3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
4 * 2002, 2003, 2005, 2006, 2007, 2008 by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * A Elbereth Gilthoniel,
13 * silivren penna míriel
14 * o menel aglar elenath!
15 * Na-chaered palan-díriel
16 * o galadhremmin ennorath,
17 * Fanuilos, le linnathon
18 * nef aear, si nef aearon!
20 * [p.238 of _The Lord of the Rings_, II/i: "Many Meetings"]
23 /* utility functions for handling locale-specific stuff like what
24 * character represents the decimal point.
26 * All C programs have an underlying locale. Perl code generally doesn't pay
27 * any attention to it except within the scope of a 'use locale'. For most
28 * categories, it accomplishes this by just using different operations if it is
29 * in such scope than if not. However, various libc functions called by Perl
30 * are affected by the LC_NUMERIC category, so there are macros in perl.h that
31 * are used to toggle between the current locale and the C locale depending on
32 * the desired behavior of those functions at the moment. And, LC_MESSAGES is
33 * switched to the C locale for outputting the message unless within the scope
36 * There is more than the typical amount of variation between platforms with
37 * regard to locale handling. At the end of these introductory comments, are
38 * listed various relevent Configuration options, including some that can be
39 * used to pretend to some extent that this is being developed on a different
40 * platform than it actually is. This allows you to make changes and catch
41 * some errors without having access to those other platforms.
43 * This code now has multi-thread-safe locale handling on systems that support
44 * that. This is completely transparent to most XS code. On earlier systems,
45 * it would be possible to emulate thread-safe locales, but this likely would
46 * involve a lot of locale switching, and would require XS code changes.
47 * Macros could be written so that the code wouldn't have to know which type of
48 * system is being used.
50 * Table-driven code is used for simplicity and clarity, as many operations
51 * differ only in which category is being worked on. However the system
52 * categories need not be small contiguous integers, so do not lend themselves
53 * to table lookup. Instead we have created our own equivalent values which
54 * are all small contiguous non-negative integers, and translation functions
55 * between the two sets. For category 'LC_foo', the name of our index is
56 * LC_foo_INDEX_. Various parallel tables, indexed by these, are used for the
57 * translation. The tables are generated at compile-time based on platform
58 * characteristics and Configure options. They hide from the code many of the
59 * vagaries of the different locale implementations out there.
61 * On unthreaded perls, most operations expand out to just the basic
62 * setlocale() calls. That sort of is true on threaded perls on modern Windows
63 * systems where the same API, after set up, is used for thread-safe locale
64 * handling. (But there are complications on Windows due to internal character
65 * set issues.) On other systems, there is a completely different API,
66 * specified in POSIX 2008, to do thread-safe locales. On these systems, our
67 * bool_setlocale_2008_i() function is used to hide the different API from the
68 * outside. This makes it completely transparent to most XS code.
70 * A huge complicating factor is that the LC_NUMERIC category is normally held
71 * in the C locale, except during those relatively rare times when it needs to
72 * be in the underlying locale. There is a bunch of code to accomplish this,
73 * and to allow easy switches from one state to the other.
75 * In addition, the setlocale equivalents have versions for the return context,
76 * 'void' and 'bool', besides the full return value. This can present
77 * opportunities for avoiding work. We don't have to necessarily create a safe
78 * copy to return if no return is desired.
80 * There are 3.5 major implementations here; which one chosen depends on what
81 * the platform has available, and Configuration options.
83 * 1) Raw posix_setlocale(). This implementation is basically the libc
84 * setlocale(), with possibly minor tweaks. This is used for startup, and
85 * always for unthreaded perls, and when the API for safe locale threading
86 * is identical to the unsafe API (Windows, currently).
88 * This implementation is composed of two layers:
89 * a) posix_setlocale() implements the libc setlocale(). In most cases,
90 * it is just an alias for the libc version. But Windows doesn't
91 * fully conform to the POSIX standard, and this is a layer on top of
92 * libc to bring it more into conformance. And in Configurations
93 * where perl is to ignore some locale categories that the libc
94 * setlocale() knows about, there is a layer to cope with that.
95 * b) stdized_setlocale() is a layer above a) that fixes some vagaries in
96 * the return value of the libc setlocale(). On most platforms this
97 * layer is empty; it requires perl to be Configured with a parameter
98 * indicating the platform's defect, in order to be activated. The
99 * current ones are listed at the definition of the macro.
101 * 2) An implementation that adds a minimal layer above implementation 1),
102 * making that implementation uninterruptible and returning a
103 * per-thread/per-category value.
105 * 3a and 3b) An implementation of POSIX 2008 thread-safe locale handling,
106 * hiding from the programmer the completely different API for this.
107 * This automatically makes almost all code thread-safe without need for
108 * changes. This implementation is chosen on threaded perls when the
109 * platform properly supports the POSIX 2008 functions, and when there is no
110 * manual override to the contrary passed to Configure.
112 * 3a) is when the platform has a documented reliable querylocale() function
113 * or equivalent that is selected to be used.
114 * 3b) is when we have to emulate that functionality.
116 * Unfortunately, it seems that some platforms that claim to support these
117 * are buggy, in one way or another. There are workarounds encoded here,
118 * where feasible, for platforms where the bugs are amenable to that
119 * (glibc, for example). But other platforms instead don't use this
122 * z/OS (os390) is an outlier. Locales really don't work under threads when
123 * either the radix character isn't a dot, or attempts are made to change
124 * locales after the first thread is created. The reason is that IBM has made
125 * it thread-safe by refusing to change locales (returning failure if
126 * attempted) any time after an application has called pthread_create() to
127 * create another thread. The expectation is that an application will set up
128 * its locale information before the first fork, and be stable thereafter. But
129 * perl toggles LC_NUMERIC if the locale's radix character isn't a dot, as do
130 * the other toggles, which are less common.
132 * Associated with each implementation are three sets of macros that translate
133 * a consistent API into what that implementation needs. Each set consists of
134 * three macros with the suffixes:
135 * _c Means the argument is a locale category number known at compile time.
136 * An example would be LC_TIME. This token is a compile-time constant
137 * and can be passed to a '_c' macro.
138 * _r Means the argument is a locale category number whose value might not be
139 * known until runtime
140 * _i Means the argument is our internal index of a locale category
142 * The three sets are: ('_X' means one of '_c', '_r', '_i')
143 * 1) bool_setlocale_X()
144 * This calls the appropriate setlocale()-equivalent for the
145 * implementation, with the category and new locale. The input locale is
146 * not necessarily valid, so the return is true or false depending on
147 * whether or not the setlocale() succeeded. This is not used for
148 * querying the locale, so the input locale must not be NULL.
150 * This macro is suitable for toggling the locale back and forth during an
151 * operation. For example, the names of days and months under LC_TIME are
152 * strings that are also subject to LC_CTYPE. If the locales of these two
153 * categories differ, mojibake can result on many platforms. The code
154 * here will toggle LC_CTYPE into the locale of LC_TIME temporarily to
157 * Several categories require extra work when their locale is changed.
158 * LC_CTYPE, for example, requires the calculation of the table of which
159 * characters fold to which others under /i pattern matching or fc(), as
160 * folding is not a concept in POSIX. This table isn't needed when the
161 * LC_CTYPE locale gets toggled during an operation, and will be toggled
162 * back before return to the caller. To save work that would be
163 * discarded, the bool_setlocale_X() implementations don't do this extra
164 * work. Instead, there is a separate function for just this purpose to
165 * be done before control is transferred back to the external caller. All
166 * categories that have such requirements have such a function. The
167 * update_functions[] array contains pointers to them (or NULL for
168 * categories which don't need a function).
170 * Care must be taken to remember to call the separate function before
171 * returning to an external caller, and to not use things it updates
172 * before its call. An alternative approach would be to have
173 * bool_setlocale_X() always call the update, which would return
174 * immediately if a flag wasn't set indicating it was time to actually
177 * 2) void_setlocale_X()
178 * This is like bool_setlocale_X(), but it is used only when it is
179 * expected that the call must succeed, or something is seriously wrong.
180 * A panic is issued if it fails. The caller uses this form when it just
181 * wants to assume things worked.
184 * This returns a string that specifies the current locale for the given
185 * category given by the input argument. The string is safe from other
186 * threads zapping it, and the caller need not worry about freeing it, but
187 * it may be mortalized, so must be copied if you need to preserve it
188 * across calls, or long term. This returns the actual current locale,
189 * not the nominal. These differ, for example, when LC_NUMERIC is
190 * supposed to be a locale whose decimal radix character is a comma. As
191 * mentioned above, Perl actually keeps this category set to C in such
192 * circumstances so that XS code can just assume a dot radix character.
193 * querylocale_X() returns the locale that libc has stored at this moment,
194 * so most of the time will return a locale whose radix character is a
195 * dot. The macro query_nominal_locale_i() can be used to get the nominal
196 * locale that an external caller would expect, for all categories except
197 * LC_ALL. For that, you can use the function
198 * S_calculate_LC_ALL_string(). Or S_native_querylocale_i() will operate
201 * The underlying C API that this implements uses category numbers, hence the
202 * code is structured to use '_r' at the API level to convert to indexes, which
203 * are then used internally with the '_i' forms.
205 * The splitting apart into setting vs querying means that the return value of
206 * the bool macros is not subject to potential clashes with other threads,
207 * eliminating any need for the calling code to worry about that and get it
208 * wrong. Whereas, you do have to think about thread interactions when using a
211 * Additionally, for the implementations where there aren't any complications,
212 * a setlocale_i() is defined that is like plain setlocale(), returning the new
213 * locale. Thus it combines a bool_setlocale_X() with a querylocale_X(). It
214 * is used only for performance on implementations that allow it, such as
215 * non-threaded perls.
217 * There are also a few other macros herein that use this naming convention to
218 * describe their category parameter.
220 * Relevant Configure options
222 * -Accflags=-DNO_LOCALE
223 * This compiles perl to always use the C locale, ignoring any
224 * attempts to change it. This could be useful on platforms with a
225 * crippled locale implementation.
227 * -Accflags=-DNO_THREAD_SAFE_LOCALE
228 * Even if thread-safe operations are available on this platform and
229 * would otherwise be used (because this is a perl with multiplicity),
230 * perl is compiled to not use them. This could be useful on
231 * platforms where the libc is buggy.
233 * -Accflags=-DNO_POSIX_2008_LOCALE
234 * Even if the libc locale operations specified by the Posix 2008
235 * Standard are available on this platform and would otherwise be used
236 * (because this is a perl with multiplicity), perl is compiled to not
237 * use them. This could be useful on platforms where the libc is
238 * buggy. This is like NO_THREAD_SAFE_LOCALE, but has no effect on
239 * platforms that don't have these functions.
241 * -Accflags=-DUSE_POSIX_2008_LOCALE
242 * Normally, setlocale() is used for locale operations on perls
243 * compiled without multiplicity. This option causes the locale
244 * operations defined by the Posix 2008 Standard to always be used
245 * instead. This could be useful on platforms where the libc
246 * setlocale() is buggy.
248 * -Accflags=-DNO_THREAD_SAFE_QUERYLOCALE
249 * This applies only to platforms that have a querylocale() libc
250 * function. perl assumes that that function is thread-safe, unless
251 * overridden by this, typically in a hints file. When overridden,
252 * querylocale() is called only while the locale mutex is locked, and
253 * the result is copied to a per-thread place before unlocking.
255 * -Accflags=-DNO_LOCALE_CTYE
256 * -Accflags=-DNO_LOCALE_NUMERIC
259 * If the named category(ies) does(do) not exist on this platform,
260 * these have no effect. Otherwise they cause perl to be compiled to
261 * always keep the named category(ies) in the C locale XXX
263 * -Accflags=-DHAS_BROKEN_SETLOCALE_QUERY_LC_ALL
264 * This would be set in a hints file to tell perl that doing a libc
265 * setlocale(LC_ALL, NULL)
266 * can give erroneous results, and perl will compensate to get the
267 * correct results. This is known to be a problem in earlier AIX
270 * -Accflags=-DHAS_LF_IN_SETLOCALE_RETURN
271 * This would be set in a hints file to tell perl that a libc
272 * setlocale() can return results containing \n characters that need
273 * to be stripped off. khw believes there aren't any such platforms
274 * still in existence.
276 * -Accflags=USE_FAKE_LC_ALL_POSITIONAL_NOTATION
277 * This is used when developing Perl on a platform that uses
278 * 'name=value;' notation to represent LC_ALL when not all categories
279 * are the same. When so compiled, much of the code gets compiled
280 * and exercised that applies to platforms that instead use positional
281 * notation. This allows for finding many bugs in that portion of the
282 * implementation, without having to access such a platform.
284 * -Accflags=-DWIN32_USE_FAKE_OLD_MINGW_LOCALES
285 * This is used when developing Perl on a non-Windows platform to
286 * compile and exercise much of the locale-related code that instead
287 * applies to MingW platforms that don't use the more modern UCRT
288 * library. This allows for finding many bugs in that portion of the
289 * implementation, without having to access such a platform.
292 /* If the environment says to, we can output debugging information during
293 * initialization. This is done before option parsing, and before any thread
294 * creation, so can be a file-level static. (Must come before #including
298 /* Returns the Unix errno portion; ignoring any others. This is a macro here
299 * instead of putting it into perl.h, because unclear to khw what should be
301 #define GET_ERRNO saved_errno
304 static int debug_initialization = 0;
305 # define DEBUG_INITIALIZATION_set(v) (debug_initialization = v)
306 # define DEBUG_LOCALE_INITIALIZATION_ debug_initialization
308 # ifdef HAS_EXTENDED_OS_ERRNO
309 /* Output the non-zero errno and/or the non-zero extended errno */
310 # define DEBUG_ERRNO \
312 int extended = get_extended_os_errno(); \
313 const char * errno_string; \
314 if (GET_ERRNO == 0) { /* Skip output if both errno types are 0 */ \
315 if (LIKELY(extended == 0)) errno_string = ""; \
316 else errno_string = Perl_form(aTHX_ "; $^E=%d", extended); \
318 else if (LIKELY(extended == GET_ERRNO)) \
319 errno_string = Perl_form(aTHX_ "; $!=%d", GET_ERRNO); \
320 else errno_string = Perl_form(aTHX_ "; $!=%d, $^E=%d", \
321 GET_ERRNO, extended);
323 /* Output the errno, if non-zero */
324 # define DEBUG_ERRNO \
326 const char * errno_string = ""; \
327 if (GET_ERRNO != 0) { \
329 errno_string = Perl_form(aTHX_ "; $!=%d", GET_ERRNO); \
333 /* Automatically include the caller's file, and line number in debugging output;
334 * and the errno (and/or extended errno) if non-zero. On threaded perls add
336 # if defined(USE_ITHREADS) && ! defined(NO_LOCALE_THREADS)
337 # define DEBUG_PRE_STMTS \
339 PerlIO_printf(Perl_debug_log, "\n%s: %" LINE_Tf ": 0x%p%s: ", \
340 __FILE__, (line_t)__LINE__, aTHX_ \
343 # define DEBUG_PRE_STMTS \
345 PerlIO_printf(Perl_debug_log, "\n%s: %" LINE_Tf "%s: ", \
346 __FILE__, (line_t)__LINE__, \
349 # define DEBUG_POST_STMTS RESTORE_ERRNO;
351 # define debug_initialization 0
352 # define DEBUG_INITIALIZATION_set(v)
353 # define DEBUG_PRE_STMTS
354 # define DEBUG_POST_STMTS
358 #define PERL_IN_LOCALE_C
361 #ifdef WIN32_USE_FAKE_OLD_MINGW_LOCALES
363 /* Use -Accflags=-DWIN32_USE_FAKE_OLD_MINGW_LOCALES on a POSIX or *nix box
364 * to get a semblance of pretending the locale handling is that of a MingW
365 * that doesn't use UCRT (hence 'OLD' in the name). This exercizes code
366 * paths that are not compiled on non-Windows boxes, and allows for ASAN.
367 * This is thus a way to see if locale.c on Windows is likely going to
368 * compile, without having to use a real Win32 box. And running the test
369 * suite will verify to a large extent our logic and memory allocation
370 * handling for such boxes. And access to ASAN and PERL_MEMLOG Of course the underlying calls are to the POSIX
371 * libc, so any differences in implementation between those and the Windows
372 * versions will not be caught by this. */
375 # undef P_CS_PRECEDES
376 # undef CURRENCY_SYMBOL
378 # undef _configthreadlocale
379 # define _configthreadlocale(arg) NOOP
381 # define MultiByteToWideChar(cp, flags, byte_string, m1, wstring, req_size) \
382 (mbsrtowcs(wstring, &(byte_string), req_size, NULL) + 1)
383 # define WideCharToMultiByte(cp, flags, wstring, m1, byte_string, \
384 req_size, default_char, found_default_char) \
385 (wcsrtombs(byte_string, &(wstring), req_size, NULL) + 1)
389 static const wchar_t * wsetlocale_buf = NULL;
390 static Size_t wsetlocale_buf_size = 0;
391 static PerlInterpreter * wsetlocale_buf_aTHX = NULL;
395 S_wsetlocale(const int category, const wchar_t * wlocale)
397 /* Windows uses a setlocale that takes a wchar_t* locale. Other boxes
398 * don't have this, so this Windows replacement converts the wchar_t input
399 * to plain 'char*', calls plain setlocale(), and converts the result back
402 const char * byte_locale = NULL;
404 byte_locale = Win_wstring_to_byte_string(CP_UTF8, wlocale);
407 const char * byte_result = setlocale(category, byte_locale);
408 Safefree(byte_locale);
409 if (byte_result == NULL) {
413 const wchar_t * wresult = Win_byte_string_to_wstring(CP_UTF8, byte_result);
419 /* Emulate a global static memory return from wsetlocale(). This currently
420 * leaks at process end; would require changing LOCALE_TERM to fix that */
421 Size_t string_size = wcslen(wresult) + 1;
423 if (wsetlocale_buf_size == 0) {
424 Newx(wsetlocale_buf, string_size, wchar_t);
425 wsetlocale_buf_size = string_size;
430 wsetlocale_buf_aTHX = aTHX;
435 else if (string_size > wsetlocale_buf_size) {
436 Renew(wsetlocale_buf, string_size, wchar_t);
437 wsetlocale_buf_size = string_size;
440 Copy(wresult, wsetlocale_buf, string_size, wchar_t);
443 return wsetlocale_buf;
446 # define _wsetlocale(category, wlocale) S_wsetlocale(category, wlocale)
448 #endif /* WIN32_USE_FAKE_OLD_MINGW_LOCALES */
450 /* 'for' loop headers to hide the necessary casts */
451 #define for_all_individual_category_indexes(i) \
452 for (locale_category_index i = (locale_category_index) 0; \
454 i = (locale_category_index) ((int) i + 1))
456 #define for_all_but_0th_individual_category_indexes(i) \
457 for (locale_category_index i = (locale_category_index) 1; \
459 i = (locale_category_index) ((int) i + 1))
461 #define for_all_category_indexes(i) \
462 for (locale_category_index i = (locale_category_index) 0; \
463 i <= LC_ALL_INDEX_; \
464 i = (locale_category_index) ((int) i + 1))
467 # if defined(USE_FAKE_LC_ALL_POSITIONAL_NOTATION) && defined(LC_ALL)
469 /* This simulates an underlying positional notation for LC_ALL when compiled on
470 * a system that uses name=value notation. Use this to develop on Linux and
471 * make a quick check that things have some chance of working on a positional
472 * box. Enable by adding to the Congfigure parameters:
473 * -Accflags=USE_FAKE_LC_ALL_POSITIONAL_NOTATION
475 * NOTE it redefines setlocale() and usequerylocale()
479 S_positional_name_value_xlation(const char * locale, bool direction)
480 { /* direction == 1 is from name=value to positional
481 direction == 0 is from positional to name=value */
485 const char * individ_locales[LC_ALL_INDEX_] = { NULL };
487 /* This parses either notation */
488 switch (parse_LC_ALL_string(locale,
489 (const char **) &individ_locales,
490 no_override, /* Handled by other code */
491 false, /* Return only [0] if suffices */
492 false, /* Don't panic on error */
495 default: /* Some compilers don't realize that below is the complete
496 list of the available enum values */
503 SAVEFREEPV(individ_locales[0]);
504 return individ_locales[0];
507 calc_LC_ALL_format format = (direction)
508 ? EXTERNAL_FORMAT_FOR_SET
510 const char * retval = calculate_LC_ALL_string(individ_locales,
515 for_all_individual_category_indexes(i) {
516 Safefree(individ_locales[i]);
525 S_positional_setlocale(int cat, const char * locale)
527 if (cat != LC_ALL) return setlocale(cat, locale);
529 if (locale && strNE(locale, "")) {
530 locale = S_positional_name_value_xlation(locale, 0);
531 if (! locale) return NULL;
534 locale = setlocale(cat, locale);
535 if (locale == NULL) return NULL;
536 return S_positional_name_value_xlation(locale, 1);
540 # define setlocale(a,b) S_positional_setlocale(a,b)
541 # ifdef USE_POSIX_2008_LOCALE
544 S_positional_newlocale(int mask, const char * locale, locale_t base)
548 if (mask != LC_ALL_MASK) return newlocale(mask, locale, base);
550 if (strNE(locale, "")) locale = S_positional_name_value_xlation(locale, 0);
551 if (locale == NULL) return NULL;
552 return newlocale(LC_ALL_MASK, locale, base);
556 # define newlocale(a,b,c) S_positional_newlocale(a,b,c)
559 #endif /* End of fake positional notation */
570 /* The main errno that gets used is this one, on platforms that support it */
572 # define SET_EINVAL SETERRNO(EINVAL, LIB_INVARG)
577 /* If we have any of these library functions, we can reliably determine is a
578 * locale is a UTF-8 one or not. And if we aren't using locales at all, we act
579 * as if everything is the C locale, so the answer there is always "No, it
580 * isn't UTF-8"; this too is reliably accurate */
581 #if defined(HAS_SOME_LANGINFO) || defined(HAS_MBTOWC) \
582 || defined(HAS_MBRTOWC) || ! defined(USE_LOCALE)
583 # define HAS_RELIABLE_UTF8NESS_DETERMINATION
586 /* This is a starting guess as to when this is true. It definititely isn't
587 * true on *BSD where positional LC_ALL notation is used. Likely this will end
588 * up being defined in hints files. */
589 #ifdef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
590 # define NEWLOCALE_HANDLES_DISPARATE_LC_ALL
593 /* But regardless, we have to look at individual categories if some are
595 #ifdef HAS_IGNORED_LOCALE_CATEGORIES_
596 # undef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
600 /* Not all categories need be set to the same locale. This macro determines if
601 * 'name' which represents LC_ALL is uniform or disparate. There are two
602 * situations: 1) the platform uses unordered name=value pairs; 2) the platform
603 * uses ordered positional values, with a separator string between them */
604 # ifdef PERL_LC_ALL_SEPARATOR /* positional */
605 # define is_disparate_LC_ALL(name) cBOOL(instr(name, PERL_LC_ALL_SEPARATOR))
606 # else /* name=value */
608 /* In the, hopefully never occurring, event that the platform doesn't use
609 * either mechanism for disparate LC_ALL's, assume the name=value pairs
610 * form, rather than taking the extreme step of refusing to compile. Many
611 * programs won't have disparate locales, so will generally work */
612 # define PERL_LC_ALL_SEPARATOR ";"
613 # define is_disparate_LC_ALL(name) cBOOL( strchr(name, ';') \
614 && strchr(name, '='))
617 /* It is possible to compile perl to always keep any individual category in the
618 * C locale. This would be done where the implementation on a platform is
619 * flawed or incomplete. At the time of this writing, for example, OpenBSD has
620 * not implemented LC_COLLATE beyond the C locale. The 'category_available[]'
621 * table is a bool that says whether a category is changeable, or must be kept
622 * in C. This macro substitutes C for the locale appropriately, expanding to
623 * nothing on the more typical case where all possible categories present on
624 * the platform are handled. */
625 # ifdef HAS_IGNORED_LOCALE_CATEGORIES_
626 # define need_to_override_category(i) (! category_available[i])
627 # define override_ignored_category(i, new_locale) \
628 ((need_to_override_category(i)) ? "C" : (new_locale))
630 # define need_to_override_category(i) 0
631 # define override_ignored_category(i, new_locale) (new_locale)
634 PERL_STATIC_INLINE const char *
635 S_mortalized_pv_copy(pTHX_ const char * const pv)
637 PERL_ARGS_ASSERT_MORTALIZED_PV_COPY;
639 /* Copies the input pv, and arranges for it to be freed at an unspecified
646 const char * copy = savepv(pv);
653 /* Default values come from the C locale */
654 #define C_codeset "ANSI_X3.4-1968" /* Only in some Configurations, and usually
655 a single instance, so is a #define */
656 static const char C_decimal_point[] = ".";
658 #if (defined(USE_LOCALE_NUMERIC) && ! defined(TS_W32_BROKEN_LOCALECONV)) \
659 || ! ( defined(USE_LOCALE_NUMERIC) \
660 && (defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV)))
661 static const char C_thousands_sep[] = "";
664 /* Is the C string input 'name' "C" or "POSIX"? If so, and 'name' is the
665 * return of setlocale(), then this is extremely likely to be the C or POSIX
666 * locale. However, the output of setlocale() is documented to be opaque, but
667 * the odds are extremely small that it would return these two strings for some
668 * other locale. Note that VMS includes many non-ASCII characters in these two
669 * locales as controls and punctuation (below are hex bytes):
671 * punct: A1-A3 A5 A7-AB B0-B3 B5-B7 B9-BD BF-CF D1-DD DF-EF F1-FD
672 * Oddly, none there are listed as alphas, though some represent alphabetics
673 * http://www.nntp.perl.org/group/perl.perl5.porters/2013/02/msg198753.html */
674 #define isNAME_C_OR_POSIX(name) \
676 && (( *(name) == 'C' && (*(name + 1)) == '\0') \
677 || strEQ((name), "POSIX")))
679 #if defined(HAS_NL_LANGINFO_L) || defined(HAS_NL_LANGINFO)
680 # define HAS_SOME_LANGINFO
683 #define my_langinfo_c(item, category, locale, retbufp, retbuf_sizep, utf8ness) \
684 my_langinfo_i(item, category##_INDEX_, locale, retbufp, \
685 retbuf_sizep, utf8ness)
688 # define setlocale_debug_string_i(index, locale, result) \
689 my_setlocale_debug_string_i(index, locale, result, __LINE__)
690 # define setlocale_debug_string_c(category, locale, result) \
691 setlocale_debug_string_i(category##_INDEX_, locale, result)
692 # define setlocale_debug_string_r(category, locale, result) \
693 setlocale_debug_string_i(get_category_index(category), \
697 # define toggle_locale_i(index, locale) \
698 S_toggle_locale_i(aTHX_ index, locale, __LINE__)
699 # define toggle_locale_c(cat, locale) toggle_locale_i(cat##_INDEX_, locale)
700 # define restore_toggled_locale_i(index, locale) \
701 S_restore_toggled_locale_i(aTHX_ index, locale, __LINE__)
702 # define restore_toggled_locale_c(cat, locale) \
703 restore_toggled_locale_i(cat##_INDEX_, locale)
705 /* On systems without LC_ALL, pretending it exists anyway simplifies things.
706 * Choose a value for it that is very unlikely to clash with any actual
708 # define FAKE_LC_ALL PERL_INT_MIN
710 /* Below are parallel arrays for locale information indexed by our mapping of
711 * category numbers into small non-negative indexes. locale_table.h contains
712 * an entry like this for each individual category used on this system:
713 * PERL_LOCALE_TABLE_ENTRY(CTYPE, S_new_ctype)
715 * Each array redefines PERL_LOCALE_TABLE_ENTRY to generate the information
716 * needed for that array, and #includes locale_table.h to get the valid
719 * An entry for the conglomerate category LC_ALL is added here, immediately
720 * following the individual categories. (The treatment for it varies, so can't
721 * be in locale_table.h.)
723 * Following this, each array ends with an entry for illegal categories. All
724 * category numbers unknown to perl get mapped to this entry. This is likely
725 * to be a parameter error from the calling program; but it could be that this
726 * platform has a category we don't know about, in which case it needs to be
727 * added, using the paradigm of one of the existing categories. */
729 /* The first array is the locale categories perl uses on this system, used to
730 * map our index back to the system's category number. */
731 STATIC const int categories[] = {
733 # undef PERL_LOCALE_TABLE_ENTRY
734 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) LC_ ## name,
735 # include "locale_table.h"
743 (FAKE_LC_ALL + 1) /* Entry for unknown category; this number is unlikely
744 to clash with a real category */
747 # define GET_NAME_AS_STRING(token) # token
748 # define GET_LC_NAME_AS_STRING(token) GET_NAME_AS_STRING(LC_ ## token)
750 /* The second array is the category names. */
751 STATIC const char * const category_names[] = {
753 # undef PERL_LOCALE_TABLE_ENTRY
754 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) GET_LC_NAME_AS_STRING(name),
755 # include "locale_table.h"
758 # define LC_ALL_STRING "LC_ALL"
760 # define LC_ALL_STRING "If you see this, it is a bug in perl;" \
761 " please report it via perlbug"
766 # define LC_UNKNOWN_STRING "Locale category unknown to Perl; if you see" \
767 " this, it is a bug in perl; please report it" \
772 STATIC const Size_t category_name_lengths[] = {
774 # undef PERL_LOCALE_TABLE_ENTRY
775 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) \
776 STRLENs(GET_LC_NAME_AS_STRING(name)),
777 # include "locale_table.h"
779 STRLENs(LC_ALL_STRING),
780 STRLENs(LC_UNKNOWN_STRING)
783 /* Each entry includes space for the '=' and ';' */
784 # undef PERL_LOCALE_TABLE_ENTRY
785 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) \
786 + STRLENs(GET_LC_NAME_AS_STRING(name)) + 2
788 STATIC const Size_t lc_all_boiler_plate_length = 1 /* space for trailing NUL */
789 # include "locale_table.h"
792 /* A few categories require additional setup when they are changed. This table
793 * points to the functions that do that setup */
794 STATIC void (*update_functions[]) (pTHX_ const char *, bool force) = {
796 # undef PERL_LOCALE_TABLE_ENTRY
797 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) call_back,
798 # include "locale_table.h"
801 NULL, /* No update for unknown category */
804 # if defined(HAS_IGNORED_LOCALE_CATEGORIES_)
806 /* Indicates if each category on this platform is available to use not in
808 STATIC const bool category_available[] = {
810 # undef PERL_LOCALE_TABLE_ENTRY
811 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) LC_ ## name ## _AVAIL_,
812 # include "locale_table.h"
820 false /* LC_UNKNOWN_AVAIL_ */
824 # if defined(USE_POSIX_2008_LOCALE)
826 STATIC const int category_masks[] = {
828 # undef PERL_LOCALE_TABLE_ENTRY
829 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) LC_ ## name ## _MASK,
830 # include "locale_table.h"
832 LC_ALL_MASK, /* Will rightly refuse to compile unless this is defined */
833 0 /* Empty mask for unknown category */
837 # if ! defined(PERL_LC_ALL_USES_NAME_VALUE_PAIRS)
839 /* On platforms that use positional notation for expressing LC_ALL, this maps
840 * the position of each category to our corresponding internal index for it.
841 * This is initialized at run time if needed. LC_ALL_INDEX_ is not legal for
842 * an individual locale, hence marks the elements here as not actually
846 map_LC_ALL_position_to_index[LC_ALL_INDEX_] = { LC_ALL_INDEX_ };
850 #if defined(USE_LOCALE) || defined(DEBUGGING)
853 S_get_displayable_string(pTHX_
854 const char * const s,
855 const char * const e,
858 PERL_ARGS_ASSERT_GET_DISPLAYABLE_STRING;
865 bool prev_was_printable = TRUE;
866 bool first_time = TRUE;
869 /* Worst case scenario: All are non-printable so have a blank between each.
870 * If UTF-8, all are the largest possible code point; otherwise all are a
871 * single byte. '(2 + 1)' is from each byte takes 2 characters to
872 * display, and a blank (or NUL for the final one) after it */
873 const Size_t size = (e - s) * (2 + 1) * ((is_utf8) ? UVSIZE : 1);
874 Newxz(ret, size, char);
879 ? utf8_to_uvchr_buf((U8 *) t, e, NULL)
882 if (! prev_was_printable) {
883 my_strlcat(ret, " ", size);
886 /* Escape these to avoid any ambiguity */
887 if (cp == ' ' || cp == '\\') {
888 my_strlcat(ret, "\\", size);
890 my_strlcat(ret, Perl_form(aTHX_ "%c", (U8) cp), size);
891 prev_was_printable = TRUE;
895 my_strlcat(ret, " ", size);
897 my_strlcat(ret, Perl_form(aTHX_ "%02" UVXf, cp), size);
898 prev_was_printable = FALSE;
900 t += (is_utf8) ? UTF8SKIP(t) : 1;
910 # define get_category_index(cat) get_category_index_helper(cat, NULL, __LINE__)
912 STATIC locale_category_index
913 S_get_category_index_helper(pTHX_ const int category, bool * succeeded,
914 const line_t caller_line)
916 PERL_ARGS_ASSERT_GET_CATEGORY_INDEX_HELPER;
918 /* Given a category, return the equivalent internal index we generally use
919 * instead, warn or panic if not found. */
921 locale_category_index i;
923 # undef PERL_LOCALE_TABLE_ENTRY
924 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) \
925 case LC_ ## name: i = LC_ ## name ## _INDEX_; break;
929 # include "locale_table.h"
931 case LC_ALL: i = LC_ALL_INDEX_; break;
934 default: goto unknown_locale;
937 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
938 "index of category %d (%s) is %d;"
939 " called from %" LINE_Tf "\n",
940 category, category_names[i], i, caller_line));
952 return LC_ALL_INDEX_; /* Arbitrary */
955 locale_panic_via_(Perl_form(aTHX_ "Unknown locale category %d", category),
956 __FILE__, caller_line);
957 NOT_REACHED; /* NOTREACHED */
960 #endif /* ifdef USE_LOCALE */
963 Perl_force_locale_unlock(pTHX)
965 /* Remove any locale mutex, in preperation for an inglorious termination,
966 * typically a panic */
968 #if defined(USE_LOCALE_THREADS)
970 /* If recursively locked, clear all at once */
971 if (PL_locale_mutex_depth > 1) {
972 PL_locale_mutex_depth = 1;
975 if (PL_locale_mutex_depth > 0) {
983 #ifdef USE_POSIX_2008_LOCALE
986 S_use_curlocale_scratch(pTHX)
988 /* This function is used to hide from the caller the case where the current
989 * locale_t object in POSIX 2008 is the global one, which is illegal in
990 * many of the P2008 API calls. This checks for that and, if necessary
991 * creates a proper P2008 object. Any prior object is deleted, as is any
992 * remaining object during global destruction. */
994 locale_t cur = uselocale((locale_t) 0);
996 if (cur != LC_GLOBAL_LOCALE) {
1000 if (PL_scratch_locale_obj) {
1001 freelocale(PL_scratch_locale_obj);
1004 PL_scratch_locale_obj = duplocale(LC_GLOBAL_LOCALE);
1005 return PL_scratch_locale_obj;
1011 Perl_locale_panic(const char * msg,
1012 const line_t immediate_caller_line,
1013 const char * const higher_caller_file,
1014 const line_t higher_caller_line)
1016 PERL_ARGS_ASSERT_LOCALE_PANIC;
1020 force_locale_unlock();
1022 #ifdef USE_C_BACKTRACE
1023 dump_c_backtrace(Perl_debug_log, 20, 1);
1026 const char * called_by = "";
1027 if ( strNE(__FILE__, higher_caller_file)
1028 || immediate_caller_line != higher_caller_line)
1030 called_by = Perl_form(aTHX_ "\nCalled by %s: %" LINE_Tf "\n",
1031 higher_caller_file, higher_caller_line);
1036 const char * errno_text;
1038 #ifdef HAS_EXTENDED_OS_ERRNO
1040 const int extended_errnum = get_extended_os_errno();
1041 if (errno != extended_errnum) {
1042 errno_text = Perl_form(aTHX_ "; errno=%d, $^E=%d",
1043 errno, extended_errnum);
1050 errno_text = Perl_form(aTHX_ "; errno=%d", errno);
1053 /* diag_listed_as: panic: %s */
1054 Perl_croak(aTHX_ "%s: %" LINE_Tf ": panic: %s%s%s\n",
1055 __FILE__, immediate_caller_line,
1056 msg, errno_text, called_by);
1059 /* Macros to report and croak on an unexpected failure to set the locale. The
1060 * via version has more stack trace information */
1061 #define setlocale_failure_panic_i(i, cur, fail, line, higher_line) \
1062 setlocale_failure_panic_via_i(i, cur, fail, __LINE__, line, \
1063 __FILE__, higher_line)
1065 #define setlocale_failure_panic_c(cat, cur, fail, line, higher_line) \
1066 setlocale_failure_panic_i(cat##_INDEX_, cur, fail, line, higher_line)
1068 #if defined(LC_ALL) && defined(USE_LOCALE)
1070 /* Expands to the code to
1071 * result = savepvn(s, len)
1072 * if the category whose internal index is 'i' doesn't need to be kept in the C
1073 * locale on this system, or if 'action is 'no_override'. Otherwise it expands
1075 * result = savepv("C")
1076 * unless 'action' isn't 'check_that_overridden', in which case if the string
1077 * 's' isn't already "C" it panics */
1078 # ifndef HAS_IGNORED_LOCALE_CATEGORIES_
1079 # define OVERRIDE_AND_SAVEPV(s, len, result, i, action) \
1080 result = savepvn(s, len)
1082 # define OVERRIDE_AND_SAVEPV(s, len, result, i, action) \
1084 if (LIKELY( ! need_to_override_category(i) \
1085 || action == no_override)) { \
1086 result = savepvn(s, len); \
1089 const char * temp = savepvn(s, len); \
1090 result = savepv(override_ignored_category(i, temp)); \
1091 if (action == check_that_overridden && strNE(result, temp)) { \
1092 locale_panic_(Perl_form(aTHX_ \
1093 "%s expected to be '%s', instead is '%s'", \
1094 category_names[i], result, temp)); \
1101 STATIC parse_LC_ALL_string_return
1102 S_parse_LC_ALL_string(pTHX_ const char * string,
1103 const char ** output,
1104 const parse_LC_ALL_STRING_action override,
1105 bool always_use_full_array,
1106 const bool panic_on_error,
1107 const line_t caller_line)
1109 /* This function parses the value of the input 'string' which is expected
1110 * to be the representation of an LC_ALL locale, and splits the result into
1111 * the values for the individual component categories, returning those in
1112 * the 'output' array. Each array value will be a savepv() copy that is
1113 * the responsibility of the caller to make sure gets freed
1115 * The locale for each category is independent of the other categories.
1116 * Often, they are all the same, but certainly not always. Perl, in fact,
1117 * usually keeps LC_NUMERIC in the C locale, regardless of the underlying
1118 * locale. LC_ALL has to be able to represent the case of when not all
1119 * categories have the same locale. Platforms have differing ways of
1120 * representing this. Internally, this file uses the 'name=value;'
1121 * representation found on some platforms, so this function always looks
1122 * for and parses that. Other platforms use a positional notation. On
1123 * those platforms, this function also parses that form. It examines the
1124 * input to see which form is being parsed.
1126 * Often, all categories will have the same locale. This is special cased
1127 * if 'always_use_full_array' is false on input:
1128 * 1) If the input 'string' is a single value, this function doesn't
1129 * store anything into 'output', and returns 'no_array'
1130 * 2) Some platforms will return multiple occurrences of the same
1131 * value rather than coalescing them down to a single one. HP-UX
1132 * is such a one. This function will do that collapsing for you,
1133 * returning 'only_element_0' and saving the single value in
1134 * output[0], which the caller will need to arrange to be freed.
1135 * The rest of output[] is undefined, and does not need to be
1138 * Otherwise, the input 'string' may not be valid. This function looks
1139 * mainly for syntactic errors, and if found, returns 'invalid'. 'output'
1140 * will not be filled in in that case, but the input state of it isn't
1141 * necessarily preserved. Turning on -DL debugging will give details as to
1142 * the error. If 'panic_on_error' is 'true', the function panics instead
1143 * of returning on error, with a message giving the details.
1145 * Otherwise, output[] will be filled with the individual locale names for
1146 * all categories on the system, 'full_array' will be returned, and the
1147 * caller needs to arrange for each to be freed. This means that either at
1148 * least one category differed from the others, or 'always_use_full_array' was
1151 * perl may be configured to ignore changes to a category's locale to
1152 * non-C. The parameter 'override' tells this function what to do when
1153 * encountering such an illegal combination:
1155 * no_override indicates to take no special action
1156 * override_if_ignored, indicates to return 'C' instead of what the
1157 * input string actually says.
1158 * check_that_overridden indicates to panic if the string says the
1159 * category is not 'C'. This is used when
1160 * non-C is very unexpected behavior.
1163 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1164 "Entering parse_LC_ALL_string; called from %" \
1165 LINE_Tf "\nnew='%s'\n", caller_line, string));
1167 # ifdef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
1169 const char separator[] = ";";
1170 const Size_t separator_len = 1;
1171 const bool single_component = (strchr(string, ';') == NULL);
1175 /* It's possible (but quite unlikely) that the separator string is an '='
1176 * or a ';'. Requiring both to be present for using the 'name=value;' form
1177 * properly handles those possibilities */
1178 const bool name_value = strchr(string, '=') && strchr(string, ';');
1179 const char * separator;
1180 Size_t separator_len;
1181 bool single_component;
1185 single_component = false; /* Since has both [;=], must be multi */
1188 separator = PERL_LC_ALL_SEPARATOR;
1189 separator_len = STRLENs(PERL_LC_ALL_SEPARATOR);
1190 single_component = instr(string, separator) == NULL;
1193 Size_t component_number = 0; /* Position in the parsing loop below */
1196 # ifndef HAS_IGNORED_LOCALE_CATEGORIES_
1197 PERL_UNUSED_ARG(override);
1200 /* Any ignored categories are to be set to "C", so if this single-component
1201 * LC_ALL isn't to C, it has both "C" and non-C, so isn't really a single
1202 * component. All the non-ignored categories are set to the input
1203 * component, but the ignored ones are overridden to be C.
1205 * This incidentally handles the case where the string is "". The return
1206 * will be C for each ignored category and "" for the others. Then the
1207 * caller can individually set each category, and get the right answer. */
1208 if (single_component && ! isNAME_C_OR_POSIX(string)) {
1209 for_all_individual_category_indexes(i) {
1210 OVERRIDE_AND_SAVEPV(string, strlen(string), output[i], i, override);
1218 if (single_component) {
1219 if (! always_use_full_array) {
1223 for_all_individual_category_indexes(i) {
1224 output[i] = savepv(string);
1230 /* Here the input is multiple components. Parse through them. (It is
1231 * possible that these components are all the same, so we check, and if so,
1232 * return just the 0th component (unless 'always_use_full_array' is true)
1234 * This enum notes the possible errors findable in parsing */
1239 contains_LC_ALL_element
1242 /* Keep track of the categories we have encountered so far */
1243 bool seen[LC_ALL_INDEX_] = { false };
1245 Size_t index; /* Our internal index for the current category */
1246 const char * s = string;
1247 const char * e = s + strlen(string);
1248 const char * category_end = NULL;
1249 const char * saved_first = NULL;
1251 /* Parse the input locale string */
1254 /* 'separator' has been set up to delimit the components */
1255 const char * next_sep = instr(s, separator);
1256 if (! next_sep) { /* At the end of the input */
1260 # ifndef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
1263 /* Get the index of the category in this position */
1264 index = map_LC_ALL_position_to_index[component_number++];
1270 { /* Get the category part when each component is the
1271 * 'category=locale' form */
1273 category_end = strchr(s, '=');
1275 /* The '=' terminates the category name. If no '=', is improper
1277 if (! category_end) {
1282 /* Find our internal index of the category name; uses a linear
1283 * search. (XXX This could be avoided by various means, but the
1284 * maximum likely search is 6 items, and khw doesn't think the
1285 * added complexity would save very much at all.) */
1286 const unsigned int name_len = (unsigned int) (category_end - s);
1287 for (index = 0; index < C_ARRAY_LENGTH(category_names); index++) {
1288 if ( name_len == category_name_lengths[index]
1289 && memEQ(s, category_names[index], name_len))
1291 goto found_category;
1295 /* Here, the category is not in our list. */
1296 error = unknown_category;
1299 found_category: /* The system knows about this category. */
1301 if (index == LC_ALL_INDEX_) {
1302 error = contains_LC_ALL_element;
1306 /* The locale name starts just beyond the '=' */
1307 s = category_end + 1;
1309 /* Linux (and maybe others) doesn't treat a duplicate category in
1310 * the string as an error. Instead it uses the final occurrence as
1311 * the intended value. So if this is a duplicate, free the former
1312 * value before setting the new one */
1314 Safefree(output[index]);
1321 /* Here, 'index' contains our internal index number for the current
1322 * category, and 's' points to the beginning of the locale name for
1324 OVERRIDE_AND_SAVEPV(s, next_sep - s, output[index], index, override);
1326 if (! always_use_full_array) {
1327 if (! saved_first) {
1328 saved_first = output[index];
1331 if (strNE(saved_first, output[index])) {
1332 always_use_full_array = true;
1337 /* Next time start from the new position */
1338 s = next_sep + separator_len;
1341 /* Finished looping through all the categories
1343 * Check if the input was incomplete. */
1345 # ifndef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
1347 if (! name_value) { /* Positional notation */
1348 if (component_number != LC_ALL_INDEX_) {
1357 { /* Here is the name=value notation */
1358 for_all_individual_category_indexes(i) {
1366 /* In the loop above, we changed 'always_use_full_array' to true iff not all
1367 * categories have the same locale. Hence, if it is still 'false', all of
1368 * them are the same. */
1369 if (always_use_full_array) {
1373 /* Free the dangling ones */
1374 for_all_but_0th_individual_category_indexes(i) {
1375 Safefree(output[i]);
1379 return only_element_0;
1383 /* Don't leave memory dangling that we allocated before the failure */
1384 for_all_individual_category_indexes(i) {
1386 Safefree(output[i]);
1392 const char * display_start = s;
1393 const char * display_end = e;
1397 msg = "doesn't list every locale category";
1398 display_start = string;
1401 msg = "needs an '=' to split name=value";
1403 case unknown_category:
1404 msg = "is an unknown category";
1405 display_end = (category_end && category_end > display_start)
1409 case contains_LC_ALL_element:
1410 msg = "has LC_ALL, which is illegal here";
1414 msg = Perl_form(aTHX_ "'%.*s' %s\n",
1415 (int) (display_end - display_start),
1416 display_start, msg);
1418 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s", msg));
1420 if (panic_on_error) {
1421 locale_panic_via_(msg, __FILE__, caller_line);
1427 # undef OVERRIDE_AND_SAVEPV
1430 /*==========================================================================
1431 * Here starts the code that gives a uniform interface to its callers, hiding
1432 * the differences between platforms.
1434 * base_posix_setlocale_() presents a consistent POSIX-compliant interface to
1435 * setlocale(). Windows requres a customized base-level setlocale(). This
1436 * layer should only be used by the next level up: the plain posix_setlocale
1437 * layer. Any necessary mutex locking needs to be done at a higher level. The
1438 * return may be overwritten by the next call to this function */
1440 # define base_posix_setlocale_(cat, locale) win32_setlocale(cat, locale)
1442 # define base_posix_setlocale_(cat, locale) \
1443 ((const char *) setlocale(cat, locale))
1446 /*==========================================================================
1447 * Here is the main posix layer. It is the same as the base one unless the
1448 * system is lacking LC_ALL, or there are categories that we ignore, but that
1449 * the system libc knows about */
1451 #if ! defined(USE_LOCALE) \
1452 || (defined(LC_ALL) && ! defined(HAS_IGNORED_LOCALE_CATEGORIES_))
1453 # define posix_setlocale(cat, locale) base_posix_setlocale_(cat, locale)
1455 # define posix_setlocale(cat, locale) \
1456 S_posix_setlocale_with_complications(aTHX_ cat, locale, __LINE__)
1459 S_posix_setlocale_with_complications(pTHX_ const int cat,
1460 const char * new_locale,
1461 const line_t caller_line)
1463 /* This implements the posix layer above the base posix layer.
1464 * It is needed to reconcile our internal records that reflect only a
1465 * proper subset of the categories known by the system. */
1467 /* Querying the current locale returns the real value */
1468 if (new_locale == NULL) {
1469 new_locale = base_posix_setlocale_(cat, NULL);
1474 const char * locale_on_entry = NULL;
1476 /* If setting from the environment, actually do the set to get the system's
1477 * idea of what that means; we may have to override later. */
1478 if (strEQ(new_locale, "")) {
1479 locale_on_entry = base_posix_setlocale_(cat, NULL);
1480 assert(locale_on_entry);
1481 new_locale = base_posix_setlocale_(cat, "");
1490 const char * new_locales[LC_ALL_INDEX_] = { NULL };
1492 if (cat == LC_ALL) {
1493 switch (parse_LC_ALL_string(new_locale,
1494 (const char **) &new_locales,
1495 override_if_ignored, /* Override any
1498 false, /* Return only [0] if suffices */
1499 false, /* Don't panic on error */
1509 case only_element_0:
1510 new_locale = new_locales[0];
1511 SAVEFREEPV(new_locale);
1516 /* Turn the array into a string that the libc setlocale() should
1517 * understand. (Another option would be to loop, setting the
1518 * individual locales, and then return base(cat, NULL) */
1519 new_locale = calculate_LC_ALL_string(new_locales,
1520 EXTERNAL_FORMAT_FOR_SET,
1524 for_all_individual_category_indexes(i) {
1525 Safefree(new_locales[i]);
1528 /* And call the libc setlocale. We could avoid this call if
1529 * locale_on_entry is set and eq the new_locale. But that would be
1530 * only for the relatively rare case of the desired locale being
1531 * "", and the time spent in doing the string compare might be more
1532 * than that of just setting it unconditionally */
1533 new_locale = base_posix_setlocale_(cat, new_locale);
1544 /* Here, 'new_locale' is a single value, not an aggregation. Just set it.
1547 base_posix_setlocale_(cat,
1548 override_ignored_category(
1549 get_category_index(cat), new_locale));
1558 /* 'locale_on_entry' being set indicates there has likely been a change in
1559 * locale which needs to be restored */
1560 if (locale_on_entry) {
1561 if (! base_posix_setlocale_(cat, locale_on_entry)) {
1562 setlocale_failure_panic_i(get_category_index(cat),
1563 NULL, locale_on_entry,
1564 __LINE__, caller_line);
1574 /* End of posix layer
1575 *==========================================================================
1577 * The next layer up is to catch vagaries and bugs in the libc setlocale return
1578 * value. The return is not guaranteed to be stable.
1580 * Any necessary mutex locking needs to be done at a higher level.
1582 * On most platforms this layer is empty, expanding to just the layer
1583 * below. To enable it, call Configure with either or both:
1584 * -Accflags=-DHAS_LF_IN_SETLOCALE_RETURN
1585 * to indicate that extraneous \n characters can be returned
1587 * -Accflags=-DHAS_BROKEN_SETLOCALE_QUERY_LC_ALL
1588 * to indicate that setlocale(LC_ALL, NULL) cannot be relied
1592 #define STDIZED_SETLOCALE_LOCK POSIX_SETLOCALE_LOCK
1593 #define STDIZED_SETLOCALE_UNLOCK POSIX_SETLOCALE_UNLOCK
1594 #if ! defined(USE_LOCALE) \
1595 || ! ( defined(HAS_LF_IN_SETLOCALE_RETURN) \
1596 || defined(HAS_BROKEN_SETLOCALE_QUERY_LC_ALL))
1597 # define stdized_setlocale(cat, locale) posix_setlocale(cat, locale)
1598 # define stdize_locale(cat, locale) (locale)
1600 # define stdized_setlocale(cat, locale) \
1601 S_stdize_locale(aTHX_ cat, posix_setlocale(cat, locale), __LINE__)
1604 S_stdize_locale(pTHX_ const int category,
1605 const char *input_locale,
1606 const line_t caller_line)
1608 /* The return value of setlocale() is opaque, but is required to be usable
1609 * as input to a future setlocale() to create the same state.
1610 * Unfortunately not all systems are compliant. This function brings those
1611 * outliers into conformance. It is based on what problems have arisen in
1614 * This has similar constraints as the posix layer. You need to lock
1615 * around it until its return is safely copied or no longer needed. (The
1616 * return may point to a global static buffer or may be mortalized.)
1618 * The current things this corrects are:
1619 * 1) A new-line. This function chops any \n characters
1620 * 2) A broken 'setlocale(LC_ALL, foo)' This constructs a proper returned
1621 * string from the constituent categories
1623 * If no changes were made, the input is returned as-is */
1625 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1626 "Entering stdize_locale(%d, '%s');"
1627 " called from %" LINE_Tf "\n",
1628 category, input_locale, caller_line));
1630 if (input_locale == NULL) {
1635 char * retval = (char *) input_locale;
1637 # if defined(LC_ALL) && defined(HAS_BROKEN_SETLOCALE_QUERY_LC_ALL)
1639 /* If setlocale(LC_ALL, NULL) is broken, compute what the system
1640 * actually thinks it should be from its individual components */
1641 if (category == LC_ALL) {
1642 retval = (char *) calculate_LC_ALL_string(
1643 NULL, /* query each individ locale */
1644 EXTERNAL_FORMAT_FOR_SET,
1650 # ifdef HAS_NL_IN_SETLOCALE_RETURN
1652 char * first_bad = NULL;
1656 PERL_UNUSED_ARG(category);
1657 PERL_UNUSED_ARG(caller_line);
1659 # define INPUT_LOCALE retval
1660 # define MARK_CHANGED
1663 char * individ_locales[LC_ALL_INDEX_] = { NULL };
1664 bool made_changes = false;
1666 if (category != LC_ALL) {
1667 individ_locales[0] = retval;
1672 /* And parse the locale string, splitting into its individual
1674 switch (parse_LC_ALL_string(retval,
1675 (const char **) &individ_locales,
1676 check_that_overridden, /* ignored
1680 false, /* Return only [0] if suffices */
1681 false, /* Don't panic on error */
1688 case full_array: /* Loop below through all the component categories.
1690 upper = LC_ALL_INDEX_ - 1;
1694 /* All categories here are set to the same locale, and the parse
1695 * didn't fill in any of 'individ_locales'. Set the 0th element to
1697 individ_locales[0] = retval;
1700 case only_element_0: /* Element 0 is the only element we need to look
1707 for (unsigned int i = 0; i <= upper; i++)
1709 # define INPUT_LOCALE individ_locales[i]
1710 # define MARK_CHANGED made_changes = true;
1711 # endif /* Has LC_ALL */
1714 first_bad = (char *) strchr(INPUT_LOCALE, '\n');
1716 /* Most likely, there isn't a problem with the input */
1717 if (UNLIKELY(first_bad)) {
1719 /* This element will need to be adjusted. Create a modifiable
1722 retval = savepv(INPUT_LOCALE);
1725 /* Translate the found position into terms of the copy */
1726 first_bad = retval + (first_bad - INPUT_LOCALE);
1728 /* Get rid of the \n and what follows. (Originally, only a
1729 * trailing \n was stripped. Unsure what to do if not trailing) */
1730 *((char *) first_bad) = '\0';
1731 } /* End of needs adjusting */
1732 } /* End of looking for problems */
1736 /* If we had multiple elements, extra work is required */
1739 /* If no changes were made to the input, 'retval' already contains it
1743 /* But if did make changes, need to calculate the new value */
1744 retval = (char *) calculate_LC_ALL_string(
1745 (const char **) &individ_locales,
1746 EXTERNAL_FORMAT_FOR_SET,
1751 /* And free the no-longer needed memory */
1752 for (unsigned int i = 0; i <= upper; i++) {
1753 Safefree(individ_locales[i]);
1758 # undef INPUT_LOCALE
1759 # undef MARK_CHANGED
1760 # endif /* HAS_NL_IN_SETLOCALE_RETURN */
1762 return (const char *) retval;
1765 #endif /* USE_LOCALE */
1767 /* End of stdize_locale layer
1769 * ==========================================================================
1771 * The next many lines form several implementations of a layer above the
1772 * close-to-the-metal 'posix' and 'stdized' macros. They are used to present a
1773 * uniform API to the rest of the code in this file in spite of the disparate
1774 * underlying implementations. Which implementation gets compiled depends on
1775 * the platform capabilities (and some user choice) as determined by Configure.
1777 * As more fully described in the introductory comments in this file, the
1778 * API of each implementation consists of three sets of macros. Each set has
1779 * three variants with suffixes '_c', '_r', and '_i'. In the list below '_X'
1780 * is to be replaced by any of these suffixes.
1782 * 1) bool_setlocale_X attempts to set the given category's locale to the
1783 * given value, returning if it worked or not.
1784 * 2) void_setlocale_X is like the corresponding bool_setlocale, but used when
1785 * success is the only sane outcome, so failure causes it
1787 * 3) querylocale_X to see what the given category's locale is
1789 * 4) setlocale_i() is defined only in those implementations where the bool
1790 * and query forms are essentially the same, and can be
1791 * combined to save CPU time.
1793 * Each implementation below is separated by ==== lines, and includes bool,
1794 * void, and query macros. The query macros are first, followed by any
1795 * functions needed to implement them. Then come the bool, again followed by
1796 * any implementing functions Then are the void macros; next is setlocale_i if
1797 * present on this implementation. Finally are any helper functions. The sets
1798 * in each implementation are separated by ---- lines.
1800 * The returned strings from all the querylocale...() forms in all
1801 * implementations are thread-safe, and the caller should not free them,
1802 * but each may be a mortalized copy. If you need something stable across
1803 * calls, you need to savepv() the result yourself.
1805 *===========================================================================*/
1807 #if (! defined(USE_LOCALE_THREADS) && ! defined(USE_POSIX_2008_LOCALE)) \
1808 || ( defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE))
1810 /* For non-threaded perls, the implementation just expands to the base-level
1811 * functions (except if we are Configured to nonetheless use the POSIX 2008
1812 * interface) This implementation is also used on threaded perls where
1813 * threading is invisible to us. Currently this is only on later Windows
1816 # define querylocale_r(cat) mortalized_pv_copy(stdized_setlocale(cat, NULL))
1817 # define querylocale_c(cat) querylocale_r(cat)
1818 # define querylocale_i(i) querylocale_c(categories[i])
1820 /*---------------------------------------------------------------------------*/
1822 # define bool_setlocale_r(cat, locale) cBOOL(posix_setlocale(cat, locale))
1823 # define bool_setlocale_i(i, locale) \
1824 bool_setlocale_c(categories[i], locale)
1825 # define bool_setlocale_c(cat, locale) bool_setlocale_r(cat, locale)
1827 /*---------------------------------------------------------------------------*/
1829 # define void_setlocale_r_with_caller(cat, locale, file, line) \
1831 if (! bool_setlocale_r(cat, locale)) \
1832 setlocale_failure_panic_via_i(get_category_index(cat), \
1833 NULL, locale, __LINE__, 0, \
1837 # define void_setlocale_c_with_caller(cat, locale, file, line) \
1838 void_setlocale_r_with_caller(cat, locale, file, line)
1840 # define void_setlocale_i_with_caller(i, locale, file, line) \
1841 void_setlocale_r_with_caller(categories[i], locale, file, line)
1843 # define void_setlocale_r(cat, locale) \
1844 void_setlocale_r_with_caller(cat, locale, __FILE__, __LINE__)
1845 # define void_setlocale_c(cat, locale) void_setlocale_r(cat, locale)
1846 # define void_setlocale_i(i, locale) void_setlocale_r(categories[i], locale)
1848 /*---------------------------------------------------------------------------*/
1850 /* setlocale_i is only defined for Configurations where the libc setlocale()
1851 * doesn't need any tweaking. It allows for some shortcuts */
1852 # ifndef USE_LOCALE_THREADS
1853 # define setlocale_i(i, locale) stdized_setlocale(categories[i], locale)
1855 # elif defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE)
1857 /* On Windows, we don't know at compile time if we are in thread-safe mode or
1858 * not. If we are, we can just return the result of the layer below us. If we
1859 * are in unsafe mode, we need to first copy that result to a safe place while
1860 * in a critical section */
1862 # define setlocale_i(i, locale) S_setlocale_i(aTHX_ categories[i], locale)
1865 S_setlocale_i(pTHX_ const int category, const char * locale)
1867 if (LIKELY(_configthreadlocale(0) == _ENABLE_PER_THREAD_LOCALE)) {
1868 return stdized_setlocale(category, locale);
1872 const char * retval = save_to_buffer(stdized_setlocale(category, locale),
1874 &PL_setlocale_bufsize);
1882 /*===========================================================================*/
1883 #elif defined(USE_LOCALE_THREADS) \
1884 && ! defined(USE_THREAD_SAFE_LOCALE)
1886 /* Here, there are threads, and there is no support for thread-safe
1887 * operation. This is a dangerous situation, which perl is documented as
1888 * not supporting, but it arises in practice. We can do a modicum of
1889 * automatic mitigation by making sure there is a per-thread return from
1890 * setlocale(), and that a mutex protects it from races */
1892 # define querylocale_r(cat) \
1893 mortalized_pv_copy(less_dicey_setlocale_r(cat, NULL))
1894 # define querylocale_c(cat) querylocale_r(cat)
1895 # define querylocale_i(i) querylocale_r(categories[i])
1898 S_less_dicey_setlocale_r(pTHX_ const int category, const char * locale)
1900 const char * retval;
1902 PERL_ARGS_ASSERT_LESS_DICEY_SETLOCALE_R;
1904 STDIZED_SETLOCALE_LOCK;
1906 retval = save_to_buffer(stdized_setlocale(category, locale),
1907 &PL_less_dicey_locale_buf,
1908 &PL_less_dicey_locale_bufsize);
1910 STDIZED_SETLOCALE_UNLOCK;
1915 /*---------------------------------------------------------------------------*/
1917 # define bool_setlocale_r(cat, locale) \
1918 less_dicey_bool_setlocale_r(cat, locale)
1919 # define bool_setlocale_i(i, locale) \
1920 bool_setlocale_r(categories[i], locale)
1921 # define bool_setlocale_c(cat, locale) bool_setlocale_r(cat, locale)
1924 S_less_dicey_bool_setlocale_r(pTHX_ const int cat, const char * locale)
1928 PERL_ARGS_ASSERT_LESS_DICEY_BOOL_SETLOCALE_R;
1930 /* Unlikely, but potentially possible that another thread could zap the
1931 * buffer from true to false or vice-versa, so need to lock here */
1932 POSIX_SETLOCALE_LOCK;
1933 retval = cBOOL(posix_setlocale(cat, locale));
1934 POSIX_SETLOCALE_UNLOCK;
1939 /*---------------------------------------------------------------------------*/
1941 # define void_setlocale_r_with_caller(cat, locale, file, line) \
1943 if (! bool_setlocale_r(cat, locale)) \
1944 setlocale_failure_panic_via_i(get_category_index(cat), \
1945 NULL, locale, __LINE__, 0, \
1949 # define void_setlocale_c_with_caller(cat, locale, file, line) \
1950 void_setlocale_r_with_caller(cat, locale, file, line)
1952 # define void_setlocale_i_with_caller(i, locale, file, line) \
1953 void_setlocale_r_with_caller(categories[i], locale, file, line)
1955 # define void_setlocale_r(cat, locale) \
1956 void_setlocale_r_with_caller(cat, locale, __FILE__, __LINE__)
1957 # define void_setlocale_c(cat, locale) void_setlocale_r(cat, locale)
1958 # define void_setlocale_i(i, locale) void_setlocale_r(categories[i], locale)
1960 /*===========================================================================*/
1962 #elif defined(USE_POSIX_2008_LOCALE)
1964 # error This code assumes that LC_ALL is available on a system modern enough to have POSIX 2008
1967 /* Here, there is a completely different API to get thread-safe locales. We
1968 * emulate the setlocale() API with our own function(s). setlocale categories,
1969 * like LC_NUMERIC, are not valid here for the POSIX 2008 API. Instead, there
1970 * are equivalents, like LC_NUMERIC_MASK, which we use instead, which we find
1971 * by table lookup. */
1973 # if defined(__GLIBC__) && defined(USE_LOCALE_MESSAGES)
1974 /* https://sourceware.org/bugzilla/show_bug.cgi?id=24936 */
1975 # define HAS_GLIBC_LC_MESSAGES_BUG
1976 # include <libintl.h>
1979 # define querylocale_i(i) querylocale_2008_i(i, __LINE__)
1980 # define querylocale_c(cat) querylocale_i(cat##_INDEX_)
1981 # define querylocale_r(cat) querylocale_i(get_category_index(cat))
1984 S_querylocale_2008_i(pTHX_ const locale_category_index index,
1985 const line_t caller_line)
1987 PERL_ARGS_ASSERT_QUERYLOCALE_2008_I;
1988 assert(index <= LC_ALL_INDEX_);
1990 /* This function returns the name of the locale category given by the input
1991 * 'index' into our parallel tables of them.
1993 * POSIX 2008, for some sick reason, chose not to provide a method to find
1994 * the category name of a locale, disregarding a basic linguistic tenet
1995 * that for any object, people will create a name for it. (The next
1996 * version of the POSIX standard is proposed to fix this.) Some vendors
1997 * have created a querylocale() function to do this in the meantime. On
1998 * systems without querylocale(), we have to keep track of what the locale
1999 * has been set to, so that we can return its name so as to emulate
2000 * setlocale(). There are potential problems with this:
2002 * 1) We don't know what calling newlocale() with the locale argument ""
2003 * actually does. It gets its values from the program's environment.
2004 * find_locale_from_environment() is used to work around this. But it
2005 * isn't fool-proof. See the comments for that function for details.
2006 * 2) It's possible for C code in some library to change the locale
2007 * without us knowing it, and thus our records become wrong;
2008 * querylocale() would catch this. But as of September 2017, there
2009 * are no occurrences in CPAN of uselocale(). Some libraries do use
2010 * setlocale(), but that changes the global locale, and threads using
2011 * per-thread locales will just ignore those changes.
2012 * 3) Many systems have multiple names for the same locale. Generally,
2013 * there is an underlying base name, with aliases that evaluate to it.
2014 * On some systems, if you set the locale to an alias, and then
2015 * retrieve the name, you get the alias as expected; but on others you
2016 * get the base name, not the alias you used. And sometimes the
2017 * charade is incomplete. See
2018 * https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269375.
2020 * The code is structured so that the returned locale name when the
2021 * locale is changed is whatever the result of querylocale() on the
2022 * new locale is. This effectively gives the result the system
2023 * expects. Without querylocale, the name returned is always the
2024 * input name. Theoretically this could cause problems, but khw knows
2025 * of none so far, but mentions it here in case you are trying to
2026 * debug something. (This could be worked around by messing with the
2027 * global locale temporarily, using setlocale() to get the base name;
2028 * but that could cause a race. The comments for
2029 * find_locale_from_environment() give details on the potential race.)
2032 const locale_t cur_obj = uselocale((locale_t) 0);
2033 const char * retval;
2035 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "querylocale_2008_i(%s) on %p;"
2036 " called from %" LINE_Tf "\n",
2037 category_names[index], cur_obj,
2040 if (UNLIKELY(cur_obj == LC_GLOBAL_LOCALE)) {
2042 /* Even on platforms that have querylocale(), it is unclear if they
2043 * work in the global locale, and we have the means to get the correct
2044 * answer anyway. khw is unsure this situation even comes up these
2045 * days, hence the branch prediction */
2046 POSIX_SETLOCALE_LOCK;
2047 retval = mortalized_pv_copy(posix_setlocale(categories[index], NULL));
2048 POSIX_SETLOCALE_UNLOCK;
2051 /* Here we have handled the case of the the current locale being the global
2052 * one. Below is the 'else' case of that. There are two different
2053 * implementations, depending on USE_PL_CURLOCALES */
2055 # ifdef USE_PL_CURLOCALES
2059 /* PL_curlocales[] is kept up-to-date for all categories except LC_ALL,
2060 * which may have been invalidated by setting it to NULL, and if so,
2061 * should now be calculated. (The called function updates that
2063 if (index == LC_ALL_INDEX_ && PL_curlocales[LC_ALL_INDEX_] == NULL) {
2064 calculate_LC_ALL_string((const char **) &PL_curlocales,
2070 if (cur_obj == PL_C_locale_obj) {
2072 /* If the current locale object is the C object, then the answer is
2073 * "C" or POSIX, regardless of the category. Handling this
2074 * reasonably likely case specially shortcuts extra effort, and
2075 * hides some bugs from us in OS's that alias other locales to C,
2076 * but do so incompletely. If our records say it is POSIX, use
2077 * that; otherwise use C. See
2078 * https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269375 */
2079 retval = mortalized_pv_copy((strEQ(PL_curlocales[index], "POSIX"))
2084 retval = mortalized_pv_copy(PL_curlocales[index]);
2090 /* Below is the implementation of the 'else' clause which handles the case
2091 * of the current locale not being the global one on platforms where
2092 * USE_PL_CURLOCALES is NOT in effect. That means the system must have
2093 * some form of querylocale. But these have varying characteristics, so
2094 * first create some #defines to make the actual 'else' clause uniform.
2096 * First, glibc has a function that implements querylocale(), but is called
2097 * something else, and takes the category number; the others take the mask.
2099 # if defined(USE_QUERYLOCALE) && ( defined(_NL_LOCALE_NAME) \
2100 && defined(HAS_NL_LANGINFO_L))
2101 # define my_querylocale(index, cur_obj) \
2102 nl_langinfo_l(_NL_LOCALE_NAME(categories[index]), cur_obj)
2104 /* Experience so far shows it is thread-safe, as well as glibc's
2105 * nl_langinfo_l(), so unless overridden, mark it so */
2106 # ifdef NO_THREAD_SAFE_QUERYLOCALE
2107 # undef HAS_THREAD_SAFE_QUERYLOCALE
2109 # define HAS_THREAD_SAFE_QUERYLOCALE
2111 # else /* below, ! glibc */
2113 /* Otherwise, use the system's querylocale(). */
2114 # define my_querylocale(index, cur_obj) \
2115 querylocale(category_masks[index], cur_obj)
2117 /* There is no standard for this function, and khw has never seen
2118 * anything beyond minimal vendor documentation, lacking important
2119 * details. Experience has shown that some implementations have race
2120 * condiions, and their returns may not be thread safe. It would be
2121 * unreliable to test for complete thread safety in Configure. What we
2122 * do instead is to assume that it is thread-safe, unless overriden by,
2123 * say, a hints file specifying
2124 * -Accflags='-DNO_THREAD_SAFE_QUERYLOCALE */
2125 # ifdef NO_THREAD_SAFE_QUERYLOCALE
2126 # undef HAS_THREAD_SAFE_QUERYLOCALE
2128 # define HAS_THREAD_SAFE_QUERYLOCALE
2132 /* Here, we have set up enough information to know if this querylocale()
2133 * is thread-safe, or needs to use a mutex */
2134 # ifdef HAS_THREAD_SAFE_QUERYLOCALE
2135 # define QUERYLOCALE_LOCK
2136 # define QUERYLOCALE_UNLOCK
2138 # define QUERYLOCALE_LOCK gwLOCALE_LOCK
2139 # define QUERYLOCALE_UNLOCK gwLOCALE_UNLOCK
2142 /* Finally, everything is ready, so here is the 'else' clause to implement
2143 * the case of the current locale not being the global one on systems that
2144 * have some form of querylocale(). (POSIX will presumably eventually
2145 * publish their next version in their pipeline, which will define a
2146 * precisely specified querylocale equivalent, and there can be a new
2147 * #ifdef to use it without having to guess at its characteristics) */
2150 /* We don't keep records when there is querylocale(), so as to avoid the
2151 * pitfalls mentioned at the beginning of this function.
2153 * That means LC_ALL has to be calculated from all its constituent
2154 * categories each time, since the querylocale() forms on many (if not
2155 * all) platforms only work on individual categories */
2156 if (index == LC_ALL_INDEX_) {
2157 retval = calculate_LC_ALL_string(NULL, INTERNAL_FORMAT,
2164 retval = savepv(my_querylocale(index, cur_obj));
2167 /* querylocale() may conflate the C locale with something that
2168 * isn't exactly the same. See for example
2169 * https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269375
2170 * We know that if the locale object is the C one, we
2171 * are in the C locale, which may go by the name POSIX, as both, by
2172 * definition, are equivalent. But we consider any other name
2173 * spurious, so override with "C". As in the PL_CURLOCALES case
2174 * above, this hides those glitches, for the most part, from the
2175 * rest of our code. (The code is ordered this way so that if the
2176 * system distinugishes "C" from "POSIX", we do too.) */
2177 if (cur_obj == PL_C_locale_obj && ! isNAME_C_OR_POSIX(retval)) {
2179 retval = savepv("C");
2186 # undef QUERYLOCALE_LOCK
2187 # undef QUERYLOCALE_UNLOCK
2190 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2191 "querylocale_2008_i(%s) returning '%s'\n",
2192 category_names[index], retval));
2193 assert(strNE(retval, ""));
2197 /*---------------------------------------------------------------------------*/
2199 # define bool_setlocale_i(i, locale) \
2200 bool_setlocale_2008_i(i, locale, __LINE__)
2201 # define bool_setlocale_c(cat, locale) \
2202 bool_setlocale_i(cat##_INDEX_, locale)
2203 # define bool_setlocale_r(cat, locale) \
2204 bool_setlocale_i(get_category_index(cat), locale)
2206 /* If this doesn't exist on this platform, make it a no-op (to save #ifdefs) */
2207 # ifndef update_PL_curlocales_i
2208 # define update_PL_curlocales_i(index, new_locale, caller_line)
2212 S_bool_setlocale_2008_i(pTHX_
2214 /* Our internal index of the 'category' setlocale is called with */
2215 const locale_category_index index,
2216 const char * new_locale, /* The locale to set the category to */
2217 const line_t caller_line /* Called from this line number */
2220 PERL_ARGS_ASSERT_BOOL_SETLOCALE_2008_I;
2221 assert(index <= LC_ALL_INDEX_);
2223 /* This function effectively performs a setlocale() on just the current
2224 * thread; thus it is thread-safe. It does this by using the POSIX 2008
2225 * locale functions to emulate the behavior of setlocale(). Similar to
2226 * regular setlocale(), the return from this function points to memory that
2227 * can be overwritten by other system calls, so needs to be copied
2228 * immediately if you need to retain it. The difference here is that
2229 * system calls besides another setlocale() can overwrite it.
2231 * By doing this, most locale-sensitive functions become thread-safe. The
2232 * exceptions are mostly those that return a pointer to static memory.
2235 int mask = category_masks[index];
2236 const locale_t entry_obj = uselocale((locale_t) 0);
2237 const char * locale_on_entry = querylocale_i(index);
2239 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2240 "bool_setlocale_2008_i: input=%d (%s), mask=0x%x,"
2241 " new locale=\"%s\", current locale=\"%s\","
2242 " index=%d, entry object=%p;"
2243 " called from %" LINE_Tf "\n",
2244 categories[index], category_names[index], mask,
2245 ((new_locale == NULL) ? "(nil)" : new_locale),
2246 locale_on_entry, index, entry_obj, caller_line));
2248 /* Here, trying to change the locale, but it is a no-op if the new boss is
2249 * the same as the old boss. Except this routine is called when converting
2250 * from the global locale, so in that case we will create a per-thread
2251 * locale below (with the current values). It also seemed that newlocale()
2252 * could free up the basis locale memory if we called it with the new and
2253 * old being the same, but khw now thinks that this was due to some other
2254 * bug, since fixed, as there are other places where newlocale() gets
2255 * similarly called without problems. */
2256 if ( entry_obj != LC_GLOBAL_LOCALE
2258 && strEQ(new_locale, locale_on_entry))
2260 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2261 "bool_setlocale_2008_i: no-op to change to"
2262 " what it already was\n"));
2266 # ifndef USE_QUERYLOCALE
2268 /* Without a querylocale() mechanism, we have to figure out ourselves what
2269 * happens with setting a locale to "" */
2271 if (strEQ(new_locale, "")) {
2272 new_locale = find_locale_from_environment(index);
2280 # ifdef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
2282 const bool need_loop = false;
2286 bool need_loop = false;
2287 const char * new_locales[LC_ALL_INDEX_] = { NULL };
2289 /* If we're going to have to parse the LC_ALL string, might as well do it
2290 * now before we have made changes that we would have to back out of if the
2292 if (index == LC_ALL_INDEX_) {
2293 switch (parse_LC_ALL_string(new_locale,
2294 (const char **) &new_locales,
2295 override_if_ignored,
2296 false, /* Return only [0] if suffices */
2297 false, /* Don't panic on error */
2308 case only_element_0:
2309 SAVEFREEPV(new_locales[0]);
2310 new_locale = new_locales[0];
2321 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
2323 /* For this bug, if the LC_MESSAGES locale changes, we have to do an
2324 * expensive workaround. Save the current value so we can later determine
2326 const char * old_messages_locale = NULL;
2327 if ( (index == LC_MESSAGES_INDEX_ || index == LC_ALL_INDEX_)
2328 && LIKELY(PL_phase != PERL_PHASE_CONSTRUCT))
2330 old_messages_locale = querylocale_c(LC_MESSAGES);
2335 assert(PL_C_locale_obj);
2337 /* Now ready to switch to the input 'new_locale' */
2339 /* Switching locales generally entails freeing the current one's space (at
2340 * the C library's discretion), hence we can't be using that locale at the
2341 * time of the switch (this wasn't obvious to khw from the man pages). So
2342 * switch to a known locale object that we don't otherwise mess with. */
2343 if (! uselocale(PL_C_locale_obj)) {
2345 /* Not being able to change to the C locale is severe; don't keep
2347 setlocale_failure_panic_i(index, locale_on_entry, "C",
2348 __LINE__, caller_line);
2349 NOT_REACHED; /* NOTREACHED */
2352 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2353 "bool_setlocale_2008_i: now using C"
2354 " object=%p\n", PL_C_locale_obj));
2356 /* These two objects are special:
2357 * LC_GLOBAL_LOCALE because it is undefined behavior to call
2358 * newlocale() with it as a parameter.
2359 * PL_C_locale_obj because newlocale() generally destroys its locale
2360 * object parameter when it succeeds; and we don't
2361 * want that happening to this immutable object.
2362 * Copies will be made for them to use instead if we get so far as to call
2364 bool entry_obj_is_special = ( entry_obj == LC_GLOBAL_LOCALE
2365 || entry_obj == PL_C_locale_obj);
2368 /* PL_C_locale_obj is LC_ALL set to the C locale. If this call is to
2369 * switch to LC_ALL => C, simply use that object. But in fact, we already
2370 * have switched to it just above, in preparation for the general case.
2371 * Since we're already there, no need to do further switching. */
2372 if (mask == LC_ALL_MASK && isNAME_C_OR_POSIX(new_locale)) {
2373 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2374 "bool_setlocale_2008_i: will stay in C"
2376 new_obj = PL_C_locale_obj;
2378 /* 'entry_obj' is now dangling, of no further use to anyone (unless it
2379 * is one of the special ones). Free it to avoid a leak */
2380 if (! entry_obj_is_special) {
2381 freelocale(entry_obj);
2384 update_PL_curlocales_i(index, new_locale, caller_line);
2386 else { /* Here is the general case, not to LC_ALL => C */
2388 /* The newlocale() call(s) below take a basis object to build upon to
2389 * create the changed locale, trashing it iff successful.
2391 * For the objects that are not to be modified by this function, we
2392 * create a duplicate that gets trashed instead.
2394 * Also if we will have to loop doing multiple newlocale()s, there is a
2395 * chance we will succeed for the first few, and then fail, having to
2396 * back out. We need to duplicate 'entry_obj' in this case as well, so
2397 * it remains valid as something to back out to. */
2398 locale_t basis_obj = entry_obj;
2400 if (entry_obj_is_special || need_loop) {
2401 basis_obj = duplocale(basis_obj);
2403 locale_panic_via_("duplocale failed", __FILE__, caller_line);
2404 NOT_REACHED; /* NOTREACHED */
2407 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2408 "bool_setlocale_2008_i created %p by"
2409 " duping the input\n", basis_obj));
2412 # define DEBUG_NEW_OBJECT_CREATED(category, locale, new, old, caller_line) \
2413 DEBUG_Lv(PerlIO_printf(Perl_debug_log, \
2414 "bool_setlocale_2008_i(%s, %s): created %p" \
2415 " while freeing %p; called from %" LINE_Tf \
2416 " via %" LINE_Tf "\n", \
2417 category, locale, new, old, \
2418 caller_line, __LINE__))
2419 # define DEBUG_NEW_OBJECT_FAILED(category, locale, basis_obj) \
2420 DEBUG_L(PerlIO_printf(Perl_debug_log, \
2421 "bool_setlocale_2008_i: creating new object" \
2422 " for (%s '%s') from %p failed; called from %" \
2423 LINE_Tf " via %" LINE_Tf "\n", \
2424 category, locale, basis_obj, \
2425 caller_line, __LINE__));
2427 /* Ready to create a new locale by modification of the existing one.
2429 * NOTE: This code may incorrectly show up as a leak under the address
2430 * sanitizer. We do not free this object under normal teardown, however
2431 * you can set PERL_DESTRUCT_LEVEL=2 to cause it to be freed.
2434 # ifdef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
2436 /* Some platforms have a newlocale() that can handle disparate LC_ALL
2437 * input, so on these a single call to newlocale() always works */
2440 /* If a single call to newlocale() will do */
2446 new_obj = newlocale(mask,
2447 override_ignored_category(index, new_locale),
2450 DEBUG_NEW_OBJECT_FAILED(category_names[index], new_locale,
2453 /* Since the call failed, it didn't trash 'basis_obj', which is
2454 * a dup for these objects, and hence would leak if we don't
2455 * free it. XXX However, something is seriously wrong if we
2456 * can't switch to C or the global locale, so maybe should
2458 if (entry_obj_is_special) {
2459 freelocale(basis_obj);
2462 goto must_restore_state;
2465 DEBUG_NEW_OBJECT_CREATED(category_names[index], new_locale,
2466 new_obj, basis_obj, caller_line);
2468 update_PL_curlocales_i(index, new_locale, caller_line);
2471 # ifndef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
2473 else { /* Need multiple newlocale() calls */
2475 /* Loop through the individual categories, setting the locale of
2476 * each to the corresponding name previously populated into
2477 * newlocales[]. Each iteration builds on the previous one, adding
2478 * its category to what's already been calculated, and taking as a
2479 * basis for what's been calculated 'basis_obj', which is updated
2480 * each iteration to be the result of the previous one. Upon
2481 * success, newlocale() trashes the 'basis_obj' parameter to it.
2482 * If any iteration fails, we immediately give up, restore the
2483 * locale to what it was at the time this function was called
2484 * (saved in 'entry_obj'), and return failure. */
2486 /* Loop, using the previous iteration's result as the basis for the
2487 * next one. (The first time we effectively use the locale in
2488 * force upon entry to this function.) */
2489 for_all_individual_category_indexes(i) {
2490 new_obj = newlocale(category_masks[i],
2494 DEBUG_NEW_OBJECT_CREATED(category_names[i],
2498 basis_obj = new_obj;
2502 /* Failed. Likely this is because the proposed new locale
2503 * isn't valid on this system. */
2505 DEBUG_NEW_OBJECT_FAILED(category_names[i],
2509 /* newlocale() didn't trash this, since the function call
2511 freelocale(basis_obj);
2513 for_all_individual_category_indexes(j) {
2514 Safefree(new_locales[j]);
2517 goto must_restore_state;
2520 /* Success for all categories. */
2521 for_all_individual_category_indexes(i) {
2522 update_PL_curlocales_i(i, new_locales[i], caller_line);
2523 Safefree(new_locales[i]);
2526 /* We dup'd entry_obj in case we had to fall back to it. The
2527 * newlocale() above destroyed the dup when it first succeeded, but
2528 * entry_obj itself is left dangling, so free it */
2529 if (! entry_obj_is_special) {
2530 freelocale(entry_obj);
2534 # endif /* End of newlocale can't handle disparate LC_ALL input */
2538 # undef DEBUG_NEW_OBJECT_CREATED
2539 # undef DEBUG_NEW_OBJECT_FAILED
2541 /* Here, successfully created an object representing the desired locale;
2542 * now switch into it */
2543 if (! uselocale(new_obj)) {
2544 freelocale(new_obj);
2545 locale_panic_(Perl_form(aTHX_ "(called from %" LINE_Tf "):"
2546 " bool_setlocale_2008_i: switching"
2547 " into new locale failed",
2551 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2552 "bool_setlocale_2008_i: now using %p\n", new_obj));
2554 # ifdef MULTIPLICITY /* Unlikely, but POSIX 2008 functions could be
2555 Configured to be used on unthreaded perls, in which
2556 case this object doesn't exist */
2558 if (DEBUG_Lv_TEST) {
2559 if (PL_cur_locale_obj != new_obj) {
2560 PerlIO_printf(Perl_debug_log,
2561 "bool_setlocale_2008_i: PL_cur_locale_obj"
2562 " was %p, now is %p\n",
2563 PL_cur_locale_obj, new_obj);
2567 /* Update the current object */
2568 PL_cur_locale_obj = new_obj;
2571 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
2573 /* Invalidate the glibc cache of loaded translations if the locale has
2574 * changed, see [perl #134264] and
2575 * https://sourceware.org/bugzilla/show_bug.cgi?id=24936 */
2576 if (old_messages_locale) {
2577 if (strNE(old_messages_locale, querylocale_c(LC_MESSAGES))) {
2578 textdomain(textdomain(NULL));
2588 /* We earlier switched to the LC_ALL => C locale in anticipation of it
2589 * succeeding, Now have to switch back to the state upon entry. */
2590 if (! uselocale(entry_obj)) {
2591 setlocale_failure_panic_i(index, "switching back to",
2592 locale_on_entry, __LINE__, caller_line);
2598 /*---------------------------------------------------------------------------*/
2600 # define void_setlocale_i_with_caller(i, locale, file, line) \
2602 if (! bool_setlocale_i(i, locale)) \
2603 setlocale_failure_panic_via_i(i, NULL, locale, __LINE__, 0, \
2607 # define void_setlocale_r_with_caller(cat, locale, file, line) \
2608 void_setlocale_i_with_caller(get_category_index(cat), locale, \
2611 # define void_setlocale_c_with_caller(cat, locale, file, line) \
2612 void_setlocale_i_with_caller(cat##_INDEX_, locale, file, line)
2614 # define void_setlocale_i(i, locale) \
2615 void_setlocale_i_with_caller(i, locale, __FILE__, __LINE__)
2616 # define void_setlocale_c(cat, locale) \
2617 void_setlocale_i(cat##_INDEX_, locale)
2618 # define void_setlocale_r(cat, locale) \
2619 void_setlocale_i(get_category_index(cat), locale)
2621 /*===========================================================================*/
2624 # error Unexpected Configuration
2625 #endif /* End of the various implementations of the setlocale and
2626 querylocale macros used in the remainder of this program */
2628 /* query_nominal_locale_i() is used when the caller needs the locale that an
2629 * external caller would be expecting, and not what we're secretly using
2630 * behind the scenes. It deliberately doesn't handle LC_ALL; use
2631 * calculate_LC_ALL_string() for that. */
2632 #ifdef USE_LOCALE_NUMERIC
2633 # define query_nominal_locale_i(i) \
2634 (__ASSERT_(i != LC_ALL_INDEX_) \
2635 ((i == LC_NUMERIC_INDEX_) ? PL_numeric_name : querylocale_i(i)))
2637 # define query_nominal_locale_i(i) \
2638 (__ASSERT_(i != LC_ALL_INDEX_) querylocale_i(i))
2641 #ifdef USE_PL_CURLOCALES
2644 S_update_PL_curlocales_i(pTHX_
2645 const locale_category_index index,
2646 const char * new_locale,
2647 const line_t caller_line)
2649 /* Update PL_curlocales[], which is parallel to the other ones indexed by
2650 * our mapping of libc category number to our internal equivalents. */
2652 PERL_ARGS_ASSERT_UPDATE_PL_CURLOCALES_I;
2653 assert(index <= LC_ALL_INDEX_);
2655 if (index == LC_ALL_INDEX_) {
2657 /* For LC_ALL, we change all individual categories to correspond,
2658 * including the LC_ALL element */
2659 for (unsigned int i = 0; i <= LC_ALL_INDEX_; i++) {
2660 Safefree(PL_curlocales[i]);
2661 PL_curlocales[i] = NULL;
2664 switch (parse_LC_ALL_string(new_locale,
2665 (const char **) &PL_curlocales,
2666 check_that_overridden, /* things should
2670 true, /* Always fill array */
2671 true, /* Panic if fails, as to get here
2672 it earlier had to have succeeded
2678 case only_element_0:
2679 locale_panic_via_("Unexpected return from parse_LC_ALL_string",
2680 __FILE__, caller_line);
2683 /* parse_LC_ALL_string() has already filled PL_curlocales properly,
2684 * except for the LC_ALL element, which should be set to
2686 PL_curlocales[LC_ALL_INDEX_] = savepv(new_locale);
2689 else { /* Not LC_ALL */
2691 /* Update the single category's record */
2692 Safefree(PL_curlocales[index]);
2693 PL_curlocales[index] = savepv(new_locale);
2695 /* Invalidate LC_ALL */
2696 Safefree(PL_curlocales[LC_ALL_INDEX_]);
2697 PL_curlocales[LC_ALL_INDEX_] = NULL;
2701 # endif /* Need PL_curlocales[] */
2703 /*===========================================================================*/
2705 #if defined(USE_LOCALE)
2707 /* This paradigm is needed in several places in the function below. We have to
2708 * substitute the nominal locale for LC_NUMERIC when returning a value for
2709 * external consumption */
2710 # ifndef USE_LOCALE_NUMERIC
2711 # define ENTRY(i, array, format) array[i]
2713 # define ENTRY(i, array, format) \
2714 (UNLIKELY( format == EXTERNAL_FORMAT_FOR_QUERY \
2715 && i == LC_NUMERIC_INDEX_) \
2722 S_calculate_LC_ALL_string(pTHX_ const char ** category_locales_list,
2723 const calc_LC_ALL_format format,
2724 const calc_LC_ALL_return returning,
2725 const line_t caller_line)
2727 PERL_ARGS_ASSERT_CALCULATE_LC_ALL_STRING;
2729 /* NOTE: On Configurations that have PL_curlocales[], this function has the
2730 * side effect of updating the LC_ALL_INDEX_ element with its result.
2732 * This function calculates a string that defines the locale(s) LC_ALL is
2733 * set to, in either:
2734 * 1) Our internal format if 'format' is set to INTERNAL_FORMAT.
2735 * 2) The external format returned by Perl_setlocale() if 'format' is set
2736 * to EXTERNAL_FORMAT_FOR_QUERY or EXTERNAL_FORMAT_FOR_SET.
2738 * These two are distinguished by:
2739 * a) EXTERNAL_FORMAT_FOR_SET returns the actual locale currently in
2741 * b) EXTERNAL_FORMAT_FOR_QUERY returns the nominal locale.
2742 * Currently this can differ only from the actual locale in the
2743 * LC_NUMERIC category when it is set to a locale whose radix is
2744 * not a dot. (The actual locale is kept as a dot to accommodate
2745 * the large corpus of XS code that expects it to be that;
2746 * switched to a non-dot temporarily during certain operations
2747 * that require the actual radix.)
2749 * In both 1) and 2), LC_ALL's values are passed to this function by
2750 * 'category_locales_list' which is either:
2751 * 1) a pointer to an array of strings with up-to-date values of all the
2752 * individual categories; or
2753 * 2) NULL, to indicate to use querylocale_i() to get each individual
2756 * The caller sets 'returning' to
2757 * WANT_TEMP_PV the function returns the calculated string
2758 * as a mortalized temporary, so the caller
2759 * doesn't have to worry about it being
2760 * per-thread, nor needs to arrange for its
2762 * WANT_VOID NULL is returned. This is used when the
2763 * function is being called only for its side
2764 * effect of updating
2765 * PL_curlocales[LC_ALL_INDEX_]
2767 * querylocale(), on systems that have it, doesn't tend to work for LC_ALL.
2768 * So we have to construct the answer ourselves based on the passed in
2771 * If all individual categories are the same locale, we can just set LC_ALL
2772 * to that locale. But if not, we have to create an aggregation of all the
2773 * categories on the system. Platforms differ as to the syntax they use
2774 * for these non-uniform locales for LC_ALL. Some, like glibc and Windows,
2775 * use an unordered series of name=value pairs, like
2776 * LC_NUMERIC=C;LC_TIME=en_US.UTF-8;...
2777 * to specify LC_ALL; others, like *BSD, use a positional notation with a
2778 * delimitter, typically a single '/' character:
2781 * When the external format is desired, this function returns whatever the
2782 * system expects. The internal format is always name=value pairs.
2784 * For systems that have categories we don't know about, the algorithm
2785 * below won't know about those missing categories, leading to potential
2786 * bugs for code that looks at them. If there is an environment variable
2787 * that sets that category, we won't know to look for it, and so our use of
2788 * LANG or "C" improperly overrides it. On the other hand, if we don't do
2789 * what is done here, and there is no environment variable, the category's
2790 * locale should be set to LANG or "C". So there is no good solution. khw
2791 * thinks the best is to make sure we have a complete list of possible
2792 * categories, adding new ones as they show up on obscure platforms.
2795 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2796 "Entering calculate_LC_ALL_string(%s);"
2797 " called from %" LINE_Tf "\n",
2798 ((format == EXTERNAL_FORMAT_FOR_QUERY)
2799 ? "EXTERNAL_FORMAT_FOR_QUERY"
2800 : ((format == EXTERNAL_FORMAT_FOR_SET)
2801 ? "EXTERNAL_FORMAT_FOR_SET"
2802 : "INTERNAL_FORMAT")),
2805 bool input_list_was_NULL = (category_locales_list == NULL);
2807 /* If there was no input category list, construct a temporary one
2809 const char * my_category_locales_list[LC_ALL_INDEX_];
2810 const char ** locales_list = category_locales_list;
2811 if (locales_list == NULL) {
2812 locales_list = my_category_locales_list;
2814 if (format == EXTERNAL_FORMAT_FOR_QUERY) {
2815 for_all_individual_category_indexes(i) {
2816 locales_list[i] = query_nominal_locale_i(i);
2820 for_all_individual_category_indexes(i) {
2821 locales_list[i] = querylocale_i(i);
2826 /* While we are calculating LC_ALL, we see if every category's locale is
2827 * the same as every other's or not. */
2828 # ifndef HAS_IGNORED_LOCALE_CATEGORIES_
2830 /* When we pay attention to all categories, we assume they are all the same
2831 * until proven different */
2832 bool disparate = false;
2836 /* But if there are ignored categories, those will be set to "C", so try an
2837 * arbitrary category, and if it isn't C, we know immediately that the
2838 * locales are disparate. (The #if conditionals are to handle the case
2839 * where LC_NUMERIC_INDEX_ is 0. We don't want to use LC_NUMERIC to
2840 * compare, as that may be different between external and internal forms.)
2842 # if ! defined(USE_LOCALE_NUMERIC)
2844 bool disparate = ! isNAME_C_OR_POSIX(locales_list[0]);
2846 # elif LC_NUMERIC_INDEX_ != 0
2848 bool disparate = ! isNAME_C_OR_POSIX(locales_list[0]);
2852 /* Would need revision to handle the very unlikely case where only a single
2853 * category, LC_NUMERIC, is defined */
2854 assert(LOCALE_CATEGORIES_COUNT_ > 0);
2856 bool disparate = ! isNAME_C_OR_POSIX(locales_list[1]);
2861 /* Calculate the needed size for the string listing the individual locales.
2862 * Initialize with values known at compile time. */
2864 const char *separator;
2866 # ifdef PERL_LC_ALL_USES_NAME_VALUE_PAIRS /* Positional formatted LC_ALL */
2867 PERL_UNUSED_ARG(format);
2870 if (format != INTERNAL_FORMAT) {
2872 /* Here, we will be using positional notation. it includes n-1
2874 total_len = ( LOCALE_CATEGORIES_COUNT_ - 1)
2875 * STRLENs(PERL_LC_ALL_SEPARATOR)
2876 + 1; /* And a trailing NUL */
2877 separator = PERL_LC_ALL_SEPARATOR;
2884 /* name=value output is always used in internal format, and when
2885 * positional isn't available on the platform. */
2886 total_len = lc_all_boiler_plate_length;
2890 /* The total length then is just the sum of the above boiler-plate plus the
2891 * total strlen()s of the locale name of each individual category. */
2892 for_all_individual_category_indexes(i) {
2893 const char * entry = ENTRY(i, locales_list, format);
2895 total_len += strlen(entry);
2896 if (! disparate && strNE(entry, locales_list[0])) {
2901 bool free_if_void_return = false;
2902 const char * retval;
2904 /* If all categories have the same locale, we already know the answer */
2906 retval = locales_list[0];
2908 /* If a temporary is wanted for the return, and we had to create
2909 * the input list ourselves, we created it into such a temporary,
2910 * so no further work is needed; but otherwise, make a mortal copy
2911 * of this passed-in list element */
2912 if (returning == WANT_TEMP_PV && ! input_list_was_NULL) {
2913 retval = savepv(retval);
2917 /* In all cases here, there's nothing we create that needs to be
2918 * freed, so leave 'free_if_void_return' set to the default
2921 else { /* Here, not all categories have the same locale */
2925 Newx(constructed, total_len, char);
2927 /* If returning the new memory, it must be set up to be freed
2928 * later; otherwise at the end of this function */
2929 if (returning == WANT_TEMP_PV) {
2930 SAVEFREEPV(constructed);
2933 free_if_void_return = true;
2936 constructed[0] = '\0';
2938 /* Loop through all the categories */
2939 for_all_individual_category_indexes(j) {
2941 /* Add a separator, except before the first one */
2943 my_strlcat(constructed, separator, total_len);
2950 # ifndef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
2952 if (UNLIKELY(format != INTERNAL_FORMAT)) {
2954 /* In positional notation 'j' means the position, and we have
2955 * to convert to the index 'i' */
2956 i = map_LC_ALL_position_to_index[j];
2958 entry = ENTRY(i, locales_list, format);
2959 needed_len = my_strlcat(constructed, entry, total_len);
2965 /* Below, we are to use name=value notation, either because
2966 * that's what the platform uses, or because this is the
2967 * internal format, which uses that notation regardless of the
2970 entry = ENTRY(i, locales_list, format);
2972 /* "name=locale;" */
2973 my_strlcat(constructed, category_names[i], total_len);
2974 my_strlcat(constructed, "=", total_len);
2975 needed_len = my_strlcat(constructed, entry, total_len);
2978 if (LIKELY(needed_len <= total_len)) {
2982 /* If would have overflowed, panic */
2983 locale_panic_via_(Perl_form(aTHX_
2984 "Internal length calculation wrong.\n"
2985 "\"%s\" was not entirely added to"
2986 " \"%.*s\"; needed=%zu, had=%zu",
2987 entry, (int) total_len,
2989 needed_len, total_len),
2992 } /* End of loop through the categories */
2994 retval = constructed;
2995 } /* End of the categories' locales are displarate */
2997 # if defined(USE_PL_CURLOCALES) && defined(LC_ALL)
2999 if (format == INTERNAL_FORMAT) {
3001 /* PL_curlocales[LC_ALL_INDEX_] is updated as a side-effect of this
3002 * function for internal format. */
3003 Safefree(PL_curlocales[LC_ALL_INDEX_]);
3004 PL_curlocales[LC_ALL_INDEX_] = savepv(retval);
3009 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3010 "calculate_LC_ALL_string calculated '%s'\n",
3013 if (returning == WANT_VOID) {
3014 if (free_if_void_return) {
3024 # if defined(WIN32) || ( defined(USE_POSIX_2008_LOCALE) \
3025 && ! defined(USE_QUERYLOCALE))
3028 S_find_locale_from_environment(pTHX_ const locale_category_index index)
3030 /* NB: This function may actually change the locale on Windows. It
3031 * currently is designed to be called only from setting the locale on
3032 * Windows, and POSIX 2008
3034 * This function returns the locale specified by the program's environment
3035 * for the category specified by our internal index number 'index'. It
3036 * therefore simulates:
3037 * setlocale(cat, "")
3038 * but, except for some cases in Windows, doesn't actually change the
3039 * locale; merely returns it.
3041 * The return need not be freed by the caller. This
3042 * promise relies on PerlEnv_getenv() returning a mortalized copy to us.
3044 * The simulation is needed only on certain platforms; otherwise, libc is
3045 * called with "" to get the actual value(s). The simulation is needed
3048 * 1) On Windows systems, the concept of the POSIX ordering of
3049 * environment variables is missing. To increase portability of
3050 * programs across platforms, the POSIX ordering is emulated on
3053 * 2) On POSIX 2008 systems without querylocale(), it is problematic
3054 * getting the results of the POSIX 2008 equivalent of
3056 * setlocale(category, "")
3058 * To ensure that we know exactly what those values are, we do the
3059 * setting ourselves, using the documented algorithm specified by the
3060 * POSIX standard (assuming the platform follows the Standard) rather
3061 * than use "" as the locale. This will lead to results that differ
3062 * from native behavior if the native behavior differs from the
3063 * Standard's documented value, but khw believes it is better to know
3064 * what's going on, even if different from native, than to just guess.
3066 * glibc systems differ from this standard in having a LANGUAGE
3067 * environment variable used for just LC_MESSAGES. This function does
3070 * Another option for the POSIX 2008 case would be, in a critical
3071 * section, to save the global locale's current value, and do a
3072 * straight setlocale(LC_ALL, ""). That would return our desired
3073 * values, destroying the global locale's, which we would then
3074 * restore. But that could cause races with any other thread that is
3075 * using the global locale and isn't using the mutex. And, the only
3076 * reason someone would have done that is because they are calling a
3077 * library function, like in gtk, that calls setlocale(), and which
3078 * can't be changed to use the mutex. That wouldn't be a problem if
3079 * this were to be done before any threads had switched, say during
3080 * perl construction time. But this code would still be needed for
3083 * The Windows and POSIX 2008 differ in that the ultimate fallback is "C"
3084 * in POSIX, and is the system default locale in Windows. To get that
3085 * system default value, we actually have to call setlocale() on Windows.
3088 const char * const lc_all = PerlEnv_getenv("LC_ALL");
3089 const char * locale_names[LC_ALL_INDEX_] = { NULL };
3091 /* Use any "LC_ALL" environment variable, as it overrides everything else.
3093 if (lc_all && strNE(lc_all, "")) {
3097 /* Here, no usable LC_ALL environment variable. We have to handle each
3098 * category separately. If all categories are desired, we loop through
3099 * them all. If only an individual category is desired, to avoid
3100 * duplicating logic, we use the same loop, but set up the limits so it is
3101 * only executed once, for that particular category. */
3102 locale_category_index lower, upper, offset;
3103 if (index == LC_ALL_INDEX_) {
3104 lower = (locale_category_index) 0;
3105 upper = (locale_category_index) ((int) LC_ALL_INDEX_ - 1);
3106 offset = (locale_category_index) 0;
3112 /* 'offset' is used so that the result of the single loop iteration is
3113 * stored into output[0] */
3117 /* When no LC_ALL environment variable, LANG is used as a default, but
3118 * overridden for individual categories that have corresponding environment
3119 * variables. If no LANG exists, the default is "C" on POSIX 2008, or the
3120 * system default for the category on Windows. */
3121 const char * env_lang = NULL;
3123 /* For each desired category, use any corresponding environment variable;
3124 * or the default if none such exists. */
3125 bool is_disparate = false; /* Assume is uniform until proven otherwise */
3126 for (unsigned i = lower; i <= upper; i++) {
3127 const char * const env_override = PerlEnv_getenv(category_names[i]);
3128 unsigned int j = i - offset;
3130 if (env_override && strNE(env_override, "")) {
3131 locale_names[j] = env_override;
3133 else { /* Here, no corresponding environment variable, see if LANG
3134 exists and is usable. Done this way to avoid fetching LANG
3135 unless it is actually needed */
3136 if (env_lang == NULL) {
3137 env_lang = PerlEnv_getenv("LANG");
3139 /* If not usable, set it to a non-NULL illegal value so won't
3140 * try to use it below */
3141 if (env_lang == NULL || strEQ(env_lang, "")) {
3142 env_lang = (const char *) 1;
3146 /* If a usable LANG exists, use it. */
3147 if (env_lang != NULL && env_lang != (const char *) 1) {
3148 locale_names[j] = env_lang;
3153 /* If no LANG, use the system default on Windows. */
3154 locale_names[j] = wrap_wsetlocale(categories[i], ".ACP");
3155 if (locale_names[j]) {
3156 SAVEFREEPV(locale_names[j]);
3160 { /* If nothing was found or worked, use C */
3161 locale_names[j] = "C";
3166 if (j > 0 && ! is_disparate && strNE(locale_names[0], locale_names[j]))
3168 is_disparate = true;
3171 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3172 "find_locale_from_environment i=%u, j=%u, name=%s,"
3173 " locale=%s, locale of 0th category=%s, disparate=%d\n",
3174 i, j, category_names[i],
3175 locale_names[j], locale_names[0], is_disparate));
3178 if (! is_disparate) {
3179 return locale_names[0];
3182 return calculate_LC_ALL_string(locale_names, INTERNAL_FORMAT,
3188 # if defined(DEBUGGING) || defined(USE_PERL_SWITCH_LOCALE_CONTEXT)
3191 S_get_LC_ALL_display(pTHX)
3193 return calculate_LC_ALL_string(NULL, INTERNAL_FORMAT,
3201 S_setlocale_failure_panic_via_i(pTHX_
3202 const locale_category_index cat_index,
3203 const char * current,
3204 const char * failed,
3205 const line_t proxy_caller_line,
3206 const line_t immediate_caller_line,
3207 const char * const higher_caller_file,
3208 const line_t higher_caller_line)
3210 PERL_ARGS_ASSERT_SETLOCALE_FAILURE_PANIC_VIA_I;
3212 /* Called to panic when a setlocale form unexpectedly failed for the
3213 * category determined by 'cat_index', and the locale that was in effect
3214 * (and likely still is) is 'current'. 'current' may be NULL, which causes
3215 * this function to query what it is.
3217 * The extra caller information is used for when a function acts as a
3218 * stand-in for another function, which a typical reader would more likely
3219 * think would be the caller
3221 * If a line number is 0, its stack (sort-of) frame is omitted; same if
3222 * it's the same line number as the next higher caller. */
3224 const int cat = categories[cat_index];
3225 const char * name = category_names[cat_index];
3229 if (current == NULL) {
3230 current = querylocale_i(cat_index);
3233 const char * proxy_text = "";
3234 if (proxy_caller_line != 0 && proxy_caller_line != immediate_caller_line)
3236 proxy_text = Perl_form(aTHX_ "\nCalled via %s: %" LINE_Tf,
3237 __FILE__, proxy_caller_line);
3239 if ( strNE(__FILE__, higher_caller_file)
3240 || ( immediate_caller_line != 0
3241 && immediate_caller_line != higher_caller_line))
3243 proxy_text = Perl_form(aTHX_ "%s\nCalled via %s: %" LINE_Tf,
3244 proxy_text, __FILE__,
3245 immediate_caller_line);
3248 /* 'false' in the get_displayable_string() calls makes it not think the
3249 * locale is UTF-8, so just dumps bytes. Actually figuring it out can be
3250 * too complicated for a panic situation. */
3251 const char * msg = Perl_form(aTHX_
3252 "Can't change locale for %s (%d) from '%s' to '%s'"
3255 get_displayable_string(current,
3256 current + strlen(current),
3258 get_displayable_string(failed,
3259 failed + strlen(failed),
3264 Perl_locale_panic(msg, __LINE__, higher_caller_file, higher_caller_line);
3265 NOT_REACHED; /* NOTREACHED */
3268 /* Any of these will allow us to find the RADIX */
3269 # if defined(USE_LOCALE_NUMERIC) && ( defined(HAS_SOME_LANGINFO) \
3270 || defined(HAS_LOCALECONV) \
3271 || defined(HAS_SNPRINTF))
3272 # define CAN_CALCULATE_RADIX
3274 # ifdef USE_LOCALE_NUMERIC
3277 S_new_numeric(pTHX_ const char *newnum, bool force)
3279 PERL_ARGS_ASSERT_NEW_NUMERIC;
3281 /* Called after each libc setlocale() or uselocale() call affecting
3282 * LC_NUMERIC, to tell core Perl this and that 'newnum' is the name of the
3283 * new locale, and we are switched into it. It installs this locale as the
3284 * current underlying default, and then switches to the C locale, if
3285 * necessary, so that the code that has traditionally expected the radix
3286 * character to be a dot may continue to do so.
3288 * The default locale and the C locale can be toggled between by use of the
3289 * set_numeric_underlying() and set_numeric_standard() functions, which
3290 * should probably not be called directly, but only via macros like
3291 * SET_NUMERIC_STANDARD() in perl.h.
3293 * The toggling is necessary mainly so that a non-dot radix decimal point
3294 * character can be input and output, while allowing internal calculations
3297 * This sets several interpreter-level variables:
3298 * PL_numeric_name The underlying locale's name: a copy of 'newnum'
3299 * PL_numeric_underlying A boolean indicating if the toggled state is
3300 * such that the current locale is the program's
3302 * PL_numeric_standard An int indicating if the toggled state is such
3303 * that the current locale is the C locale or
3304 * indistinguishable from the C locale. If non-zero, it
3305 * is in C; if > 1, it means it may not be toggled away
3307 * PL_numeric_underlying_is_standard A bool kept by this function
3308 * indicating that the underlying locale and the standard
3309 * C locale are indistinguishable for the purposes of
3310 * LC_NUMERIC. This happens when both of the above two
3311 * variables are true at the same time. (Toggling is a
3312 * no-op under these circumstances.) This variable is
3313 * used to avoid having to recalculate.
3314 * PL_numeric_radix_sv Contains the string that code should use for the
3315 * decimal point. It is set to either a dot or the
3316 * program's underlying locale's radix character string,
3317 * depending on the situation.
3318 * PL_underlying_radix_sv Contains the program's underlying locale's
3319 * radix character string. This is copied into
3320 * PL_numeric_radix_sv when the situation warrants. It
3321 * exists to avoid having to recalculate it when toggling.
3324 DEBUG_L( PerlIO_printf(Perl_debug_log,
3325 "Called new_numeric with %s, PL_numeric_name=%s\n",
3326 newnum, PL_numeric_name));
3328 /* If not forcing this procedure, and there isn't actually a change from
3329 * our records, do nothing. (Our records can be wrong when sync'ing to the
3330 * locale set up by an external library, hence the 'force' parameter) */
3331 if (! force && strEQ(PL_numeric_name, newnum)) {
3335 Safefree(PL_numeric_name);
3336 PL_numeric_name = savepv(newnum);
3338 /* Handle the trivial case. Since this is called at process
3339 * initialization, be aware that this bit can't rely on much being
3341 if (isNAME_C_OR_POSIX(PL_numeric_name)) {
3342 PL_numeric_standard = TRUE;
3343 PL_numeric_underlying_is_standard = TRUE;
3344 PL_numeric_underlying = TRUE;
3345 sv_setpv(PL_numeric_radix_sv, C_decimal_point);
3346 SvUTF8_off(PL_numeric_radix_sv);
3347 sv_setpv(PL_underlying_radix_sv, C_decimal_point);
3348 SvUTF8_off(PL_underlying_radix_sv);
3352 /* We are in the underlying locale until changed at the end of this
3354 PL_numeric_underlying = TRUE;
3356 char * radix = NULL;
3357 utf8ness_t utf8ness = UTF8NESS_IMMATERIAL;
3359 /* Find and save this locale's radix character. */
3360 my_langinfo_c(RADIXCHAR, LC_NUMERIC, PL_numeric_name,
3361 &radix, NULL, &utf8ness);
3362 sv_setpv(PL_underlying_radix_sv, radix);
3364 if (utf8ness == UTF8NESS_YES) {
3365 SvUTF8_on(PL_underlying_radix_sv);
3368 SvUTF8_off(PL_underlying_radix_sv);
3371 DEBUG_L(PerlIO_printf(Perl_debug_log,
3372 "Locale radix is '%s', ?UTF-8=%d\n",
3373 SvPVX(PL_underlying_radix_sv),
3374 cBOOL(SvUTF8(PL_underlying_radix_sv))));
3376 /* This locale is indistinguishable from C (for numeric purposes) if both
3377 * the radix character and the thousands separator are the same as C's.
3378 * Start with the radix. */
3379 PL_numeric_underlying_is_standard = strEQ(C_decimal_point, radix);
3382 # ifndef TS_W32_BROKEN_LOCALECONV
3384 /* If the radix isn't the same as C's, we know it is distinguishable from
3385 * C; otherwise check the thousands separator too. Only if both are the
3386 * same as C's is the locale indistinguishable from C.
3388 * But on earlier Windows versions, there is a potential race. This code
3389 * knows that localeconv() (elsewhere in this file) will be used to extract
3390 * the needed value, and localeconv() was buggy for quite a while, and that
3391 * code in this file hence uses a workaround. And that workaround may have
3392 * an (unlikely) race. Gathering the radix uses a different workaround on
3393 * Windows that doesn't involve a race. It might be possible to do the
3394 * same for this (patches welcome).
3396 * Until then khw doesn't think it's worth even the small risk of a race to
3397 * get this value, which doesn't appear to be used in any of the Microsoft
3398 * library routines anyway. */
3400 char * scratch_buffer = NULL;
3401 if (PL_numeric_underlying_is_standard) {
3402 PL_numeric_underlying_is_standard = strEQ(C_thousands_sep,
3403 my_langinfo_c(THOUSEP, LC_NUMERIC,
3408 Safefree(scratch_buffer);
3412 PL_numeric_standard = PL_numeric_underlying_is_standard;
3414 /* Keep LC_NUMERIC so that it has the C locale radix and thousands
3415 * separator. This is for XS modules, so they don't have to worry about
3416 * the radix being a non-dot. (Core operations that need the underlying
3417 * locale change to it temporarily). */
3418 if (! PL_numeric_standard) {
3419 set_numeric_standard(__FILE__, __LINE__);
3426 Perl_set_numeric_standard(pTHX_ const char * const file, const line_t line)
3428 PERL_ARGS_ASSERT_SET_NUMERIC_STANDARD;
3429 PERL_UNUSED_ARG(file); /* Some Configurations ignore these */
3430 PERL_UNUSED_ARG(line);
3432 # ifdef USE_LOCALE_NUMERIC
3434 /* Unconditionally toggle the LC_NUMERIC locale to the C locale
3436 * Most code should use the macro SET_NUMERIC_STANDARD() in perl.h
3437 * instead of calling this directly. The macro avoids calling this routine
3438 * if toggling isn't necessary according to our records (which could be
3439 * wrong if some XS code has changed the locale behind our back) */
3441 DEBUG_L(PerlIO_printf(Perl_debug_log, "Setting LC_NUMERIC locale to"
3442 " standard C; called from %s: %"
3443 LINE_Tf "\n", file, line));
3445 void_setlocale_c_with_caller(LC_NUMERIC, "C", file, line);
3446 PL_numeric_standard = TRUE;
3447 sv_setpv(PL_numeric_radix_sv, C_decimal_point);
3448 SvUTF8_off(PL_numeric_radix_sv);
3450 PL_numeric_underlying = PL_numeric_underlying_is_standard;
3452 # endif /* USE_LOCALE_NUMERIC */
3457 Perl_set_numeric_underlying(pTHX_ const char * const file, const line_t line)
3459 PERL_ARGS_ASSERT_SET_NUMERIC_UNDERLYING;
3460 PERL_UNUSED_ARG(file); /* Some Configurations ignore these */
3461 PERL_UNUSED_ARG(line);
3463 # ifdef USE_LOCALE_NUMERIC
3465 /* Unconditionally toggle the LC_NUMERIC locale to the current underlying
3468 * Most code should use the macro SET_NUMERIC_UNDERLYING() in perl.h
3469 * instead of calling this directly. The macro avoids calling this routine
3470 * if toggling isn't necessary according to our records (which could be
3471 * wrong if some XS code has changed the locale behind our back) */
3473 DEBUG_L(PerlIO_printf(Perl_debug_log, "Setting LC_NUMERIC locale to %s;"
3474 " called from %s: %" LINE_Tf "\n",
3475 PL_numeric_name, file, line));
3476 /* Maybe not in init? assert(PL_locale_mutex_depth > 0);*/
3478 void_setlocale_c_with_caller(LC_NUMERIC, PL_numeric_name, file, line);
3479 PL_numeric_underlying = TRUE;
3480 sv_setsv_nomg(PL_numeric_radix_sv, PL_underlying_radix_sv);
3482 PL_numeric_standard = PL_numeric_underlying_is_standard;
3484 # endif /* USE_LOCALE_NUMERIC */
3488 # ifdef USE_LOCALE_CTYPE
3491 S_new_ctype(pTHX_ const char *newctype, bool force)
3493 PERL_ARGS_ASSERT_NEW_CTYPE;
3494 PERL_UNUSED_ARG(force);
3496 /* Called after each libc setlocale() call affecting LC_CTYPE, to tell
3497 * core Perl this and that 'newctype' is the name of the new locale.
3499 * This function sets up the folding arrays for all 256 bytes, assuming
3500 * that tofold() is tolc() since fold case is not a concept in POSIX,
3503 DEBUG_L(PerlIO_printf(Perl_debug_log, "Entering new_ctype(%s)\n",
3506 /* No change means no-op */
3507 if (strEQ(PL_ctype_name, newctype)) {
3511 /* We will replace any bad locale warning with
3512 * 1) nothing if the new one is ok; or
3513 * 2) a new warning for the bad new locale */
3514 if (PL_warn_locale) {
3515 SvREFCNT_dec_NN(PL_warn_locale);
3516 PL_warn_locale = NULL;
3520 Safefree(PL_ctype_name);
3523 PL_in_utf8_turkic_locale = FALSE;
3525 /* For the C locale, just use the standard folds, and we know there are no
3526 * glitches possible, so return early. Since this is called at process
3527 * initialization, be aware that this bit can't rely on much being
3529 if (isNAME_C_OR_POSIX(newctype)) {
3530 Copy(PL_fold, PL_fold_locale, 256, U8);
3531 PL_ctype_name = savepv(newctype);
3532 PL_in_utf8_CTYPE_locale = FALSE;
3536 /* The cache being cleared signals the called function to compute a new
3538 PL_in_utf8_CTYPE_locale = is_locale_utf8(newctype);
3540 PL_ctype_name = savepv(newctype);
3541 bool maybe_utf8_turkic = FALSE;
3543 /* Don't check for problems if we are suppressing the warnings */
3544 bool check_for_problems = ckWARN_d(WARN_LOCALE) || UNLIKELY(DEBUG_L_TEST);
3546 if (PL_in_utf8_CTYPE_locale) {
3548 /* A UTF-8 locale gets standard rules. But note that code still has to
3549 * handle this specially because of the three problematic code points
3551 Copy(PL_fold_latin1, PL_fold_locale, 256, U8);
3553 /* UTF-8 locales can have special handling for 'I' and 'i' if they are
3554 * Turkic. Make sure these two are the only anomalies. (We don't
3555 * require towupper and towlower because they aren't in C89.) */
3557 # if defined(HAS_TOWUPPER) && defined (HAS_TOWLOWER)
3559 if (towupper('i') == 0x130 && towlower('I') == 0x131)
3563 if (toU8_UPPER_LC('i') == 'i' && toU8_LOWER_LC('I') == 'I')
3568 /* This is how we determine it really is Turkic */
3569 check_for_problems = TRUE;
3570 maybe_utf8_turkic = TRUE;
3573 else { /* Not a canned locale we know the values for. Compute them */
3577 bool has_non_ascii_fold = FALSE;
3578 bool found_unexpected = FALSE;
3580 /* Under -DLv, see if there are any folds outside the ASCII range.
3581 * This factoid is used below */
3582 if (DEBUG_Lv_TEST) {
3583 for (unsigned i = 128; i < 256; i++) {
3584 int j = LATIN1_TO_NATIVE(i);
3585 if (toU8_LOWER_LC(j) != j || toU8_UPPER_LC(j) != j) {
3586 has_non_ascii_fold = TRUE;
3594 for (unsigned i = 0; i < 256; i++) {
3595 if (isU8_UPPER_LC(i))
3596 PL_fold_locale[i] = (U8) toU8_LOWER_LC(i);
3597 else if (isU8_LOWER_LC(i))
3598 PL_fold_locale[i] = (U8) toU8_UPPER_LC(i);
3600 PL_fold_locale[i] = (U8) i;
3604 /* Most locales these days are supersets of ASCII. When debugging
3605 * with -DLv, it is helpful to know what the exceptions to that are
3607 if (DEBUG_Lv_TEST) {
3608 bool unexpected = FALSE;
3610 if (isUPPER_L1(i)) {
3612 if (PL_fold_locale[i] != toLOWER_A(i)) {
3616 else if (has_non_ascii_fold) {
3617 if (PL_fold_locale[i] != toLOWER_L1(i)) {
3621 else if (PL_fold_locale[i] != i) {
3625 else if ( isLOWER_L1(i)
3626 && i != LATIN_SMALL_LETTER_SHARP_S
3630 if (PL_fold_locale[i] != toUPPER_A(i)) {
3634 else if (has_non_ascii_fold) {
3635 if (PL_fold_locale[i] != toUPPER_LATIN1_MOD(i)) {
3639 else if (PL_fold_locale[i] != i) {
3643 else if (PL_fold_locale[i] != i) {
3648 found_unexpected = TRUE;
3649 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3650 "For %s, fold of %02x is %02x\n",
3651 newctype, i, PL_fold_locale[i]));
3656 if (found_unexpected) {
3657 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3658 "All bytes not mentioned above either fold to"
3659 " themselves or are the expected ASCII or"
3663 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3664 "No nonstandard folds were found\n"));
3672 /* We only handle single-byte locales (outside of UTF-8 ones); so if this
3673 * locale requires more than one byte, there are going to be BIG problems.
3676 const int mb_cur_max = MB_CUR_MAX;
3678 if (mb_cur_max > 1 && ! PL_in_utf8_CTYPE_locale
3680 /* Some platforms return MB_CUR_MAX > 1 for even the "C" locale.
3681 * Just assume that the implementation for them (plus for POSIX) is
3682 * correct and the > 1 value is spurious. (Since these are
3683 * specially handled to never be considered UTF-8 locales, as long
3684 * as this is the only problem, everything should work fine */
3685 && ! isNAME_C_OR_POSIX(newctype))
3687 DEBUG_L(PerlIO_printf(Perl_debug_log,
3688 "Unsupported, MB_CUR_MAX=%d\n", mb_cur_max));
3690 Perl_ck_warner_d(aTHX_ packWARN(WARN_LOCALE),
3691 "Locale '%s' is unsupported, and may crash the"
3698 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "check_for_problems=%d\n",
3699 check_for_problems));
3701 /* We don't populate the other lists if a UTF-8 locale, but do check that
3702 * everything works as expected, unless checking turned off */
3703 if (check_for_problems) {
3704 /* Assume enough space for every character being bad. 4 spaces each
3705 * for the 94 printable characters that are output like "'x' "; and 5
3706 * spaces each for "'\\' ", "'\t' ", and "'\n' "; plus a terminating
3708 char bad_chars_list[ (94 * 4) + (3 * 5) + 1 ] = { '\0' };
3709 unsigned int bad_count = 0; /* Count of bad characters */
3711 for (unsigned i = 0; i < 256; i++) {
3713 /* If checking for locale problems, see if the native ASCII-range
3714 * printables plus \n and \t are in their expected categories in
3715 * the new locale. If not, this could mean big trouble, upending
3716 * Perl's and most programs' assumptions, like having a
3717 * metacharacter with special meaning become a \w. Fortunately,
3718 * it's very rare to find locales that aren't supersets of ASCII
3719 * nowadays. It isn't a problem for most controls to be changed
3720 * into something else; we check only \n and \t, though perhaps \r
3721 * could be an issue as well. */
3722 if (isGRAPH_A(i) || isBLANK_A(i) || i == '\n') {
3723 bool is_bad = FALSE;
3724 char name[4] = { '\0' };
3726 /* Convert the name into a string */
3731 else if (i == '\n') {
3732 my_strlcpy(name, "\\n", sizeof(name));
3734 else if (i == '\t') {
3735 my_strlcpy(name, "\\t", sizeof(name));
3739 my_strlcpy(name, "' '", sizeof(name));
3742 /* Check each possibe class */
3743 if (UNLIKELY(cBOOL(isU8_ALPHANUMERIC_LC(i)) !=
3744 cBOOL(isALPHANUMERIC_A(i))))
3747 DEBUG_L(PerlIO_printf(Perl_debug_log,
3748 "isalnum('%s') unexpectedly is %x\n",
3749 name, cBOOL(isU8_ALPHANUMERIC_LC(i))));
3751 if (UNLIKELY(cBOOL(isU8_ALPHA_LC(i)) != cBOOL(isALPHA_A(i)))) {
3753 DEBUG_L(PerlIO_printf(Perl_debug_log,
3754 "isalpha('%s') unexpectedly is %x\n",
3755 name, cBOOL(isU8_ALPHA_LC(i))));
3757 if (UNLIKELY(cBOOL(isU8_DIGIT_LC(i)) != cBOOL(isDIGIT_A(i)))) {
3759 DEBUG_L(PerlIO_printf(Perl_debug_log,
3760 "isdigit('%s') unexpectedly is %x\n",
3761 name, cBOOL(isU8_DIGIT_LC(i))));
3763 if (UNLIKELY(cBOOL(isU8_GRAPH_LC(i)) != cBOOL(isGRAPH_A(i)))) {
3765 DEBUG_L(PerlIO_printf(Perl_debug_log,
3766 "isgraph('%s') unexpectedly is %x\n",
3767 name, cBOOL(isU8_GRAPH_LC(i))));
3769 if (UNLIKELY(cBOOL(isU8_LOWER_LC(i)) != cBOOL(isLOWER_A(i)))) {
3771 DEBUG_L(PerlIO_printf(Perl_debug_log,
3772 "islower('%s') unexpectedly is %x\n",
3773 name, cBOOL(isU8_LOWER_LC(i))));
3775 if (UNLIKELY(cBOOL(isU8_PRINT_LC(i)) != cBOOL(isPRINT_A(i)))) {
3777 DEBUG_L(PerlIO_printf(Perl_debug_log,
3778 "isprint('%s') unexpectedly is %x\n",
3779 name, cBOOL(isU8_PRINT_LC(i))));
3781 if (UNLIKELY(cBOOL(isU8_PUNCT_LC(i)) != cBOOL(isPUNCT_A(i)))) {
3783 DEBUG_L(PerlIO_printf(Perl_debug_log,
3784 "ispunct('%s') unexpectedly is %x\n",
3785 name, cBOOL(isU8_PUNCT_LC(i))));
3787 if (UNLIKELY(cBOOL(isU8_SPACE_LC(i)) != cBOOL(isSPACE_A(i)))) {
3789 DEBUG_L(PerlIO_printf(Perl_debug_log,
3790 "isspace('%s') unexpectedly is %x\n",
3791 name, cBOOL(isU8_SPACE_LC(i))));
3793 if (UNLIKELY(cBOOL(isU8_UPPER_LC(i)) != cBOOL(isUPPER_A(i)))) {
3795 DEBUG_L(PerlIO_printf(Perl_debug_log,
3796 "isupper('%s') unexpectedly is %x\n",
3797 name, cBOOL(isU8_UPPER_LC(i))));
3799 if (UNLIKELY(cBOOL(isU8_XDIGIT_LC(i))!= cBOOL(isXDIGIT_A(i)))) {
3801 DEBUG_L(PerlIO_printf(Perl_debug_log,
3802 "isxdigit('%s') unexpectedly is %x\n",
3803 name, cBOOL(isU8_XDIGIT_LC(i))));
3805 if (UNLIKELY(toU8_LOWER_LC(i) != (int) toLOWER_A(i))) {
3807 DEBUG_L(PerlIO_printf(Perl_debug_log,
3808 "tolower('%s')=0x%x instead of the expected 0x%x\n",
3809 name, toU8_LOWER_LC(i), (int) toLOWER_A(i)));
3811 if (UNLIKELY(toU8_UPPER_LC(i) != (int) toUPPER_A(i))) {
3813 DEBUG_L(PerlIO_printf(Perl_debug_log,
3814 "toupper('%s')=0x%x instead of the expected 0x%x\n",
3815 name, toU8_UPPER_LC(i), (int) toUPPER_A(i)));
3817 if (UNLIKELY((i == '\n' && ! isCNTRL_LC(i)))) {
3819 DEBUG_L(PerlIO_printf(Perl_debug_log,
3820 "'\\n' (=%02X) is not a control\n", (int) i));
3823 /* Add to the list; Separate multiple entries with a blank */
3826 my_strlcat(bad_chars_list, " ", sizeof(bad_chars_list));
3828 my_strlcat(bad_chars_list, name, sizeof(bad_chars_list));
3834 if (bad_count == 2 && maybe_utf8_turkic) {
3836 *bad_chars_list = '\0';
3838 /* The casts are because otherwise some compilers warn:
3839 gcc.gnu.org/bugzilla/show_bug.cgi?id=99950
3840 gcc.gnu.org/bugzilla/show_bug.cgi?id=94182
3842 PL_fold_locale[ (U8) 'I' ] = 'I';
3843 PL_fold_locale[ (U8) 'i' ] = 'i';
3844 PL_in_utf8_turkic_locale = TRUE;
3845 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s is turkic\n", newctype));
3848 /* If we found problems and we want them output, do so */
3849 if ( (UNLIKELY(bad_count))
3850 && (LIKELY(ckWARN_d(WARN_LOCALE)) || UNLIKELY(DEBUG_L_TEST)))
3852 /* WARNING. If you change the wording of these; be sure to update
3853 * t/loc_tools.pl correspondingly */
3855 if (PL_in_utf8_CTYPE_locale) {
3856 PL_warn_locale = Perl_newSVpvf(aTHX_
3857 "Locale '%s' contains (at least) the following characters"
3858 " which have\nunexpected meanings: %s\nThe Perl program"
3859 " will use the expected meanings",
3860 newctype, bad_chars_list);
3865 "\nThe following characters (and maybe"
3866 " others) may not have the same meaning as"
3867 " the Perl program expects: %s\n",
3872 # if defined(HAS_SOME_LANGINFO) || defined(WIN32)
3874 char * scratch_buffer = NULL;
3875 Perl_sv_catpvf(aTHX_ PL_warn_locale, "; codeset=%s",
3876 my_langinfo_c(CODESET, LC_CTYPE,
3878 &scratch_buffer, NULL,
3880 Safefree(scratch_buffer);
3884 Perl_sv_catpvf(aTHX_ PL_warn_locale, "\n");
3886 /* If we are actually in the scope of the locale or are debugging,
3887 * output the message now. If not in that scope, we save the
3888 * message to be output at the first operation using this locale,
3889 * if that actually happens. Most programs don't use locales, so
3890 * they are immune to bad ones. */
3891 if (IN_LC(LC_CTYPE) || UNLIKELY(DEBUG_L_TEST)) {
3893 /* The '0' below suppresses a bogus gcc compiler warning */
3894 Perl_warner(aTHX_ packWARN(WARN_LOCALE), SvPVX(PL_warn_locale),
3896 if (IN_LC(LC_CTYPE)) {
3897 SvREFCNT_dec_NN(PL_warn_locale);
3898 PL_warn_locale = NULL;
3906 Perl_warn_problematic_locale()
3910 /* Core-only function that outputs the message in PL_warn_locale,
3911 * and then NULLS it. Should be called only through the macro
3912 * CHECK_AND_WARN_PROBLEMATIC_LOCALE_ */
3914 if (PL_warn_locale) {
3915 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
3916 SvPVX(PL_warn_locale),
3917 0 /* dummy to avoid compiler warning */ );
3918 SvREFCNT_dec_NN(PL_warn_locale);
3919 PL_warn_locale = NULL;
3923 # endif /* USE_LOCALE_CTYPE */
3929 S_new_LC_ALL(pTHX_ const char *lc_all, bool force)
3933 S_new_LC_ALL(pTHX_ const char ** individ_locales, bool force)
3938 PERL_ARGS_ASSERT_NEW_LC_ALL;
3940 /* new_LC_ALL() updates all the things we care about. Note that this is
3941 * called just after a change, so uses the actual underlying locale just
3942 * set, and not the nominal one (should they differ, as they may in
3947 const char * individ_locales[LC_ALL_INDEX_] = { NULL };
3949 switch (parse_LC_ALL_string(lc_all,
3951 override_if_ignored, /* Override any ignored
3953 true, /* Always fill array */
3954 true, /* Panic if fails, as to get here it
3955 earlier had to have succeeded */
3960 case only_element_0:
3961 locale_panic_("Unexpected return from parse_LC_ALL_string");
3969 for_all_individual_category_indexes(i) {
3970 if (update_functions[i]) {
3971 const char * this_locale = individ_locales[i];
3972 update_functions[i](aTHX_ this_locale, force);
3977 Safefree(individ_locales[i]);
3984 # ifdef USE_LOCALE_COLLATE
3987 S_new_collate(pTHX_ const char *newcoll, bool force)
3989 PERL_ARGS_ASSERT_NEW_COLLATE;
3990 PERL_UNUSED_ARG(force);
3992 /* Called after each libc setlocale() call affecting LC_COLLATE, to tell
3993 * core Perl this and that 'newcoll' is the name of the new locale.
3995 * The design of locale collation is that every locale change is given an
3996 * index 'PL_collation_ix'. The first time a string participates in an
3997 * operation that requires collation while locale collation is active, it
3998 * is given PERL_MAGIC_collxfrm magic (via sv_collxfrm_flags()). That
3999 * magic includes the collation index, and the transformation of the string
4000 * by strxfrm(), q.v. That transformation is used when doing comparisons,
4001 * instead of the string itself. If a string changes, the magic is
4002 * cleared. The next time the locale changes, the index is incremented,
4003 * and so we know during a comparison that the transformation is not
4004 * necessarily still valid, and so is recomputed. Note that if the locale
4005 * changes enough times, the index could wrap, and it is possible that a
4006 * transformation would improperly be considered valid, leading to an
4007 * unlikely bug. The value is declared to the widest possible type on this
4010 /* Return if the locale isn't changing */
4011 if (strEQ(PL_collation_name, newcoll)) {
4015 Safefree(PL_collation_name);
4016 PL_collation_name = savepv(newcoll);
4019 /* Set the new one up if trivial. Since this is called at process
4020 * initialization, be aware that this bit can't rely on much being
4022 PL_collation_standard = isNAME_C_OR_POSIX(newcoll);
4023 if (PL_collation_standard) {
4024 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4025 "Setting PL_collation name='%s'\n",
4026 PL_collation_name));
4027 PL_collxfrm_base = 0;
4028 PL_collxfrm_mult = 2;
4029 PL_in_utf8_COLLATE_locale = FALSE;
4030 PL_strxfrm_NUL_replacement = '\0';
4031 PL_strxfrm_max_cp = 0;
4035 /* Flag that the remainder of the set up is being deferred until first
4037 PL_collxfrm_mult = 0;
4038 PL_collxfrm_base = 0;
4042 # endif /* USE_LOCALE_COLLATE */
4047 S_Win_byte_string_to_wstring(const UINT code_page, const char * byte_string)
4049 /* Caller must arrange to free the returned string */
4051 int req_size = MultiByteToWideChar(code_page, 0, byte_string, -1, NULL, 0);
4058 Newx(wstring, req_size, wchar_t);
4060 if (! MultiByteToWideChar(code_page, 0, byte_string, -1, wstring, req_size))
4070 # define Win_utf8_string_to_wstring(s) \
4071 Win_byte_string_to_wstring(CP_UTF8, (s))
4074 S_Win_wstring_to_byte_string(const UINT code_page, const wchar_t * wstring)
4076 /* Caller must arrange to free the returned string */
4079 WideCharToMultiByte(code_page, 0, wstring, -1, NULL, 0, NULL, NULL);
4082 Newx(byte_string, req_size, char);
4084 if (! WideCharToMultiByte(code_page, 0, wstring, -1, byte_string,
4085 req_size, NULL, NULL))
4087 Safefree(byte_string);
4095 # define Win_wstring_to_utf8_string(ws) \
4096 Win_wstring_to_byte_string(CP_UTF8, (ws))
4099 S_wrap_wsetlocale(pTHX_ const int category, const char *locale)
4101 PERL_ARGS_ASSERT_WRAP_WSETLOCALE;
4103 /* Calls _wsetlocale(), converting the parameters/return to/from
4104 * Perl-expected forms as if plain setlocale() were being called instead.
4106 * Caller must arrange for the returned PV to be freed.
4109 const wchar_t * wlocale = NULL;
4112 wlocale = Win_utf8_string_to_wstring(locale);
4119 const wchar_t * wresult = _wsetlocale(category, wlocale);
4127 const char * result = Win_wstring_to_utf8_string(wresult);
4135 S_win32_setlocale(pTHX_ int category, const char* locale)
4137 /* This, for Windows, emulates POSIX setlocale() behavior. There is no
4138 * difference between the two unless the input locale is "", which normally
4139 * means on Windows to get the machine default, which is set via the
4140 * computer's "Regional and Language Options" (or its current equivalent).
4141 * In POSIX, it instead means to find the locale from the user's
4142 * environment. This routine changes the Windows behavior to try the POSIX
4143 * behavior first. Further details are in the called function
4144 * find_locale_from_environment().
4147 if (locale != NULL && strEQ(locale, "")) {
4148 /* Note this function may change the locale, but that's ok because we
4149 * are about to change it anyway */
4150 locale = find_locale_from_environment(get_category_index(category));
4151 if (locale == NULL) {
4157 const char * result = wrap_wsetlocale(category, locale);
4158 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n",
4159 setlocale_debug_string_r(category, locale, result)));
4166 save_to_buffer(result, &PL_setlocale_buf, &PL_setlocale_bufsize);
4168 # ifndef USE_PL_CUR_LC_ALL
4174 /* Here, we need to keep track of LC_ALL, so store the new value. but if
4175 * the input locale is NULL, we were just querying, so the original value
4177 if (locale == NULL) {
4182 /* If we set LC_ALL directly above, we already know its new value; but
4183 * if we changed just an individual category, find the new LC_ALL */
4184 if (category != LC_ALL) {
4186 result = wrap_wsetlocale(LC_ALL, NULL);
4189 Safefree(PL_cur_LC_ALL);
4190 PL_cur_LC_ALL = result;
4193 DEBUG_L(PerlIO_printf(Perl_debug_log, "new PL_cur_LC_ALL=%s\n",
4197 return PL_setlocale_buf;
4203 S_native_querylocale_i(pTHX_ const locale_category_index cat_index)
4205 /* Determine the current locale and return it in the form the platform's
4206 * native locale handling understands. This is different only from our
4207 * internal form for the LC_ALL category, as platforms differ in how they
4210 * This is only called from Perl_setlocale(). As such it returns in
4211 * PL_setlocale_buf */
4213 # ifdef USE_LOCALE_NUMERIC
4215 /* We have the LC_NUMERIC name saved, because we are normally switched into
4216 * the C locale (or equivalent) for it. */
4217 if (cat_index == LC_NUMERIC_INDEX_) {
4219 /* We don't have to copy this return value, as it is a per-thread
4220 * variable, and won't change until a future setlocale */
4221 return PL_numeric_name;
4227 if (cat_index != LC_ALL_INDEX_)
4232 /* Here, not LC_ALL, and not LC_NUMERIC: the actual and native values
4235 # ifdef setlocale_i /* Can shortcut if this is defined */
4237 return setlocale_i(cat_index, NULL);
4241 return save_to_buffer(querylocale_i(cat_index),
4242 &PL_setlocale_buf, &PL_setlocale_bufsize);
4247 /* Below, querying LC_ALL */
4250 # ifdef USE_PL_CURLOCALES
4251 # define LC_ALL_ARG PL_curlocales
4253 # define LC_ALL_ARG NULL /* Causes calculate_LC_ALL_string() to find the
4254 locale using a querylocale function */
4257 return calculate_LC_ALL_string(LC_ALL_ARG, EXTERNAL_FORMAT_FOR_QUERY,
4258 WANT_PL_setlocale_buf,
4261 # endif /* has LC_ALL */
4265 #endif /* USE_LOCALE */
4268 =for apidoc Perl_setlocale
4270 This is an (almost) drop-in replacement for the system L<C<setlocale(3)>>,
4271 taking the same parameters, and returning the same information, except that it
4272 returns the correct underlying C<LC_NUMERIC> locale. Regular C<setlocale> will
4273 instead return C<C> if the underlying locale has a non-dot decimal point
4274 character, or a non-empty thousands separator for displaying floating point
4275 numbers. This is because perl keeps that locale category such that it has a
4276 dot and empty separator, changing the locale briefly during the operations
4277 where the underlying one is required. C<Perl_setlocale> knows about this, and
4278 compensates; regular C<setlocale> doesn't.
4280 Another reason it isn't completely a drop-in replacement is that it is
4281 declared to return S<C<const char *>>, whereas the system setlocale omits the
4282 C<const> (presumably because its API was specified long ago, and can't be
4283 updated; it is illegal to change the information C<setlocale> returns; doing
4284 so leads to segfaults.)
4286 Finally, C<Perl_setlocale> works under all circumstances, whereas plain
4287 C<setlocale> can be completely ineffective on some platforms under some
4290 Changing the locale is not a good idea when more than one thread is running,
4291 except on systems where the predefined variable C<${^SAFE_LOCALES}> is
4292 non-zero. This is because on such systems the locale is global to the whole
4293 process and not local to just the thread calling the function. So changing it
4294 in one thread instantaneously changes it in all. On some such systems, the
4295 system C<setlocale()> is ineffective, returning the wrong information, and
4296 failing to actually change the locale. z/OS refuses to try to change the
4297 locale once a second thread is created. C<Perl_setlocale>, should give you
4298 accurate results of what actually happened on these problematic platforms,
4299 returning NULL if the system forbade the locale change.
4301 The return points to a per-thread static buffer, which is overwritten the next
4302 time C<Perl_setlocale> is called from the same thread.
4309 Perl_setlocale(const int category, const char * locale)
4311 /* This wraps POSIX::setlocale() */
4315 PERL_UNUSED_ARG(category);
4316 PERL_UNUSED_ARG(locale);
4324 DEBUG_L(PerlIO_printf(Perl_debug_log,
4325 "Entering Perl_setlocale(%d, \"%s\")\n",
4328 bool valid_category;
4329 locale_category_index cat_index = get_category_index_helper(category,
4332 if (! valid_category) {
4333 if (ckWARN(WARN_LOCALE)) {
4334 const char * conditional_warn_text;
4335 if (locale == NULL) {
4336 conditional_warn_text = "";
4340 conditional_warn_text = "; can't set it to ";
4343 /* diag_listed_as: Unknown locale category %d; can't set it to %s */
4345 packWARN(WARN_LOCALE),
4346 "Unknown locale category %d%s%s",
4347 category, conditional_warn_text, locale);
4356 /* setlocale_i() gets defined only on Configurations that use setlocale()
4357 * in a simple manner that adequately handles all cases. If this category
4358 * doesn't have any perl complications, just do that. */
4359 if (! update_functions[cat_index]) {
4360 return setlocale_i(cat_index, locale);
4365 /* Get current locale */
4366 const char * current_locale = native_querylocale_i(cat_index);
4368 /* A NULL locale means only query what the current one is. */
4369 if (locale == NULL) {
4370 return current_locale;
4373 if (strEQ(current_locale, locale)) {
4374 DEBUG_L(PerlIO_printf(Perl_debug_log,
4375 "Already in requested locale: no action taken\n"));
4376 return current_locale;
4379 /* Here, an actual change is being requested. Do it */
4380 if (! bool_setlocale_i(cat_index, locale)) {
4381 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n",
4382 setlocale_debug_string_i(cat_index, locale, "NULL")));
4386 /* At this point, the locale has been changed based on the requested value,
4387 * and the querylocale_i() will return the actual new value that the system
4388 * has for the category. That may not be the same as the input, as libc
4389 * may have returned a synonymous locale name instead of the input one; or,
4390 * if there are locale categories that we are compiled to ignore, any
4391 * attempt to change them away from "C" is overruled */
4392 current_locale = querylocale_i(cat_index);
4394 /* But certain categories need further work. For example we may need to
4395 * calculate new folding or collation rules. And for LC_NUMERIC, we have
4396 * to switch into a locale that has a dot radix. */
4397 if (update_functions[cat_index]) {
4398 update_functions[cat_index](aTHX_ current_locale,
4399 /* No need to force recalculation, as
4400 * aren't coming from a situation
4401 * where Perl hasn't been controlling
4402 * the locale, so has accurate
4407 /* Make sure the result is in a stable buffer for the caller's use, and is
4408 * in the expected format */
4409 current_locale = native_querylocale_i(cat_index);
4411 DEBUG_L(PerlIO_printf(Perl_debug_log, "returning '%s'\n", current_locale));
4413 return current_locale;
4419 #if defined(USE_LOCALE) || defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV)
4422 S_get_locale_string_utf8ness_i(pTHX_ const char * string,
4423 const locale_utf8ness_t known_utf8,
4424 const char * locale,
4425 const locale_category_index cat_index)
4427 PERL_ARGS_ASSERT_GET_LOCALE_STRING_UTF8NESS_I;
4432 PERL_UNUSED_ARG(string);
4433 PERL_UNUSED_ARG(known_utf8);
4434 PERL_UNUSED_ARG(locale);
4435 PERL_UNUSED_ARG(cat_index);
4439 assert(cat_index <= LC_ALL_INDEX_);
4441 /* Return to indicate if 'string' in the locale given by the input
4442 * arguments should be considered UTF-8 or not.
4444 * If the input 'locale' is not NULL, use that for the locale; otherwise
4445 * use the current locale for the category specified by 'cat_index'.
4448 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4449 "Entering get_locale_string_utf8ness_i; locale=%s,"
4450 " index=%u(%s), string=%s, known_utf8=%d\n",
4451 locale, cat_index, category_names[cat_index],
4453 ? _byte_dump_string((U8 *) string,
4458 if (string == NULL) {
4459 return UTF8NESS_IMMATERIAL;
4462 if (IN_BYTES) { /* respect 'use bytes' */
4466 Size_t len = strlen(string);
4468 /* UTF8ness is immaterial if the representation doesn't vary */
4469 const U8 * first_variant = NULL;
4470 if (is_utf8_invariant_string_loc((U8 *) string, len, &first_variant)) {
4471 return UTF8NESS_IMMATERIAL;
4474 /* Can't be UTF-8 if invalid */
4475 if (! is_utf8_string((U8 *) first_variant,
4476 len - ((char *) first_variant - string)))
4481 /* Here and below, we know the string is legal UTF-8, containing at least
4482 * one character requiring a sequence of two or more bytes. It is quite
4483 * likely to be UTF-8. But it pays to be paranoid and do further checking.
4485 * If we already know the UTF-8ness of the locale, then we immediately know
4486 * what the string is */
4487 if (UNLIKELY(known_utf8 != LOCALE_UTF8NESS_UNKNOWN)) {
4488 if (known_utf8 == LOCALE_IS_UTF8) {
4489 return UTF8NESS_YES;
4496 # ifdef HAS_RELIABLE_UTF8NESS_DETERMINATION
4498 /* Here, we have available the libc functions that can be used to
4499 * accurately determine the UTF8ness of the underlying locale. If it is a
4500 * UTF-8 locale, the string is UTF-8; otherwise it was coincidental that
4501 * the string is legal UTF-8
4503 * However, if the perl is compiled to not pay attention to the category
4504 * being passed in, you might think that that locale is essentially always
4505 * the C locale, so it would make sense to say it isn't UTF-8. But to get
4506 * here, the string has to contain characters unknown in the C locale. And
4507 * in fact, Windows boxes are compiled without LC_MESSAGES, as their
4508 * message catalog isn't really a part of the locale system. But those
4509 * messages really could be UTF-8, and given that the odds are rather small
4510 * of something not being UTF-8 but being syntactically valid UTF-8, khw
4511 * has decided to call such strings as UTF-8. */
4513 if (locale == NULL) {
4514 locale = querylocale_i(cat_index);
4517 if (is_locale_utf8(locale)) {
4518 return UTF8NESS_YES;
4525 /* Here, we have a valid UTF-8 string containing non-ASCII characters, and
4526 * don't have access to functions to check if the locale is UTF-8 or not.
4527 * Assume that it is. khw tried adding a check that the string is entirely
4528 * in a single Unicode script, but discovered the strftime() timezone is
4529 * user-settable through the environment, which may be in a different
4530 * script than the locale-expected value. */
4531 PERL_UNUSED_ARG(locale);
4532 PERL_UNUSED_ARG(cat_index);
4534 return UTF8NESS_YES;
4542 S_is_locale_utf8(pTHX_ const char * locale)
4544 /* Returns TRUE if the locale 'locale' is UTF-8; FALSE otherwise. It uses
4545 * my_langinfo(), which employs various methods to get this information
4546 * if nl_langinfo() isn't available, using heuristics as a last resort, in
4547 * which case, the result will very likely be correct for locales for
4548 * languages that have commonly used non-ASCII characters, but for notably
4549 * English, it comes down to if the locale's name ends in something like
4550 * "UTF-8". It errs on the side of not being a UTF-8 locale.
4552 * Systems conforming to C99 should have the needed libc calls to give us a
4553 * completely reliable result. */
4555 # if ! defined(USE_LOCALE) \
4556 || ! defined(USE_LOCALE_CTYPE) \
4557 || defined(EBCDIC) /* There aren't any real UTF-8 locales at this time */
4559 PERL_UNUSED_ARG(locale);
4565 char * scratch_buffer = NULL;
4566 const char * codeset;
4569 PERL_ARGS_ASSERT_IS_LOCALE_UTF8;
4571 if (strEQ(locale, PL_ctype_name)) {
4572 return PL_in_utf8_CTYPE_locale;
4575 codeset = my_langinfo_c(CODESET, LC_CTYPE, locale,
4576 &scratch_buffer, NULL, NULL);
4577 retval = is_codeset_name_UTF8(codeset);
4579 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4580 "found codeset=%s, is_utf8=%d\n", codeset, retval));
4582 Safefree(scratch_buffer);
4584 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "is_locale_utf8(%s) returning %d\n",
4596 S_set_save_buffer_min_size(pTHX_ Size_t min_len,
4598 Size_t * buf_cursize)
4600 /* Make sure the buffer pointed to by *buf is at least as large 'min_len';
4601 * *buf_cursize is the size of 'buf' upon entry; it will be updated to the
4602 * new size on exit. 'buf_cursize' being NULL is to be used when this is a
4603 * single use buffer, which will shortly be freed by the caller. */
4605 if (buf_cursize == NULL) {
4606 Newx(*buf, min_len, char);
4608 else if (*buf_cursize == 0) {
4609 Newx(*buf, min_len, char);
4610 *buf_cursize = min_len;
4612 else if (min_len > *buf_cursize) {
4613 Renew(*buf, min_len, char);
4614 *buf_cursize = min_len;
4619 S_save_to_buffer(pTHX_ const char * string, char **buf, Size_t *buf_size)
4621 PERL_ARGS_ASSERT_SAVE_TO_BUFFER;
4623 /* Copy the NUL-terminated 'string' to a buffer whose address before this
4624 * call began at *buf, and whose available length before this call was
4627 * If the length of 'string' is greater than the space available, the
4628 * buffer is grown accordingly, which may mean that it gets relocated.
4629 * *buf and *buf_size will be updated to reflect this.
4631 * Regardless, the function returns a pointer to where 'string' is now
4634 * 'string' may be NULL, which means no action gets taken, and NULL is
4637 * 'buf_size' being NULL is to be used when this is a single use buffer,
4638 * which will shortly be freed by the caller.
4640 * If *buf or 'buf_size' are NULL or *buf_size is 0, the buffer is assumed
4641 * empty, and memory is malloc'd.
4648 /* No-op to copy over oneself */
4649 if (string == *buf) {
4653 Size_t string_size = strlen(string) + 1;
4654 set_save_buffer_min_size(string_size, buf, buf_size);
4658 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4659 "Copying '%s' to %p\n",
4660 ((is_utf8_string((U8 *) string, 0))
4662 :_byte_dump_string((U8 *) string, strlen(string), 0)),
4665 /* Catch glitches. Usually this is because LC_CTYPE needs to be the same
4666 * locale as whatever is being worked on */
4667 if (UNLIKELY(instr(string, REPLACEMENT_CHARACTER_UTF8))) {
4668 locale_panic_(Perl_form(aTHX_
4669 "Unexpected REPLACEMENT_CHARACTER in '%s'\n%s",
4670 string, get_LC_ALL_display()));
4675 Copy(string, *buf, string_size, char);
4682 Perl_get_win32_message_utf8ness(pTHX_ const char * string)
4684 /* This is because Windows doesn't have LC_MESSAGES. */
4686 # ifdef USE_LOCALE_CTYPE
4688 return get_locale_string_utf8ness_i(string, LOCALE_IS_UTF8,
4689 NULL, LC_CTYPE_INDEX_);
4699 #endif /* USE_LOCALE */
4702 Perl_mbtowc_(pTHX_ const wchar_t * pwc, const char * s, const Size_t len)
4705 #if ! defined(HAS_MBRTOWC) && ! defined(HAS_MBTOWC)
4707 PERL_UNUSED_ARG(pwc);
4709 PERL_UNUSED_ARG(len);
4712 #else /* Below we have some form of mbtowc() */
4713 # if defined(HAS_MBRTOWC) \
4714 && (defined(USE_LOCALE_THREADS) || ! defined(HAS_MBTOWC))
4715 # define USE_MBRTOWC
4722 if (s == NULL) { /* Initialize the shift state to all zeros in
4725 # if defined(USE_MBRTOWC)
4727 memzero(&PL_mbrtowc_ps, sizeof(PL_mbrtowc_ps));
4734 retval = mbtowc(NULL, NULL, 0);
4742 # if defined(USE_MBRTOWC)
4746 retval = (SSize_t) mbrtowc((wchar_t *) pwc, s, len, &PL_mbrtowc_ps);
4751 /* Locking prevents races, but locales can be switched out without locking,
4752 * so this isn't a cure all */
4755 retval = mbtowc((wchar_t *) pwc, s, len);
4767 =for apidoc Perl_localeconv
4769 This is a thread-safe version of the libc L<localeconv(3)>. It is the same as
4770 L<POSIX::localeconv|POSIX/localeconv> (returning a hash of the C<localeconv()>
4771 fields), but directly callable from XS code.
4777 Perl_localeconv(pTHX)
4780 #if ! defined(HAS_LOCALECONV)
4786 return my_localeconv(0);
4792 #if defined(HAS_LOCALECONV)
4795 S_my_localeconv(pTHX_ const int item)
4797 PERL_ARGS_ASSERT_MY_LOCALECONV;
4799 /* This returns a mortalized hash containing all or certain elements
4800 * returned by localeconv(). It is used by Perl_localeconv() and
4801 * POSIX::localeconv() and is thread-safe.
4803 * There are two use cases:
4804 * 1) Called from POSIX::locale_conv(). This returns the lconv structure
4805 * copied to a hash, based on the current underlying locales for
4806 * LC_NUMERIC and LC_MONETARY. An input item==0 signifies this case, or
4807 * on many platforms it is the only use case compiled.
4808 * 2) Certain items that nl_langinfo() provides are also derivable from
4809 * the return of localeconv(). Windows notably doesn't have
4810 * nl_langinfo(), so on that, and actually any platform lacking it,
4811 * my_localeconv() is used also to emulate it for those particular
4812 * items. The code to do this is compiled only on such platforms.
4813 * Rather than going to the expense of creating a full hash when only
4814 * one item is needed, the returned hash has just the desired item in
4817 * To access all the localeconv() struct lconv fields, there is a data
4818 * structure that contains every commonly documented field in it. (Maybe
4819 * some minority platforms have extra fields. Those could be added here
4820 * without harm; they would just be ignored on platforms lacking them.)
4822 * Our structure is compiled to make looping through the fields easier by
4823 * pointing each name to its value's offset within lconv, e.g.,
4824 { "thousands_sep", STRUCT_OFFSET(struct lconv, thousands_sep) }
4826 # define LCONV_ENTRY(name) \
4827 {STRINGIFY(name), STRUCT_OFFSET(struct lconv, name)}
4829 /* These synonyms are just for clarity, and to make it easier in case
4830 * something needs to change in the future */
4831 # define LCONV_NUMERIC_ENTRY(name) LCONV_ENTRY(name)
4832 # define LCONV_MONETARY_ENTRY(name) LCONV_ENTRY(name)
4834 /* There are just a few fields for NUMERIC strings */
4835 const lconv_offset_t lconv_numeric_strings[] = {
4836 # ifndef NO_LOCALECONV_GROUPING
4837 LCONV_NUMERIC_ENTRY(grouping),
4839 LCONV_NUMERIC_ENTRY(thousands_sep),
4840 LCONV_NUMERIC_ENTRY(decimal_point),
4844 /* When used to implement nl_langinfo(), we save time by only populating
4845 * the hash with the field(s) needed. Thus we would need a data structure
4847 * LCONV_NUMERIC_ENTRY(decimal_point),
4850 * By placing the decimal_point field last in the full structure, we can
4851 * use just the tail for this bit of it, saving space. This macro yields
4852 * the address of the sub structure. */
4853 # define DECIMAL_POINT_ADDRESS \
4854 &lconv_numeric_strings[(C_ARRAY_LENGTH(lconv_numeric_strings) - 2)]
4856 /* And the MONETARY string fields */
4857 const lconv_offset_t lconv_monetary_strings[] = {
4858 LCONV_MONETARY_ENTRY(int_curr_symbol),
4859 LCONV_MONETARY_ENTRY(mon_decimal_point),
4860 # ifndef NO_LOCALECONV_MON_THOUSANDS_SEP
4861 LCONV_MONETARY_ENTRY(mon_thousands_sep),
4863 # ifndef NO_LOCALECONV_MON_GROUPING
4864 LCONV_MONETARY_ENTRY(mon_grouping),
4866 LCONV_MONETARY_ENTRY(positive_sign),
4867 LCONV_MONETARY_ENTRY(negative_sign),
4868 LCONV_MONETARY_ENTRY(currency_symbol),
4872 /* Like above, this field being last can be used as a sub structure */
4873 # define CURRENCY_SYMBOL_ADDRESS \
4874 &lconv_monetary_strings[(C_ARRAY_LENGTH(lconv_monetary_strings) - 2)]
4876 /* Finally there are integer fields, all are for monetary purposes */
4877 const lconv_offset_t lconv_integers[] = {
4878 LCONV_ENTRY(int_frac_digits),
4879 LCONV_ENTRY(frac_digits),
4880 LCONV_ENTRY(p_sep_by_space),
4881 LCONV_ENTRY(n_cs_precedes),
4882 LCONV_ENTRY(n_sep_by_space),
4883 LCONV_ENTRY(p_sign_posn),
4884 LCONV_ENTRY(n_sign_posn),
4885 # ifdef HAS_LC_MONETARY_2008
4886 LCONV_ENTRY(int_p_cs_precedes),
4887 LCONV_ENTRY(int_p_sep_by_space),
4888 LCONV_ENTRY(int_n_cs_precedes),
4889 LCONV_ENTRY(int_n_sep_by_space),
4890 LCONV_ENTRY(int_p_sign_posn),
4891 LCONV_ENTRY(int_n_sign_posn),
4893 LCONV_ENTRY(p_cs_precedes),
4897 /* Like above, this field being last can be used as a sub structure */
4898 # define P_CS_PRECEDES_ADDRESS \
4899 &lconv_integers[(C_ARRAY_LENGTH(lconv_integers) - 2)]
4901 /* If we aren't paying attention to a given category, use LC_CTYPE instead;
4902 * If not paying attention to that either, the code below should end up not
4903 * using this. Make sure that things blow up if that avoidance gets lost,
4904 * by setting the category to an out-of-bounds value */
4905 locale_category_index numeric_index;
4906 locale_category_index monetary_index;
4908 # ifdef USE_LOCALE_NUMERIC
4909 numeric_index = LC_NUMERIC_INDEX_;
4910 # elif defined(USE_LOCALE_CTYPE)
4911 numeric_index = LC_CTYPE_INDEX_;
4913 numeric_index = LC_ALL_INDEX_; /* Out-of-bounds */
4915 # ifdef USE_LOCALE_MONETARY
4916 monetary_index = LC_MONETARY_INDEX_;
4917 # elif defined(USE_LOCALE_CTYPE)
4918 monetary_index = LC_CTYPE_INDEX_;
4920 monetary_index = LC_ALL_INDEX_; /* Out-of-bounds */
4923 /* Some platforms, for correct non-mojibake results, require LC_CTYPE's
4924 * locale to match LC_NUMERIC's for the numeric fields, and LC_MONETARY's
4925 * for the monetary ones. What happens if LC_NUMERIC and LC_MONETARY
4926 * aren't compatible? Wrong results. To avoid that, we call localeconv()
4927 * twice, once for each locale, setting LC_CTYPE to match the category.
4928 * But if the locales of both categories are the same, there is no need for
4929 * a second call. Assume this is the case unless overridden below */
4930 bool requires_2nd_localeconv = false;
4932 /* The actual hash populating is done by S_populate_hash_from_localeconv().
4933 * It gets passed an array of length two containing the data structure it
4934 * is supposed to use to get the key names to fill the hash with. One
4935 * element is always for the NUMERIC strings (or NULL if none to use), and
4936 * the other element similarly for the MONETARY ones. */
4937 # define NUMERIC_STRING_OFFSET 0
4938 # define MONETARY_STRING_OFFSET 1
4939 const lconv_offset_t * strings[2] = { NULL, NULL };
4941 /* This is a mask, with one bit to tell S_populate_hash_from_localeconv to
4942 * populate the NUMERIC items; another bit for the MONETARY ones. This way
4943 * it can choose which (or both) to populate from */
4946 /* This converts from a locale index to its bit position in the above mask.
4948 # define INDEX_TO_BIT(i) (1 << (i))
4950 /* The two categories can have disparate locales. Initialize them to C and
4951 * override later whichever one(s) we pay attention to */
4952 const char * numeric_locale = "C";
4953 const char * monetary_locale = "C";
4955 /* This will be either 'numeric_locale' or 'monetary_locale' depending on
4956 * what we are working on at the moment */
4957 const char * locale;
4959 /* The LC_MONETARY category also has some integer-valued fields, whose
4960 * information is kept in a separate list */
4961 const lconv_offset_t * integers;
4963 # ifdef HAS_SOME_LANGINFO
4965 /* If the only use-case for this is the full localeconv(), the 'item'
4966 * parameter is ignored. */
4967 PERL_UNUSED_ARG(item);
4971 /* This only gets compiled for the use-case of using localeconv() to
4972 * emulate an nl_langinfo() missing from the platform. */
4974 # ifdef USE_LOCALE_NUMERIC
4976 /* We need this substructure to only return this field for the THOUSEP
4977 * item. The other items also need substructures, but they were handled
4978 * above by placing the substructure's item at the end of the full one, so
4979 * the data structure could do double duty. However, both this and
4980 * RADIXCHAR would need to be in the final position of the same full
4981 * structure; an impossibility. So make this into a separate structure */
4982 const lconv_offset_t thousands_sep_string[] = {
4983 LCONV_NUMERIC_ENTRY(thousands_sep),
4989 /* End of all the initialization of data structures. Now for actual code.
4991 * Without nl_langinfo(), the call to my_localeconv() could be for just one
4992 * of the following 3 items to emulate nl_langinfo(). This is compiled
4993 * only when using perl_langinfo.h, which we control, and it has been
4994 * constructed so that no item is numbered 0.
4996 * For each, set up the appropriate parameters for the call below to
4997 * S_populate_hash_from_localeconv() */
4998 if (item != 0) switch (item) {
5000 locale_panic_(Perl_form(aTHX_
5001 "Unexpected item passed to my_localeconv: %d", item));
5004 # ifdef USE_LOCALE_NUMERIC
5007 locale = numeric_locale = PL_numeric_name;
5008 index_bits = INDEX_TO_BIT(LC_NUMERIC_INDEX_);
5009 strings[NUMERIC_STRING_OFFSET] = DECIMAL_POINT_ADDRESS;
5014 index_bits = INDEX_TO_BIT(LC_NUMERIC_INDEX_);
5015 locale = numeric_locale = PL_numeric_name;
5016 strings[NUMERIC_STRING_OFFSET] = thousands_sep_string;
5021 # ifdef USE_LOCALE_MONETARY
5024 index_bits = INDEX_TO_BIT(LC_MONETARY_INDEX_);
5025 locale = monetary_locale = querylocale_i(LC_MONETARY_INDEX_);
5027 /* This item needs the values for both the currency symbol, and another
5028 * one used to construct the nl_langino()-compatible return */
5029 strings[MONETARY_STRING_OFFSET] = CURRENCY_SYMBOL_ADDRESS;
5030 integers = P_CS_PRECEDES_ADDRESS;
5035 } /* End of switch() */
5037 else /* End of for just one item to emulate nl_langinfo() */
5041 { /* Here, the call is for all of localeconv(). It has a bunch of
5042 * items. As in the individual item case, set up the parameters for
5043 * S_populate_hash_from_localeconv(); */
5045 # ifdef USE_LOCALE_NUMERIC
5046 numeric_locale = PL_numeric_name;
5047 # elif defined(USE_LOCALE_CTYPE)
5048 numeric_locale = querylocale_i(numeric_index);
5050 # if defined(USE_LOCALE_MONETARY) || defined(USE_LOCALE_CTYPE)
5051 monetary_locale = querylocale_i(monetary_index);
5054 /* The first call to S_populate_hash_from_localeconv() will be for the
5055 * MONETARY values */
5056 index_bits = INDEX_TO_BIT(monetary_index);
5057 locale = monetary_locale;
5059 /* And if the locales for the two categories are the same, we can also
5060 * do the NUMERIC values in the same call */
5061 if (strEQ(numeric_locale, monetary_locale)) {
5062 index_bits |= INDEX_TO_BIT(numeric_index);
5065 requires_2nd_localeconv = true;
5068 /* We always pass both sets of strings. 'index_bits' tells
5069 * S_populate_hash_from_localeconv which to actually look at */
5070 strings[NUMERIC_STRING_OFFSET] = lconv_numeric_strings;
5071 strings[MONETARY_STRING_OFFSET] = lconv_monetary_strings;
5073 /* And pass the integer values to populate; again 'index_bits' will
5074 * say to use them or not */
5075 integers = lconv_integers;
5077 } /* End of call is for localeconv() */
5079 /* The code above has determined the parameters to
5080 S_populate_hash_from_localeconv() for both cases of an individual item
5081 and for the entire structure. Below is code common to both */
5083 HV * hv = newHV(); /* The returned hash, initially empty */
5084 sv_2mortal((SV*)hv);
5086 /* Call localeconv() and copy its results into the hash. All the
5087 * parameters have been initialized above */
5088 populate_hash_from_localeconv(hv,
5095 /* The above call may have done all the hash fields, but not always, as
5096 * already explained. If we need a second call it is always for the
5098 if (requires_2nd_localeconv) {
5099 populate_hash_from_localeconv(hv,
5101 INDEX_TO_BIT(numeric_index),
5103 NULL /* There are no NUMERIC integer
5108 /* Here, the hash has been completely populated.
5110 * Now go through all the items and:
5111 * a) For string items, see if they should be marked as UTF-8 or not.
5112 * This would have been more convenient and faster to do while
5113 * populating the hash in the first place, but that operation has to be
5114 * done within a critical section, keeping other threads from
5115 * executing, so only the minimal amount of work necessary is done at
5117 * b) For integer items, convert the C CHAR_MAX value into -1. Again,
5118 * this could have been done in the critical section, but was deferred
5119 * to here to keep to the bare minimum amount the time spent owning the
5120 * processor. CHAR_MAX is a C concept for an 8-bit character type.
5121 * Perl has no such type; the closest fit is a -1.
5123 * XXX On unthreaded perls, this code could be #ifdef'd out, and the
5124 * corrections determined at hash population time, at an extra maintenance
5125 * cost which khw doesn't think is worth it
5127 for (unsigned int i = 0; i < 2; i++) { /* Try both types of strings */
5128 if (! strings[i]) { /* Skip if no strings of this type */
5132 locale = (i == NUMERIC_STRING_OFFSET)
5136 locale_utf8ness_t locale_is_utf8 = LOCALE_UTF8NESS_UNKNOWN;
5138 # ifdef HAS_RELIABLE_UTF8NESS_DETERMINATION
5140 /* It saves time in the loop below to have predetermined the UTF8ness
5141 * of the locale. But only do so if the platform reliably has this
5142 * information; otherwise it's better to do it only it should become
5143 * necessary, which happens on a per-element basis in the loop. */
5145 locale_is_utf8 = (is_locale_utf8(locale))
5149 if (locale_is_utf8 == LOCALE_NOT_UTF8) {
5150 continue; /* No string can be UTF-8 if the locale isn't */
5155 /* Examine each string */
5157 const char * name = strings[i]->name;
5159 if (! name) { /* Reached the end */
5163 /* 'value' will contain the string that may need to be marked as
5165 SV ** value = hv_fetch(hv, name, strlen(name), true);
5170 /* Determine if the string should be marked as UTF-8. */
5171 if (UTF8NESS_YES == (get_locale_string_utf8ness_i(SvPVX(*value),
5174 (locale_category_index) 0)))
5179 strings[i]++; /* Iterate */
5181 } /* End of fixing up UTF8ness */
5184 /* Examine each integer */
5185 if (integers) while (1) {
5186 const char * name = integers->name;
5188 if (! name) { /* Reached the end */
5192 SV ** value = hv_fetch(hv, name, strlen(name), true);
5197 /* Change CHAR_MAX to -1 */
5198 if (SvIV(*value) == CHAR_MAX) {
5199 sv_setiv(*value, -1);
5202 integers++; /* Iterate */
5209 S_populate_hash_from_localeconv(pTHX_ HV * hv,
5211 /* Switch to this locale to run
5212 * localeconv() from */
5213 const char * locale,
5215 /* bit mask of which categories to
5217 const U32 which_mask,
5219 /* strings[0] points to the numeric
5220 * string fields; [1] to the monetary */
5221 const lconv_offset_t * strings[2],
5223 /* And to the monetary integer fields */
5224 const lconv_offset_t * integers)
5226 PERL_ARGS_ASSERT_POPULATE_HASH_FROM_LOCALECONV;
5227 PERL_UNUSED_ARG(which_mask); /* Some configurations don't use this;
5228 complicated to figure out which */
5230 PERL_UNUSED_ARG(locale);
5233 /* Run localeconv() and copy some or all of its results to the input 'hv'
5234 * hash. Most localeconv() implementations return the values in a global
5235 * static buffer, so the operation must be performed in a critical section,
5236 * ending only after the copy is completed. There are so many locks
5237 * because localeconv() deals with two categories, and returns in a single
5238 * global static buffer. Some locks might be no-ops on this platform, but
5239 * not others. We need to lock if any one isn't a no-op. */
5241 # ifdef USE_LOCALE_CTYPE
5243 /* Some platforms require LC_CTYPE to be congruent with the category we are
5245 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
5248 # ifdef USE_LOCALE_NUMERIC
5250 /* We need to toggle to the underlying NUMERIC locale if we are getting
5251 * NUMERIC strings */
5252 const char * orig_NUMERIC_locale = NULL;
5253 if (which_mask & INDEX_TO_BIT(LC_NUMERIC_INDEX_)) {
5258 /* There is a bug in Windows in which setting LC_CTYPE after the others
5259 * doesn't actually take effect for localeconv(). See commit
5260 * 418efacd1950763f74ed3cc22f8cf9206661b892 for details. Thus we have
5261 * to make sure that the locale we want is set after LC_CTYPE. We
5262 * unconditionally toggle away from and back to the current locale
5263 * prior to calling localeconv().
5265 * This code will have no effect if we already are in C, but khw
5266 * hasn't seen any cases where this causes problems when we are in the
5268 orig_NUMERIC_locale = toggle_locale_i(LC_NUMERIC_INDEX_, "C");
5269 toggle_locale_i(LC_NUMERIC_INDEX_, locale);
5273 /* No need for the extra toggle when not on Windows */
5274 orig_NUMERIC_locale = toggle_locale_i(LC_NUMERIC_INDEX_, locale);
5281 # if defined(USE_LOCALE_MONETARY) && defined(WIN32)
5283 /* Same Windows bug as described just above for NUMERIC. Otherwise, no
5284 * need to toggle LC_MONETARY, as it is kept in the underlying locale */
5285 const char * orig_MONETARY_locale = NULL;
5286 if (which_mask & INDEX_TO_BIT(LC_MONETARY_INDEX_)) {
5287 orig_MONETARY_locale = toggle_locale_i(LC_MONETARY_INDEX_, "C");
5288 toggle_locale_i(LC_MONETARY_INDEX_, locale);
5293 /* Finally ready to do the actual localeconv(). Lock to prevent other
5294 * accesses until we have made a copy of its returned static buffer */
5297 # if defined(TS_W32_BROKEN_LOCALECONV) && defined(USE_THREAD_SAFE_LOCALE)
5299 /* This is a workaround for another bug in Windows. localeconv() was
5300 * broken with thread-safe locales prior to VS 15. It looks at the global
5301 * locale instead of the thread one. As a work-around, we toggle to the
5302 * global locale; populate the return; then toggle back. We have to use
5303 * LC_ALL instead of the individual categories because of yet another bug
5304 * in Windows. And this all has to be done in a critical section.
5306 * This introduces a potential race with any other thread that has also
5307 * converted to use the global locale, and doesn't protect its locale calls
5308 * with mutexes. khw can't think of any reason for a thread to do so on
5309 * Windows, as the locale API is the same regardless of thread-safety,
5310 * except if the code is ported from working on another platform where
5311 * there might be some reason to do this. But this is typically due to
5312 * some alien-to-Perl library that thinks it owns locale setting. Such a
5313 * library isn't likely to exist on Windows, so such an application is
5314 * unlikely to be run on Windows
5316 bool restore_per_thread = FALSE;
5318 /* Save the per-thread locale state */
5319 const char * save_thread = querylocale_c(LC_ALL);
5321 /* Change to the global locale, and note if we already were there */
5322 int config_return = _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
5323 if (config_return != _DISABLE_PER_THREAD_LOCALE) {
5324 if (config_return == -1) {
5325 locale_panic_("_configthreadlocale returned an error");
5328 restore_per_thread = TRUE;
5331 /* Save the state of the global locale; then convert to our desired
5333 const char * save_global = querylocale_c(LC_ALL);
5334 void_setlocale_c(LC_ALL, save_thread);
5336 # endif /* TS_W32_BROKEN_LOCALECONV */
5338 /* Finally, do the actual localeconv */
5339 const char *lcbuf_as_string = (const char *) localeconv();
5341 /* Fill in the string fields of the HV* */
5342 for (unsigned int i = 0; i < 2; i++) {
5344 /* One iteration is only for the numeric string fields. Skip these
5345 * unless we are compiled to care about those fields and the input
5346 * parameters indicate we want their values */
5347 if ( i == NUMERIC_STRING_OFFSET
5349 # ifdef USE_LOCALE_NUMERIC
5351 && (which_mask & INDEX_TO_BIT(LC_NUMERIC_INDEX_)) == 0
5359 /* The other iteration is only for the monetary string fields. Again
5360 * skip it unless we want those values */
5361 if ( i == MONETARY_STRING_OFFSET
5363 # ifdef USE_LOCALE_MONETARY
5365 && (which_mask & INDEX_TO_BIT(LC_MONETARY_INDEX_)) == 0
5373 /* For each field for the given category ... */
5374 const lconv_offset_t * category_strings = strings[i];
5376 const char * name = category_strings->name;
5377 if (! name) { /* Quit at the end */
5381 /* we have set things up so that we know where in the returned
5382 * structure, when viewed as a string, the corresponding value is.
5384 const char *value = *((const char **)( lcbuf_as_string
5385 + category_strings->offset));
5387 /* Set to get next string on next iteration */
5390 /* Skip if this platform doesn't have this field. */
5395 /* Copy to the hash */
5398 newSVpv(value, strlen(value)),
5402 /* Add any int fields to the HV* */
5403 if (i == MONETARY_STRING_OFFSET && integers) {
5404 while (integers->name) {
5405 const char value = *((const char *)( lcbuf_as_string
5406 + integers->offset));
5407 (void) hv_store(hv, integers->name,
5408 strlen(integers->name), newSViv(value), 0);
5412 } /* End of loop through the fields */
5414 /* Done with copying to the hash. Can unwind the critical section locks */
5416 # if defined(TS_W32_BROKEN_LOCALECONV) && defined(USE_THREAD_SAFE_LOCALE)
5418 /* Restore the global locale's prior state */
5419 void_setlocale_c(LC_ALL, save_global);
5421 /* And back to per-thread locales */
5422 if (restore_per_thread) {
5423 if (_configthreadlocale(_ENABLE_PER_THREAD_LOCALE) == -1) {
5424 locale_panic_("_configthreadlocale returned an error");
5428 /* Restore the per-thread locale state */
5429 void_setlocale_c(LC_ALL, save_thread);
5431 # endif /* TS_W32_BROKEN_LOCALECONV */
5433 gwLOCALE_UNLOCK; /* Finished with the critical section of a
5434 globally-accessible buffer */
5436 # if defined(USE_LOCALE_MONETARY) && defined(WIN32)
5438 restore_toggled_locale_i(LC_MONETARY_INDEX_, orig_MONETARY_locale);
5441 # ifdef USE_LOCALE_NUMERIC
5443 restore_toggled_locale_i(LC_NUMERIC_INDEX_, orig_NUMERIC_locale);
5444 if (which_mask & INDEX_TO_BIT(LC_NUMERIC_INDEX_)) {
5449 # ifdef USE_LOCALE_CTYPE
5451 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
5457 #endif /* defined(HAS_LOCALECONV) */
5458 #ifndef HAS_SOME_LANGINFO
5460 typedef int nl_item; /* Substitute 'int' for emulated nl_langinfo() */
5466 =for apidoc Perl_langinfo
5467 =for apidoc_item Perl_langinfo8
5469 C<Perl_langinfo> is an (almost) drop-in replacement for the system
5470 C<L<nl_langinfo(3)>>, taking the same C<item> parameter values, and returning
5471 the same information. But it is more thread-safe than regular
5472 C<nl_langinfo()>, and hides the quirks of Perl's locale handling from your
5473 code, and can be used on systems that lack a native C<nl_langinfo>.
5475 However, you should instead use the improved version of this:
5476 L</Perl_langinfo8>, which behaves identically except for an additional
5477 parameter, a pointer to a variable declared as L</C<utf8ness_t>>, into which it
5478 returns to you how you should treat the returned string with regards to it
5479 being encoded in UTF-8 or not.
5481 Concerning the differences between these and plain C<nl_langinfo()>:
5487 C<Perl_langinfo8> has an extra parameter, described above. Besides this, the
5488 other reason they aren't quite a drop-in replacement is actually an advantage.
5489 The C<const>ness of the return allows the compiler to catch attempts to write
5490 into the returned buffer, which is illegal and could cause run-time crashes.
5494 They deliver the correct results for the C<RADIXCHAR> and C<THOUSEP> items,
5495 without you having to write extra code. The reason for the extra code would be
5496 because these are from the C<LC_NUMERIC> locale category, which is normally
5497 kept set by Perl so that the radix is a dot, and the separator is the empty
5498 string, no matter what the underlying locale is supposed to be, and so to get
5499 the expected results, you have to temporarily toggle into the underlying
5500 locale, and later toggle back. (You could use plain C<nl_langinfo> and
5501 C<L</STORE_LC_NUMERIC_FORCE_TO_UNDERLYING>> for this but then you wouldn't get
5502 the other advantages of C<Perl_langinfo()>; not keeping C<LC_NUMERIC> in the C
5503 (or equivalent) locale would break a lot of CPAN, which is expecting the radix
5504 (decimal point) character to be a dot.)
5508 The system function they replace can have its static return buffer trashed,
5509 not only by a subsequent call to that function, but by a C<freelocale>,
5510 C<setlocale>, or other locale change. The returned buffer of these functions
5511 is not changed until the next call to one or the other, so the buffer is never
5516 The return buffer is per-thread, so it also is never overwritten by a call to
5517 these functions from another thread; unlike the function it replaces.
5521 But most importantly, they work on systems that don't have C<nl_langinfo>, such
5522 as Windows, hence making your code more portable. Of the fifty-some possible
5523 items specified by the POSIX 2008 standard,
5524 L<http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/langinfo.h.html>,
5525 only one is completely unimplemented, though on non-Windows platforms, another
5526 significant one is not fully implemented). They use various techniques to
5527 recover the other items, including calling C<L<localeconv(3)>>, and
5528 C<L<strftime(3)>>, both of which are specified in C89, so should be always be
5529 available. Later C<strftime()> versions have additional capabilities.
5530 If an item is not available on your system, this returns either the value
5531 associated with the C locale, or simply C<"">, whichever is more appropriate.
5533 It is important to note that, when called with an item that is recovered by
5534 using C<localeconv>, the buffer from any previous explicit call to
5535 C<L<localeconv(3)>> will be overwritten. But you shouldn't be using
5536 C<localeconv> anyway because it is is very much not thread-safe, and suffers
5537 from the same problems outlined in item 'b.' above for the fields it returns
5538 that are controlled by the LC_NUMERIC locale category. Instead, avoid all of
5539 those problems by calling L</Perl_localeconv>, which is thread-safe; or by
5540 using the methods given in L<perlcall> to call
5541 L<C<POSIX::localeconv()>|POSIX/localeconv>, which is also thread-safe.
5545 The details for those items which may deviate from what this emulation returns
5546 and what a native C<nl_langinfo()> would return are specified in
5549 When using C<Perl_langinfo8> (or plain C<Perl_langinfo>) on systems that don't
5550 have a native C<nl_langinfo()>, you must
5552 #include "perl_langinfo.h"
5554 before the C<perl.h> C<#include>. You can replace your F<langinfo.h>
5555 C<#include> with this one. (Doing it this way keeps out the symbols that plain
5556 F<langinfo.h> would try to import into the namespace for code that doesn't need
5564 Perl_langinfo(const nl_item item)
5566 return Perl_langinfo8(item, NULL);
5570 Perl_langinfo8(const nl_item item, utf8ness_t * utf8ness)
5573 locale_category_index cat_index;
5575 PERL_ARGS_ASSERT_PERL_LANGINFO8;
5577 if (utf8ness) { /* Assume for now */
5578 *utf8ness = UTF8NESS_IMMATERIAL;
5581 /* Find the locale category that controls the input 'item'. If we are not
5582 * paying attention to that category, instead return a default value. Also
5583 * return the default value if there is no way for us to figure out the
5584 * correct value. If we have some form of nl_langinfo(), we can always
5585 * figure it out, but lacking that, there may be alternative methods that
5586 * can be used to recover most of the possible items. Some of those
5587 * methods need libc functions, which may or may not be available. If
5588 * unavailable, we can't compute the correct value, so must here return the
5594 #ifdef USE_LOCALE_CTYPE
5596 cat_index = LC_CTYPE_INDEX_;
5602 #if defined(USE_LOCALE_MESSAGES) && defined(HAS_SOME_LANGINFO)
5604 case YESEXPR: case YESSTR: case NOEXPR: case NOSTR:
5605 cat_index = LC_MESSAGES_INDEX_;
5608 case YESEXPR: return "^[+1yY]";
5609 case YESSTR: return "yes";
5610 case NOEXPR: return "^[-0nN]";
5611 case NOSTR: return "no";
5616 #if defined(USE_LOCALE_MONETARY) \
5617 && (defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV))
5619 cat_index = LC_MONETARY_INDEX_;
5627 #ifdef CAN_CALCULATE_RADIX
5629 cat_index = LC_NUMERIC_INDEX_;
5632 return C_decimal_point;
5637 #if defined(USE_LOCALE_NUMERIC) \
5638 && (defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV))
5640 cat_index = LC_NUMERIC_INDEX_;
5643 return C_thousands_sep;
5646 /* The other possible items are all in LC_TIME. */
5647 #ifdef USE_LOCALE_TIME
5650 cat_index = LC_TIME_INDEX_;
5654 #if ! defined(USE_LOCALE_TIME) || ! defined(HAS_SOME_LANGINFO)
5656 /* If not using LC_TIME, hard code the rest. Or, if there is no
5657 * nl_langinfo(), we use strftime() as an alternative, and it is missing
5658 * functionality to get every single one, so hard-code those */
5660 case ERA: return ""; /* Unimplemented; for use with strftime() %E
5663 /* These formats are defined by C89, so we assume that strftime supports
5664 * them, and so are returned unconditionally; they may not be what the
5665 * locale actually says, but should give good enough results for someone
5666 * using them as formats (as opposed to trying to parse them to figure
5667 * out what the locale says). The other format items are actually tested
5668 * to verify they work on the platform */
5669 case D_FMT: return "%x";
5670 case T_FMT: return "%X";
5671 case D_T_FMT: return "%c";
5673 # if defined(WIN32) || ! defined(USE_LOCALE_TIME)
5675 /* strftime() on Windows doesn't have the POSIX (beyond C89) extensions
5676 * that would allow it to recover these */
5677 case ERA_D_FMT: return "%x";
5678 case ERA_T_FMT: return "%X";
5679 case ERA_D_T_FMT: return "%c";
5680 case ALT_DIGITS: return "0";
5683 # ifndef USE_LOCALE_TIME
5685 case T_FMT_AMPM: return "%r";
5686 case ABDAY_1: return "Sun";
5687 case ABDAY_2: return "Mon";
5688 case ABDAY_3: return "Tue";
5689 case ABDAY_4: return "Wed";
5690 case ABDAY_5: return "Thu";
5691 case ABDAY_6: return "Fri";
5692 case ABDAY_7: return "Sat";
5693 case AM_STR: return "AM";
5694 case PM_STR: return "PM";
5695 case ABMON_1: return "Jan";
5696 case ABMON_2: return "Feb";
5697 case ABMON_3: return "Mar";
5698 case ABMON_4: return "Apr";
5699 case ABMON_5: return "May";
5700 case ABMON_6: return "Jun";
5701 case ABMON_7: return "Jul";
5702 case ABMON_8: return "Aug";
5703 case ABMON_9: return "Sep";
5704 case ABMON_10: return "Oct";
5705 case ABMON_11: return "Nov";
5706 case ABMON_12: return "Dec";
5707 case DAY_1: return "Sunday";
5708 case DAY_2: return "Monday";
5709 case DAY_3: return "Tuesday";
5710 case DAY_4: return "Wednesday";
5711 case DAY_5: return "Thursday";
5712 case DAY_6: return "Friday";
5713 case DAY_7: return "Saturday";
5714 case MON_1: return "January";
5715 case MON_2: return "February";
5716 case MON_3: return "March";
5717 case MON_4: return "April";
5718 case MON_5: return "May";
5719 case MON_6: return "June";
5720 case MON_7: return "July";
5721 case MON_8: return "August";
5722 case MON_9: return "September";
5723 case MON_10: return "October";
5724 case MON_11: return "November";
5725 case MON_12: return "December";
5730 } /* End of switch on item */
5734 Perl_croak_nocontext("panic: Unexpected nl_langinfo() item %d", item);
5735 NOT_REACHED; /* NOTREACHED */
5736 PERL_UNUSED_VAR(cat_index);
5740 return my_langinfo_i(item, cat_index, query_nominal_locale_i(cat_index),
5741 &PL_langinfo_buf, &PL_langinfo_bufsize, utf8ness);
5747 Perl_my_strftime(pTHX_ const char *fmt, int sec, int min, int hour,
5748 int mday, int mon, int year, int wday, int yday,
5750 { /* Documented above */
5751 PERL_ARGS_ASSERT_MY_STRFTIME;
5754 ints_to_tm(&mytm, sec, min, hour, mday, mon, year, wday, yday, isdst);
5755 char * ret = strftime_tm(fmt, &mytm);
5760 Perl_sv_strftime_tm(pTHX_ SV * fmt, const struct tm * mytm)
5761 { /* Documented above */
5762 PERL_ARGS_ASSERT_SV_STRFTIME_TM;
5764 utf8ness_t fmt_utf8ness = (SvUTF8(fmt) && LIKELY(! IN_BYTES))
5768 utf8ness_t result_utf8ness;
5769 char * retval = strftime8(SvPV_nolen(fmt),
5773 true /* calling from sv_strftime */
5777 STRLEN len = strlen(retval);
5779 sv_usepvn_flags(sv, retval, len, SV_HAS_TRAILING_NUL);
5781 if (result_utf8ness == UTF8NESS_YES) {
5790 Perl_sv_strftime_ints(pTHX_ SV * fmt, int sec, int min, int hour,
5791 int mday, int mon, int year, int wday,
5792 int yday, int isdst)
5793 { /* Documented above */
5794 PERL_ARGS_ASSERT_SV_STRFTIME_INTS;
5797 ints_to_tm(&mytm, sec, min, hour, mday, mon, year, wday, yday, isdst);
5798 SV * ret = sv_strftime_tm(fmt, &mytm);
5804 /* There are several implementations of my_langinfo, depending on the
5805 * Configuration. They all share the same beginning of the function */
5807 S_my_langinfo_i(pTHX_
5808 const nl_item item, /* The item to look up */
5809 const locale_category_index cat_index, /* The locale category
5811 /* The locale to look up 'item' in. */
5812 const char * locale,
5814 /* Where to store the result, and where the size of that buffer
5815 * is stored, updated on exit. retbuf_sizep may be NULL for an
5816 * empty-on-entry, single use buffer whose size we don't need
5817 * to keep track of */
5819 Size_t * retbuf_sizep,
5821 /* If not NULL, the location to store the UTF8-ness of 'item's
5822 * value, as documented */
5823 utf8ness_t * utf8ness)
5825 const char * retval = NULL;
5827 PERL_ARGS_ASSERT_MY_LANGINFO_I;
5828 assert(cat_index < LC_ALL_INDEX_);
5830 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
5831 "Entering my_langinfo item=%ld, using locale %s\n",
5832 (long) item, locale));
5833 /*--------------------------------------------------------------------------*/
5834 /* Above is the common beginning to all the implementations of my_langinfo().
5835 * Below are the various completions.
5837 * Some platforms don't deal well with non-ASCII strings in locale X when
5838 * LC_CTYPE is not in X. (Actually it is probably when X is UTF-8 and LC_CTYPE
5839 * isn't, or vice versa). There is explicit code to bring the categories into
5840 * sync. This doesn't seem to be a problem with nl_langinfo(), so that
5841 * implementation doesn't currently worry about it. But it is a problem on
5842 * Windows boxes, which don't have nl_langinfo(). */
5844 /*--------------------------------------------------------------------------*/
5845 # if defined(HAS_NL_LANGINFO) /* nl_langinfo() is available. */
5846 # ifdef USE_LOCALE_CTYPE
5848 /* This function sorts out if things actually have to be switched or not,
5849 * for both save and restore. */
5850 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
5854 const char * orig_switched_locale = toggle_locale_i(cat_index, locale);
5857 retval = save_to_buffer(nl_langinfo(item), retbufp, retbuf_sizep);
5861 *utf8ness = get_locale_string_utf8ness_i(retval,
5862 LOCALE_UTF8NESS_UNKNOWN,
5866 restore_toggled_locale_i(cat_index, orig_switched_locale);
5868 # ifdef USE_LOCALE_CTYPE
5870 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
5875 /*--------------------------------------------------------------------------*/
5876 # else /* Below, emulate nl_langinfo as best we can */
5878 /* The other completion is where we have to emulate nl_langinfo(). There
5879 * are various possibilities depending on the Configuration */
5881 # ifdef USE_LOCALE_CTYPE
5883 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
5887 const char * orig_switched_locale = toggle_locale_i(cat_index, locale);
5889 /* Here, we are in the locale we want information about */
5891 /* Almost all the items will have ASCII return values. Set that here, and
5892 * override if necessary */
5893 utf8ness_t is_utf8 = UTF8NESS_IMMATERIAL;
5897 assert(item < 0); /* Make sure using perl_langinfo.h */
5903 # if defined(HAS_SNPRINTF) \
5904 && (! defined(HAS_LOCALECONV) || defined(TS_W32_BROKEN_LOCALECONV))
5907 /* snprintf() can be used to find the radix character by outputting
5908 * a known simple floating point number to a buffer, and parsing
5909 * it, inferring the radix as the bytes separating the integer and
5910 * fractional parts. But localeconv() is more direct, not
5911 * requiring inference, so use it instead of the code just below,
5912 * if (likely) it is available and works ok */
5914 char * floatbuf = NULL;
5915 const Size_t initial_size = 10;
5917 Newx(floatbuf, initial_size, char);
5919 /* 1.5 is exactly representable on binary computers */
5920 Size_t needed_size = snprintf(floatbuf, initial_size, "%.1f", 1.5);
5922 /* If our guess wasn't big enough, increase and try again, based on
5923 * the real number that snprintf() is supposed to return */
5924 if (UNLIKELY(needed_size >= initial_size)) {
5925 needed_size++; /* insurance */
5926 Renew(floatbuf, needed_size, char);
5927 Size_t new_needed = snprintf(floatbuf, needed_size, "%.1f", 1.5);
5928 assert(new_needed <= needed_size);
5929 needed_size = new_needed;
5932 char * s = floatbuf;
5933 char * e = floatbuf + needed_size;
5936 while (s < e && *s != '1') {
5940 if (LIKELY(s < e)) {
5945 char * item_start = s;
5946 while (s < e && *s != '5') {
5950 /* Everything in between is the radix string */
5951 if (LIKELY(s < e)) {
5953 retval = save_to_buffer(item_start, retbufp, retbuf_sizep);
5957 is_utf8 = get_locale_string_utf8ness_i(retval,
5958 LOCALE_UTF8NESS_UNKNOWN,
5968 # ifdef HAS_LOCALECONV /* snprintf() failed; drop down to use
5973 # else /* snprintf() failed and no localeconv() */
5975 retval = C_decimal_point;
5980 # ifdef HAS_LOCALECONV
5982 /* These items are available from localeconv(). (To avoid using
5983 * TS_W32_BROKEN_LOCALECONV, one could use GetNumberFormat and
5984 * GetCurrencyFormat; patches welcome) */
5986 # define P_CS_PRECEDES "p_cs_precedes"
5987 # define CURRENCY_SYMBOL "currency_symbol"
5989 /* case RADIXCHAR: // May drop down to here in some configurations */
5994 /* The hash gets populated with just the field(s) related to 'item'. */
5995 HV * result_hv = my_localeconv(item);
5998 if (item != CRNCYSTR) {
6000 /* These items have been populated with just one key => value */
6001 (void) hv_iterinit(result_hv);
6002 HE * entry = hv_iternext(result_hv);
6003 string = hv_iterval(result_hv, entry);
6007 /* But CRNCYSTR localeconv() returns a slightly different value
6008 * than the nl_langinfo() API calls for, so have to modify this one
6009 * to conform. We need another value from localeconv() to know
6010 * what to change it to. my_localeconv() has populated the hash
6011 * with exactly both fields. Delete this one, leaving just the
6012 * CRNCYSTR one in the hash */
6013 SV* precedes = hv_delete(result_hv,
6014 P_CS_PRECEDES, STRLENs(P_CS_PRECEDES),
6017 locale_panic_("my_localeconv() unexpectedly didn't return"
6018 " a value for " P_CS_PRECEDES);
6021 /* The modification is to prefix the localeconv() return with a
6022 * single byte, calculated as follows: */
6023 char prefix = (LIKELY(SvIV(precedes) != -1))
6024 ? ((precedes != 0) ? '-' : '+')
6026 /* khw couldn't find any documentation that
6027 * CHAR_MAX (which we modify to -1) is the signal,
6028 * but cygwin uses it thusly, and it makes sense
6029 * given that CHAR_MAX indicates the value isn't
6030 * used, so it neither precedes nor succeeds */
6033 /* Now get CRNCYSTR */
6034 (void) hv_iterinit(result_hv);
6035 HE * entry = hv_iternext(result_hv);
6036 string = hv_iterval(result_hv, entry);
6038 /* And perform the modification */
6039 Perl_sv_setpvf(aTHX_ string, "%c%s", prefix, SvPV_nolen(string));
6042 /* Here, 'string' contains the value we want to return */
6043 retval = save_to_buffer(SvPV_nolen(string), retbufp, retbuf_sizep);
6046 is_utf8 = (SvUTF8(string))
6048 : (is_utf8_invariant_string( (U8 *) retval,
6050 ? UTF8NESS_IMMATERIAL
6058 # endif /* Some form of localeconv */
6059 # ifdef HAS_STRFTIME
6061 /* These formats are only available in later strftime's */
6062 case ERA_D_FMT: case ERA_T_FMT: case ERA_D_T_FMT: case T_FMT_AMPM:
6064 /* The rest can be gotten from most versions of strftime(). */
6065 case ABDAY_1: case ABDAY_2: case ABDAY_3:
6066 case ABDAY_4: case ABDAY_5: case ABDAY_6: case ABDAY_7:
6068 case AM_STR: case PM_STR:
6069 case ABMON_1: case ABMON_2: case ABMON_3: case ABMON_4:
6070 case ABMON_5: case ABMON_6: case ABMON_7: case ABMON_8:
6071 case ABMON_9: case ABMON_10: case ABMON_11: case ABMON_12:
6072 case DAY_1: case DAY_2: case DAY_3: case DAY_4:
6073 case DAY_5: case DAY_6: case DAY_7:
6074 case MON_1: case MON_2: case MON_3: case MON_4:
6075 case MON_5: case MON_6: case MON_7: case MON_8:
6076 case MON_9: case MON_10: case MON_11: case MON_12:
6078 const char * format;
6079 bool return_format = FALSE;
6084 GCC_DIAG_IGNORE_STMT(-Wimplicit-fallthrough);
6088 locale_panic_(Perl_form(aTHX_ "switch case: %d problem", item));
6089 NOT_REACHED; /* NOTREACHED */
6091 case PM_STR: hour = 18;
6095 case ABDAY_7: mday++;
6096 case ABDAY_6: mday++;
6097 case ABDAY_5: mday++;
6098 case ABDAY_4: mday++;
6099 case ABDAY_3: mday++;
6100 case ABDAY_2: mday++;
6113 case ABMON_12: mon++;
6114 case ABMON_11: mon++;
6115 case ABMON_10: mon++;
6116 case ABMON_9: mon++;
6117 case ABMON_8: mon++;
6118 case ABMON_7: mon++;
6119 case ABMON_6: mon++;
6120 case ABMON_5: mon++;
6121 case ABMON_4: mon++;
6122 case ABMON_3: mon++;
6123 case ABMON_2: mon++;
6143 return_format = TRUE;
6147 return_format = TRUE;
6151 return_format = TRUE;
6155 return_format = TRUE;
6158 format = "%Ow"; /* Find the alternate digit for 0 */
6162 GCC_DIAG_RESTORE_STMT;
6164 /* The year was deliberately chosen so that January 1 is on the
6165 * first day of the week. Since we're only getting one thing at a
6166 * time, it all works */
6168 ints_to_tm(&mytm, 30, 30, hour, mday, mon, 2011, 0, 0, 0);
6171 temp = strftime8(format,
6173 UTF8NESS_IMMATERIAL, /* All possible formats
6177 false /* not calling from sv_strftime */
6181 temp = strftime_tm(format, &mytm);
6184 retval = save_to_buffer(temp, retbufp, retbuf_sizep);
6187 /* If the item is 'ALT_DIGITS', '*retbuf' contains the alternate
6188 * format for wday 0. If the value is the same as the normal 0,
6189 * there isn't an alternate, so clear the buffer.
6191 * (wday was chosen because its range is all a single digit.
6192 * Things like tm_sec have two digits as the minimum: '00'.) */
6193 if (item == ALT_DIGITS && strEQ(*retbufp, "0")) {
6198 /* ALT_DIGITS is problematic. Experiments on it showed that
6199 * strftime() did not always work properly when going from alt-9 to
6200 * alt-10. Only a few locales have this item defined, and in all
6201 * of them on Linux that khw was able to find, nl_langinfo() merely
6202 * returned the alt-0 character, possibly doubled. Most Unicode
6203 * digits are in blocks of 10 consecutive code points, so that is
6204 * sufficient information for such scripts, as we can infer alt-1,
6205 * alt-2, .... But for a Japanese locale, a CJK ideographic 0 is
6206 * returned, and the CJK digits are not in code point order, so you
6207 * can't really infer anything. The localedef for this locale did
6208 * specify the succeeding digits, so that strftime() works properly
6209 * on them, without needing to infer anything. But the
6210 * nl_langinfo() return did not give sufficient information for the
6211 * caller to understand what's going on. So until there is
6212 * evidence that it should work differently, this returns the alt-0
6213 * string for ALT_DIGITS. */
6215 if (return_format) {
6217 /* If to return the format, not the value, overwrite the buffer
6218 * with it. But some strftime()s will keep the original format
6219 * if illegal, so change those to "" */
6220 if (strEQ(*retbufp, format)) {
6227 /* A format is always in ASCII */
6228 is_utf8 = UTF8NESS_IMMATERIAL;
6235 # ifdef USE_LOCALE_CTYPE
6239 /* The trivial case */
6240 if (isNAME_C_OR_POSIX(locale)) {
6247 /* This function retrieves the code page. It is subject to change, but
6248 * is documented and has been stable for many releases */
6249 UINT ___lc_codepage_func(void);
6251 # ifndef WIN32_USE_FAKE_OLD_MINGW_LOCALES
6253 retval = save_to_buffer(Perl_form(aTHX_ "%d", ___lc_codepage_func()),
6254 retbufp, retbuf_sizep);
6257 retval = save_to_buffer(nl_langinfo(CODESET),
6258 retbufp, retbuf_sizep);
6261 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "locale='%s' cp=%s\n",
6267 /* The codeset is important, but khw did not figure out a way for it to
6268 * be retrieved on non-Windows boxes without nl_langinfo(). But even
6269 * if we can't get it directly, we can usually determine if it is a
6270 * UTF-8 locale or not. If it is UTF-8, we (correctly) use that for
6273 # if defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)
6275 /* If libc mbtowc() evaluates the bytes that form the REPLACEMENT
6276 * CHARACTER as that Unicode code point, this has to be a UTF-8 locale.
6279 (void) Perl_mbtowc_(aTHX_ NULL, NULL, 0);/* Reset shift state */
6280 int mbtowc_ret = Perl_mbtowc_(aTHX_ &wc,
6281 STR_WITH_LEN(REPLACEMENT_CHARACTER_UTF8));
6282 if (mbtowc_ret >= 0 && wc == UNICODE_REPLACEMENT) {
6283 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
6284 "mbtowc returned REPLACEMENT\n"));
6289 /* Here, it isn't a UTF-8 locale. */
6291 # else /* mbtowc() is not available. The chances of this code getting
6292 compiled are very small, as it is a C99 required function,
6293 and we are now requiring C99; perhaps if it is a defective
6294 implementation. But if so, there are other libc functions
6295 that could be used instead. */
6297 /* Sling together several possibilities, depending on platform
6298 * capabilities and what we found.
6300 * For non-English locales or non-dollar currency locales, we likely
6301 * will find out whether a locale is UTF-8 or not */
6303 utf8ness_t is_utf8 = UTF8NESS_UNKNOWN;
6304 const char * scratch_buf = NULL;
6306 # if defined(USE_LOCALE_MONETARY) && defined(HAS_LOCALECONV)
6307 # define LANGINFO_RECURSED_MONETARY 0x1
6308 # define LANGINFO_RECURSED_TIME 0x2
6310 /* Can't use this method unless localeconv() is available, as that's
6311 * the way we find out the currency symbol.
6313 * First try looking at the currency symbol (via a recursive call) to
6314 * see if it disambiguates things. Often that will be in the native
6315 * script, and if the symbol isn't legal UTF-8, we know that the locale
6318 * The recursion calls my_localeconv() to find CRNCYSTR, and that can
6319 * call is_locale_utf8() which will call my_langinfo(CODESET) which
6320 * will get to here again, ad infinitum. The guard prevents that.
6322 if ((PL_langinfo_recursed & LANGINFO_RECURSED_MONETARY) == 0) {
6323 PL_langinfo_recursed |= LANGINFO_RECURSED_MONETARY;
6324 (void) my_langinfo_c(CRNCYSTR, LC_MONETARY, locale, &scratch_buf,
6326 PL_langinfo_recursed &= ~LANGINFO_RECURSED_MONETARY;
6329 Safefree(scratch_buf);
6332 # ifdef USE_LOCALE_TIME
6334 /* If we have ruled out being UTF-8, no point in checking further. */
6335 if ( is_utf8 != UTF8NESS_NO
6336 && (PL_langinfo_recursed & LANGINFO_RECURSED_TIME) == 0)
6338 /* But otherwise do check more. This is done even if the currency
6339 * symbol looks to be UTF-8, just in case that's a false positive.
6341 * Look at the LC_TIME entries, like the names of the months or
6342 * weekdays. We quit at the first one that is illegal UTF-8
6344 * The recursion guard is because the recursed my_langinfo_c() will
6345 * call strftime8() to find the LC_TIME value passed to it, and
6346 * that will call my_langinfo(CODESET) for non-ASCII returns,
6347 * which will get here again, ad infinitum
6350 utf8ness_t this_is_utf8 = UTF8NESS_UNKNOWN;
6351 const int times[] = {
6352 DAY_1, DAY_2, DAY_3, DAY_4, DAY_5, DAY_6, DAY_7,
6353 MON_1, MON_2, MON_3, MON_4, MON_5, MON_6, MON_7, MON_8,
6354 MON_9, MON_10, MON_11, MON_12,
6355 ALT_DIGITS, AM_STR, PM_STR,
6356 ABDAY_1, ABDAY_2, ABDAY_3, ABDAY_4, ABDAY_5, ABDAY_6,
6358 ABMON_1, ABMON_2, ABMON_3, ABMON_4, ABMON_5, ABMON_6,
6359 ABMON_7, ABMON_8, ABMON_9, ABMON_10, ABMON_11, ABMON_12
6362 /* The code in the recursive call can handle switching the locales,
6363 * but by doing it here, we avoid switching each iteration of the
6365 const char * orig_TIME_locale = toggle_locale_c(LC_TIME, locale);
6367 PL_langinfo_recursed |= LANGINFO_RECURSED_TIME;
6368 for (PERL_UINT_FAST8_T i = 0; i < C_ARRAY_LENGTH(times); i++) {
6370 (void) my_langinfo_c(times[i], LC_TIME, locale, &scratch_buf,
6371 NULL, &this_is_utf8);
6372 Safefree(scratch_buf);
6373 if (this_is_utf8 == UTF8NESS_NO) {
6374 is_utf8 = UTF8NESS_NO;
6378 if (this_is_utf8 == UTF8NESS_YES) {
6379 is_utf8 = UTF8NESS_YES;
6382 PL_langinfo_recursed &= ~LANGINFO_RECURSED_TIME;
6384 /* Here we have gone through all the LC_TIME elements. is_utf8 has
6385 * been set as follows:
6386 * UTF8NESS_NO If at least one isn't legal UTF-8
6387 * UTF8NESS_IMMMATERIAL If all are ASCII
6388 * UTF8NESS_YES If all are legal UTF-8 (including
6389 * ASCII), and at least one isn't
6392 restore_toggled_locale_c(LC_TIME, orig_TIME_locale);
6395 # endif /* LC_TIME */
6397 /* If nothing examined above rules out it being UTF-8, and at least one
6398 * thing fits as UTF-8 (and not plain ASCII), assume the codeset is
6400 if (is_utf8 == UTF8NESS_YES) {
6405 /* Here, nothing examined indicates that the codeset is UTF-8. But
6406 * what is it? The other locale categories are not likely to be of
6409 * LC_NUMERIC Only a few locales in the world have a non-ASCII radix
6410 * or group separator.
6411 * LC_CTYPE This code wouldn't be compiled if mbtowc() existed and
6412 * was reliable. This is unlikely in C99. There are
6413 * other functions that could be used instead, but are
6414 * they going to exist, and be able to distinguish between
6415 * UTF-8 and 8859-1? Deal with this only if it becomes
6417 * LC_MESSAGES The strings returned from strerror() would seem likely
6418 * candidates, but experience has shown that many systems
6419 * don't actually have translations installed for them.
6420 * They are instead always in English, so everything in
6421 * them is ASCII, which is of no help to us. A Configure
6422 * probe could possibly be written to see if this platform
6423 * has non-ASCII error messages. But again, wait until it
6424 * turns out to be an actual problem. */
6426 # endif /* ! mbtowc() */
6428 /* Rejoin the mbtowc available/not-available cases.
6430 * We got here only because we haven't been able to find the codeset.
6431 * The only other option khw could think of is to see if the codeset is
6432 * part of the locale name. This is very less than ideal; often there
6433 * is no code set in the name; and at other times they even lie.
6435 * But there is an XPG standard syntax, which many locales follow:
6437 * language[_territory[.codeset]][@modifier]
6439 * So we take the part between the dot and any '@' */
6440 retval = (const char *) strchr(locale, '.');
6442 retval = ""; /* Alas, no dot */
6446 /* Don't include the dot */
6449 /* And stop before any '@' */
6450 const char * modifier = strchr(retval, '@');
6452 char * code_set_name;
6453 const Size_t name_len = modifier - retval;
6454 Newx(code_set_name, name_len + 1, char); /* +1 for NUL */
6455 my_strlcpy(code_set_name, retval, name_len + 1);
6456 SAVEFREEPV(code_set_name);
6457 retval = code_set_name;
6460 # if defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)
6462 /* When these functions, are available, they were tried earlier and
6463 * indicated that the locale did not act like a proper UTF-8 one. So
6464 * if it claims to be UTF-8, it is a lie */
6465 if (is_codeset_name_UTF8(retval)) {
6472 /* Otherwise the code set name is considered to be everything between
6473 * the dot and the '@' */
6474 retval = save_to_buffer(retval, retbufp, retbuf_sizep);
6478 # endif /* ! WIN32 */
6479 # endif /* USE_LOCALE_CTYPE */
6481 } /* Giant switch() of nl_langinfo() items */
6483 restore_toggled_locale_i(cat_index, orig_switched_locale);
6485 # ifdef USE_LOCALE_CTYPE
6486 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
6490 *utf8ness = is_utf8;
6495 # endif /* All the implementations of my_langinfo() */
6497 /*--------------------------------------------------------------------------*/
6499 } /* my_langinfo() */
6501 #endif /* USE_LOCALE */
6504 =for apidoc_section $time
6505 =for apidoc sv_strftime_tm
6506 =for apidoc_item sv_strftime_ints
6507 =for apidoc_item my_strftime
6509 These implement the libc strftime(), but with a different API so that the return
6510 value is a pointer to the formatted result (which MUST be arranged to be FREED
6511 BY THE CALLER). This allows these functions to increase the buffer size as
6512 needed, so that the caller doesn't have to worry about that.
6514 On failure they return NULL, and set errno to C<EINVAL>.
6516 C<sv_strftime_tm> and C<sv_strftime_ints> are preferred, as they transparently
6517 handle the UTF-8ness of the current locale, the input C<fmt>, and the returned
6518 result. Only if the current C<LC_TIME> locale is a UTF-8 one (and S<C<use
6519 bytes>> is not in effect) will the result be marked as UTF-8. These differ
6520 only in the form of their inputs. C<sv_strftime_tm> takes a filled-in
6521 S<C<struct tm>> parameter. C<sv_strftime_ints> takes a bunch of integer
6522 parameters that together completely define a given time.
6524 C<my_strftime> is kept for backwards compatibility. Knowing if the result
6525 should be considered UTF-8 or not requires significant extra logic.
6527 Note that C<yday> and C<wday> effectively are ignored by C<sv_strftime_ints>
6528 and C<my_strftime>, as mini_mktime() overwrites them
6530 Also note that all three functions are always executed in the underlying
6531 C<LC_TIME> locale of the program, giving results based on that locale.
6537 S_ints_to_tm(pTHX_ struct tm * mytm,
6538 int sec, int min, int hour, int mday, int mon, int year,
6539 int wday, int yday, int isdst)
6541 /* Create a struct tm structure from the input time-related integer
6544 /* Override with the passed-in values */
6545 Zero(mytm, 1, struct tm);
6548 mytm->tm_hour = hour;
6549 mytm->tm_mday = mday;
6551 mytm->tm_year = year;
6552 mytm->tm_wday = wday;
6553 mytm->tm_yday = yday;
6554 mytm->tm_isdst = isdst;
6557 /* use libc to get the values for tm_gmtoff and tm_zone on platforms that
6558 * have them [perl #18238] */
6559 #if defined(HAS_MKTIME) \
6560 && (defined(HAS_TM_TM_GMTOFF) || defined(HAS_TM_TM_ZONE))
6561 struct tm mytm2 = *mytm;
6565 # ifdef HAS_TM_TM_GMTOFF
6566 mytm->tm_gmtoff = mytm2.tm_gmtoff;
6568 # ifdef HAS_TM_TM_ZONE
6569 mytm->tm_zone = mytm2.tm_zone;
6577 S_strftime_tm(pTHX_ const char *fmt, const struct tm *mytm)
6579 PERL_ARGS_ASSERT_STRFTIME_TM;
6581 /* Execute strftime() based on the input struct tm */
6583 /* An empty format yields an empty result */
6584 const int fmtlen = strlen(fmt);
6587 Newxz (ret, 1, char);
6591 #ifndef HAS_STRFTIME
6592 Perl_croak(aTHX_ "panic: no strftime");
6594 # if defined(USE_LOCALE_CTYPE) && defined(USE_LOCALE_TIME)
6596 const char * orig_CTYPE_LOCALE = toggle_locale_c(LC_CTYPE,
6597 querylocale_c(LC_TIME));
6600 /* Guess an initial size for the returned string based on an expansion
6601 * factor of the input format, but with a minimum that should handle most
6602 * common cases. If this guess is too small, we will try again with a
6604 int bufsize = MAX(fmtlen * 2, 64);
6606 char *buf = NULL; /* Makes Renew() act as Newx() on the first iteration */
6608 Renew(buf, bufsize, char);
6610 /* allowing user-supplied (rather than literal) formats is normally
6611 * frowned upon as a potential security risk; but this is part of the
6612 * API so we have to allow it (and the available formats have a much
6613 * lower chance of doing something bad than the ones for printf etc. */
6614 GCC_DIAG_IGNORE_STMT(-Wformat-nonliteral);
6617 int len = strftime(buf, bufsize, fmt, mytm);
6620 GCC_DIAG_RESTORE_STMT;
6622 /* A non-zero return indicates success. But to make sure we're not
6623 * dealing with some rogue strftime that returns how much space it
6624 * needs instead of 0 when there isn't enough, check that the return
6625 * indicates we have at least one byte of spare space (which will be
6626 * used for the terminating NUL). */
6627 if (inRANGE(len, 1, bufsize - 1)) {
6628 goto strftime_return;
6631 /* There are several possible reasons for a 0 return code for a
6632 * non-empty format, and they are not trivial to tease apart. This
6633 * issue is a known bug in the strftime() API. What we do to cope is
6634 * to assume that the reason is not enough space in the buffer, so
6635 * increase it and try again. */
6638 /* But don't just keep increasing the size indefinitely. Stop when it
6639 * becomes obvious that the reason for failure is something besides not
6640 * enough space. The most likely largest expanding format is %c. On
6641 * khw's Linux box, the maximum result of this is 67 characters, in the
6642 * km_KH locale. If a new script comes along that uses 4 UTF-8 bytes
6643 * per character, and with a similar expansion factor, that would be a
6644 * 268:2 byte ratio, or a bit more than 128:1 = 2**7:1. Some strftime
6645 * implementations allow you to say %1000c to pad to 1000 bytes. This
6646 * shows that it is impossible to implement this without a heuristic
6647 * (which can fail). But it indicates we need to be generous in the
6648 * upper limit before failing. The previous heuristic used was too
6649 * stingy. Since the size doubles per iteration, it doesn't take many
6650 * to reach the limit */
6651 } while (bufsize < ((1 << 11) + 1) * fmtlen);
6653 /* Here, strftime() returned 0, and it likely wasn't for lack of space.
6654 * There are two possible reasons:
6656 * First is that the result is legitimately 0 length. This can happen
6657 * when the format is precisely "%p". That is the only documented format
6658 * that can have an empty result. */
6659 if (strEQ(fmt, "%p")) {
6660 Renew(buf, 1, char);
6662 goto strftime_return;
6665 /* The other reason is that the format string is malformed. Probably it is
6666 * that the string is syntactically invalid for the locale. On some
6667 * platforms an invalid conversion specifier '%?' (for all illegal '?') is
6668 * treated as a literal, but others may fail when '?' is illegal */
6675 # if defined(USE_LOCALE_CTYPE) && defined(USE_LOCALE_TIME)
6677 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_LOCALE);
6688 S_strftime8(pTHX_ const char * fmt,
6689 const struct tm * mytm,
6690 const utf8ness_t fmt_utf8ness,
6691 utf8ness_t * result_utf8ness,
6692 const bool came_from_sv)
6694 PERL_ARGS_ASSERT_STRFTIME8;
6696 /* Wrap strftime_tm, taking into account the input and output UTF-8ness */
6698 #ifdef USE_LOCALE_TIME
6699 # define INDEX_TO_USE LC_TIME_INDEX_
6701 const char * locale = querylocale_c(LC_TIME);
6702 locale_utf8ness_t locale_utf8ness = LOCALE_UTF8NESS_UNKNOWN;
6705 # define INDEX_TO_USE LC_ALL_INDEX_ /* Effectively out of bounds */
6707 const char * locale = "C";
6708 locale_utf8ness_t locale_utf8ness = LOCALE_NOT_UTF8;
6712 switch (fmt_utf8ness) {
6713 case UTF8NESS_IMMATERIAL:
6716 case UTF8NESS_NO: /* Known not to be UTF-8; must not be UTF-8 locale */
6717 if (is_locale_utf8(locale)) {
6722 locale_utf8ness = LOCALE_NOT_UTF8;
6725 case UTF8NESS_YES: /* Known to be UTF-8; must be UTF-8 locale if can't
6726 downgrade. But downgrading assumes the locale
6727 is latin 1. Maybe just fail XXX */
6728 if (! is_locale_utf8(locale)) {
6729 locale_utf8ness = LOCALE_NOT_UTF8;
6731 bool is_utf8 = true;
6732 Size_t fmt_len = strlen(fmt);
6733 fmt = (char *) bytes_from_utf8((U8 *) fmt, &fmt_len, &is_utf8);
6742 locale_utf8ness = LOCALE_IS_UTF8;
6747 case UTF8NESS_UNKNOWN:
6748 if (! is_locale_utf8(locale)) {
6749 locale_utf8ness = LOCALE_NOT_UTF8;
6752 locale_utf8ness = LOCALE_IS_UTF8;
6755 /* Upgrade 'fmt' to UTF-8 for a UTF-8 locale. Otherwise the
6756 * locale would find any UTF-8 variant characters to be
6758 Size_t fmt_len = strlen(fmt);
6759 fmt = (char *) bytes_to_utf8((U8 *) fmt, &fmt_len);
6767 char * retval = strftime_tm(fmt, mytm);
6768 *result_utf8ness = get_locale_string_utf8ness_i(retval,
6772 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
6773 "fmt=%s, retval=%s; utf8ness=%d",
6775 ((is_utf8_string((U8 *) retval, 0))
6777 :_byte_dump_string((U8 *) retval, strlen(retval),0)),
6788 S_give_perl_locale_control(pTHX_
6790 const char * lc_all_string,
6792 const char ** locales,
6794 const line_t caller_line)
6796 PERL_UNUSED_ARG(caller_line);
6798 /* This is called when the program is in the global locale and are
6799 * switching to per-thread (if available). And it is called at
6800 * initialization time to do the same.
6803 # if defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE)
6805 /* On Windows, convert to per-thread behavior. This isn't necessary in
6806 * POSIX 2008, as the conversion gets done automatically in the
6807 * void_setlocale_i() calls below. */
6808 if (_configthreadlocale(_ENABLE_PER_THREAD_LOCALE) == -1) {
6809 locale_panic_("_configthreadlocale returned an error");
6813 # if ! defined(USE_THREAD_SAFE_LOCALE) \
6814 && ! defined(USE_POSIX_2008_LOCALE)
6815 # if defined(LC_ALL)
6816 PERL_UNUSED_ARG(lc_all_string);
6818 PERL_UNUSED_ARG(locales);
6822 /* This platform has per-thread locale handling. Do the conversion. */
6824 # if defined(LC_ALL)
6826 void_setlocale_c_with_caller(LC_ALL, lc_all_string, __FILE__, caller_line);
6830 for_all_individual_category_indexes(i) {
6831 void_setlocale_i_with_caller(i, locales[i], __FILE__, caller_line);
6837 /* Finally, update our remaining records. 'true' => force recalculation.
6838 * This is needed because we don't know what's happened while Perl hasn't
6839 * had control, so we need to figure out the current state */
6841 # if defined(LC_ALL)
6843 new_LC_ALL(lc_all_string, true);
6847 new_LC_ALL(locales, true);
6853 S_output_check_environment_warning(pTHX_ const char * const language,
6854 const char * const lc_all,
6855 const char * const lang)
6857 PerlIO_printf(Perl_error_log,
6858 "perl: warning: Please check that your locale settings:\n");
6862 PerlIO_printf(Perl_error_log, "\tLANGUAGE = %c%s%c,\n",
6863 language ? '"' : '(',
6864 language ? language : "unset",
6865 language ? '"' : ')');
6867 PERL_UNUSED_ARG(language);
6870 PerlIO_printf(Perl_error_log, "\tLC_ALL = %c%s%c,\n",
6872 lc_all ? lc_all : "unset",
6873 lc_all ? '"' : ')');
6875 for_all_individual_category_indexes(i) {
6876 const char * value = PerlEnv_getenv(category_names[i]);
6877 PerlIO_printf(Perl_error_log,
6881 value ? value : "unset",
6885 PerlIO_printf(Perl_error_log, "\tLANG = %c%s%c\n",
6887 lang ? lang : "unset",
6889 PerlIO_printf(Perl_error_log,
6890 " are supported and installed on your system.\n");
6895 /* A helper macro for the next function. Needed because would be called in two
6896 * places. Knows about the internal workings of the function */
6897 #define GET_DESCRIPTION(trial, name) \
6898 ((isNAME_C_OR_POSIX(name)) \
6899 ? "the standard locale" \
6900 : ((trial == (system_default_trial) \
6901 ? "the system default locale" \
6902 : "a fallback locale")))
6905 * Initialize locale awareness.
6908 Perl_init_i18nl10n(pTHX_ int printwarn)
6911 * 0 if not to output warning when setup locale is bad
6912 * 1 if to output warning based on value of PERL_BADLANG
6913 * >1 if to output regardless of PERL_BADLANG
6916 * 1 = set ok or not applicable,
6917 * 0 = fallback to a locale of lower priority
6918 * -1 = fallback to all locales failed, not even to the C locale
6920 * Under -DDEBUGGING, if the environment variable PERL_DEBUG_LOCALE_INIT is
6921 * set, debugging information is output.
6923 * This routine effectively does the following in most cases:
6925 * basic initialization;
6926 * asserts that the compiled tables are consistent;
6927 * initialize data structures;
6928 * make sure we are in the global locale;
6929 * setlocale(LC_ALL, "");
6930 * switch to per-thread locale if applicable;
6932 * The "" causes the locale to be set to what the environment variables at
6933 * the time say it should be.
6935 * To handle possible failures, the setlocale is expanded to be like:
6937 * trial_locale = pre-first-trial;
6938 * while (has_another_trial()) {
6939 * trial_locale = next_trial();
6940 * if setlocale(LC_ALL, trial_locale) {
6945 * had_failure = true;
6949 * if (had_failure) {
6951 * if (! ok) warn_still_more();
6954 * The first trial is either:
6955 * "" to examine the environment variables for the locale
6956 * NULL to use the values already set for the locale by the program
6957 * embedding this perl instantiation.
6959 * Something is wrong if this trial fails, but there is a sequence of
6960 * fallbacks to try should that happen. They are given in the enum below.
6962 * If there is no LC_ALL defined on the system, the setlocale() above is
6963 * replaced by a loop setting each individual category separately.
6965 * In a non-embeded environment, this code is executed exactly once. It
6966 * sets up the global locale environment. At the end, if some sort of
6967 * thread-safety is in effect, it will turn thread 0 into using that, with
6968 * the same locale as the global initially. thread 0 can then change its
6969 * locale at will without affecting the global one.
6971 * At destruction time, thread 0 will revert to the global locale as the
6972 * other threads die.
6974 * Care must be taken in an embedded environment. This code will be
6975 * executed for each instantiation. Since it changes the global locale, it
6976 * could clash with another running instantiation that isn't using
6977 * per-thread locales. perlembed suggests having the controlling program
6978 * set each instantiation's locale and set PERL_SKIP_LOCALE_INIT so this
6979 * code uses that without actually changing anything. Then the onus is on
6980 * the controlling program to prevent any races. The code below does
6981 * enough locking so as to prevent system calls from overwriting data
6982 * before it is safely copied here, but that isn't a general solution.
6987 PERL_UNUSED_ARG(printwarn);
6990 #else /* USE_LOCALE to near the end of the routine */
6996 const char * const language = PerlEnv_getenv("LANGUAGE");
6999 const char * const language = NULL; /* Unused placeholder */
7002 /* A later getenv() could zap this, so only use here */
7003 const char * const bad_lang_use_once = PerlEnv_getenv("PERL_BADLANG");
7005 const bool locwarn = (printwarn > 1
7007 && ( ! bad_lang_use_once
7009 /* disallow with "" or "0" */
7011 && strNE("0", bad_lang_use_once)))));
7014 # define DEBUG_LOCALE_INIT(a,b,c)
7017 DEBUG_INITIALIZATION_set(cBOOL(PerlEnv_getenv("PERL_DEBUG_LOCALE_INIT")));
7019 # define DEBUG_LOCALE_INIT(cat_index, locale, result) \
7020 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n", \
7021 setlocale_debug_string_i(cat_index, locale, result)));
7024 assert(categories[LC_ALL_INDEX_] == LC_ALL);
7025 assert(strEQ(category_names[LC_ALL_INDEX_], "LC_ALL"));
7026 # ifdef USE_POSIX_2008_LOCALE
7027 assert(category_masks[LC_ALL_INDEX_] == LC_ALL_MASK);
7031 for_all_individual_category_indexes(i) {
7032 assert(category_name_lengths[i] == strlen(category_names[i]));
7035 # endif /* DEBUGGING */
7037 /* Initialize the per-thread mbrFOO() state variables. See POSIX.xs for
7038 * why these particular incantations are used. */
7040 memzero(&PL_mbrlen_ps, sizeof(PL_mbrlen_ps));
7043 memzero(&PL_mbrtowc_ps, sizeof(PL_mbrtowc_ps));
7046 wcrtomb(NULL, L'\0', &PL_wcrtomb_ps);
7048 # ifdef USE_PL_CURLOCALES
7050 for (unsigned int i = 0; i <= LC_ALL_INDEX_; i++) {
7051 PL_curlocales[i] = savepv("C");
7055 # ifdef USE_PL_CUR_LC_ALL
7057 PL_cur_LC_ALL = savepv("C");
7060 # if ! defined(PERL_LC_ALL_USES_NAME_VALUE_PAIRS) && defined(LC_ALL)
7064 /* If we haven't done so already, translate the LC_ALL positions of
7065 * categories into our internal indices. */
7066 if (map_LC_ALL_position_to_index[0] == LC_ALL_INDEX_) {
7068 /* Use this array, initialized by a config.h constant */
7069 int lc_all_category_positions[] = PERL_LC_ALL_CATEGORY_POSITIONS_INIT;
7070 STATIC_ASSERT_STMT( C_ARRAY_LENGTH(lc_all_category_positions)
7073 for (unsigned int i = 0;
7074 i < C_ARRAY_LENGTH(lc_all_category_positions);
7077 map_LC_ALL_position_to_index[i] =
7078 get_category_index(lc_all_category_positions[i]);
7085 # ifdef USE_POSIX_2008_LOCALE
7087 /* This is a global, so be sure to keep another instance from zapping it */
7089 if (PL_C_locale_obj) {
7093 PL_C_locale_obj = newlocale(LC_ALL_MASK, "C", (locale_t) 0);
7094 if (! PL_C_locale_obj) {
7096 locale_panic_(Perl_form(aTHX_
7097 "Cannot create POSIX 2008 C locale object"));
7101 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "created C object %p\n",
7105 /* Switch to using the POSIX 2008 interface now. This would happen below
7106 * anyway, but deferring it can lead to leaks of memory that would also get
7107 * malloc'd in the interim. We arbitrarily switch to the C locale,
7108 * overridden below */
7109 if (! uselocale(PL_C_locale_obj)) {
7110 locale_panic_(Perl_form(aTHX_
7111 "Can't uselocale(%p), LC_ALL supposed to"
7116 # ifdef MULTIPLICITY
7118 PL_cur_locale_obj = PL_C_locale_obj;
7123 /* Now initialize some data structures. This is entirely so that
7124 * later-executed code doesn't have to concern itself with things not being
7125 * initialized. Arbitrarily use the C locale (which we know has to exist
7126 * on the system). */
7128 # ifdef USE_LOCALE_NUMERIC
7130 PL_numeric_radix_sv = newSV(1);
7131 PL_underlying_radix_sv = newSV(1);
7132 Newxz(PL_numeric_name, 1, char); /* Single NUL character */
7135 # ifdef USE_LOCALE_COLLATE
7137 Newxz(PL_collation_name, 1, char);
7140 # ifdef USE_LOCALE_CTYPE
7142 Newxz(PL_ctype_name, 1, char);
7146 new_LC_ALL("C", true /* Don't shortcut */);
7148 /*===========================================================================*/
7150 /* Now ready to override the initialization with the values that the user
7151 * wants. This is done in the global locale as explained in the
7152 * introductory comments to this function */
7153 switch_to_global_locale();
7155 const char * const lc_all = PerlEnv_getenv("LC_ALL");
7156 const char * const lang = PerlEnv_getenv("LANG");
7158 /* We try each locale in the enum, in order, until we get one that works,
7159 * or exhaust the list. Normally the loop is executed just once.
7161 * Each enum value is +1 from the previous */
7164 environment_trial = 0, /* "" or NULL; code below assumes value
7165 0 is the first real trial */
7166 LC_ALL_trial, /* ENV{LC_ALL} */
7167 LANG_trial, /* ENV{LANG} */
7168 system_default_trial, /* Windows .ACP */
7169 C_trial, /* C locale */
7174 SSize_t already_checked = 0;
7175 const char * checked[C_trial];
7178 const char * lc_all_string;
7180 const char * curlocales[LC_ALL_INDEX_];
7183 /* Loop through the initial setting and all the possible fallbacks,
7184 * breaking out of the loop on success */
7185 trial = dummy_trial;
7186 while (trial != beyond_final_trial) {
7188 /* Each time through compute the next trial to use based on the one in
7189 * the previous iteration and switch to the new one. This enforces the
7190 * order in which the fallbacks are applied */
7192 trial = (trials) ((int) trial + 1); /* Casts are needed for g++ */
7194 const char * locale = NULL;
7196 /* Set up the parameters for this trial */
7199 locale_panic_("Unexpectedly got 'dummy_trial");
7202 case environment_trial:
7203 /* This is either "" to get the values from the environment, or
7204 * NULL if the calling program has initialized the values already.
7206 locale = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
7212 if (! lc_all || strEQ(lc_all, "")) {
7213 continue; /* No-op */
7220 if (! lang || strEQ(lang, "")) {
7221 continue; /* No-op */
7227 case system_default_trial:
7229 # if ! defined(WIN32) || ! defined(LC_ALL)
7231 continue; /* No-op */
7234 /* For Windows, we also try the system default locale before "C".
7235 * (If there exists a Windows without LC_ALL we skip this because
7236 * it gets too complicated. For those, "C" is the next fallback
7246 case beyond_final_trial:
7247 continue; /* No-op, causes loop to exit */
7250 /* If the locale is a substantive name, don't try the same locale
7252 if (locale && strNE(locale, "")) {
7253 for (unsigned int i = 0; i < already_checked; i++) {
7254 if (strEQ(checked[i], locale)) {
7259 /* And, for future iterations, indicate we've tried this locale */
7260 checked[already_checked] = savepv(locale);
7261 SAVEFREEPV(checked[already_checked]);
7267 STDIZED_SETLOCALE_LOCK;
7268 lc_all_string = savepv(stdized_setlocale(LC_ALL, locale));
7269 STDIZED_SETLOCALE_UNLOCK;
7271 DEBUG_LOCALE_INIT(LC_ALL_INDEX_, locale, lc_all_string);
7273 if (LIKELY(lc_all_string)) { /* Succeeded */
7278 if (trial == 0 && locwarn) {
7279 PerlIO_printf(Perl_error_log,
7280 "perl: warning: Setting locale failed.\n");
7281 output_check_environment_warning(language, lc_all, lang);
7284 # else /* Below is ! LC_ALL */
7286 bool setlocale_failure = FALSE; /* This trial hasn't failed so far */
7287 bool dowarn = trial == 0 && locwarn;
7289 for_all_individual_category_indexes(j) {
7290 STDIZED_SETLOCALE_LOCK;
7291 curlocales[j] = savepv(stdized_setlocale(categories[j], locale));
7292 STDIZED_SETLOCALE_UNLOCK;
7294 DEBUG_LOCALE_INIT(j, locale, curlocales[j]);
7296 if (UNLIKELY(! curlocales[j])) {
7297 setlocale_failure = TRUE;
7299 /* If are going to warn below, continue to loop so all failures
7300 * are included in the message */
7307 if (LIKELY(! setlocale_failure)) { /* All succeeded */
7309 break; /* Exit trial_locales loop */
7312 /* Here, this trial failed */
7315 PerlIO_printf(Perl_error_log,
7316 "perl: warning: Setting locale failed for the categories:\n");
7318 for_all_individual_category_indexes(j) {
7319 if (! curlocales[j]) {
7320 PerlIO_printf(Perl_error_log, "\t%s\n", category_names[j]);
7324 output_check_environment_warning(language, lc_all, lang);
7325 } /* end of warning on first failure */
7327 # endif /* LC_ALL */
7329 } /* end of looping through the trial locales */
7331 /* If we had to do more than the first trial, it means that one failed, and
7332 * we may need to output a warning, and, if none worked, do more */
7333 if (UNLIKELY(trial != 0)) {
7335 const char * description = "a fallback locale";
7336 const char * name = NULL;;
7338 /* If we didn't find a good fallback, list all we tried */
7339 if (! ok && already_checked > 0) {
7340 PerlIO_printf(Perl_error_log, "perl: warning: Failed to fall"
7342 if (already_checked > 1) { /* more than one was tried */
7343 PerlIO_printf(Perl_error_log, "any of:\n");
7346 while (already_checked > 0) {
7347 name = checked[--already_checked];
7348 description = GET_DESCRIPTION(trial, name);
7349 PerlIO_printf(Perl_error_log, "%s (\"%s\")\n",
7356 /* Here, a fallback worked. So we have saved its name, and the
7357 * trial that succeeded is still valid */
7359 const char * individ_locales[LC_ALL_INDEX_] = { NULL };
7361 /* Even though we know the valid string for LC_ALL that worked,
7362 * translate it into our internal format, which is the
7363 * name=value pairs notation. This is easier for a human to
7364 * decipher than the positional notation. Some platforms
7365 * can return "C C C C C C" for LC_ALL. This code also
7366 * standardizes that result into plain "C". */
7367 switch (parse_LC_ALL_string(lc_all_string,
7368 (const char **) &individ_locales,
7370 false, /* Return only [0] if
7372 false, /* Don't panic on error */
7377 /* Here, the parse failed, which shouldn't happen, but if
7378 * it does, we have an easy fallback that allows us to keep
7380 name = lc_all_string;
7383 case no_array: /* The original is a single locale */
7384 name = lc_all_string;
7387 case only_element_0: /* element[0] is a single locale valid
7388 for all categories */
7389 SAVEFREEPV(individ_locales[0]);
7390 name = individ_locales[0];
7394 name = calculate_LC_ALL_string(individ_locales,
7398 for_all_individual_category_indexes(j) {
7399 Safefree(individ_locales[j]);
7403 name = calculate_LC_ALL_string(curlocales,
7408 description = GET_DESCRIPTION(trial, name);
7412 /* Nothing seems to be working, yet we want to continue
7413 * executing. It may well be that locales are mostly
7414 * irrelevant to this particular program, and there must be
7415 * some locale underlying the program. Figure it out as best
7416 * we can, by querying the system's current locale */
7420 STDIZED_SETLOCALE_LOCK;
7421 name = stdized_setlocale(LC_ALL, NULL);
7422 STDIZED_SETLOCALE_UNLOCK;
7424 if (UNLIKELY(! name)) {
7425 name = "locale name not determinable";
7428 # else /* Below is ! LC_ALL */
7430 const char * system_locales[LC_ALL_INDEX_] = { NULL };
7432 for_all_individual_category_indexes(j) {
7433 STDIZED_SETLOCALE_LOCK;
7434 system_locales[j] = savepv(stdized_setlocale(categories[j],
7436 STDIZED_SETLOCALE_UNLOCK;
7438 if (UNLIKELY(! system_locales[j])) {
7439 system_locales[j] = "not determinable";
7443 /* We use the name=value form for the string, as that is more
7444 * human readable than the positional notation */
7445 name = calculate_LC_ALL_string(system_locales,
7449 description = "what the system says";
7451 for_all_individual_category_indexes(j) {
7452 Safefree(system_locales[j]);
7457 PerlIO_printf(Perl_error_log,
7458 "perl: warning: Falling back to %s (\"%s\").\n",
7461 /* Here, ok being true indicates that the first attempt failed, but
7462 * a fallback succeeded; false => nothing working. Translate to
7463 * API return values. */
7470 give_perl_locale_control(lc_all_string, __LINE__);
7471 Safefree(lc_all_string);
7475 give_perl_locale_control((const char **) &curlocales, __LINE__);
7477 for_all_individual_category_indexes(j) {
7478 Safefree(curlocales[j]);
7482 # if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
7484 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
7485 * locale is UTF-8. give_perl_locale_control() just above has already
7486 * calculated the latter value and saved it in PL_in_utf8_CTYPE_locale. If
7487 * both PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
7488 * are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer on
7489 * STDIN, STDOUT, STDERR, _and_ the default open discipline. */
7490 PL_utf8locale = PL_in_utf8_CTYPE_locale;
7492 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
7493 This is an alternative to using the -C command line switch
7494 (the -C if present will override this). */
7496 const char *p = PerlEnv_getenv("PERL_UNICODE");
7497 PL_unicode = p ? parse_unicode_opts(&p) : 0;
7498 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
7503 # if defined(USE_POSIX_2008_LOCALE) && defined(MULTIPLICITY)
7504 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
7505 "finished Perl_init_i18nl10n; actual obj=%p,"
7506 " expected obj=%p, initial=%s\n",
7507 uselocale(0), PL_cur_locale_obj,
7508 get_LC_ALL_display()));
7511 /* So won't continue to output stuff */
7512 DEBUG_INITIALIZATION_set(FALSE);
7514 #endif /* USE_LOCALE */
7519 #undef GET_DESCRIPTION
7520 #ifdef USE_LOCALE_COLLATE
7523 S_compute_collxfrm_coefficients(pTHX)
7526 /* A locale collation definition includes primary, secondary, tertiary,
7527 * etc. weights for each character. To sort, the primary weights are used,
7528 * and only if they compare equal, then the secondary weights are used, and
7529 * only if they compare equal, then the tertiary, etc.
7531 * strxfrm() works by taking the input string, say ABC, and creating an
7532 * output transformed string consisting of first the primary weights,
7533 * A¹B¹C¹ followed by the secondary ones, A²B²C²; and then the tertiary,
7534 * etc, yielding A¹B¹C¹ A²B²C² A³B³C³ .... Some characters may not have
7535 * weights at every level. In our example, let's say B doesn't have a
7536 * tertiary weight, and A doesn't have a secondary weight. The constructed
7537 * string is then going to be
7538 * A¹B¹C¹ B²C² A³C³ ....
7539 * This has the desired effect that strcmp() will look at the secondary or
7540 * tertiary weights only if the strings compare equal at all higher
7541 * priority weights. The spaces shown here, like in
7543 * are not just for readability. In the general case, these must actually
7544 * be bytes, which we will call here 'separator weights'; and they must be
7545 * smaller than any other weight value, but since these are C strings, only
7546 * the terminating one can be a NUL (some implementations may include a
7547 * non-NUL separator weight just before the NUL). Implementations tend to
7548 * reserve 01 for the separator weights. They are needed so that a shorter
7549 * string's secondary weights won't be misconstrued as primary weights of a
7550 * longer string, etc. By making them smaller than any other weight, the
7551 * shorter string will sort first. (Actually, if all secondary weights are
7552 * smaller than all primary ones, there is no need for a separator weight
7553 * between those two levels, etc.)
7555 * The length of the transformed string is roughly a linear function of the
7556 * input string. It's not exactly linear because some characters don't
7557 * have weights at all levels. When we call strxfrm() we have to allocate
7558 * some memory to hold the transformed string. The calculations below try
7559 * to find coefficients 'm' and 'b' for this locale so that m*x + b equals
7560 * how much space we need, given the size of the input string in 'x'. If
7561 * we calculate too small, we increase the size as needed, and call
7562 * strxfrm() again, but it is better to get it right the first time to
7563 * avoid wasted expensive string transformations.
7565 * We use the string below to find how long the transformation of it is.
7566 * Almost all locales are supersets of ASCII, or at least the ASCII
7567 * letters. We use all of them, half upper half lower, because if we used
7568 * fewer, we might hit just the ones that are outliers in a particular
7569 * locale. Most of the strings being collated will contain a preponderance
7570 * of letters, and even if they are above-ASCII, they are likely to have
7571 * the same number of weight levels as the ASCII ones. It turns out that
7572 * digits tend to have fewer levels, and some punctuation has more, but
7573 * those are relatively sparse in text, and khw believes this gives a
7574 * reasonable result, but it could be changed if experience so dictates. */
7575 const char longer[] = "ABCDEFGHIJKLMnopqrstuvwxyz";
7576 char * x_longer; /* Transformed 'longer' */
7577 Size_t x_len_longer; /* Length of 'x_longer' */
7579 char * x_shorter; /* We also transform a substring of 'longer' */
7580 Size_t x_len_shorter;
7582 PL_in_utf8_COLLATE_locale = (PL_collation_standard)
7584 : is_locale_utf8(PL_collation_name);
7585 PL_strxfrm_NUL_replacement = '\0';
7586 PL_strxfrm_max_cp = 0;
7588 /* mem_collxfrm_() is used get the transformation (though here we are
7589 * interested only in its length). It is used because it has the
7590 * intelligence to handle all cases, but to work, it needs some values of
7591 * 'm' and 'b' to get it started. For the purposes of this calculation we
7592 * use a very conservative estimate of 'm' and 'b'. This assumes a weight
7593 * can be multiple bytes, enough to hold any UV on the platform, and there
7594 * are 5 levels, 4 weight bytes, and a trailing NUL. */
7595 PL_collxfrm_base = 5;
7596 PL_collxfrm_mult = 5 * sizeof(UV);
7598 /* Find out how long the transformation really is */
7599 x_longer = mem_collxfrm_(longer,
7603 /* We avoid converting to UTF-8 in the called
7604 * function by telling it the string is in UTF-8
7605 * if the locale is a UTF-8 one. Since the string
7606 * passed here is invariant under UTF-8, we can
7607 * claim it's UTF-8 even if it isn't. */
7608 PL_in_utf8_COLLATE_locale);
7611 /* Find out how long the transformation of a substring of 'longer' is.
7612 * Together the lengths of these transformations are sufficient to
7613 * calculate 'm' and 'b'. The substring is all of 'longer' except the
7614 * first character. This minimizes the chances of being swayed by outliers
7616 x_shorter = mem_collxfrm_(longer + 1,
7619 PL_in_utf8_COLLATE_locale);
7620 Safefree(x_shorter);
7622 /* If the results are nonsensical for this simple test, the whole locale
7623 * definition is suspect. Mark it so that locale collation is not active
7624 * at all for it. XXX Should we warn? */
7625 if ( x_len_shorter == 0
7626 || x_len_longer == 0
7627 || x_len_shorter >= x_len_longer)
7629 PL_collxfrm_mult = 0;
7630 PL_collxfrm_base = 1;
7631 DEBUG_L(PerlIO_printf(Perl_debug_log,
7632 "Disabling locale collation for LC_COLLATE='%s';"
7633 " length for shorter sample=%zu; longer=%zu\n",
7634 PL_collation_name, x_len_shorter, x_len_longer));
7637 SSize_t base; /* Temporary */
7639 /* We have both: m * strlen(longer) + b = x_len_longer
7640 * m * strlen(shorter) + b = x_len_shorter;
7641 * subtracting yields:
7642 * m * (strlen(longer) - strlen(shorter))
7643 * = x_len_longer - x_len_shorter
7644 * But we have set things up so that 'shorter' is 1 byte smaller than
7646 * m = x_len_longer - x_len_shorter
7648 * But if something went wrong, make sure the multiplier is at least 1.
7650 if (x_len_longer > x_len_shorter) {
7651 PL_collxfrm_mult = (STRLEN) x_len_longer - x_len_shorter;
7654 PL_collxfrm_mult = 1;
7659 * but in case something has gone wrong, make sure it is non-negative
7661 base = x_len_longer - PL_collxfrm_mult * (sizeof(longer) - 1);
7666 /* Add 1 for the trailing NUL */
7667 PL_collxfrm_base = base + 1;
7670 DEBUG_L(PerlIO_printf(Perl_debug_log,
7671 "?UTF-8 locale=%d; x_len_shorter=%zu, "
7673 " collate multipler=%zu, collate base=%zu\n",
7674 PL_in_utf8_COLLATE_locale,
7675 x_len_shorter, x_len_longer,
7676 PL_collxfrm_mult, PL_collxfrm_base));
7680 Perl_mem_collxfrm_(pTHX_ const char *input_string,
7681 STRLEN len, /* Length of 'input_string' */
7682 STRLEN *xlen, /* Set to length of returned string
7683 (not including the collation index
7685 bool utf8 /* Is the input in UTF-8? */
7688 /* mem_collxfrm_() is like strxfrm() but with two important differences.
7689 * First, it handles embedded NULs. Second, it allocates a bit more memory
7690 * than needed for the transformed data itself. The real transformed data
7691 * begins at offset COLLXFRM_HDR_LEN. *xlen is set to the length of that,
7692 * and doesn't include the collation index size.
7694 * It is the caller's responsibility to eventually free the memory returned
7697 * Please see sv_collxfrm() to see how this is used. */
7699 # define COLLXFRM_HDR_LEN sizeof(PL_collation_ix)
7701 char * s = (char *) input_string;
7702 STRLEN s_strlen = strlen(input_string);
7704 STRLEN xAlloc; /* xalloc is a reserved word in VC */
7705 STRLEN length_in_chars;
7706 bool first_time = TRUE; /* Cleared after first loop iteration */
7708 # ifdef USE_LOCALE_CTYPE
7709 const char * orig_CTYPE_locale = NULL;
7712 # if defined(USE_POSIX_2008_LOCALE) && defined HAS_STRXFRM_L
7713 locale_t constructed_locale = (locale_t) 0;
7716 PERL_ARGS_ASSERT_MEM_COLLXFRM_;
7718 /* Must be NUL-terminated */
7719 assert(*(input_string + len) == '\0');
7721 if (PL_collxfrm_mult == 0) { /* unknown or bad */
7722 if (PL_collxfrm_base != 0) { /* bad collation => skip */
7723 DEBUG_L(PerlIO_printf(Perl_debug_log,
7724 "mem_collxfrm_: locale's collation is defective\n"));
7728 /* (mult, base) == (0,0) means we need to calculate mult and base
7729 * before proceeding */
7730 S_compute_collxfrm_coefficients(aTHX);
7733 /* Replace any embedded NULs with the control that sorts before any others.
7734 * This will give as good as possible results on strings that don't
7735 * otherwise contain that character, but otherwise there may be
7736 * less-than-perfect results with that character and NUL. This is
7737 * unavoidable unless we replace strxfrm with our own implementation. */
7738 if (UNLIKELY(s_strlen < len)) { /* Only execute if there is an embedded
7742 STRLEN sans_nuls_len;
7743 int try_non_controls;
7744 char this_replacement_char[] = "?\0"; /* Room for a two-byte string,
7745 making sure 2nd byte is NUL.
7747 STRLEN this_replacement_len;
7749 /* If we don't know what non-NUL control character sorts lowest for
7750 * this locale, find it */
7751 if (PL_strxfrm_NUL_replacement == '\0') {
7753 char * cur_min_x = NULL; /* The min_char's xfrm, (except it also
7754 includes the collation index
7757 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "Looking to replace NUL\n"));
7759 /* Unlikely, but it may be that no control will work to replace
7760 * NUL, in which case we instead look for any character. Controls
7761 * are preferred because collation order is, in general, context
7762 * sensitive, with adjoining characters affecting the order, and
7763 * controls are less likely to have such interactions, allowing the
7764 * NUL-replacement to stand on its own. (Another way to look at it
7765 * is to imagine what would happen if the NUL were replaced by a
7766 * combining character; it wouldn't work out all that well.) */
7767 for (try_non_controls = 0;
7768 try_non_controls < 2;
7772 # ifdef USE_LOCALE_CTYPE
7774 /* In this case we use isCNTRL_LC() below, which relies on
7775 * LC_CTYPE, so that must be switched to correspond with the
7776 * LC_COLLATE locale */
7777 if (! try_non_controls && ! PL_in_utf8_COLLATE_locale) {
7778 orig_CTYPE_locale = toggle_locale_c(LC_CTYPE,
7782 /* Look through all legal code points (NUL isn't) */
7783 for (j = 1; j < 256; j++) {
7784 char * x; /* j's xfrm plus collation index */
7785 STRLEN x_len; /* length of 'x' */
7786 STRLEN trial_len = 1;
7787 char cur_source[] = { '\0', '\0' };
7789 /* Skip non-controls the first time through the loop. The
7790 * controls in a UTF-8 locale are the L1 ones */
7791 if (! try_non_controls && (PL_in_utf8_COLLATE_locale)
7798 /* Create a 1-char string of the current code point */
7799 cur_source[0] = (char) j;
7801 /* Then transform it */
7802 x = mem_collxfrm_(cur_source, trial_len, &x_len,
7803 0 /* The string is not in UTF-8 */);
7805 /* Ignore any character that didn't successfully transform.
7811 /* If this character's transformation is lower than
7812 * the current lowest, this one becomes the lowest */
7813 if ( cur_min_x == NULL
7814 || strLT(x + COLLXFRM_HDR_LEN,
7815 cur_min_x + COLLXFRM_HDR_LEN))
7817 PL_strxfrm_NUL_replacement = j;
7818 Safefree(cur_min_x);
7824 } /* end of loop through all 255 characters */
7826 # ifdef USE_LOCALE_CTYPE
7827 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
7830 /* Stop looking if found */
7835 /* Unlikely, but possible, if there aren't any controls that
7836 * work in the locale, repeat the loop, looking for any
7837 * character that works */
7838 DEBUG_L(PerlIO_printf(Perl_debug_log,
7839 "mem_collxfrm_: No control worked. Trying non-controls\n"));
7840 } /* End of loop to try first the controls, then any char */
7843 DEBUG_L(PerlIO_printf(Perl_debug_log,
7844 "mem_collxfrm_: Couldn't find any character to replace"
7845 " embedded NULs in locale %s with", PL_collation_name));
7849 DEBUG_L(PerlIO_printf(Perl_debug_log,
7850 "mem_collxfrm_: Replacing embedded NULs in locale %s with "
7851 "0x%02X\n", PL_collation_name, PL_strxfrm_NUL_replacement));
7853 Safefree(cur_min_x);
7854 } /* End of determining the character that is to replace NULs */
7856 /* If the replacement is variant under UTF-8, it must match the
7857 * UTF8-ness of the original */
7858 if ( ! UVCHR_IS_INVARIANT(PL_strxfrm_NUL_replacement) && utf8) {
7859 this_replacement_char[0] =
7860 UTF8_EIGHT_BIT_HI(PL_strxfrm_NUL_replacement);
7861 this_replacement_char[1] =
7862 UTF8_EIGHT_BIT_LO(PL_strxfrm_NUL_replacement);
7863 this_replacement_len = 2;
7866 this_replacement_char[0] = PL_strxfrm_NUL_replacement;
7867 /* this_replacement_char[1] = '\0' was done at initialization */
7868 this_replacement_len = 1;
7871 /* The worst case length for the replaced string would be if every
7872 * character in it is NUL. Multiply that by the length of each
7873 * replacement, and allow for a trailing NUL */
7874 sans_nuls_len = (len * this_replacement_len) + 1;
7875 Newx(sans_nuls, sans_nuls_len, char);
7878 /* Replace each NUL with the lowest collating control. Loop until have
7879 * exhausted all the NULs */
7880 while (s + s_strlen < e) {
7881 my_strlcat(sans_nuls, s, sans_nuls_len);
7883 /* Do the actual replacement */
7884 my_strlcat(sans_nuls, this_replacement_char, sans_nuls_len);
7886 /* Move past the input NUL */
7888 s_strlen = strlen(s);
7891 /* And add anything that trails the final NUL */
7892 my_strlcat(sans_nuls, s, sans_nuls_len);
7894 /* Switch so below we transform this modified string */
7897 } /* End of replacing NULs */
7899 /* Make sure the UTF8ness of the string and locale match */
7900 if (utf8 != PL_in_utf8_COLLATE_locale) {
7901 /* XXX convert above Unicode to 10FFFF? */
7902 const char * const t = s; /* Temporary so we can later find where the
7905 /* Here they don't match. Change the string's to be what the locale is
7908 if (! utf8) { /* locale is UTF-8, but input isn't; upgrade the input */
7909 s = (char *) bytes_to_utf8((const U8 *) s, &len);
7912 else { /* locale is not UTF-8; but input is; downgrade the input */
7914 s = (char *) bytes_from_utf8((const U8 *) s, &len, &utf8);
7916 /* If the downgrade was successful we are done, but if the input
7917 * contains things that require UTF-8 to represent, have to do
7918 * damage control ... */
7919 if (UNLIKELY(utf8)) {
7921 /* What we do is construct a non-UTF-8 string with
7922 * 1) the characters representable by a single byte converted
7923 * to be so (if necessary);
7924 * 2) and the rest converted to collate the same as the
7925 * highest collating representable character. That makes
7926 * them collate at the end. This is similar to how we
7927 * handle embedded NULs, but we use the highest collating
7928 * code point instead of the smallest. Like the NUL case,
7929 * this isn't perfect, but is the best we can reasonably
7930 * do. Every above-255 code point will sort the same as
7931 * the highest-sorting 0-255 code point. If that code
7932 * point can combine in a sequence with some other code
7933 * points for weight calculations, us changing something to
7934 * be it can adversely affect the results. But in most
7935 * cases, it should work reasonably. And note that this is
7936 * really an illegal situation: using code points above 255
7937 * on a locale where only 0-255 are valid. If two strings
7938 * sort entirely equal, then the sort order for the
7939 * above-255 code points will be in code point order. */
7943 /* If we haven't calculated the code point with the maximum
7944 * collating order for this locale, do so now */
7945 if (! PL_strxfrm_max_cp) {
7948 /* The current transformed string that collates the
7949 * highest (except it also includes the prefixed collation
7951 char * cur_max_x = NULL;
7953 /* Look through all legal code points (NUL isn't) */
7954 for (j = 1; j < 256; j++) {
7957 char cur_source[] = { '\0', '\0' };
7959 /* Create a 1-char string of the current code point */
7960 cur_source[0] = (char) j;
7962 /* Then transform it */
7963 x = mem_collxfrm_(cur_source, 1, &x_len, FALSE);
7965 /* If something went wrong (which it shouldn't), just
7966 * ignore this code point */
7971 /* If this character's transformation is higher than
7972 * the current highest, this one becomes the highest */
7973 if ( cur_max_x == NULL
7974 || strGT(x + COLLXFRM_HDR_LEN,
7975 cur_max_x + COLLXFRM_HDR_LEN))
7977 PL_strxfrm_max_cp = j;
7978 Safefree(cur_max_x);
7987 DEBUG_L(PerlIO_printf(Perl_debug_log,
7988 "mem_collxfrm_: Couldn't find any character to"
7989 " replace above-Latin1 chars in locale %s with",
7990 PL_collation_name));
7994 DEBUG_L(PerlIO_printf(Perl_debug_log,
7995 "mem_collxfrm_: highest 1-byte collating character"
7996 " in locale %s is 0x%02X\n",
7998 PL_strxfrm_max_cp));
8000 Safefree(cur_max_x);
8003 /* Here we know which legal code point collates the highest.
8004 * We are ready to construct the non-UTF-8 string. The length
8005 * will be at least 1 byte smaller than the input string
8006 * (because we changed at least one 2-byte character into a
8007 * single byte), but that is eaten up by the trailing NUL */
8013 char * e = (char *) t + len;
8015 for (i = 0; i < len; i+= UTF8SKIP(t + i)) {
8017 if (UTF8_IS_INVARIANT(cur_char)) {
8020 else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(t + i, e)) {
8021 s[d++] = EIGHT_BIT_UTF8_TO_NATIVE(cur_char, t[i+1]);
8023 else { /* Replace illegal cp with highest collating
8025 s[d++] = PL_strxfrm_max_cp;
8029 Renew(s, d, char); /* Free up unused space */
8034 /* Here, we have constructed a modified version of the input. It could
8035 * be that we already had a modified copy before we did this version.
8036 * If so, that copy is no longer needed */
8037 if (t != input_string) {
8042 length_in_chars = (utf8)
8043 ? utf8_length((U8 *) s, (U8 *) s + len)
8046 /* The first element in the output is the collation id, used by
8047 * sv_collxfrm(); then comes the space for the transformed string. The
8048 * equation should give us a good estimate as to how much is needed */
8049 xAlloc = COLLXFRM_HDR_LEN
8051 + (PL_collxfrm_mult * length_in_chars);
8052 Newx(xbuf, xAlloc, char);
8053 if (UNLIKELY(! xbuf)) {
8054 DEBUG_L(PerlIO_printf(Perl_debug_log,
8055 "mem_collxfrm_: Couldn't malloc %zu bytes\n", xAlloc));
8059 /* Store the collation id */
8060 *(PERL_UINTMAX_T *)xbuf = PL_collation_ix;
8062 # if defined(USE_POSIX_2008_LOCALE) && defined HAS_STRXFRM_L
8063 # ifdef USE_LOCALE_CTYPE
8065 constructed_locale = newlocale(LC_CTYPE_MASK, PL_collation_name,
8066 duplocale(use_curlocale_scratch()));
8069 constructed_locale = duplocale(use_curlocale_scratch());
8072 # define my_strxfrm(dest, src, n) strxfrm_l(dest, src, n, \
8074 # define CLEANUP_STRXFRM \
8076 if (constructed_locale != (locale_t) 0) \
8077 freelocale(constructed_locale); \
8080 # define my_strxfrm(dest, src, n) strxfrm(dest, src, n)
8081 # ifdef USE_LOCALE_CTYPE
8083 orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, PL_collation_name);
8085 # define CLEANUP_STRXFRM \
8086 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale)
8088 # define CLEANUP_STRXFRM NOOP
8092 /* Then the transformation of the input. We loop until successful, or we
8097 *xlen = my_strxfrm(xbuf + COLLXFRM_HDR_LEN,
8099 xAlloc - COLLXFRM_HDR_LEN);
8102 /* If the transformed string occupies less space than we told strxfrm()
8103 * was available, it means it transformed the whole string. */
8104 if (*xlen < xAlloc - COLLXFRM_HDR_LEN) {
8106 /* But there still could have been a problem */
8108 DEBUG_L(PerlIO_printf(Perl_debug_log,
8109 "strxfrm failed for LC_COLLATE=%s; errno=%d, input=%s\n",
8110 PL_collation_name, errno,
8111 _byte_dump_string((U8 *) s, len, 0)));
8115 /* Here, the transformation was successful. Some systems include a
8116 * trailing NUL in the returned length. Ignore it, using a loop in
8117 * case multiple trailing NULs are returned. */
8119 && *(xbuf + COLLXFRM_HDR_LEN + (*xlen) - 1) == '\0')
8124 /* If the first try didn't get it, it means our prediction was low.
8125 * Modify the coefficients so that we predict a larger value in any
8126 * future transformations */
8128 STRLEN needed = *xlen + 1; /* +1 For trailing NUL */
8129 STRLEN computed_guess = PL_collxfrm_base
8130 + (PL_collxfrm_mult * length_in_chars);
8132 /* On zero-length input, just keep current slope instead of
8134 const STRLEN new_m = (length_in_chars != 0)
8135 ? needed / length_in_chars
8138 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8139 "initial size of %zu bytes for a length "
8140 "%zu string was insufficient, %zu needed\n",
8141 computed_guess, length_in_chars, needed));
8143 /* If slope increased, use it, but discard this result for
8144 * length 1 strings, as we can't be sure that it's a real slope
8146 if (length_in_chars > 1 && new_m > PL_collxfrm_mult) {
8150 STRLEN old_m = PL_collxfrm_mult;
8151 STRLEN old_b = PL_collxfrm_base;
8155 PL_collxfrm_mult = new_m;
8156 PL_collxfrm_base = 1; /* +1 For trailing NUL */
8157 computed_guess = PL_collxfrm_base
8158 + (PL_collxfrm_mult * length_in_chars);
8159 if (computed_guess < needed) {
8160 PL_collxfrm_base += needed - computed_guess;
8163 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8164 "slope is now %zu; was %zu, base "
8165 "is now %zu; was %zu\n",
8166 PL_collxfrm_mult, old_m,
8167 PL_collxfrm_base, old_b));
8169 else { /* Slope didn't change, but 'b' did */
8170 const STRLEN new_b = needed
8173 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8174 "base is now %zu; was %zu\n", new_b, PL_collxfrm_base));
8175 PL_collxfrm_base = new_b;
8182 if (UNLIKELY(*xlen >= PERL_INT_MAX)) {
8183 DEBUG_L(PerlIO_printf(Perl_debug_log,
8184 "mem_collxfrm_: Needed %zu bytes, max permissible is %u\n",
8185 *xlen, PERL_INT_MAX));
8189 /* A well-behaved strxfrm() returns exactly how much space it needs
8190 * (usually not including the trailing NUL) when it fails due to not
8191 * enough space being provided. Assume that this is the case unless
8192 * it's been proven otherwise */
8193 if (LIKELY(PL_strxfrm_is_behaved) && first_time) {
8194 xAlloc = *xlen + COLLXFRM_HDR_LEN + 1;
8196 else { /* Here, either:
8197 * 1) The strxfrm() has previously shown bad behavior; or
8198 * 2) It isn't the first time through the loop, which means
8199 * that the strxfrm() is now showing bad behavior, because
8200 * we gave it what it said was needed in the previous
8201 * iteration, and it came back saying it needed still more.
8202 * (Many versions of cygwin fit this. When the buffer size
8203 * isn't sufficient, they return the input size instead of
8204 * how much is needed.)
8205 * Increase the buffer size by a fixed percentage and try again.
8207 xAlloc += (xAlloc / 4) + 1;
8208 PL_strxfrm_is_behaved = FALSE;
8210 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8211 "mem_collxfrm_ required more space than previously"
8212 " calculated for locale %s, trying again with new"
8214 PL_collation_name, COLLXFRM_HDR_LEN,
8215 xAlloc - COLLXFRM_HDR_LEN));
8218 Renew(xbuf, xAlloc, char);
8219 if (UNLIKELY(! xbuf)) {
8220 DEBUG_L(PerlIO_printf(Perl_debug_log,
8221 "mem_collxfrm_: Couldn't realloc %zu bytes\n", xAlloc));
8230 DEBUG_L(print_collxfrm_input_and_return(s, s + len, xbuf, *xlen, utf8));
8232 /* Free up unneeded space; retain enough for trailing NUL */
8233 Renew(xbuf, COLLXFRM_HDR_LEN + *xlen + 1, char);
8235 if (s != input_string) {
8244 DEBUG_L(print_collxfrm_input_and_return(s, s + len, NULL, 0, utf8));
8247 if (s != input_string) {
8258 S_print_collxfrm_input_and_return(pTHX_
8266 PERL_ARGS_ASSERT_PRINT_COLLXFRM_INPUT_AND_RETURN;
8268 PerlIO_printf(Perl_debug_log,
8269 "mem_collxfrm_[ix %" UVuf "] for locale '%s':\n"
8270 " input=%s\n return=%s\n return len=%zu\n",
8271 (UV) PL_collation_ix, PL_collation_name,
8272 get_displayable_string(s, e, is_utf8),
8277 : _byte_dump_string((U8 *) xbuf + COLLXFRM_HDR_LEN,
8282 # endif /* DEBUGGING */
8285 Perl_strxfrm(pTHX_ SV * src)
8287 PERL_ARGS_ASSERT_STRXFRM;
8289 /* For use by POSIX::strxfrm(). If they differ, toggle LC_CTYPE to
8290 * LC_COLLATE to avoid potential mojibake.
8292 * If we can't calculate a collation, 'src' is instead returned, so that
8293 * future comparisons will be by code point order */
8295 # ifdef USE_LOCALE_CTYPE
8297 const char * orig_ctype = toggle_locale_c(LC_CTYPE,
8298 querylocale_c(LC_COLLATE));
8304 const char *p = SvPV_const(src, srclen);
8305 const U32 utf8_flag = SvUTF8(src);
8306 char *d = mem_collxfrm_(p, srclen, &dstlen, cBOOL(utf8_flag));
8308 assert(utf8_flag == 0 || utf8_flag == SVf_UTF8);
8312 dst =newSVpvn_flags(d + COLLXFRM_HDR_LEN,
8313 dstlen, SVs_TEMP|utf8_flag);
8317 # ifdef USE_LOCALE_CTYPE
8319 restore_toggled_locale_c(LC_CTYPE, orig_ctype);
8326 #endif /* USE_LOCALE_COLLATE */
8330 S_toggle_locale_i(pTHX_ const locale_category_index cat_index,
8331 const char * new_locale,
8332 const line_t caller_line)
8334 PERL_ARGS_ASSERT_TOGGLE_LOCALE_I;
8335 assert(cat_index <= LC_ALL_INDEX_);
8337 /* Changes the locale for the category specified by 'index' to 'new_locale,
8338 * if they aren't already the same.
8340 * Returns a copy of the name of the original locale for 'cat_index'
8341 * so can be switched back to with the companion function
8342 * restore_toggled_locale_i(), (NULL if no restoral is necessary.) */
8344 /* Find the original locale of the category we may need to change, so that
8345 * it can be restored to later */
8346 const char * locale_to_restore_to = querylocale_i(cat_index);
8348 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8349 "(%" LINE_Tf "): toggle_locale_i: index=%d(%s), wanted=%s,"
8351 caller_line, cat_index, category_names[cat_index],
8352 new_locale, locale_to_restore_to));
8354 if (! locale_to_restore_to) {
8355 locale_panic_via_(Perl_form(aTHX_
8356 "Could not find current %s locale",
8357 category_names[cat_index]),
8358 __FILE__, caller_line);
8361 /* If the locales are the same, there's nothing to do */
8362 if (strEQ(locale_to_restore_to, new_locale)) {
8363 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8364 "(%" LINE_Tf "): %s locale unchanged as %s\n",
8365 caller_line, category_names[cat_index],
8371 /* Finally, change the locale to the new one */
8372 void_setlocale_i_with_caller(cat_index, new_locale, __FILE__, caller_line);
8374 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8375 "(%" LINE_Tf "): %s locale switched to %s\n",
8376 caller_line, category_names[cat_index], new_locale));
8378 return locale_to_restore_to;
8381 PERL_UNUSED_ARG(caller_line);
8387 S_restore_toggled_locale_i(pTHX_ const locale_category_index cat_index,
8388 const char * restore_locale,
8389 const line_t caller_line)
8391 /* Restores the locale for LC_category corresponding to cat_index to
8392 * 'restore_locale' (which is a copy that will be freed by this function),
8393 * or do nothing if the latter parameter is NULL */
8395 PERL_ARGS_ASSERT_RESTORE_TOGGLED_LOCALE_I;
8396 assert(cat_index <= LC_ALL_INDEX_);
8398 if (restore_locale == NULL) {
8399 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8400 "(%" LINE_Tf "): No need to restore %s\n",
8401 caller_line, category_names[cat_index]));
8405 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8406 "(%" LINE_Tf "): %s restoring locale to %s\n",
8407 caller_line, category_names[cat_index],
8410 void_setlocale_i_with_caller(cat_index, restore_locale,
8411 __FILE__, caller_line);
8414 PERL_UNUSED_ARG(caller_line);
8419 # ifdef USE_LOCALE_CTYPE
8422 S_is_codeset_name_UTF8(const char * name)
8424 /* Return a boolean as to if the passed-in name indicates it is a UTF-8
8425 * code set. Several variants are possible */
8426 const Size_t len = strlen(name);
8428 PERL_ARGS_ASSERT_IS_CODESET_NAME_UTF8;
8432 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
8433 if (memENDs(name, len, "65001")) {
8438 /* 'UTF8' or 'UTF-8' */
8439 return ( inRANGE(len, 4, 5)
8440 && name[len-1] == '8'
8441 && ( memBEGINs(name, len, "UTF")
8442 || memBEGINs(name, len, "utf"))
8443 && (len == 4 || name[3] == '-'));
8447 #endif /* USE_LOCALE */
8450 Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
8452 /* Internal function which returns if we are in the scope of a pragma that
8453 * enables the locale category 'category'. 'compiling' should indicate if
8454 * this is during the compilation phase (TRUE) or not (FALSE). */
8456 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
8458 SV *these_categories = cop_hints_fetch_pvs(cop, "locale", 0);
8459 if (! these_categories || these_categories == &PL_sv_placeholder) {
8463 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
8464 * a valid unsigned */
8465 assert(category >= -1);
8466 return cBOOL(SvUV(these_categories) & (1U << (category + 1)));
8469 /* my_strerror() returns a mortalized copy of the text of the error message
8470 * associated with 'errnum'.
8472 * If not called from within the scope of 'use locale', it uses the text from
8473 * the C locale. If Perl is compiled to not pay attention to LC_CTYPE nor
8474 * LC_MESSAGES, it uses whatever strerror() returns. Otherwise the text is
8475 * derived from the locale, LC_MESSAGES if we have that; LC_CTYPE if not.
8477 * It returns in *utf8ness the result's UTF-8ness
8479 * The function just calls strerror(), but temporarily switches locales, if
8480 * needed. Many platforms require LC_CTYPE and LC_MESSAGES to be in the same
8481 * CODESET in order for the return from strerror() to not contain '?' symbols,
8482 * or worse, mojibaked. It's cheaper to just use the stricter criteria of
8483 * being in the same locale. So the code below uses a common locale for both
8484 * categories. Again, that is C if not within 'use locale' scope; or the
8485 * LC_MESSAGES locale if in scope and we have that category; and LC_CTYPE if we
8486 * don't have LC_MESSAGES; and whatever strerror returns if we don't have
8489 * There are two sets of implementations. The first below is if we have
8490 * strerror_l(). This is the simpler. We just use the already-built C locale
8491 * object if not in locale scope, or build up a custom one otherwise.
8493 * When strerror_l() is not available, we may have to swap locales temporarily
8494 * to bring the two categories into sync with each other, and possibly to the C
8497 * Because the prepropessing directives to conditionally compile this function
8498 * would greatly obscure the logic of the various implementations, the whole
8499 * function is repeated for each configuration, with some common macros. */
8501 /* Used to shorten the definitions of the following implementations of
8503 #define DEBUG_STRERROR_ENTER(errnum, in_locale) \
8504 DEBUG_Lv(PerlIO_printf(Perl_debug_log, \
8505 "my_strerror called with errnum %d;" \
8506 " Within locale scope=%d\n", \
8509 #define DEBUG_STRERROR_RETURN(errstr, utf8ness) \
8510 DEBUG_Lv(PerlIO_printf(Perl_debug_log, \
8511 "Strerror returned; saving a copy: '%s';" \
8513 get_displayable_string(errstr, \
8514 errstr + strlen(errstr), \
8518 /* On platforms that have precisely one of these categories (Windows
8519 * qualifies), these yield the correct one */
8520 #if defined(USE_LOCALE_CTYPE)
8521 # define WHICH_LC_INDEX LC_CTYPE_INDEX_
8522 #elif defined(USE_LOCALE_MESSAGES)
8523 # define WHICH_LC_INDEX LC_MESSAGES_INDEX_
8526 /*===========================================================================*/
8527 /* First set of implementations, when have strerror_l() */
8529 #if defined(USE_POSIX_2008_LOCALE) && defined(HAS_STRERROR_L)
8531 # if ! defined(USE_LOCALE_CTYPE) && ! defined(USE_LOCALE_MESSAGES)
8533 /* Here, neither category is defined: use the C locale */
8535 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8537 PERL_ARGS_ASSERT_MY_STRERROR;
8539 DEBUG_STRERROR_ENTER(errnum, 0);
8541 const char *errstr = savepv(strerror_l(errnum, PL_C_locale_obj));
8542 *utf8ness = UTF8NESS_IMMATERIAL;
8544 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8550 # elif ! defined(USE_LOCALE_CTYPE) || ! defined(USE_LOCALE_MESSAGES)
8552 /*--------------------------------------------------------------------------*/
8554 /* Here one or the other of CTYPE or MESSAGES is defined, but not both. If we
8555 * are not within 'use locale' scope of the only one defined, we use the C
8556 * locale; otherwise use the current locale object */
8559 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8561 PERL_ARGS_ASSERT_MY_STRERROR;
8563 DEBUG_STRERROR_ENTER(errnum, IN_LC(categories[WHICH_LC_INDEX]));
8565 /* Use C if not within locale scope; Otherwise, use current locale */
8566 const locale_t which_obj = (IN_LC(categories[WHICH_LC_INDEX]))
8568 : use_curlocale_scratch();
8570 const char *errstr = savepv(strerror_l(errnum, which_obj));
8571 *utf8ness = get_locale_string_utf8ness_i(errstr, LOCALE_UTF8NESS_UNKNOWN,
8572 NULL, WHICH_LC_INDEX);
8573 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8579 /*--------------------------------------------------------------------------*/
8580 # else /* Are using both categories. Place them in the same CODESET,
8581 * either C or the LC_MESSAGES locale */
8584 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8586 PERL_ARGS_ASSERT_MY_STRERROR;
8588 DEBUG_STRERROR_ENTER(errnum, IN_LC(LC_MESSAGES));
8591 if (! IN_LC(LC_MESSAGES)) { /* Use C if not within locale scope */
8592 errstr = savepv(strerror_l(errnum, PL_C_locale_obj));
8593 *utf8ness = UTF8NESS_IMMATERIAL;
8595 else { /* Otherwise, use the LC_MESSAGES locale, making sure LC_CTYPE
8597 locale_t cur = duplocale(use_curlocale_scratch());
8599 cur = newlocale(LC_CTYPE_MASK, querylocale_c(LC_MESSAGES), cur);
8600 errstr = savepv(strerror_l(errnum, cur));
8601 *utf8ness = get_locale_string_utf8ness_i(errstr,
8602 LOCALE_UTF8NESS_UNKNOWN,
8603 NULL, LC_MESSAGES_INDEX_);
8607 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8612 # endif /* Above is using strerror_l */
8613 /*===========================================================================*/
8614 #else /* Below is not using strerror_l */
8615 # if ! defined(USE_LOCALE_CTYPE) && ! defined(USE_LOCALE_MESSAGES)
8617 /* If not using using either of the categories, return plain, unadorned
8621 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8623 PERL_ARGS_ASSERT_MY_STRERROR;
8625 DEBUG_STRERROR_ENTER(errnum, 0);
8627 const char *errstr = savepv(Strerror(errnum));
8628 *utf8ness = UTF8NESS_IMMATERIAL;
8630 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8636 /*--------------------------------------------------------------------------*/
8637 # elif ! defined(USE_LOCALE_CTYPE) || ! defined(USE_LOCALE_MESSAGES)
8639 /* Here one or the other of CTYPE or MESSAGES is defined, but not both. If we
8640 * are not within 'use locale' scope of the only one defined, we use the C
8641 * locale; otherwise use the current locale */
8644 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8646 PERL_ARGS_ASSERT_MY_STRERROR;
8648 DEBUG_STRERROR_ENTER(errnum, IN_LC(categories[WHICH_LC_INDEX]));
8651 if (IN_LC(categories[WHICH_LC_INDEX])) {
8652 errstr = savepv(Strerror(errnum));
8653 *utf8ness = get_locale_string_utf8ness_i(errstr,
8654 LOCALE_UTF8NESS_UNKNOWN,
8655 NULL, WHICH_LC_INDEX);
8661 const char * orig_locale = toggle_locale_i(WHICH_LC_INDEX, "C");
8663 errstr = savepv(Strerror(errnum));
8665 restore_toggled_locale_i(WHICH_LC_INDEX, orig_locale);
8669 *utf8ness = UTF8NESS_IMMATERIAL;
8672 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8678 /*--------------------------------------------------------------------------*/
8681 /* Below, have both LC_CTYPE and LC_MESSAGES. Place them in the same CODESET,
8682 * either C or the LC_MESSAGES locale */
8685 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8687 PERL_ARGS_ASSERT_MY_STRERROR;
8689 DEBUG_STRERROR_ENTER(errnum, IN_LC(LC_MESSAGES));
8691 const char * desired_locale = (IN_LC(LC_MESSAGES))
8692 ? querylocale_c(LC_MESSAGES)
8694 /* XXX Can fail on z/OS */
8698 const char* orig_CTYPE_locale = toggle_locale_c(LC_CTYPE,
8700 const char* orig_MESSAGES_locale = toggle_locale_c(LC_MESSAGES,
8702 const char *errstr = savepv(Strerror(errnum));
8704 restore_toggled_locale_c(LC_MESSAGES, orig_MESSAGES_locale);
8705 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
8709 *utf8ness = get_locale_string_utf8ness_i(errstr, LOCALE_UTF8NESS_UNKNOWN,
8710 NULL, LC_MESSAGES_INDEX_);
8711 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8717 /*--------------------------------------------------------------------------*/
8718 # endif /* end of not using strerror_l() */
8719 #endif /* end of all the my_strerror() implementations */
8723 =for apidoc switch_to_global_locale
8725 This function copies the locale state of the calling thread into the program's
8726 global locale, and converts the thread to use that global locale.
8728 It is intended so that Perl can safely be used with C libraries that access the
8729 global locale and which can't be converted to not access it. Effectively, this
8730 means libraries that call C<L<setlocale(3)>> on non-Windows systems. (For
8731 portability, it is a good idea to use it on Windows as well.)
8733 A downside of using it is that it disables the services that Perl provides to
8734 hide locale gotchas from your code. The service you most likely will miss
8735 regards the radix character (decimal point) in floating point numbers. Code
8736 executed after this function is called can no longer just assume that this
8737 character is correct for the current circumstances.
8739 To return to Perl control, and restart the gotcha prevention services, call
8740 C<L</sync_locale>>. Behavior is undefined for any pure Perl code that executes
8741 while the switch is in effect.
8743 The global locale and the per-thread locales are independent. As long as just
8744 one thread converts to the global locale, everything works smoothly. But if
8745 more than one does, they can easily interfere with each other, and races are
8746 likely. On Windows systems prior to Visual Studio 15 (at which point Microsoft
8747 fixed a bug), races can occur (even if only one thread has been converted to
8748 the global locale), but only if you use the following operations:
8752 =item L<POSIX::localeconv|POSIX/localeconv>
8754 =item L<I18N::Langinfo>, items C<CRNCYSTR> and C<THOUSEP>
8756 =item L<perlapi/Perl_langinfo>, items C<CRNCYSTR> and C<THOUSEP>
8760 The first item is not fixable (except by upgrading to a later Visual Studio
8761 release), but it would be possible to work around the latter two items by
8762 having Perl change its algorithm for calculating these to use Windows API
8763 functions (likely C<GetNumberFormat> and C<GetCurrencyFormat>); patches
8766 XS code should never call plain C<setlocale>, but should instead be converted
8767 to either call L<C<Perl_setlocale>|perlapi/Perl_setlocale> (which is a drop-in
8768 for the system C<setlocale>) or use the methods given in L<perlcall> to call
8769 L<C<POSIX::setlocale>|POSIX/setlocale>. Either one will transparently properly
8770 handle all cases of single- vs multi-thread, POSIX 2008-supported or not.
8775 #if defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE)
8776 # define CHANGE_SYSTEM_LOCALE_TO_GLOBAL \
8778 if (_configthreadlocale(_DISABLE_PER_THREAD_LOCALE) == -1) { \
8779 locale_panic_("_configthreadlocale returned an error"); \
8782 #elif defined(USE_POSIX_2008_LOCALE)
8783 # define CHANGE_SYSTEM_LOCALE_TO_GLOBAL \
8785 locale_t old_locale = uselocale(LC_GLOBAL_LOCALE); \
8786 if (! old_locale) { \
8787 locale_panic_("Could not change to global locale"); \
8790 /* Free the per-thread memory */ \
8791 if ( old_locale != LC_GLOBAL_LOCALE \
8792 && old_locale != PL_C_locale_obj) \
8794 freelocale(old_locale); \
8798 # define CHANGE_SYSTEM_LOCALE_TO_GLOBAL
8802 Perl_switch_to_global_locale(pTHX)
8807 DEBUG_L(PerlIO_printf(Perl_debug_log, "Entering switch_to_global; %s\n",
8808 get_LC_ALL_display()));
8810 /* In these cases, we use the system state to determine if we are in the
8811 * global locale or not. */
8812 # ifdef USE_POSIX_2008_LOCALE
8814 const bool perl_controls = (LC_GLOBAL_LOCALE != uselocale((locale_t) 0));
8816 # elif defined(USE_THREAD_SAFE_LOCALE) && defined(WIN32)
8818 int config_return = _configthreadlocale(0);
8819 if (config_return == -1) {
8820 locale_panic_("_configthreadlocale returned an error");
8822 const bool perl_controls = (config_return == _ENABLE_PER_THREAD_LOCALE);
8826 const bool perl_controls = false;
8830 /* No-op if already in global */
8831 if (! perl_controls) {
8837 const char * thread_locale = calculate_LC_ALL_string(NULL,
8838 EXTERNAL_FORMAT_FOR_SET,
8841 CHANGE_SYSTEM_LOCALE_TO_GLOBAL;
8842 posix_setlocale(LC_ALL, thread_locale);
8844 # else /* Must be USE_POSIX_2008_LOCALE) */
8846 const char * cur_thread_locales[LC_ALL_INDEX_];
8848 /* Save each category's current per-thread state */
8849 for_all_individual_category_indexes(i) {
8850 cur_thread_locales[i] = querylocale_i(i);
8853 CHANGE_SYSTEM_LOCALE_TO_GLOBAL;
8855 /* Set the global to what was our per-thread state */
8856 POSIX_SETLOCALE_LOCK;
8857 for_all_individual_category_indexes(i) {
8858 posix_setlocale(categories[i], cur_thread_locales[i]);
8860 POSIX_SETLOCALE_UNLOCK;
8863 # ifdef USE_LOCALE_NUMERIC
8865 /* Switch to the underlying C numeric locale; the application is on its
8867 POSIX_SETLOCALE_LOCK;
8868 posix_setlocale(LC_NUMERIC, PL_numeric_name);
8869 POSIX_SETLOCALE_UNLOCK;
8878 =for apidoc sync_locale
8880 This function copies the state of the program global locale into the calling
8881 thread, and converts that thread to using per-thread locales, if it wasn't
8882 already, and the platform supports them. The LC_NUMERIC locale is toggled into
8883 the standard state (using the C locale's conventions), if not within the
8884 lexical scope of S<C<use locale>>.
8886 Perl will now consider itself to have control of the locale.
8888 Since unthreaded perls have only a global locale, this function is a no-op
8891 This function is intended for use with C libraries that do locale manipulation.
8892 It allows Perl to accommodate the use of them. Call this function before
8893 transferring back to Perl space so that it knows what state the C code has left
8896 XS code should not manipulate the locale on its own. Instead,
8897 L<C<Perl_setlocale>|perlapi/Perl_setlocale> can be used at any time to query or
8898 change the locale (though changing the locale is antisocial and dangerous on
8899 multi-threaded systems that don't have multi-thread safe locale operations.
8900 (See L<perllocale/Multi-threaded operation>).
8902 Using the libc L<C<setlocale(3)>> function should be avoided. Nevertheless,
8903 certain non-Perl libraries called from XS, do call it, and their behavior may
8904 not be able to be changed. This function, along with
8905 C<L</switch_to_global_locale>>, can be used to get seamless behavior in these
8906 circumstances, as long as only one thread is involved.
8908 If the library has an option to turn off its locale manipulation, doing that is
8909 preferable to using this mechanism. C<Gtk> is such a library.
8911 The return value is a boolean: TRUE if the global locale at the time of call
8912 was in effect for the caller; and FALSE if a per-thread locale was in effect.
8918 Perl_sync_locale(pTHX)
8927 bool was_in_global = TRUE;
8929 # ifdef USE_THREAD_SAFE_LOCALE
8932 int config_return = _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
8933 if (config_return == -1) {
8934 locale_panic_("_configthreadlocale returned an error");
8936 was_in_global = (config_return == _DISABLE_PER_THREAD_LOCALE);
8938 # elif defined(USE_POSIX_2008_LOCALE)
8940 was_in_global = (LC_GLOBAL_LOCALE == uselocale(LC_GLOBAL_LOCALE));
8943 # error Unexpected Configuration
8945 # endif /* USE_THREAD_SAFE_LOCALE */
8947 /* Here, we are in the global locale. Get and save the values for each
8948 * category, and convert the current thread to use them */
8952 STDIZED_SETLOCALE_LOCK;
8953 const char * lc_all_string = savepv(stdized_setlocale(LC_ALL, NULL));
8954 STDIZED_SETLOCALE_UNLOCK;
8956 give_perl_locale_control(lc_all_string, __LINE__);
8957 Safefree(lc_all_string);
8961 const char * current_globals[LC_ALL_INDEX_];
8962 for_all_individual_category_indexes(i) {
8963 STDIZED_SETLOCALE_LOCK;
8964 current_globals[i] = savepv(stdized_setlocale(categories[i], NULL));
8965 STDIZED_SETLOCALE_UNLOCK;
8968 give_perl_locale_control((const char **) ¤t_globals, __LINE__);
8970 for_all_individual_category_indexes(i) {
8971 Safefree(current_globals[i]);
8976 return was_in_global;
8982 #if defined(DEBUGGING) && defined(USE_LOCALE)
8985 S_my_setlocale_debug_string_i(pTHX_
8986 const locale_category_index cat_index,
8987 const char* locale, /* Optional locale name */
8989 /* return value from setlocale() when attempting
8990 * to set 'category' to 'locale' */
8995 /* Returns a pointer to a NUL-terminated string in static storage with
8996 * added text about the info passed in. This is not thread safe and will
8997 * be overwritten by the next call, so this should be used just to
8998 * formulate a string to immediately print or savepv() on. */
9000 const char * locale_quote;
9001 const char * retval_quote;
9003 assert(cat_index <= LC_ALL_INDEX_);
9005 if (locale == NULL) {
9010 locale_quote = "\"";
9013 if (retval == NULL) {
9018 retval_quote = "\"";
9021 # ifdef USE_LOCALE_THREADS
9022 # define THREAD_FORMAT "%p:"
9023 # define THREAD_ARGUMENT aTHX_
9025 # define THREAD_FORMAT
9026 # define THREAD_ARGUMENT
9029 return Perl_form(aTHX_
9030 "%s:%" LINE_Tf ": " THREAD_FORMAT
9031 " setlocale(%s[%d], %s%s%s) returned %s%s%s\n",
9033 __FILE__, line, THREAD_ARGUMENT
9034 category_names[cat_index], categories[cat_index],
9035 locale_quote, locale, locale_quote,
9036 retval_quote, retval, retval_quote);
9040 #ifdef USE_PERL_SWITCH_LOCALE_CONTEXT
9043 Perl_switch_locale_context(pTHX)
9045 /* libc keeps per-thread locale status information in some configurations.
9046 * So, we can't just switch out aTHX to switch to a new thread. libc has
9047 * to follow along. This routine does that based on per-interpreter
9048 * variables we keep just for this purpose.
9050 * There are two implementations where this is an issue. For the other
9051 * implementations, it doesn't matter because libc is using global values
9052 * that all threads know about.
9054 * The two implementations are where libc keeps thread-specific information
9055 * on its own. These are
9057 * POSIX 2008: The current locale is kept by libc as an object. We save
9058 * a copy of that in the per-thread PL_cur_locale_obj, and so
9059 * this routine uses that copy to tell the thread it should be
9060 * operating with that object
9061 * Windows thread-safe locales: A given thread in Windows can be being run
9062 * with per-thread locales, or not. When the thread context
9063 * changes, libc doesn't automatically know if the thread is
9064 * using per-thread locales, nor does it know what the new
9065 * thread's locale is. We keep that information in the
9066 * per-thread variables:
9067 * PL_controls_locale indicates if this thread is using
9068 * per-thread locales or not
9069 * PL_cur_LC_ALL indicates what the the locale
9070 * should be if it is a per-thread
9074 if (UNLIKELY( PL_veto_switch_non_tTHX_context
9075 || PL_phase == PERL_PHASE_CONSTRUCT))
9080 # ifdef USE_POSIX_2008_LOCALE
9082 if (! uselocale(PL_cur_locale_obj)) {
9083 locale_panic_(Perl_form(aTHX_
9084 "Can't uselocale(%p), LC_ALL supposed to"
9086 PL_cur_locale_obj, get_LC_ALL_display()));
9089 # elif defined(WIN32)
9091 if (! bool_setlocale_c(LC_ALL, PL_cur_LC_ALL)) {
9092 locale_panic_(Perl_form(aTHX_ "Can't setlocale(%s)", PL_cur_LC_ALL));
9102 Perl_thread_locale_init(pTHX)
9105 #ifdef USE_THREAD_SAFE_LOCALE
9106 # ifdef USE_POSIX_2008_LOCALE
9108 /* Called from a thread on startup.
9110 * The operations here have to be done from within the calling thread, as
9111 * they affect libc's knowledge of the thread; libc has no knowledge of
9114 DEBUG_L(PerlIO_printf(Perl_debug_log,
9115 "new thread, initial locale is %s;"
9116 " calling setlocale(LC_ALL, \"C\")\n",
9117 get_LC_ALL_display()));
9119 if (! uselocale(PL_C_locale_obj)) {
9121 /* Not being able to change to the C locale is severe; don't keep
9123 locale_panic_(Perl_form(aTHX_
9124 "Can't uselocale(%p), 'C'", PL_C_locale_obj));
9125 NOT_REACHED; /* NOTREACHED */
9128 # ifdef MULTIPLICITY
9130 PL_cur_locale_obj = PL_C_locale_obj;
9133 # elif defined(WIN32)
9135 /* On Windows, make sure new thread has per-thread locales enabled */
9136 if (_configthreadlocale(_ENABLE_PER_THREAD_LOCALE) == -1) {
9137 locale_panic_("_configthreadlocale returned an error");
9139 void_setlocale_c(LC_ALL, "C");
9147 Perl_thread_locale_term(pTHX)
9149 /* Called from a thread as it gets ready to terminate.
9151 * The operations here have to be done from within the calling thread, as
9152 * they affect libc's knowledge of the thread; libc has no knowledge of
9155 #if defined(USE_POSIX_2008_LOCALE) && defined(USE_THREADS)
9157 /* Switch to the global locale, so can free up the per-thread object */
9158 locale_t actual_obj = uselocale(LC_GLOBAL_LOCALE);
9159 if (actual_obj != LC_GLOBAL_LOCALE && actual_obj != PL_C_locale_obj) {
9160 freelocale(actual_obj);
9163 /* Prevent leaks even if something has gone wrong */
9164 locale_t expected_obj = PL_cur_locale_obj;
9165 if (UNLIKELY( expected_obj != actual_obj
9166 && expected_obj != LC_GLOBAL_LOCALE
9167 && expected_obj != PL_C_locale_obj))
9169 freelocale(expected_obj);
9172 PL_cur_locale_obj = LC_GLOBAL_LOCALE;
9175 #ifdef WIN32_USE_FAKE_OLD_MINGW_LOCALES
9177 /* When faking the mingw implementation, we coerce this function into doing
9178 * something completely different from its intent -- namely to free up our
9179 * static buffer to avoid a leak. This function gets called for each
9180 * thread that is terminating, so will give us a chance to free the buffer
9181 * from the appropriate pool. On unthreaded systems, it gets called by the
9182 * mutex termination code. */
9184 # ifdef MULTIPLICITY
9186 if (aTHX != wsetlocale_buf_aTHX) {
9192 if (wsetlocale_buf_size > 0) {
9193 Safefree(wsetlocale_buf);
9194 wsetlocale_buf_size = 0;
9202 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob