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
38 #define PERL_IN_LOCALE_C
39 #include "perl_langinfo.h"
48 /* If the environment says to, we can output debugging information during
49 * initialization. This is done before option parsing, and before any thread
50 * creation, so can be a file-level static */
51 #if ! defined(DEBUGGING) || defined(PERL_GLOBAL_STRUCT)
52 # define debug_initialization 0
53 # define DEBUG_INITIALIZATION_set(v)
55 static bool debug_initialization = FALSE;
56 # define DEBUG_INITIALIZATION_set(v) (debug_initialization = v)
60 /* Returns the Unix errno portion; ignoring any others. This is a macro here
61 * instead of putting it into perl.h, because unclear to khw what should be
63 #define GET_ERRNO saved_errno
65 /* strlen() of a literal string constant. We might want this more general,
66 * but using it in just this file for now. A problem with more generality is
67 * the compiler warnings about comparing unlike signs */
68 #define STRLENs(s) (sizeof("" s "") - 1)
70 /* Is the C string input 'name' "C" or "POSIX"? If so, and 'name' is the
71 * return of setlocale(), then this is extremely likely to be the C or POSIX
72 * locale. However, the output of setlocale() is documented to be opaque, but
73 * the odds are extremely small that it would return these two strings for some
74 * other locale. Note that VMS in these two locales includes many non-ASCII
75 * characters as controls and punctuation (below are hex bytes):
77 * punct: A1-A3 A5 A7-AB B0-B3 B5-B7 B9-BD BF-CF D1-DD DF-EF F1-FD
78 * Oddly, none there are listed as alphas, though some represent alphabetics
79 * http://www.nntp.perl.org/group/perl.perl5.porters/2013/02/msg198753.html */
80 #define isNAME_C_OR_POSIX(name) \
82 && (( *(name) == 'C' && (*(name + 1)) == '\0') \
83 || strEQ((name), "POSIX")))
87 /* This code keeps a LRU cache of the UTF-8ness of the locales it has so-far
88 * looked up. This is in the form of a C string: */
90 #define UTF8NESS_SEP "\v"
91 #define UTF8NESS_PREFIX "\f"
93 /* So, the string looks like:
95 * \vC\a0\vPOSIX\a0\vam_ET\a0\vaf_ZA.utf8\a1\ven_US.UTF-8\a1\0
97 * where the digit 0 after the \a indicates that the locale starting just
98 * after the preceding \v is not UTF-8, and the digit 1 mean it is. */
100 STATIC_ASSERT_DECL(STRLENs(UTF8NESS_SEP) == 1);
101 STATIC_ASSERT_DECL(STRLENs(UTF8NESS_PREFIX) == 1);
103 #define C_and_POSIX_utf8ness UTF8NESS_SEP "C" UTF8NESS_PREFIX "0" \
104 UTF8NESS_SEP "POSIX" UTF8NESS_PREFIX "0"
106 /* The cache is initialized to C_and_POSIX_utf8ness at start up. These are
107 * kept there always. The remining portion of the cache is LRU, with the
108 * oldest looked-up locale at the tail end */
111 S_stdize_locale(pTHX_ char *locs)
113 /* Standardize the locale name from a string returned by 'setlocale',
114 * possibly modifying that string.
116 * The typical return value of setlocale() is either
117 * (1) "xx_YY" if the first argument of setlocale() is not LC_ALL
118 * (2) "xa_YY xb_YY ..." if the first argument of setlocale() is LC_ALL
119 * (the space-separated values represent the various sublocales,
120 * in some unspecified order). This is not handled by this function.
122 * In some platforms it has a form like "LC_SOMETHING=Lang_Country.866\n",
123 * which is harmful for further use of the string in setlocale(). This
124 * function removes the trailing new line and everything up through the '='
127 const char * const s = strchr(locs, '=');
130 PERL_ARGS_ASSERT_STDIZE_LOCALE;
133 const char * const t = strchr(s, '.');
136 const char * const u = strchr(t, '\n');
137 if (u && (u[1] == 0)) {
138 const STRLEN len = u - s;
139 Move(s + 1, locs, len, char);
147 Perl_croak(aTHX_ "Can't fix broken locale name \"%s\"", locs);
152 /* Two parallel arrays; first the locale categories Perl uses on this system;
153 * the second array is their names. These arrays are in mostly arbitrary
156 const int categories[] = {
158 # ifdef USE_LOCALE_NUMERIC
161 # ifdef USE_LOCALE_CTYPE
164 # ifdef USE_LOCALE_COLLATE
167 # ifdef USE_LOCALE_TIME
170 # ifdef USE_LOCALE_MESSAGES
173 # ifdef USE_LOCALE_MONETARY
176 # ifdef USE_LOCALE_ADDRESS
179 # ifdef USE_LOCALE_IDENTIFICATION
182 # ifdef USE_LOCALE_MEASUREMENT
185 # ifdef USE_LOCALE_PAPER
188 # ifdef USE_LOCALE_TELEPHONE
194 -1 /* Placeholder because C doesn't allow a
195 trailing comma, and it would get complicated
196 with all the #ifdef's */
199 /* The top-most real element is LC_ALL */
201 const char * category_names[] = {
203 # ifdef USE_LOCALE_NUMERIC
206 # ifdef USE_LOCALE_CTYPE
209 # ifdef USE_LOCALE_COLLATE
212 # ifdef USE_LOCALE_TIME
215 # ifdef USE_LOCALE_MESSAGES
218 # ifdef USE_LOCALE_MONETARY
221 # ifdef USE_LOCALE_ADDRESS
224 # ifdef USE_LOCALE_IDENTIFICATION
227 # ifdef USE_LOCALE_MEASUREMENT
230 # ifdef USE_LOCALE_PAPER
233 # ifdef USE_LOCALE_TELEPHONE
239 NULL /* Placeholder */
244 /* On systems with LC_ALL, it is kept in the highest index position. (-2
245 * to account for the final unused placeholder element.) */
246 # define NOMINAL_LC_ALL_INDEX (C_ARRAY_LENGTH(categories) - 2)
250 /* On systems without LC_ALL, we pretend it is there, one beyond the real
251 * top element, hence in the unused placeholder element. */
252 # define NOMINAL_LC_ALL_INDEX (C_ARRAY_LENGTH(categories) - 1)
256 /* Pretending there is an LC_ALL element just above allows us to avoid most
257 * special cases. Most loops through these arrays in the code below are
258 * written like 'for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++)'. They will work
259 * on either type of system. But the code must be written to not access the
260 * element at 'LC_ALL_INDEX' except on platforms that have it. This can be
261 * checked for at compile time by using the #define LC_ALL_INDEX which is only
262 * defined if we do have LC_ALL. */
265 S_category_name(const int category)
271 if (category == LC_ALL) {
277 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
278 if (category == categories[i]) {
279 return category_names[i];
284 const char suffix[] = " (unknown)";
286 Size_t length = sizeof(suffix) + 1;
295 /* Calculate the number of digits */
301 Newx(unknown, length, char);
302 my_snprintf(unknown, length, "%d%s", category, suffix);
308 /* Now create LC_foo_INDEX #defines for just those categories on this system */
309 # ifdef USE_LOCALE_NUMERIC
310 # define LC_NUMERIC_INDEX 0
311 # define _DUMMY_NUMERIC LC_NUMERIC_INDEX
313 # define _DUMMY_NUMERIC -1
315 # ifdef USE_LOCALE_CTYPE
316 # define LC_CTYPE_INDEX _DUMMY_NUMERIC + 1
317 # define _DUMMY_CTYPE LC_CTYPE_INDEX
319 # define _DUMMY_CTYPE _DUMMY_NUMERIC
321 # ifdef USE_LOCALE_COLLATE
322 # define LC_COLLATE_INDEX _DUMMY_CTYPE + 1
323 # define _DUMMY_COLLATE LC_COLLATE_INDEX
325 # define _DUMMY_COLLATE _DUMMY_COLLATE
327 # ifdef USE_LOCALE_TIME
328 # define LC_TIME_INDEX _DUMMY_COLLATE + 1
329 # define _DUMMY_TIME LC_TIME_INDEX
331 # define _DUMMY_TIME _DUMMY_COLLATE
333 # ifdef USE_LOCALE_MESSAGES
334 # define LC_MESSAGES_INDEX _DUMMY_TIME + 1
335 # define _DUMMY_MESSAGES LC_MESSAGES_INDEX
337 # define _DUMMY_MESSAGES _DUMMY_TIME
339 # ifdef USE_LOCALE_MONETARY
340 # define LC_MONETARY_INDEX _DUMMY_MESSAGES + 1
341 # define _DUMMY_MONETARY LC_MONETARY_INDEX
343 # define _DUMMY_MONETARY _DUMMY_MESSAGES
345 # ifdef USE_LOCALE_ADDRESS
346 # define LC_ADDRESS_INDEX _DUMMY_MONETARY + 1
347 # define _DUMMY_ADDRESS LC_ADDRESS_INDEX
349 # define _DUMMY_ADDRESS _DUMMY_MONETARY
351 # ifdef USE_LOCALE_IDENTIFICATION
352 # define LC_IDENTIFICATION_INDEX _DUMMY_ADDRESS + 1
353 # define _DUMMY_IDENTIFICATION LC_IDENTIFICATION_INDEX
355 # define _DUMMY_IDENTIFICATION _DUMMY_ADDRESS
357 # ifdef USE_LOCALE_MEASUREMENT
358 # define LC_MEASUREMENT_INDEX _DUMMY_IDENTIFICATION + 1
359 # define _DUMMY_MEASUREMENT LC_MEASUREMENT_INDEX
361 # define _DUMMY_MEASUREMENT _DUMMY_IDENTIFICATION
363 # ifdef USE_LOCALE_PAPER
364 # define LC_PAPER_INDEX _DUMMY_MEASUREMENT + 1
365 # define _DUMMY_PAPER LC_PAPER_INDEX
367 # define _DUMMY_PAPER _DUMMY_MEASUREMENT
369 # ifdef USE_LOCALE_TELEPHONE
370 # define LC_TELEPHONE_INDEX _DUMMY_PAPER + 1
371 # define _DUMMY_TELEPHONE LC_TELEPHONE_INDEX
373 # define _DUMMY_TELEPHONE _DUMMY_PAPER
376 # define LC_ALL_INDEX _DUMMY_TELEPHONE + 1
378 #endif /* ifdef USE_LOCALE */
380 /* Windows requres a customized base-level setlocale() */
382 # define my_setlocale(cat, locale) win32_setlocale(cat, locale)
384 # define my_setlocale(cat, locale) setlocale(cat, locale)
387 /* Just placeholders for now. "_c" is intended to be called when the category
388 * is a constant known at compile time; "_r", not known until run time */
389 # define do_setlocale_c(category, locale) my_setlocale(category, locale)
390 # define do_setlocale_r(category, locale) my_setlocale(category, locale)
393 S_set_numeric_radix(pTHX_ const bool use_locale)
395 /* If 'use_locale' is FALSE, set to use a dot for the radix character. If
396 * TRUE, use the radix character derived from the current locale */
398 #if defined(USE_LOCALE_NUMERIC) && ( defined(HAS_LOCALECONV) \
399 || defined(HAS_NL_LANGINFO))
401 const char * radix = (use_locale)
402 ? my_nl_langinfo(PERL_RADIXCHAR, FALSE)
403 /* FALSE => already in dest locale */
406 sv_setpv(PL_numeric_radix_sv, radix);
408 /* If this is valid UTF-8 that isn't totally ASCII, and we are in
409 * a UTF-8 locale, then mark the radix as being in UTF-8 */
410 if (is_utf8_non_invariant_string((U8 *) SvPVX(PL_numeric_radix_sv),
411 SvCUR(PL_numeric_radix_sv))
412 && _is_cur_LC_category_utf8(LC_NUMERIC))
414 SvUTF8_on(PL_numeric_radix_sv);
419 if (DEBUG_L_TEST || debug_initialization) {
420 PerlIO_printf(Perl_debug_log, "Locale radix is '%s', ?UTF-8=%d\n",
421 SvPVX(PL_numeric_radix_sv),
422 cBOOL(SvUTF8(PL_numeric_radix_sv)));
426 #endif /* USE_LOCALE_NUMERIC and can find the radix char */
431 S_new_numeric(pTHX_ const char *newnum)
434 #ifndef USE_LOCALE_NUMERIC
436 PERL_UNUSED_ARG(newnum);
440 /* Called after each libc setlocale() call affecting LC_NUMERIC, to tell
441 * core Perl this and that 'newnum' is the name of the new locale.
442 * It installs this locale as the current underlying default.
444 * The default locale and the C locale can be toggled between by use of the
445 * set_numeric_underlying() and set_numeric_standard() functions, which
446 * should probably not be called directly, but only via macros like
447 * SET_NUMERIC_STANDARD() in perl.h.
449 * The toggling is necessary mainly so that a non-dot radix decimal point
450 * character can be output, while allowing internal calculations to use a
453 * This sets several interpreter-level variables:
454 * PL_numeric_name The underlying locale's name: a copy of 'newnum'
455 * PL_numeric_underlying A boolean indicating if the toggled state is such
456 * that the current locale is the program's underlying
458 * PL_numeric_standard An int indicating if the toggled state is such
459 * that the current locale is the C locale or
460 * indistinguishable from the C locale. If non-zero, it
461 * is in C; if > 1, it means it may not be toggled away
463 * PL_numeric_underlying_is_standard A bool kept by this function
464 * indicating that the underlying locale and the standard
465 * C locale are indistinguishable for the purposes of
466 * LC_NUMERIC. This happens when both of the above two
467 * variables are true at the same time. (Toggling is a
468 * no-op under these circumstances.) This variable is
469 * used to avoid having to recalculate.
475 Safefree(PL_numeric_name);
476 PL_numeric_name = NULL;
477 PL_numeric_standard = TRUE;
478 PL_numeric_underlying = TRUE;
479 PL_numeric_underlying_is_standard = TRUE;
483 save_newnum = stdize_locale(savepv(newnum));
484 PL_numeric_underlying = TRUE;
485 PL_numeric_standard = isNAME_C_OR_POSIX(save_newnum);
487 /* If its name isn't C nor POSIX, it could still be indistinguishable from
489 if (! PL_numeric_standard) {
490 PL_numeric_standard = cBOOL(strEQ(".", my_nl_langinfo(PERL_RADIXCHAR,
491 FALSE /* Don't toggle locale */ ))
492 && strEQ("", my_nl_langinfo(PERL_THOUSEP,
496 /* Save the new name if it isn't the same as the previous one, if any */
497 if (! PL_numeric_name || strNE(PL_numeric_name, save_newnum)) {
498 Safefree(PL_numeric_name);
499 PL_numeric_name = save_newnum;
502 Safefree(save_newnum);
505 PL_numeric_underlying_is_standard = PL_numeric_standard;
507 # ifdef HAS_POSIX_2008_LOCALE
509 PL_underlying_numeric_obj = newlocale(LC_NUMERIC_MASK,
511 PL_underlying_numeric_obj);
515 if (DEBUG_L_TEST || debug_initialization) {
516 PerlIO_printf(Perl_debug_log, "Called new_numeric with %s, PL_numeric_name=%s\n", newnum, PL_numeric_name);
519 /* Keep LC_NUMERIC in the C locale. This is for XS modules, so they don't
520 * have to worry about the radix being a non-dot. (Core operations that
521 * need the underlying locale change to it temporarily). */
522 set_numeric_standard();
524 #endif /* USE_LOCALE_NUMERIC */
529 Perl_set_numeric_standard(pTHX)
532 #ifdef USE_LOCALE_NUMERIC
534 /* Toggle the LC_NUMERIC locale to C. Most code should use the macros like
535 * SET_NUMERIC_STANDARD() in perl.h instead of calling this directly. The
536 * macro avoids calling this routine if toggling isn't necessary according
537 * to our records (which could be wrong if some XS code has changed the
538 * locale behind our back) */
540 do_setlocale_c(LC_NUMERIC, "C");
541 PL_numeric_standard = TRUE;
542 PL_numeric_underlying = PL_numeric_underlying_is_standard;
543 set_numeric_radix(0);
547 if (DEBUG_L_TEST || debug_initialization) {
548 PerlIO_printf(Perl_debug_log,
549 "LC_NUMERIC locale now is standard C\n");
553 #endif /* USE_LOCALE_NUMERIC */
558 Perl_set_numeric_underlying(pTHX)
561 #ifdef USE_LOCALE_NUMERIC
563 /* Toggle the LC_NUMERIC locale to the current underlying default. Most
564 * code should use the macros like SET_NUMERIC_UNDERLYING() in perl.h
565 * instead of calling this directly. The macro avoids calling this routine
566 * if toggling isn't necessary according to our records (which could be
567 * wrong if some XS code has changed the locale behind our back) */
569 do_setlocale_c(LC_NUMERIC, PL_numeric_name);
570 PL_numeric_standard = PL_numeric_underlying_is_standard;
571 PL_numeric_underlying = TRUE;
572 set_numeric_radix(! PL_numeric_standard);
576 if (DEBUG_L_TEST || debug_initialization) {
577 PerlIO_printf(Perl_debug_log,
578 "LC_NUMERIC locale now is %s\n",
583 #endif /* USE_LOCALE_NUMERIC */
588 * Set up for a new ctype locale.
591 S_new_ctype(pTHX_ const char *newctype)
594 #ifndef USE_LOCALE_CTYPE
596 PERL_ARGS_ASSERT_NEW_CTYPE;
597 PERL_UNUSED_ARG(newctype);
602 /* Called after each libc setlocale() call affecting LC_CTYPE, to tell
603 * core Perl this and that 'newctype' is the name of the new locale.
605 * This function sets up the folding arrays for all 256 bytes, assuming
606 * that tofold() is tolc() since fold case is not a concept in POSIX,
608 * Any code changing the locale (outside this file) should use
609 * Perl_setlocale or POSIX::setlocale, which call this function. Therefore
610 * this function should be called directly only from this file and from
611 * POSIX::setlocale() */
616 /* Don't check for problems if we are suppressing the warnings */
617 bool check_for_problems = ckWARN_d(WARN_LOCALE) || UNLIKELY(DEBUG_L_TEST);
619 PERL_ARGS_ASSERT_NEW_CTYPE;
621 /* We will replace any bad locale warning with 1) nothing if the new one is
622 * ok; or 2) a new warning for the bad new locale */
623 if (PL_warn_locale) {
624 SvREFCNT_dec_NN(PL_warn_locale);
625 PL_warn_locale = NULL;
628 PL_in_utf8_CTYPE_locale = _is_cur_LC_category_utf8(LC_CTYPE);
630 /* A UTF-8 locale gets standard rules. But note that code still has to
631 * handle this specially because of the three problematic code points */
632 if (PL_in_utf8_CTYPE_locale) {
633 Copy(PL_fold_latin1, PL_fold_locale, 256, U8);
636 /* We don't populate the other lists if a UTF-8 locale, but do check that
637 * everything works as expected, unless checking turned off */
638 if (check_for_problems || ! PL_in_utf8_CTYPE_locale) {
639 /* Assume enough space for every character being bad. 4 spaces each
640 * for the 94 printable characters that are output like "'x' "; and 5
641 * spaces each for "'\\' ", "'\t' ", and "'\n' "; plus a terminating
643 char bad_chars_list[ (94 * 4) + (3 * 5) + 1 ] = { '\0' };
644 bool multi_byte_locale = FALSE; /* Assume is a single-byte locale
646 unsigned int bad_count = 0; /* Count of bad characters */
648 for (i = 0; i < 256; i++) {
649 if (! PL_in_utf8_CTYPE_locale) {
651 PL_fold_locale[i] = (U8) tolower(i);
653 PL_fold_locale[i] = (U8) toupper(i);
655 PL_fold_locale[i] = (U8) i;
658 /* If checking for locale problems, see if the native ASCII-range
659 * printables plus \n and \t are in their expected categories in
660 * the new locale. If not, this could mean big trouble, upending
661 * Perl's and most programs' assumptions, like having a
662 * metacharacter with special meaning become a \w. Fortunately,
663 * it's very rare to find locales that aren't supersets of ASCII
664 * nowadays. It isn't a problem for most controls to be changed
665 * into something else; we check only \n and \t, though perhaps \r
666 * could be an issue as well. */
667 if ( check_for_problems
668 && (isGRAPH_A(i) || isBLANK_A(i) || i == '\n'))
671 char name[3] = { '\0' };
673 /* Convert the name into a string */
678 else if (i == '\n') {
679 my_strlcpy(name, "\n", sizeof(name));
682 my_strlcpy(name, "\t", sizeof(name));
685 /* Check each possibe class */
686 if (UNLIKELY(cBOOL(isalnum(i)) != cBOOL(isALPHANUMERIC_A(i)))) {
688 DEBUG_L(PerlIO_printf(Perl_debug_log,
689 "isalnum('%s') unexpectedly is %d\n",
690 name, cBOOL(isalnum(i))));
692 if (UNLIKELY(cBOOL(isalpha(i)) != cBOOL(isALPHA_A(i)))) {
694 DEBUG_L(PerlIO_printf(Perl_debug_log,
695 "isalpha('%s') unexpectedly is %d\n",
696 name, cBOOL(isalpha(i))));
698 if (UNLIKELY(cBOOL(isdigit(i)) != cBOOL(isDIGIT_A(i)))) {
700 DEBUG_L(PerlIO_printf(Perl_debug_log,
701 "isdigit('%s') unexpectedly is %d\n",
702 name, cBOOL(isdigit(i))));
704 if (UNLIKELY(cBOOL(isgraph(i)) != cBOOL(isGRAPH_A(i)))) {
706 DEBUG_L(PerlIO_printf(Perl_debug_log,
707 "isgraph('%s') unexpectedly is %d\n",
708 name, cBOOL(isgraph(i))));
710 if (UNLIKELY(cBOOL(islower(i)) != cBOOL(isLOWER_A(i)))) {
712 DEBUG_L(PerlIO_printf(Perl_debug_log,
713 "islower('%s') unexpectedly is %d\n",
714 name, cBOOL(islower(i))));
716 if (UNLIKELY(cBOOL(isprint(i)) != cBOOL(isPRINT_A(i)))) {
718 DEBUG_L(PerlIO_printf(Perl_debug_log,
719 "isprint('%s') unexpectedly is %d\n",
720 name, cBOOL(isprint(i))));
722 if (UNLIKELY(cBOOL(ispunct(i)) != cBOOL(isPUNCT_A(i)))) {
724 DEBUG_L(PerlIO_printf(Perl_debug_log,
725 "ispunct('%s') unexpectedly is %d\n",
726 name, cBOOL(ispunct(i))));
728 if (UNLIKELY(cBOOL(isspace(i)) != cBOOL(isSPACE_A(i)))) {
730 DEBUG_L(PerlIO_printf(Perl_debug_log,
731 "isspace('%s') unexpectedly is %d\n",
732 name, cBOOL(isspace(i))));
734 if (UNLIKELY(cBOOL(isupper(i)) != cBOOL(isUPPER_A(i)))) {
736 DEBUG_L(PerlIO_printf(Perl_debug_log,
737 "isupper('%s') unexpectedly is %d\n",
738 name, cBOOL(isupper(i))));
740 if (UNLIKELY(cBOOL(isxdigit(i))!= cBOOL(isXDIGIT_A(i)))) {
742 DEBUG_L(PerlIO_printf(Perl_debug_log,
743 "isxdigit('%s') unexpectedly is %d\n",
744 name, cBOOL(isxdigit(i))));
746 if (UNLIKELY(tolower(i) != (int) toLOWER_A(i))) {
748 DEBUG_L(PerlIO_printf(Perl_debug_log,
749 "tolower('%s')=0x%x instead of the expected 0x%x\n",
750 name, tolower(i), (int) toLOWER_A(i)));
752 if (UNLIKELY(toupper(i) != (int) toUPPER_A(i))) {
754 DEBUG_L(PerlIO_printf(Perl_debug_log,
755 "toupper('%s')=0x%x instead of the expected 0x%x\n",
756 name, toupper(i), (int) toUPPER_A(i)));
758 if (UNLIKELY((i == '\n' && ! isCNTRL_LC(i)))) {
760 DEBUG_L(PerlIO_printf(Perl_debug_log,
761 "'\\n' (=%02X) is not a control\n", (int) i));
764 /* Add to the list; Separate multiple entries with a blank */
767 my_strlcat(bad_chars_list, " ", sizeof(bad_chars_list));
769 my_strlcat(bad_chars_list, name, sizeof(bad_chars_list));
777 /* We only handle single-byte locales (outside of UTF-8 ones; so if
778 * this locale requires more than one byte, there are going to be
780 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
781 "%s:%d: check_for_problems=%d, MB_CUR_MAX=%d\n",
782 __FILE__, __LINE__, check_for_problems, (int) MB_CUR_MAX));
784 if ( check_for_problems && MB_CUR_MAX > 1
785 && ! PL_in_utf8_CTYPE_locale
787 /* Some platforms return MB_CUR_MAX > 1 for even the "C"
788 * locale. Just assume that the implementation for them (plus
789 * for POSIX) is correct and the > 1 value is spurious. (Since
790 * these are specially handled to never be considered UTF-8
791 * locales, as long as this is the only problem, everything
792 * should work fine */
793 && strNE(newctype, "C") && strNE(newctype, "POSIX"))
795 multi_byte_locale = TRUE;
800 if (UNLIKELY(bad_count) || UNLIKELY(multi_byte_locale)) {
801 if (UNLIKELY(bad_count) && PL_in_utf8_CTYPE_locale) {
802 PL_warn_locale = Perl_newSVpvf(aTHX_
803 "Locale '%s' contains (at least) the following characters"
804 " which have\nnon-standard meanings: %s\nThe Perl program"
805 " will use the standard meanings",
806 newctype, bad_chars_list);
810 PL_warn_locale = Perl_newSVpvf(aTHX_
811 "Locale '%s' may not work well.%s%s%s\n",
814 ? " Some characters in it are not recognized by"
818 ? "\nThe following characters (and maybe others)"
819 " may not have the same meaning as the Perl"
820 " program expects:\n"
828 # ifdef HAS_NL_LANGINFO
830 Perl_sv_catpvf(aTHX_ PL_warn_locale, "; codeset=%s",
831 /* parameter FALSE is a don't care here */
832 my_nl_langinfo(PERL_CODESET, FALSE));
836 Perl_sv_catpvf(aTHX_ PL_warn_locale, "\n");
838 /* If we are actually in the scope of the locale or are debugging,
839 * output the message now. If not in that scope, we save the
840 * message to be output at the first operation using this locale,
841 * if that actually happens. Most programs don't use locales, so
842 * they are immune to bad ones. */
843 if (IN_LC(LC_CTYPE) || UNLIKELY(DEBUG_L_TEST)) {
845 /* The '0' below suppresses a bogus gcc compiler warning */
846 Perl_warner(aTHX_ packWARN(WARN_LOCALE), SvPVX(PL_warn_locale), 0);
848 if (IN_LC(LC_CTYPE)) {
849 SvREFCNT_dec_NN(PL_warn_locale);
850 PL_warn_locale = NULL;
856 #endif /* USE_LOCALE_CTYPE */
861 Perl__warn_problematic_locale()
864 #ifdef USE_LOCALE_CTYPE
868 /* Internal-to-core function that outputs the message in PL_warn_locale,
869 * and then NULLS it. Should be called only through the macro
870 * _CHECK_AND_WARN_PROBLEMATIC_LOCALE */
872 if (PL_warn_locale) {
873 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
874 SvPVX(PL_warn_locale),
875 0 /* dummy to avoid compiler warning */ );
876 SvREFCNT_dec_NN(PL_warn_locale);
877 PL_warn_locale = NULL;
885 S_new_collate(pTHX_ const char *newcoll)
888 #ifndef USE_LOCALE_COLLATE
890 PERL_UNUSED_ARG(newcoll);
895 /* Called after each libc setlocale() call affecting LC_COLLATE, to tell
896 * core Perl this and that 'newcoll' is the name of the new locale.
898 * The design of locale collation is that every locale change is given an
899 * index 'PL_collation_ix'. The first time a string particpates in an
900 * operation that requires collation while locale collation is active, it
901 * is given PERL_MAGIC_collxfrm magic (via sv_collxfrm_flags()). That
902 * magic includes the collation index, and the transformation of the string
903 * by strxfrm(), q.v. That transformation is used when doing comparisons,
904 * instead of the string itself. If a string changes, the magic is
905 * cleared. The next time the locale changes, the index is incremented,
906 * and so we know during a comparison that the transformation is not
907 * necessarily still valid, and so is recomputed. Note that if the locale
908 * changes enough times, the index could wrap (a U32), and it is possible
909 * that a transformation would improperly be considered valid, leading to
913 if (PL_collation_name) {
915 Safefree(PL_collation_name);
916 PL_collation_name = NULL;
918 PL_collation_standard = TRUE;
919 is_standard_collation:
920 PL_collxfrm_base = 0;
921 PL_collxfrm_mult = 2;
922 PL_in_utf8_COLLATE_locale = FALSE;
923 PL_strxfrm_NUL_replacement = '\0';
924 PL_strxfrm_max_cp = 0;
928 /* If this is not the same locale as currently, set the new one up */
929 if (! PL_collation_name || strNE(PL_collation_name, newcoll)) {
931 Safefree(PL_collation_name);
932 PL_collation_name = stdize_locale(savepv(newcoll));
933 PL_collation_standard = isNAME_C_OR_POSIX(newcoll);
934 if (PL_collation_standard) {
935 goto is_standard_collation;
938 PL_in_utf8_COLLATE_locale = _is_cur_LC_category_utf8(LC_COLLATE);
939 PL_strxfrm_NUL_replacement = '\0';
940 PL_strxfrm_max_cp = 0;
942 /* A locale collation definition includes primary, secondary, tertiary,
943 * etc. weights for each character. To sort, the primary weights are
944 * used, and only if they compare equal, then the secondary weights are
945 * used, and only if they compare equal, then the tertiary, etc.
947 * strxfrm() works by taking the input string, say ABC, and creating an
948 * output transformed string consisting of first the primary weights,
949 * A¹B¹C¹ followed by the secondary ones, A²B²C²; and then the
950 * tertiary, etc, yielding A¹B¹C¹ A²B²C² A³B³C³ .... Some characters
951 * may not have weights at every level. In our example, let's say B
952 * doesn't have a tertiary weight, and A doesn't have a secondary
953 * weight. The constructed string is then going to be
954 * A¹B¹C¹ B²C² A³C³ ....
955 * This has the desired effect that strcmp() will look at the secondary
956 * or tertiary weights only if the strings compare equal at all higher
957 * priority weights. The spaces shown here, like in
959 * are not just for readability. In the general case, these must
960 * actually be bytes, which we will call here 'separator weights'; and
961 * they must be smaller than any other weight value, but since these
962 * are C strings, only the terminating one can be a NUL (some
963 * implementations may include a non-NUL separator weight just before
964 * the NUL). Implementations tend to reserve 01 for the separator
965 * weights. They are needed so that a shorter string's secondary
966 * weights won't be misconstrued as primary weights of a longer string,
967 * etc. By making them smaller than any other weight, the shorter
968 * string will sort first. (Actually, if all secondary weights are
969 * smaller than all primary ones, there is no need for a separator
970 * weight between those two levels, etc.)
972 * The length of the transformed string is roughly a linear function of
973 * the input string. It's not exactly linear because some characters
974 * don't have weights at all levels. When we call strxfrm() we have to
975 * allocate some memory to hold the transformed string. The
976 * calculations below try to find coefficients 'm' and 'b' for this
977 * locale so that m*x + b equals how much space we need, given the size
978 * of the input string in 'x'. If we calculate too small, we increase
979 * the size as needed, and call strxfrm() again, but it is better to
980 * get it right the first time to avoid wasted expensive string
981 * transformations. */
984 /* We use the string below to find how long the tranformation of it
985 * is. Almost all locales are supersets of ASCII, or at least the
986 * ASCII letters. We use all of them, half upper half lower,
987 * because if we used fewer, we might hit just the ones that are
988 * outliers in a particular locale. Most of the strings being
989 * collated will contain a preponderance of letters, and even if
990 * they are above-ASCII, they are likely to have the same number of
991 * weight levels as the ASCII ones. It turns out that digits tend
992 * to have fewer levels, and some punctuation has more, but those
993 * are relatively sparse in text, and khw believes this gives a
994 * reasonable result, but it could be changed if experience so
996 const char longer[] = "ABCDEFGHIJKLMnopqrstuvwxyz";
997 char * x_longer; /* Transformed 'longer' */
998 Size_t x_len_longer; /* Length of 'x_longer' */
1000 char * x_shorter; /* We also transform a substring of 'longer' */
1001 Size_t x_len_shorter;
1003 /* _mem_collxfrm() is used get the transformation (though here we
1004 * are interested only in its length). It is used because it has
1005 * the intelligence to handle all cases, but to work, it needs some
1006 * values of 'm' and 'b' to get it started. For the purposes of
1007 * this calculation we use a very conservative estimate of 'm' and
1008 * 'b'. This assumes a weight can be multiple bytes, enough to
1009 * hold any UV on the platform, and there are 5 levels, 4 weight
1010 * bytes, and a trailing NUL. */
1011 PL_collxfrm_base = 5;
1012 PL_collxfrm_mult = 5 * sizeof(UV);
1014 /* Find out how long the transformation really is */
1015 x_longer = _mem_collxfrm(longer,
1019 /* We avoid converting to UTF-8 in the
1020 * called function by telling it the
1021 * string is in UTF-8 if the locale is a
1022 * UTF-8 one. Since the string passed
1023 * here is invariant under UTF-8, we can
1024 * claim it's UTF-8 even though it isn't.
1026 PL_in_utf8_COLLATE_locale);
1029 /* Find out how long the transformation of a substring of 'longer'
1030 * is. Together the lengths of these transformations are
1031 * sufficient to calculate 'm' and 'b'. The substring is all of
1032 * 'longer' except the first character. This minimizes the chances
1033 * of being swayed by outliers */
1034 x_shorter = _mem_collxfrm(longer + 1,
1037 PL_in_utf8_COLLATE_locale);
1038 Safefree(x_shorter);
1040 /* If the results are nonsensical for this simple test, the whole
1041 * locale definition is suspect. Mark it so that locale collation
1042 * is not active at all for it. XXX Should we warn? */
1043 if ( x_len_shorter == 0
1044 || x_len_longer == 0
1045 || x_len_shorter >= x_len_longer)
1047 PL_collxfrm_mult = 0;
1048 PL_collxfrm_base = 0;
1051 SSize_t base; /* Temporary */
1053 /* We have both: m * strlen(longer) + b = x_len_longer
1054 * m * strlen(shorter) + b = x_len_shorter;
1055 * subtracting yields:
1056 * m * (strlen(longer) - strlen(shorter))
1057 * = x_len_longer - x_len_shorter
1058 * But we have set things up so that 'shorter' is 1 byte smaller
1059 * than 'longer'. Hence:
1060 * m = x_len_longer - x_len_shorter
1062 * But if something went wrong, make sure the multiplier is at
1065 if (x_len_longer > x_len_shorter) {
1066 PL_collxfrm_mult = (STRLEN) x_len_longer - x_len_shorter;
1069 PL_collxfrm_mult = 1;
1074 * but in case something has gone wrong, make sure it is
1076 base = x_len_longer - PL_collxfrm_mult * (sizeof(longer) - 1);
1081 /* Add 1 for the trailing NUL */
1082 PL_collxfrm_base = base + 1;
1087 if (DEBUG_L_TEST || debug_initialization) {
1088 PerlIO_printf(Perl_debug_log,
1089 "%s:%d: ?UTF-8 locale=%d; x_len_shorter=%zu, "
1091 " collate multipler=%zu, collate base=%zu\n",
1093 PL_in_utf8_COLLATE_locale,
1094 x_len_shorter, x_len_longer,
1095 PL_collxfrm_mult, PL_collxfrm_base);
1102 #endif /* USE_LOCALE_COLLATE */
1109 S_win32_setlocale(pTHX_ int category, const char* locale)
1111 /* This, for Windows, emulates POSIX setlocale() behavior. There is no
1112 * difference between the two unless the input locale is "", which normally
1113 * means on Windows to get the machine default, which is set via the
1114 * computer's "Regional and Language Options" (or its current equivalent).
1115 * In POSIX, it instead means to find the locale from the user's
1116 * environment. This routine changes the Windows behavior to first look in
1117 * the environment, and, if anything is found, use that instead of going to
1118 * the machine default. If there is no environment override, the machine
1119 * default is used, by calling the real setlocale() with "".
1121 * The POSIX behavior is to use the LC_ALL variable if set; otherwise to
1122 * use the particular category's variable if set; otherwise to use the LANG
1125 bool override_LC_ALL = FALSE;
1129 if (locale && strEQ(locale, "")) {
1133 locale = PerlEnv_getenv("LC_ALL");
1135 if (category == LC_ALL) {
1136 override_LC_ALL = TRUE;
1142 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
1143 if (category == categories[i]) {
1144 locale = PerlEnv_getenv(category_names[i]);
1149 locale = PerlEnv_getenv("LANG");
1165 result = setlocale(category, locale);
1166 DEBUG_L(STMT_START {
1168 PerlIO_printf(Perl_debug_log, "%s:%d: %s\n", __FILE__, __LINE__,
1169 setlocale_debug_string(category, locale, result));
1173 if (! override_LC_ALL) {
1177 /* Here the input category was LC_ALL, and we have set it to what is in the
1178 * LANG variable or the system default if there is no LANG. But these have
1179 * lower priority than the other LC_foo variables, so override it for each
1180 * one that is set. (If they are set to "", it means to use the same thing
1181 * we just set LC_ALL to, so can skip) */
1183 for (i = 0; i < LC_ALL_INDEX; i++) {
1184 result = PerlEnv_getenv(category_names[i]);
1185 if (result && strNE(result, "")) {
1186 setlocale(categories[i], result);
1187 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
1189 setlocale_debug_string(categories[i], result, "not captured")));
1193 result = setlocale(LC_ALL, NULL);
1194 DEBUG_L(STMT_START {
1196 PerlIO_printf(Perl_debug_log, "%s:%d: %s\n",
1198 setlocale_debug_string(LC_ALL, NULL, result));
1209 =head1 Locale-related functions and macros
1211 =for apidoc Perl_setlocale
1213 This is an (almost) drop-in replacement for the system L<C<setlocale(3)>>,
1214 taking the same parameters, and returning the same information, except that it
1215 returns the correct underlying C<LC_NUMERIC> locale, instead of C<C> always, as
1216 perl keeps that locale category as C<C>, changing it briefly during the
1217 operations where the underlying one is required.
1219 The other reason it isn't completely a drop-in replacement is that it is
1220 declared to return S<C<const char *>>, whereas the system setlocale omits the
1221 C<const>. (If it were being written today, plain setlocale would be declared
1222 const, since it is illegal to change the information it returns; doing so leads
1225 C<Perl_setlocale> should not be used to change the locale except on systems
1226 where the predefined variable C<${^SAFE_LOCALES}> is 1.
1228 The return points to a per-thread static buffer, which is overwritten the next
1229 time C<Perl_setlocale> is called from the same thread.
1236 Perl_setlocale(const int category, const char * locale)
1238 /* This wraps POSIX::setlocale() */
1240 const char * retval;
1241 const char * newlocale;
1243 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
1246 #ifdef USE_LOCALE_NUMERIC
1248 /* A NULL locale means only query what the current one is. We have the
1249 * LC_NUMERIC name saved, because we are normally switched into the C
1250 * locale for it. For an LC_ALL query, switch back to get the correct
1251 * results. All other categories don't require special handling */
1252 if (locale == NULL) {
1253 if (category == LC_NUMERIC) {
1255 /* We don't have to copy this return value, as it is a per-thread
1256 * variable, and won't change until a future setlocale */
1257 return PL_numeric_name;
1262 else if (category == LC_ALL) {
1263 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
1272 retval = do_setlocale_r(category, locale);
1275 #if defined(USE_LOCALE_NUMERIC) && defined(LC_ALL)
1277 if (locale == NULL && category == LC_ALL) {
1278 RESTORE_LC_NUMERIC();
1283 DEBUG_L(PerlIO_printf(Perl_debug_log,
1284 "%s:%d: %s\n", __FILE__, __LINE__,
1285 setlocale_debug_string(category, locale, retval)));
1293 save_to_buffer(retval, &PL_setlocale_buf, &PL_setlocale_bufsize, 0);
1294 retval = PL_setlocale_buf;
1296 /* If locale == NULL, we are just querying the state */
1297 if (locale == NULL) {
1301 /* Now that have switched locales, we have to update our records to
1306 #ifdef USE_LOCALE_CTYPE
1313 #ifdef USE_LOCALE_COLLATE
1316 new_collate(retval);
1320 #ifdef USE_LOCALE_NUMERIC
1323 new_numeric(retval);
1331 /* LC_ALL updates all the things we care about. The values may not
1332 * be the same as 'retval', as the locale "" may have set things
1335 # ifdef USE_LOCALE_CTYPE
1337 newlocale = do_setlocale_c(LC_CTYPE, NULL);
1338 new_ctype(newlocale);
1340 # endif /* USE_LOCALE_CTYPE */
1341 # ifdef USE_LOCALE_COLLATE
1343 newlocale = do_setlocale_c(LC_COLLATE, NULL);
1344 new_collate(newlocale);
1347 # ifdef USE_LOCALE_NUMERIC
1349 newlocale = do_setlocale_c(LC_NUMERIC, NULL);
1350 new_numeric(newlocale);
1352 # endif /* USE_LOCALE_NUMERIC */
1363 PERL_STATIC_INLINE const char *
1364 S_save_to_buffer(const char * string, char **buf, Size_t *buf_size, const Size_t offset)
1366 /* Copy the NUL-terminated 'string' to 'buf' + 'offset'. 'buf' has size 'buf_size',
1367 * growing it if necessary */
1369 const Size_t string_size = strlen(string) + offset + 1;
1371 PERL_ARGS_ASSERT_SAVE_TO_BUFFER;
1373 if (*buf_size == 0) {
1374 Newx(*buf, string_size, char);
1375 *buf_size = string_size;
1377 else if (string_size > *buf_size) {
1378 Renew(*buf, string_size, char);
1379 *buf_size = string_size;
1382 Copy(string, *buf + offset, string_size - offset, char);
1388 =for apidoc Perl_langinfo
1390 This is an (almost ª) drop-in replacement for the system C<L<nl_langinfo(3)>>,
1391 taking the same C<item> parameter values, and returning the same information.
1392 But it is more thread-safe than regular C<nl_langinfo()>, and hides the quirks
1393 of Perl's locale handling from your code, and can be used on systems that lack
1394 a native C<nl_langinfo>.
1402 It delivers the correct results for the C<RADIXCHAR> and C<THOUSESEP> items,
1403 without you having to write extra code. The reason for the extra code would be
1404 because these are from the C<LC_NUMERIC> locale category, which is normally
1405 kept set to the C locale by Perl, no matter what the underlying locale is
1406 supposed to be, and so to get the expected results, you have to temporarily
1407 toggle into the underlying locale, and later toggle back. (You could use
1408 plain C<nl_langinfo> and C<L</STORE_LC_NUMERIC_FORCE_TO_UNDERLYING>> for this
1409 but then you wouldn't get the other advantages of C<Perl_langinfo()>; not
1410 keeping C<LC_NUMERIC> in the C locale would break a lot of CPAN, which is
1411 expecting the radix (decimal point) character to be a dot.)
1415 Depending on C<item>, it works on systems that don't have C<nl_langinfo>, hence
1416 makes your code more portable. Of the fifty-some possible items specified by
1417 the POSIX 2008 standard,
1418 L<http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/langinfo.h.html>,
1419 only two are completely unimplemented. It uses various techniques to recover
1420 the other items, including calling C<L<localeconv(3)>>, and C<L<strftime(3)>>,
1421 both of which are specified in C89, so should be always be available. Later
1422 C<strftime()> versions have additional capabilities; C<""> is returned for
1423 those not available on your system.
1425 It is important to note that on such systems, this calls C<localeconv>, and so
1426 overwrites the static buffer returned from previous explicit calls to that
1427 function. Thus, if the program doesn't use or save the information from an
1428 explicit C<localeconv> call (which good practice suggests should be done
1429 anyway), use of this function can break it.
1431 The details for those items which may differ from what this emulation returns
1432 and what a native C<nl_langinfo()> would return are:
1440 Unimplemented, so returns C<"">.
1450 Only the values for English are returned. C<YESSTR> and C<NOSTR> have been
1451 removed from POSIX 2008, and are retained for backwards compatibility. Your
1452 platform's C<nl_langinfo> may not support them.
1456 Always evaluates to C<%x>, the locale's appropriate date representation.
1460 Always evaluates to C<%X>, the locale's appropriate time representation.
1464 Always evaluates to C<%c>, the locale's appropriate date and time
1469 The return may be incorrect for those rare locales where the currency symbol
1470 replaces the radix character.
1471 Send email to L<mailto:perlbug@perl.org> if you have examples of it needing
1472 to work differently.
1476 Currently this gives the same results as Linux does.
1477 Send email to L<mailto:perlbug@perl.org> if you have examples of it needing
1478 to work differently.
1484 =item C<ERA_D_T_FMT>
1488 These are derived by using C<strftime()>, and not all versions of that function
1489 know about them. C<""> is returned for these on such systems.
1493 When using C<Perl_langinfo> on systems that don't have a native
1494 C<nl_langinfo()>, you must
1496 #include "perl_langinfo.h"
1498 before the C<perl.h> C<#include>. You can replace your C<langinfo.h>
1499 C<#include> with this one. (Doing it this way keeps out the symbols that plain
1500 C<langinfo.h> imports into the namespace for code that doesn't need it.)
1502 You also should not use the bare C<langinfo.h> item names, but should preface
1503 them with C<PERL_>, so use C<PERL_RADIXCHAR> instead of plain C<RADIXCHAR>.
1504 The C<PERL_I<foo>> versions will also work for this function on systems that do
1505 have a native C<nl_langinfo>.
1509 It is thread-friendly, returning its result in a buffer that won't be
1510 overwritten by another thread, so you don't have to code for that possibility.
1511 The buffer can be overwritten by the next call to C<nl_langinfo> or
1512 C<Perl_langinfo> in the same thread.
1516 ª It returns S<C<const char *>>, whereas plain C<nl_langinfo()> returns S<C<char
1517 *>>, but you are (only by documentation) forbidden to write into the buffer.
1518 By declaring this C<const>, the compiler enforces this restriction. The extra
1519 C<const> is why this isn't an unequivocal drop-in replacement for
1524 The original impetus for C<Perl_langinfo()> was so that code that needs to
1525 find out the current currency symbol, floating point radix character, or digit
1526 grouping separator can use, on all systems, the simpler and more
1527 thread-friendly C<nl_langinfo> API instead of C<L<localeconv(3)>> which is a
1528 pain to make thread-friendly. For other fields returned by C<localeconv>, it
1529 is better to use the methods given in L<perlcall> to call
1530 L<C<POSIX::localeconv()>|POSIX/localeconv>, which is thread-friendly.
1537 #ifdef HAS_NL_LANGINFO
1538 Perl_langinfo(const nl_item item)
1540 Perl_langinfo(const int item)
1543 return my_nl_langinfo(item, TRUE);
1547 #ifdef HAS_NL_LANGINFO
1548 S_my_nl_langinfo(const nl_item item, bool toggle)
1550 S_my_nl_langinfo(const int item, bool toggle)
1555 /* We only need to toggle into the underlying LC_NUMERIC locale for these
1556 * two items, and only if not already there */
1557 if (toggle && (( item != PERL_RADIXCHAR && item != PERL_THOUSEP)
1558 || PL_numeric_underlying))
1563 #if defined(HAS_NL_LANGINFO) /* nl_langinfo() is available. */
1564 #if ! defined(HAS_POSIX_2008_LOCALE)
1566 /* Here, use plain nl_langinfo(), switching to the underlying LC_NUMERIC
1567 * for those items dependent on it. This must be copied to a buffer before
1568 * switching back, as some systems destroy the buffer when setlocale() is
1572 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
1575 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
1578 LOCALE_LOCK; /* Prevent interference from another thread executing
1579 this code section (the only call to nl_langinfo in
1582 save_to_buffer(nl_langinfo(item), &PL_langinfo_buf,
1583 &PL_langinfo_bufsize, 0);
1588 RESTORE_LC_NUMERIC();
1592 # else /* Use nl_langinfo_l(), avoiding both a mutex and changing the locale */
1595 bool do_free = FALSE;
1596 locale_t cur = uselocale((locale_t) 0);
1598 if (cur == LC_GLOBAL_LOCALE) {
1599 cur = duplocale(LC_GLOBAL_LOCALE);
1604 if (PL_underlying_numeric_obj) {
1605 cur = PL_underlying_numeric_obj;
1608 cur = newlocale(LC_NUMERIC_MASK, PL_numeric_name, cur);
1613 save_to_buffer(nl_langinfo_l(item, cur),
1614 &PL_langinfo_buf, &PL_langinfo_bufsize, 0);
1622 if (strEQ(PL_langinfo_buf, "")) {
1623 if (item == PERL_YESSTR) {
1626 if (item == PERL_NOSTR) {
1631 return PL_langinfo_buf;
1633 #else /* Below, emulate nl_langinfo as best we can */
1637 # ifdef HAS_LOCALECONV
1639 const struct lconv* lc;
1640 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
1643 # ifdef HAS_STRFTIME
1646 bool return_format = FALSE; /* Return the %format, not the value */
1647 const char * format;
1651 /* We copy the results to a per-thread buffer, even if not
1652 * multi-threaded. This is in part to simplify this code, and partly
1653 * because we need a buffer anyway for strftime(), and partly because a
1654 * call of localeconv() could otherwise wipe out the buffer, and the
1655 * programmer would not be expecting this, as this is a nl_langinfo()
1656 * substitute after all, so s/he might be thinking their localeconv()
1657 * is safe until another localeconv() call. */
1661 const char * retval;
1663 /* These 2 are unimplemented */
1665 case PERL_ERA: /* For use with strftime() %E modifier */
1670 /* We use only an English set, since we don't know any more */
1671 case PERL_YESEXPR: return "^[+1yY]";
1672 case PERL_YESSTR: return "yes";
1673 case PERL_NOEXPR: return "^[-0nN]";
1674 case PERL_NOSTR: return "no";
1676 # ifdef HAS_LOCALECONV
1680 /* We don't bother with localeconv_l() because any system that
1681 * has it is likely to also have nl_langinfo() */
1683 LOCALE_LOCK; /* Prevent interference with other threads
1684 using localeconv() */
1688 || ! lc->currency_symbol
1689 || strEQ("", lc->currency_symbol))
1695 /* Leave the first spot empty to be filled in below */
1696 save_to_buffer(lc->currency_symbol, &PL_langinfo_buf,
1697 &PL_langinfo_bufsize, 1);
1698 if (lc->mon_decimal_point && strEQ(lc->mon_decimal_point, ""))
1699 { /* khw couldn't figure out how the localedef specifications
1700 would show that the $ should replace the radix; this is
1701 just a guess as to how it might work.*/
1702 *PL_langinfo_buf = '.';
1704 else if (lc->p_cs_precedes) {
1705 *PL_langinfo_buf = '-';
1708 *PL_langinfo_buf = '+';
1714 case PERL_RADIXCHAR:
1718 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
1721 LOCALE_LOCK; /* Prevent interference with other threads
1722 using localeconv() */
1729 retval = (item == PERL_RADIXCHAR)
1731 : lc->thousands_sep;
1737 save_to_buffer(retval, &PL_langinfo_buf,
1738 &PL_langinfo_bufsize, 0);
1743 RESTORE_LC_NUMERIC();
1749 # ifdef HAS_STRFTIME
1751 /* These are defined by C89, so we assume that strftime supports
1752 * them, and so are returned unconditionally; they may not be what
1753 * the locale actually says, but should give good enough results
1754 * for someone using them as formats (as opposed to trying to parse
1755 * them to figure out what the locale says). The other format
1756 * items are actually tested to verify they work on the platform */
1757 case PERL_D_FMT: return "%x";
1758 case PERL_T_FMT: return "%X";
1759 case PERL_D_T_FMT: return "%c";
1761 /* These formats are only available in later strfmtime's */
1762 case PERL_ERA_D_FMT: case PERL_ERA_T_FMT: case PERL_ERA_D_T_FMT:
1763 case PERL_T_FMT_AMPM:
1765 /* The rest can be gotten from most versions of strftime(). */
1766 case PERL_ABDAY_1: case PERL_ABDAY_2: case PERL_ABDAY_3:
1767 case PERL_ABDAY_4: case PERL_ABDAY_5: case PERL_ABDAY_6:
1769 case PERL_ALT_DIGITS:
1770 case PERL_AM_STR: case PERL_PM_STR:
1771 case PERL_ABMON_1: case PERL_ABMON_2: case PERL_ABMON_3:
1772 case PERL_ABMON_4: case PERL_ABMON_5: case PERL_ABMON_6:
1773 case PERL_ABMON_7: case PERL_ABMON_8: case PERL_ABMON_9:
1774 case PERL_ABMON_10: case PERL_ABMON_11: case PERL_ABMON_12:
1775 case PERL_DAY_1: case PERL_DAY_2: case PERL_DAY_3: case PERL_DAY_4:
1776 case PERL_DAY_5: case PERL_DAY_6: case PERL_DAY_7:
1777 case PERL_MON_1: case PERL_MON_2: case PERL_MON_3: case PERL_MON_4:
1778 case PERL_MON_5: case PERL_MON_6: case PERL_MON_7: case PERL_MON_8:
1779 case PERL_MON_9: case PERL_MON_10: case PERL_MON_11:
1784 init_tm(&tm); /* Precaution against core dumps */
1788 tm.tm_year = 2017 - 1900;
1795 "panic: %s: %d: switch case: %d problem",
1796 __FILE__, __LINE__, item);
1797 NOT_REACHED; /* NOTREACHED */
1799 case PERL_PM_STR: tm.tm_hour = 18;
1804 case PERL_ABDAY_7: tm.tm_wday++;
1805 case PERL_ABDAY_6: tm.tm_wday++;
1806 case PERL_ABDAY_5: tm.tm_wday++;
1807 case PERL_ABDAY_4: tm.tm_wday++;
1808 case PERL_ABDAY_3: tm.tm_wday++;
1809 case PERL_ABDAY_2: tm.tm_wday++;
1814 case PERL_DAY_7: tm.tm_wday++;
1815 case PERL_DAY_6: tm.tm_wday++;
1816 case PERL_DAY_5: tm.tm_wday++;
1817 case PERL_DAY_4: tm.tm_wday++;
1818 case PERL_DAY_3: tm.tm_wday++;
1819 case PERL_DAY_2: tm.tm_wday++;
1824 case PERL_ABMON_12: tm.tm_mon++;
1825 case PERL_ABMON_11: tm.tm_mon++;
1826 case PERL_ABMON_10: tm.tm_mon++;
1827 case PERL_ABMON_9: tm.tm_mon++;
1828 case PERL_ABMON_8: tm.tm_mon++;
1829 case PERL_ABMON_7: tm.tm_mon++;
1830 case PERL_ABMON_6: tm.tm_mon++;
1831 case PERL_ABMON_5: tm.tm_mon++;
1832 case PERL_ABMON_4: tm.tm_mon++;
1833 case PERL_ABMON_3: tm.tm_mon++;
1834 case PERL_ABMON_2: tm.tm_mon++;
1839 case PERL_MON_12: tm.tm_mon++;
1840 case PERL_MON_11: tm.tm_mon++;
1841 case PERL_MON_10: tm.tm_mon++;
1842 case PERL_MON_9: tm.tm_mon++;
1843 case PERL_MON_8: tm.tm_mon++;
1844 case PERL_MON_7: tm.tm_mon++;
1845 case PERL_MON_6: tm.tm_mon++;
1846 case PERL_MON_5: tm.tm_mon++;
1847 case PERL_MON_4: tm.tm_mon++;
1848 case PERL_MON_3: tm.tm_mon++;
1849 case PERL_MON_2: tm.tm_mon++;
1854 case PERL_T_FMT_AMPM:
1856 return_format = TRUE;
1859 case PERL_ERA_D_FMT:
1861 return_format = TRUE;
1864 case PERL_ERA_T_FMT:
1866 return_format = TRUE;
1869 case PERL_ERA_D_T_FMT:
1871 return_format = TRUE;
1874 case PERL_ALT_DIGITS:
1876 format = "%Ow"; /* Find the alternate digit for 0 */
1880 /* We can't use my_strftime() because it doesn't look at
1882 while (0 == strftime(PL_langinfo_buf, PL_langinfo_bufsize,
1885 /* A zero return means one of:
1886 * a) there wasn't enough space in PL_langinfo_buf
1887 * b) the format, like a plain %p, returns empty
1888 * c) it was an illegal format, though some
1889 * implementations of strftime will just return the
1890 * illegal format as a plain character sequence.
1892 * To quickly test for case 'b)', try again but precede
1893 * the format with a plain character. If that result is
1894 * still empty, the problem is either 'a)' or 'c)' */
1896 Size_t format_size = strlen(format) + 1;
1897 Size_t mod_size = format_size + 1;
1901 Newx(mod_format, mod_size, char);
1902 Newx(temp_result, PL_langinfo_bufsize, char);
1904 my_strlcpy(mod_format + 1, format, mod_size);
1905 len = strftime(temp_result,
1906 PL_langinfo_bufsize,
1908 Safefree(mod_format);
1909 Safefree(temp_result);
1911 /* If 'len' is non-zero, it means that we had a case like
1912 * %p which means the current locale doesn't use a.m. or
1913 * p.m., and that is valid */
1916 /* Here, still didn't work. If we get well beyond a
1917 * reasonable size, bail out to prevent an infinite
1920 if (PL_langinfo_bufsize > 100 * format_size) {
1921 *PL_langinfo_buf = '\0';
1924 /* Double the buffer size to retry; Add 1 in case
1925 * original was 0, so we aren't stuck at 0. */
1926 PL_langinfo_bufsize *= 2;
1927 PL_langinfo_bufsize++;
1928 Renew(PL_langinfo_buf, PL_langinfo_bufsize, char);
1936 /* Here, we got a result.
1938 * If the item is 'ALT_DIGITS', PL_langinfo_buf contains the
1939 * alternate format for wday 0. If the value is the same as
1940 * the normal 0, there isn't an alternate, so clear the buffer.
1942 if ( item == PERL_ALT_DIGITS
1943 && strEQ(PL_langinfo_buf, "0"))
1945 *PL_langinfo_buf = '\0';
1948 /* ALT_DIGITS is problematic. Experiments on it showed that
1949 * strftime() did not always work properly when going from
1950 * alt-9 to alt-10. Only a few locales have this item defined,
1951 * and in all of them on Linux that khw was able to find,
1952 * nl_langinfo() merely returned the alt-0 character, possibly
1953 * doubled. Most Unicode digits are in blocks of 10
1954 * consecutive code points, so that is sufficient information
1955 * for those scripts, as we can infer alt-1, alt-2, .... But
1956 * for a Japanese locale, a CJK ideographic 0 is returned, and
1957 * the CJK digits are not in code point order, so you can't
1958 * really infer anything. The localedef for this locale did
1959 * specify the succeeding digits, so that strftime() works
1960 * properly on them, without needing to infer anything. But
1961 * the nl_langinfo() return did not give sufficient information
1962 * for the caller to understand what's going on. So until
1963 * there is evidence that it should work differently, this
1964 * returns the alt-0 string for ALT_DIGITS.
1966 * wday was chosen because its range is all a single digit.
1967 * Things like tm_sec have two digits as the minimum: '00' */
1971 /* If to return the format, not the value, overwrite the buffer
1972 * with it. But some strftime()s will keep the original format
1973 * if illegal, so change those to "" */
1974 if (return_format) {
1975 if (strEQ(PL_langinfo_buf, format)) {
1976 *PL_langinfo_buf = '\0';
1979 save_to_buffer(format, &PL_langinfo_buf,
1980 &PL_langinfo_bufsize, 0);
1991 return PL_langinfo_buf;
1998 * Initialize locale awareness.
2001 Perl_init_i18nl10n(pTHX_ int printwarn)
2005 * 0 if not to output warning when setup locale is bad
2006 * 1 if to output warning based on value of PERL_BADLANG
2007 * >1 if to output regardless of PERL_BADLANG
2010 * 1 = set ok or not applicable,
2011 * 0 = fallback to a locale of lower priority
2012 * -1 = fallback to all locales failed, not even to the C locale
2014 * Under -DDEBUGGING, if the environment variable PERL_DEBUG_LOCALE_INIT is
2015 * set, debugging information is output.
2017 * This looks more complicated than it is, mainly due to the #ifdefs.
2019 * We try to set LC_ALL to the value determined by the environment. If
2020 * there is no LC_ALL on this platform, we try the individual categories we
2021 * know about. If this works, we are done.
2023 * But if it doesn't work, we have to do something else. We search the
2024 * environment variables ourselves instead of relying on the system to do
2025 * it. We look at, in order, LC_ALL, LANG, a system default locale (if we
2026 * think there is one), and the ultimate fallback "C". This is all done in
2027 * the same loop as above to avoid duplicating code, but it makes things
2028 * more complex. The 'trial_locales' array is initialized with just one
2029 * element; it causes the behavior described in the paragraph above this to
2030 * happen. If that fails, we add elements to 'trial_locales', and do extra
2031 * loop iterations to cause the behavior described in this paragraph.
2033 * On Ultrix, the locale MUST come from the environment, so there is
2034 * preliminary code to set it. I (khw) am not sure that it is necessary,
2035 * and that this couldn't be folded into the loop, but barring any real
2036 * platforms to test on, it's staying as-is
2038 * A slight complication is that in embedded Perls, the locale may already
2039 * be set-up, and we don't want to get it from the normal environment
2040 * variables. This is handled by having a special environment variable
2041 * indicate we're in this situation. We simply set setlocale's 2nd
2042 * parameter to be a NULL instead of "". That indicates to setlocale that
2043 * it is not to change anything, but to return the current value,
2044 * effectively initializing perl's db to what the locale already is.
2046 * We play the same trick with NULL if a LC_ALL succeeds. We call
2047 * setlocale() on the individual categores with NULL to get their existing
2048 * values for our db, instead of trying to change them.
2055 PERL_UNUSED_ARG(printwarn);
2057 #else /* USE_LOCALE */
2060 const char * const language = savepv(PerlEnv_getenv("LANGUAGE"));
2064 /* NULL uses the existing already set up locale */
2065 const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
2068 const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
2069 unsigned int trial_locales_count;
2070 const char * const lc_all = savepv(PerlEnv_getenv("LC_ALL"));
2071 const char * const lang = savepv(PerlEnv_getenv("LANG"));
2072 bool setlocale_failure = FALSE;
2075 /* A later getenv() could zap this, so only use here */
2076 const char * const bad_lang_use_once = PerlEnv_getenv("PERL_BADLANG");
2078 const bool locwarn = (printwarn > 1
2080 && ( ! bad_lang_use_once
2082 /* disallow with "" or "0" */
2084 && strNE("0", bad_lang_use_once)))));
2086 /* setlocale() return vals; not copied so must be looked at immediately */
2087 const char * sl_result[NOMINAL_LC_ALL_INDEX + 1];
2089 /* current locale for given category; should have been copied so aren't
2091 const char * curlocales[NOMINAL_LC_ALL_INDEX + 1];
2095 /* In some systems you can find out the system default locale
2096 * and use that as the fallback locale. */
2097 # define SYSTEM_DEFAULT_LOCALE
2099 # ifdef SYSTEM_DEFAULT_LOCALE
2101 const char *system_default_locale = NULL;
2106 # define DEBUG_LOCALE_INIT(a,b,c)
2109 DEBUG_INITIALIZATION_set(cBOOL(PerlEnv_getenv("PERL_DEBUG_LOCALE_INIT")));
2111 # define DEBUG_LOCALE_INIT(category, locale, result) \
2113 if (debug_initialization) { \
2114 PerlIO_printf(Perl_debug_log, \
2116 __FILE__, __LINE__, \
2117 setlocale_debug_string(category, \
2123 /* Make sure the parallel arrays are properly set up */
2124 # ifdef USE_LOCALE_NUMERIC
2125 assert(categories[LC_NUMERIC_INDEX] == LC_NUMERIC);
2126 assert(strEQ(category_names[LC_NUMERIC_INDEX], "LC_NUMERIC"));
2128 # ifdef USE_LOCALE_CTYPE
2129 assert(categories[LC_CTYPE_INDEX] == LC_CTYPE);
2130 assert(strEQ(category_names[LC_CTYPE_INDEX], "LC_CTYPE"));
2132 # ifdef USE_LOCALE_COLLATE
2133 assert(categories[LC_COLLATE_INDEX] == LC_COLLATE);
2134 assert(strEQ(category_names[LC_COLLATE_INDEX], "LC_COLLATE"));
2136 # ifdef USE_LOCALE_TIME
2137 assert(categories[LC_TIME_INDEX] == LC_TIME);
2138 assert(strEQ(category_names[LC_TIME_INDEX], "LC_TIME"));
2140 # ifdef USE_LOCALE_MESSAGES
2141 assert(categories[LC_MESSAGES_INDEX] == LC_MESSAGES);
2142 assert(strEQ(category_names[LC_MESSAGES_INDEX], "LC_MESSAGES"));
2144 # ifdef USE_LOCALE_MONETARY
2145 assert(categories[LC_MONETARY_INDEX] == LC_MONETARY);
2146 assert(strEQ(category_names[LC_MONETARY_INDEX], "LC_MONETARY"));
2148 # ifdef USE_LOCALE_ADDRESS
2149 assert(categories[LC_ADDRESS_INDEX] == LC_ADDRESS);
2150 assert(strEQ(category_names[LC_ADDRESS_INDEX], "LC_ADDRESS"));
2152 # ifdef USE_LOCALE_IDENTIFICATION
2153 assert(categories[LC_IDENTIFICATION_INDEX] == LC_IDENTIFICATION);
2154 assert(strEQ(category_names[LC_IDENTIFICATION_INDEX], "LC_IDENTIFICATION"));
2156 # ifdef USE_LOCALE_MEASUREMENT
2157 assert(categories[LC_MEASUREMENT_INDEX] == LC_MEASUREMENT);
2158 assert(strEQ(category_names[LC_MEASUREMENT_INDEX], "LC_MEASUREMENT"));
2160 # ifdef USE_LOCALE_PAPER
2161 assert(categories[LC_PAPER_INDEX] == LC_PAPER);
2162 assert(strEQ(category_names[LC_PAPER_INDEX], "LC_PAPER"));
2164 # ifdef USE_LOCALE_TELEPHONE
2165 assert(categories[LC_TELEPHONE_INDEX] == LC_TELEPHONE);
2166 assert(strEQ(category_names[LC_TELEPHONE_INDEX], "LC_TELEPHONE"));
2169 assert(categories[LC_ALL_INDEX] == LC_ALL);
2170 assert(strEQ(category_names[LC_ALL_INDEX], "LC_ALL"));
2171 assert(NOMINAL_LC_ALL_INDEX == LC_ALL_INDEX);
2173 # endif /* DEBUGGING */
2175 /* Initialize the cache of the program's UTF-8ness for the always known
2176 * locales C and POSIX */
2177 my_strlcpy(PL_locale_utf8ness, C_and_POSIX_utf8ness,
2178 sizeof(PL_locale_utf8ness));
2180 PL_numeric_radix_sv = newSVpvs(".");
2182 # ifdef LOCALE_ENVIRON_REQUIRED
2185 * Ultrix setlocale(..., "") fails if there are no environment
2186 * variables from which to get a locale name.
2190 # error Ultrix without LC_ALL not implemented
2196 sl_result[LC_ALL_INDEX] = do_setlocale_c(LC_ALL, setlocale_init);
2197 DEBUG_LOCALE_INIT(LC_ALL, setlocale_init, sl_result[LC_ALL_INDEX]);
2198 if (sl_result[LC_ALL_INDEX])
2201 setlocale_failure = TRUE;
2203 if (! setlocale_failure) {
2204 const char * locale_param;
2205 for (i = 0; i < LC_ALL_INDEX; i++) {
2206 locale_param = (! done && (lang || PerlEnv_getenv(category_names[i])))
2209 sl_result[i] = do_setlocale_r(categories[i], locale_param);
2210 if (! sl_result[i]) {
2211 setlocale_failure = TRUE;
2213 DEBUG_LOCALE_INIT(categories[i], locale_param, sl_result[i]);
2218 # endif /* LC_ALL */
2219 # endif /* LOCALE_ENVIRON_REQUIRED */
2221 /* We try each locale in the list until we get one that works, or exhaust
2222 * the list. Normally the loop is executed just once. But if setting the
2223 * locale fails, inside the loop we add fallback trials to the array and so
2224 * will execute the loop multiple times */
2225 trial_locales[0] = setlocale_init;
2226 trial_locales_count = 1;
2228 for (i= 0; i < trial_locales_count; i++) {
2229 const char * trial_locale = trial_locales[i];
2233 /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
2234 * when i==0, but I (khw) don't think that behavior makes much
2236 setlocale_failure = FALSE;
2238 # ifdef SYSTEM_DEFAULT_LOCALE
2239 # ifdef WIN32 /* Note that assumes Win32 has LC_ALL */
2241 /* On Windows machines, an entry of "" after the 0th means to use
2242 * the system default locale, which we now proceed to get. */
2243 if (strEQ(trial_locale, "")) {
2246 /* Note that this may change the locale, but we are going to do
2247 * that anyway just below */
2248 system_default_locale = do_setlocale_c(LC_ALL, "");
2249 DEBUG_LOCALE_INIT(LC_ALL, "", system_default_locale);
2251 /* Skip if invalid or if it's already on the list of locales to
2253 if (! system_default_locale) {
2254 goto next_iteration;
2256 for (j = 0; j < trial_locales_count; j++) {
2257 if (strEQ(system_default_locale, trial_locales[j])) {
2258 goto next_iteration;
2262 trial_locale = system_default_locale;
2265 # error SYSTEM_DEFAULT_LOCALE only implemented for Win32
2267 # endif /* SYSTEM_DEFAULT_LOCALE */
2273 sl_result[LC_ALL_INDEX] = do_setlocale_c(LC_ALL, trial_locale);
2274 DEBUG_LOCALE_INIT(LC_ALL, trial_locale, sl_result[LC_ALL_INDEX]);
2275 if (! sl_result[LC_ALL_INDEX]) {
2276 setlocale_failure = TRUE;
2279 /* Since LC_ALL succeeded, it should have changed all the other
2280 * categories it can to its value; so we massage things so that the
2281 * setlocales below just return their category's current values.
2282 * This adequately handles the case in NetBSD where LC_COLLATE may
2283 * not be defined for a locale, and setting it individually will
2284 * fail, whereas setting LC_ALL succeeds, leaving LC_COLLATE set to
2285 * the POSIX locale. */
2286 trial_locale = NULL;
2289 # endif /* LC_ALL */
2291 if (! setlocale_failure) {
2293 for (j = 0; j < NOMINAL_LC_ALL_INDEX; j++) {
2295 = savepv(do_setlocale_r(categories[j], trial_locale));
2296 if (! curlocales[j]) {
2297 setlocale_failure = TRUE;
2299 DEBUG_LOCALE_INIT(categories[j], trial_locale, curlocales[j]);
2302 if (! setlocale_failure) { /* All succeeded */
2303 break; /* Exit trial_locales loop */
2307 /* Here, something failed; will need to try a fallback. */
2313 if (locwarn) { /* Output failure info only on the first one */
2317 PerlIO_printf(Perl_error_log,
2318 "perl: warning: Setting locale failed.\n");
2320 # else /* !LC_ALL */
2322 PerlIO_printf(Perl_error_log,
2323 "perl: warning: Setting locale failed for the categories:\n\t");
2325 for (j = 0; j < NOMINAL_LC_ALL_INDEX; j++) {
2326 if (! curlocales[j]) {
2327 PerlIO_printf(Perl_error_log, category_names[j]);
2330 Safefree(curlocales[j]);
2334 # endif /* LC_ALL */
2336 PerlIO_printf(Perl_error_log,
2337 "perl: warning: Please check that your locale settings:\n");
2341 PerlIO_printf(Perl_error_log,
2342 "\tLANGUAGE = %c%s%c,\n",
2343 language ? '"' : '(',
2344 language ? language : "unset",
2345 language ? '"' : ')');
2348 PerlIO_printf(Perl_error_log,
2349 "\tLC_ALL = %c%s%c,\n",
2351 lc_all ? lc_all : "unset",
2352 lc_all ? '"' : ')');
2354 # if defined(USE_ENVIRON_ARRAY)
2359 /* Look through the environment for any variables of the
2360 * form qr/ ^ LC_ [A-Z]+ = /x, except LC_ALL which was
2361 * already handled above. These are assumed to be locale
2362 * settings. Output them and their values. */
2363 for (e = environ; *e; e++) {
2364 const STRLEN prefix_len = sizeof("LC_") - 1;
2367 if ( strBEGINs(*e, "LC_")
2368 && ! strBEGINs(*e, "LC_ALL=")
2369 && (uppers_len = strspn(*e + prefix_len,
2370 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"))
2371 && ((*e)[prefix_len + uppers_len] == '='))
2373 PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
2374 (int) (prefix_len + uppers_len), *e,
2375 *e + prefix_len + uppers_len + 1);
2382 PerlIO_printf(Perl_error_log,
2383 "\t(possibly more locale environment variables)\n");
2387 PerlIO_printf(Perl_error_log,
2388 "\tLANG = %c%s%c\n",
2390 lang ? lang : "unset",
2393 PerlIO_printf(Perl_error_log,
2394 " are supported and installed on your system.\n");
2397 /* Calculate what fallback locales to try. We have avoided this
2398 * until we have to, because failure is quite unlikely. This will
2399 * usually change the upper bound of the loop we are in.
2401 * Since the system's default way of setting the locale has not
2402 * found one that works, We use Perl's defined ordering: LC_ALL,
2403 * LANG, and the C locale. We don't try the same locale twice, so
2404 * don't add to the list if already there. (On POSIX systems, the
2405 * LC_ALL element will likely be a repeat of the 0th element "",
2406 * but there's no harm done by doing it explicitly.
2408 * Note that this tries the LC_ALL environment variable even on
2409 * systems which have no LC_ALL locale setting. This may or may
2410 * not have been originally intentional, but there's no real need
2411 * to change the behavior. */
2413 for (j = 0; j < trial_locales_count; j++) {
2414 if (strEQ(lc_all, trial_locales[j])) {
2418 trial_locales[trial_locales_count++] = lc_all;
2423 for (j = 0; j < trial_locales_count; j++) {
2424 if (strEQ(lang, trial_locales[j])) {
2428 trial_locales[trial_locales_count++] = lang;
2432 # if defined(WIN32) && defined(LC_ALL)
2434 /* For Windows, we also try the system default locale before "C".
2435 * (If there exists a Windows without LC_ALL we skip this because
2436 * it gets too complicated. For those, the "C" is the next
2437 * fallback possibility). The "" is the same as the 0th element of
2438 * the array, but the code at the loop above knows to treat it
2439 * differently when not the 0th */
2440 trial_locales[trial_locales_count++] = "";
2444 for (j = 0; j < trial_locales_count; j++) {
2445 if (strEQ("C", trial_locales[j])) {
2449 trial_locales[trial_locales_count++] = "C";
2452 } /* end of first time through the loop */
2460 } /* end of looping through the trial locales */
2462 if (ok < 1) { /* If we tried to fallback */
2464 if (! setlocale_failure) { /* fallback succeeded */
2465 msg = "Falling back to";
2467 else { /* fallback failed */
2470 /* We dropped off the end of the loop, so have to decrement i to
2471 * get back to the value the last time through */
2475 msg = "Failed to fall back to";
2477 /* To continue, we should use whatever values we've got */
2479 for (j = 0; j < NOMINAL_LC_ALL_INDEX; j++) {
2480 Safefree(curlocales[j]);
2481 curlocales[j] = savepv(do_setlocale_r(categories[j], NULL));
2482 DEBUG_LOCALE_INIT(categories[j], NULL, curlocales[j]);
2487 const char * description;
2488 const char * name = "";
2489 if (strEQ(trial_locales[i], "C")) {
2490 description = "the standard locale";
2494 # ifdef SYSTEM_DEFAULT_LOCALE
2496 else if (strEQ(trial_locales[i], "")) {
2497 description = "the system default locale";
2498 if (system_default_locale) {
2499 name = system_default_locale;
2503 # endif /* SYSTEM_DEFAULT_LOCALE */
2506 description = "a fallback locale";
2507 name = trial_locales[i];
2509 if (name && strNE(name, "")) {
2510 PerlIO_printf(Perl_error_log,
2511 "perl: warning: %s %s (\"%s\").\n", msg, description, name);
2514 PerlIO_printf(Perl_error_log,
2515 "perl: warning: %s %s.\n", msg, description);
2518 } /* End of tried to fallback */
2520 /* Done with finding the locales; update our records */
2522 # ifdef USE_LOCALE_CTYPE
2524 new_ctype(curlocales[LC_CTYPE_INDEX]);
2527 # ifdef USE_LOCALE_COLLATE
2529 new_collate(curlocales[LC_COLLATE_INDEX]);
2532 # ifdef USE_LOCALE_NUMERIC
2534 new_numeric(curlocales[LC_NUMERIC_INDEX]);
2538 for (i = 0; i < NOMINAL_LC_ALL_INDEX; i++) {
2540 # if defined(USE_ITHREADS)
2542 /* This caches whether each category's locale is UTF-8 or not. This
2543 * may involve changing the locale. It is ok to do this at
2544 * initialization time before any threads have started, but not later.
2545 * Caching means that if the program heeds our dictate not to change
2546 * locales in threaded applications, this data will remain valid, and
2547 * it may get queried without having to change locales. If the
2548 * environment is such that all categories have the same locale, this
2549 * isn't needed, as the code will not change the locale; but this
2550 * handles the uncommon case where the environment has disparate
2551 * locales for the categories */
2552 (void) _is_cur_LC_category_utf8(categories[i]);
2556 Safefree(curlocales[i]);
2559 # if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
2561 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
2562 * locale is UTF-8. The call to new_ctype() just above has already
2563 * calculated the latter value and saved it in PL_in_utf8_CTYPE_locale. If
2564 * both PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
2565 * are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer on
2566 * STDIN, STDOUT, STDERR, _and_ the default open discipline. */
2567 PL_utf8locale = PL_in_utf8_CTYPE_locale;
2569 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
2570 This is an alternative to using the -C command line switch
2571 (the -C if present will override this). */
2573 const char *p = PerlEnv_getenv("PERL_UNICODE");
2574 PL_unicode = p ? parse_unicode_opts(&p) : 0;
2575 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
2589 #endif /* USE_LOCALE */
2592 /* So won't continue to output stuff */
2593 DEBUG_INITIALIZATION_set(FALSE);
2600 #ifdef USE_LOCALE_COLLATE
2603 Perl__mem_collxfrm(pTHX_ const char *input_string,
2604 STRLEN len, /* Length of 'input_string' */
2605 STRLEN *xlen, /* Set to length of returned string
2606 (not including the collation index
2608 bool utf8 /* Is the input in UTF-8? */
2612 /* _mem_collxfrm() is a bit like strxfrm() but with two important
2613 * differences. First, it handles embedded NULs. Second, it allocates a bit
2614 * more memory than needed for the transformed data itself. The real
2615 * transformed data begins at offset COLLXFRM_HDR_LEN. *xlen is set to
2616 * the length of that, and doesn't include the collation index size.
2617 * Please see sv_collxfrm() to see how this is used. */
2619 #define COLLXFRM_HDR_LEN sizeof(PL_collation_ix)
2621 char * s = (char *) input_string;
2622 STRLEN s_strlen = strlen(input_string);
2624 STRLEN xAlloc; /* xalloc is a reserved word in VC */
2625 STRLEN length_in_chars;
2626 bool first_time = TRUE; /* Cleared after first loop iteration */
2628 PERL_ARGS_ASSERT__MEM_COLLXFRM;
2630 /* Must be NUL-terminated */
2631 assert(*(input_string + len) == '\0');
2633 /* If this locale has defective collation, skip */
2634 if (PL_collxfrm_base == 0 && PL_collxfrm_mult == 0) {
2635 DEBUG_L(PerlIO_printf(Perl_debug_log,
2636 "_mem_collxfrm: locale's collation is defective\n"));
2640 /* Replace any embedded NULs with the control that sorts before any others.
