3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
4 * 2002, 2003, 2005, 2006, 2007, 2008 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!
20 * [p.238 of _The Lord of the Rings_, II/i: "Many Meetings"]
23 /* utility functions for handling locale-specific stuff like what
24 * character represents the decimal point.
26 * All C programs have an underlying locale. Perl code generally doesn't pay
27 * any attention to it except within the scope of a 'use locale'. For most
28 * categories, it accomplishes this by just using different operations if it is
29 * in such scope than if not. However, various libc functions called by Perl
30 * are affected by the LC_NUMERIC category, so there are macros in perl.h that
31 * are used to toggle between the current locale and the C locale depending on
32 * the desired behavior of those functions at the moment. And, LC_MESSAGES is
33 * switched to the C locale for outputting the message unless within the scope
36 * This code now has multi-thread-safe locale handling on systems that support
37 * that. This is completely transparent to most XS code. On earlier systems,
38 * it would be possible to emulate thread-safe locales, but this likely would
39 * involve a lot of locale switching, and would require XS code changes.
40 * Macros could be written so that the code wouldn't have to know which type of
41 * system is being used.
43 * Table-driven code is used for simplicity and clarity, as many operations
44 * differ only in which category is being worked on. However the system
45 * categories need not be small contiguous integers, so do not lend themselves
46 * to table lookup. Instead we have created our own equivalent values which
47 * are all small contiguous non-negative integers, and translation functions
48 * between the two sets. For category 'LC_foo', the name of our index is
49 * LC_foo_INDEX_. Various parallel tables, indexed by these, are used.
51 * Many of the macros and functions in this file have one of the suffixes '_c',
52 * '_r', or '_i'. khw found these useful in remembering what type of locale
53 * category to use as their parameter. '_r' takes an int category number as
54 * passed to setlocale(), like LC_ALL, LC_CTYPE, etc. The 'r' indicates that
55 * the value isn't known until runtime. '_c' also indicates such a category
56 * number, but its value is known at compile time. These are both converted
57 * into unsigned indexes into various tables of category information, where the
58 * real work is generally done. The tables are generated at compile-time based
59 * on platform characteristics and Configure options. They hide from the code
60 * many of the vagaries of the different locale implementations out there. You
61 * may have already guessed that '_i' indicates the parameter is such an
62 * unsigned index. Converting from '_r' to '_i' requires run-time lookup.
63 * '_c' is used to get cpp to do this at compile time. To avoid the runtime
64 * expense, the code is structured to use '_r' at the API level, and once
65 * converted, everything possible is done using the table indexes.
67 * On unthreaded perls, most operations expand out to just the basic
68 * setlocale() calls. The same is true on threaded perls on modern Windows
69 * systems where the same API, after set up, is used for thread-safe locale
70 * handling. On other systems, there is a completely different API, specified
71 * in POSIX 2008, to do thread-safe locales. On these systems, our
72 * emulate_setlocale_i() function is used to hide the different API from the
73 * outside. This makes it completely transparent to most XS code.
75 * A huge complicating factor is that the LC_NUMERIC category is normally held
76 * in the C locale, except during those relatively rare times when it needs to
77 * be in the underlying locale. There is a bunch of code to accomplish this,
78 * and to allow easy switches from one state to the other.
80 * z/OS (os390) is an outlier. Locales really don't work under threads when
81 * either the radix character isn't a dot, or attempts are made to change
82 * locales after the first thread is created. The reason is that IBM has made
83 * it thread-safe by refusing to change locales (returning failure if
84 * attempted) any time after an application has called pthread_create() to
85 * create another thread. The expectation is that an application will set up
86 * its locale information before the first fork, and be stable thereafter. But
87 * perl toggles LC_NUMERIC if the locale's radix character isn't a dot, as do
88 * the other toggles, which are less common.
91 /* If the environment says to, we can output debugging information during
92 * initialization. This is done before option parsing, and before any thread
93 * creation, so can be a file-level static. (Must come before #including
96 static int debug_initialization = 0;
97 # define DEBUG_INITIALIZATION_set(v) (debug_initialization = v)
98 # define DEBUG_LOCALE_INITIALIZATION_ debug_initialization
100 # define debug_initialization 0
101 # define DEBUG_INITIALIZATION_set(v)
104 #define DEBUG_PRE_STMTS dSAVE_ERRNO; \
105 PerlIO_printf(Perl_debug_log, "%s: %" LINE_Tf ": ", __FILE__, __LINE__);
106 #define DEBUG_POST_STMTS RESTORE_ERRNO;
109 #define PERL_IN_LOCALE_C
110 #include "perl_langinfo.h"
123 /* Returns the Unix errno portion; ignoring any others. This is a macro here
124 * instead of putting it into perl.h, because unclear to khw what should be
126 #define GET_ERRNO saved_errno
128 /* Is the C string input 'name' "C" or "POSIX"? If so, and 'name' is the
129 * return of setlocale(), then this is extremely likely to be the C or POSIX
130 * locale. However, the output of setlocale() is documented to be opaque, but
131 * the odds are extremely small that it would return these two strings for some
132 * other locale. Note that VMS in these two locales includes many non-ASCII
133 * characters as controls and punctuation (below are hex bytes):
135 * punct: A1-A3 A5 A7-AB B0-B3 B5-B7 B9-BD BF-CF D1-DD DF-EF F1-FD
136 * Oddly, none there are listed as alphas, though some represent alphabetics
137 * http://www.nntp.perl.org/group/perl.perl5.porters/2013/02/msg198753.html */
138 #define isNAME_C_OR_POSIX(name) \
140 && (( *(name) == 'C' && (*(name + 1)) == '\0') \
141 || strEQ((name), "POSIX")))
145 /* This code keeps a LRU cache of the UTF-8ness of the locales it has so-far
146 * looked up. This is in the form of a C string: */
148 # define UTF8NESS_SEP "\v"
149 # define UTF8NESS_PREFIX "\f"
151 /* So, the string looks like:
153 * \vC\a0\vPOSIX\a0\vam_ET\a0\vaf_ZA.utf8\a1\ven_US.UTF-8\a1\0
155 * where the digit 0 after the \a indicates that the locale starting just
156 * after the preceding \v is not UTF-8, and the digit 1 mean it is. */
158 STATIC_ASSERT_DECL(STRLENs(UTF8NESS_SEP) == 1);
159 STATIC_ASSERT_DECL(STRLENs(UTF8NESS_PREFIX) == 1);
161 # define C_and_POSIX_utf8ness UTF8NESS_SEP "C" UTF8NESS_PREFIX "0" \
162 UTF8NESS_SEP "POSIX" UTF8NESS_PREFIX "0"
164 /* The cache is initialized to C_and_POSIX_utf8ness at start up. These are
165 * kept there always. The remining portion of the cache is LRU, with the
166 * oldest looked-up locale at the tail end */
169 # define setlocale_debug_string_c(category, locale, result) \
170 setlocale_debug_string_i(category##_INDEX_, locale, result)
171 # define setlocale_debug_string_r(category, locale, result) \
172 setlocale_debug_string_i(get_category_index(category, locale), \
176 /* Two parallel arrays indexed by our mapping of category numbers into small
177 * non-negative indexes; first the locale categories Perl uses on this system,
178 * used to do the inverse mapping. The second array is their names. These
179 * arrays are in mostly arbitrary order. */
181 STATIC const int categories[] = {
183 # ifdef USE_LOCALE_CTYPE
186 # ifdef USE_LOCALE_NUMERIC
189 # ifdef USE_LOCALE_COLLATE
192 # ifdef USE_LOCALE_TIME
195 # ifdef USE_LOCALE_MESSAGES
198 # ifdef USE_LOCALE_MONETARY
201 # ifdef USE_LOCALE_ADDRESS
204 # ifdef USE_LOCALE_IDENTIFICATION
207 # ifdef USE_LOCALE_MEASUREMENT
210 # ifdef USE_LOCALE_PAPER
213 # ifdef USE_LOCALE_TELEPHONE
216 # ifdef USE_LOCALE_SYNTAX
219 # ifdef USE_LOCALE_TOD
226 /* Placeholder as a precaution if code fails to check the return of
227 * get_category_index(), which returns this element to indicate an error */
231 /* The top-most real element is LC_ALL */
233 STATIC const char * const category_names[] = {
235 # ifdef USE_LOCALE_CTYPE
238 # ifdef USE_LOCALE_NUMERIC
241 # ifdef USE_LOCALE_COLLATE
244 # ifdef USE_LOCALE_TIME
247 # ifdef USE_LOCALE_MESSAGES
250 # ifdef USE_LOCALE_MONETARY
253 # ifdef USE_LOCALE_ADDRESS
256 # ifdef USE_LOCALE_IDENTIFICATION
259 # ifdef USE_LOCALE_MEASUREMENT
262 # ifdef USE_LOCALE_PAPER
265 # ifdef USE_LOCALE_TELEPHONE
268 # ifdef USE_LOCALE_SYNTAX
271 # ifdef USE_LOCALE_TOD
278 /* Placeholder as a precaution if code fails to check the return of
279 * get_category_index(), which returns this element to indicate an error */
283 /* A few categories require additional setup when they are changed. This table
284 * points to the functions that do that setup */
285 STATIC void (*update_functions[]) (pTHX_ const char *) = {
286 # ifdef USE_LOCALE_CTYPE
289 # ifdef USE_LOCALE_NUMERIC
292 # ifdef USE_LOCALE_COLLATE
295 # ifdef USE_LOCALE_TIME
298 # ifdef USE_LOCALE_MESSAGES
301 # ifdef USE_LOCALE_MONETARY
304 # ifdef USE_LOCALE_ADDRESS
307 # ifdef USE_LOCALE_IDENTIFICATION
310 # ifdef USE_LOCALE_MEASUREMENT
313 # ifdef USE_LOCALE_PAPER
316 # ifdef USE_LOCALE_TELEPHONE
319 # ifdef USE_LOCALE_SYNTAX
322 # ifdef USE_LOCALE_TOD
325 /* No harm done to have this even without an LC_ALL */
328 /* Placeholder as a precaution if code fails to check the return of
329 * get_category_index(), which returns this element to indicate an error */
335 /* On systems with LC_ALL, it is kept in the highest index position. (-2
336 * to account for the final unused placeholder element.) */
337 # define NOMINAL_LC_ALL_INDEX (C_ARRAY_LENGTH(categories) - 2)
340 /* On systems without LC_ALL, we pretend it is there, one beyond the real
341 * top element, hence in the unused placeholder element. */
342 # define NOMINAL_LC_ALL_INDEX (C_ARRAY_LENGTH(categories) - 1)
345 /* Pretending there is an LC_ALL element just above allows us to avoid most
346 * special cases. Most loops through these arrays in the code below are
347 * written like 'for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++)'. They will work
348 * on either type of system. But the code must be written to not access the
349 * element at 'LC_ALL_INDEX_' except on platforms that have it. This can be
350 * checked for at compile time by using the #define LC_ALL_INDEX_ which is only
351 * defined if we do have LC_ALL. */
354 S_get_category_index(const int category, const char * locale)
356 /* Given a category, return the equivalent internal index we generally use
359 * 'locale' is for use in any generated diagnostics, and may be NULL
361 * Some sort of hash could be used instead of this loop, but the number of
362 * elements is so far at most 12 */
365 const char * conditional_warn_text = "; can't set it to ";
367 PERL_ARGS_ASSERT_GET_CATEGORY_INDEX;
370 for (i = 0; i <= LC_ALL_INDEX_; i++)
372 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++)
375 if (category == categories[i]) {
377 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
378 "index of category %d (%s) is %d\n",
379 category, category_names[i], i));
384 /* Here, we don't know about this category, so can't handle it. */
388 conditional_warn_text = "";
391 /* diag_listed_as: Unknown locale category %d; can't set it to %s */
392 Perl_warner_nocontext(packWARN(WARN_LOCALE),
393 "Unknown locale category %d%s%s",
394 category, conditional_warn_text, locale);
398 SETERRNO(EINVAL, LIB_INVARG);
402 /* Return an out-of-bounds value */
403 return NOMINAL_LC_ALL_INDEX + 1;
407 S_category_name(const int category)
411 index = get_category_index(category, NULL);
413 if (index <= NOMINAL_LC_ALL_INDEX) {
414 return category_names[index];
417 return Perl_form_nocontext("%d (unknown)", category);
420 #endif /* ifdef USE_LOCALE */
423 Perl_force_locale_unlock()
426 #if defined(USE_LOCALE_THREADS)
429 # ifdef LOCALE_UNLOCK_
437 #ifdef USE_POSIX_2008_LOCALE
440 S_use_curlocale_scratch(pTHX)
442 /* This function is used to hide from the caller the case where the current
443 * locale_t object in POSIX 2008 is the global one, which is illegal in
444 * many of the P2008 API calls. This checks for that and, if necessary
445 * creates a proper P2008 object. Any prior object is deleted, as is any
446 * remaining object during global destruction. */
448 locale_t cur = uselocale((locale_t) 0);
450 if (cur != LC_GLOBAL_LOCALE) {
454 if (PL_scratch_locale_obj) {
455 freelocale(PL_scratch_locale_obj);
458 PL_scratch_locale_obj = duplocale(LC_GLOBAL_LOCALE);
459 return PL_scratch_locale_obj;
465 Perl_locale_panic(const char * msg,
466 const char * file_name,
472 PERL_ARGS_ASSERT_LOCALE_PANIC;
474 force_locale_unlock();
476 #ifdef USE_C_BACKTRACE
477 dump_c_backtrace(Perl_debug_log, 20, 1);
480 /* diag_listed_as: panic: %s */
481 Perl_croak(aTHX_ "%s: %d: panic: %s; errno=%d\n",
482 file_name, line, msg, errnum);
485 #define setlocale_failure_panic_c( \
486 cat, current, failed, caller_0_line, caller_1_line) \
487 setlocale_failure_panic_i(cat##_INDEX_, current, failed, \
488 caller_0_line, caller_1_line)
490 /* porcelain_setlocale() presents a consistent POSIX-compliant interface to
491 * setlocale(). Windows requres a customized base-level setlocale() */
493 # define porcelain_setlocale(cat, locale) win32_setlocale(cat, locale)
495 # define porcelain_setlocale(cat, locale) \
496 ((const char *) setlocale(cat, locale))
499 /* The next layer up is to catch vagaries and bugs in the libc setlocale return
502 # define stdized_setlocale(cat, locale) \
503 stdize_locale(cat, porcelain_setlocale(cat, locale), \
504 &PL_stdize_locale_buf, &PL_stdize_locale_bufsize, __LINE__)
506 # define stdized_setlocale(cat, locale) porcelain_setlocale(cat, locale)
509 /* The next many lines form a layer above the close-to-the-metal 'porcelain'
510 * and 'stdized' macros. They are used to present a uniform API to the rest of
511 * the code in this file in spite of the disparate underlying implementations.
514 #ifndef USE_POSIX_2008_LOCALE
516 /* For non-threaded perls (which we are not to use the POSIX 2008 API on), or a
517 * thread-safe Windows one in which threading is invisible to us, the added
518 * layer just calls the base-level functions. See the introductory comments in
519 * this file for the meaning of the suffixes '_c', '_r', '_i'. */
521 # define setlocale_r(cat, locale) stdized_setlocale(cat, locale)
522 # define setlocale_i(i, locale) setlocale_r(categories[i], locale)
523 # define setlocale_c(cat, locale) setlocale_r(cat, locale)
525 # define void_setlocale_i(i, locale) \
527 if (! porcelain_setlocale(categories[i], locale)) { \
528 setlocale_failure_panic_i(i, NULL, locale, __LINE__, 0); \
529 NOT_REACHED; /* NOTREACHED */ \
532 # define void_setlocale_c(cat, locale) \
533 void_setlocale_i(cat##_INDEX_, locale)
534 # define void_setlocale_r(cat, locale) \
535 void_setlocale_i(get_category_index(cat, locale), locale)
537 # define bool_setlocale_r(cat, locale) \
538 cBOOL(porcelain_setlocale(cat, locale))
539 # define bool_setlocale_i(i, locale) \
540 bool_setlocale_c(categories[i], locale)
541 # define bool_setlocale_c(cat, locale) bool_setlocale_r(cat, locale)
543 # define querylocale_r(cat) setlocale_r(cat, NULL)
544 # define querylocale_c(cat) querylocale_r(cat)
545 # define querylocale_i(i) querylocale_c(categories[i])
547 #else /* Below is defined(POSIX 2008) */
549 /* Here, there is a completely different API to get thread-safe locales. We
550 * emulate the setlocale() API with our own function(s). setlocale categories,
551 * like LC_NUMERIC, are not valid here for the POSIX 2008 API. Instead, there
552 * are equivalents, like LC_NUMERIC_MASK, which we use instead, converting to
553 * by using get_category_index() followed by table lookup. */
555 # define emulate_setlocale_c(cat, locale, recalc_LC_ALL, line) \
556 emulate_setlocale_i(cat##_INDEX_, locale, recalc_LC_ALL, line)
558 /* A wrapper for the macros below. */
559 # define common_emulate_setlocale(i, locale) \
560 emulate_setlocale_i(i, locale, YES_RECALC_LC_ALL, __LINE__)
562 # define setlocale_i(i, locale) common_emulate_setlocale(i, locale)
563 # define setlocale_c(cat, locale) setlocale_i(cat##_INDEX_, locale)
564 # define setlocale_r(cat, locale) \
565 setlocale_i(get_category_index(cat, locale), locale)
567 # define void_setlocale_i(i, locale) ((void) setlocale_i(i, locale))
568 # define void_setlocale_c(cat, locale) \
569 void_setlocale_i(cat##_INDEX_, locale)
570 # define void_setlocale_r(cat, locale) ((void) setlocale_r(cat, locale))
572 # define bool_setlocale_i(i, locale) cBOOL(setlocale_i(i, locale))
573 # define bool_setlocale_c(cat, locale) \
574 bool_setlocale_i(cat##_INDEX_, locale)
575 # define bool_setlocale_r(cat, locale) cBOOL(setlocale_r(cat, locale))
577 # define querylocale_i(i) my_querylocale_i(i)
578 # define querylocale_c(cat) querylocale_i(cat##_INDEX_)
579 # define querylocale_r(cat) querylocale_i(get_category_index(cat,NULL))
581 # ifndef USE_QUERYLOCALE
582 # define USE_PL_CURLOCALES
584 # define isSINGLE_BIT_SET(mask) isPOWER_OF_2(mask)
586 /* This code used to think querylocale() was valid on LC_ALL. Make sure
587 * all instances of that have been removed */
588 # define QUERYLOCALE_ASSERT(index) \
589 __ASSERT_(isSINGLE_BIT_SET(category_masks[index]))
590 # if ! defined(HAS_QUERYLOCALE) && defined(_NL_LOCALE_NAME)
591 # define querylocale_l(index, locale_obj) \
592 (QUERYLOCALE_ASSERT(index) \
593 nl_langinfo_l(_NL_LOCALE_NAME(categories[index]), locale_obj))
595 # define querylocale_l(index, locale_obj) \
596 (QUERYLOCALE_ASSERT(index) \
597 querylocale(category_masks[index], locale_obj))
600 # if defined(__GLIBC__) && defined(USE_LOCALE_MESSAGES)
601 # define HAS_GLIBC_LC_MESSAGES_BUG
602 # include <libintl.h>
605 /* A fourth array, parallel to the ones above to map from category to its
607 STATIC const int category_masks[] = {
608 # ifdef USE_LOCALE_CTYPE
611 # ifdef USE_LOCALE_NUMERIC
614 # ifdef USE_LOCALE_COLLATE
617 # ifdef USE_LOCALE_TIME
620 # ifdef USE_LOCALE_MESSAGES
623 # ifdef USE_LOCALE_MONETARY
626 # ifdef USE_LOCALE_ADDRESS
629 # ifdef USE_LOCALE_IDENTIFICATION
630 LC_IDENTIFICATION_MASK,
632 # ifdef USE_LOCALE_MEASUREMENT
635 # ifdef USE_LOCALE_PAPER
638 # ifdef USE_LOCALE_TELEPHONE
641 # ifdef USE_LOCALE_SYNTAX
644 # ifdef USE_LOCALE_TOD
647 /* LC_ALL can't be turned off by a Configure
648 * option, and in Posix 2008, should always be
649 * here, so compile it in unconditionally.
650 * This could catch some glitches at compile
654 /* Placeholder as a precaution if code fails to check the return of
655 * get_category_index(), which returns this element to indicate an error */
659 # define my_querylocale_c(cat) my_querylocale_i(cat##_INDEX_)
662 S_my_querylocale_i(pTHX_ const unsigned int index)
664 /* This function returns the name of the locale category given by the input
665 * index into our parallel tables of them.
667 * POSIX 2008, for some sick reason, chose not to provide a method to find
668 * the category name of a locale, discarding a basic linguistic tenet that
669 * for any object, people will create a name for it. Some vendors have
670 * created a querylocale() function to do just that. This function is a
671 * lot simpler to implement on systems that have this. Otherwise, we have
672 * to keep track of what the locale has been set to, so that we can return
673 * its name so as to emulate setlocale(). It's also possible for C code in
674 * some library to change the locale without us knowing it, though as of
675 * September 2017, there are no occurrences in CPAN of uselocale(). Some
676 * libraries do use setlocale(), but that changes the global locale, and
677 * threads using per-thread locales will just ignore those changes. */
680 const locale_t cur_obj = uselocale((locale_t) 0);
683 PERL_ARGS_ASSERT_MY_QUERYLOCALE_I;
684 assert(index <= NOMINAL_LC_ALL_INDEX);
686 category = categories[index];
688 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "my_querylocale_i(%s) on %p\n",
689 category_names[index], cur_obj));
690 if (cur_obj == LC_GLOBAL_LOCALE) {
691 retval = porcelain_setlocale(category, NULL);
695 # ifdef USE_QUERYLOCALE
697 /* We don't currently keep records when there is querylocale(), so have
698 * to get it anew each time */
699 retval = (index == LC_ALL_INDEX_)
700 ? calculate_LC_ALL(cur_obj)
701 : querylocale_l(index, cur_obj);
705 /* But we do have up-to-date values when we keep our own records */
706 retval = PL_curlocales[index];
712 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
713 "my_querylocale_i(%s) returning '%s'\n",
714 category_names[index], retval));
718 # ifdef USE_PL_CURLOCALES
721 S_update_PL_curlocales_i(pTHX_
722 const unsigned int index,
723 const char * new_locale,
724 recalc_lc_all_t recalc_LC_ALL)
726 /* This is a helper function for emulate_setlocale_i(), mostly used to
727 * make that function easier to read. */
729 PERL_ARGS_ASSERT_UPDATE_PL_CURLOCALES_I;
730 assert(index <= NOMINAL_LC_ALL_INDEX);
732 if (index == LC_ALL_INDEX_) {
735 /* For LC_ALL, we change all individual categories to correspond */
736 /* PL_curlocales is a parallel array, so has same
737 * length as 'categories' */
738 for (i = 0; i < LC_ALL_INDEX_; i++) {
739 Safefree(PL_curlocales[i]);
740 PL_curlocales[i] = savepv(new_locale);
743 recalc_LC_ALL = YES_RECALC_LC_ALL;
747 /* Update the single category's record */
748 Safefree(PL_curlocales[index]);
749 PL_curlocales[index] = savepv(new_locale);
751 if (recalc_LC_ALL == RECALCULATE_LC_ALL_ON_FINAL_INTERATION) {
752 recalc_LC_ALL = (index == NOMINAL_LC_ALL_INDEX - 1)
754 : DONT_RECALC_LC_ALL;
758 if (recalc_LC_ALL == YES_RECALC_LC_ALL) {
759 Safefree(PL_curlocales[LC_ALL_INDEX_]);
760 PL_curlocales[LC_ALL_INDEX_] =
761 savepv(calculate_LC_ALL(PL_curlocales));
764 return PL_curlocales[index];
767 # endif /* Need PL_curlocales[] */
770 S_setlocale_from_aggregate_LC_ALL(pTHX_ const char * locale, const line_t line)
772 /* This function parses the value of the LC_ALL locale, assuming glibc
773 * syntax, and sets each individual category on the system to the proper
776 * This is likely to only ever be called from one place, so exists to make
777 * the calling function easier to read by moving this ancillary code out of
780 * The locale for each category is independent of the other categories.
781 * Often, they are all the same, but certainly not always. Perl, in fact,
782 * usually keeps LC_NUMERIC in the C locale, regardless of the underlying
783 * locale. LC_ALL has to be able to represent the case of when there are
784 * varying locales. Platforms have differing ways of representing this.
785 * Because of this, the code in this file goes to lengths to avoid the
786 * issue, generally looping over the component categories instead of
787 * referring to them in the aggregate, wherever possible. However, there
788 * are cases where we have to parse our own constructed aggregates, which use
789 * the glibc syntax. */
791 const char * locale_on_entry = savepv(querylocale_c(LC_ALL));
793 PERL_ARGS_ASSERT_SETLOCALE_FROM_AGGREGATE_LC_ALL;
795 /* If the string that gives what to set doesn't include all categories,
796 * the omitted ones get set to "C". To get this behavior, first set
797 * all the individual categories to "C", and override the furnished
798 * ones below. FALSE => No need to recalculate LC_ALL, as this is a
800 if (! emulate_setlocale_c(LC_ALL, "C", DONT_RECALC_LC_ALL, line)) {
801 setlocale_failure_panic_c(LC_ALL, locale_on_entry,
802 "C", __LINE__, line);
803 NOT_REACHED; /* NOTREACHED */
806 const char * s = locale;
807 const char * e = locale + strlen(locale);
811 /* Parse through the category */
812 while (isWORDCHAR(*p)) {
816 const char * category_end = p;
819 locale_panic_(Perl_form(aTHX_
820 "Unexpected character in locale category name '%02X", *(p-1)));
823 /* Parse through the locale name */
824 const char * name_start = p;
825 while (p < e && *p != ';') {
827 locale_panic_(Perl_form(aTHX_
828 "Unexpected character in locale name '%02X", *p));
833 const char * name_end = p;
835 /* Space past the semi-colon */
840 /* Find the index of the category name in our lists */
841 for (PERL_UINT_FAST8_T i = 0; i < LC_ALL_INDEX_; i++) {
843 /* Keep going if this index doesn't point to the category being
844 * parsed. The strnNE() avoids a Perl_form(), but would fail if
845 * ever a category name could be a substring of another one, e.g.,
846 * if there were a "LC_TIME_DATE" */
847 if strnNE(s, category_names[i], category_end - s) {
851 /* Here i points to the category being parsed. Now isolate the
852 * locale it is being changed to */
853 const char * individ_locale = Perl_form(aTHX_ "%.*s",
854 (int) (name_end - name_start), name_start);
856 /* And do the change. FALSE => Don't recalculate LC_ALL; we'll do
857 * it ourselves after the loop */
858 if (! emulate_setlocale_i(i, individ_locale,
859 DONT_RECALC_LC_ALL, line))
862 /* But if we have to back out, do fix up LC_ALL */
863 if (! emulate_setlocale_c(LC_ALL, locale_on_entry,
864 YES_RECALC_LC_ALL, line))
866 Safefree(locale_on_entry);
867 setlocale_failure_panic_i(i, individ_locale,
868 locale, __LINE__, line);
869 NOT_REACHED; /* NOTREACHED */
872 Safefree(locale_on_entry);
874 /* Reverting to the entry value succeeded, but the operation
875 * failed to go to the requested locale. */
879 /* Found and handled the desired category. Quit the inner loop to
880 * try the next category */
884 /* Finished with this category; iterate to the next one in the input */
888 # ifdef USE_PL_CURLOCALES
890 /* Here we have set all the individual categories. Update the LC_ALL entry
891 * as well. We can't just use the input 'locale' as the value may omit
892 * categories whose locale is 'C'. khw thinks it's better to store a
893 * complete LC_ALL. So calculate it. */
894 const char * retval = savepv(calculate_LC_ALL(PL_curlocales));
895 Safefree(PL_curlocales[LC_ALL_INDEX_]);
896 PL_curlocales[LC_ALL_INDEX_] = retval;
900 const char * retval = querylocale_c(LC_ALL);
904 Safefree(locale_on_entry);
908 # ifndef USE_QUERYLOCALE
911 S_find_locale_from_environment(pTHX_ const unsigned int index)
913 /* On systems without querylocale(), it is problematic getting the results
914 * of the POSIX 2008 equivalent of setlocale(category, "") (which gets the
915 * locale from the environment).
917 * To ensure that we know exactly what those values are, we do the setting
918 * ourselves, using the documented algorithm (assuming the documentation is
919 * correct) rather than use "" as the locale. This will lead to results
920 * that differ from native behavior if the native behavior differs from the
921 * standard documented value, but khw believes it is better to know what's
922 * going on, even if different from native, than to just guess.
924 * Another option would be, in a critical section, to save the global
925 * locale's current value, and do a straight setlocale(LC_ALL, ""). That
926 * would return our desired values, destroying the global locale's, which
927 * we would then restore. But that could cause races with any other thread
928 * that is using the global locale and isn't using the mutex. And, the
929 * only reason someone would have done that is because they are calling a
930 * library function, like in gtk, that calls setlocale(), and which can't
931 * be changed to use the mutex. That wouldn't be a problem if this were to
932 * be done before any threads had switched, say during perl construction
933 * time. But this code would still be needed for the general case. */
935 const char * default_name;
937 const char * locale_names[LC_ALL_INDEX_];
939 /* We rely on PerlEnv_getenv() returning a mortalized copy */
940 const char * const lc_all = PerlEnv_getenv("LC_ALL");
942 /* Use any "LC_ALL" environment variable, as it overrides everything
944 if (lc_all && strNE(lc_all, "")) {
948 /* Otherwise, we need to dig deeper. Unless overridden, the default is
949 * the LANG environment variable; "C" if it doesn't exist. */
950 default_name = PerlEnv_getenv("LANG");
951 if (! default_name || strEQ(default_name, "")) {
955 /* If setting an individual category, use its corresponding value found in
956 * the environment, if any; otherwise use the default we already
958 if (index != LC_ALL_INDEX_) {
959 const char * const new_value = PerlEnv_getenv(category_names[index]);
961 return (new_value && strNE(new_value, ""))
966 /* Here, we are getting LC_ALL. Any categories that don't have a
967 * corresponding environment variable set should be set to 'default_name'
969 * Simply find the values for all categories, and call the function to
971 for (i = 0; i < LC_ALL_INDEX_; i++) {
972 const char * const env_override = PerlEnv_getenv(category_names[i]);
974 locale_names[i] = (env_override && strNE(env_override, ""))
978 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
979 "find_locale_from_environment i=%d, name=%s, locale=%s\n",
980 i, category_names[i], locale_names[i]));
983 return calculate_LC_ALL(locale_names);
989 S_emulate_setlocale_i(pTHX_
991 /* Our internal index of the 'category' setlocale is
993 const unsigned int index,
995 const char * new_locale, /* The locale to set the category to */
996 const recalc_lc_all_t recalc_LC_ALL, /* Explained below */
997 const line_t line /* Called from this line number */
1000 PERL_ARGS_ASSERT_EMULATE_SETLOCALE_I;
1001 assert(index <= NOMINAL_LC_ALL_INDEX);
1003 /* This function effectively performs a setlocale() on just the current
1004 * thread; thus it is thread-safe. It does this by using the POSIX 2008
1005 * locale functions to emulate the behavior of setlocale(). Similar to
1006 * regular setlocale(), the return from this function points to memory that
1007 * can be overwritten by other system calls, so needs to be copied
1008 * immediately if you need to retain it. The difference here is that
1009 * system calls besides another setlocale() can overwrite it.
1011 * By doing this, most locale-sensitive functions become thread-safe. The
1012 * exceptions are mostly those that return a pointer to static memory.
1014 * This function may be called in a tight loop that iterates over all
1015 * categories. Because LC_ALL is not a "real" category, but merely the sum
1016 * of all the other ones, such loops don't include LC_ALL. On systems that
1017 * have querylocale() or similar, the current LC_ALL value is immediately
1018 * retrievable; on systems lacking that feature, we have to keep track of
1019 * LC_ALL ourselves. We could do that on each iteration, only to throw it
1020 * away on the next, but the calculation is more than a trivial amount of
1021 * work. Instead, the 'recalc_LC_ALL' parameter is set to
1022 * RECALCULATE_LC_ALL_ON_FINAL_INTERATION to only do the calculation once.
1023 * This function calls itself recursively in such a loop.
1025 * When not in such a loop, the parameter is set to the other enum values
1026 * DONT_RECALC_LC_ALL or YES_RECALC_LC_ALL. */
1028 int mask = category_masks[index];
1029 const locale_t entry_obj = uselocale((locale_t) 0);
1030 const char * locale_on_entry = querylocale_i(index);
1032 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1033 "emulate_setlocale_i input=%d (%s), mask=0x%x,"
1034 " new locale=\"%s\", current locale=\"%s\","
1035 "index=%d, object=%p\n",
1036 categories[index], category_name(categories[index]), mask,
1037 ((new_locale == NULL) ? "(nil)" : new_locale),
1038 locale_on_entry, index, entry_obj));
1040 /* Return the already-calculated info if just querying what the existing
1042 if (new_locale == NULL) {
1043 return locale_on_entry;
1046 /* Here, trying to change the locale, but it is a no-op if the new boss is
1047 * the same as the old boss. Except this routine is called when converting
1048 * from the global locale, so in that case we will create a per-thread
1049 * locale below (with the current values). Bitter experience also
1050 * indicates that newlocale() can free up the basis locale memory if we
1051 * call it with the new and old being the same. */
1052 if ( entry_obj != LC_GLOBAL_LOCALE
1054 && strEQ(new_locale, locale_on_entry))
1056 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1057 "(%" LINE_Tf "): emulate_setlocale_i"
1058 " no-op to change to what it already was\n",
1061 # ifdef USE_PL_CURLOCALES
1063 /* On the final iteration of a loop that needs to recalculate LC_ALL, do
1064 * so. If no iteration changed anything, LC_ALL also doesn't change,
1065 * but khw believes the complexity needed to keep track of that isn't
1067 if (UNLIKELY( recalc_LC_ALL == RECALCULATE_LC_ALL_ON_FINAL_INTERATION
1068 && index == NOMINAL_LC_ALL_INDEX - 1))
1070 Safefree(PL_curlocales[LC_ALL_INDEX_]);
1071 PL_curlocales[LC_ALL_INDEX_] =
1072 savepv(calculate_LC_ALL(PL_curlocales));
1077 return locale_on_entry;
1080 # ifndef USE_QUERYLOCALE
1082 /* Without a querylocale() mechanism, we have to figure out ourselves what
1083 * happens with setting a locale to "" */
1084 if (strEQ(new_locale, "")) {
1085 new_locale = find_locale_from_environment(index);
1090 /* So far, it has worked that a semi-colon in the locale name means that
1091 * the category is LC_ALL and it subsumes categories which don't all have
1092 * the same locale. This is the glibc syntax. */
1093 if (strchr(new_locale, ';')) {
1094 assert(index == LC_ALL_INDEX_);
1095 return setlocale_from_aggregate_LC_ALL(new_locale, line);
1098 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
1100 /* For this bug, if the LC_MESSAGES locale changes, we have to do an
1101 * expensive workaround. Save the current value so we can later determine
1103 const char * old_messages_locale = NULL;
1104 if ( (index == LC_MESSAGES_INDEX_ || index == LC_ALL_INDEX_)
1105 && LIKELY(PL_phase != PERL_PHASE_CONSTRUCT))
1107 old_messages_locale = savepv(querylocale_c(LC_MESSAGES));
1112 assert(PL_C_locale_obj);
1114 /* Now ready to switch to the input 'new_locale' */
1116 /* Switching locales generally entails freeing the current one's space (at
1117 * the C library's discretion), hence we can't be using that locale at the
1118 * time of the switch (this wasn't obvious to khw from the man pages). So
1119 * switch to a known locale object that we don't otherwise mess with. */
1120 if (! uselocale(PL_C_locale_obj)) {
1122 /* Not being able to change to the C locale is severe; don't keep
1124 setlocale_failure_panic_i(index, locale_on_entry, "C", __LINE__, line);
1125 NOT_REACHED; /* NOTREACHED */
1128 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1129 "(%" LINE_Tf "): emulate_setlocale_i now using C"
1130 " object=%p\n", line, PL_C_locale_obj));
1134 /* We created a (never changing) object at start-up for LC_ALL being in the
1135 * C locale. If this call is to switch to LC_ALL=>C, simply use that
1136 * object. But in fact, we already have switched to it just above, in
1137 * preparation for the general case. Since we're already there, no need to
1138 * do further switching. */
1139 if (mask == LC_ALL_MASK && isNAME_C_OR_POSIX(new_locale)) {
1140 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "(%" LINE_Tf "):"
1141 " emulate_setlocale_i will stay"
1142 " in C object\n", line));
1143 new_obj = PL_C_locale_obj;
1145 /* And free the old object if it isn't a special one */
1146 if (entry_obj != LC_GLOBAL_LOCALE && entry_obj != PL_C_locale_obj) {
1147 freelocale(entry_obj);
1150 else { /* Here is the general case, not to LC_ALL=>C */
1151 locale_t basis_obj = entry_obj;
1153 /* Specially handle two objects */
1154 if (basis_obj == LC_GLOBAL_LOCALE || basis_obj == PL_C_locale_obj) {
1156 /* For these two objects, we make duplicates to hand to newlocale()
1157 * below. For LC_GLOBAL_LOCALE, this is because newlocale()
1158 * doesn't necessarily accept it as input (the results are
1159 * undefined). For PL_C_locale_obj, it is so that it never gets
1160 * modified, as otherwise newlocale() is free to do so */
1161 basis_obj = duplocale(entry_obj);
1163 locale_panic_(Perl_form(aTHX_ "(%" LINE_Tf "): duplocale failed",
1165 NOT_REACHED; /* NOTREACHED */
1168 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1169 "(%" LINE_Tf "): emulate_setlocale_i"
1170 " created %p by duping the input\n",
1174 /* Ready to create a new locale by modification of the exising one */
1175 new_obj = newlocale(mask, new_locale, basis_obj);
1178 DEBUG_L(PerlIO_printf(Perl_debug_log,
1179 " (%" LINE_Tf "): emulate_setlocale_i"
1180 " creating new object from %p failed:"
1182 line, basis_obj, GET_ERRNO));
1184 /* Failed. Likely this is because the proposed new locale isn't
1185 * valid on this system. But we earlier switched to the LC_ALL=>C
1186 * locale in anticipation of it succeeding, Now have to switch
1187 * back to the state upon entry */
1188 if (! uselocale(entry_obj)) {
1189 setlocale_failure_panic_i(index, "switching back to",
1190 locale_on_entry, __LINE__, line);
1191 NOT_REACHED; /* NOTREACHED */
1194 # ifdef USE_PL_CURLOCALES
1196 if (entry_obj == LC_GLOBAL_LOCALE) {
1197 /* Here, we are back in the global locale. We may never have
1198 * set PL_curlocales. If the locale change had succeeded, the
1199 * code would have then set them up, but since it didn't, do so
1200 * here. khw isn't sure if this prevents some issues or not,
1201 * but tis is defensive coding. The system setlocale() returns
1202 * the desired information. This will calculate LC_ALL's entry
1203 * only on the final iteration */
1204 for (PERL_UINT_FAST8_T i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
1205 update_PL_curlocales_i(i,
1206 porcelain_setlocale(categories[i], NULL),
1207 RECALCULATE_LC_ALL_ON_FINAL_INTERATION);
1212 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
1213 Safefree(old_messages_locale);
1219 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1220 "(%" LINE_Tf "): emulate_setlocale_i created %p"
1221 " while freeing %p\n", line, new_obj, basis_obj));
1223 /* Here, successfully created an object representing the desired
1224 * locale; now switch into it */
1225 if (! uselocale(new_obj)) {
1226 freelocale(new_obj);
1227 locale_panic_(Perl_form(aTHX_ "(%" LINE_Tf "): emulate_setlocale_i"
1228 " switching into new locale failed",
1233 /* Here, we are using 'new_obj' which matches the input 'new_locale'. */
1234 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1235 "(%" LINE_Tf "): emulate_setlocale_i now using %p\n", line, new_obj));
1237 /* We are done, except for updating our records (if the system doesn't keep
1238 * them) and in the case of locale "", we don't actually know what the
1239 * locale that got switched to is, as it came from the environment. So
1240 * have to find it */
1242 # ifdef USE_QUERYLOCALE
1244 if (strEQ(new_locale, "")) {
1245 new_locale = querylocale_i(index);
1248 PERL_UNUSED_ARG(recalc_LC_ALL);
1252 new_locale = update_PL_curlocales_i(index, new_locale, recalc_LC_ALL);
1255 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
1257 /* Invalidate the glibc cache of loaded translations if the locale has changed,
1258 * see [perl #134264] */
1259 if (old_messages_locale) {
1260 if (strNE(old_messages_locale, my_querylocale_c(LC_MESSAGES))) {
1261 textdomain(textdomain(NULL));
1264 Safefree(old_messages_locale);
1272 #endif /* End of the various implementations of the setlocale and
1273 querylocale macros used in the remainder of this program */
1277 /* So far, the locale strings returned by modern 2008-compliant systems have
1281 S_stdize_locale(pTHX_ const int category,
1282 const char *input_locale,
1285 const line_t caller_line)
1287 /* The return value of setlocale() is opaque, but is required to be usable
1288 * as input to a future setlocale() to create the same state.
1289 * Unfortunately not all systems are compliant. But most often they are of
1290 * a very restricted set of forms that this file has been coded to expect.
1292 * There are some outliers, though, that this function tries to tame:
1294 * 1) A new-line. This function chomps any \n characters
1295 * 2) foo=bar. 'bar' is what is generally meant, and the foo= part is
1296 * stripped. This form is legal for LC_ALL. When found in
1297 * that category group, the function calls itself
1298 * recursively on each possible component category to make
1299 * sure the individual categories are ok.
1301 * If no changes to the input were made, it is returned; otherwise the
1302 * changed version is stored into memory at *buf, with *buf_size set to its
1303 * new value, and *buf is returned.
1306 const char * first_bad;
1307 const char * retval;
1309 PERL_ARGS_ASSERT_STDIZE_LOCALE;
1311 if (input_locale == NULL) {
1315 first_bad = strpbrk(input_locale, "=\n");
1317 /* Most likely, there isn't a problem with the input */
1318 if (LIKELY(! first_bad)) {
1319 return input_locale;
1324 /* But if there is, and the category is LC_ALL, we have to look at each
1325 * component category */
1326 if (category == LC_ALL) {
1327 const char * individ_locales[LC_ALL_INDEX_];
1328 bool made_changes = FALSE;
1331 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
1332 Size_t this_size = 0;
1333 individ_locales[i] = stdize_locale(categories[i],
1334 porcelain_setlocale(categories[i],
1336 &individ_locales[i],
1340 /* If the size didn't change, it means this category did not have
1341 * to be adjusted, and individ_locales[i] points to the buffer
1342 * returned by porcelain_setlocale(); we have to copy that before
1343 * it's called again in the next iteration */
1344 if (this_size == 0) {
1345 individ_locales[i] = savepv(individ_locales[i]);
1348 made_changes = TRUE;
1352 /* If all the individual categories were ok as-is, this was a false
1353 * alarm. We must have seen an '=' which was a legal occurrence in
1354 * this combination locale */
1355 if (! made_changes) {
1356 retval = input_locale; /* The input can be returned unchanged */
1359 retval = save_to_buffer(querylocale_c(LC_ALL), buf, buf_size, 0);
1362 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
1363 Safefree(individ_locales[i]);
1371 /* Here, there was a problem in an individual category. This means that at
1372 * least one adjustment will be necessary. Create a modifiable copy */
1373 retval = save_to_buffer(input_locale, buf, buf_size, 0);
1375 if (*first_bad != '=') {
1377 /* Translate the found position into terms of the copy */
1378 first_bad = retval + (first_bad - input_locale);
1382 /* It is unlikely that the return is so screwed-up that it contains
1383 * multiple equals signs, but handle that case by stripping all of
1385 const char * final_equals = strrchr(retval, '=');
1387 /* The length passed here causes the move to include the terminating
1389 Move(final_equals + 1, retval, strlen(final_equals), char);
1391 /* See if there are additional problems; if not, we're good to return.
1393 first_bad = strpbrk(retval, "\n");
1400 /* Here, the problem must be a \n. Get rid of it and what follows.
1401 * (Originally, only a trailing \n was stripped. Unsure what to do if not
1403 *((char *) first_bad) = '\0';
1407 #if defined(USE_POSIX_2008_LOCALE)
1412 # ifdef USE_QUERYLOCALE
1413 S_calculate_LC_ALL(pTHX_ const locale_t cur_obj)
1415 S_calculate_LC_ALL(pTHX_ const char ** individ_locales)
1419 /* For POSIX 2008, we have to figure out LC_ALL ourselves when needed.
1420 * querylocale(), on systems that have it, doesn't tend to work for LC_ALL.
1421 * So we have to construct the answer ourselves based on the passed in
1422 * data, which is either a locale_t object, for systems with querylocale(),
1423 * or an array we keep updated to the proper values, otherwise.
1425 * This returns a mortalized string containing the locale name(s) of
1428 * If all individual categories are the same locale, we can just set LC_ALL
1429 * to that locale. But if not, we have to create an aggregation of all the
1430 * categories on the system. Platforms differ as to the syntax they use
1431 * for these non-uniform locales for LC_ALL. Some use a '/' or other
1432 * delimiter of the locales with a predetermined order of categories; a
1433 * Configure probe would be needed to tell us how to decipher those. glibc
1434 * uses a series of name=value pairs, like
1435 * LC_NUMERIC=C;LC_TIME=en_US.UTF-8;...
1436 * The syntax we use for our aggregation doesn't much matter, as we take
1437 * care not to use the native setlocale() function on whatever style is
1438 * chosen. But, it would be possible for someone to call Perl_setlocale()
1439 * using a native style we don't understand. So far no one has complained.
1441 * For systems that have categories we don't know about, the algorithm
1442 * below won't know about those missing categories, leading to potential
1443 * bugs for code that looks at them. If there is an environment variable
1444 * that sets that category, we won't know to look for it, and so our use of
1445 * LANG or "C" improperly overrides it. On the other hand, if we don't do
1446 * what is done here, and there is no environment variable, the category's
1447 * locale should be set to LANG or "C". So there is no good solution. khw
1448 * thinks the best is to make sure we have a complete list of possible
1449 * categories, adding new ones as they show up on obscure platforms.
1453 Size_t names_len = 0;
1454 bool are_all_categories_the_same_locale = TRUE;
1455 char * aggregate_locale;
1456 char * previous_start = NULL;
1457 char * this_start = NULL;
1458 Size_t entry_len = 0;
1460 PERL_ARGS_ASSERT_CALCULATE_LC_ALL;
1462 /* First calculate the needed size for the string listing the categories
1463 * and their locales. */
1464 for (i = 0; i < LC_ALL_INDEX_; i++) {
1466 # ifdef USE_QUERYLOCALE
1467 const char * entry = querylocale_l(i, cur_obj);
1469 const char * entry = individ_locales[i];
1472 names_len += strlen(category_names[i])
1478 names_len++; /* Trailing '\0' */
1480 /* Allocate enough space for the aggregated string */
1481 SAVEFREEPV(Newxz(aggregate_locale, names_len, char));
1483 /* Then fill it in */
1484 for (i = 0; i < LC_ALL_INDEX_; i++) {
1487 # ifdef USE_QUERYLOCALE
1488 const char * entry = querylocale_l(i, cur_obj);
1490 const char * entry = individ_locales[i];
1493 new_len = my_strlcat(aggregate_locale, category_names[i], names_len);
1494 assert(new_len <= names_len);
1495 new_len = my_strlcat(aggregate_locale, "=", names_len);
1496 assert(new_len <= names_len);
1498 this_start = aggregate_locale + strlen(aggregate_locale);
1499 entry_len = strlen(entry);
1501 new_len = my_strlcat(aggregate_locale, entry, names_len);
1502 assert(new_len <= names_len);
1503 new_len = my_strlcat(aggregate_locale, ";", names_len);
1504 assert(new_len <= names_len);
1505 PERL_UNUSED_VAR(new_len); /* Only used in DEBUGGING */
1508 && are_all_categories_the_same_locale
1509 && memNE(previous_start, this_start, entry_len + 1))
1511 are_all_categories_the_same_locale = FALSE;
1514 previous_start = this_start;
1518 /* If they are all the same, just return any one of them */
1519 if (are_all_categories_the_same_locale) {
1520 aggregate_locale = this_start;
1521 aggregate_locale[entry_len] = '\0';
1524 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1525 "calculate_LC_ALL returning '%s'\n",
1528 return aggregate_locale;
1530 #endif /*defined(USE_POSIX_2008_LOCALE)*/
1533 S_setlocale_failure_panic_i(pTHX_
1534 const unsigned int cat_index,
1535 const char * current,
1536 const char * failed,
1537 const line_t caller_0_line,
1538 const line_t caller_1_line)
1541 const int cat = categories[cat_index];
1542 const char * name = category_names[cat_index];
1544 PERL_ARGS_ASSERT_SETLOCALE_FAILURE_PANIC_I;
1546 if (current == NULL) {
1547 current = querylocale_i(cat_index);
1550 Perl_locale_panic(Perl_form(aTHX_ "(%" LINE_Tf
1551 "): Can't change locale for %s(%d)"
1552 " from '%s' to '%s'",
1553 caller_1_line, name, cat,
1555 __FILE__, caller_0_line, GET_ERRNO);
1556 NOT_REACHED; /* NOTREACHED */
1560 S_set_numeric_radix(pTHX_ const bool use_locale)
1562 /* If 'use_locale' is FALSE, set to use a dot for the radix character. If
1563 * TRUE, use the radix character derived from the current locale */
1565 # if defined(USE_LOCALE_NUMERIC) && ( defined(HAS_LOCALECONV) \
1566 || defined(HAS_NL_LANGINFO))
1568 const char * radix = (use_locale)
1569 ? my_nl_langinfo(RADIXCHAR, FALSE)
1570 /* FALSE => already in dest locale */
1573 sv_setpv(PL_numeric_radix_sv, radix);
1575 /* If this is valid UTF-8 that isn't totally ASCII, and we are in
1576 * a UTF-8 locale, then mark the radix as being in UTF-8 */
1577 if (is_utf8_non_invariant_string((U8 *) SvPVX(PL_numeric_radix_sv),
1578 SvCUR(PL_numeric_radix_sv))
1579 && _is_cur_LC_category_utf8(LC_NUMERIC))
1581 SvUTF8_on(PL_numeric_radix_sv);
1584 DEBUG_L(PerlIO_printf(Perl_debug_log, "Locale radix is '%s', ?UTF-8=%d\n",
1585 SvPVX(PL_numeric_radix_sv),
1586 cBOOL(SvUTF8(PL_numeric_radix_sv))));
1589 PERL_UNUSED_ARG(use_locale);
1591 # endif /* USE_LOCALE_NUMERIC and can find the radix char */
1596 S_new_numeric(pTHX_ const char *newnum)
1599 # ifndef USE_LOCALE_NUMERIC
1601 PERL_UNUSED_ARG(newnum);
1605 /* Called after each libc setlocale() call affecting LC_NUMERIC, to tell
1606 * core Perl this and that 'newnum' is the name of the new locale, and we
1607 * are switched into it. It installs this locale as the current underlying
1608 * default, and then switches to the C locale, if necessary, so that the
1609 * code that has traditionally expected the radix character to be a dot may
1610 * continue to do so.
1612 * The default locale and the C locale can be toggled between by use of the
1613 * set_numeric_underlying() and set_numeric_standard() functions, which
1614 * should probably not be called directly, but only via macros like
1615 * SET_NUMERIC_STANDARD() in perl.h.
1617 * The toggling is necessary mainly so that a non-dot radix decimal point
1618 * character can be input and output, while allowing internal calculations
1621 * This sets several interpreter-level variables:
1622 * PL_numeric_name The underlying locale's name: a copy of 'newnum'
1623 * PL_numeric_underlying A boolean indicating if the toggled state is such
1624 * that the current locale is the program's underlying
1626 * PL_numeric_standard An int indicating if the toggled state is such
1627 * that the current locale is the C locale or
1628 * indistinguishable from the C locale. If non-zero, it
1629 * is in C; if > 1, it means it may not be toggled away
1631 * PL_numeric_underlying_is_standard A bool kept by this function
1632 * indicating that the underlying locale and the standard
1633 * C locale are indistinguishable for the purposes of
1634 * LC_NUMERIC. This happens when both of the above two
1635 * variables are true at the same time. (Toggling is a
1636 * no-op under these circumstances.) This variable is
1637 * used to avoid having to recalculate.
1638 * PL_numeric_radix_sv Contains the string that code should use for the
1639 * decimal point. It is set to either a dot or the
1640 * program's underlying locale's radix character string,
1641 * depending on the situation.
1642 * PL_underlying_numeric_obj = (only on POSIX 2008 platforms) An object
1643 * with everything set up properly so as to avoid work on
1650 Safefree(PL_numeric_name);
1651 PL_numeric_name = NULL;
1652 PL_numeric_standard = TRUE;
1653 PL_numeric_underlying = TRUE;
1654 PL_numeric_underlying_is_standard = TRUE;
1658 save_newnum = savepv(newnum);
1659 PL_numeric_underlying = TRUE;
1660 PL_numeric_standard = isNAME_C_OR_POSIX(save_newnum);
1662 # ifndef TS_W32_BROKEN_LOCALECONV
1664 /* If its name isn't C nor POSIX, it could still be indistinguishable from
1665 * them. But on broken Windows systems calling my_nl_langinfo() for
1666 * THOUSEP can currently (but rarely) cause a race, so avoid doing that,
1667 * and just always change the locale if not C nor POSIX on those systems */
1668 if (! PL_numeric_standard) {
1669 PL_numeric_standard = cBOOL(strEQ(".", my_nl_langinfo(RADIXCHAR,
1670 FALSE /* Don't toggle locale */ ))
1671 && strEQ("", my_nl_langinfo(THOUSEP, FALSE)));
1676 /* Save the new name if it isn't the same as the previous one, if any */
1677 if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) {
1678 /* Save the locale name for future use */
1679 Safefree(PL_numeric_name);
1680 PL_numeric_name = save_newnum;
1683 Safefree(save_newnum);
1686 PL_numeric_underlying_is_standard = PL_numeric_standard;
1688 # ifdef USE_POSIX_2008_LOCALE
1690 /* We keep a special object for easy switching to */
1691 PL_underlying_numeric_obj = newlocale(LC_NUMERIC_MASK,
1693 PL_underlying_numeric_obj);
1697 DEBUG_L( PerlIO_printf(Perl_debug_log,
1698 "Called new_numeric with %s, PL_numeric_name=%s\n",
1699 newnum, PL_numeric_name));
1701 /* Keep LC_NUMERIC so that it has the C locale radix and thousands
1702 * separator. This is for XS modules, so they don't have to worry about
1703 * the radix being a non-dot. (Core operations that need the underlying
1704 * locale change to it temporarily). */
1705 if (PL_numeric_standard) {
1706 set_numeric_radix(0);
1709 set_numeric_standard();
1717 Perl_set_numeric_standard(pTHX)
1720 # ifdef USE_LOCALE_NUMERIC
1722 /* Unconditionally toggle the LC_NUMERIC locale to the current underlying
1725 * Most code should use the macro SET_NUMERIC_STANDARD() in perl.h
1726 * instead of calling this directly. The macro avoids calling this routine
1727 * if toggling isn't necessary according to our records (which could be
1728 * wrong if some XS code has changed the locale behind our back) */
1730 DEBUG_L(PerlIO_printf(Perl_debug_log,
1731 "Setting LC_NUMERIC locale to standard C\n"));
1733 void_setlocale_c(LC_NUMERIC, "C");
1734 PL_numeric_standard = TRUE;
1735 PL_numeric_underlying = PL_numeric_underlying_is_standard;
1736 set_numeric_radix(0);
1738 # endif /* USE_LOCALE_NUMERIC */
1743 Perl_set_numeric_underlying(pTHX)
1746 # ifdef USE_LOCALE_NUMERIC
1748 /* Unconditionally toggle the LC_NUMERIC locale to the current underlying
1751 * Most code should use the macro SET_NUMERIC_UNDERLYING() in perl.h
1752 * instead of calling this directly. The macro avoids calling this routine
1753 * if toggling isn't necessary according to our records (which could be
1754 * wrong if some XS code has changed the locale behind our back) */
1756 DEBUG_L(PerlIO_printf(Perl_debug_log, "Setting LC_NUMERIC locale to %s\n",
1759 void_setlocale_c(LC_NUMERIC, PL_numeric_name);
1760 PL_numeric_standard = PL_numeric_underlying_is_standard;
1761 PL_numeric_underlying = TRUE;
1762 set_numeric_radix(! PL_numeric_standard);
1764 # endif /* USE_LOCALE_NUMERIC */
1769 * Set up for a new ctype locale.
1772 S_new_ctype(pTHX_ const char *newctype)
1775 # ifndef USE_LOCALE_CTYPE
1777 PERL_UNUSED_ARG(newctype);
1778 PERL_UNUSED_CONTEXT;
1782 /* Called after each libc setlocale() call affecting LC_CTYPE, to tell
1783 * core Perl this and that 'newctype' is the name of the new locale.
1785 * This function sets up the folding arrays for all 256 bytes, assuming
1786 * that tofold() is tolc() since fold case is not a concept in POSIX,
1788 * Any code changing the locale (outside this file) should use
1789 * Perl_setlocale or POSIX::setlocale, which call this function. Therefore
1790 * this function should be called directly only from this file and from
1791 * POSIX::setlocale() */
1795 /* Don't check for problems if we are suppressing the warnings */
1796 bool check_for_problems = ckWARN_d(WARN_LOCALE) || UNLIKELY(DEBUG_L_TEST);
1797 bool maybe_utf8_turkic = FALSE;
1799 PERL_ARGS_ASSERT_NEW_CTYPE;
1801 DEBUG_L(PerlIO_printf(Perl_debug_log, "Entering new_ctype(%s)\n", newctype));
1803 /* We will replace any bad locale warning with 1) nothing if the new one is
1804 * ok; or 2) a new warning for the bad new locale */
1805 if (PL_warn_locale) {
1806 SvREFCNT_dec_NN(PL_warn_locale);
1807 PL_warn_locale = NULL;
1810 PL_in_utf8_CTYPE_locale = _is_cur_LC_category_utf8(LC_CTYPE);
1812 /* A UTF-8 locale gets standard rules. But note that code still has to
1813 * handle this specially because of the three problematic code points */
1814 if (PL_in_utf8_CTYPE_locale) {
1815 Copy(PL_fold_latin1, PL_fold_locale, 256, U8);
1817 /* UTF-8 locales can have special handling for 'I' and 'i' if they are
1818 * Turkic. Make sure these two are the only anomalies. (We don't
1819 * require towupper and towlower because they aren't in C89.) */
1821 # if defined(HAS_TOWUPPER) && defined (HAS_TOWLOWER)
1823 if (towupper('i') == 0x130 && towlower('I') == 0x131)
1827 if (toU8_UPPER_LC('i') == 'i' && toU8_LOWER_LC('I') == 'I')
1832 /* This is how we determine it really is Turkic */
1833 check_for_problems = TRUE;
1834 maybe_utf8_turkic = TRUE;
1838 /* We don't populate the other lists if a UTF-8 locale, but do check that
1839 * everything works as expected, unless checking turned off */
1840 if (check_for_problems || ! PL_in_utf8_CTYPE_locale) {
1841 /* Assume enough space for every character being bad. 4 spaces each
1842 * for the 94 printable characters that are output like "'x' "; and 5
1843 * spaces each for "'\\' ", "'\t' ", and "'\n' "; plus a terminating
1845 char bad_chars_list[ (94 * 4) + (3 * 5) + 1 ] = { '\0' };
1846 bool multi_byte_locale = FALSE; /* Assume is a single-byte locale
1848 unsigned int bad_count = 0; /* Count of bad characters */
1850 for (i = 0; i < 256; i++) {
1851 if (! PL_in_utf8_CTYPE_locale) {
1852 if (isU8_UPPER_LC(i))
1853 PL_fold_locale[i] = (U8) toU8_LOWER_LC(i);
1854 else if (isU8_LOWER_LC(i))
1855 PL_fold_locale[i] = (U8) toU8_UPPER_LC(i);
1857 PL_fold_locale[i] = (U8) i;
1860 /* If checking for locale problems, see if the native ASCII-range
1861 * printables plus \n and \t are in their expected categories in
1862 * the new locale. If not, this could mean big trouble, upending
1863 * Perl's and most programs' assumptions, like having a
1864 * metacharacter with special meaning become a \w. Fortunately,
1865 * it's very rare to find locales that aren't supersets of ASCII
1866 * nowadays. It isn't a problem for most controls to be changed
1867 * into something else; we check only \n and \t, though perhaps \r
1868 * could be an issue as well. */
1869 if ( check_for_problems
1870 && (isGRAPH_A(i) || isBLANK_A(i) || i == '\n'))
1872 bool is_bad = FALSE;
1873 char name[4] = { '\0' };
1875 /* Convert the name into a string */
1880 else if (i == '\n') {
1881 my_strlcpy(name, "\\n", sizeof(name));
1883 else if (i == '\t') {
1884 my_strlcpy(name, "\\t", sizeof(name));
1888 my_strlcpy(name, "' '", sizeof(name));
1891 /* Check each possibe class */
1892 if (UNLIKELY(cBOOL(isU8_ALPHANUMERIC_LC(i)) != cBOOL(isALPHANUMERIC_A(i)))) {
1894 DEBUG_L(PerlIO_printf(Perl_debug_log,
1895 "isalnum('%s') unexpectedly is %x\n",
1896 name, cBOOL(isU8_ALPHANUMERIC_LC(i))));
1898 if (UNLIKELY(cBOOL(isU8_ALPHA_LC(i)) != cBOOL(isALPHA_A(i)))) {
1900 DEBUG_L(PerlIO_printf(Perl_debug_log,
1901 "isalpha('%s') unexpectedly is %x\n",
1902 name, cBOOL(isU8_ALPHA_LC(i))));
1904 if (UNLIKELY(cBOOL(isU8_DIGIT_LC(i)) != cBOOL(isDIGIT_A(i)))) {
1906 DEBUG_L(PerlIO_printf(Perl_debug_log,
1907 "isdigit('%s') unexpectedly is %x\n",
1908 name, cBOOL(isU8_DIGIT_LC(i))));
1910 if (UNLIKELY(cBOOL(isU8_GRAPH_LC(i)) != cBOOL(isGRAPH_A(i)))) {
1912 DEBUG_L(PerlIO_printf(Perl_debug_log,
1913 "isgraph('%s') unexpectedly is %x\n",
1914 name, cBOOL(isU8_GRAPH_LC(i))));
1916 if (UNLIKELY(cBOOL(isU8_LOWER_LC(i)) != cBOOL(isLOWER_A(i)))) {
1918 DEBUG_L(PerlIO_printf(Perl_debug_log,
1919 "islower('%s') unexpectedly is %x\n",
1920 name, cBOOL(isU8_LOWER_LC(i))));
1922 if (UNLIKELY(cBOOL(isU8_PRINT_LC(i)) != cBOOL(isPRINT_A(i)))) {
1924 DEBUG_L(PerlIO_printf(Perl_debug_log,
1925 "isprint('%s') unexpectedly is %x\n",
1926 name, cBOOL(isU8_PRINT_LC(i))));
1928 if (UNLIKELY(cBOOL(isU8_PUNCT_LC(i)) != cBOOL(isPUNCT_A(i)))) {
1930 DEBUG_L(PerlIO_printf(Perl_debug_log,
1931 "ispunct('%s') unexpectedly is %x\n",
1932 name, cBOOL(isU8_PUNCT_LC(i))));
1934 if (UNLIKELY(cBOOL(isU8_SPACE_LC(i)) != cBOOL(isSPACE_A(i)))) {
1936 DEBUG_L(PerlIO_printf(Perl_debug_log,
1937 "isspace('%s') unexpectedly is %x\n",
1938 name, cBOOL(isU8_SPACE_LC(i))));
1940 if (UNLIKELY(cBOOL(isU8_UPPER_LC(i)) != cBOOL(isUPPER_A(i)))) {
1942 DEBUG_L(PerlIO_printf(Perl_debug_log,
1943 "isupper('%s') unexpectedly is %x\n",
1944 name, cBOOL(isU8_UPPER_LC(i))));
1946 if (UNLIKELY(cBOOL(isU8_XDIGIT_LC(i))!= cBOOL(isXDIGIT_A(i)))) {
1948 DEBUG_L(PerlIO_printf(Perl_debug_log,
1949 "isxdigit('%s') unexpectedly is %x\n",
1950 name, cBOOL(isU8_XDIGIT_LC(i))));
1952 if (UNLIKELY(toU8_LOWER_LC(i) != (int) toLOWER_A(i))) {
1954 DEBUG_L(PerlIO_printf(Perl_debug_log,
1955 "tolower('%s')=0x%x instead of the expected 0x%x\n",
1956 name, toU8_LOWER_LC(i), (int) toLOWER_A(i)));
1958 if (UNLIKELY(toU8_UPPER_LC(i) != (int) toUPPER_A(i))) {
1960 DEBUG_L(PerlIO_printf(Perl_debug_log,
1961 "toupper('%s')=0x%x instead of the expected 0x%x\n",
1962 name, toU8_UPPER_LC(i), (int) toUPPER_A(i)));
1964 if (UNLIKELY((i == '\n' && ! isCNTRL_LC(i)))) {
1966 DEBUG_L(PerlIO_printf(Perl_debug_log,
1967 "'\\n' (=%02X) is not a control\n", (int) i));
1970 /* Add to the list; Separate multiple entries with a blank */
1973 my_strlcat(bad_chars_list, " ", sizeof(bad_chars_list));
1975 my_strlcat(bad_chars_list, name, sizeof(bad_chars_list));
1981 if (bad_count == 2 && maybe_utf8_turkic) {
1983 *bad_chars_list = '\0';
1984 PL_fold_locale['I'] = 'I';
1985 PL_fold_locale['i'] = 'i';
1986 PL_in_utf8_turkic_locale = TRUE;
1987 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s is turkic\n", newctype));
1990 PL_in_utf8_turkic_locale = FALSE;
1995 /* We only handle single-byte locales (outside of UTF-8 ones; so if
1996 * this locale requires more than one byte, there are going to be
1998 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1999 "check_for_problems=%d, MB_CUR_MAX=%d\n",
2000 check_for_problems, (int) MB_CUR_MAX));
2002 if ( check_for_problems && MB_CUR_MAX > 1
2003 && ! PL_in_utf8_CTYPE_locale
2005 /* Some platforms return MB_CUR_MAX > 1 for even the "C"
2006 * locale. Just assume that the implementation for them (plus
2007 * for POSIX) is correct and the > 1 value is spurious. (Since
2008 * these are specially handled to never be considered UTF-8
2009 * locales, as long as this is the only problem, everything
2010 * should work fine */
2011 && strNE(newctype, "C") && strNE(newctype, "POSIX"))
2013 multi_byte_locale = TRUE;
2018 /* If we found problems and we want them output, do so */
2019 if ( (UNLIKELY(bad_count) || UNLIKELY(multi_byte_locale))
2020 && (LIKELY(ckWARN_d(WARN_LOCALE)) || UNLIKELY(DEBUG_L_TEST)))
2022 if (UNLIKELY(bad_count) && PL_in_utf8_CTYPE_locale) {
2023 PL_warn_locale = Perl_newSVpvf(aTHX_
2024 "Locale '%s' contains (at least) the following characters"
2025 " which have\nunexpected meanings: %s\nThe Perl program"
2026 " will use the expected meanings",
2027 newctype, bad_chars_list);
2030 PL_warn_locale = Perl_newSVpvf(aTHX_
2031 "Locale '%s' may not work well.%s%s%s\n",
2034 ? " Some characters in it are not recognized by"
2038 ? "\nThe following characters (and maybe others)"
2039 " may not have the same meaning as the Perl"
2040 " program expects:\n"
2048 # ifdef HAS_NL_LANGINFO
2050 Perl_sv_catpvf(aTHX_ PL_warn_locale, "; codeset=%s",
2051 /* parameter FALSE is a don't care here */
2052 my_nl_langinfo(CODESET, FALSE));
2056 Perl_sv_catpvf(aTHX_ PL_warn_locale, "\n");
2058 /* If we are actually in the scope of the locale or are debugging,
2059 * output the message now. If not in that scope, we save the
2060 * message to be output at the first operation using this locale,
2061 * if that actually happens. Most programs don't use locales, so
2062 * they are immune to bad ones. */
2063 if (IN_LC(LC_CTYPE) || UNLIKELY(DEBUG_L_TEST)) {
2065 /* The '0' below suppresses a bogus gcc compiler warning */
2066 Perl_warner(aTHX_ packWARN(WARN_LOCALE), SvPVX(PL_warn_locale),
2069 if (IN_LC(LC_CTYPE)) {
2070 SvREFCNT_dec_NN(PL_warn_locale);
2071 PL_warn_locale = NULL;
2077 # endif /* USE_LOCALE_CTYPE */
2082 Perl__warn_problematic_locale()
2085 # ifdef USE_LOCALE_CTYPE
2089 /* Internal-to-core function that outputs the message in PL_warn_locale,
2090 * and then NULLS it. Should be called only through the macro
2091 * CHECK_AND_WARN_PROBLEMATIC_LOCALE_ */
2093 if (PL_warn_locale) {
2094 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
2095 SvPVX(PL_warn_locale),
2096 0 /* dummy to avoid compiler warning */ );
2097 SvREFCNT_dec_NN(PL_warn_locale);
2098 PL_warn_locale = NULL;
2106 S_new_LC_ALL(pTHX_ const char *unused)
2110 /* LC_ALL updates all the things we care about. */
2112 PERL_UNUSED_ARG(unused);
2114 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
2115 if (update_functions[i]) {
2116 const char * this_locale = savepv(querylocale_i(i));
2117 update_functions[i](aTHX_ this_locale);
2118 Safefree(this_locale);
2124 S_new_collate(pTHX_ const char *newcoll)
2127 # ifndef USE_LOCALE_COLLATE
2129 PERL_UNUSED_ARG(newcoll);
2130 PERL_UNUSED_CONTEXT;
2134 /* Called after each libc setlocale() call affecting LC_COLLATE, to tell
2135 * core Perl this and that 'newcoll' is the name of the new locale.
2137 * The design of locale collation is that every locale change is given an
2138 * index 'PL_collation_ix'. The first time a string particpates in an
2139 * operation that requires collation while locale collation is active, it
2140 * is given PERL_MAGIC_collxfrm magic (via sv_collxfrm_flags()). That
2141 * magic includes the collation index, and the transformation of the string
2142 * by strxfrm(), q.v. That transformation is used when doing comparisons,
2143 * instead of the string itself. If a string changes, the magic is
2144 * cleared. The next time the locale changes, the index is incremented,
2145 * and so we know during a comparison that the transformation is not
2146 * necessarily still valid, and so is recomputed. Note that if the locale
2147 * changes enough times, the index could wrap (a U32), and it is possible
2148 * that a transformation would improperly be considered valid, leading to
2149 * an unlikely bug */
2152 if (PL_collation_name) {
2154 Safefree(PL_collation_name);
2155 PL_collation_name = NULL;
2157 PL_collation_standard = TRUE;
2158 is_standard_collation:
2159 PL_collxfrm_base = 0;
2160 PL_collxfrm_mult = 2;
2161 PL_in_utf8_COLLATE_locale = FALSE;
2162 PL_strxfrm_NUL_replacement = '\0';
2163 PL_strxfrm_max_cp = 0;
2167 /* If this is not the same locale as currently, set the new one up */
2168 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
2170 Safefree(PL_collation_name);
2171 PL_collation_name = savepv(newcoll);
2172 PL_collation_standard = isNAME_C_OR_POSIX(newcoll);
2173 if (PL_collation_standard) {
2174 goto is_standard_collation;
2177 PL_in_utf8_COLLATE_locale = _is_cur_LC_category_utf8(LC_COLLATE);
2178 PL_strxfrm_NUL_replacement = '\0';
2179 PL_strxfrm_max_cp = 0;
2181 /* A locale collation definition includes primary, secondary, tertiary,
2182 * etc. weights for each character. To sort, the primary weights are
2183 * used, and only if they compare equal, then the secondary weights are
2184 * used, and only if they compare equal, then the tertiary, etc.
2186 * strxfrm() works by taking the input string, say ABC, and creating an
2187 * output transformed string consisting of first the primary weights,
2188 * A¹B¹C¹ followed by the secondary ones, A²B²C²; and then the
2189 * tertiary, etc, yielding A¹B¹C¹ A²B²C² A³B³C³ .... Some characters
2190 * may not have weights at every level. In our example, let's say B
2191 * doesn't have a tertiary weight, and A doesn't have a secondary
2192 * weight. The constructed string is then going to be
2193 * A¹B¹C¹ B²C² A³C³ ....
2194 * This has the desired effect that strcmp() will look at the secondary
2195 * or tertiary weights only if the strings compare equal at all higher
2196 * priority weights. The spaces shown here, like in
2198 * are not just for readability. In the general case, these must
2199 * actually be bytes, which we will call here 'separator weights'; and
2200 * they must be smaller than any other weight value, but since these
2201 * are C strings, only the terminating one can be a NUL (some
2202 * implementations may include a non-NUL separator weight just before
2203 * the NUL). Implementations tend to reserve 01 for the separator
2204 * weights. They are needed so that a shorter string's secondary
2205 * weights won't be misconstrued as primary weights of a longer string,
2206 * etc. By making them smaller than any other weight, the shorter
2207 * string will sort first. (Actually, if all secondary weights are
2208 * smaller than all primary ones, there is no need for a separator
2209 * weight between those two levels, etc.)
2211 * The length of the transformed string is roughly a linear function of
2212 * the input string. It's not exactly linear because some characters
2213 * don't have weights at all levels. When we call strxfrm() we have to
2214 * allocate some memory to hold the transformed string. The
2215 * calculations below try to find coefficients 'm' and 'b' for this
2216 * locale so that m*x + b equals how much space we need, given the size
2217 * of the input string in 'x'. If we calculate too small, we increase
2218 * the size as needed, and call strxfrm() again, but it is better to
2219 * get it right the first time to avoid wasted expensive string
2220 * transformations. */
2223 /* We use the string below to find how long the tranformation of it
2224 * is. Almost all locales are supersets of ASCII, or at least the
2225 * ASCII letters. We use all of them, half upper half lower,
2226 * because if we used fewer, we might hit just the ones that are
2227 * outliers in a particular locale. Most of the strings being
2228 * collated will contain a preponderance of letters, and even if
2229 * they are above-ASCII, they are likely to have the same number of
2230 * weight levels as the ASCII ones. It turns out that digits tend
2231 * to have fewer levels, and some punctuation has more, but those
2232 * are relatively sparse in text, and khw believes this gives a
2233 * reasonable result, but it could be changed if experience so
2235 const char longer[] = "ABCDEFGHIJKLMnopqrstuvwxyz";
2236 char * x_longer; /* Transformed 'longer' */
2237 Size_t x_len_longer; /* Length of 'x_longer' */
2239 char * x_shorter; /* We also transform a substring of 'longer' */
2240 Size_t x_len_shorter;
2242 /* _mem_collxfrm() is used get the transformation (though here we
2243 * are interested only in its length). It is used because it has
2244 * the intelligence to handle all cases, but to work, it needs some
2245 * values of 'm' and 'b' to get it started. For the purposes of
2246 * this calculation we use a very conservative estimate of 'm' and
2247 * 'b'. This assumes a weight can be multiple bytes, enough to
2248 * hold any UV on the platform, and there are 5 levels, 4 weight
2249 * bytes, and a trailing NUL. */
2250 PL_collxfrm_base = 5;
2251 PL_collxfrm_mult = 5 * sizeof(UV);
2253 /* Find out how long the transformation really is */
2254 x_longer = _mem_collxfrm(longer,
2258 /* We avoid converting to UTF-8 in the
2259 * called function by telling it the
2260 * string is in UTF-8 if the locale is a
2261 * UTF-8 one. Since the string passed
2262 * here is invariant under UTF-8, we can
2263 * claim it's UTF-8 even though it isn't.
2265 PL_in_utf8_COLLATE_locale);
2268 /* Find out how long the transformation of a substring of 'longer'
2269 * is. Together the lengths of these transformations are
2270 * sufficient to calculate 'm' and 'b'. The substring is all of
2271 * 'longer' except the first character. This minimizes the chances
2272 * of being swayed by outliers */
2273 x_shorter = _mem_collxfrm(longer + 1,
2276 PL_in_utf8_COLLATE_locale);
2277 Safefree(x_shorter);
2279 /* If the results are nonsensical for this simple test, the whole
2280 * locale definition is suspect. Mark it so that locale collation
2281 * is not active at all for it. XXX Should we warn? */
2282 if ( x_len_shorter == 0
2283 || x_len_longer == 0
2284 || x_len_shorter >= x_len_longer)
2286 PL_collxfrm_mult = 0;
2287 PL_collxfrm_base = 0;
2290 SSize_t base; /* Temporary */
2292 /* We have both: m * strlen(longer) + b = x_len_longer
2293 * m * strlen(shorter) + b = x_len_shorter;
2294 * subtracting yields:
2295 * m * (strlen(longer) - strlen(shorter))
2296 * = x_len_longer - x_len_shorter
2297 * But we have set things up so that 'shorter' is 1 byte smaller
2298 * than 'longer'. Hence:
2299 * m = x_len_longer - x_len_shorter
2301 * But if something went wrong, make sure the multiplier is at
2304 if (x_len_longer > x_len_shorter) {
2305 PL_collxfrm_mult = (STRLEN) x_len_longer - x_len_shorter;
2308 PL_collxfrm_mult = 1;
2313 * but in case something has gone wrong, make sure it is
2315 base = x_len_longer - PL_collxfrm_mult * (sizeof(longer) - 1);
2320 /* Add 1 for the trailing NUL */
2321 PL_collxfrm_base = base + 1;
2324 DEBUG_L(PerlIO_printf(Perl_debug_log,
2325 "?UTF-8 locale=%d; x_len_shorter=%zu, "
2327 " collate multipler=%zu, collate base=%zu\n",
2328 PL_in_utf8_COLLATE_locale,
2329 x_len_shorter, x_len_longer,
2330 PL_collxfrm_mult, PL_collxfrm_base));
2334 # endif /* USE_LOCALE_COLLATE */
2338 #endif /* USE_LOCALE */
2342 #define USE_WSETLOCALE
2344 #ifdef USE_WSETLOCALE
2347 S_wrap_wsetlocale(pTHX_ int category, const char *locale) {
2354 MultiByteToWideChar(CP_UTF8, 0, locale, -1, NULL, 0);
2361 Newx(wlocale, req_size, wchar_t);
2362 if (!MultiByteToWideChar(CP_UTF8, 0, locale, -1, wlocale, req_size)) {
2371 wresult = _wsetlocale(category, wlocale);
2375 WideCharToMultiByte(CP_UTF8, 0, wresult, -1, NULL, 0, NULL, NULL);
2376 Newx(result, req_size, char);
2377 SAVEFREEPV(result); /* is there something better we can do here? */
2378 if (!WideCharToMultiByte(CP_UTF8, 0, wresult, -1,
2379 result, req_size, NULL, NULL)) {
2394 S_win32_setlocale(pTHX_ int category, const char* locale)
2396 /* This, for Windows, emulates POSIX setlocale() behavior. There is no
2397 * difference between the two unless the input locale is "", which normally
2398 * means on Windows to get the machine default, which is set via the
2399 * computer's "Regional and Language Options" (or its current equivalent).
2400 * In POSIX, it instead means to find the locale from the user's
2401 * environment. This routine changes the Windows behavior to first look in
2402 * the environment, and, if anything is found, use that instead of going to
2403 * the machine default. If there is no environment override, the machine
2404 * default is used, by calling the real setlocale() with "".
2406 * The POSIX behavior is to use the LC_ALL variable if set; otherwise to
2407 * use the particular category's variable if set; otherwise to use the LANG
2410 bool override_LC_ALL = FALSE;
2414 if (locale && strEQ(locale, "")) {
2418 locale = PerlEnv_getenv("LC_ALL");
2420 if (category == LC_ALL) {
2421 override_LC_ALL = TRUE;
2427 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
2428 if (category == categories[i]) {
2429 locale = PerlEnv_getenv(category_names[i]);
2434 locale = PerlEnv_getenv("LANG");
2450 #ifdef USE_WSETLOCALE
2451 result = S_wrap_wsetlocale(aTHX_ category, locale);
2453 result = setlocale(category, locale);
2455 DEBUG_L(STMT_START {
2456 PerlIO_printf(Perl_debug_log, "%s\n",
2457 setlocale_debug_string_r(category, locale, result));
2460 if (! override_LC_ALL) {
2464 /* Here the input category was LC_ALL, and we have set it to what is in the
2465 * LANG variable or the system default if there is no LANG. But these have
2466 * lower priority than the other LC_foo variables, so override it for each
2467 * one that is set. (If they are set to "", it means to use the same thing
2468 * we just set LC_ALL to, so can skip) */
2470 for (i = 0; i < LC_ALL_INDEX_; i++) {
2471 result = PerlEnv_getenv(category_names[i]);
2472 if (result && strNE(result, "")) {
2473 #ifdef USE_WSETLOCALE
2474 S_wrap_wsetlocale(aTHX_ categories[i], result);
2476 setlocale(categories[i], result);
2478 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s\n",
2479 setlocale_debug_string_i(i, result, "not captured")));
2483 result = setlocale(LC_ALL, NULL);
2484 DEBUG_L(STMT_START {
2485 PerlIO_printf(Perl_debug_log, "%s\n",
2486 setlocale_debug_string_c(LC_ALL, NULL, result));
2495 =for apidoc Perl_setlocale
2497 This is an (almost) drop-in replacement for the system L<C<setlocale(3)>>,
2498 taking the same parameters, and returning the same information, except that it
2499 returns the correct underlying C<LC_NUMERIC> locale. Regular C<setlocale> will
2500 instead return C<C> if the underlying locale has a non-dot decimal point
2501 character, or a non-empty thousands separator for displaying floating point
2502 numbers. This is because perl keeps that locale category such that it has a
2503 dot and empty separator, changing the locale briefly during the operations
2504 where the underlying one is required. C<Perl_setlocale> knows about this, and
2505 compensates; regular C<setlocale> doesn't.
2507 Another reason it isn't completely a drop-in replacement is that it is
2508 declared to return S<C<const char *>>, whereas the system setlocale omits the
2509 C<const> (presumably because its API was specified long ago, and can't be
2510 updated; it is illegal to change the information C<setlocale> returns; doing
2511 so leads to segfaults.)
2513 Finally, C<Perl_setlocale> works under all circumstances, whereas plain
2514 C<setlocale> can be completely ineffective on some platforms under some
2517 C<Perl_setlocale> should not be used to change the locale except on systems
2518 where the predefined variable C<${^SAFE_LOCALES}> is 1. On some such systems,
2519 the system C<setlocale()> is ineffective, returning the wrong information, and
2520 failing to actually change the locale. C<Perl_setlocale>, however works
2521 properly in all circumstances.
2523 The return points to a per-thread static buffer, which is overwritten the next
2524 time C<Perl_setlocale> is called from the same thread.
2531 Perl_setlocale(const int category, const char * locale)
2533 /* This wraps POSIX::setlocale() */
2537 PERL_UNUSED_ARG(category);
2538 PERL_UNUSED_ARG(locale);
2544 const char * retval;
2547 DEBUG_L(PerlIO_printf(Perl_debug_log,
2548 "Entering Perl_setlocale(%d, \"%s\")\n",
2551 /* A NULL locale means only query what the current one is. */
2552 if (locale == NULL) {
2554 # ifndef USE_LOCALE_NUMERIC
2556 /* Without LC_NUMERIC, it's trivial; we just return the value */
2557 return querylocale_r(category);
2561 /* We have the LC_NUMERIC name saved, because we are normally switched
2562 * into the C locale (or equivalent) for it. */
2563 if (category == LC_NUMERIC) {
2564 DEBUG_L(PerlIO_printf(Perl_debug_log,
2565 "Perl_setlocale(LC_NUMERIC, NULL) returning stashed '%s'\n",
2568 /* We don't have to copy this return value, as it is a per-thread
2569 * variable, and won't change until a future setlocale */
2570 return PL_numeric_name;
2575 /* Without LC_ALL, just return the value */
2576 return querylocale_r(category);
2580 /* Here, LC_ALL is available on this platform. It's the one
2581 * complicating category (because it can contain a toggled LC_NUMERIC
2582 * value), for all the remaining ones (we took care of LC_NUMERIC
2583 * above), just return the value */
2584 if (category != LC_ALL) {
2585 return querylocale_r(category);
2588 bool toggled = FALSE;
2590 /* For an LC_ALL query, switch back to the underlying numeric locale
2591 * (if we aren't there already) so as to get the correct results. Our
2592 * records for all the other categories are valid without switching */
2593 if (! PL_numeric_underlying) {
2594 set_numeric_underlying();
2598 retval = querylocale_c(LC_ALL);
2602 /* This toggling back could destroy 'retval' */
2603 retval = save_to_buffer(retval,
2604 &PL_setlocale_buf, &PL_setlocale_bufsize, 0);
2605 set_numeric_standard();
2608 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n",
2609 setlocale_debug_string_r(category, locale, retval)));
2612 # endif /* Has LC_ALL */
2613 # endif /* Has LC_NUMERIC */
2615 } /* End of querying the current locale */
2617 unsigned int cat_index = get_category_index(category, NULL);
2618 retval = save_to_buffer(setlocale_i(cat_index, locale),
2619 &PL_setlocale_buf, &PL_setlocale_bufsize, 0);
2621 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n",
2622 setlocale_debug_string_i(cat_index, locale, "NULL")));
2626 /* Now that have changed locales, we have to update our records to
2627 * correspond. Only certain categories have extra work to update. */
2628 if (update_functions[cat_index]) {
2629 update_functions[cat_index](aTHX_ retval);
2632 DEBUG_L(PerlIO_printf(Perl_debug_log, "returning '%s'\n", retval));
2640 PERL_STATIC_INLINE const char *
2641 S_save_to_buffer(const char * string, const char **buf, Size_t *buf_size,
2642 const Size_t offset)
2644 /* Copy the NUL-terminated 'string' to 'buf' + 'offset'. 'buf' has size
2645 * 'buf_size', growing it if necessary */
2649 PERL_ARGS_ASSERT_SAVE_TO_BUFFER;
2655 string_size = strlen(string) + offset + 1;
2657 if (*buf_size == 0) {
2658 Newx(*buf, string_size, char);
2659 *buf_size = string_size;
2661 else if (string_size > *buf_size) {
2662 Renew(*buf, string_size, char);
2663 *buf_size = string_size;
2666 Copy(string, *buf + offset, string_size - offset, char);
2672 =for apidoc Perl_langinfo
2674 This is an (almost) drop-in replacement for the system C<L<nl_langinfo(3)>>,
2675 taking the same C<item> parameter values, and returning the same information.
2676 But it is more thread-safe than regular C<nl_langinfo()>, and hides the quirks
2677 of Perl's locale handling from your code, and can be used on systems that lack
2678 a native C<nl_langinfo>.
2686 The reason it isn't quite a drop-in replacement is actually an advantage. The
2687 only difference is that it returns S<C<const char *>>, whereas plain
2688 C<nl_langinfo()> returns S<C<char *>>, but you are (only by documentation)
2689 forbidden to write into the buffer. By declaring this C<const>, the compiler
2690 enforces this restriction, so if it is violated, you know at compilation time,
2691 rather than getting segfaults at runtime.
2695 It delivers the correct results for the C<RADIXCHAR> and C<THOUSEP> items,
2696 without you having to write extra code. The reason for the extra code would be
2697 because these are from the C<LC_NUMERIC> locale category, which is normally
2698 kept set by Perl so that the radix is a dot, and the separator is the empty
2699 string, no matter what the underlying locale is supposed to be, and so to get
2700 the expected results, you have to temporarily toggle into the underlying
2701 locale, and later toggle back. (You could use plain C<nl_langinfo> and
2702 C<L</STORE_LC_NUMERIC_FORCE_TO_UNDERLYING>> for this but then you wouldn't get
2703 the other advantages of C<Perl_langinfo()>; not keeping C<LC_NUMERIC> in the C
2704 (or equivalent) locale would break a lot of CPAN, which is expecting the radix
2705 (decimal point) character to be a dot.)
2709 The system function it replaces can have its static return buffer trashed,
2710 not only by a subsequent call to that function, but by a C<freelocale>,
2711 C<setlocale>, or other locale change. The returned buffer of this function is
2712 not changed until the next call to it, so the buffer is never in a trashed
2717 Its return buffer is per-thread, so it also is never overwritten by a call to
2718 this function from another thread; unlike the function it replaces.
2722 But most importantly, it works on systems that don't have C<nl_langinfo>, such
2723 as Windows, hence makes your code more portable. Of the fifty-some possible
2724 items specified by the POSIX 2008 standard,
2725 L<http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/langinfo.h.html>,
2726 only one is completely unimplemented, though on non-Windows platforms, another
2727 significant one is also not implemented). It uses various techniques to
2728 recover the other items, including calling C<L<localeconv(3)>>, and
2729 C<L<strftime(3)>>, both of which are specified in C89, so should be always be
2730 available. Later C<strftime()> versions have additional capabilities; C<""> is
2731 returned for those not available on your system.
2733 It is important to note that when called with an item that is recovered by
2734 using C<localeconv>, the buffer from any previous explicit call to
2735 C<localeconv> will be overwritten. This means you must save that buffer's
2736 contents if you need to access them after a call to this function. (But note
2737 that you might not want to be using C<localeconv()> directly anyway, because of
2738 issues like the ones listed in the second item of this list (above) for
2739 C<RADIXCHAR> and C<THOUSEP>. You can use the methods given in L<perlcall> to
2740 call L<POSIX/localeconv> and avoid all the issues, but then you have a hash to
2743 The details for those items which may deviate from what this emulation returns
2744 and what a native C<nl_langinfo()> would return are specified in
2749 When using C<Perl_langinfo> on systems that don't have a native
2750 C<nl_langinfo()>, you must
2752 #include "perl_langinfo.h"
2754 before the C<perl.h> C<#include>. You can replace your C<langinfo.h>
2755 C<#include> with this one. (Doing it this way keeps out the symbols that plain
2756 C<langinfo.h> would try to import into the namespace for code that doesn't need
2759 The original impetus for C<Perl_langinfo()> was so that code that needs to
2760 find out the current currency symbol, floating point radix character, or digit
2761 grouping separator can use, on all systems, the simpler and more
2762 thread-friendly C<nl_langinfo> API instead of C<L<localeconv(3)>> which is a
2763 pain to make thread-friendly. For other fields returned by C<localeconv>, it
2764 is better to use the methods given in L<perlcall> to call
2765 L<C<POSIX::localeconv()>|POSIX/localeconv>, which is thread-friendly.
2772 #ifdef HAS_NL_LANGINFO
2773 Perl_langinfo(const nl_item item)
2775 Perl_langinfo(const int item)
2778 return my_nl_langinfo(item, TRUE);
2782 # ifdef HAS_NL_LANGINFO
2783 S_my_nl_langinfo(const nl_item item, bool toggle)
2785 S_my_nl_langinfo(const int item, bool toggle)
2789 const char * retval;
2791 # ifdef USE_LOCALE_NUMERIC
2793 /* We only need to toggle into the underlying LC_NUMERIC locale for these
2794 * two items, and only if not already there */
2795 if (toggle && (( item != RADIXCHAR && item != THOUSEP)
2796 || PL_numeric_underlying))
2798 # endif /* No toggling needed if not using LC_NUMERIC */
2802 /*--------------------------------------------------------------------------*/
2803 /* Above is the common beginning to all the implementations of my_langinfo().
2804 * Below are the various completions */
2805 # if defined(HAS_NL_LANGINFO) /* nl_langinfo() is available. */
2806 # if ! defined(HAS_THREAD_SAFE_NL_LANGINFO_L) \
2807 || ! defined(USE_POSIX_2008_LOCALE)
2809 /* Here, use plain nl_langinfo(), switching to the underlying LC_NUMERIC
2810 * for those items dependent on it. This must be copied to a buffer before
2811 * switching back, as some systems destroy the buffer when setlocale() is
2815 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
2818 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
2821 /* Prevent interference from another thread executing this code
2825 /* Copy to a per-thread buffer, which is also one that won't be
2826 * destroyed by a subsequent setlocale(), such as the
2827 * RESTORE_LC_NUMERIC may do just below. */
2828 retval = save_to_buffer(nl_langinfo(item),
2829 ((const char **) &PL_langinfo_buf),
2830 &PL_langinfo_bufsize, 0);
2834 RESTORE_LC_NUMERIC();
2837 /*--------------------------------------------------------------------------*/
2838 # else /* Use nl_langinfo_l(), avoiding both a mutex and changing the
2842 locale_t cur = use_curlocale_scratch();
2844 # ifdef USE_LOCALE_NUMERIC
2847 if (PL_underlying_numeric_obj) {
2848 cur = PL_underlying_numeric_obj;
2851 cur = newlocale(LC_NUMERIC_MASK, PL_numeric_name, cur);
2857 /* We have to save it to a buffer, because the freelocale() just below
2858 * can invalidate the internal one */
2859 retval = save_to_buffer(nl_langinfo_l(item, cur),
2860 ((const char **) &PL_langinfo_buf),
2861 &PL_langinfo_bufsize, 0);
2866 /* We can return 'yes' and 'no' even if we didn't get a result */
2867 if (strEQ(retval, "")) {
2868 if (item == YESSTR) {
2871 if (item == NOSTR) {
2877 /*--------------------------------------------------------------------------*/
2878 # else /* Below, emulate nl_langinfo as best we can */
2882 # ifdef HAS_LOCALECONV
2884 const struct lconv* lc;
2886 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
2888 # ifdef TS_W32_BROKEN_LOCALECONV
2890 const char * save_global;
2891 const char * save_thread;
2899 # ifdef HAS_STRFTIME
2902 bool return_format = FALSE; /* Return the %format, not the value */
2903 const char * format;
2907 /* We copy the results to a per-thread buffer, even if not
2908 * multi-threaded. This is in part to simplify this code, and partly
2909 * because we need a buffer anyway for strftime(), and partly because a
2910 * call of localeconv() could otherwise wipe out the buffer, and the
2911 * programmer would not be expecting this, as this is a nl_langinfo()
2912 * substitute after all, so s/he might be thinking their localeconv()
2913 * is safe until another localeconv() call. */
2918 /* This is unimplemented */
2919 case ERA: /* For use with strftime() %E modifier */
2924 /* We use only an English set, since we don't know any more */
2925 case YESEXPR: return "^[+1yY]";
2926 case YESSTR: return "yes";
2927 case NOEXPR: return "^[-0nN]";
2928 case NOSTR: return "no";
2934 /* On non-windows, this is unimplemented, in part because of
2935 * inconsistencies between vendors. The Darwin native
2936 * nl_langinfo() implementation simply looks at everything past
2937 * any dot in the name, but that doesn't work for other
2938 * vendors. Many Linux locales that don't have UTF-8 in their
2939 * names really are UTF-8, for example; z/OS locales that do
2940 * have UTF-8 in their names, aren't really UTF-8 */
2945 { /* But on Windows, the name does seem to be consistent, so
2950 const char * name = querylocale_c(LC_CTYPE);
2952 if (isNAME_C_OR_POSIX(name)) {
2953 return "ANSI_X3.4-1968";
2956 /* Find the dot in the locale name */
2957 first = (const char *) strchr(name, '.');
2963 /* Look at everything past the dot */
2968 if (! isDIGIT(*p)) {
2975 /* Here everything past the dot is a digit. Treat it as a
2977 retval = save_to_buffer("CP",
2978 ((const char **) &PL_langinfo_buf),
2979 &PL_langinfo_bufsize, 0);
2980 offset = STRLENs("CP");
2984 retval = save_to_buffer(first,
2985 ((const char **) &PL_langinfo_buf),
2986 &PL_langinfo_bufsize, offset);
2992 # ifdef HAS_LOCALECONV
2996 /* We don't bother with localeconv_l() because any system that
2997 * has it is likely to also have nl_langinfo() */
2999 LOCALECONV_LOCK; /* Prevent interference with other threads
3000 using localeconv() */
3002 # ifdef TS_W32_BROKEN_LOCALECONV
3004 /* This is a workaround for a Windows bug prior to VS 15.
3005 * What we do here is, while locked, switch to the global
3006 * locale so localeconv() works; then switch back just before
3007 * the unlock. This can screw things up if some thread is
3008 * already using the global locale while assuming no other is.
3009 * A different workaround would be to call GetCurrencyFormat on
3010 * a known value, and parse it; patches welcome
3012 * We have to use LC_ALL instead of LC_MONETARY because of
3013 * another bug in Windows */
3015 save_thread = savepv(querylocale_c(LC_ALL));
3016 _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
3017 save_global= savepv(querylocale_c(LC_ALL));
3018 void_setlocale_c(LC_ALL, save_thread);
3025 || ! lc->currency_symbol
3026 || strEQ("", lc->currency_symbol))
3032 /* Leave the first spot empty to be filled in below */
3033 retval = save_to_buffer(lc->currency_symbol,
3034 ((const char **) &PL_langinfo_buf),
3035 &PL_langinfo_bufsize, 1);
3036 if (lc->mon_decimal_point && strEQ(lc->mon_decimal_point, ""))
3037 { /* khw couldn't figure out how the localedef specifications
3038 would show that the $ should replace the radix; this is
3039 just a guess as to how it might work.*/
3040 PL_langinfo_buf[0] = '.';
3042 else if (lc->p_cs_precedes) {
3043 PL_langinfo_buf[0] = '-';
3046 PL_langinfo_buf[0] = '+';
3049 # ifdef TS_W32_BROKEN_LOCALECONV
3051 void_setlocale_c(LC_ALL, save_global);
3052 _configthreadlocale(_ENABLE_PER_THREAD_LOCALE);
3053 void_setlocale_c(LC_ALL, save_thread);
3054 Safefree(save_global);
3055 Safefree(save_thread);
3062 # ifdef TS_W32_BROKEN_LOCALECONV
3066 /* For this, we output a known simple floating point number to
3067 * a buffer, and parse it, looking for the radix */
3070 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
3073 if (PL_langinfo_bufsize < 10) {
3074 PL_langinfo_bufsize = 10;
3075 Renew(PL_langinfo_buf, PL_langinfo_bufsize, char);
3078 needed_size = my_snprintf(PL_langinfo_buf, PL_langinfo_bufsize,
3080 if (needed_size >= (int) PL_langinfo_bufsize) {
3081 PL_langinfo_bufsize = needed_size + 1;
3082 Renew(PL_langinfo_buf, PL_langinfo_bufsize, char);
3084 = my_snprintf(PL_langinfo_buf, PL_langinfo_bufsize,
3086 assert(needed_size < (int) PL_langinfo_bufsize);
3089 ptr = PL_langinfo_buf;
3090 e = PL_langinfo_buf + PL_langinfo_bufsize;
3093 while (ptr < e && *ptr != '1') {
3100 while (ptr < e && *ptr != '5') {
3104 /* Everything in between is the radix string */
3106 PL_langinfo_buf[0] = '?';
3107 PL_langinfo_buf[1] = '\0';
3111 Move(item_start, PL_langinfo_buf, ptr - PL_langinfo_buf,
3116 RESTORE_LC_NUMERIC();
3119 retval = PL_langinfo_buf;
3124 case RADIXCHAR: /* No special handling needed */
3131 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
3134 LOCALECONV_LOCK; /* Prevent interference with other threads
3135 using localeconv() */
3137 # ifdef TS_W32_BROKEN_LOCALECONV
3139 /* This should only be for the thousands separator. A
3140 * different work around would be to use GetNumberFormat on a
3141 * known value and parse the result to find the separator */
3142 save_thread = savepv(querylocale_c(LC_ALL));
3143 _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
3144 save_global = savepv(querylocale_c(LC_ALL));
3145 void_setlocale_c(LC_ALL, save_thread);
3147 /* This is the start of code that for broken Windows replaces
3148 * the above and below code, and instead calls
3149 * GetNumberFormat() and then would parse that to find the
3150 * thousands separator. It needs to handle UTF-16 vs -8
3153 needed_size = GetNumberFormatEx(PL_numeric_name, 0, "1234.5",
3154 NULL, PL_langinfo_buf, PL_langinfo_bufsize);
3155 DEBUG_L(PerlIO_printf(Perl_debug_log,
3156 "return from GetNumber, count=%d, val=%s\n",
3157 needed_size, PL_langinfo_buf));
3167 temp = (item == RADIXCHAR)
3169 : lc->thousands_sep;
3175 retval = save_to_buffer(temp,
3176 ((const char **) &PL_langinfo_buf),
3177 &PL_langinfo_bufsize, 0);
3179 # ifdef TS_W32_BROKEN_LOCALECONV
3181 void_setlocale_c(LC_ALL, save_global);
3182 _configthreadlocale(_ENABLE_PER_THREAD_LOCALE);
3183 void_setlocale_c(LC_ALL, save_thread);
3184 Safefree(save_global);
3185 Safefree(save_thread);
3192 RESTORE_LC_NUMERIC();
3198 # ifdef HAS_STRFTIME
3200 /* These are defined by C89, so we assume that strftime supports
3201 * them, and so are returned unconditionally; they may not be what
3202 * the locale actually says, but should give good enough results
3203 * for someone using them as formats (as opposed to trying to parse
3204 * them to figure out what the locale says). The other format
3205 * items are actually tested to verify they work on the platform */
3206 case D_FMT: return "%x";
3207 case T_FMT: return "%X";
3208 case D_T_FMT: return "%c";
3210 /* These formats are only available in later strfmtime's */
3211 case ERA_D_FMT: case ERA_T_FMT: case ERA_D_T_FMT: case T_FMT_AMPM:
3213 /* The rest can be gotten from most versions of strftime(). */
3214 case ABDAY_1: case ABDAY_2: case ABDAY_3:
3215 case ABDAY_4: case ABDAY_5: case ABDAY_6: case ABDAY_7:
3217 case AM_STR: case PM_STR:
3218 case ABMON_1: case ABMON_2: case ABMON_3: case ABMON_4:
3219 case ABMON_5: case ABMON_6: case ABMON_7: case ABMON_8:
3220 case ABMON_9: case ABMON_10: case ABMON_11: case ABMON_12:
3221 case DAY_1: case DAY_2: case DAY_3: case DAY_4:
3222 case DAY_5: case DAY_6: case DAY_7:
3223 case MON_1: case MON_2: case MON_3: case MON_4:
3224 case MON_5: case MON_6: case MON_7: case MON_8:
3225 case MON_9: case MON_10: case MON_11: case MON_12:
3227 init_tm(&tm); /* Precaution against core dumps */
3231 tm.tm_year = 2017 - 1900;
3235 GCC_DIAG_IGNORE_STMT(-Wimplicit-fallthrough);
3239 locale_panic_(Perl_form(aTHX_ "switch case: %d problem", item));
3240 NOT_REACHED; /* NOTREACHED */
3242 case PM_STR: tm.tm_hour = 18;
3247 case ABDAY_7: tm.tm_wday++;
3248 case ABDAY_6: tm.tm_wday++;
3249 case ABDAY_5: tm.tm_wday++;
3250 case ABDAY_4: tm.tm_wday++;
3251 case ABDAY_3: tm.tm_wday++;
3252 case ABDAY_2: tm.tm_wday++;
3257 case DAY_7: tm.tm_wday++;
3258 case DAY_6: tm.tm_wday++;
3259 case DAY_5: tm.tm_wday++;
3260 case DAY_4: tm.tm_wday++;
3261 case DAY_3: tm.tm_wday++;
3262 case DAY_2: tm.tm_wday++;
3267 case ABMON_12: tm.tm_mon++;
3268 case ABMON_11: tm.tm_mon++;
3269 case ABMON_10: tm.tm_mon++;
3270 case ABMON_9: tm.tm_mon++;
3271 case ABMON_8: tm.tm_mon++;
3272 case ABMON_7: tm.tm_mon++;
3273 case ABMON_6: tm.tm_mon++;
3274 case ABMON_5: tm.tm_mon++;
3275 case ABMON_4: tm.tm_mon++;
3276 case ABMON_3: tm.tm_mon++;
3277 case ABMON_2: tm.tm_mon++;
3282 case MON_12: tm.tm_mon++;
3283 case MON_11: tm.tm_mon++;
3284 case MON_10: tm.tm_mon++;
3285 case MON_9: tm.tm_mon++;
3286 case MON_8: tm.tm_mon++;
3287 case MON_7: tm.tm_mon++;
3288 case MON_6: tm.tm_mon++;
3289 case MON_5: tm.tm_mon++;
3290 case MON_4: tm.tm_mon++;
3291 case MON_3: tm.tm_mon++;
3292 case MON_2: tm.tm_mon++;
3299 return_format = TRUE;
3304 return_format = TRUE;
3309 return_format = TRUE;
3314 return_format = TRUE;
3319 format = "%Ow"; /* Find the alternate digit for 0 */
3323 GCC_DIAG_RESTORE_STMT;
3325 /* We can't use my_strftime() because it doesn't look at
3327 while (0 == strftime(PL_langinfo_buf, PL_langinfo_bufsize,
3330 /* A zero return means one of:
3331 * a) there wasn't enough space in PL_langinfo_buf
3332 * b) the format, like a plain %p, returns empty
3333 * c) it was an illegal format, though some
3334 * implementations of strftime will just return the
3335 * illegal format as a plain character sequence.
3337 * To quickly test for case 'b)', try again but precede
3338 * the format with a plain character. If that result is
3339 * still empty, the problem is either 'a)' or 'c)' */
3341 Size_t format_size = strlen(format) + 1;
3342 Size_t mod_size = format_size + 1;
3346 Newx(mod_format, mod_size, char);
3347 Newx(temp_result, PL_langinfo_bufsize, char);
3349 my_strlcpy(mod_format + 1, format, mod_size);
3350 len = strftime(temp_result,
3351 PL_langinfo_bufsize,
3353 Safefree(mod_format);
3354 Safefree(temp_result);
3356 /* If 'len' is non-zero, it means that we had a case like
3357 * %p which means the current locale doesn't use a.m. or
3358 * p.m., and that is valid */
3361 /* Here, still didn't work. If we get well beyond a
3362 * reasonable size, bail out to prevent an infinite
3365 if (PL_langinfo_bufsize > 100 * format_size) {
3366 *PL_langinfo_buf = '\0';
3369 /* Double the buffer size to retry; Add 1 in case
3370 * original was 0, so we aren't stuck at 0. */
3371 PL_langinfo_bufsize *= 2;
3372 PL_langinfo_bufsize++;
3373 Renew(PL_langinfo_buf, PL_langinfo_bufsize, char);
3381 /* Here, we got a result.
3383 * If the item is 'ALT_DIGITS', 'PL_langinfo_buf' contains the
3384 * alternate format for wday 0. If the value is the same as the
3385 * normal 0, there isn't an alternate, so clear the buffer. */
3386 if (item == ALT_DIGITS && strEQ(PL_langinfo_buf, "0")) {
3387 *PL_langinfo_buf = '\0';
3390 /* ALT_DIGITS is problematic. Experiments on it showed that
3391 * strftime() did not always work properly when going from alt-9 to
3392 * alt-10. Only a few locales have this item defined, and in all
3393 * of them on Linux that khw was able to find, nl_langinfo() merely
3394 * returned the alt-0 character, possibly doubled. Most Unicode
3395 * digits are in blocks of 10 consecutive code points, so that is
3396 * sufficient information for such scripts, as we can infer alt-1,
3397 * alt-2, .... But for a Japanese locale, a CJK ideographic 0 is
3398 * returned, and the CJK digits are not in code point order, so you
3399 * can't really infer anything. The localedef for this locale did
3400 * specify the succeeding digits, so that strftime() works properly
3401 * on them, without needing to infer anything. But the
3402 * nl_langinfo() return did not give sufficient information for the
3403 * caller to understand what's going on. So until there is
3404 * evidence that it should work differently, this returns the alt-0
3405 * string for ALT_DIGITS.
3407 * wday was chosen because its range is all a single digit.
3408 * Things like tm_sec have two digits as the minimum: '00' */
3410 retval = PL_langinfo_buf;
3412 /* If to return the format, not the value, overwrite the buffer
3413 * with it. But some strftime()s will keep the original format if
3414 * illegal, so change those to "" */
3415 if (return_format) {
3416 if (strEQ(PL_langinfo_buf, format)) {
3417 *PL_langinfo_buf = '\0';
3420 retval = save_to_buffer(format,
3421 ((const char **) &PL_langinfo_buf),
3422 &PL_langinfo_bufsize, 0);
3436 /*--------------------------------------------------------------------------*/
3440 * Initialize locale awareness.
3443 Perl_init_i18nl10n(pTHX_ int printwarn)
3447 * 0 if not to output warning when setup locale is bad
3448 * 1 if to output warning based on value of PERL_BADLANG
3449 * >1 if to output regardless of PERL_BADLANG
3452 * 1 = set ok or not applicable,
3453 * 0 = fallback to a locale of lower priority
3454 * -1 = fallback to all locales failed, not even to the C locale
3456 * Under -DDEBUGGING, if the environment variable PERL_DEBUG_LOCALE_INIT is
3457 * set, debugging information is output.
3459 * This looks more complicated than it is, mainly due to the #ifdefs and
3462 * Besides some asserts, data structure initialization, and specific
3463 * platform complications, this routine is effectively represented by this
3466 * setlocale(LC_ALL, ""); x
3467 * foreach (subcategory) { x
3468 * curlocales[f(subcategory)] = setlocale(subcategory, NULL); x
3470 * if (platform_so_requires) {
3471 * foreach (subcategory) {
3472 * PL_curlocales[f(subcategory)] = curlocales[f(subcategory)]
3475 * foreach (subcategory) {
3476 * if (needs_special_handling[f(subcategory)] &this_subcat_handler
3479 * This sets all the categories to the values in the current environment,
3480 * saves them temporarily in curlocales[] until they can be handled and/or
3481 * on some platforms saved in a per-thread array PL_curlocales[].
3483 * f(foo) is a mapping from the opaque system category numbers to small
3484 * non-negative integers used most everywhere in this file as indices into
3485 * arrays (such as curlocales[]) so the program doesn't have to otherwise
3486 * deal with the opaqueness.
3488 * If the platform doesn't have LC_ALL, the lines marked 'x' above are
3489 * effectively replaced by:
3490 * foreach (subcategory) { y
3491 * curlocales[f(subcategory)] = setlocale(subcategory, ""); y
3494 * The only differences being the lack of an LC_ALL call, and using ""
3495 * instead of NULL in the setlocale calls.
3497 * But there are, of course, complications.
3499 * it has to deal with if this is an embedded perl, whose locale doesn't
3500 * come from the environment, but has been set up by the caller. This is
3501 * pretty simply handled: the "" in the setlocale calls is not a string
3502 * constant, but a variable which is set to NULL in the embedded case.
3504 * But the major complication is handling failure and doing fallback. All
3505 * the code marked 'x' or 'y' above is actually enclosed in an outer loop,
3506 * using the array trial_locales[]. On entry, trial_locales[] is
3507 * initialized to just one entry, containing the NULL or "" locale argument
3508 * shown above. If, as is almost always the case, everything works, it
3509 * exits after just the one iteration, going on to the next step.
3511 * But if there is a failure, the code tries its best to honor the
3512 * environment as much as possible. It self-modifies trial_locales[] to
3513 * have more elements, one for each of the POSIX-specified settings from
3514 * the environment, such as LANG, ending in the ultimate fallback, the C
3515 * locale. Thus if there is something bogus with a higher priority
3516 * environment variable, it will try with the next highest, until something
3517 * works. If everything fails, it limps along with whatever state it got
3520 * A further complication is that Windows has an additional fallback, the
3521 * user-default ANSI code page obtained from the operating system. This is
3522 * added as yet another loop iteration, just before the final "C"
3524 * A slight complication is that in embedded Perls, the locale may already
3525 * be set-up, and we don't want to get it from the normal environment
3526 * variables. This is handled by having a special environment variable
3527 * indicate we're in this situation. We simply set setlocale's 2nd
3528 * parameter to be a NULL instead of "". That indicates to setlocale that
3529 * it is not to change anything, but to return the current value,
3530 * effectively initializing perl's db to what the locale already is.
3532 * We play the same trick with NULL if a LC_ALL succeeds. We call
3533 * setlocale() on the individual categores with NULL to get their existing
3534 * values for our db, instead of trying to change them.
3541 PERL_UNUSED_ARG(printwarn);
3543 #else /* USE_LOCALE */
3546 const char * const language = PerlEnv_getenv("LANGUAGE");
3550 /* NULL uses the existing already set up locale */
3551 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
3554 typedef struct trial_locales_struct_s {
3555 const char* trial_locale;
3556 const char* fallback_desc;
3557 const char* fallback_name;
3558 } trial_locales_struct;
3559 /* 5 = 1 each for "", LC_ALL, LANG, (Win32) system default locale, C */
3560 trial_locales_struct trial_locales[5];
3561 unsigned int trial_locales_count;
3562 const char * const lc_all = PerlEnv_getenv("LC_ALL");
3563 const char * const lang = PerlEnv_getenv("LANG");
3564 bool setlocale_failure = FALSE;
3567 /* A later getenv() could zap this, so only use here */
3568 const char * const bad_lang_use_once = PerlEnv_getenv("PERL_BADLANG");
3570 const bool locwarn = (printwarn > 1
3572 && ( ! bad_lang_use_once
3574 /* disallow with "" or "0" */
3576 && strNE("0", bad_lang_use_once)))));
3578 /* setlocale() return vals; not copied so must be looked at immediately */
3579 const char * sl_result[NOMINAL_LC_ALL_INDEX + 1];
3581 /* current locale for given category; should have been copied so aren't
3583 const char * curlocales[NOMINAL_LC_ALL_INDEX + 1];
3586 # define DEBUG_LOCALE_INIT(a,b,c)
3589 DEBUG_INITIALIZATION_set(cBOOL(PerlEnv_getenv("PERL_DEBUG_LOCALE_INIT")));
3591 # define DEBUG_LOCALE_INIT(cat_index, locale, result) \
3592 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n", \
3593 setlocale_debug_string_i(cat_index, locale, result)));
3595 /* Make sure the parallel arrays are properly set up */
3596 # ifdef USE_LOCALE_NUMERIC
3597 assert(categories[LC_NUMERIC_INDEX_] == LC_NUMERIC);
3598 assert(strEQ(category_names[LC_NUMERIC_INDEX_], "LC_NUMERIC"));
3599 # ifdef USE_POSIX_2008_LOCALE
3600 assert(category_masks[LC_NUMERIC_INDEX_] == LC_NUMERIC_MASK);
3603 # ifdef USE_LOCALE_CTYPE
3604 assert(categories[LC_CTYPE_INDEX_] == LC_CTYPE);
3605 assert(strEQ(category_names[LC_CTYPE_INDEX_], "LC_CTYPE"));
3606 # ifdef USE_POSIX_2008_LOCALE
3607 assert(category_masks[LC_CTYPE_INDEX_] == LC_CTYPE_MASK);
3610 # ifdef USE_LOCALE_COLLATE
3611 assert(categories[LC_COLLATE_INDEX_] == LC_COLLATE);
3612 assert(strEQ(category_names[LC_COLLATE_INDEX_], "LC_COLLATE"));
3613 # ifdef USE_POSIX_2008_LOCALE
3614 assert(category_masks[LC_COLLATE_INDEX_] == LC_COLLATE_MASK);
3617 # ifdef USE_LOCALE_TIME
3618 assert(categories[LC_TIME_INDEX_] == LC_TIME);
3619 assert(strEQ(category_names[LC_TIME_INDEX_], "LC_TIME"));
3620 # ifdef USE_POSIX_2008_LOCALE
3621 assert(category_masks[LC_TIME_INDEX_] == LC_TIME_MASK);
3624 # ifdef USE_LOCALE_MESSAGES
3625 assert(categories[LC_MESSAGES_INDEX_] == LC_MESSAGES);
3626 assert(strEQ(category_names[LC_MESSAGES_INDEX_], "LC_MESSAGES"));
3627 # ifdef USE_POSIX_2008_LOCALE
3628 assert(category_masks[LC_MESSAGES_INDEX_] == LC_MESSAGES_MASK);
3631 # ifdef USE_LOCALE_MONETARY
3632 assert(categories[LC_MONETARY_INDEX_] == LC_MONETARY);
3633 assert(strEQ(category_names[LC_MONETARY_INDEX_], "LC_MONETARY"));
3634 # ifdef USE_POSIX_2008_LOCALE
3635 assert(category_masks[LC_MONETARY_INDEX_] == LC_MONETARY_MASK);
3638 # ifdef USE_LOCALE_ADDRESS
3639 assert(categories[LC_ADDRESS_INDEX_] == LC_ADDRESS);
3640 assert(strEQ(category_names[LC_ADDRESS_INDEX_], "LC_ADDRESS"));
3641 # ifdef USE_POSIX_2008_LOCALE
3642 assert(category_masks[LC_ADDRESS_INDEX_] == LC_ADDRESS_MASK);
3645 # ifdef USE_LOCALE_IDENTIFICATION
3646 assert(categories[LC_IDENTIFICATION_INDEX_] == LC_IDENTIFICATION);
3647 assert(strEQ(category_names[LC_IDENTIFICATION_INDEX_], "LC_IDENTIFICATION"));
3648 # ifdef USE_POSIX_2008_LOCALE
3649 assert(category_masks[LC_IDENTIFICATION_INDEX_] == LC_IDENTIFICATION_MASK);
3652 # ifdef USE_LOCALE_MEASUREMENT
3653 assert(categories[LC_MEASUREMENT_INDEX_] == LC_MEASUREMENT);
3654 assert(strEQ(category_names[LC_MEASUREMENT_INDEX_], "LC_MEASUREMENT"));
3655 # ifdef USE_POSIX_2008_LOCALE
3656 assert(category_masks[LC_MEASUREMENT_INDEX_] == LC_MEASUREMENT_MASK);
3659 # ifdef USE_LOCALE_PAPER
3660 assert(categories[LC_PAPER_INDEX_] == LC_PAPER);
3661 assert(strEQ(category_names[LC_PAPER_INDEX_], "LC_PAPER"));
3662 # ifdef USE_POSIX_2008_LOCALE
3663 assert(category_masks[LC_PAPER_INDEX_] == LC_PAPER_MASK);
3666 # ifdef USE_LOCALE_TELEPHONE
3667 assert(categories[LC_TELEPHONE_INDEX_] == LC_TELEPHONE);
3668 assert(strEQ(category_names[LC_TELEPHONE_INDEX_], "LC_TELEPHONE"));
3669 # ifdef USE_POSIX_2008_LOCALE
3670 assert(category_masks[LC_TELEPHONE_INDEX_] == LC_TELEPHONE_MASK);
3673 # ifdef USE_LOCALE_SYNTAX
3674 assert(categories[LC_SYNTAX_INDEX_] == LC_SYNTAX);
3675 assert(strEQ(category_names[LC_SYNTAX_INDEX_], "LC_SYNTAX"));
3676 # ifdef USE_POSIX_2008_LOCALE
3677 assert(category_masks[LC_SYNTAX_INDEX_] == LC_SYNTAX_MASK);
3680 # ifdef USE_LOCALE_TOD
3681 assert(categories[LC_TOD_INDEX_] == LC_TOD);
3682 assert(strEQ(category_names[LC_TOD_INDEX_], "LC_TOD"));
3683 # ifdef USE_POSIX_2008_LOCALE
3684 assert(category_masks[LC_TOD_INDEX_] == LC_TOD_MASK);
3688 assert(categories[LC_ALL_INDEX_] == LC_ALL);
3689 assert(strEQ(category_names[LC_ALL_INDEX_], "LC_ALL"));
3690 STATIC_ASSERT_STMT(NOMINAL_LC_ALL_INDEX == LC_ALL_INDEX_);
3691 # ifdef USE_POSIX_2008_LOCALE
3692 assert(category_masks[LC_ALL_INDEX_] == LC_ALL_MASK);
3695 # endif /* DEBUGGING */
3697 /* Initialize the per-thread mbrFOO() state variables. See POSIX.xs for
3698 * why these particular incantations are used. */
3700 memzero(&PL_mbrlen_ps, sizeof(PL_mbrlen_ps));
3703 memzero(&PL_mbrtowc_ps, sizeof(PL_mbrtowc_ps));
3706 wcrtomb(NULL, L'\0', &PL_wcrtomb_ps);
3709 /* Initialize the cache of the program's UTF-8ness for the always known
3710 * locales C and POSIX */
3711 my_strlcpy(PL_locale_utf8ness, C_and_POSIX_utf8ness,
3712 sizeof(PL_locale_utf8ness));
3714 /* See https://github.com/Perl/perl5/issues/17824 */
3715 Zero(curlocales, NOMINAL_LC_ALL_INDEX, char *);
3717 # ifdef USE_THREAD_SAFE_LOCALE
3720 _configthreadlocale(_ENABLE_PER_THREAD_LOCALE);
3724 # ifdef USE_POSIX_2008_LOCALE
3726 PL_C_locale_obj = newlocale(LC_ALL_MASK, "C", (locale_t) 0);
3727 if (! PL_C_locale_obj) {
3728 locale_panic_(Perl_form(aTHX_
3729 "Cannot create POSIX 2008 C locale object"));
3732 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "created C object %p\n",
3735 # ifdef USE_LOCALE_NUMERIC
3737 PL_numeric_radix_sv = newSVpvs(".");
3740 # ifdef USE_PL_CURLOCALES
3742 /* Initialize our records. If we have POSIX 2008, we have LC_ALL */
3743 void_setlocale_c(LC_ALL, porcelain_setlocale(LC_ALL, NULL));
3747 /* We try each locale in the list until we get one that works, or exhaust
3748 * the list. Normally the loop is executed just once. But if setting the
3749 * locale fails, inside the loop we add fallback trials to the array and so
3750 * will execute the loop multiple times */
3751 trial_locales[0] = (trial_locales_struct) {
3752 .trial_locale = setlocale_init,
3753 .fallback_desc = NULL,
3754 .fallback_name = NULL,
3756 trial_locales_count = 1;
3758 for (i= 0; i < trial_locales_count; i++) {
3759 const char * trial_locale = trial_locales[i].trial_locale;
3760 setlocale_failure = FALSE;
3764 sl_result[LC_ALL_INDEX_] = stdized_setlocale(LC_ALL, trial_locale);
3765 DEBUG_LOCALE_INIT(LC_ALL_INDEX_, trial_locale, sl_result[LC_ALL_INDEX_]);
3766 if (! sl_result[LC_ALL_INDEX_]) {
3767 setlocale_failure = TRUE;
3770 /* Since LC_ALL succeeded, it should have changed all the other
3771 * categories it can to its value; so we massage things so that the
3772 * setlocales below just return their category's current values.
3773 * This adequately handles the case in NetBSD where LC_COLLATE may
3774 * not be defined for a locale, and setting it individually will
3775 * fail, whereas setting LC_ALL succeeds, leaving LC_COLLATE set to
3776 * the POSIX locale. */
3777 trial_locale = NULL;
3780 # endif /* LC_ALL */
3782 if (! setlocale_failure) {
3784 for (j = 0; j < NOMINAL_LC_ALL_INDEX; j++) {
3785 Safefree(curlocales[j]);
3786 curlocales[j] = stdized_setlocale(categories[j], trial_locale);
3787 if (! curlocales[j]) {
3788 setlocale_failure = TRUE;
3790 curlocales[j] = savepv(curlocales[j]);
3791 DEBUG_LOCALE_INIT(j, trial_locale, curlocales[j]);
3794 if (LIKELY(! setlocale_failure)) { /* All succeeded */
3795 break; /* Exit trial_locales loop */
3799 /* Here, something failed; will need to try a fallback. */
3805 if (locwarn) { /* Output failure info only on the first one */
3809 PerlIO_printf(Perl_error_log,
3810 "perl: warning: Setting locale failed.\n");
3812 # else /* !LC_ALL */
3814 PerlIO_printf(Perl_error_log,
3815 "perl: warning: Setting locale failed for the categories:\n");
3817 for (j = 0; j < NOMINAL_LC_ALL_INDEX; j++) {
3818 if (! curlocales[j]) {
3819 PerlIO_printf(Perl_error_log, "\t%s\n", category_names[j]);
3823 # endif /* LC_ALL */
3825 PerlIO_printf(Perl_error_log,
3826 "perl: warning: Please check that your locale settings:\n");
3830 PerlIO_printf(Perl_error_log,
3831 "\tLANGUAGE = %c%s%c,\n",
3832 language ? '"' : '(',
3833 language ? language : "unset",
3834 language ? '"' : ')');
3837 PerlIO_printf(Perl_error_log,
3838 "\tLC_ALL = %c%s%c,\n",
3840 lc_all ? lc_all : "unset",
3841 lc_all ? '"' : ')');
3843 # if defined(USE_ENVIRON_ARRAY)
3848 /* Look through the environment for any variables of the
3849 * form qr/ ^ LC_ [A-Z]+ = /x, except LC_ALL which was
3850 * already handled above. These are assumed to be locale
3851 * settings. Output them and their values. */
3852 for (e = environ; *e; e++) {
3853 const STRLEN prefix_len = sizeof("LC_") - 1;
3856 if ( strBEGINs(*e, "LC_")
3857 && ! strBEGINs(*e, "LC_ALL=")
3858 && (uppers_len = strspn(*e + prefix_len,
3859 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"))
3860 && ((*e)[prefix_len + uppers_len] == '='))
3862 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
3863 (int) (prefix_len + uppers_len), *e,
3864 *e + prefix_len + uppers_len + 1);
3871 PerlIO_printf(Perl_error_log,
3872 "\t(possibly more locale environment variables)\n");
3876 PerlIO_printf(Perl_error_log,
3877 "\tLANG = %c%s%c\n",
3879 lang ? lang : "unset",
3882 PerlIO_printf(Perl_error_log,
3883 " are supported and installed on your system.\n");
3886 /* Calculate what fallback locales to try. We have avoided this
3887 * until we have to, because failure is quite unlikely. This will
3888 * usually change the upper bound of the loop we are in.
3890 * Since the system's default way of setting the locale has not
3891 * found one that works, We use Perl's defined ordering: LC_ALL,
3892 * LANG, and the C locale. We don't try the same locale twice, so
3893 * don't add to the list if already there. (On POSIX systems, the
3894 * LC_ALL element will likely be a repeat of the 0th element "",
3895 * but there's no harm done by doing it explicitly.
3897 * Note that this tries the LC_ALL environment variable even on
3898 * systems which have no LC_ALL locale setting. This may or may
3899 * not have been originally intentional, but there's no real need
3900 * to change the behavior. */
3902 for (j = 0; j < trial_locales_count; j++) {
3903 if (strEQ(lc_all, trial_locales[j].trial_locale)) {
3907 trial_locales[trial_locales_count++] = (trial_locales_struct) {
3908 .trial_locale = lc_all,
3909 .fallback_desc = (strEQ(lc_all, "C")
3910 ? "the standard locale"
3911 : "a fallback locale"),
3912 .fallback_name = lc_all,
3918 for (j = 0; j < trial_locales_count; j++) {
3919 if (strEQ(lang, trial_locales[j].trial_locale)) {
3923 trial_locales[trial_locales_count++] = (trial_locales_struct) {
3924 .trial_locale = lang,
3925 .fallback_desc = (strEQ(lang, "C")
3926 ? "the standard locale"
3927 : "a fallback locale"),
3928 .fallback_name = lang,
3933 # if defined(WIN32) && defined(LC_ALL)
3935 /* For Windows, we also try the system default locale before "C".
3936 * (If there exists a Windows without LC_ALL we skip this because
3937 * it gets too complicated. For those, the "C" is the next
3938 * fallback possibility). */
3940 /* Note that this may change the locale, but we are going to do
3942 const char *system_default_locale = stdized_setlocale(LC_ALL, "");
3943 DEBUG_LOCALE_INIT(LC_ALL_INDEX_, "", system_default_locale);
3945 /* Skip if invalid or if it's already on the list of locales to
3947 if (! system_default_locale) {
3948 goto done_system_default;
3950 for (j = 0; j < trial_locales_count; j++) {
3951 if (strEQ(system_default_locale, trial_locales[j].trial_locale)) {
3952 goto done_system_default;
3956 trial_locales[trial_locales_count++] = (trial_locales_struct) {
3957 .trial_locale = system_default_locale,
3958 .fallback_desc = (strEQ(system_default_locale, "C")
3959 ? "the standard locale"
3960 : "the system default locale"),
3961 .fallback_name = system_default_locale,
3964 done_system_default:
3968 for (j = 0; j < trial_locales_count; j++) {
3969 if (strEQ("C", trial_locales[j].trial_locale)) {
3973 trial_locales[trial_locales_count++] = (trial_locales_struct) {
3974 .trial_locale = "C",
3975 .fallback_desc = "the standard locale",
3976 .fallback_name = "C",
3980 } /* end of first time through the loop */
3988 } /* end of looping through the trial locales */
3990 if (ok < 1) { /* If we tried to fallback */
3992 if (! setlocale_failure) { /* fallback succeeded */
3993 msg = "Falling back to";
3995 else { /* fallback failed */
3998 /* We dropped off the end of the loop, so have to decrement i to
3999 * get back to the value the last time through */
4003 msg = "Failed to fall back to";
4005 /* To continue, we should use whatever values we've got */
4007 for (j = 0; j < NOMINAL_LC_ALL_INDEX; j++) {
4008 Safefree(curlocales[j]);
4009 curlocales[j] = savepv(stdized_setlocale(categories[j], NULL));
4010 DEBUG_LOCALE_INIT(j, NULL, curlocales[j]);
4015 const char * description = trial_locales[i].fallback_desc;
4016 const char * name = trial_locales[i].fallback_name;
4018 if (name && strNE(name, "")) {
4019 PerlIO_printf(Perl_error_log,
4020 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
4023 PerlIO_printf(Perl_error_log,
4024 "perl: warning: %s %s.\n", msg, description);
4027 } /* End of tried to fallback */
4029 # ifdef USE_POSIX_2008_LOCALE
4031 /* The stdized setlocales haven't affected the P2008 locales. Initialize
4032 * them now, calculating LC_ALL only on the final go round, when all have
4034 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
4035 (void) emulate_setlocale_i(i, curlocales[i],
4036 RECALCULATE_LC_ALL_ON_FINAL_INTERATION,
4042 /* Done with finding the locales; update the auxiliary records */
4045 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
4047 # if defined(USE_LOCALE_THREADS) && ! defined(USE_THREAD_SAFE_LOCALE)
4049 /* This caches whether each category's locale is UTF-8 or not. This
4050 * may involve changing the locale. It is ok to do this at
4051 * initialization time before any threads have started, but not later
4052 * unless thread-safe operations are used.
4053 * Caching means that if the program heeds our dictate not to change
4054 * locales in threaded applications, this data will remain valid, and
4055 * it may get queried without having to change locales. If the
4056 * environment is such that all categories have the same locale, this
4057 * isn't needed, as the code will not change the locale; but this
4058 * handles the uncommon case where the environment has disparate
4059 * locales for the categories */
4060 (void) _is_cur_LC_category_utf8(categories[i]);
4064 Safefree(curlocales[i]);
4067 # if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
4069 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
4070 * locale is UTF-8. The call to new_ctype() just above has already
4071 * calculated the latter value and saved it in PL_in_utf8_CTYPE_locale. If
4072 * both PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
4073 * are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer on
4074 * STDIN, STDOUT, STDERR, _and_ the default open discipline. */
4075 PL_utf8locale = PL_in_utf8_CTYPE_locale;
4077 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
4078 This is an alternative to using the -C command line switch
4079 (the -C if present will override this). */
4081 const char *p = PerlEnv_getenv("PERL_UNICODE");
4082 PL_unicode = p ? parse_unicode_opts(&p) : 0;
4083 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
4088 #endif /* USE_LOCALE */
4090 /* So won't continue to output stuff */
4091 DEBUG_INITIALIZATION_set(FALSE);
4096 #ifdef USE_LOCALE_COLLATE
4099 Perl__mem_collxfrm(pTHX_ const char *input_string,
4100 STRLEN len, /* Length of 'input_string' */
4101 STRLEN *xlen, /* Set to length of returned string
4102 (not including the collation index
4104 bool utf8 /* Is the input in UTF-8? */
4107 /* _mem_collxfrm() is like strxfrm() but with two important differences.
4108 * First, it handles embedded NULs. Second, it allocates a bit more memory
4109 * than needed for the transformed data itself. The real transformed data
4110 * begins at offset COLLXFRM_HDR_LEN. *xlen is set to the length of that,
4111 * and doesn't include the collation index size.
4113 * It is the caller's responsibility to eventually free the memory returned
4116 * Please see sv_collxfrm() to see how this is used. */
4118 # define COLLXFRM_HDR_LEN sizeof(PL_collation_ix)
4120 char * s = (char *) input_string;
4121 STRLEN s_strlen = strlen(input_string);
4123 STRLEN xAlloc; /* xalloc is a reserved word in VC */
4124 STRLEN length_in_chars;
4125 bool first_time = TRUE; /* Cleared after first loop iteration */
4127 PERL_ARGS_ASSERT__MEM_COLLXFRM;
4129 /* Must be NUL-terminated */
4130 assert(*(input_string + len) == '\0');
4132 /* If this locale has defective collation, skip */
4133 if (PL_collxfrm_base == 0 && PL_collxfrm_mult == 0) {
4134 DEBUG_L(PerlIO_printf(Perl_debug_log,
4135 "_mem_collxfrm: locale's collation is defective\n"));
4139 /* Replace any embedded NULs with the control that sorts before any others.
4140 * This will give as good as possible results on strings that don't
4141 * otherwise contain that character, but otherwise there may be
4142 * less-than-perfect results with that character and NUL. This is
4143 * unavoidable unless we replace strxfrm with our own implementation. */
4144 if (UNLIKELY(s_strlen < len)) { /* Only execute if there is an embedded
4148 STRLEN sans_nuls_len;
4149 int try_non_controls;
4150 char this_replacement_char[] = "?\0"; /* Room for a two-byte string,
4151 making sure 2nd byte is NUL.
4153 STRLEN this_replacement_len;
4155 /* If we don't know what non-NUL control character sorts lowest for
4156 * this locale, find it */
4157 if (PL_strxfrm_NUL_replacement == '\0') {
4159 char * cur_min_x = NULL; /* The min_char's xfrm, (except it also
4160 includes the collation index
4163 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "Looking to replace NUL\n"));
4165 /* Unlikely, but it may be that no control will work to replace
4166 * NUL, in which case we instead look for any character. Controls
4167 * are preferred because collation order is, in general, context
4168 * sensitive, with adjoining characters affecting the order, and
4169 * controls are less likely to have such interactions, allowing the
4170 * NUL-replacement to stand on its own. (Another way to look at it
4171 * is to imagine what would happen if the NUL were replaced by a
4172 * combining character; it wouldn't work out all that well.) */
4173 for (try_non_controls = 0;
4174 try_non_controls < 2;
4177 /* Look through all legal code points (NUL isn't) */
4178 for (j = 1; j < 256; j++) {
4179 char * x; /* j's xfrm plus collation index */
4180 STRLEN x_len; /* length of 'x' */
4181 STRLEN trial_len = 1;
4182 char cur_source[] = { '\0', '\0' };
4184 /* Skip non-controls the first time through the loop. The
4185 * controls in a UTF-8 locale are the L1 ones */
4186 if (! try_non_controls && (PL_in_utf8_COLLATE_locale)
4193 /* Create a 1-char string of the current code point */
4194 cur_source[0] = (char) j;
4196 /* Then transform it */
4197 x = _mem_collxfrm(cur_source, trial_len, &x_len,
4198 0 /* The string is not in UTF-8 */);
4200 /* Ignore any character that didn't successfully transform.
4206 /* If this character's transformation is lower than
4207 * the current lowest, this one becomes the lowest */
4208 if ( cur_min_x == NULL
4209 || strLT(x + COLLXFRM_HDR_LEN,
4210 cur_min_x + COLLXFRM_HDR_LEN))
4212 PL_strxfrm_NUL_replacement = j;
4213 Safefree(cur_min_x);
4219 } /* end of loop through all 255 characters */
4221 /* Stop looking if found */
4226 /* Unlikely, but possible, if there aren't any controls that
4227 * work in the locale, repeat the loop, looking for any
4228 * character that works */
4229 DEBUG_L(PerlIO_printf(Perl_debug_log,
4230 "_mem_collxfrm: No control worked. Trying non-controls\n"));
4231 } /* End of loop to try first the controls, then any char */
4234 DEBUG_L(PerlIO_printf(Perl_debug_log,
4235 "_mem_collxfrm: Couldn't find any character to replace"
4236 " embedded NULs in locale %s with", PL_collation_name));
4240 DEBUG_L(PerlIO_printf(Perl_debug_log,
4241 "_mem_collxfrm: Replacing embedded NULs in locale %s with "
4242 "0x%02X\n", PL_collation_name, PL_strxfrm_NUL_replacement));
4244 Safefree(cur_min_x);
4245 } /* End of determining the character that is to replace NULs */
4247 /* If the replacement is variant under UTF-8, it must match the
4248 * UTF8-ness of the original */
4249 if ( ! UVCHR_IS_INVARIANT(PL_strxfrm_NUL_replacement) && utf8) {
4250 this_replacement_char[0] =
4251 UTF8_EIGHT_BIT_HI(PL_strxfrm_NUL_replacement);
4252 this_replacement_char[1] =
4253 UTF8_EIGHT_BIT_LO(PL_strxfrm_NUL_replacement);
4254 this_replacement_len = 2;
4257 this_replacement_char[0] = PL_strxfrm_NUL_replacement;
4258 /* this_replacement_char[1] = '\0' was done at initialization */
4259 this_replacement_len = 1;
4262 /* The worst case length for the replaced string would be if every
4263 * character in it is NUL. Multiply that by the length of each
4264 * replacement, and allow for a trailing NUL */
4265 sans_nuls_len = (len * this_replacement_len) + 1;
4266 Newx(sans_nuls, sans_nuls_len, char);
4269 /* Replace each NUL with the lowest collating control. Loop until have
4270 * exhausted all the NULs */
4271 while (s + s_strlen < e) {
4272 my_strlcat(sans_nuls, s, sans_nuls_len);
4274 /* Do the actual replacement */
4275 my_strlcat(sans_nuls, this_replacement_char, sans_nuls_len);
4277 /* Move past the input NUL */
4279 s_strlen = strlen(s);
4282 /* And add anything that trails the final NUL */
4283 my_strlcat(sans_nuls, s, sans_nuls_len);
4285 /* Switch so below we transform this modified string */
4288 } /* End of replacing NULs */
4290 /* Make sure the UTF8ness of the string and locale match */
4291 if (utf8 != PL_in_utf8_COLLATE_locale) {
4292 /* XXX convert above Unicode to 10FFFF? */
4293 const char * const t = s; /* Temporary so we can later find where the
4296 /* Here they don't match. Change the string's to be what the locale is
4299 if (! utf8) { /* locale is UTF-8, but input isn't; upgrade the input */
4300 s = (char *) bytes_to_utf8((const U8 *) s, &len);
4303 else { /* locale is not UTF-8; but input is; downgrade the input */
4305 s = (char *) bytes_from_utf8((const U8 *) s, &len, &utf8);
4307 /* If the downgrade was successful we are done, but if the input
4308 * contains things that require UTF-8 to represent, have to do
4309 * damage control ... */
4310 if (UNLIKELY(utf8)) {
4312 /* What we do is construct a non-UTF-8 string with
4313 * 1) the characters representable by a single byte converted
4314 * to be so (if necessary);
4315 * 2) and the rest converted to collate the same as the
4316 * highest collating representable character. That makes
4317 * them collate at the end. This is similar to how we
4318 * handle embedded NULs, but we use the highest collating
4319 * code point instead of the smallest. Like the NUL case,
4320 * this isn't perfect, but is the best we can reasonably
4321 * do. Every above-255 code point will sort the same as
4322 * the highest-sorting 0-255 code point. If that code
4323 * point can combine in a sequence with some other code
4324 * points for weight calculations, us changing something to
4325 * be it can adversely affect the results. But in most
4326 * cases, it should work reasonably. And note that this is
4327 * really an illegal situation: using code points above 255
4328 * on a locale where only 0-255 are valid. If two strings
4329 * sort entirely equal, then the sort order for the
4330 * above-255 code points will be in code point order. */
4334 /* If we haven't calculated the code point with the maximum
4335 * collating order for this locale, do so now */
4336 if (! PL_strxfrm_max_cp) {
4339 /* The current transformed string that collates the
4340 * highest (except it also includes the prefixed collation
4342 char * cur_max_x = NULL;
4344 /* Look through all legal code points (NUL isn't) */
4345 for (j = 1; j < 256; j++) {
4348 char cur_source[] = { '\0', '\0' };
4350 /* Create a 1-char string of the current code point */
4351 cur_source[0] = (char) j;
4353 /* Then transform it */
4354 x = _mem_collxfrm(cur_source, 1, &x_len, FALSE);
4356 /* If something went wrong (which it shouldn't), just
4357 * ignore this code point */
4362 /* If this character's transformation is higher than
4363 * the current highest, this one becomes the highest */
4364 if ( cur_max_x == NULL
4365 || strGT(x + COLLXFRM_HDR_LEN,
4366 cur_max_x + COLLXFRM_HDR_LEN))
4368 PL_strxfrm_max_cp = j;
4369 Safefree(cur_max_x);
4378 DEBUG_L(PerlIO_printf(Perl_debug_log,
4379 "_mem_collxfrm: Couldn't find any character to"
4380 " replace above-Latin1 chars in locale %s with",
4381 PL_collation_name));
4385 DEBUG_L(PerlIO_printf(Perl_debug_log,
4386 "_mem_collxfrm: highest 1-byte collating character"
4387 " in locale %s is 0x%02X\n",
4389 PL_strxfrm_max_cp));
4391 Safefree(cur_max_x);
4394 /* Here we know which legal code point collates the highest.
4395 * We are ready to construct the non-UTF-8 string. The length
4396 * will be at least 1 byte smaller than the input string
4397 * (because we changed at least one 2-byte character into a
4398 * single byte), but that is eaten up by the trailing NUL */
4404 char * e = (char *) t + len;
4406 for (i = 0; i < len; i+= UTF8SKIP(t + i)) {
4408 if (UTF8_IS_INVARIANT(cur_char)) {
4411 else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(t + i, e)) {
4412 s[d++] = EIGHT_BIT_UTF8_TO_NATIVE(cur_char, t[i+1]);
4414 else { /* Replace illegal cp with highest collating
4416 s[d++] = PL_strxfrm_max_cp;
4420 Renew(s, d, char); /* Free up unused space */
4425 /* Here, we have constructed a modified version of the input. It could
4426 * be that we already had a modified copy before we did this version.
4427 * If so, that copy is no longer needed */
4428 if (t != input_string) {
4433 length_in_chars = (utf8)
4434 ? utf8_length((U8 *) s, (U8 *) s + len)
4437 /* The first element in the output is the collation id, used by
4438 * sv_collxfrm(); then comes the space for the transformed string. The
4439 * equation should give us a good estimate as to how much is needed */
4440 xAlloc = COLLXFRM_HDR_LEN
4442 + (PL_collxfrm_mult * length_in_chars);
4443 Newx(xbuf, xAlloc, char);
4444 if (UNLIKELY(! xbuf)) {
4445 DEBUG_L(PerlIO_printf(Perl_debug_log,
4446 "_mem_collxfrm: Couldn't malloc %zu bytes\n", xAlloc));
4450 /* Store the collation id */
4451 *(U32*)xbuf = PL_collation_ix;
4453 /* Then the transformation of the input. We loop until successful, or we
4458 *xlen = strxfrm(xbuf + COLLXFRM_HDR_LEN, s, xAlloc - COLLXFRM_HDR_LEN);
4460 /* If the transformed string occupies less space than we told strxfrm()
4461 * was available, it means it transformed the whole string. */
4462 if (*xlen < xAlloc - COLLXFRM_HDR_LEN) {
4464 /* But there still could have been a problem */
4466 DEBUG_L(PerlIO_printf(Perl_debug_log,
4467 "strxfrm failed for LC_COLLATE=%s; errno=%d, input=%s\n",
4468 PL_collation_name, errno,
4469 _byte_dump_string((U8 *) s, len, 0)));
4473 /* Here, the transformation was successful. Some systems include a
4474 * trailing NUL in the returned length. Ignore it, using a loop in
4475 * case multiple trailing NULs are returned. */
4477 && *(xbuf + COLLXFRM_HDR_LEN + (*xlen) - 1) == '\0')
4482 /* If the first try didn't get it, it means our prediction was low.
4483 * Modify the coefficients so that we predict a larger value in any
4484 * future transformations */
4486 STRLEN needed = *xlen + 1; /* +1 For trailing NUL */
4487 STRLEN computed_guess = PL_collxfrm_base
4488 + (PL_collxfrm_mult * length_in_chars);
4490 /* On zero-length input, just keep current slope instead of
4492 const STRLEN new_m = (length_in_chars != 0)
4493 ? needed / length_in_chars
4496 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4497 "initial size of %zu bytes for a length "
4498 "%zu string was insufficient, %zu needed\n",
4499 computed_guess, length_in_chars, needed));
4501 /* If slope increased, use it, but discard this result for
4502 * length 1 strings, as we can't be sure that it's a real slope
4504 if (length_in_chars > 1 && new_m > PL_collxfrm_mult) {
4508 STRLEN old_m = PL_collxfrm_mult;
4509 STRLEN old_b = PL_collxfrm_base;
4513 PL_collxfrm_mult = new_m;
4514 PL_collxfrm_base = 1; /* +1 For trailing NUL */
4515 computed_guess = PL_collxfrm_base
4516 + (PL_collxfrm_mult * length_in_chars);
4517 if (computed_guess < needed) {
4518 PL_collxfrm_base += needed - computed_guess;
4521 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4522 "slope is now %zu; was %zu, base "
4523 "is now %zu; was %zu\n",
4524 PL_collxfrm_mult, old_m,
4525 PL_collxfrm_base, old_b));
4527 else { /* Slope didn't change, but 'b' did */
4528 const STRLEN new_b = needed
4531 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4532 "base is now %zu; was %zu\n", new_b, PL_collxfrm_base));
4533 PL_collxfrm_base = new_b;
4540 if (UNLIKELY(*xlen >= PERL_INT_MAX)) {
4541 DEBUG_L(PerlIO_printf(Perl_debug_log,
4542 "_mem_collxfrm: Needed %zu bytes, max permissible is %u\n",
4543 *xlen, PERL_INT_MAX));
4547 /* A well-behaved strxfrm() returns exactly how much space it needs
4548 * (usually not including the trailing NUL) when it fails due to not
4549 * enough space being provided. Assume that this is the case unless
4550 * it's been proven otherwise */
4551 if (LIKELY(PL_strxfrm_is_behaved) && first_time) {
4552 xAlloc = *xlen + COLLXFRM_HDR_LEN + 1;
4554 else { /* Here, either:
4555 * 1) The strxfrm() has previously shown bad behavior; or
4556 * 2) It isn't the first time through the loop, which means
4557 * that the strxfrm() is now showing bad behavior, because
4558 * we gave it what it said was needed in the previous
4559 * iteration, and it came back saying it needed still more.
4560 * (Many versions of cygwin fit this. When the buffer size
4561 * isn't sufficient, they return the input size instead of
4562 * how much is needed.)
4563 * Increase the buffer size by a fixed percentage and try again.
4565 xAlloc += (xAlloc / 4) + 1;
4566 PL_strxfrm_is_behaved = FALSE;
4568 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4569 "_mem_collxfrm required more space than previously"
4570 " calculated for locale %s, trying again with new"
4572 PL_collation_name, COLLXFRM_HDR_LEN,
4573 xAlloc - COLLXFRM_HDR_LEN));
4576 Renew(xbuf, xAlloc, char);
4577 if (UNLIKELY(! xbuf)) {
4578 DEBUG_L(PerlIO_printf(Perl_debug_log,
4579 "_mem_collxfrm: Couldn't realloc %zu bytes\n", xAlloc));
4586 DEBUG_Lv((print_collxfrm_input_and_return(s, s + len, xlen, utf8),
4587 PerlIO_printf(Perl_debug_log, "Its xfrm is:"),
4588 PerlIO_printf(Perl_debug_log, "%s\n",
4589 _byte_dump_string((U8 *) xbuf + COLLXFRM_HDR_LEN,
4592 /* Free up unneeded space; retain enough for trailing NUL */
4593 Renew(xbuf, COLLXFRM_HDR_LEN + *xlen + 1, char);
4595 if (s != input_string) {
4603 DEBUG_Lv(print_collxfrm_input_and_return(s, s + len, NULL, utf8));
4606 if (s != input_string) {
4617 S_print_collxfrm_input_and_return(pTHX_
4618 const char * const s,
4619 const char * const e,
4620 const STRLEN * const xlen,
4624 PERL_ARGS_ASSERT_PRINT_COLLXFRM_INPUT_AND_RETURN;
4626 PerlIO_printf(Perl_debug_log, "_mem_collxfrm[%" UVuf "]: returning ",
4627 (UV)PL_collation_ix);
4629 PerlIO_printf(Perl_debug_log, "%zu", *xlen);
4632 PerlIO_printf(Perl_debug_log, "NULL");
4634 PerlIO_printf(Perl_debug_log, " for locale '%s', string='",
4636 print_bytes_for_locale(s, e, is_utf8);
4638 PerlIO_printf(Perl_debug_log, "'\n");
4641 # endif /* DEBUGGING */
4642 #endif /* USE_LOCALE_COLLATE */
4648 S_print_bytes_for_locale(pTHX_
4649 const char * const s,
4650 const char * const e,
4654 bool prev_was_printable = TRUE;
4655 bool first_time = TRUE;
4657 PERL_ARGS_ASSERT_PRINT_BYTES_FOR_LOCALE;
4661 ? utf8_to_uvchr_buf((U8 *) t, e, NULL)
4664 if (! prev_was_printable) {
4665 PerlIO_printf(Perl_debug_log, " ");
4667 PerlIO_printf(Perl_debug_log, "%c", (U8) cp);
4668 prev_was_printable = TRUE;
4672 PerlIO_printf(Perl_debug_log, " ");
4674 PerlIO_printf(Perl_debug_log, "%02" UVXf, cp);
4675 prev_was_printable = FALSE;
4677 t += (is_utf8) ? UTF8SKIP(t) : 1;
4682 # endif /* #ifdef DEBUGGING */
4685 S_switch_category_locale_to_template(pTHX_ const int switch_category,
4686 const int template_category,
4687 const char * template_locale)
4689 /* Changes the locale for LC_'switch_category" to that of
4690 * LC_'template_category', if they aren't already the same. If not NULL,
4691 * 'template_locale' is the locale that 'template_category' is in.
4693 * Returns a copy of the name of the original locale for 'switch_category'
4694 * so can be switched back to with the companion function
4695 * restore_switched_locale(), (NULL if no restoral is necessary.) */
4697 const char * restore_to_locale = NULL;
4699 if (switch_category == template_category) { /* No changes needed */
4703 /* Find the original locale of the category we may need to change, so that
4704 * it can be restored to later */
4705 restore_to_locale = querylocale_r(switch_category);
4706 if (! restore_to_locale) {
4707 locale_panic_(Perl_form(aTHX_ "Could not find current %s locale",
4708 category_name(switch_category)));
4710 restore_to_locale = savepv(restore_to_locale);
4712 /* If the locale of the template category wasn't passed in, find it now */
4713 if (template_locale == NULL) {
4714 template_locale = querylocale_r(template_category);
4715 if (! template_locale) {
4716 locale_panic_(Perl_form(aTHX_ "Could not find current %s locale\n",
4717 category_name(template_category)));
4721 /* It the locales are the same, there's nothing to do */
4722 if (strEQ(restore_to_locale, template_locale)) {
4723 Safefree(restore_to_locale);
4725 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s locale unchanged as %s\n",
4726 category_name(switch_category), template_locale));
4731 /* Finally, change the locale to the template one */
4732 if (! bool_setlocale_r(switch_category, template_locale)) {
4733 setlocale_failure_panic_i(get_category_index(switch_category,
4735 category_name(switch_category),
4741 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s locale switched to %s\n",
4742 category_name(switch_category), template_locale));
4744 return restore_to_locale;
4748 S_restore_switched_locale(pTHX_ const int category,
4749 const char * const original_locale)
4751 /* Restores the locale for LC_'category' to 'original_locale' (which is a
4752 * copy that will be freed by this function), or do nothing if the latter
4753 * parameter is NULL */
4755 if (original_locale == NULL) {
4759 if (! bool_setlocale_r(category, original_locale)) {
4760 locale_panic_(Perl_form(aTHX_ "s restoring %s to %s failed",
4761 category_name(category), original_locale));
4764 Safefree(original_locale);
4767 /* is_cur_LC_category_utf8 uses a small char buffer to avoid malloc/free */
4768 # define CUR_LC_BUFFER_SIZE 64
4771 Perl__is_cur_LC_category_utf8(pTHX_ int category)
4773 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
4774 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
4775 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
4776 * could give the wrong result. The result will very likely be correct for
4777 * languages that have commonly used non-ASCII characters, but for notably
4778 * English, it comes down to if the locale's name ends in something like
4779 * "UTF-8". It errs on the side of not being a UTF-8 locale.
4781 * If the platform is early C89, not containing mbtowc(), or we are
4782 * compiled to not pay attention to LC_CTYPE, this employs heuristics.
4783 * These work very well for non-Latin locales or those whose currency
4784 * symbol isn't a '$' nor plain ASCII text. But without LC_CTYPE and at
4785 * least MB_CUR_MAX, English locales with an ASCII currency symbol depend
4786 * on the name containing UTF-8 or not. */
4788 /* Name of current locale corresponding to the input category */
4789 const char *save_input_locale = NULL;
4791 bool is_utf8 = FALSE; /* The return value */
4793 /* The variables below are for the cache of previous lookups using this
4794 * function. The cache is a C string, described at the definition for
4795 * 'C_and_POSIX_utf8ness'.
4797 * The first part of the cache is fixed, for the C and POSIX locales. The
4798 * varying part starts just after them. */
4799 char * utf8ness_cache = PL_locale_utf8ness + STRLENs(C_and_POSIX_utf8ness);
4801 Size_t utf8ness_cache_size; /* Size of the varying portion */
4802 Size_t input_name_len; /* Length in bytes of save_input_locale */
4803 Size_t input_name_len_with_overhead; /* plus extra chars used to store
4804 the name in the cache */
4805 char * delimited; /* The name plus the delimiters used to store
4807 char buffer[CUR_LC_BUFFER_SIZE]; /* small buffer */
4808 char * name_pos; /* position of 'delimited' in the cache, or 0
4814 assert(category != LC_ALL);
4818 /* Get the desired category's locale */
4819 save_input_locale = querylocale_r(category);
4820 if (! save_input_locale) {
4821 locale_panic_(Perl_form(aTHX_ "Could not find current %s locale",
4822 category_name(category)));
4824 save_input_locale = savepv(save_input_locale);
4826 DEBUG_L(PerlIO_printf(Perl_debug_log,
4827 "Current locale for %s is %s\n",
4828 category_name(category), save_input_locale));
4830 input_name_len = strlen(save_input_locale);
4832 /* In our cache, each name is accompanied by two delimiters and a single
4834 input_name_len_with_overhead = input_name_len + 3;
4836 if ( input_name_len_with_overhead <= CUR_LC_BUFFER_SIZE ) {
4837 /* we can use the buffer, avoid a malloc */
4839 } else { /* need a malloc */
4840 /* Allocate and populate space for a copy of the name surrounded by the
4842 Newx(delimited, input_name_len_with_overhead, char);
4845 delimited[0] = UTF8NESS_SEP[0];
4846 Copy(save_input_locale, delimited + 1, input_name_len, char);
4847 delimited[input_name_len+1] = UTF8NESS_PREFIX[0];
4848 delimited[input_name_len+2] = '\0';
4850 /* And see if that is in the cache */
4851 name_pos = instr(PL_locale_utf8ness, delimited);
4853 is_utf8 = *(name_pos + input_name_len_with_overhead - 1) - '0';
4855 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4856 "UTF8ness for locale %s=%d, \n",
4857 save_input_locale, is_utf8));
4859 /* And, if not already in that position, move it to the beginning of
4860 * the non-constant portion of the list, since it is the most recently
4861 * used. (We don't have to worry about overflow, since just moving
4862 * existing names around) */
4863 if (name_pos > utf8ness_cache) {
4864 Move(utf8ness_cache,
4865 utf8ness_cache + input_name_len_with_overhead,
4866 name_pos - utf8ness_cache, char);
4869 input_name_len_with_overhead - 1, char);
4870 utf8ness_cache[input_name_len_with_overhead - 1] = is_utf8 + '0';
4873 /* free only when not using the buffer */
4874 if ( delimited != buffer ) Safefree(delimited);
4875 Safefree(save_input_locale);
4879 /* Here we don't have stored the utf8ness for the input locale. We have to
4882 # if defined(USE_LOCALE_CTYPE) \
4883 && ( defined(HAS_NL_LANGINFO) \
4884 || (defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)))
4887 const char *original_ctype_locale
4888 = switch_category_locale_to_template(LC_CTYPE,
4892 /* Here the current LC_CTYPE is set to the locale of the category whose
4893 * information is desired. This means that nl_langinfo() and mbtowc()
4894 * should give the correct results */
4896 # ifdef MB_CUR_MAX /* But we can potentially rule out UTF-8ness, avoiding
4897 calling the functions if we have this */
4899 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
4900 * Unicode code point. */
4902 DEBUG_L(PerlIO_printf(Perl_debug_log, "MB_CUR_MAX=%d\n",
4904 if ((unsigned) MB_CUR_MAX < STRLENs(MAX_UNICODE_UTF8)) {
4906 restore_switched_locale(LC_CTYPE, original_ctype_locale);
4907 goto finish_and_return;
4911 # if defined(HAS_NL_LANGINFO)
4913 { /* The task is easiest if the platform has this POSIX 2001 function.
4914 Except on some platforms it can wrongly return "", so have to have
4915 a fallback. And it can return that it's UTF-8, even if there are
4916 variances from that. For example, Turkish locales may use the
4917 alternate dotted I rules, and sometimes it appears to be a
4918 defective locale definition. XXX We should probably check for
4919 these in the Latin1 range and warn (but on glibc, requires
4920 iswalnum() etc. due to their not handling 80-FF correctly */
4921 const char *codeset = my_nl_langinfo(CODESET, FALSE);
4922 /* FALSE => already in dest locale */
4924 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4925 "\tnllanginfo returned CODESET '%s'\n", codeset));
4927 if (codeset && strNE(codeset, "")) {
4929 /* If the implementation of foldEQ() somehow were
4930 * to change to not go byte-by-byte, this could
4931 * read past end of string, as only one length is
4932 * checked. But currently, a premature NUL will
4933 * compare false, and it will stop there */
4934 is_utf8 = cBOOL( foldEQ(codeset, "UTF-8", STRLENs("UTF-8"))
4935 || foldEQ(codeset, "UTF8", STRLENs("UTF8")));
4937 DEBUG_L(PerlIO_printf(Perl_debug_log,
4938 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
4940 restore_switched_locale(LC_CTYPE, original_ctype_locale);
4941 goto finish_and_return;
4946 # if defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)
4947 /* We can see if this is a UTF-8-like locale if have mbtowc(). It was a
4948 * late adder to C89, so very likely to have it. However, testing has
4949 * shown that, like nl_langinfo() above, there are locales that are not
4950 * strictly UTF-8 that this will return that they are */
4957 # if defined(HAS_MBRTOWC) && defined(USE_LOCALE_THREADS)
4963 /* mbrtowc() and mbtowc() convert a byte string to a wide
4964 * character. Feed a byte string to one of them and check that the
4965 * result is the expected Unicode code point */
4967 # if defined(HAS_MBRTOWC) && defined(USE_LOCALE_THREADS)
4968 /* Prefer this function if available, as it's reentrant */
4970 memzero(&ps, sizeof(ps));;
4971 PERL_UNUSED_RESULT(mbrtowc(&wc, NULL, 0, &ps)); /* Reset any shift
4974 len = mbrtowc(&wc, STR_WITH_LEN(REPLACEMENT_CHARACTER_UTF8), &ps);
4980 PERL_UNUSED_RESULT(mbtowc(&wc, NULL, 0));/* Reset any shift state */
4982 len = mbtowc(&wc, STR_WITH_LEN(REPLACEMENT_CHARACTER_UTF8));
4989 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4990 "\treturn from mbtowc; len=%d; code_point=%x; errno=%d\n",
4991 len, (unsigned int) wc, GET_ERRNO));
4993 is_utf8 = cBOOL( len == STRLENs(REPLACEMENT_CHARACTER_UTF8)
4994 && wc == (wchar_t) UNICODE_REPLACEMENT);
4999 restore_switched_locale(LC_CTYPE, original_ctype_locale);
5000 goto finish_and_return;
5005 /* Here, we must have a C89 compiler that doesn't have mbtowc(). Next
5006 * try looking at the currency symbol to see if it disambiguates
5007 * things. Often that will be in the native script, and if the symbol
5008 * isn't in UTF-8, we know that the locale isn't. If it is non-ASCII
5009 * UTF-8, we infer that the locale is too, as the odds of a non-UTF8
5010 * string being valid UTF-8 are quite small */
5012 # ifdef USE_LOCALE_MONETARY
5014 /* If have LC_MONETARY, we can look at the currency symbol. Often that
5015 * will be in the native script. We do this one first because there is
5016 * just one string to examine, so potentially avoids work */
5019 const char *original_monetary_locale
5020 = switch_category_locale_to_template(LC_MONETARY,
5023 bool only_ascii = FALSE;
5024 const U8 * currency_string
5025 = (const U8 *) my_nl_langinfo(CRNCYSTR, FALSE);
5026 /* 2nd param not relevant for this item */
5027 const U8 * first_variant;
5029 assert( *currency_string == '-'
5030 || *currency_string == '+'
5031 || *currency_string == '.');
5035 if (is_utf8_invariant_string_loc(currency_string, 0, &first_variant))
5037 DEBUG_L(PerlIO_printf(Perl_debug_log,
5038 "Couldn't get currency symbol for %s, or contains"
5039 " only ASCII; can't use for determining if UTF-8"
5040 " locale\n", save_input_locale));
5044 is_utf8 = is_strict_utf8_string(first_variant, 0);
5047 restore_switched_locale(LC_MONETARY, original_monetary_locale);
5051 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8;
5052 * otherwise assume the locale is UTF-8 if and only if the symbol
5053 * is non-ascii UTF-8. */
5054 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5055 "\t?Currency symbol for %s is UTF-8=%d\n",
5056 save_input_locale, is_utf8));
5057 goto finish_and_return;
5061 # endif /* USE_LOCALE_MONETARY */
5062 # if defined(HAS_STRFTIME) && defined(USE_LOCALE_TIME)
5064 /* Still haven't found a non-ASCII string to disambiguate UTF-8 or not. Try
5065 * the names of the months and weekdays, timezone, and am/pm indicator */
5067 const char *original_time_locale
5068 = switch_category_locale_to_template(LC_TIME,
5072 bool is_dst = FALSE;
5076 char * formatted_time;
5078 /* Here the current LC_TIME is set to the locale of the category
5079 * whose information is desired. Look at all the days of the week
5080 * and month names, and the timezone and am/pm indicator for UTF-8
5081 * variant characters. The first such a one found will tell us if
5082 * the locale is UTF-8 or not */
5084 for (i = 0; i < 7 + 12; i++) { /* 7 days; 12 months */
5085 formatted_time = my_strftime("%A %B %Z %p",
5086 0, 0, hour, dom, month, 2012 - 1900, 0, 0, is_dst);
5087 if ( ! formatted_time
5088 || is_utf8_invariant_string((U8 *) formatted_time, 0))
5091 /* Here, we didn't find a non-ASCII. Try the next time
5092 * through with the complemented dst and am/pm, and try
5093 * with the next weekday. After we have gotten all
5094 * weekdays, try the next month */
5096 hour = (hour + 12) % 24;
5101 Safefree(formatted_time);
5105 /* Here, we have a non-ASCII. Return TRUE is it is valid UTF8;
5106 * false otherwise. But first, restore LC_TIME to its original
5107 * locale if we changed it */
5108 restore_switched_locale(LC_TIME, original_time_locale);
5110 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5111 "\t?time-related strings for %s are UTF-8=%d\n",
5113 is_utf8_string((U8 *) formatted_time, 0)));
5114 is_utf8 = is_utf8_string((U8 *) formatted_time, 0);
5115 Safefree(formatted_time);
5116 goto finish_and_return;
5119 /* Falling off the end of the loop indicates all the names were just
5120 * ASCII. Go on to the next test. If we changed it, restore LC_TIME
5121 * to its original locale */
5122 restore_switched_locale(LC_TIME, original_time_locale);
5123 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5124 "All time-related words for %s contain only ASCII;"
5125 " can't use for determining if UTF-8 locale\n",
5126 save_input_locale));
5131 # if 0 && defined(USE_LOCALE_MESSAGES) && defined(HAS_SYS_ERRLIST)
5133 /* This code is ifdefd out because it was found to not be necessary in
5134 * testing on our dromedary test machine, which has over 700 locales.
5135 * There, this added no value to looking at the currency symbol and the
5136 * time strings. I left it in so as to avoid rewriting it if real-world
5137 * experience indicates that dromedary is an outlier. Essentially, instead
5138 * of returning abpve if we haven't found illegal utf8, we continue on and
5139 * examine all the strerror() messages on the platform for utf8ness. If
5140 * all are ASCII, we still don't know the answer; but otherwise we have a
5141 * pretty good indication of the utf8ness. The reason this doesn't help
5142 * much is that the messages may not have been translated into the locale.
5143 * The currency symbol and time strings are much more likely to have been
5147 bool non_ascii = FALSE;
5148 const char *original_messages_locale
5149 = switch_category_locale_to_template(LC_MESSAGES,
5152 const char * errmsg = NULL;
5154 /* Here the current LC_MESSAGES is set to the locale of the category
5155 * whose information is desired. Look through all the messages. We
5156 * can't use Strerror() here because it may expand to code that
5157 * segfaults in miniperl */
5159 for (e = 0; e <= sys_nerr; e++) {
5161 errmsg = sys_errlist[e];
5162 if (errno || !errmsg) {
5165 errmsg = savepv(errmsg);
5166 if (! is_utf8_invariant_string((U8 *) errmsg, 0)) {
5168 is_utf8 = is_utf8_string((U8 *) errmsg, 0);
5174 restore_switched_locale(LC_MESSAGES, original_messages_locale);
5178 /* Any non-UTF-8 message means not a UTF-8 locale; if all are
5179 * valid, any non-ascii means it is one; otherwise we assume it
5181 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5182 "\t?error messages for %s are UTF-8=%d\n",
5185 goto finish_and_return;
5188 DEBUG_L(PerlIO_printf(Perl_debug_log,
5189 "All error messages for %s contain only ASCII;"
5190 " can't use for determining if UTF-8 locale\n",
5191 save_input_locale));
5195 # ifndef EBCDIC /* On os390, even if the name ends with "UTF-8', it isn't a
5198 /* As a last resort, look at the locale name to see if it matches
5199 * qr/UTF -? * 8 /ix, or some other common locale names. This "name", the
5200 * return of setlocale(), is actually defined to be opaque, so we can't
5201 * really rely on the absence of various substrings in the name to indicate
5202 * its UTF-8ness, but if it has UTF8 in the name, it is extremely likely to
5203 * be a UTF-8 locale. Similarly for the other common names */
5206 const Size_t final_pos = strlen(save_input_locale) - 1;
5208 if (final_pos >= 3) {
5209 const char *name = save_input_locale;
5211 /* Find next 'U' or 'u' and look from there */
5212 while ((name += strcspn(name, "Uu") + 1)
5213 <= save_input_locale + final_pos - 2)
5215 if ( isALPHA_FOLD_NE(*name, 't')
5216 || isALPHA_FOLD_NE(*(name + 1), 'f'))
5221 if (*(name) == '-') {
5222 if ((name > save_input_locale + final_pos - 1)) {
5227 if (*(name) == '8') {
5228 DEBUG_L(PerlIO_printf(Perl_debug_log,
5229 "Locale %s ends with UTF-8 in name\n",
5230 save_input_locale));
5232 goto finish_and_return;
5235 DEBUG_L(PerlIO_printf(Perl_debug_log,
5236 "Locale %s doesn't end with UTF-8 in name\n",
5237 save_input_locale));
5242 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
5243 if (memENDs(save_input_locale, final_pos, "65001")) {
5244 DEBUG_L(PerlIO_printf(Perl_debug_log,
5245 "Locale %s ends with 65001 in name, is UTF-8 locale\n",
5246 save_input_locale));
5248 goto finish_and_return;
5255 /* Other common encodings are the ISO 8859 series, which aren't UTF-8. But
5256 * since we are about to return FALSE anyway, there is no point in doing
5257 * this extra work */
5260 if (instr(save_input_locale, "8859")) {
5261 DEBUG_L(PerlIO_printf(Perl_debug_log,
5262 "Locale %s has 8859 in name, not UTF-8 locale\n",
5263 save_input_locale));
5265 goto finish_and_return;
5269 DEBUG_L(PerlIO_printf(Perl_debug_log,
5270 "Assuming locale %s is not a UTF-8 locale\n",
5271 save_input_locale));
5274 # endif /* the code that is compiled when no modern LC_CTYPE */
5278 /* Cache this result so we don't have to go through all this next time. */
5279 utf8ness_cache_size = sizeof(PL_locale_utf8ness)
5280 - (utf8ness_cache - PL_locale_utf8ness);
5282 /* But we can't save it if it is too large for the total space available */
5283 if (LIKELY(input_name_len_with_overhead < utf8ness_cache_size)) {
5284 Size_t utf8ness_cache_len = strlen(utf8ness_cache);
5286 /* Here it can fit, but we may need to clear out the oldest cached
5287 * result(s) to do so. Check */
5288 if (utf8ness_cache_len + input_name_len_with_overhead
5289 >= utf8ness_cache_size)
5291 /* Here we have to clear something out to make room for this.
5292 * Start looking at the rightmost place where it could fit and find
5293 * the beginning of the entry that extends past that. */
5294 char * cutoff = (char *) my_memrchr(utf8ness_cache,
5297 - input_name_len_with_overhead);
5300 assert(cutoff >= utf8ness_cache);
5302 /* This and all subsequent entries must be removed */
5304 utf8ness_cache_len = strlen(utf8ness_cache);
5307 /* Make space for the new entry */
5308 Move(utf8ness_cache,
5309 utf8ness_cache + input_name_len_with_overhead,
5310 utf8ness_cache_len + 1 /* Incl. trailing NUL */, char);
5313 Copy(delimited, utf8ness_cache, input_name_len_with_overhead - 1, char);
5314 utf8ness_cache[input_name_len_with_overhead - 1] = is_utf8 + '0';
5316 if ((PL_locale_utf8ness[strlen(PL_locale_utf8ness)-1] & ~1) != '0') {
5317 locale_panic_(Perl_form(aTHX_
5318 "Corrupt utf8ness_cache=%s\nlen=%zu,"
5319 " inserted_name=%s, its_len=%zu",
5320 PL_locale_utf8ness, strlen(PL_locale_utf8ness),
5321 delimited, input_name_len_with_overhead));
5327 if (DEBUG_Lv_TEST) {
5328 const char * s = PL_locale_utf8ness;
5330 /* Audit the structure */
5331 while (s < PL_locale_utf8ness + strlen(PL_locale_utf8ness)) {
5334 if (*s != UTF8NESS_SEP[0]) {
5335 locale_panic_(Perl_form(aTHX_
5336 "Corrupt utf8ness_cache: missing"
5337 " separator %.*s<-- HERE %s",
5338 (int) (s - PL_locale_utf8ness),
5343 e = strchr(s, UTF8NESS_PREFIX[0]);
5345 e = PL_locale_utf8ness + strlen(PL_locale_utf8ness);
5346 locale_panic_(Perl_form(aTHX_
5347 "Corrupt utf8ness_cache: missing"
5348 " separator %.*s<-- HERE %s",
5349 (int) (e - PL_locale_utf8ness),
5354 if (*e != '0' && *e != '1') {
5355 locale_panic_(Perl_form(aTHX_
5356 "Corrupt utf8ness_cache: utf8ness"
5357 " must be [01] %.*s<-- HERE %s",
5358 (int) (e + 1 - PL_locale_utf8ness),
5362 if (ninstr(PL_locale_utf8ness, s, s-1, e)) {
5363 locale_panic_(Perl_form(aTHX_
5364 "Corrupt utf8ness_cache: entry"
5365 " has duplicate %.*s<-- HERE %s",
5366 (int) (e - PL_locale_utf8ness),
5374 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5375 "PL_locale_utf8ness is now %s; returning %d\n",
5376 PL_locale_utf8ness, is_utf8));
5380 /* free only when not using the buffer */
5381 if ( delimited != buffer ) Safefree(delimited);
5382 Safefree(save_input_locale);
5389 Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
5391 /* Internal function which returns if we are in the scope of a pragma that
5392 * enables the locale category 'category'. 'compiling' should indicate if
5393 * this is during the compilation phase (TRUE) or not (FALSE). */
5395 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
5397 SV *these_categories = cop_hints_fetch_pvs(cop, "locale", 0);
5398 if (! these_categories || these_categories == &PL_sv_placeholder) {
5402 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
5403 * a valid unsigned */
5404 assert(category >= -1);
5405 return cBOOL(SvUV(these_categories) & (1U << (category + 1)));
5409 Perl_my_strerror(pTHX_ const int errnum)
5411 /* Returns a mortalized copy of the text of the error message associated
5412 * with 'errnum'. It uses the current locale's text unless the platform
5413 * doesn't have the LC_MESSAGES category or we are not being called from
5414 * within the scope of 'use locale'. In the former case, it uses whatever
5415 * strerror returns; in the latter case it uses the text from the C locale.
5417 * The function just calls strerror(), but temporarily switches, if needed,
5418 * to the C locale */
5422 #ifndef USE_LOCALE_MESSAGES
5424 /* If platform doesn't have messages category, we don't do any switching to
5425 * the C locale; we just use whatever strerror() returns */
5427 errstr = savepv(Strerror(errnum));
5429 #else /* Has locale messages */
5431 const bool within_locale_scope = IN_LC(LC_MESSAGES);
5433 # ifndef USE_LOCALE_THREADS
5435 /* This function is trivial without threads. */
5436 if (within_locale_scope) {
5437 errstr = savepv(Strerror(errnum));
5440 const char * save_locale = savepv(querylocale_c(LC_MESSAGES));
5442 void_setlocale_c(LC_MESSAGES, "C");
5443 errstr = savepv(Strerror(errnum));
5444 void_setlocale_c(LC_MESSAGES, save_locale);
5445 Safefree(save_locale);
5448 # elif defined(USE_POSIX_2008_LOCALE) && defined(HAS_STRERROR_L)
5450 /* This function is also trivial if we don't have to worry about thread
5451 * safety and have strerror_l(), as it handles the switch of locales so we
5452 * don't have to deal with that. We don't have to worry about thread
5453 * safety if strerror_r() is also available. Both it and strerror_l() are
5454 * thread-safe. Plain strerror() isn't thread safe. But on threaded
5455 * builds when strerror_r() is available, the apparent call to strerror()
5456 * below is actually a macro that behind-the-scenes calls strerror_r(). */
5458 # ifdef HAS_STRERROR_R
5460 if (within_locale_scope) {
5461 errstr = savepv(Strerror(errnum));
5464 errstr = savepv(strerror_l(errnum, PL_C_locale_obj));
5469 /* Here we have strerror_l(), but not strerror_r() and we are on a
5470 * threaded-build. We use strerror_l() for everything, constructing a
5471 * locale to pass to it if necessary */
5473 locale_t locale_to_use;
5475 if (within_locale_scope) {
5476 locale_to_use = use_curlocale_scratch();
5478 else { /* Use C locale if not within 'use locale' scope */
5479 locale_to_use = PL_C_locale_obj;
5482 errstr = savepv(strerror_l(errnum, locale_to_use));
5485 # else /* Doesn't have strerror_l() */
5487 const char * save_locale = NULL;
5488 bool locale_is_C = FALSE;
5490 /* We have a critical section to prevent another thread from executing this
5491 * same code at the same time. (On thread-safe perls, the LOCK is a
5492 * no-op.) Since this is the only place in core that changes LC_MESSAGES
5493 * (unless the user has called setlocale(), this works to prevent races. */
5496 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5497 "my_strerror called with errnum %d\n", errnum));
5498 if (! within_locale_scope) {
5499 save_locale = querylocale_c(LC_MESSAGES);
5500 if (! save_locale) {
5502 locale_panic_("Could not find current LC_MESSAGES locale");
5505 locale_is_C = isNAME_C_OR_POSIX(save_locale);
5507 /* Switch to the C locale if not already in it */
5508 if (! locale_is_C) {
5510 /* The setlocale() just below likely will zap 'save_locale', so
5512 save_locale = savepv(save_locale);
5513 if (! bool_setlocale_c(LC_MESSAGES, "C")) {
5515 /* If, for some reason, the locale change failed, we
5516 * soldier on as best as possible under the circumstances,
5517 * using the current locale, and clear save_locale, so we
5518 * don't try to change back. On z/0S, all setlocale()
5519 * calls fail after you've created a thread. This is their
5520 * way of making sure the entire process is always a single
5521 * locale. This means that 'use locale' is always in place
5522 * for messages under these circumstances. */
5523 Safefree(save_locale);
5528 } /* end of ! within_locale_scope */
5530 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "WITHIN locale scope\n"));
5533 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5534 "Any locale change has been done; about to call Strerror\n"));
5535 errstr = savepv(Strerror(errnum));
5537 if (! within_locale_scope) {
5538 if (save_locale && ! locale_is_C) {
5539 if (! bool_setlocale_c(LC_MESSAGES, save_locale)) {
5541 locale_panic_(Perl_form(aTHX_
5542 "setlocale restore to '%s' failed",
5545 Safefree(save_locale);
5551 # endif /* End of doesn't have strerror_l */
5553 DEBUG_Lv((PerlIO_printf(Perl_debug_log,
5554 "Strerror returned; saving a copy: '"),
5555 print_bytes_for_locale(errstr, errstr + strlen(errstr), 0),
5556 PerlIO_printf(Perl_debug_log, "'\n")));
5558 #endif /* End of does have locale messages */
5566 =for apidoc switch_to_global_locale
5568 On systems without locale support, or on typical single-threaded builds, or on
5569 platforms that do not support per-thread locale operations, this function does
5570 nothing. On such systems that do have locale support, only a locale global to
5571 the whole program is available.
5573 On multi-threaded builds on systems that do have per-thread locale operations,
5574 this function converts the thread it is running in to use the global locale.
5575 This is for code that has not yet or cannot be updated to handle multi-threaded
5576 locale operation. As long as only a single thread is so-converted, everything
5577 works fine, as all the other threads continue to ignore the global one, so only
5578 this thread looks at it.
5580 However, on Windows systems this isn't quite true prior to Visual Studio 15,
5581 at which point Microsoft fixed a bug. A race can occur if you use the
5582 following operations on earlier Windows platforms:
5586 =item L<POSIX::localeconv|POSIX/localeconv>
5588 =item L<I18N::Langinfo>, items C<CRNCYSTR> and C<THOUSEP>
5590 =item L<perlapi/Perl_langinfo>, items C<CRNCYSTR> and C<THOUSEP>
5594 The first item is not fixable (except by upgrading to a later Visual Studio
5595 release), but it would be possible to work around the latter two items by using
5596 the Windows API functions C<GetNumberFormat> and C<GetCurrencyFormat>; patches
5599 Without this function call, threads that use the L<C<setlocale(3)>> system
5600 function will not work properly, as all the locale-sensitive functions will
5601 look at the per-thread locale, and C<setlocale> will have no effect on this
5604 Perl code should convert to either call
5605 L<C<Perl_setlocale>|perlapi/Perl_setlocale> (which is a drop-in for the system
5606 C<setlocale>) or use the methods given in L<perlcall> to call
5607 L<C<POSIX::setlocale>|POSIX/setlocale>. Either one will transparently properly
5608 handle all cases of single- vs multi-thread, POSIX 2008-supported or not.
5610 Non-Perl libraries, such as C<gtk>, that call the system C<setlocale> can
5611 continue to work if this function is called before transferring control to the
5614 Upon return from the code that needs to use the global locale,
5615 L<C<sync_locale()>|perlapi/sync_locale> should be called to restore the safe
5616 multi-thread operation.
5622 Perl_switch_to_global_locale()
5626 #ifdef USE_THREAD_SAFE_LOCALE
5629 _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
5636 for (i = 0; i < LC_ALL_INDEX_; i++) {
5637 setlocale(categories[i], querylocale_i(i));
5641 uselocale(LC_GLOBAL_LOCALE);
5650 =for apidoc sync_locale
5652 L<C<Perl_setlocale>|perlapi/Perl_setlocale> can be used at any time to query or
5653 change the locale (though changing the locale is antisocial and dangerous on
5654 multi-threaded systems that don't have multi-thread safe locale operations.
5655 (See L<perllocale/Multi-threaded operation>). Using the system
5656 L<C<setlocale(3)>> should be avoided. Nevertheless, certain non-Perl libraries
5657 called from XS, such as C<Gtk> do so, and this can't be changed. When the
5658 locale is changed by XS code that didn't use
5659 L<C<Perl_setlocale>|perlapi/Perl_setlocale>, Perl needs to be told that the
5660 locale has changed. Use this function to do so, before returning to Perl.
5662 The return value is a boolean: TRUE if the global locale at the time of call
5663 was in effect; and FALSE if a per-thread locale was in effect. This can be
5664 used by the caller that needs to restore things as-they-were to decide whether
5666 L<C<Perl_switch_to_global_locale>|perlapi/switch_to_global_locale>.
5681 const char * newlocale;
5684 # ifdef USE_POSIX_2008_LOCALE
5686 bool was_in_global_locale = FALSE;
5687 locale_t cur_obj = uselocale((locale_t) 0);
5689 /* On Windows, unless the foreign code has turned off the thread-safe
5690 * locale setting, any plain setlocale() will have affected what we see, so
5691 * no need to worry. Otherwise, If the foreign code has done a plain
5692 * setlocale(), it will only affect the global locale on POSIX systems, but
5693 * will affect the */
5694 if (cur_obj == LC_GLOBAL_LOCALE) {
5696 # ifdef HAS_QUERY_LOCALE
5698 void_setlocale_c(LC_ALL, querylocale_c(LC_ALL));
5704 /* We can't trust that we can read the LC_ALL format on the
5705 * platform, so do them individually */
5706 for (i = 0; i < LC_ALL_INDEX_; i++) {
5707 void_setlocale_i(i, querylocale_i(i));
5712 was_in_global_locale = TRUE;
5717 bool was_in_global_locale = TRUE;
5720 # ifdef USE_LOCALE_CTYPE
5722 newlocale = savepv(querylocale_c(LC_CTYPE));
5723 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5724 "%s\n", setlocale_debug_string_c(LC_CTYPE, NULL, newlocale)));
5725 new_ctype(newlocale);
5726 Safefree(newlocale);
5728 # endif /* USE_LOCALE_CTYPE */
5729 # ifdef USE_LOCALE_COLLATE
5731 newlocale = savepv(querylocale_c(LC_COLLATE));
5732 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5733 "%s\n", setlocale_debug_string_c(LC_COLLATE, NULL, newlocale)));
5734 new_collate(newlocale);
5735 Safefree(newlocale);
5738 # ifdef USE_LOCALE_NUMERIC
5740 newlocale = savepv(querylocale_c(LC_NUMERIC));
5741 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5742 "%s\n", setlocale_debug_string_c(LC_NUMERIC, NULL, newlocale)));
5743 new_numeric(newlocale);
5744 Safefree(newlocale);
5746 # endif /* USE_LOCALE_NUMERIC */
5748 return was_in_global_locale;
5754 #if defined(DEBUGGING) && defined(USE_LOCALE)
5757 S_setlocale_debug_string_i(const unsigned cat_index,
5758 const char* const locale, /* Optional locale name */
5760 /* return value from setlocale() when attempting to
5761 * set 'category' to 'locale' */
5762 const char* const retval)
5764 /* Returns a pointer to a NUL-terminated string in static storage with
5765 * added text about the info passed in. This is not thread safe and will
5766 * be overwritten by the next call, so this should be used just to
5767 * formulate a string to immediately print or savepv() on. */
5769 static char ret[1024];
5770 assert(cat_index <= NOMINAL_LC_ALL_INDEX);
5772 my_strlcpy(ret, "setlocale(", sizeof(ret));
5773 my_strlcat(ret, category_names[cat_index], sizeof(ret));
5774 my_strlcat(ret, ", ", sizeof(ret));
5777 my_strlcat(ret, "\"", sizeof(ret));
5778 my_strlcat(ret, locale, sizeof(ret));
5779 my_strlcat(ret, "\"", sizeof(ret));
5782 my_strlcat(ret, "NULL", sizeof(ret));
5785 my_strlcat(ret, ") returned ", sizeof(ret));
5788 my_strlcat(ret, "\"", sizeof(ret));
5789 my_strlcat(ret, retval, sizeof(ret));
5790 my_strlcat(ret, "\"", sizeof(ret));
5793 my_strlcat(ret, "NULL", sizeof(ret));
5796 assert(strlen(ret) < sizeof(ret));
5804 Perl_thread_locale_init()
5806 /* Called from a thread on startup*/
5808 #ifdef USE_THREAD_SAFE_LOCALE
5813 DEBUG_L(PerlIO_printf(Perl_debug_log,
5814 "new thread, initial locale is %s; calling setlocale\n",
5815 setlocale(LC_ALL, NULL)));
5819 /* On Windows, make sure new thread has per-thread locales enabled */
5820 _configthreadlocale(_ENABLE_PER_THREAD_LOCALE);
5824 /* This thread starts off in the C locale */
5825 Perl_setlocale(LC_ALL, "C");
5833 Perl_thread_locale_term()
5835 /* Called from a thread as it gets ready to terminate */
5837 #ifdef USE_POSIX_2008_LOCALE
5839 /* C starts the new thread in the global C locale. If we are thread-safe,
5840 * we want to not be in the global locale */
5843 locale_t cur_obj = uselocale(LC_GLOBAL_LOCALE);
5844 if (cur_obj != LC_GLOBAL_LOCALE && cur_obj != PL_C_locale_obj) {
5845 freelocale(cur_obj);
5854 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob