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 | ||
8ef6e574 KW |
37 | #ifdef USE_LOCALE |
38 | ||
98994639 HS |
39 | /* |
40 | * Standardize the locale name from a string returned by 'setlocale'. | |
41 | * | |
0ef2a2b2 | 42 | * The typical return value of setlocale() is either |
98994639 HS |
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, | |
0ef2a2b2 | 46 | * in some unspecified order). This is not handled by this function. |
98994639 HS |
47 | * |
48 | * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n", | |
0ef2a2b2 KW |
49 | * which is harmful for further use of the string in setlocale(). This |
50 | * function removes the trailing new line and everything up through the '=' | |
98994639 HS |
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 | } | |
80 | ||
8ef6e574 KW |
81 | #endif |
82 | ||
98994639 HS |
83 | void |
84 | Perl_set_numeric_radix(pTHX) | |
85 | { | |
86 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 87 | dVAR; |
98994639 | 88 | # ifdef HAS_LOCALECONV |
7452cf6a | 89 | const struct lconv* const lc = localeconv(); |
98994639 | 90 | |
98994639 HS |
91 | if (lc && lc->decimal_point) { |
92 | if (lc->decimal_point[0] == '.' && lc->decimal_point[1] == 0) { | |
93 | SvREFCNT_dec(PL_numeric_radix_sv); | |
a0714e2c | 94 | PL_numeric_radix_sv = NULL; |
98994639 HS |
95 | } |
96 | else { | |
97 | if (PL_numeric_radix_sv) | |
98 | sv_setpv(PL_numeric_radix_sv, lc->decimal_point); | |
99 | else | |
100 | PL_numeric_radix_sv = newSVpv(lc->decimal_point, 0); | |
28acfe03 KW |
101 | if (! is_ascii_string((U8 *) lc->decimal_point, 0) |
102 | && is_utf8_string((U8 *) lc->decimal_point, 0) | |
103 | && is_cur_LC_category_utf8(LC_NUMERIC)) | |
104 | { | |
105 | SvUTF8_on(PL_numeric_radix_sv); | |
106 | } | |
98994639 HS |
107 | } |
108 | } | |
109 | else | |
a0714e2c | 110 | PL_numeric_radix_sv = NULL; |
98994639 HS |
111 | # endif /* HAS_LOCALECONV */ |
112 | #endif /* USE_LOCALE_NUMERIC */ | |
113 | } | |
114 | ||
115 | /* | |
116 | * Set up for a new numeric locale. | |
117 | */ | |
118 | void | |
8772537c | 119 | Perl_new_numeric(pTHX_ const char *newnum) |
98994639 HS |
120 | { |
121 | #ifdef USE_LOCALE_NUMERIC | |
b03f34cf | 122 | char *save_newnum; |
97aff369 | 123 | dVAR; |
98994639 HS |
124 | |
125 | if (! newnum) { | |
43c5f42d NC |
126 | Safefree(PL_numeric_name); |
127 | PL_numeric_name = NULL; | |
98994639 HS |
128 | PL_numeric_standard = TRUE; |
129 | PL_numeric_local = TRUE; | |
130 | return; | |
131 | } | |
132 | ||
b03f34cf KW |
133 | save_newnum = stdize_locale(savepv(newnum)); |
134 | if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) { | |
98994639 | 135 | Safefree(PL_numeric_name); |
b03f34cf KW |
136 | PL_numeric_name = save_newnum; |
137 | PL_numeric_standard = ((*save_newnum == 'C' && save_newnum[1] == '\0') | |
138 | || strEQ(save_newnum, "POSIX")); | |
98994639 HS |
139 | PL_numeric_local = TRUE; |
140 | set_numeric_radix(); | |
141 | } | |
b03f34cf KW |
142 | else { |
143 | Safefree(save_newnum); | |
144 | } | |
98994639 HS |
145 | |
146 | #endif /* USE_LOCALE_NUMERIC */ | |
147 | } | |
148 | ||
149 | void | |
150 | Perl_set_numeric_standard(pTHX) | |
151 | { | |
152 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 153 | dVAR; |
98994639 HS |
154 | |
155 | if (! PL_numeric_standard) { | |
156 | setlocale(LC_NUMERIC, "C"); | |
157 | PL_numeric_standard = TRUE; | |
158 | PL_numeric_local = FALSE; | |
159 | set_numeric_radix(); | |
160 | } | |
161 | ||
162 | #endif /* USE_LOCALE_NUMERIC */ | |
163 | } | |
164 | ||
165 | void | |
166 | Perl_set_numeric_local(pTHX) | |
167 | { | |
168 | #ifdef USE_LOCALE_NUMERIC | |
97aff369 | 169 | dVAR; |
98994639 HS |
170 | |
171 | if (! PL_numeric_local) { | |
172 | setlocale(LC_NUMERIC, PL_numeric_name); | |
173 | PL_numeric_standard = FALSE; | |
174 | PL_numeric_local = TRUE; | |
175 | set_numeric_radix(); | |
176 | } | |
177 | ||
178 | #endif /* USE_LOCALE_NUMERIC */ | |
179 | } | |
180 | ||
181 | /* | |
182 | * Set up for a new ctype locale. | |
183 | */ | |
184 | void | |
8772537c | 185 | Perl_new_ctype(pTHX_ const char *newctype) |
98994639 HS |
186 | { |
187 | #ifdef USE_LOCALE_CTYPE | |
27da23d5 | 188 | dVAR; |
68067e4e | 189 | UV i; |
98994639 | 190 | |
7918f24d NC |
191 | PERL_ARGS_ASSERT_NEW_CTYPE; |
192 | ||
77806dea KW |
193 | for (i = 0; i < 256; i++) { |
194 | if (isUPPER_LC((U8) i)) | |
195 | PL_fold_locale[i] = toLOWER_LC((U8) i); | |
196 | else if (isLOWER_LC((U8) i)) | |
197 | PL_fold_locale[i] = toUPPER_LC((U8) i); | |
98994639 HS |
198 | else |
199 | PL_fold_locale[i] = i; | |
200 | } | |
201 | ||
202 | #endif /* USE_LOCALE_CTYPE */ | |
7918f24d | 203 | PERL_ARGS_ASSERT_NEW_CTYPE; |
8772537c | 204 | PERL_UNUSED_ARG(newctype); |
96a5add6 | 205 | PERL_UNUSED_CONTEXT; |
98994639 HS |
206 | } |
207 | ||
208 | /* | |
209 | * Set up for a new collation locale. | |
210 | */ | |
211 | void | |
8772537c | 212 | Perl_new_collate(pTHX_ const char *newcoll) |
98994639 HS |
213 | { |
214 | #ifdef USE_LOCALE_COLLATE | |
97aff369 | 215 | dVAR; |
98994639 HS |
216 | |
217 | if (! newcoll) { | |
218 | if (PL_collation_name) { | |
219 | ++PL_collation_ix; | |
220 | Safefree(PL_collation_name); | |
221 | PL_collation_name = NULL; | |
222 | } | |
223 | PL_collation_standard = TRUE; | |
224 | PL_collxfrm_base = 0; | |
225 | PL_collxfrm_mult = 2; | |
226 | return; | |
227 | } | |
228 | ||
229 | if (! PL_collation_name || strNE(PL_collation_name, newcoll)) { | |
230 | ++PL_collation_ix; | |
231 | Safefree(PL_collation_name); | |
232 | PL_collation_name = stdize_locale(savepv(newcoll)); | |
770526c1 NC |
233 | PL_collation_standard = ((*newcoll == 'C' && newcoll[1] == '\0') |
234 | || strEQ(newcoll, "POSIX")); | |
98994639 HS |
235 | |
236 | { | |
237 | /* 2: at most so many chars ('a', 'b'). */ | |
238 | /* 50: surely no system expands a char more. */ | |
239 | #define XFRMBUFSIZE (2 * 50) | |
240 | char xbuf[XFRMBUFSIZE]; | |
8772537c AL |
241 | const Size_t fa = strxfrm(xbuf, "a", XFRMBUFSIZE); |
242 | const Size_t fb = strxfrm(xbuf, "ab", XFRMBUFSIZE); | |
243 | const SSize_t mult = fb - fa; | |
e0601d3c | 244 | if (mult < 1 && !(fa == 0 && fb == 0)) |
ad49ad39 NC |
245 | Perl_croak(aTHX_ "panic: strxfrm() gets absurd - a => %"UVuf", ab => %"UVuf, |
246 | (UV) fa, (UV) fb); | |
eb160463 | 247 | PL_collxfrm_base = (fa > (Size_t)mult) ? (fa - mult) : 0; |
98994639 HS |
248 | PL_collxfrm_mult = mult; |
249 | } | |
250 | } | |
251 | ||
252 | #endif /* USE_LOCALE_COLLATE */ | |
253 | } | |
254 | ||
255 | /* | |
256 | * Initialize locale awareness. | |
257 | */ | |
258 | int | |
259 | Perl_init_i18nl10n(pTHX_ int printwarn) | |
260 | { | |
261 | int ok = 1; | |
262 | /* returns | |
263 | * 1 = set ok or not applicable, | |
264 | * 0 = fallback to C locale, | |
265 | * -1 = fallback to C locale failed | |
266 | */ | |
267 | ||
268 | #if defined(USE_LOCALE) | |
97aff369 | 269 | dVAR; |
98994639 HS |
270 | |
271 | #ifdef USE_LOCALE_CTYPE | |
272 | char *curctype = NULL; | |
273 | #endif /* USE_LOCALE_CTYPE */ | |
274 | #ifdef USE_LOCALE_COLLATE | |
275 | char *curcoll = NULL; | |
276 | #endif /* USE_LOCALE_COLLATE */ | |
277 | #ifdef USE_LOCALE_NUMERIC | |
278 | char *curnum = NULL; | |
279 | #endif /* USE_LOCALE_NUMERIC */ | |
280 | #ifdef __GLIBC__ | |
7452cf6a | 281 | char * const language = PerlEnv_getenv("LANGUAGE"); |
98994639 | 282 | #endif |
ccd65d51 KW |
283 | /* NULL uses the existing already set up locale */ |
284 | const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT")) | |
285 | ? NULL | |
286 | : ""; | |
7452cf6a AL |
287 | char * const lc_all = PerlEnv_getenv("LC_ALL"); |
288 | char * const lang = PerlEnv_getenv("LANG"); | |
98994639 HS |
289 | bool setlocale_failure = FALSE; |
290 | ||
291 | #ifdef LOCALE_ENVIRON_REQUIRED | |
292 | ||
293 | /* | |
294 | * Ultrix setlocale(..., "") fails if there are no environment | |
295 | * variables from which to get a locale name. | |
296 | */ | |
297 | ||
298 | bool done = FALSE; | |
299 | ||
b3e384bf | 300 | # ifdef LC_ALL |
98994639 | 301 | if (lang) { |
ccd65d51 | 302 | if (setlocale(LC_ALL, setlocale_init)) |
98994639 HS |
303 | done = TRUE; |
304 | else | |
305 | setlocale_failure = TRUE; | |
306 | } | |
307 | if (!setlocale_failure) { | |
b3e384bf | 308 | # ifdef USE_LOCALE_CTYPE |
56279a21 | 309 | Safefree(curctype); |
98994639 HS |
310 | if (! (curctype = |
311 | setlocale(LC_CTYPE, | |
312 | (!done && (lang || PerlEnv_getenv("LC_CTYPE"))) | |
ccd65d51 | 313 | ? setlocale_init : NULL))) |
98994639 HS |
314 | setlocale_failure = TRUE; |
315 | else | |
316 | curctype = savepv(curctype); | |
b3e384bf KW |
317 | # endif /* USE_LOCALE_CTYPE */ |
318 | # ifdef USE_LOCALE_COLLATE | |
56279a21 | 319 | Safefree(curcoll); |
98994639 HS |
320 | if (! (curcoll = |
321 | setlocale(LC_COLLATE, | |
322 | (!done && (lang || PerlEnv_getenv("LC_COLLATE"))) | |
ccd65d51 | 323 | ? setlocale_init : NULL))) |
98994639 HS |
324 | setlocale_failure = TRUE; |
325 | else | |
326 | curcoll = savepv(curcoll); | |
b3e384bf KW |
327 | # endif /* USE_LOCALE_COLLATE */ |
328 | # ifdef USE_LOCALE_NUMERIC | |
56279a21 | 329 | Safefree(curnum); |
98994639 HS |
330 | if (! (curnum = |
331 | setlocale(LC_NUMERIC, | |
332 | (!done && (lang || PerlEnv_getenv("LC_NUMERIC"))) | |
ccd65d51 | 333 | ? setlocale_init : NULL))) |
98994639 HS |
334 | setlocale_failure = TRUE; |
335 | else | |
336 | curnum = savepv(curnum); | |
b3e384bf | 337 | # endif /* USE_LOCALE_NUMERIC */ |
98994639 HS |
338 | } |
339 | ||
b3e384bf | 340 | # endif /* LC_ALL */ |
98994639 HS |
341 | |
342 | #endif /* !LOCALE_ENVIRON_REQUIRED */ | |
343 | ||
344 | #ifdef LC_ALL | |
ccd65d51 | 345 | if (! setlocale(LC_ALL, setlocale_init)) |
98994639 HS |
346 | setlocale_failure = TRUE; |
347 | #endif /* LC_ALL */ | |
348 | ||
349 | if (!setlocale_failure) { | |
350 | #ifdef USE_LOCALE_CTYPE | |
6cb43dbf | 351 | Safefree(curctype); |
ccd65d51 | 352 | if (! (curctype = setlocale(LC_CTYPE, setlocale_init))) |
98994639 HS |
353 | setlocale_failure = TRUE; |
354 | else | |
355 | curctype = savepv(curctype); | |
356 | #endif /* USE_LOCALE_CTYPE */ | |
357 | #ifdef USE_LOCALE_COLLATE | |
6cb43dbf | 358 | Safefree(curcoll); |
ccd65d51 | 359 | if (! (curcoll = setlocale(LC_COLLATE, setlocale_init))) |
98994639 HS |
360 | setlocale_failure = TRUE; |
361 | else | |
362 | curcoll = savepv(curcoll); | |
363 | #endif /* USE_LOCALE_COLLATE */ | |
364 | #ifdef USE_LOCALE_NUMERIC | |
6cb43dbf | 365 | Safefree(curnum); |
ccd65d51 | 366 | if (! (curnum = setlocale(LC_NUMERIC, setlocale_init))) |
98994639 HS |
367 | setlocale_failure = TRUE; |
368 | else | |
369 | curnum = savepv(curnum); | |
370 | #endif /* USE_LOCALE_NUMERIC */ | |
371 | } | |
372 | ||
373 | if (setlocale_failure) { | |
374 | char *p; | |
0bd48802 | 375 | const bool locwarn = (printwarn > 1 || |
98994639 HS |
376 | (printwarn && |
377 | (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p)))); | |
378 | ||
379 | if (locwarn) { | |
380 | #ifdef LC_ALL | |
381 | ||
382 | PerlIO_printf(Perl_error_log, | |
383 | "perl: warning: Setting locale failed.\n"); | |
384 | ||
385 | #else /* !LC_ALL */ | |
386 | ||
387 | PerlIO_printf(Perl_error_log, | |
388 | "perl: warning: Setting locale failed for the categories:\n\t"); | |
389 | #ifdef USE_LOCALE_CTYPE | |
390 | if (! curctype) | |
391 | PerlIO_printf(Perl_error_log, "LC_CTYPE "); | |
392 | #endif /* USE_LOCALE_CTYPE */ | |
393 | #ifdef USE_LOCALE_COLLATE | |
394 | if (! curcoll) | |
395 | PerlIO_printf(Perl_error_log, "LC_COLLATE "); | |
396 | #endif /* USE_LOCALE_COLLATE */ | |
397 | #ifdef USE_LOCALE_NUMERIC | |
398 | if (! curnum) | |
399 | PerlIO_printf(Perl_error_log, "LC_NUMERIC "); | |
400 | #endif /* USE_LOCALE_NUMERIC */ | |
401 | PerlIO_printf(Perl_error_log, "\n"); | |
402 | ||
403 | #endif /* LC_ALL */ | |
404 | ||
405 | PerlIO_printf(Perl_error_log, | |
406 | "perl: warning: Please check that your locale settings:\n"); | |
407 | ||
408 | #ifdef __GLIBC__ | |
409 | PerlIO_printf(Perl_error_log, | |
410 | "\tLANGUAGE = %c%s%c,\n", | |
411 | language ? '"' : '(', | |
412 | language ? language : "unset", | |
413 | language ? '"' : ')'); | |
414 | #endif | |
415 | ||
416 | PerlIO_printf(Perl_error_log, | |
417 | "\tLC_ALL = %c%s%c,\n", | |
418 | lc_all ? '"' : '(', | |
419 | lc_all ? lc_all : "unset", | |
420 | lc_all ? '"' : ')'); | |
421 | ||
422 | #if defined(USE_ENVIRON_ARRAY) | |
423 | { | |
424 | char **e; | |
425 | for (e = environ; *e; e++) { | |
426 | if (strnEQ(*e, "LC_", 3) | |
427 | && strnNE(*e, "LC_ALL=", 7) | |
428 | && (p = strchr(*e, '='))) | |
429 | PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n", | |
430 | (int)(p - *e), *e, p + 1); | |
431 | } | |
432 | } | |
433 | #else | |
434 | PerlIO_printf(Perl_error_log, | |
435 | "\t(possibly more locale environment variables)\n"); | |
436 | #endif | |
437 | ||
438 | PerlIO_printf(Perl_error_log, | |
439 | "\tLANG = %c%s%c\n", | |
440 | lang ? '"' : '(', | |
441 | lang ? lang : "unset", | |
442 | lang ? '"' : ')'); | |
443 | ||
444 | PerlIO_printf(Perl_error_log, | |
445 | " are supported and installed on your system.\n"); | |
446 | } | |
447 | ||
448 | #ifdef LC_ALL | |
449 | ||
450 | if (setlocale(LC_ALL, "C")) { | |
451 | if (locwarn) | |
452 | PerlIO_printf(Perl_error_log, | |
453 | "perl: warning: Falling back to the standard locale (\"C\").\n"); | |
454 | ok = 0; | |
455 | } | |
456 | else { | |
457 | if (locwarn) | |
458 | PerlIO_printf(Perl_error_log, | |
459 | "perl: warning: Failed to fall back to the standard locale (\"C\").\n"); | |
460 | ok = -1; | |
461 | } | |
462 | ||
463 | #else /* ! LC_ALL */ | |
464 | ||
465 | if (0 | |
466 | #ifdef USE_LOCALE_CTYPE | |
467 | || !(curctype || setlocale(LC_CTYPE, "C")) | |
468 | #endif /* USE_LOCALE_CTYPE */ | |
469 | #ifdef USE_LOCALE_COLLATE | |
470 | || !(curcoll || setlocale(LC_COLLATE, "C")) | |
471 | #endif /* USE_LOCALE_COLLATE */ | |
472 | #ifdef USE_LOCALE_NUMERIC | |
473 | || !(curnum || setlocale(LC_NUMERIC, "C")) | |
474 | #endif /* USE_LOCALE_NUMERIC */ | |
475 | ) | |
476 | { | |
477 | if (locwarn) | |
478 | PerlIO_printf(Perl_error_log, | |
479 | "perl: warning: Cannot fall back to the standard locale (\"C\").\n"); | |
480 | ok = -1; | |
481 | } | |
482 | ||
483 | #endif /* ! LC_ALL */ | |
484 | ||
485 | #ifdef USE_LOCALE_CTYPE | |
56279a21 | 486 | Safefree(curctype); |
bd61b366 | 487 | curctype = savepv(setlocale(LC_CTYPE, NULL)); |
98994639 HS |
488 | #endif /* USE_LOCALE_CTYPE */ |
489 | #ifdef USE_LOCALE_COLLATE | |
56279a21 | 490 | Safefree(curcoll); |
bd61b366 | 491 | curcoll = savepv(setlocale(LC_COLLATE, NULL)); |
98994639 HS |
492 | #endif /* USE_LOCALE_COLLATE */ |
493 | #ifdef USE_LOCALE_NUMERIC | |
56279a21 | 494 | Safefree(curnum); |
bd61b366 | 495 | curnum = savepv(setlocale(LC_NUMERIC, NULL)); |
98994639 HS |
496 | #endif /* USE_LOCALE_NUMERIC */ |
497 | } | |
498 | else { | |
499 | ||
500 | #ifdef USE_LOCALE_CTYPE | |
501 | new_ctype(curctype); | |
502 | #endif /* USE_LOCALE_CTYPE */ | |
503 | ||
504 | #ifdef USE_LOCALE_COLLATE | |
505 | new_collate(curcoll); | |
506 | #endif /* USE_LOCALE_COLLATE */ | |
507 | ||
508 | #ifdef USE_LOCALE_NUMERIC | |
509 | new_numeric(curnum); | |
510 | #endif /* USE_LOCALE_NUMERIC */ | |
b310b053 | 511 | |
98994639 HS |
512 | } |
513 | ||
8ef6e574 | 514 | #if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE) |
b310b053 | 515 | { |
fde18df1 | 516 | /* Set PL_utf8locale to TRUE if using PerlIO _and_ |
7d74bb61 | 517 | the current LC_CTYPE locale is UTF-8. |
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 | */ |
7d74bb61 | 522 | PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE); |
fde18df1 | 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; |
5a22a2bb NC |
530 | if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG) |
531 | PL_utf8cache = -1; | |
b310b053 | 532 | } |
ec71e770 | 533 | #endif |
b310b053 | 534 | |
98994639 | 535 | #ifdef USE_LOCALE_CTYPE |
43c5f42d | 536 | Safefree(curctype); |
98994639 HS |
537 | #endif /* USE_LOCALE_CTYPE */ |
538 | #ifdef USE_LOCALE_COLLATE | |
43c5f42d | 539 | Safefree(curcoll); |
98994639 HS |
540 | #endif /* USE_LOCALE_COLLATE */ |
541 | #ifdef USE_LOCALE_NUMERIC | |
43c5f42d | 542 | Safefree(curnum); |
98994639 | 543 | #endif /* USE_LOCALE_NUMERIC */ |
8ef6e574 KW |
544 | |
545 | #endif /* USE_LOCALE */ | |
546 | ||
98994639 HS |
547 | return ok; |
548 | } | |
549 | ||
98994639 HS |
550 | #ifdef USE_LOCALE_COLLATE |
551 | ||
552 | /* | |
553 | * mem_collxfrm() is a bit like strxfrm() but with two important | |
554 | * differences. First, it handles embedded NULs. Second, it allocates | |
555 | * a bit more memory than needed for the transformed data itself. | |
556 | * The real transformed data begins at offset sizeof(collationix). | |
557 | * Please see sv_collxfrm() to see how this is used. | |
558 | */ | |
166f8a29 | 559 | |
98994639 HS |
560 | char * |
561 | Perl_mem_collxfrm(pTHX_ const char *s, STRLEN len, STRLEN *xlen) | |
562 | { | |
97aff369 | 563 | dVAR; |
98994639 HS |
564 | char *xbuf; |
565 | STRLEN xAlloc, xin, xout; /* xalloc is a reserved word in VC */ | |
566 | ||
7918f24d NC |
567 | PERL_ARGS_ASSERT_MEM_COLLXFRM; |
568 | ||
98994639 HS |
569 | /* the first sizeof(collationix) bytes are used by sv_collxfrm(). */ |
570 | /* the +1 is for the terminating NUL. */ | |
571 | ||
572 | xAlloc = sizeof(PL_collation_ix) + PL_collxfrm_base + (PL_collxfrm_mult * len) + 1; | |
a02a5408 | 573 | Newx(xbuf, xAlloc, char); |
98994639 HS |
574 | if (! xbuf) |
575 | goto bad; | |
576 | ||
577 | *(U32*)xbuf = PL_collation_ix; | |
578 | xout = sizeof(PL_collation_ix); | |
579 | for (xin = 0; xin < len; ) { | |
224e8ef5 | 580 | Size_t xused; |
98994639 HS |
581 | |
582 | for (;;) { | |
583 | xused = strxfrm(xbuf + xout, s + xin, xAlloc - xout); | |
224e8ef5 | 584 | if (xused >= PERL_INT_MAX) |
98994639 | 585 | goto bad; |
eb160463 | 586 | if ((STRLEN)xused < xAlloc - xout) |
98994639 HS |
587 | break; |
588 | xAlloc = (2 * xAlloc) + 1; | |
589 | Renew(xbuf, xAlloc, char); | |
590 | if (! xbuf) | |
591 | goto bad; | |
592 | } | |
593 | ||
594 | xin += strlen(s + xin) + 1; | |
595 | xout += xused; | |
596 | ||
597 | /* Embedded NULs are understood but silently skipped | |
598 | * because they make no sense in locale collation. */ | |
599 | } | |
600 | ||
601 | xbuf[xout] = '\0'; | |
602 | *xlen = xout - sizeof(PL_collation_ix); | |
603 | return xbuf; | |
604 | ||
605 | bad: | |
606 | Safefree(xbuf); | |
607 | *xlen = 0; | |
608 | return NULL; | |
609 | } | |
610 | ||
611 | #endif /* USE_LOCALE_COLLATE */ | |
612 | ||
8ef6e574 KW |
613 | #ifdef USE_LOCALE |
614 | ||
637917bb | 615 | STATIC bool |
7d74bb61 KW |
616 | S_is_cur_LC_category_utf8(pTHX_ int category) |
617 | { | |
618 | /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE | |
619 | * otherwise. 'category' may not be LC_ALL. If the platform doesn't have | |
620 | * nl_langinfo(), this employs a heuristic, which hence could give the | |
621 | * wrong result. It errs on the side of not being a UTF-8 locale. */ | |
622 | ||
623 | char *save_input_locale = NULL; | |
7d74bb61 KW |
624 | STRLEN final_pos; |
625 | ||
8ef6e574 | 626 | #ifdef LC_ALL |
7d74bb61 | 627 | assert(category != LC_ALL); |
8ef6e574 | 628 | #endif |
7d74bb61 KW |
629 | |
630 | /* First dispose of the trivial cases */ | |
631 | save_input_locale = stdize_locale(setlocale(category, NULL)); | |
632 | if (! save_input_locale) { | |
633 | return FALSE; /* XXX maybe should croak */ | |
634 | } | |
635 | if ((*save_input_locale == 'C' && save_input_locale[1] == '\0') | |
636 | || strEQ(save_input_locale, "POSIX")) | |
637 | { | |
638 | return FALSE; | |
639 | } | |
640 | ||
641 | save_input_locale = savepv(save_input_locale); | |
642 | ||
643 | #if defined(HAS_NL_LANGINFO) && defined(CODESET) && defined(USE_LOCALE_CTYPE) | |
644 | ||
645 | { /* Next try nl_langinfo if available */ | |
646 | ||
647 | char *save_ctype_locale = NULL; | |
648 | char *codeset = NULL; | |
649 | ||
650 | if (category != LC_CTYPE) { /* nl_langinfo works only on LC_CTYPE */ | |
651 | ||
652 | /* Get the current LC_CTYPE locale */ | |
653 | save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL))); | |
654 | if (! save_ctype_locale) { | |
655 | goto cant_use_nllanginfo; | |
656 | } | |
657 | ||
658 | /* If LC_CTYPE and the desired category use the same locale, this | |
659 | * means that finding the value for LC_CTYPE is the same as finding | |
660 | * the value for the desired category. Otherwise, switch LC_CTYPE | |
661 | * to the desired category's locale */ | |
662 | if (strEQ(save_ctype_locale, save_input_locale)) { | |
663 | Safefree(save_ctype_locale); | |
664 | save_ctype_locale = NULL; | |
665 | } | |
666 | else if (! setlocale(LC_CTYPE, save_input_locale)) { | |
667 | Safefree(save_ctype_locale); | |
668 | goto cant_use_nllanginfo; | |
669 | } | |
670 | } | |
671 | ||
672 | /* Here the current LC_CTYPE is set to the locale of the category whose | |
673 | * information is desired. This means that nl_langinfo() should give | |
674 | * the correct results */ | |
675 | codeset = savepv(nl_langinfo(CODESET)); | |
676 | if (codeset) { | |
677 | bool is_utf8; | |
678 | ||
679 | /* If we switched LC_CTYPE, switch back */ | |
680 | if (save_ctype_locale) { | |
681 | setlocale(LC_CTYPE, save_ctype_locale); | |
682 | Safefree(save_ctype_locale); | |
683 | } | |
684 | ||
685 | is_utf8 = foldEQ(codeset, STR_WITH_LEN("UTF-8")) | |
686 | || foldEQ(codeset, STR_WITH_LEN("UTF8")); | |
687 | ||
688 | Safefree(codeset); | |
689 | Safefree(save_input_locale); | |
690 | return is_utf8; | |
691 | } | |
692 | ||
693 | } | |
694 | cant_use_nllanginfo: | |
695 | ||
696 | #endif /* HAS_NL_LANGINFO etc */ | |
697 | ||
698 | /* nl_langinfo not available or failed somehow. Look at the locale name to | |
751e9426 | 699 | * see if it matches qr/UTF -? 8 /ix */ |
7d74bb61 KW |
700 | |
701 | final_pos = strlen(save_input_locale) - 1; | |
751e9426 KW |
702 | if (final_pos >= 3) { |
703 | char *name = save_input_locale; | |
704 | ||
705 | /* Find next 'U' or 'u' and look from there */ | |
706 | while ((name += strcspn(name, "Uu") + 1) | |
707 | <= save_input_locale + final_pos - 2) | |
7d74bb61 | 708 | { |
751e9426 KW |
709 | if (toFOLD(*(name)) != 't' |
710 | || toFOLD(*(name + 1)) != 'f') | |
711 | { | |
712 | continue; | |
713 | } | |
714 | name += 2; | |
715 | if (*(name) == '-') { | |
716 | if ((name > save_input_locale + final_pos - 1)) { | |
717 | break; | |
718 | } | |
719 | name++; | |
720 | } | |
721 | if (*(name) == '8') { | |
722 | Safefree(save_input_locale); | |
723 | return TRUE; | |
724 | } | |
7d74bb61 KW |
725 | } |
726 | } | |
727 | ||
728 | #ifdef WIN32 | |
729 | /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */ | |
730 | if (final_pos >= 4 | |
731 | && *(save_input_locale + final_pos - 0) == '1' | |
732 | && *(save_input_locale + final_pos - 1) == '0' | |
733 | && *(save_input_locale + final_pos - 2) == '0' | |
734 | && *(save_input_locale + final_pos - 3) == '5' | |
735 | && *(save_input_locale + final_pos - 4) == '6') | |
736 | { | |
737 | Safefree(save_input_locale); | |
738 | return TRUE; | |
739 | } | |
740 | #endif | |
741 | ||
fa9b773e KW |
742 | /* Other common encodings are the ISO 8859 series, which aren't UTF-8 */ |
743 | if (instr(save_input_locale, "8859")) { | |
744 | Safefree(save_input_locale); | |
745 | return FALSE; | |
746 | } | |
747 | ||
748 | #ifdef HAS_LOCALECONV | |
749 | ||
750 | # ifdef USE_LOCALE_MONETARY | |
751 | ||
752 | /* Here, there is nothing in the locale name to indicate whether the locale | |
753 | * is UTF-8 or not. This "name", the return of setlocale(), is actually | |
754 | * defined to be opaque, so we can't really rely on the absence of various | |
755 | * substrings in the name to indicate its UTF-8ness. Look at the locale's | |
756 | * currency symbol. Often that will be in the native script, and if the | |
757 | * symbol isn't in UTF-8, we know that the locale isn't. If it is | |
758 | * non-ASCII UTF-8, we infer that the locale is too. | |
759 | * To do this, like above for LC_CTYPE, we first set LC_MONETARY to the | |
760 | * locale of the desired category, if it isn't that locale already */ | |
761 | ||
762 | { | |
763 | char *save_monetary_locale = NULL; | |
764 | bool illegal_utf8 = FALSE; | |
765 | bool only_ascii = FALSE; | |
766 | const struct lconv* const lc = localeconv(); | |
767 | ||
768 | if (category != LC_MONETARY) { | |
769 | ||
770 | save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY, | |
771 | NULL))); | |
772 | if (! save_monetary_locale) { | |
773 | goto cant_use_monetary; | |
774 | } | |
775 | ||
776 | if (strNE(save_monetary_locale, save_input_locale)) { | |
777 | if (! setlocale(LC_MONETARY, save_input_locale)) { | |
778 | Safefree(save_monetary_locale); | |
779 | goto cant_use_monetary; | |
780 | } | |
781 | } | |
782 | } | |
783 | ||
784 | /* Here the current LC_MONETARY is set to the locale of the category | |
785 | * whose information is desired. */ | |
786 | ||
787 | if (lc && lc->currency_symbol) { | |
788 | if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) { | |
789 | illegal_utf8 = TRUE; | |
790 | } | |
791 | else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) { | |
792 | only_ascii = TRUE; | |
793 | } | |
794 | } | |
795 | ||
796 | /* If we changed it, restore LC_MONETARY to its original locale */ | |
797 | if (save_monetary_locale) { | |
798 | setlocale(LC_MONETARY, save_monetary_locale); | |
799 | Safefree(save_monetary_locale); | |
800 | } | |
801 | ||
802 | Safefree(save_input_locale); | |
803 | ||
804 | /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise | |
805 | * assume the locale is UTF-8 if and only if the symbol is non-ascii | |
806 | * UTF-8. (We can't really tell if the locale is UTF-8 or not if the | |
807 | * symbol is just a '$', so we err on the side of it not being UTF-8) | |
808 | * */ | |
809 | return (illegal_utf8) | |
810 | ? FALSE | |
811 | : ! only_ascii; | |
812 | ||
813 | } | |
814 | cant_use_monetary: | |
815 | ||
816 | # endif /* USE_LOCALE_MONETARY */ | |
817 | #endif /* HAS_LOCALECONV */ | |
818 | ||
819 | #if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES) | |
820 | ||
821 | /* This code is ifdefd out because it was found to not be necessary in testing | |
822 | * on our dromedary test machine, which has over 700 locales. There, looking | |
823 | * at just the currency symbol gave essentially the same results as doing this | |
824 | * extra work. Executing this also caused segfaults in miniperl. I left it in | |
825 | * so as to avoid rewriting it if real-world experience indicates that | |
826 | * dromedary is an outlier. Essentially, instead of returning abpve if we | |
827 | * haven't found illegal utf8, we continue on and examine all the strerror() | |
828 | * messages on the platform for utf8ness. If all are ASCII, we still don't | |
829 | * know the answer; but otherwise we have a pretty good indication of the | |
830 | * utf8ness. The reason this doesn't necessarily help much is that the | |
831 | * messages may not have been translated into the locale. The currency symbol | |
832 | * is much more likely to have been translated. The code below would need to | |
833 | * be altered somewhat to just be a continuation of testing the currency | |
834 | * symbol. */ | |
835 | int e; | |
836 | unsigned int failures = 0, non_ascii = 0; | |
837 | char *save_messages_locale = NULL; | |
838 | ||
839 | /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the | |
840 | * desired category, if it isn't that locale already */ | |
841 | ||
842 | if (category != LC_MESSAGES) { | |
843 | ||
844 | save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES, | |
845 | NULL))); | |
846 | if (! save_messages_locale) { | |
847 | goto cant_use_messages; | |
848 | } | |
849 | ||
850 | if (strEQ(save_messages_locale, save_input_locale)) { | |
851 | Safefree(save_input_locale); | |
852 | } | |
853 | else if (! setlocale(LC_MESSAGES, save_input_locale)) { | |
854 | Safefree(save_messages_locale); | |
855 | goto cant_use_messages; | |
856 | } | |
857 | } | |
858 | ||
859 | /* Here the current LC_MESSAGES is set to the locale of the category | |
860 | * whose information is desired. Look through all the messages */ | |
861 | ||
862 | for (e = 0; | |
863 | #ifdef HAS_SYS_ERRLIST | |
864 | e <= sys_nerr | |
865 | #endif | |
866 | ; e++) | |
867 | { | |
868 | const U8* const errmsg = (U8 *) Strerror(e) ; | |
869 | if (!errmsg) | |
870 | break; | |
871 | if (! is_utf8_string(errmsg, 0)) { | |
872 | failures++; | |
873 | break; | |
874 | } | |
875 | else if (! is_ascii_string(errmsg, 0)) { | |
876 | non_ascii++; | |
877 | } | |
878 | } | |
879 | ||
880 | /* And, if we changed it, restore LC_MESSAGES to its original locale */ | |
881 | if (save_messages_locale) { | |
882 | setlocale(LC_MESSAGES, save_messages_locale); | |
883 | Safefree(save_messages_locale); | |
884 | } | |
885 | ||
886 | /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid, | |
887 | * any non-ascii means it is one; otherwise we assume it isn't */ | |
888 | return (failures) ? FALSE : non_ascii; | |
889 | ||
890 | } | |
891 | cant_use_messages: | |
892 | ||
893 | #endif | |
894 | ||
895 | Safefree(save_input_locale); | |
7d74bb61 KW |
896 | return FALSE; |
897 | } | |
898 | ||
8ef6e574 | 899 | #endif |
7d74bb61 | 900 | |
66610fdd RGS |
901 | /* |
902 | * Local variables: | |
903 | * c-indentation-style: bsd | |
904 | * c-basic-offset: 4 | |
14d04a33 | 905 | * indent-tabs-mode: nil |
66610fdd RGS |
906 | * End: |
907 | * | |
14d04a33 | 908 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 909 | */ |