2641 * This will give as good as possible results on strings that don't
2642 * otherwise contain that character, but otherwise there may be
2643 * less-than-perfect results with that character and NUL. This is
2644 * unavoidable unless we replace strxfrm with our own implementation. */
2645 if (UNLIKELY(s_strlen < len)) { /* Only execute if there is an embedded
2649 STRLEN sans_nuls_len;
2650 int try_non_controls;
2651 char this_replacement_char[] = "?\0"; /* Room for a two-byte string,
2652 making sure 2nd byte is NUL.
2654 STRLEN this_replacement_len;
2656 /* If we don't know what non-NUL control character sorts lowest for
2657 * this locale, find it */
2658 if (PL_strxfrm_NUL_replacement == '\0') {
2660 char * cur_min_x = NULL; /* The min_char's xfrm, (except it also
2661 includes the collation index
2664 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "Looking to replace NUL\n"));
2666 /* Unlikely, but it may be that no control will work to replace
2667 * NUL, in which case we instead look for any character. Controls
2668 * are preferred because collation order is, in general, context
2669 * sensitive, with adjoining characters affecting the order, and
2670 * controls are less likely to have such interactions, allowing the
2671 * NUL-replacement to stand on its own. (Another way to look at it
2672 * is to imagine what would happen if the NUL were replaced by a
2673 * combining character; it wouldn't work out all that well.) */
2674 for (try_non_controls = 0;
2675 try_non_controls < 2;
2678 /* Look through all legal code points (NUL isn't) */
2679 for (j = 1; j < 256; j++) {
2680 char * x; /* j's xfrm plus collation index */
2681 STRLEN x_len; /* length of 'x' */
2682 STRLEN trial_len = 1;
2683 char cur_source[] = { '\0', '\0' };
2685 /* Skip non-controls the first time through the loop. The
2686 * controls in a UTF-8 locale are the L1 ones */
2687 if (! try_non_controls && (PL_in_utf8_COLLATE_locale)
2694 /* Create a 1-char string of the current code point */
2695 cur_source[0] = (char) j;
2697 /* Then transform it */
2698 x = _mem_collxfrm(cur_source, trial_len, &x_len,
2699 0 /* The string is not in UTF-8 */);
2701 /* Ignore any character that didn't successfully transform.
2707 /* If this character's transformation is lower than
2708 * the current lowest, this one becomes the lowest */
2709 if ( cur_min_x == NULL
2710 || strLT(x + COLLXFRM_HDR_LEN,
2711 cur_min_x + COLLXFRM_HDR_LEN))
2713 PL_strxfrm_NUL_replacement = j;
2719 } /* end of loop through all 255 characters */
2721 /* Stop looking if found */
2726 /* Unlikely, but possible, if there aren't any controls that
2727 * work in the locale, repeat the loop, looking for any
2728 * character that works */
2729 DEBUG_L(PerlIO_printf(Perl_debug_log,
2730 "_mem_collxfrm: No control worked. Trying non-controls\n"));
2731 } /* End of loop to try first the controls, then any char */
2734 DEBUG_L(PerlIO_printf(Perl_debug_log,
2735 "_mem_collxfrm: Couldn't find any character to replace"
2736 " embedded NULs in locale %s with", PL_collation_name));
2740 DEBUG_L(PerlIO_printf(Perl_debug_log,
2741 "_mem_collxfrm: Replacing embedded NULs in locale %s with "
2742 "0x%02X\n", PL_collation_name, PL_strxfrm_NUL_replacement));
2744 Safefree(cur_min_x);
2745 } /* End of determining the character that is to replace NULs */
2747 /* If the replacement is variant under UTF-8, it must match the
2748 * UTF8-ness of the original */
2749 if ( ! UVCHR_IS_INVARIANT(PL_strxfrm_NUL_replacement) && utf8) {
2750 this_replacement_char[0] =
2751 UTF8_EIGHT_BIT_HI(PL_strxfrm_NUL_replacement);
2752 this_replacement_char[1] =
2753 UTF8_EIGHT_BIT_LO(PL_strxfrm_NUL_replacement);
2754 this_replacement_len = 2;
2757 this_replacement_char[0] = PL_strxfrm_NUL_replacement;
2758 /* this_replacement_char[1] = '\0' was done at initialization */
2759 this_replacement_len = 1;
2762 /* The worst case length for the replaced string would be if every
2763 * character in it is NUL. Multiply that by the length of each
2764 * replacement, and allow for a trailing NUL */
2765 sans_nuls_len = (len * this_replacement_len) + 1;
2766 Newx(sans_nuls, sans_nuls_len, char);
2769 /* Replace each NUL with the lowest collating control. Loop until have
2770 * exhausted all the NULs */
2771 while (s + s_strlen < e) {
2772 my_strlcat(sans_nuls, s, sans_nuls_len);
2774 /* Do the actual replacement */
2775 my_strlcat(sans_nuls, this_replacement_char, sans_nuls_len);
2777 /* Move past the input NUL */
2779 s_strlen = strlen(s);
2782 /* And add anything that trails the final NUL */
2783 my_strlcat(sans_nuls, s, sans_nuls_len);
2785 /* Switch so below we transform this modified string */
2788 } /* End of replacing NULs */
2790 /* Make sure the UTF8ness of the string and locale match */
2791 if (utf8 != PL_in_utf8_COLLATE_locale) {
2792 /* XXX convert above Unicode to 10FFFF? */
2793 const char * const t = s; /* Temporary so we can later find where the
2796 /* Here they don't match. Change the string's to be what the locale is
2799 if (! utf8) { /* locale is UTF-8, but input isn't; upgrade the input */
2800 s = (char *) bytes_to_utf8((const U8 *) s, &len);
2803 else { /* locale is not UTF-8; but input is; downgrade the input */
2805 s = (char *) bytes_from_utf8((const U8 *) s, &len, &utf8);
2807 /* If the downgrade was successful we are done, but if the input
2808 * contains things that require UTF-8 to represent, have to do
2809 * damage control ... */
2810 if (UNLIKELY(utf8)) {
2812 /* What we do is construct a non-UTF-8 string with
2813 * 1) the characters representable by a single byte converted
2814 * to be so (if necessary);
2815 * 2) and the rest converted to collate the same as the
2816 * highest collating representable character. That makes
2817 * them collate at the end. This is similar to how we
2818 * handle embedded NULs, but we use the highest collating
2819 * code point instead of the smallest. Like the NUL case,
2820 * this isn't perfect, but is the best we can reasonably
2821 * do. Every above-255 code point will sort the same as
2822 * the highest-sorting 0-255 code point. If that code
2823 * point can combine in a sequence with some other code
2824 * points for weight calculations, us changing something to
2825 * be it can adversely affect the results. But in most
2826 * cases, it should work reasonably. And note that this is
2827 * really an illegal situation: using code points above 255
2828 * on a locale where only 0-255 are valid. If two strings
2829 * sort entirely equal, then the sort order for the
2830 * above-255 code points will be in code point order. */
2834 /* If we haven't calculated the code point with the maximum
2835 * collating order for this locale, do so now */
2836 if (! PL_strxfrm_max_cp) {
2839 /* The current transformed string that collates the
2840 * highest (except it also includes the prefixed collation
2842 char * cur_max_x = NULL;
2844 /* Look through all legal code points (NUL isn't) */
2845 for (j = 1; j < 256; j++) {
2848 char cur_source[] = { '\0', '\0' };
2850 /* Create a 1-char string of the current code point */
2851 cur_source[0] = (char) j;
2853 /* Then transform it */
2854 x = _mem_collxfrm(cur_source, 1, &x_len, FALSE);
2856 /* If something went wrong (which it shouldn't), just
2857 * ignore this code point */
2862 /* If this character's transformation is higher than
2863 * the current highest, this one becomes the highest */
2864 if ( cur_max_x == NULL
2865 || strGT(x + COLLXFRM_HDR_LEN,
2866 cur_max_x + COLLXFRM_HDR_LEN))
2868 PL_strxfrm_max_cp = j;
2877 DEBUG_L(PerlIO_printf(Perl_debug_log,
2878 "_mem_collxfrm: Couldn't find any character to"
2879 " replace above-Latin1 chars in locale %s with",
2880 PL_collation_name));
2884 DEBUG_L(PerlIO_printf(Perl_debug_log,
2885 "_mem_collxfrm: highest 1-byte collating character"
2886 " in locale %s is 0x%02X\n",
2888 PL_strxfrm_max_cp));
2890 Safefree(cur_max_x);
2893 /* Here we know which legal code point collates the highest.
2894 * We are ready to construct the non-UTF-8 string. The length
2895 * will be at least 1 byte smaller than the input string
2896 * (because we changed at least one 2-byte character into a
2897 * single byte), but that is eaten up by the trailing NUL */
2903 char * e = (char *) t + len;
2905 for (i = 0; i < len; i+= UTF8SKIP(t + i)) {
2907 if (UTF8_IS_INVARIANT(cur_char)) {
2910 else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(t + i, e)) {
2911 s[d++] = EIGHT_BIT_UTF8_TO_NATIVE(cur_char, t[i+1]);
2913 else { /* Replace illegal cp with highest collating
2915 s[d++] = PL_strxfrm_max_cp;
2919 Renew(s, d, char); /* Free up unused space */
2924 /* Here, we have constructed a modified version of the input. It could
2925 * be that we already had a modified copy before we did this version.
2926 * If so, that copy is no longer needed */
2927 if (t != input_string) {
2932 length_in_chars = (utf8)
2933 ? utf8_length((U8 *) s, (U8 *) s + len)
2936 /* The first element in the output is the collation id, used by
2937 * sv_collxfrm(); then comes the space for the transformed string. The
2938 * equation should give us a good estimate as to how much is needed */
2939 xAlloc = COLLXFRM_HDR_LEN
2941 + (PL_collxfrm_mult * length_in_chars);
2942 Newx(xbuf, xAlloc, char);
2943 if (UNLIKELY(! xbuf)) {
2944 DEBUG_L(PerlIO_printf(Perl_debug_log,
2945 "_mem_collxfrm: Couldn't malloc %zu bytes\n", xAlloc));
2949 /* Store the collation id */
2950 *(U32*)xbuf = PL_collation_ix;
2952 /* Then the transformation of the input. We loop until successful, or we
2956 *xlen = strxfrm(xbuf + COLLXFRM_HDR_LEN, s, xAlloc - COLLXFRM_HDR_LEN);
2958 /* If the transformed string occupies less space than we told strxfrm()
2959 * was available, it means it successfully transformed the whole
2961 if (*xlen < xAlloc - COLLXFRM_HDR_LEN) {
2963 /* Some systems include a trailing NUL in the returned length.
2964 * Ignore it, using a loop in case multiple trailing NULs are
2967 && *(xbuf + COLLXFRM_HDR_LEN + (*xlen) - 1) == '\0')
2972 /* If the first try didn't get it, it means our prediction was low.
2973 * Modify the coefficients so that we predict a larger value in any
2974 * future transformations */
2976 STRLEN needed = *xlen + 1; /* +1 For trailing NUL */
2977 STRLEN computed_guess = PL_collxfrm_base
2978 + (PL_collxfrm_mult * length_in_chars);
2980 /* On zero-length input, just keep current slope instead of
2982 const STRLEN new_m = (length_in_chars != 0)
2983 ? needed / length_in_chars
2986 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2987 "%s: %d: initial size of %zu bytes for a length "
2988 "%zu string was insufficient, %zu needed\n",
2990 computed_guess, length_in_chars, needed));
2992 /* If slope increased, use it, but discard this result for
2993 * length 1 strings, as we can't be sure that it's a real slope
2995 if (length_in_chars > 1 && new_m > PL_collxfrm_mult) {
2999 STRLEN old_m = PL_collxfrm_mult;
3000 STRLEN old_b = PL_collxfrm_base;
3004 PL_collxfrm_mult = new_m;
3005 PL_collxfrm_base = 1; /* +1 For trailing NUL */
3006 computed_guess = PL_collxfrm_base
3007 + (PL_collxfrm_mult * length_in_chars);
3008 if (computed_guess < needed) {
3009 PL_collxfrm_base += needed - computed_guess;
3012 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3013 "%s: %d: slope is now %zu; was %zu, base "
3014 "is now %zu; was %zu\n",
3016 PL_collxfrm_mult, old_m,
3017 PL_collxfrm_base, old_b));
3019 else { /* Slope didn't change, but 'b' did */
3020 const STRLEN new_b = needed
3023 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3024 "%s: %d: base is now %zu; was %zu\n",
3026 new_b, PL_collxfrm_base));
3027 PL_collxfrm_base = new_b;
3034 if (UNLIKELY(*xlen >= PERL_INT_MAX)) {
3035 DEBUG_L(PerlIO_printf(Perl_debug_log,
3036 "_mem_collxfrm: Needed %zu bytes, max permissible is %u\n",
3037 *xlen, PERL_INT_MAX));
3041 /* A well-behaved strxfrm() returns exactly how much space it needs
3042 * (usually not including the trailing NUL) when it fails due to not
3043 * enough space being provided. Assume that this is the case unless
3044 * it's been proven otherwise */
3045 if (LIKELY(PL_strxfrm_is_behaved) && first_time) {
3046 xAlloc = *xlen + COLLXFRM_HDR_LEN + 1;
3048 else { /* Here, either:
3049 * 1) The strxfrm() has previously shown bad behavior; or
3050 * 2) It isn't the first time through the loop, which means
3051 * that the strxfrm() is now showing bad behavior, because
3052 * we gave it what it said was needed in the previous
3053 * iteration, and it came back saying it needed still more.
3054 * (Many versions of cygwin fit this. When the buffer size
3055 * isn't sufficient, they return the input size instead of
3056 * how much is needed.)
3057 * Increase the buffer size by a fixed percentage and try again.
3059 xAlloc += (xAlloc / 4) + 1;
3060 PL_strxfrm_is_behaved = FALSE;
3064 if (DEBUG_Lv_TEST || debug_initialization) {
3065 PerlIO_printf(Perl_debug_log,
3066 "_mem_collxfrm required more space than previously calculated"
3067 " for locale %s, trying again with new guess=%d+%zu\n",
3068 PL_collation_name, (int) COLLXFRM_HDR_LEN,
3069 xAlloc - COLLXFRM_HDR_LEN);
3076 Renew(xbuf, xAlloc, char);
3077 if (UNLIKELY(! xbuf)) {
3078 DEBUG_L(PerlIO_printf(Perl_debug_log,
3079 "_mem_collxfrm: Couldn't realloc %zu bytes\n", xAlloc));
3089 if (DEBUG_Lv_TEST || debug_initialization) {
3091 print_collxfrm_input_and_return(s, s + len, xlen, utf8);
3092 PerlIO_printf(Perl_debug_log, "Its xfrm is:");
3093 PerlIO_printf(Perl_debug_log, "%s\n",
3094 _byte_dump_string((U8 *) xbuf + COLLXFRM_HDR_LEN,
3100 /* Free up unneeded space; retain ehough for trailing NUL */
3101 Renew(xbuf, COLLXFRM_HDR_LEN + *xlen + 1, char);
3103 if (s != input_string) {
3111 if (s != input_string) {
3118 if (DEBUG_Lv_TEST || debug_initialization) {
3119 print_collxfrm_input_and_return(s, s + len, NULL, utf8);
3130 S_print_collxfrm_input_and_return(pTHX_
3131 const char * const s,
3132 const char * const e,
3133 const STRLEN * const xlen,
3137 PERL_ARGS_ASSERT_PRINT_COLLXFRM_INPUT_AND_RETURN;
3139 PerlIO_printf(Perl_debug_log, "_mem_collxfrm[%" UVuf "]: returning ",
3140 (UV)PL_collation_ix);
3142 PerlIO_printf(Perl_debug_log, "%zu", *xlen);
3145 PerlIO_printf(Perl_debug_log, "NULL");
3147 PerlIO_printf(Perl_debug_log, " for locale '%s', string='",
3149 print_bytes_for_locale(s, e, is_utf8);
3151 PerlIO_printf(Perl_debug_log, "'\n");
3155 S_print_bytes_for_locale(pTHX_
3156 const char * const s,
3157 const char * const e,
3161 bool prev_was_printable = TRUE;
3162 bool first_time = TRUE;
3164 PERL_ARGS_ASSERT_PRINT_BYTES_FOR_LOCALE;
3168 ? utf8_to_uvchr_buf((U8 *) t, e, NULL)
3171 if (! prev_was_printable) {
3172 PerlIO_printf(Perl_debug_log, " ");
3174 PerlIO_printf(Perl_debug_log, "%c", (U8) cp);
3175 prev_was_printable = TRUE;
3179 PerlIO_printf(Perl_debug_log, " ");
3181 PerlIO_printf(Perl_debug_log, "%02" UVXf, cp);
3182 prev_was_printable = FALSE;
3184 t += (is_utf8) ? UTF8SKIP(t) : 1;
3189 # endif /* #ifdef DEBUGGING */
3190 #endif /* USE_LOCALE_COLLATE */
3195 S_switch_category_locale_to_template(pTHX_ const int switch_category, const int template_category, const char * template_locale)
3197 /* Changes the locale for LC_'switch_category" to that of
3198 * LC_'template_category', if they aren't already the same. If not NULL,
3199 * 'template_locale' is the locale that 'template_category' is in.
3201 * Returns a copy of the name of the original locale for 'switch_category'
3202 * so can be switched back to with the companion function
3203 * restore_switched_locale(), (NULL if no restoral is necessary.) */
3205 char * restore_to_locale = NULL;
3207 if (switch_category == template_category) { /* No changes needed */
3211 /* Find the original locale of the category we may need to change, so that
3212 * it can be restored to later */
3213 restore_to_locale = stdize_locale(savepv(do_setlocale_r(switch_category,
3215 if (! restore_to_locale) {
3217 "panic: %s: %d: Could not find current %s locale, errno=%d\n",
3218 __FILE__, __LINE__, category_name(switch_category), errno);
3221 /* If the locale of the template category wasn't passed in, find it now */
3222 if (template_locale == NULL) {
3223 template_locale = do_setlocale_r(template_category, NULL);
3224 if (! template_locale) {
3226 "panic: %s: %d: Could not find current %s locale, errno=%d\n",
3227 __FILE__, __LINE__, category_name(template_category), errno);
3231 /* It the locales are the same, there's nothing to do */
3232 if (strEQ(restore_to_locale, template_locale)) {
3233 Safefree(restore_to_locale);
3235 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s locale unchanged as %s\n",
3236 category_name(switch_category), restore_to_locale));
3241 /* Finally, change the locale to the template one */
3242 if (! do_setlocale_r(switch_category, template_locale)) {
3244 "panic: %s: %d: Could not change %s locale to %s, errno=%d\n",
3245 __FILE__, __LINE__, category_name(switch_category),
3246 template_locale, errno);
3249 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s locale switched to %s\n",
3250 category_name(switch_category), template_locale));
3252 return restore_to_locale;
3256 S_restore_switched_locale(pTHX_ const int category, const char * const original_locale)
3258 /* Restores the locale for LC_'category' to 'original_locale' (which is a
3259 * copy that will be freed by this function), or do nothing if the latter
3260 * parameter is NULL */
3262 if (original_locale == NULL) {
3266 if (! do_setlocale_r(category, original_locale)) {
3268 "panic: %s: %d: setlocale %s restore to %s failed, errno=%d\n",
3270 category_name(category), original_locale, errno);
3273 Safefree(original_locale);
3277 Perl__is_cur_LC_category_utf8(pTHX_ int category)
3279 /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE
3280 * otherwise. 'category' may not be LC_ALL. If the platform doesn't have
3281 * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence
3282 * could give the wrong result. The result will very likely be correct for
3283 * languages that have commonly used non-ASCII characters, but for notably
3284 * English, it comes down to if the locale's name ends in something like
3285 * "UTF-8". It errs on the side of not being a UTF-8 locale.
3287 * If the platform is early C89, not containing mbtowc(), or we are
3288 * compiled to not pay attention to LC_CTYPE, this employs heuristics.
3289 * These work very well for non-Latin locales or those whose currency
3290 * symbol isn't a '$' nor plain ASCII text. But without LC_CTYPE and at
3291 * least MB_CUR_MAX, English locales with an ASCII currency symbol depend
3292 * on the name containing UTF-8 or not. */
3294 /* Name of current locale corresponding to the input category */
3295 const char *save_input_locale = NULL;
3297 bool is_utf8 = FALSE; /* The return value */
3299 /* The variables below are for the cache of previous lookups using this
3300 * function. The cache is a C string, described at the definition for
3301 * 'C_and_POSIX_utf8ness'.
3303 * The first part of the cache is fixed, for the C and POSIX locales. The
3304 * varying part starts just after them. */
3305 char * utf8ness_cache = PL_locale_utf8ness + STRLENs(C_and_POSIX_utf8ness);
3307 Size_t utf8ness_cache_size; /* Size of the varying portion */
3308 Size_t input_name_len; /* Length in bytes of save_input_locale */
3309 Size_t input_name_len_with_overhead; /* plus extra chars used to store
3310 the name in the cache */
3311 char * delimited; /* The name plus the delimiters used to store
3313 char * name_pos; /* position of 'delimited' in the cache, or 0
3319 assert(category != LC_ALL);
3323 /* Get the desired category's locale */
3324 save_input_locale = stdize_locale(savepv(do_setlocale_r(category, NULL)));
3325 if (! save_input_locale) {
3327 "panic: %s: %d: Could not find current %s locale, errno=%d\n",
3328 __FILE__, __LINE__, category_name(category), errno);
3331 DEBUG_L(PerlIO_printf(Perl_debug_log,
3332 "Current locale for %s is %s\n",
3333 category_name(category), save_input_locale));
3335 input_name_len = strlen(save_input_locale);
3337 /* In our cache, each name is accompanied by two delimiters and a single
3339 input_name_len_with_overhead = input_name_len + 3;
3341 /* Allocate and populate space for a copy of the name surrounded by the
3343 Newx(delimited, input_name_len_with_overhead, char);
3344 delimited[0] = UTF8NESS_SEP[0];
3345 Copy(save_input_locale, delimited + 1, input_name_len, char);
3346 delimited[input_name_len+1] = UTF8NESS_PREFIX[0];
3347 delimited[input_name_len+2] = '\0';
3349 /* And see if that is in the cache */
3350 name_pos = instr(PL_locale_utf8ness, delimited);
3352 is_utf8 = *(name_pos + input_name_len_with_overhead - 1) - '0';
3356 if (DEBUG_Lv_TEST || debug_initialization) {
3357 PerlIO_printf(Perl_debug_log, "UTF8ness for locale %s=%d, \n",
3358 save_input_locale, is_utf8);
3363 /* And, if not already in that position, move it to the beginning of
3364 * the non-constant portion of the list, since it is the most recently
3365 * used. (We don't have to worry about overflow, since just moving
3366 * existing names around) */
3367 if (name_pos > utf8ness_cache) {
3368 Move(utf8ness_cache,
3369 utf8ness_cache + input_name_len_with_overhead,
3370 name_pos - utf8ness_cache, char);
3373 input_name_len_with_overhead - 1, char);
3374 utf8ness_cache[input_name_len_with_overhead - 1] = is_utf8 + '0';
3377 Safefree(delimited);
3378 Safefree(save_input_locale);
3382 /* Here we don't have stored the utf8ness for the input locale. We have to
3385 # if defined(USE_LOCALE_CTYPE) \
3386 && ( (defined(HAS_NL_LANGINFO) && defined(CODESET)) \
3387 || (defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)))
3390 const char *original_ctype_locale
3391 = switch_category_locale_to_template(LC_CTYPE,
3395 /* Here the current LC_CTYPE is set to the locale of the category whose
3396 * information is desired. This means that nl_langinfo() and mbtowc()
3397 * should give the correct results */
3399 # ifdef MB_CUR_MAX /* But we can potentially rule out UTF-8ness, avoiding
3400 calling the functions if we have this */
3402 /* Standard UTF-8 needs at least 4 bytes to represent the maximum
3403 * Unicode code point. */
3405 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s: %d: MB_CUR_MAX=%d\n",
3406 __FILE__, __LINE__, (int) MB_CUR_MAX));
3407 if ((unsigned) MB_CUR_MAX < STRLENs(MAX_UNICODE_UTF8)) {
3409 restore_switched_locale(LC_CTYPE, original_ctype_locale);
3410 goto finish_and_return;
3414 # if defined(HAS_NL_LANGINFO) && defined(CODESET)
3416 { /* The task is easiest if the platform has this POSIX 2001 function.
3417 Except on some platforms it can wrongly return "", so have to have
3418 a fallback. And it can return that it's UTF-8, even if there are
3419 variances from that. For example, Turkish locales may use the
3420 alternate dotted I rules, and sometimes it appears to be a
3421 defective locale definition. XXX We should probably check for
3422 these in the Latin1 range and warn (but on glibc, requires
3423 iswalnum() etc. due to their not handling 80-FF correctly */
3424 const char *codeset = my_nl_langinfo(PERL_CODESET, FALSE);
3425 /* FALSE => already in dest locale */
3427 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3428 "\tnllanginfo returned CODESET '%s'\n", codeset));
3430 if (codeset && strNE(codeset, "")) {
3432 /* If the implementation of foldEQ() somehow were
3433 * to change to not go byte-by-byte, this could
3434 * read past end of string, as only one length is
3435 * checked. But currently, a premature NUL will
3436 * compare false, and it will stop there */
3437 is_utf8 = cBOOL( foldEQ(codeset, STR_WITH_LEN("UTF-8"))
3438 || foldEQ(codeset, STR_WITH_LEN("UTF8")));
3440 DEBUG_L(PerlIO_printf(Perl_debug_log,
3441 "\tnllanginfo returned CODESET '%s'; ?UTF8 locale=%d\n",
3443 restore_switched_locale(LC_CTYPE, original_ctype_locale);
3444 goto finish_and_return;
3449 # if defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)
3450 /* We can see if this is a UTF-8-like locale if have mbtowc(). It was a
3451 * late adder to C89, so very likely to have it. However, testing has
3452 * shown that, like nl_langinfo() above, there are locales that are not
3453 * strictly UTF-8 that this will return that they are */
3460 # if defined(HAS_MBRTOWC) && defined(USE_ITHREADS)
3466 /* mbrtowc() and mbtowc() convert a byte string to a wide
3467 * character. Feed a byte string to one of them and check that the
3468 * result is the expected Unicode code point */
3470 # if defined(HAS_MBRTOWC) && defined(USE_ITHREADS)
3471 /* Prefer this function if available, as it's reentrant */
3473 memset(&ps, 0, sizeof(ps));;
3474 PERL_UNUSED_RESULT(mbrtowc(&wc, NULL, 0, &ps)); /* Reset any shift
3477 len = mbrtowc(&wc, STR_WITH_LEN(REPLACEMENT_CHARACTER_UTF8), &ps);
3483 PERL_UNUSED_RESULT(mbtowc(&wc, NULL, 0));/* Reset any shift state */
3485 len = mbtowc(&wc, STR_WITH_LEN(REPLACEMENT_CHARACTER_UTF8));
3492 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3493 "\treturn from mbtowc; len=%d; code_point=%x; errno=%d\n",
3494 len, (unsigned int) wc, GET_ERRNO));
3496 is_utf8 = cBOOL( len == STRLENs(REPLACEMENT_CHARACTER_UTF8)
3497 && wc == (wchar_t) UNICODE_REPLACEMENT);
3500 restore_switched_locale(LC_CTYPE, original_ctype_locale);
3501 goto finish_and_return;
3507 /* Here, we must have a C89 compiler that doesn't have mbtowc(). Next
3508 * try looking at the currency symbol to see if it disambiguates
3509 * things. Often that will be in the native script, and if the symbol
3510 * isn't in UTF-8, we know that the locale isn't. If it is non-ASCII
3511 * UTF-8, we infer that the locale is too, as the odds of a non-UTF8
3512 * string being valid UTF-8 are quite small */
3514 # ifdef USE_LOCALE_MONETARY
3516 /* If have LC_MONETARY, we can look at the currency symbol. Often that
3517 * will be in the native script. We do this one first because there is
3518 * just one string to examine, so potentially avoids work */
3521 const char *original_monetary_locale
3522 = switch_category_locale_to_template(LC_MONETARY,
3525 bool only_ascii = FALSE;
3526 const U8 * currency_string
3527 = (const U8 *) my_nl_langinfo(PERL_CRNCYSTR, FALSE);
3528 /* 2nd param not relevant for this item */
3529 const U8 * first_variant;
3531 assert( *currency_string == '-'
3532 || *currency_string == '+'
3533 || *currency_string == '.');
3537 if (is_utf8_invariant_string_loc(currency_string, 0, &first_variant))
3539 DEBUG_L(PerlIO_printf(Perl_debug_log, "Couldn't get currency symbol for %s, or contains only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
3543 is_utf8 = is_strict_utf8_string(first_variant, 0);
3546 restore_switched_locale(LC_MONETARY, original_monetary_locale);
3550 /* It isn't a UTF-8 locale if the symbol is not legal UTF-8;
3551 * otherwise assume the locale is UTF-8 if and only if the symbol
3552 * is non-ascii UTF-8. */
3553 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "\t?Currency symbol for %s is UTF-8=%d\n",
3554 save_input_locale, is_utf8));
3555 goto finish_and_return;
3559 # endif /* USE_LOCALE_MONETARY */
3560 # if defined(HAS_STRFTIME) && defined(USE_LOCALE_TIME)
3562 /* Still haven't found a non-ASCII string to disambiguate UTF-8 or not. Try
3563 * the names of the months and weekdays, timezone, and am/pm indicator */
3565 const char *original_time_locale
3566 = switch_category_locale_to_template(LC_TIME,
3570 bool is_dst = FALSE;
3574 char * formatted_time;
3576 /* Here the current LC_TIME is set to the locale of the category
3577 * whose information is desired. Look at all the days of the week and
3578 * month names, and the timezone and am/pm indicator for UTF-8 variant
3579 * characters. The first such a one found will tell us if the locale
3580 * is UTF-8 or not */
3582 for (i = 0; i < 7 + 12; i++) { /* 7 days; 12 months */
3583 formatted_time = my_strftime("%A %B %Z %p",
3584 0, 0, hour, dom, month, 2012 - 1900, 0, 0, is_dst);
3585 if ( ! formatted_time
3586 || is_utf8_invariant_string((U8 *) formatted_time, 0))
3589 /* Here, we didn't find a non-ASCII. Try the next time through
3590 * with the complemented dst and am/pm, and try with the next
3591 * weekday. After we have gotten all weekdays, try the next
3594 hour = (hour + 12) % 24;
3602 /* Here, we have a non-ASCII. Return TRUE is it is valid UTF8;
3603 * false otherwise. But first, restore LC_TIME to its original
3604 * locale if we changed it */
3605 restore_switched_locale(LC_TIME, original_time_locale);
3607 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "\t?time-related strings for %s are UTF-8=%d\n",
3609 is_utf8_string((U8 *) formatted_time, 0)));
3610 is_utf8 = is_utf8_string((U8 *) formatted_time, 0);
3611 goto finish_and_return;
3614 /* Falling off the end of the loop indicates all the names were just
3615 * ASCII. Go on to the next test. If we changed it, restore LC_TIME
3616 * to its original locale */
3617 restore_switched_locale(LC_TIME, original_time_locale);
3618 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "All time-related words for %s contain only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
3623 # if 0 && defined(USE_LOCALE_MESSAGES) && defined(HAS_SYS_ERRLIST)
3625 /* This code is ifdefd out because it was found to not be necessary in testing
3626 * on our dromedary test machine, which has over 700 locales. There, this
3627 * added no value to looking at the currency symbol and the time strings. I
3628 * left it in so as to avoid rewriting it if real-world experience indicates
3629 * that dromedary is an outlier. Essentially, instead of returning abpve if we
3630 * haven't found illegal utf8, we continue on and examine all the strerror()
3631 * messages on the platform for utf8ness. If all are ASCII, we still don't
3632 * know the answer; but otherwise we have a pretty good indication of the
3633 * utf8ness. The reason this doesn't help much is that the messages may not
3634 * have been translated into the locale. The currency symbol and time strings
3635 * are much more likely to have been translated. */
3638 bool non_ascii = FALSE;
3639 const char *original_messages_locale
3640 = switch_category_locale_to_template(LC_MESSAGES,
3643 const char * errmsg = NULL;
3645 /* Here the current LC_MESSAGES is set to the locale of the category
3646 * whose information is desired. Look through all the messages. We
3647 * can't use Strerror() here because it may expand to code that
3648 * segfaults in miniperl */
3650 for (e = 0; e <= sys_nerr; e++) {
3652 errmsg = sys_errlist[e];
3653 if (errno || !errmsg) {
3656 errmsg = savepv(errmsg);
3657 if (! is_utf8_invariant_string((U8 *) errmsg, 0)) {
3659 is_utf8 = is_utf8_string((U8 *) errmsg, 0);
3665 restore_switched_locale(LC_MESSAGES, original_messages_locale);
3669 /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid,
3670 * any non-ascii means it is one; otherwise we assume it isn't */
3671 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "\t?error messages for %s are UTF-8=%d\n",
3674 goto finish_and_return;
3677 DEBUG_L(PerlIO_printf(Perl_debug_log, "All error messages for %s contain only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale));
3681 # ifndef EBCDIC /* On os390, even if the name ends with "UTF-8', it isn't a
3684 /* As a last resort, look at the locale name to see if it matches
3685 * qr/UTF -? * 8 /ix, or some other common locale names. This "name", the
3686 * return of setlocale(), is actually defined to be opaque, so we can't
3687 * really rely on the absence of various substrings in the name to indicate
3688 * its UTF-8ness, but if it has UTF8 in the name, it is extremely likely to
3689 * be a UTF-8 locale. Similarly for the other common names */
3692 const Size_t final_pos = strlen(save_input_locale) - 1;
3694 if (final_pos >= 3) {
3695 const char *name = save_input_locale;
3697 /* Find next 'U' or 'u' and look from there */
3698 while ((name += strcspn(name, "Uu") + 1)
3699 <= save_input_locale + final_pos - 2)
3701 if ( isALPHA_FOLD_NE(*name, 't')
3702 || isALPHA_FOLD_NE(*(name + 1), 'f'))
3707 if (*(name) == '-') {
3708 if ((name > save_input_locale + final_pos - 1)) {
3713 if (*(name) == '8') {
3714 DEBUG_L(PerlIO_printf(Perl_debug_log,
3715 "Locale %s ends with UTF-8 in name\n",
3716 save_input_locale));
3718 goto finish_and_return;
3721 DEBUG_L(PerlIO_printf(Perl_debug_log,
3722 "Locale %s doesn't end with UTF-8 in name\n",
3723 save_input_locale));
3728 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
3729 if (memENDs(save_input_locale, final_pos, "65001")) {
3730 DEBUG_L(PerlIO_printf(Perl_debug_log,
3731 "Locale %s ends with 65001 in name, is UTF-8 locale\n",
3732 save_input_locale));
3734 goto finish_and_return;
3741 /* Other common encodings are the ISO 8859 series, which aren't UTF-8. But
3742 * since we are about to return FALSE anyway, there is no point in doing
3743 * this extra work */
3746 if (instr(save_input_locale, "8859")) {
3747 DEBUG_L(PerlIO_printf(Perl_debug_log,
3748 "Locale %s has 8859 in name, not UTF-8 locale\n",
3749 save_input_locale));
3751 goto finish_and_return;
3755 DEBUG_L(PerlIO_printf(Perl_debug_log,
3756 "Assuming locale %s is not a UTF-8 locale\n",
3757 save_input_locale));
3760 # endif /* the code that is compiled when no modern LC_CTYPE */
3764 /* Cache this result so we don't have to go through all this next time. */
3765 utf8ness_cache_size = sizeof(PL_locale_utf8ness)
3766 - (utf8ness_cache - PL_locale_utf8ness);
3768 /* But we can't save it if it is too large for the total space available */
3769 if (LIKELY(input_name_len_with_overhead < utf8ness_cache_size)) {
3770 Size_t utf8ness_cache_len = strlen(utf8ness_cache);
3772 /* Here it can fit, but we may need to clear out the oldest cached
3773 * result(s) to do so. Check */
3774 if (utf8ness_cache_len + input_name_len_with_overhead
3775 >= utf8ness_cache_size)
3777 /* Here we have to clear something out to make room for this.
3778 * Start looking at the rightmost place where it could fit and find
3779 * the beginning of the entry that extends past that. */
3780 char * cutoff = (char *) my_memrchr(utf8ness_cache,
3783 - input_name_len_with_overhead);
3786 assert(cutoff >= utf8ness_cache);
3788 /* This and all subsequent entries must be removed */
3790 utf8ness_cache_len = strlen(utf8ness_cache);
3793 /* Make space for the new entry */
3794 Move(utf8ness_cache,
3795 utf8ness_cache + input_name_len_with_overhead,
3796 utf8ness_cache_len + 1 /* Incl. trailing NUL */, char);
3799 Copy(delimited, utf8ness_cache, input_name_len_with_overhead - 1, char);
3800 utf8ness_cache[input_name_len_with_overhead - 1] = is_utf8 + '0';
3802 if ((PL_locale_utf8ness[strlen(PL_locale_utf8ness)-1]
3803 & (PERL_UINTMAX_T) ~1) != '0')
3806 "panic: %s: %d: Corrupt utf8ness_cache=%s\nlen=%zu,"
3807 " inserted_name=%s, its_len=%zu\n",
3809 PL_locale_utf8ness, strlen(PL_locale_utf8ness),
3810 delimited, input_name_len_with_overhead);
3816 if (DEBUG_Lv_TEST) {
3817 const char * s = PL_locale_utf8ness;
3819 /* Audit the structure */
3820 while (s < PL_locale_utf8ness + strlen(PL_locale_utf8ness)) {
3823 if (*s != UTF8NESS_SEP[0]) {
3825 "panic: %s: %d: Corrupt utf8ness_cache: missing"
3826 " separator %.*s<-- HERE %s\n",
3828 (int) (s - PL_locale_utf8ness), PL_locale_utf8ness,
3832 e = strchr(s, UTF8NESS_PREFIX[0]);
3835 "panic: %s: %d: Corrupt utf8ness_cache: missing"
3836 " separator %.*s<-- HERE %s\n",
3838 (int) (e - PL_locale_utf8ness), PL_locale_utf8ness,
3842 if (*e != '0' && *e != '1') {
3844 "panic: %s: %d: Corrupt utf8ness_cache: utf8ness"
3845 " must be [01] %.*s<-- HERE %s\n",
3847 (int) (e + 1 - PL_locale_utf8ness),
3848 PL_locale_utf8ness, e + 1);
3850 if (ninstr(PL_locale_utf8ness, s, s-1, e)) {
3852 "panic: %s: %d: Corrupt utf8ness_cache: entry"
3853 " has duplicate %.*s<-- HERE %s\n",
3855 (int) (e - PL_locale_utf8ness), PL_locale_utf8ness,
3862 if (DEBUG_Lv_TEST || debug_initialization) {
3864 PerlIO_printf(Perl_debug_log,
3865 "PL_locale_utf8ness is now %s; returning %d\n",
3866 PL_locale_utf8ness, is_utf8);
3871 Safefree(delimited);
3872 Safefree(save_input_locale);
3879 Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
3882 /* Internal function which returns if we are in the scope of a pragma that
3883 * enables the locale category 'category'. 'compiling' should indicate if
3884 * this is during the compilation phase (TRUE) or not (FALSE). */
3886 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
3888 SV *categories = cop_hints_fetch_pvs(cop, "locale", 0);
3889 if (! categories || categories == &PL_sv_placeholder) {
3893 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
3894 * a valid unsigned */
3895 assert(category >= -1);
3896 return cBOOL(SvUV(categories) & (1U << (category + 1)));
3900 Perl_my_strerror(pTHX_ const int errnum)
3902 /* Returns a mortalized copy of the text of the error message associated
3903 * with 'errnum'. It uses the current locale's text unless the platform
3904 * doesn't have the LC_MESSAGES category or we are not being called from
3905 * within the scope of 'use locale'. In the former case, it uses whatever
3906 * strerror returns; in the latter case it uses the text from the C locale.
3908 * The function just calls strerror(), but temporarily switches, if needed,
3909 * to the C locale */
3914 #ifndef USE_LOCALE_MESSAGES
3916 /* If platform doesn't have messages category, we don't do any switching to
3917 * the C locale; we just use whatever strerror() returns */
3919 errstr = savepv(Strerror(errnum));
3921 #else /* Has locale messages */
3923 const bool within_locale_scope = IN_LC(LC_MESSAGES);
3925 # if defined(HAS_POSIX_2008_LOCALE) && defined(HAS_STRERROR_L)
3927 /* This function is trivial if we don't have to worry about thread safety
3928 * and have strerror_l(), as it handles the switch of locales so we don't
3929 * have to deal with that. We don't have to worry about thread safety if
3930 * this is an unthreaded build, or if strerror_r() is also available. Both
3931 * it and strerror_l() are thread-safe. Plain strerror() isn't thread
3932 * safe. But on threaded builds when strerror_r() is available, the
3933 * apparent call to strerror() below is actually a macro that
3934 * behind-the-scenes calls strerror_r().
3937 # if ! defined(USE_ITHREADS) || defined(HAS_STRERROR_R)
3939 if (within_locale_scope) {
3940 errstr = savepv(strerror(errnum));
3943 errstr = savepv(strerror_l(errnum, PL_C_locale_obj));
3948 /* Here we have strerror_l(), but not strerror_r() and we are on a
3949 * threaded-build. We use strerror_l() for everything, constructing a
3950 * locale to pass to it if necessary */
3952 bool do_free = FALSE;
3953 locale_t locale_to_use;
3955 if (within_locale_scope) {
3956 locale_to_use = uselocale((locale_t) 0);
3957 if (locale_to_use == LC_GLOBAL_LOCALE) {
3958 locale_to_use = duplocale(LC_GLOBAL_LOCALE);
3962 else { /* Use C locale if not within 'use locale' scope */
3963 locale_to_use = PL_C_locale_obj;
3966 errstr = savepv(strerror_l(errnum, locale_to_use));
3969 freelocale(locale_to_use);
3973 # else /* Doesn't have strerror_l() */
3975 # ifdef USE_POSIX_2008_LOCALE
3977 locale_t save_locale = NULL;
3981 const char * save_locale = NULL;
3982 bool locale_is_C = FALSE;
3984 /* We have a critical section to prevent another thread from executing this
3985 * same code at the same time. (On unthreaded perls, the LOCK is a
3986 * no-op.) Since this is the only place in core that changes LC_MESSAGES
3987 * (unless the user has called setlocale(), this works to prevent races. */
3992 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3993 "my_strerror called with errnum %d\n", errnum));
3994 if (! within_locale_scope) {
3997 # ifdef USE_POSIX_2008_LOCALE /* Use the thread-safe locale functions */
3999 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4000 "Not within locale scope, about to call"
4001 " uselocale(0x%p)\n", PL_C_locale_obj));
4002 save_locale = uselocale(PL_C_locale_obj);
4003 if (! save_locale) {
4004 DEBUG_L(PerlIO_printf(Perl_debug_log,
4005 "uselocale failed, errno=%d\n", errno));
4008 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4009 "uselocale returned 0x%p\n", save_locale));
4012 # else /* Not thread-safe build */
4014 save_locale = do_setlocale_c(LC_MESSAGES, NULL);
4015 if (! save_locale) {
4017 "panic: %s: %d: Could not find current LC_MESSAGES locale,"
4018 " errno=%d\n", __FILE__, __LINE__, errno);
4021 locale_is_C = isNAME_C_OR_POSIX(save_locale);
4023 /* Switch to the C locale if not already in it */
4024 if (! locale_is_C) {
4026 /* The setlocale() just below likely will zap 'save_locale', so
4028 save_locale = savepv(save_locale);
4029 do_setlocale_c(LC_MESSAGES, "C");
4035 } /* end of ! within_locale_scope */
4037 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s: %d: WITHIN locale scope\n",
4038 __FILE__, __LINE__));
4041 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4042 "Any locale change has been done; about to call Strerror\n"));
4043 errstr = savepv(Strerror(errnum));
4045 if (! within_locale_scope) {
4048 # ifdef USE_POSIX_2008_LOCALE
4050 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4051 "%s: %d: not within locale scope, restoring the locale\n",
4052 __FILE__, __LINE__));
4053 if (save_locale && ! uselocale(save_locale)) {
4054 DEBUG_L(PerlIO_printf(Perl_debug_log,
4055 "uselocale restore failed, errno=%d\n", errno));
4061 if (save_locale && ! locale_is_C) {
4062 if (! do_setlocale_c(LC_MESSAGES, save_locale)) {
4064 "panic: %s: %d: setlocale restore failed, errno=%d\n",
4065 __FILE__, __LINE__, errno);
4067 Safefree(save_locale);
4074 # endif /* End of doesn't have strerror_l */
4075 #endif /* End of does have locale messages */
4079 if (DEBUG_Lv_TEST) {
4080 PerlIO_printf(Perl_debug_log, "Strerror returned; saving a copy: '");
4081 print_bytes_for_locale(errstr, errstr + strlen(errstr), 0);
4082 PerlIO_printf(Perl_debug_log, "'\n");
4093 =for apidoc sync_locale
4095 Changing the program's locale should be avoided by XS code. Nevertheless,
4096 certain non-Perl libraries called from XS, such as C<Gtk> do so. When this
4097 happens, Perl needs to be told that the locale has changed. Use this function
4098 to do so, before returning to Perl.
4104 Perl_sync_locale(pTHX)
4108 #ifdef USE_LOCALE_CTYPE
4110 newlocale = do_setlocale_c(LC_CTYPE, NULL);
4111 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4112 "%s:%d: %s\n", __FILE__, __LINE__,
4113 setlocale_debug_string(LC_CTYPE, NULL, newlocale)));
4114 new_ctype(newlocale);
4116 #endif /* USE_LOCALE_CTYPE */
4117 #ifdef USE_LOCALE_COLLATE
4119 newlocale = do_setlocale_c(LC_COLLATE, NULL);
4120 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4121 "%s:%d: %s\n", __FILE__, __LINE__,
4122 setlocale_debug_string(LC_COLLATE, NULL, newlocale)));
4123 new_collate(newlocale);
4126 #ifdef USE_LOCALE_NUMERIC
4128 newlocale = do_setlocale_c(LC_NUMERIC, NULL);
4129 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4130 "%s:%d: %s\n", __FILE__, __LINE__,
4131 setlocale_debug_string(LC_NUMERIC, NULL, newlocale)));
4132 new_numeric(newlocale);
4134 #endif /* USE_LOCALE_NUMERIC */
4138 #if defined(DEBUGGING) && defined(USE_LOCALE)
4141 S_setlocale_debug_string(const int category, /* category number,
4143 const char* const locale, /* locale name */
4145 /* return value from setlocale() when attempting to
4146 * set 'category' to 'locale' */
4147 const char* const retval)
4149 /* Returns a pointer to a NUL-terminated string in static storage with
4150 * added text about the info passed in. This is not thread safe and will
4151 * be overwritten by the next call, so this should be used just to
4152 * formulate a string to immediately print or savepv() on. */
4154 /* initialise to a non-null value to keep it out of BSS and so keep
4155 * -DPERL_GLOBAL_STRUCT_PRIVATE happy */
4156 static char ret[128] = "If you can read this, thank your buggy C"
4157 " library strlcpy(), and change your hints file"
4160 my_strlcpy(ret, "setlocale(", sizeof(ret));
4161 my_strlcat(ret, category_name(category), sizeof(ret));
4162 my_strlcat(ret, ", ", sizeof(ret));
4165 my_strlcat(ret, "\"", sizeof(ret));
4166 my_strlcat(ret, locale, sizeof(ret));
4167 my_strlcat(ret, "\"", sizeof(ret));
4170 my_strlcat(ret, "NULL", sizeof(ret));
4173 my_strlcat(ret, ") returned ", sizeof(ret));
4176 my_strlcat(ret, "\"", sizeof(ret));
4177 my_strlcat(ret, retval, sizeof(ret));
4178 my_strlcat(ret, "\"", sizeof(ret));
4181 my_strlcat(ret, "NULL", sizeof(ret));
4184 assert(strlen(ret) < sizeof(ret));
4193 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob