utf8locale = (foldEQ(lc_all, STR_WITH_LEN("UTF-8")) |
544 | || foldEQ(lc_all, STR_WITH_LEN("UTF8") )); | |
61de9fb5 | 545 | } |
fde18df1 JH |
546 | #endif /* USE_LOCALE */ |
547 | if (utf8locale) | |
548 | PL_utf8locale = TRUE; | |
549 | } | |
a05d7ebb | 550 | /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO. |
fde18df1 JH |
551 | This is an alternative to using the -C command line switch |
552 | (the -C if present will override this). */ | |
553 | { | |
dd374669 | 554 | const char *p = PerlEnv_getenv("PERL_UNICODE"); |
a05d7ebb | 555 | PL_unicode = p ? parse_unicode_opts(&p) : 0; |
5a22a2bb NC |
556 | if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG) |
557 | PL_utf8cache = -1; | |
b310b053 | 558 | } |
ec71e770 | 559 | #endif |
b310b053 | 560 | |
98994639 | 561 | #ifdef USE_LOCALE_CTYPE |
43c5f42d | 562 | Safefree(curctype); |
98994639 HS |
563 | #endif /* USE_LOCALE_CTYPE */ |
564 | #ifdef USE_LOCALE_COLLATE | |
43c5f42d | 565 | Safefree(curcoll); |
98994639 HS |
566 | #endif /* USE_LOCALE_COLLATE */ |
567 | #ifdef USE_LOCALE_NUMERIC | |
43c5f42d | 568 | Safefree(curnum); |
98994639 HS |
569 | #endif /* USE_LOCALE_NUMERIC */ |
570 | return ok; | |
571 | } | |
572 | ||
98994639 HS |
573 | #ifdef USE_LOCALE_COLLATE |
574 | ||
575 | /* | |
576 | * mem_collxfrm() is a bit like strxfrm() but with two important | |
577 | * differences. First, it handles embedded NULs. Second, it allocates | |
578 | * a bit more memory than needed for the transformed data itself. | |
579 | * The real transformed data begins at offset sizeof(collationix). | |
580 | * Please see sv_collxfrm() to see how this is used. | |
581 | */ | |
166f8a29 | 582 | |
98994639 HS |
583 | char * |
584 | Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen) | |
585 | { | |
97aff369 | 586 | dVAR; |
98994639 HS |
587 | char *xbuf; |
588 | STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */ | |
589 | ||
7918f24d NC |
590 | PERL_ARGS_ASSERT_MEM_COLLXFRM; |
591 | ||
98994639 HS |
592 | /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */ |
593 | /* the +1 is for the terminating NUL. */ | |
594 | ||
595 | xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1; | |
a02a5408 | 596 | Newx(xbuf, xAlloc, char); |
98994639 HS |
597 | if (! xbuf) |
598 | goto bad; | |
599 | ||
600 | *(U32*)xbuf = PL_collation_ix; | |
601 | xout = sizeof(PL_collation_ix); | |
602 | for (xin = 0; xin < len; ) { | |
224e8ef5 | 603 | Size_t xused; |
98994639 HS |
604 | |
605 | for (;;) { | |
606 | xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout); | |
224e8ef5 | 607 | if (xused >= PERL_INT_MAX) |
98994639 | 608 | goto bad; |
eb160463 | 609 | if ((STRLEN)xused < xAlloc - xout) |
98994639 HS |
610 | break; |
611 | xAlloc = (2 * xAlloc) + 1; | |
612 | Renew(xbuf, xAlloc, char); | |
613 | if (! xbuf) | |
614 | goto bad; | |
615 | } | |
616 | ||
617 | xin += strlen(s + xin) + 1; | |
618 | xout += xused; | |
619 | ||
620 | /* Embedded NULs are understood but silently skipped | |
621 | * because they make no sense in locale collation. */ | |
622 | } | |
623 | ||
624 | xbuf[xout] = '\0'; | |
625 | *xlen = xout - sizeof(PL_collation_ix); | |
626 | return xbuf; | |
627 | ||
628 | bad: | |
629 | Safefree(xbuf); | |
630 | *xlen = 0; | |
631 | return NULL; | |
632 | } | |
633 | ||
634 | #endif /* USE_LOCALE_COLLATE */ | |
635 | ||
66610fdd RGS |
636 | /* |
637 | * Local variables: | |
638 | * c-indentation-style: bsd | |
639 | * c-basic-offset: 4 | |
14d04a33 | 640 | * indent-tabs-mode: nil |
66610fdd RGS |
641 | * End: |
642 | * | |
14d04a33 | 643 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 644 | */ |