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; in order to be activated, it requires perl to be
98 * Configured with a parameter indicating the platform's defect. 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=-DUSE_NL_LOCALE_NAME
256 * glibc has an undocumented equivalent function to querylocale(). It
257 * currently isn't used by default because it is undocumented. But
258 * testing hasn't found any problems with it. Using this Configure
259 * option enables it on systems that have it (with no effect on
260 * systems lacking it). Enabling this removes the need for perl
261 * to keep its own records, hence is more efficient and guaranteed to
264 * -Accflags=-DNO_LOCALE_CTYPE
265 * -Accflags=-DNO_LOCALE_NUMERIC
268 * If the named category(ies) does(do) not exist on this platform,
269 * these have no effect. Otherwise they cause perl to be compiled to
270 * always keep the named category(ies) in the C locale.
272 * -Accflags=-DHAS_BROKEN_SETLOCALE_QUERY_LC_ALL
273 * This would be set in a hints file to tell perl that doing a libc
274 * setlocale(LC_ALL, NULL)
275 * can give erroneous results, and perl will compensate to get the
276 * correct results. This is known to be a problem in earlier AIX
279 * -Accflags=-DHAS_LF_IN_SETLOCALE_RETURN
280 * This would be set in a hints file to tell perl that a libc
281 * setlocale() can return results containing \n characters that need
282 * to be stripped off. khw believes there aren't any such platforms
283 * still in existence.
285 * -Accflags=-DLIBC_HANDLES_MISMATCHED_CTYPE
286 * Consider the name of a month in some language, Chinese for example.
287 * If LC_TIME has been set to a Chinese locale, strftime() can be used
288 * to generate the Chinese month name for any given date, by using the
289 * %B format. But also suppose that LC_CTYPE is set to, say, "C".
290 * The return from strftime() on many platforms will be mojibake given
291 * that no Chinese month name is composed of just ASCII characters.
292 * Perl handles this for you by automatically toggling LC_CTYPE to
293 * whatever LC_TIME is during the execution of strftime(), and
294 * afterwards restoring it to its prior value. But the strftime()
295 * (and similar functions) in some libc implementations already do
296 * this toggle, meaning perl's action is redundant. You can tell perl
297 * that a libc does this by setting this Configure option, and it will
298 * skip its syncing LC_CTYPE and whatever the other locale is.
299 * Currently, perl ignores this Configuration option and syncs anyway
300 * for LC_COLLATE-related operations, due to perl's internal needs.
302 * -Accflags=USE_FAKE_LC_ALL_POSITIONAL_NOTATION
303 * This is used when developing Perl on a platform that uses
304 * 'name=value;' notation to represent LC_ALL when not all categories
305 * are the same. When so compiled, much of the code gets compiled
306 * and exercised that applies to platforms that instead use positional
307 * notation. This allows for finding many bugs in that portion of the
308 * implementation, without having to access such a platform.
310 * -Accflags=-DWIN32_USE_FAKE_OLD_MINGW_LOCALES
311 * This is used when developing Perl on a non-Windows platform to
312 * compile and exercise much of the locale-related code that instead
313 * applies to MingW platforms that don't use the more modern UCRT
314 * library. This allows for finding many bugs in that portion of the
315 * implementation, without having to access such a platform.
318 /* If the environment says to, we can output debugging information during
319 * initialization. This is done before option parsing, and before any thread
320 * creation, so can be a file-level static. (Must come before #including
324 /* Returns the Unix errno portion; ignoring any others. This is a macro here
325 * instead of putting it into perl.h, because unclear to khw what should be
327 #define GET_ERRNO saved_errno
330 static int debug_initialization = 0;
331 # define DEBUG_INITIALIZATION_set(v) (debug_initialization = v)
332 # define DEBUG_LOCALE_INITIALIZATION_ debug_initialization
334 # ifdef HAS_EXTENDED_OS_ERRNO
335 /* Output the non-zero errno and/or the non-zero extended errno */
336 # define DEBUG_ERRNO \
338 int extended = get_extended_os_errno(); \
339 const char * errno_string; \
340 if (GET_ERRNO == 0) { /* Skip output if both errno types are 0 */ \
341 if (LIKELY(extended == 0)) errno_string = ""; \
342 else errno_string = Perl_form(aTHX_ "; $^E=%d", extended); \
344 else if (LIKELY(extended == GET_ERRNO)) \
345 errno_string = Perl_form(aTHX_ "; $!=%d", GET_ERRNO); \
346 else errno_string = Perl_form(aTHX_ "; $!=%d, $^E=%d", \
347 GET_ERRNO, extended);
349 /* Output the errno, if non-zero */
350 # define DEBUG_ERRNO \
352 const char * errno_string = ""; \
353 if (GET_ERRNO != 0) { \
355 errno_string = Perl_form(aTHX_ "; $!=%d", GET_ERRNO); \
359 /* Automatically include the caller's file, and line number in debugging output;
360 * and the errno (and/or extended errno) if non-zero. On threaded perls add
362 # if defined(USE_ITHREADS) && ! defined(NO_LOCALE_THREADS)
363 # define DEBUG_PRE_STMTS \
365 PerlIO_printf(Perl_debug_log, "\n%s: %" LINE_Tf ": 0x%p%s: ", \
366 __FILE__, (line_t)__LINE__, aTHX_ \
369 # define DEBUG_PRE_STMTS \
371 PerlIO_printf(Perl_debug_log, "\n%s: %" LINE_Tf "%s: ", \
372 __FILE__, (line_t)__LINE__, \
375 # define DEBUG_POST_STMTS RESTORE_ERRNO;
377 # define debug_initialization 0
378 # define DEBUG_INITIALIZATION_set(v)
379 # define DEBUG_PRE_STMTS
380 # define DEBUG_POST_STMTS
384 #define PERL_IN_LOCALE_C
387 /* Some platforms require LC_CTYPE to be congruent with the category we are
388 * looking for. XXX This still presumes that we have to match COLLATE and
389 * CTYPE even on platforms that apparently handle this. */
390 #if defined(USE_LOCALE_CTYPE) && ! defined(LIBC_HANDLES_MISMATCHED_CTYPE)
391 # define WE_MUST_DEAL_WITH_MISMATCHED_CTYPE
394 #if PERL_VERSION_GT(5,39,9)
395 # error Revert the commit that added this line
398 #ifdef WIN32_USE_FAKE_OLD_MINGW_LOCALES
400 /* Use -Accflags=-DWIN32_USE_FAKE_OLD_MINGW_LOCALES on a POSIX or *nix box
401 * to get a semblance of pretending the locale handling is that of a MingW
402 * that doesn't use UCRT (hence 'OLD' in the name). This exercizes code
403 * paths that are not compiled on non-Windows boxes, and allows for ASAN
404 * and PERL_MEMLOG. This is thus a way to see if locale.c on Windows is
405 * likely going to compile, without having to use a real Win32 box. And
406 * running the test suite will verify to a large extent our logic and memory
407 * allocation handling for such boxes. Of course the underlying calls are
408 * to the POSIX libc, so any differences in implementation between those and
409 * the Windows versions will not be caught by this. */
412 # undef P_CS_PRECEDES
413 # undef CURRENCY_SYMBOL
415 # undef _configthreadlocale
416 # define _configthreadlocale(arg) NOOP
418 # define MultiByteToWideChar(cp, flags, byte_string, m1, wstring, req_size) \
419 (PERL_UNUSED_ARG(cp), \
420 mbsrtowcs(wstring, &(byte_string), req_size, NULL) + 1)
421 # define WideCharToMultiByte(cp, flags, wstring, m1, byte_string, \
422 req_size, default_char, found_default_char) \
423 (PERL_UNUSED_ARG(cp), \
424 wcsrtombs(byte_string, &(wstring), req_size, NULL) + 1)
428 static const wchar_t * wsetlocale_buf = NULL;
429 static Size_t wsetlocale_buf_size = 0;
433 static PerlInterpreter * wsetlocale_buf_aTHX = NULL;
439 S_wsetlocale(const int category, const wchar_t * wlocale)
441 /* Windows uses a setlocale that takes a wchar_t* locale. Other boxes
442 * don't have this, so this Windows replacement converts the wchar_t input
443 * to plain 'char*', calls plain setlocale(), and converts the result back
446 const char * byte_locale = NULL;
448 byte_locale = Win_wstring_to_byte_string(CP_UTF8, wlocale);
451 const char * byte_result = setlocale(category, byte_locale);
452 Safefree(byte_locale);
453 if (byte_result == NULL) {
457 const wchar_t * wresult = Win_byte_string_to_wstring(CP_UTF8, byte_result);
463 /* Emulate a global static memory return from wsetlocale(). This currently
464 * leaks at process end; would require changing LOCALE_TERM to fix that */
465 Size_t string_size = wcslen(wresult) + 1;
467 if (wsetlocale_buf_size == 0) {
468 Newx(wsetlocale_buf, string_size, wchar_t);
469 wsetlocale_buf_size = string_size;
474 wsetlocale_buf_aTHX = aTHX;
479 else if (string_size > wsetlocale_buf_size) {
480 Renew(wsetlocale_buf, string_size, wchar_t);
481 wsetlocale_buf_size = string_size;
484 Copy(wresult, wsetlocale_buf, string_size, wchar_t);
487 return wsetlocale_buf;
490 # define _wsetlocale(category, wlocale) S_wsetlocale(category, wlocale)
492 #endif /* WIN32_USE_FAKE_OLD_MINGW_LOCALES */
494 /* 'for' loop headers to hide the necessary casts */
495 #define for_all_individual_category_indexes(i) \
496 for (locale_category_index i = (locale_category_index) 0; \
498 i = (locale_category_index) ((int) i + 1))
500 #define for_all_but_0th_individual_category_indexes(i) \
501 for (locale_category_index i = (locale_category_index) 1; \
503 i = (locale_category_index) ((int) i + 1))
505 #define for_all_category_indexes(i) \
506 for (locale_category_index i = (locale_category_index) 0; \
507 i <= LC_ALL_INDEX_; \
508 i = (locale_category_index) ((int) i + 1))
511 # if defined(USE_FAKE_LC_ALL_POSITIONAL_NOTATION) && defined(LC_ALL)
513 /* This simulates an underlying positional notation for LC_ALL when compiled on
514 * a system that uses name=value notation. Use this to develop on Linux and
515 * make a quick check that things have some chance of working on a positional
516 * box. Enable by adding to the Congfigure parameters:
517 * -Accflags=USE_FAKE_LC_ALL_POSITIONAL_NOTATION
519 * NOTE it redefines setlocale() and usequerylocale()
523 S_positional_name_value_xlation(const char * locale, bool direction)
524 { /* direction == 1 is from name=value to positional
525 direction == 0 is from positional to name=value */
529 const char * individ_locales[LC_ALL_INDEX_] = { NULL };
531 /* This parses either notation */
532 switch (parse_LC_ALL_string(locale,
533 (const char **) &individ_locales,
534 no_override, /* Handled by other code */
535 false, /* Return only [0] if suffices */
536 false, /* Don't panic on error */
539 default: /* Some compilers don't realize that below is the complete
540 list of the available enum values */
547 SAVEFREEPV(individ_locales[0]);
548 return individ_locales[0];
551 calc_LC_ALL_format format = (direction)
552 ? EXTERNAL_FORMAT_FOR_SET
554 const char * retval = calculate_LC_ALL_string(individ_locales,
559 for_all_individual_category_indexes(i) {
560 Safefree(individ_locales[i]);
569 S_positional_setlocale(int cat, const char * locale)
571 if (cat != LC_ALL) return setlocale(cat, locale);
573 if (locale && strNE(locale, "")) {
574 locale = S_positional_name_value_xlation(locale, 0);
575 if (! locale) return NULL;
578 locale = setlocale(cat, locale);
579 if (locale == NULL) return NULL;
580 return S_positional_name_value_xlation(locale, 1);
584 # define setlocale(a,b) S_positional_setlocale(a,b)
585 # ifdef USE_POSIX_2008_LOCALE
588 S_positional_newlocale(int mask, const char * locale, locale_t base)
592 if (mask != LC_ALL_MASK) return newlocale(mask, locale, base);
594 if (strNE(locale, "")) locale = S_positional_name_value_xlation(locale, 0);
595 if (locale == NULL) return NULL;
596 return newlocale(LC_ALL_MASK, locale, base);
600 # define newlocale(a,b,c) S_positional_newlocale(a,b,c)
603 #endif /* End of fake positional notation */
614 /* The main errno that gets used is this one, on platforms that support it */
616 # define SET_EINVAL SETERRNO(EINVAL, LIB_INVARG)
621 /* This is a starting guess as to when this is true. It definititely isn't
622 * true on *BSD where positional LC_ALL notation is used. Likely this will end
623 * up being defined in hints files. */
624 #ifdef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
625 # define NEWLOCALE_HANDLES_DISPARATE_LC_ALL
628 /* But regardless, we have to look at individual categories if some are
630 #ifdef HAS_IGNORED_LOCALE_CATEGORIES_
631 # undef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
635 /* Not all categories need be set to the same locale. This macro determines if
636 * 'name' which represents LC_ALL is uniform or disparate. There are two
637 * situations: 1) the platform uses unordered name=value pairs; 2) the platform
638 * uses ordered positional values, with a separator string between them */
639 # ifdef PERL_LC_ALL_SEPARATOR /* positional */
640 # define is_disparate_LC_ALL(name) cBOOL(instr(name, PERL_LC_ALL_SEPARATOR))
641 # else /* name=value */
643 /* In the, hopefully never occurring, event that the platform doesn't use
644 * either mechanism for disparate LC_ALL's, assume the name=value pairs
645 * form, rather than taking the extreme step of refusing to compile. Many
646 * programs won't have disparate locales, so will generally work */
647 # define PERL_LC_ALL_SEPARATOR ";"
648 # define is_disparate_LC_ALL(name) cBOOL( strchr(name, ';') \
649 && strchr(name, '='))
652 /* It is possible to compile perl to always keep any individual category in the
653 * C locale. This would be done where the implementation on a platform is
654 * flawed or incomplete. At the time of this writing, for example, OpenBSD has
655 * not implemented LC_COLLATE beyond the C locale. The 'category_available[]'
656 * table is a bool that says whether a category is changeable, or must be kept
657 * in C. This macro substitutes C for the locale appropriately, expanding to
658 * nothing on the more typical case where all possible categories present on
659 * the platform are handled. */
660 # ifdef HAS_IGNORED_LOCALE_CATEGORIES_
661 # define need_to_override_category(i) (! category_available[i])
662 # define override_ignored_category(i, new_locale) \
663 ((need_to_override_category(i)) ? "C" : (new_locale))
665 # define need_to_override_category(i) 0
666 # define override_ignored_category(i, new_locale) (new_locale)
669 PERL_STATIC_INLINE const char *
670 S_mortalized_pv_copy(pTHX_ const char * const pv)
672 PERL_ARGS_ASSERT_MORTALIZED_PV_COPY;
674 /* Copies the input pv, and arranges for it to be freed at an unspecified
681 const char * copy = savepv(pv);
688 /* Default values come from the C locale */
689 #define C_codeset "ANSI_X3.4-1968" /* Only in some Configurations, and usually
690 a single instance, so is a #define */
691 static const char C_decimal_point[] = ".";
693 #if (defined(USE_LOCALE_NUMERIC) && ! defined(TS_W32_BROKEN_LOCALECONV)) \
694 || ! ( defined(USE_LOCALE_NUMERIC) \
695 && (defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV)))
696 static const char C_thousands_sep[] = "";
699 /* Is the C string input 'name' "C" or "POSIX"? If so, and 'name' is the
700 * return of setlocale(), then this is extremely likely to be the C or POSIX
701 * locale. However, the output of setlocale() is documented to be opaque, but
702 * the odds are extremely small that it would return these two strings for some
703 * other locale. Note that VMS includes many non-ASCII characters in these two
704 * locales as controls and punctuation (below are hex bytes):
706 * punct: A1-A3 A5 A7-AB B0-B3 B5-B7 B9-BD BF-CF D1-DD DF-EF F1-FD
707 * Oddly, none there are listed as alphas, though some represent alphabetics
708 * http://www.nntp.perl.org/group/perl.perl5.porters/2013/02/msg198753.html */
709 #define isNAME_C_OR_POSIX(name) \
711 && (( *(name) == 'C' && (*(name + 1)) == '\0') \
712 || strEQ((name), "POSIX")))
714 #if defined(HAS_NL_LANGINFO_L) || defined(HAS_NL_LANGINFO)
715 # define HAS_SOME_LANGINFO
718 #define my_langinfo_c(item, category, locale, retbufp, retbuf_sizep, utf8ness) \
719 my_langinfo_i(item, category##_INDEX_, locale, retbufp, \
720 retbuf_sizep, utf8ness)
723 # define setlocale_debug_string_i(index, locale, result) \
724 my_setlocale_debug_string_i(index, locale, result, __LINE__)
725 # define setlocale_debug_string_c(category, locale, result) \
726 setlocale_debug_string_i(category##_INDEX_, locale, result)
727 # define setlocale_debug_string_r(category, locale, result) \
728 setlocale_debug_string_i(get_category_index(category), \
732 # define toggle_locale_i(index, locale) \
733 S_toggle_locale_i(aTHX_ index, locale, __LINE__)
734 # define toggle_locale_c(cat, locale) toggle_locale_i(cat##_INDEX_, locale)
735 # define restore_toggled_locale_i(index, locale) \
736 S_restore_toggled_locale_i(aTHX_ index, locale, __LINE__)
737 # define restore_toggled_locale_c(cat, locale) \
738 restore_toggled_locale_i(cat##_INDEX_, locale)
740 /* On systems without LC_ALL, pretending it exists anyway simplifies things.
741 * Choose a value for it that is very unlikely to clash with any actual
743 # define FAKE_LC_ALL PERL_INT_MIN
745 /* Below are parallel arrays for locale information indexed by our mapping of
746 * category numbers into small non-negative indexes. locale_table.h contains
747 * an entry like this for each individual category used on this system:
748 * PERL_LOCALE_TABLE_ENTRY(CTYPE, S_new_ctype)
750 * Each array redefines PERL_LOCALE_TABLE_ENTRY to generate the information
751 * needed for that array, and #includes locale_table.h to get the valid
754 * An entry for the conglomerate category LC_ALL is added here, immediately
755 * following the individual categories. (The treatment for it varies, so can't
756 * be in locale_table.h.)
758 * Following this, each array ends with an entry for illegal categories. All
759 * category numbers unknown to perl get mapped to this entry. This is likely
760 * to be a parameter error from the calling program; but it could be that this
761 * platform has a category we don't know about, in which case it needs to be
762 * added, using the paradigm of one of the existing categories. */
764 /* The first array is the locale categories perl uses on this system, used to
765 * map our index back to the system's category number. */
766 STATIC const int categories[] = {
768 # undef PERL_LOCALE_TABLE_ENTRY
769 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) LC_ ## name,
770 # include "locale_table.h"
778 (FAKE_LC_ALL + 1) /* Entry for unknown category; this number is unlikely
779 to clash with a real category */
782 # define GET_NAME_AS_STRING(token) # token
783 # define GET_LC_NAME_AS_STRING(token) GET_NAME_AS_STRING(LC_ ## token)
785 /* The second array is the category names. */
786 STATIC const char * const category_names[] = {
788 # undef PERL_LOCALE_TABLE_ENTRY
789 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) GET_LC_NAME_AS_STRING(name),
790 # include "locale_table.h"
793 # define LC_ALL_STRING "LC_ALL"
795 # define LC_ALL_STRING "If you see this, it is a bug in perl;" \
796 " please report it via perlbug"
801 # define LC_UNKNOWN_STRING "Locale category unknown to Perl; if you see" \
802 " this, it is a bug in perl; please report it" \
807 STATIC const Size_t category_name_lengths[] = {
809 # undef PERL_LOCALE_TABLE_ENTRY
810 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) \
811 STRLENs(GET_LC_NAME_AS_STRING(name)),
812 # include "locale_table.h"
814 STRLENs(LC_ALL_STRING),
815 STRLENs(LC_UNKNOWN_STRING)
818 /* Each entry includes space for the '=' and ';' */
819 # undef PERL_LOCALE_TABLE_ENTRY
820 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) \
821 + STRLENs(GET_LC_NAME_AS_STRING(name)) + 2
823 STATIC const Size_t lc_all_boiler_plate_length = 1 /* space for trailing NUL */
824 # include "locale_table.h"
827 /* A few categories require additional setup when they are changed. This table
828 * points to the functions that do that setup */
829 STATIC void (*update_functions[]) (pTHX_ const char *, bool force) = {
831 # undef PERL_LOCALE_TABLE_ENTRY
832 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) call_back,
833 # include "locale_table.h"
836 NULL, /* No update for unknown category */
839 # if defined(HAS_IGNORED_LOCALE_CATEGORIES_)
841 /* Indicates if each category on this platform is available to use not in
843 STATIC const bool category_available[] = {
845 # undef PERL_LOCALE_TABLE_ENTRY
846 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) LC_ ## name ## _AVAIL_,
847 # include "locale_table.h"
855 false /* LC_UNKNOWN_AVAIL_ */
859 # if defined(USE_POSIX_2008_LOCALE)
861 STATIC const int category_masks[] = {
863 # undef PERL_LOCALE_TABLE_ENTRY
864 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) LC_ ## name ## _MASK,
865 # include "locale_table.h"
867 LC_ALL_MASK, /* Will rightly refuse to compile unless this is defined */
868 0 /* Empty mask for unknown category */
872 # if ! defined(PERL_LC_ALL_USES_NAME_VALUE_PAIRS)
874 /* On platforms that use positional notation for expressing LC_ALL, this maps
875 * the position of each category to our corresponding internal index for it.
876 * This is initialized at run time if needed. LC_ALL_INDEX_ is not legal for
877 * an individual locale, hence marks the elements here as not actually
881 map_LC_ALL_position_to_index[LC_ALL_INDEX_] = { LC_ALL_INDEX_ };
885 #if defined(USE_LOCALE) || defined(DEBUGGING)
888 S_get_displayable_string(pTHX_
889 const char * const s,
890 const char * const e,
893 PERL_ARGS_ASSERT_GET_DISPLAYABLE_STRING;
900 bool prev_was_printable = TRUE;
901 bool first_time = TRUE;
904 /* Worst case scenario: All are non-printable so have a blank between each.
905 * If UTF-8, all are the largest possible code point; otherwise all are a
906 * single byte. '(2 + 1)' is from each byte takes 2 characters to
907 * display, and a blank (or NUL for the final one) after it */
908 const Size_t size = (e - s) * (2 + 1) * ((is_utf8) ? UVSIZE : 1);
909 Newxz(ret, size, char);
914 ? utf8_to_uvchr_buf((U8 *) t, e, NULL)
917 if (! prev_was_printable) {
918 my_strlcat(ret, " ", size);
921 /* Escape these to avoid any ambiguity */
922 if (cp == ' ' || cp == '\\') {
923 my_strlcat(ret, "\\", size);
925 my_strlcat(ret, Perl_form(aTHX_ "%c", (U8) cp), size);
926 prev_was_printable = TRUE;
930 my_strlcat(ret, " ", size);
932 my_strlcat(ret, Perl_form(aTHX_ "%02" UVXf, cp), size);
933 prev_was_printable = FALSE;
935 t += (is_utf8) ? UTF8SKIP(t) : 1;
945 # define get_category_index(cat) get_category_index_helper(cat, NULL, __LINE__)
947 STATIC locale_category_index
948 S_get_category_index_helper(pTHX_ const int category, bool * succeeded,
949 const line_t caller_line)
951 PERL_ARGS_ASSERT_GET_CATEGORY_INDEX_HELPER;
953 /* Given a category, return the equivalent internal index we generally use
954 * instead, warn or panic if not found. */
956 locale_category_index i;
958 # undef PERL_LOCALE_TABLE_ENTRY
959 # define PERL_LOCALE_TABLE_ENTRY(name, call_back) \
960 case LC_ ## name: i = LC_ ## name ## _INDEX_; break;
964 # include "locale_table.h"
966 case LC_ALL: i = LC_ALL_INDEX_; break;
969 default: goto unknown_locale;
972 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
973 "index of category %d (%s) is %d;"
974 " called from %" LINE_Tf "\n",
975 category, category_names[i], i, caller_line));
987 return LC_ALL_INDEX_; /* Arbitrary */
990 locale_panic_via_(Perl_form(aTHX_ "Unknown locale category %d", category),
991 __FILE__, caller_line);
992 NOT_REACHED; /* NOTREACHED */
995 #endif /* ifdef USE_LOCALE */
998 Perl_force_locale_unlock(pTHX)
1000 /* Remove any locale mutex, in preperation for an inglorious termination,
1001 * typically a panic */
1003 #if defined(USE_LOCALE_THREADS)
1005 /* If recursively locked, clear all at once */
1006 if (PL_locale_mutex_depth > 1) {
1007 PL_locale_mutex_depth = 1;
1010 if (PL_locale_mutex_depth > 0) {
1018 #ifdef USE_POSIX_2008_LOCALE
1021 S_use_curlocale_scratch(pTHX)
1023 /* This function is used to hide from the caller the case where the current
1024 * locale_t object in POSIX 2008 is the global one, which is illegal in
1025 * many of the P2008 API calls. This checks for that and, if necessary
1026 * creates a proper P2008 object. Any prior object is deleted, as is any
1027 * remaining object during global destruction. */
1029 locale_t cur = uselocale((locale_t) 0);
1031 if (cur != LC_GLOBAL_LOCALE) {
1035 if (PL_scratch_locale_obj) {
1036 freelocale(PL_scratch_locale_obj);
1039 PL_scratch_locale_obj = duplocale(LC_GLOBAL_LOCALE);
1040 return PL_scratch_locale_obj;
1046 Perl_locale_panic(const char * msg,
1047 const line_t immediate_caller_line,
1048 const char * const higher_caller_file,
1049 const line_t higher_caller_line)
1051 PERL_ARGS_ASSERT_LOCALE_PANIC;
1055 force_locale_unlock();
1057 #ifdef USE_C_BACKTRACE
1058 dump_c_backtrace(Perl_debug_log, 20, 1);
1061 const char * called_by = "";
1062 if ( strNE(__FILE__, higher_caller_file)
1063 || immediate_caller_line != higher_caller_line)
1065 called_by = Perl_form(aTHX_ "\nCalled by %s: %" LINE_Tf "\n",
1066 higher_caller_file, higher_caller_line);
1071 const char * errno_text;
1073 #ifdef HAS_EXTENDED_OS_ERRNO
1075 const int extended_errnum = get_extended_os_errno();
1076 if (errno != extended_errnum) {
1077 errno_text = Perl_form(aTHX_ "; errno=%d, $^E=%d",
1078 errno, extended_errnum);
1085 errno_text = Perl_form(aTHX_ "; errno=%d", errno);
1088 /* diag_listed_as: panic: %s */
1089 Perl_croak(aTHX_ "%s: %" LINE_Tf ": panic: %s%s%s\n",
1090 __FILE__, immediate_caller_line,
1091 msg, errno_text, called_by);
1094 /* Macros to report and croak on an unexpected failure to set the locale. The
1095 * via version has more stack trace information */
1096 #define setlocale_failure_panic_i(i, cur, fail, line, higher_line) \
1097 setlocale_failure_panic_via_i(i, cur, fail, __LINE__, line, \
1098 __FILE__, higher_line)
1100 #define setlocale_failure_panic_c(cat, cur, fail, line, higher_line) \
1101 setlocale_failure_panic_i(cat##_INDEX_, cur, fail, line, higher_line)
1103 #if defined(USE_LOCALE)
1105 /* Expands to the code to
1106 * result = savepvn(s, len)
1107 * if the category whose internal index is 'i' doesn't need to be kept in the C
1108 * locale on this system, or if 'action is 'no_override'. Otherwise it expands
1110 * result = savepv("C")
1111 * unless 'action' isn't 'check_that_overridden', in which case if the string
1112 * 's' isn't already "C" it panics */
1113 # ifndef HAS_IGNORED_LOCALE_CATEGORIES_
1114 # define OVERRIDE_AND_SAVEPV(s, len, result, i, action) \
1115 result = savepvn(s, len)
1117 # define OVERRIDE_AND_SAVEPV(s, len, result, i, action) \
1119 if (LIKELY( ! need_to_override_category(i) \
1120 || action == no_override)) { \
1121 result = savepvn(s, len); \
1124 const char * temp = savepvn(s, len); \
1125 result = savepv(override_ignored_category(i, temp)); \
1126 if (action == check_that_overridden && strNE(result, temp)) { \
1127 locale_panic_(Perl_form(aTHX_ \
1128 "%s expected to be '%s', instead is '%s'", \
1129 category_names[i], result, temp)); \
1136 STATIC parse_LC_ALL_string_return
1137 S_parse_LC_ALL_string(pTHX_ const char * string,
1138 const char ** output,
1139 const parse_LC_ALL_STRING_action override,
1140 bool always_use_full_array,
1141 const bool panic_on_error,
1142 const line_t caller_line)
1144 /* This function parses the value of the input 'string' which is expected
1145 * to be the representation of an LC_ALL locale, and splits the result into
1146 * the values for the individual component categories, returning those in
1147 * the 'output' array. Each array value will be a savepv() copy that is
1148 * the responsibility of the caller to make sure gets freed
1150 * The locale for each category is independent of the other categories.
1151 * Often, they are all the same, but certainly not always. Perl, in fact,
1152 * usually keeps LC_NUMERIC in the C locale, regardless of the underlying
1153 * locale. LC_ALL has to be able to represent the case of when not all
1154 * categories have the same locale. Platforms have differing ways of
1155 * representing this. Internally, this file uses the 'name=value;'
1156 * representation found on some platforms, so this function always looks
1157 * for and parses that. Other platforms use a positional notation. On
1158 * those platforms, this function also parses that form. It examines the
1159 * input to see which form is being parsed.
1161 * Often, all categories will have the same locale. This is special cased
1162 * if 'always_use_full_array' is false on input:
1163 * 1) If the input 'string' is a single value, this function doesn't
1164 * store anything into 'output', and returns 'no_array'
1165 * 2) Some platforms will return multiple occurrences of the same
1166 * value rather than coalescing them down to a single one. HP-UX
1167 * is such a one. This function will do that collapsing for you,
1168 * returning 'only_element_0' and saving the single value in
1169 * output[0], which the caller will need to arrange to be freed.
1170 * The rest of output[] is undefined, and does not need to be
1173 * Otherwise, the input 'string' may not be valid. This function looks
1174 * mainly for syntactic errors, and if found, returns 'invalid'. 'output'
1175 * will not be filled in in that case, but the input state of it isn't
1176 * necessarily preserved. Turning on -DL debugging will give details as to
1177 * the error. If 'panic_on_error' is 'true', the function panics instead
1178 * of returning on error, with a message giving the details.
1180 * Otherwise, output[] will be filled with the individual locale names for
1181 * all categories on the system, 'full_array' will be returned, and the
1182 * caller needs to arrange for each to be freed. This means that either at
1183 * least one category differed from the others, or 'always_use_full_array' was
1186 * perl may be configured to ignore changes to a category's locale to
1187 * non-C. The parameter 'override' tells this function what to do when
1188 * encountering such an illegal combination:
1190 * no_override indicates to take no special action
1191 * override_if_ignored, indicates to return 'C' instead of what the
1192 * input string actually says.
1193 * check_that_overridden indicates to panic if the string says the
1194 * category is not 'C'. This is used when
1195 * non-C is very unexpected behavior.
1198 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1199 "Entering parse_LC_ALL_string; called from %" \
1200 LINE_Tf "\nnew='%s'\n", caller_line, string));
1202 # ifdef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
1204 const char separator[] = ";";
1205 const Size_t separator_len = 1;
1206 const bool single_component = (strchr(string, ';') == NULL);
1210 /* It's possible (but quite unlikely) that the separator string is an '='
1211 * or a ';'. Requiring both to be present for using the 'name=value;' form
1212 * properly handles those possibilities */
1213 const bool name_value = strchr(string, '=') && strchr(string, ';');
1214 const char * separator;
1215 Size_t separator_len;
1216 bool single_component;
1220 single_component = false; /* Since has both [;=], must be multi */
1223 separator = PERL_LC_ALL_SEPARATOR;
1224 separator_len = STRLENs(PERL_LC_ALL_SEPARATOR);
1225 single_component = instr(string, separator) == NULL;
1228 Size_t component_number = 0; /* Position in the parsing loop below */
1231 # ifndef HAS_IGNORED_LOCALE_CATEGORIES_
1232 PERL_UNUSED_ARG(override);
1235 /* Any ignored categories are to be set to "C", so if this single-component
1236 * LC_ALL isn't to C, it has both "C" and non-C, so isn't really a single
1237 * component. All the non-ignored categories are set to the input
1238 * component, but the ignored ones are overridden to be C.
1240 * This incidentally handles the case where the string is "". The return
1241 * will be C for each ignored category and "" for the others. Then the
1242 * caller can individually set each category, and get the right answer. */
1243 if (single_component && ! isNAME_C_OR_POSIX(string)) {
1244 for_all_individual_category_indexes(i) {
1245 OVERRIDE_AND_SAVEPV(string, strlen(string), output[i], i, override);
1253 if (single_component) {
1254 if (! always_use_full_array) {
1258 for_all_individual_category_indexes(i) {
1259 output[i] = savepv(string);
1265 /* Here the input is multiple components. Parse through them. (It is
1266 * possible that these components are all the same, so we check, and if so,
1267 * return just the 0th component (unless 'always_use_full_array' is true)
1269 * This enum notes the possible errors findable in parsing */
1274 contains_LC_ALL_element
1277 /* Keep track of the categories we have encountered so far */
1278 bool seen[LC_ALL_INDEX_] = { false };
1280 Size_t index; /* Our internal index for the current category */
1281 const char * s = string;
1282 const char * e = s + strlen(string);
1283 const char * category_end = NULL;
1284 const char * saved_first = NULL;
1286 /* Parse the input locale string */
1289 /* 'separator' has been set up to delimit the components */
1290 const char * next_sep = instr(s, separator);
1291 if (! next_sep) { /* At the end of the input */
1295 # ifndef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
1298 /* Get the index of the category in this position */
1299 index = map_LC_ALL_position_to_index[component_number++];
1305 { /* Get the category part when each component is the
1306 * 'category=locale' form */
1308 category_end = strchr(s, '=');
1310 /* The '=' terminates the category name. If no '=', is improper
1312 if (! category_end) {
1317 /* Find our internal index of the category name; uses a linear
1318 * search. (XXX This could be avoided by various means, but the
1319 * maximum likely search is 6 items, and khw doesn't think the
1320 * added complexity would save very much at all.) */
1321 const unsigned int name_len = (unsigned int) (category_end - s);
1322 for (index = 0; index < C_ARRAY_LENGTH(category_names); index++) {
1323 if ( name_len == category_name_lengths[index]
1324 && memEQ(s, category_names[index], name_len))
1326 goto found_category;
1330 /* Here, the category is not in our list. */
1331 error = unknown_category;
1334 found_category: /* The system knows about this category. */
1336 if (index == LC_ALL_INDEX_) {
1337 error = contains_LC_ALL_element;
1341 /* The locale name starts just beyond the '=' */
1342 s = category_end + 1;
1344 /* Linux (and maybe others) doesn't treat a duplicate category in
1345 * the string as an error. Instead it uses the final occurrence as
1346 * the intended value. So if this is a duplicate, free the former
1347 * value before setting the new one */
1349 Safefree(output[index]);
1356 /* Here, 'index' contains our internal index number for the current
1357 * category, and 's' points to the beginning of the locale name for
1359 OVERRIDE_AND_SAVEPV(s, next_sep - s, output[index], index, override);
1361 if (! always_use_full_array) {
1362 if (! saved_first) {
1363 saved_first = output[index];
1366 if (strNE(saved_first, output[index])) {
1367 always_use_full_array = true;
1372 /* Next time start from the new position */
1373 s = next_sep + separator_len;
1376 /* Finished looping through all the categories
1378 * Check if the input was incomplete. */
1380 # ifndef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
1382 if (! name_value) { /* Positional notation */
1383 if (component_number != LC_ALL_INDEX_) {
1392 { /* Here is the name=value notation */
1393 for_all_individual_category_indexes(i) {
1401 /* In the loop above, we changed 'always_use_full_array' to true iff not all
1402 * categories have the same locale. Hence, if it is still 'false', all of
1403 * them are the same. */
1404 if (always_use_full_array) {
1408 /* Free the dangling ones */
1409 for_all_but_0th_individual_category_indexes(i) {
1410 Safefree(output[i]);
1414 return only_element_0;
1418 /* Don't leave memory dangling that we allocated before the failure */
1419 for_all_individual_category_indexes(i) {
1421 Safefree(output[i]);
1427 const char * display_start = s;
1428 const char * display_end = e;
1432 msg = "doesn't list every locale category";
1433 display_start = string;
1436 msg = "needs an '=' to split name=value";
1438 case unknown_category:
1439 msg = "is an unknown category";
1440 display_end = (category_end && category_end > display_start)
1444 case contains_LC_ALL_element:
1445 msg = "has LC_ALL, which is illegal here";
1449 msg = Perl_form(aTHX_ "'%.*s' %s\n",
1450 (int) (display_end - display_start),
1451 display_start, msg);
1453 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s", msg));
1455 if (panic_on_error) {
1456 locale_panic_via_(msg, __FILE__, caller_line);
1462 # undef OVERRIDE_AND_SAVEPV
1465 /*==========================================================================
1466 * Here starts the code that gives a uniform interface to its callers, hiding
1467 * the differences between platforms.
1469 * base_posix_setlocale_() presents a consistent POSIX-compliant interface to
1470 * setlocale(). Windows requres a customized base-level setlocale(). This
1471 * layer should only be used by the next level up: the plain posix_setlocale
1472 * layer. Any necessary mutex locking needs to be done at a higher level. The
1473 * return may be overwritten by the next call to this function */
1475 # define base_posix_setlocale_(cat, locale) win32_setlocale(cat, locale)
1477 # define base_posix_setlocale_(cat, locale) \
1478 ((const char *) setlocale(cat, locale))
1481 /*==========================================================================
1482 * Here is the main posix layer. It is the same as the base one unless the
1483 * system is lacking LC_ALL, or there are categories that we ignore, but that
1484 * the system libc knows about */
1486 #if ! defined(USE_LOCALE) \
1487 || (defined(LC_ALL) && ! defined(HAS_IGNORED_LOCALE_CATEGORIES_))
1488 # define posix_setlocale(cat, locale) base_posix_setlocale_(cat, locale)
1490 # define posix_setlocale(cat, locale) \
1491 S_posix_setlocale_with_complications(aTHX_ cat, locale, __LINE__)
1494 S_posix_setlocale_with_complications(pTHX_ const int cat,
1495 const char * new_locale,
1496 const line_t caller_line)
1498 /* This implements the posix layer above the base posix layer.
1499 * It is needed to reconcile our internal records that reflect only a
1500 * proper subset of the categories known by the system. */
1502 /* Querying the current locale returns the real value */
1503 if (new_locale == NULL) {
1504 new_locale = base_posix_setlocale_(cat, NULL);
1509 const char * locale_on_entry = NULL;
1511 /* If setting from the environment, actually do the set to get the system's
1512 * idea of what that means; we may have to override later. */
1513 if (strEQ(new_locale, "")) {
1514 locale_on_entry = base_posix_setlocale_(cat, NULL);
1515 assert(locale_on_entry);
1516 new_locale = base_posix_setlocale_(cat, "");
1525 const char * new_locales[LC_ALL_INDEX_] = { NULL };
1527 if (cat == LC_ALL) {
1528 switch (parse_LC_ALL_string(new_locale,
1529 (const char **) &new_locales,
1530 override_if_ignored, /* Override any
1533 false, /* Return only [0] if suffices */
1534 false, /* Don't panic on error */
1544 case only_element_0:
1545 new_locale = new_locales[0];
1546 SAVEFREEPV(new_locale);
1551 /* Turn the array into a string that the libc setlocale() should
1552 * understand. (Another option would be to loop, setting the
1553 * individual locales, and then return base(cat, NULL) */
1554 new_locale = calculate_LC_ALL_string(new_locales,
1555 EXTERNAL_FORMAT_FOR_SET,
1559 for_all_individual_category_indexes(i) {
1560 Safefree(new_locales[i]);
1563 /* And call the libc setlocale. We could avoid this call if
1564 * locale_on_entry is set and eq the new_locale. But that would be
1565 * only for the relatively rare case of the desired locale being
1566 * "", and the time spent in doing the string compare might be more
1567 * than that of just setting it unconditionally */
1568 new_locale = base_posix_setlocale_(cat, new_locale);
1579 /* Here, 'new_locale' is a single value, not an aggregation. Just set it.
1582 base_posix_setlocale_(cat,
1583 override_ignored_category(
1584 get_category_index(cat), new_locale));
1593 /* 'locale_on_entry' being set indicates there has likely been a change in
1594 * locale which needs to be restored */
1595 if (locale_on_entry) {
1596 if (! base_posix_setlocale_(cat, locale_on_entry)) {
1597 setlocale_failure_panic_i(get_category_index(cat),
1598 NULL, locale_on_entry,
1599 __LINE__, caller_line);
1609 /* End of posix layer
1610 *==========================================================================
1612 * The next layer up is to catch vagaries and bugs in the libc setlocale return
1613 * value. The return is not guaranteed to be stable.
1615 * Any necessary mutex locking needs to be done at a higher level.
1617 * On most platforms this layer is empty, expanding to just the layer
1618 * below. To enable it, call Configure with either or both:
1619 * -Accflags=-DHAS_LF_IN_SETLOCALE_RETURN
1620 * to indicate that extraneous \n characters can be returned
1622 * -Accflags=-DHAS_BROKEN_SETLOCALE_QUERY_LC_ALL
1623 * to indicate that setlocale(LC_ALL, NULL) cannot be relied
1627 #define STDIZED_SETLOCALE_LOCK POSIX_SETLOCALE_LOCK
1628 #define STDIZED_SETLOCALE_UNLOCK POSIX_SETLOCALE_UNLOCK
1629 #if ! defined(USE_LOCALE) \
1630 || ! ( defined(HAS_LF_IN_SETLOCALE_RETURN) \
1631 || defined(HAS_BROKEN_SETLOCALE_QUERY_LC_ALL))
1632 # define stdized_setlocale(cat, locale) posix_setlocale(cat, locale)
1633 # define stdize_locale(cat, locale) (locale)
1635 # define stdized_setlocale(cat, locale) \
1636 S_stdize_locale(aTHX_ cat, posix_setlocale(cat, locale), __LINE__)
1639 S_stdize_locale(pTHX_ const int category,
1640 const char *input_locale,
1641 const line_t caller_line)
1643 /* The return value of setlocale() is opaque, but is required to be usable
1644 * as input to a future setlocale() to create the same state.
1645 * Unfortunately not all systems are compliant. This function brings those
1646 * outliers into conformance. It is based on what problems have arisen in
1649 * This has similar constraints as the posix layer. You need to lock
1650 * around it until its return is safely copied or no longer needed. (The
1651 * return may point to a global static buffer or may be mortalized.)
1653 * The current things this corrects are:
1654 * 1) A new-line. This function chops any \n characters
1655 * 2) A broken 'setlocale(LC_ALL, foo)' This constructs a proper returned
1656 * string from the constituent categories
1658 * If no changes were made, the input is returned as-is */
1660 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
1661 "Entering stdize_locale(%d, '%s');"
1662 " called from %" LINE_Tf "\n",
1663 category, input_locale, caller_line));
1665 if (input_locale == NULL) {
1670 char * retval = (char *) input_locale;
1672 # if defined(LC_ALL) && defined(HAS_BROKEN_SETLOCALE_QUERY_LC_ALL)
1674 /* If setlocale(LC_ALL, NULL) is broken, compute what the system
1675 * actually thinks it should be from its individual components */
1676 if (category == LC_ALL) {
1677 retval = (char *) calculate_LC_ALL_string(
1678 NULL, /* query each individ locale */
1679 EXTERNAL_FORMAT_FOR_SET,
1685 # ifdef HAS_NL_IN_SETLOCALE_RETURN
1687 char * first_bad = NULL;
1691 PERL_UNUSED_ARG(category);
1692 PERL_UNUSED_ARG(caller_line);
1694 # define INPUT_LOCALE retval
1695 # define MARK_CHANGED
1698 char * individ_locales[LC_ALL_INDEX_] = { NULL };
1699 bool made_changes = false;
1701 if (category != LC_ALL) {
1702 individ_locales[0] = retval;
1707 /* And parse the locale string, splitting into its individual
1709 switch (parse_LC_ALL_string(retval,
1710 (const char **) &individ_locales,
1711 check_that_overridden, /* ignored
1715 false, /* Return only [0] if suffices */
1716 false, /* Don't panic on error */
1723 case full_array: /* Loop below through all the component categories.
1725 upper = LC_ALL_INDEX_ - 1;
1729 /* All categories here are set to the same locale, and the parse
1730 * didn't fill in any of 'individ_locales'. Set the 0th element to
1732 individ_locales[0] = retval;
1735 case only_element_0: /* Element 0 is the only element we need to look
1742 for (unsigned int i = 0; i <= upper; i++)
1744 # define INPUT_LOCALE individ_locales[i]
1745 # define MARK_CHANGED made_changes = true;
1746 # endif /* Has LC_ALL */
1749 first_bad = (char *) strchr(INPUT_LOCALE, '\n');
1751 /* Most likely, there isn't a problem with the input */
1752 if (UNLIKELY(first_bad)) {
1754 /* This element will need to be adjusted. Create a modifiable
1757 retval = savepv(INPUT_LOCALE);
1760 /* Translate the found position into terms of the copy */
1761 first_bad = retval + (first_bad - INPUT_LOCALE);
1763 /* Get rid of the \n and what follows. (Originally, only a
1764 * trailing \n was stripped. Unsure what to do if not trailing) */
1765 *((char *) first_bad) = '\0';
1766 } /* End of needs adjusting */
1767 } /* End of looking for problems */
1771 /* If we had multiple elements, extra work is required */
1774 /* If no changes were made to the input, 'retval' already contains it
1778 /* But if did make changes, need to calculate the new value */
1779 retval = (char *) calculate_LC_ALL_string(
1780 (const char **) &individ_locales,
1781 EXTERNAL_FORMAT_FOR_SET,
1786 /* And free the no-longer needed memory */
1787 for (unsigned int i = 0; i <= upper; i++) {
1788 Safefree(individ_locales[i]);
1793 # undef INPUT_LOCALE
1794 # undef MARK_CHANGED
1795 # endif /* HAS_NL_IN_SETLOCALE_RETURN */
1797 return (const char *) retval;
1800 #endif /* USE_LOCALE */
1802 /* End of stdize_locale layer
1804 * ==========================================================================
1806 * The next many lines form several implementations of a layer above the
1807 * close-to-the-metal 'posix' and 'stdized' macros. They are used to present a
1808 * uniform API to the rest of the code in this file in spite of the disparate
1809 * underlying implementations. Which implementation gets compiled depends on
1810 * the platform capabilities (and some user choice) as determined by Configure.
1812 * As more fully described in the introductory comments in this file, the
1813 * API of each implementation consists of three sets of macros. Each set has
1814 * three variants with suffixes '_c', '_r', and '_i'. In the list below '_X'
1815 * is to be replaced by any of these suffixes.
1817 * 1) bool_setlocale_X attempts to set the given category's locale to the
1818 * given value, returning if it worked or not.
1819 * 2) void_setlocale_X is like the corresponding bool_setlocale, but used when
1820 * success is the only sane outcome, so failure causes it
1822 * 3) querylocale_X to see what the given category's locale is
1824 * 4) setlocale_i() is defined only in those implementations where the bool
1825 * and query forms are essentially the same, and can be
1826 * combined to save CPU time.
1828 * Each implementation below is separated by ==== lines, and includes bool,
1829 * void, and query macros. The query macros are first, followed by any
1830 * functions needed to implement them. Then come the bool, again followed by
1831 * any implementing functions Then are the void macros; next is setlocale_i if
1832 * present on this implementation. Finally are any helper functions. The sets
1833 * in each implementation are separated by ---- lines.
1835 * The returned strings from all the querylocale...() forms in all
1836 * implementations are thread-safe, and the caller should not free them,
1837 * but each may be a mortalized copy. If you need something stable across
1838 * calls, you need to savepv() the result yourself.
1840 *===========================================================================*/
1842 #if (! defined(USE_LOCALE_THREADS) && ! defined(USE_POSIX_2008_LOCALE)) \
1843 || ( defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE))
1845 /* For non-threaded perls, the implementation just expands to the base-level
1846 * functions (except if we are Configured to nonetheless use the POSIX 2008
1847 * interface) This implementation is also used on threaded perls where
1848 * threading is invisible to us. Currently this is only on later Windows
1851 # define querylocale_r(cat) mortalized_pv_copy(stdized_setlocale(cat, NULL))
1852 # define querylocale_c(cat) querylocale_r(cat)
1853 # define querylocale_i(i) querylocale_c(categories[i])
1855 /*---------------------------------------------------------------------------*/
1857 # define bool_setlocale_r(cat, locale) cBOOL(posix_setlocale(cat, locale))
1858 # define bool_setlocale_i(i, locale) \
1859 bool_setlocale_c(categories[i], locale)
1860 # define bool_setlocale_c(cat, locale) bool_setlocale_r(cat, locale)
1862 /*---------------------------------------------------------------------------*/
1864 # define void_setlocale_r_with_caller(cat, locale, file, line) \
1866 if (! bool_setlocale_r(cat, locale)) \
1867 setlocale_failure_panic_via_i(get_category_index(cat), \
1868 NULL, locale, __LINE__, 0, \
1872 # define void_setlocale_c_with_caller(cat, locale, file, line) \
1873 void_setlocale_r_with_caller(cat, locale, file, line)
1875 # define void_setlocale_i_with_caller(i, locale, file, line) \
1876 void_setlocale_r_with_caller(categories[i], locale, file, line)
1878 # define void_setlocale_r(cat, locale) \
1879 void_setlocale_r_with_caller(cat, locale, __FILE__, __LINE__)
1880 # define void_setlocale_c(cat, locale) void_setlocale_r(cat, locale)
1881 # define void_setlocale_i(i, locale) void_setlocale_r(categories[i], locale)
1883 /*---------------------------------------------------------------------------*/
1885 /* setlocale_i is only defined for Configurations where the libc setlocale()
1886 * doesn't need any tweaking. It allows for some shortcuts */
1887 # ifndef USE_LOCALE_THREADS
1888 # define setlocale_i(i, locale) stdized_setlocale(categories[i], locale)
1890 # elif defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE)
1892 /* On Windows, we don't know at compile time if we are in thread-safe mode or
1893 * not. If we are, we can just return the result of the layer below us. If we
1894 * are in unsafe mode, we need to first copy that result to a safe place while
1895 * in a critical section */
1897 # define setlocale_i(i, locale) S_setlocale_i(aTHX_ categories[i], locale)
1900 S_setlocale_i(pTHX_ const int category, const char * locale)
1902 if (LIKELY(_configthreadlocale(0) == _ENABLE_PER_THREAD_LOCALE)) {
1903 return stdized_setlocale(category, locale);
1907 const char * retval = save_to_buffer(stdized_setlocale(category, locale),
1909 &PL_setlocale_bufsize);
1917 /*===========================================================================*/
1918 #elif defined(USE_LOCALE_THREADS) \
1919 && ! defined(USE_THREAD_SAFE_LOCALE)
1921 /* Here, there are threads, and there is no support for thread-safe
1922 * operation. This is a dangerous situation, which perl is documented as
1923 * not supporting, but it arises in practice. We can do a modicum of
1924 * automatic mitigation by making sure there is a per-thread return from
1925 * setlocale(), and that a mutex protects it from races */
1927 # define querylocale_r(cat) \
1928 mortalized_pv_copy(less_dicey_setlocale_r(cat, NULL))
1929 # define querylocale_c(cat) querylocale_r(cat)
1930 # define querylocale_i(i) querylocale_r(categories[i])
1933 S_less_dicey_setlocale_r(pTHX_ const int category, const char * locale)
1935 const char * retval;
1937 PERL_ARGS_ASSERT_LESS_DICEY_SETLOCALE_R;
1939 STDIZED_SETLOCALE_LOCK;
1941 retval = save_to_buffer(stdized_setlocale(category, locale),
1942 &PL_less_dicey_locale_buf,
1943 &PL_less_dicey_locale_bufsize);
1945 STDIZED_SETLOCALE_UNLOCK;
1950 /*---------------------------------------------------------------------------*/
1952 # define bool_setlocale_r(cat, locale) \
1953 less_dicey_bool_setlocale_r(cat, locale)
1954 # define bool_setlocale_i(i, locale) \
1955 bool_setlocale_r(categories[i], locale)
1956 # define bool_setlocale_c(cat, locale) bool_setlocale_r(cat, locale)
1959 S_less_dicey_bool_setlocale_r(pTHX_ const int cat, const char * locale)
1963 PERL_ARGS_ASSERT_LESS_DICEY_BOOL_SETLOCALE_R;
1965 /* Unlikely, but potentially possible that another thread could zap the
1966 * buffer from true to false or vice-versa, so need to lock here */
1967 POSIX_SETLOCALE_LOCK;
1968 retval = cBOOL(posix_setlocale(cat, locale));
1969 POSIX_SETLOCALE_UNLOCK;
1974 /*---------------------------------------------------------------------------*/
1976 # define void_setlocale_r_with_caller(cat, locale, file, line) \
1978 if (! bool_setlocale_r(cat, locale)) \
1979 setlocale_failure_panic_via_i(get_category_index(cat), \
1980 NULL, locale, __LINE__, 0, \
1984 # define void_setlocale_c_with_caller(cat, locale, file, line) \
1985 void_setlocale_r_with_caller(cat, locale, file, line)
1987 # define void_setlocale_i_with_caller(i, locale, file, line) \
1988 void_setlocale_r_with_caller(categories[i], locale, file, line)
1990 # define void_setlocale_r(cat, locale) \
1991 void_setlocale_r_with_caller(cat, locale, __FILE__, __LINE__)
1992 # define void_setlocale_c(cat, locale) void_setlocale_r(cat, locale)
1993 # define void_setlocale_i(i, locale) void_setlocale_r(categories[i], locale)
1995 /*===========================================================================*/
1997 #elif defined(USE_POSIX_2008_LOCALE)
1999 # error This code assumes that LC_ALL is available on a system modern enough to have POSIX 2008
2002 /* Here, there is a completely different API to get thread-safe locales. We
2003 * emulate the setlocale() API with our own function(s). setlocale categories,
2004 * like LC_NUMERIC, are not valid here for the POSIX 2008 API. Instead, there
2005 * are equivalents, like LC_NUMERIC_MASK, which we use instead, which we find
2006 * by table lookup. */
2008 # if defined(__GLIBC__) && defined(USE_LOCALE_MESSAGES)
2009 /* https://sourceware.org/bugzilla/show_bug.cgi?id=24936 */
2010 # define HAS_GLIBC_LC_MESSAGES_BUG
2011 # include <libintl.h>
2014 # define querylocale_i(i) querylocale_2008_i(i, __LINE__)
2015 # define querylocale_c(cat) querylocale_i(cat##_INDEX_)
2016 # define querylocale_r(cat) querylocale_i(get_category_index(cat))
2019 S_querylocale_2008_i(pTHX_ const locale_category_index index,
2020 const line_t caller_line)
2022 PERL_ARGS_ASSERT_QUERYLOCALE_2008_I;
2023 assert(index <= LC_ALL_INDEX_);
2025 /* This function returns the name of the locale category given by the input
2026 * 'index' into our parallel tables of them.
2028 * POSIX 2008, for some sick reason, chose not to provide a method to find
2029 * the category name of a locale, disregarding a basic linguistic tenet
2030 * that for any object, people will create a name for it. (The next
2031 * version of the POSIX standard is proposed to fix this.) Some vendors
2032 * have created a querylocale() function to do this in the meantime. On
2033 * systems without querylocale(), we have to keep track of what the locale
2034 * has been set to, so that we can return its name so as to emulate
2035 * setlocale(). There are potential problems with this:
2037 * 1) We don't know what calling newlocale() with the locale argument ""
2038 * actually does. It gets its values from the program's environment.
2039 * find_locale_from_environment() is used to work around this. But it
2040 * isn't fool-proof. See the comments for that function for details.
2041 * 2) It's possible for C code in some library to change the locale
2042 * without us knowing it, and thus our records become wrong;
2043 * querylocale() would catch this. But as of September 2017, there
2044 * are no occurrences in CPAN of uselocale(). Some libraries do use
2045 * setlocale(), but that changes the global locale, and threads using
2046 * per-thread locales will just ignore those changes.
2047 * 3) Many systems have multiple names for the same locale. Generally,
2048 * there is an underlying base name, with aliases that evaluate to it.
2049 * On some systems, if you set the locale to an alias, and then
2050 * retrieve the name, you get the alias as expected; but on others you
2051 * get the base name, not the alias you used. And sometimes the
2052 * charade is incomplete. See
2053 * https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269375.
2055 * The code is structured so that the returned locale name when the
2056 * locale is changed is whatever the result of querylocale() on the
2057 * new locale is. This effectively gives the result the system
2058 * expects. Without querylocale, the name returned is always the
2059 * input name. Theoretically this could cause problems, but khw knows
2060 * of none so far, but mentions it here in case you are trying to
2061 * debug something. (This could be worked around by messing with the
2062 * global locale temporarily, using setlocale() to get the base name;
2063 * but that could cause a race. The comments for
2064 * find_locale_from_environment() give details on the potential race.)
2067 const locale_t cur_obj = uselocale((locale_t) 0);
2068 const char * retval;
2070 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "querylocale_2008_i(%s) on %p;"
2071 " called from %" LINE_Tf "\n",
2072 category_names[index], cur_obj,
2075 if (UNLIKELY(cur_obj == LC_GLOBAL_LOCALE)) {
2077 /* Even on platforms that have querylocale(), it is unclear if they
2078 * work in the global locale, and we have the means to get the correct
2079 * answer anyway. khw is unsure this situation even comes up these
2080 * days, hence the branch prediction */
2081 POSIX_SETLOCALE_LOCK;
2082 retval = mortalized_pv_copy(posix_setlocale(categories[index], NULL));
2083 POSIX_SETLOCALE_UNLOCK;
2086 /* Here we have handled the case of the the current locale being the global
2087 * one. Below is the 'else' case of that. There are two different
2088 * implementations, depending on USE_PL_CURLOCALES */
2090 # ifdef USE_PL_CURLOCALES
2094 /* PL_curlocales[] is kept up-to-date for all categories except LC_ALL,
2095 * which may have been invalidated by setting it to NULL, and if so,
2096 * should now be calculated. (The called function updates that
2098 if (index == LC_ALL_INDEX_ && PL_curlocales[LC_ALL_INDEX_] == NULL) {
2099 calculate_LC_ALL_string((const char **) &PL_curlocales,
2105 if (cur_obj == PL_C_locale_obj) {
2107 /* If the current locale object is the C object, then the answer is
2108 * "C" or POSIX, regardless of the category. Handling this
2109 * reasonably likely case specially shortcuts extra effort, and
2110 * hides some bugs from us in OS's that alias other locales to C,
2111 * but do so incompletely. If our records say it is POSIX, use
2112 * that; otherwise use C. See
2113 * https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269375 */
2114 retval = mortalized_pv_copy((strEQ(PL_curlocales[index], "POSIX"))
2119 retval = mortalized_pv_copy(PL_curlocales[index]);
2125 /* Below is the implementation of the 'else' clause which handles the case
2126 * of the current locale not being the global one on platforms where
2127 * USE_PL_CURLOCALES is NOT in effect. That means the system must have
2128 * some form of querylocale. But these have varying characteristics, so
2129 * first create some #defines to make the actual 'else' clause uniform.
2131 * First, glibc has a function that implements querylocale(), but is called
2132 * something else, and takes the category number; the others take the mask.
2134 # if defined(USE_QUERYLOCALE) && ( defined(_NL_LOCALE_NAME) \
2135 && defined(HAS_NL_LANGINFO_L))
2136 # define my_querylocale(index, cur_obj) \
2137 nl_langinfo_l(_NL_LOCALE_NAME(categories[index]), cur_obj)
2139 /* Experience so far shows it is thread-safe, as well as glibc's
2140 * nl_langinfo_l(), so unless overridden, mark it so */
2141 # ifdef NO_THREAD_SAFE_QUERYLOCALE
2142 # undef HAS_THREAD_SAFE_QUERYLOCALE
2144 # define HAS_THREAD_SAFE_QUERYLOCALE
2146 # else /* below, ! glibc */
2148 /* Otherwise, use the system's querylocale(). */
2149 # define my_querylocale(index, cur_obj) \
2150 querylocale(category_masks[index], cur_obj)
2152 /* There is no standard for this function, and khw has never seen
2153 * anything beyond minimal vendor documentation, lacking important
2154 * details. Experience has shown that some implementations have race
2155 * condiions, and their returns may not be thread safe. It would be
2156 * unreliable to test for complete thread safety in Configure. What we
2157 * do instead is to assume that it is thread-safe, unless overriden by,
2158 * say, a hints file specifying
2159 * -Accflags='-DNO_THREAD_SAFE_QUERYLOCALE */
2160 # ifdef NO_THREAD_SAFE_QUERYLOCALE
2161 # undef HAS_THREAD_SAFE_QUERYLOCALE
2163 # define HAS_THREAD_SAFE_QUERYLOCALE
2167 /* Here, we have set up enough information to know if this querylocale()
2168 * is thread-safe, or needs to use a mutex */
2169 # ifdef HAS_THREAD_SAFE_QUERYLOCALE
2170 # define QUERYLOCALE_LOCK
2171 # define QUERYLOCALE_UNLOCK
2173 # define QUERYLOCALE_LOCK gwLOCALE_LOCK
2174 # define QUERYLOCALE_UNLOCK gwLOCALE_UNLOCK
2177 /* Finally, everything is ready, so here is the 'else' clause to implement
2178 * the case of the current locale not being the global one on systems that
2179 * have some form of querylocale(). (POSIX will presumably eventually
2180 * publish their next version in their pipeline, which will define a
2181 * precisely specified querylocale equivalent, and there can be a new
2182 * #ifdef to use it without having to guess at its characteristics) */
2185 /* We don't keep records when there is querylocale(), so as to avoid the
2186 * pitfalls mentioned at the beginning of this function.
2188 * That means LC_ALL has to be calculated from all its constituent
2189 * categories each time, since the querylocale() forms on many (if not
2190 * all) platforms only work on individual categories */
2191 if (index == LC_ALL_INDEX_) {
2192 retval = calculate_LC_ALL_string(NULL, INTERNAL_FORMAT,
2199 retval = savepv(my_querylocale(index, cur_obj));
2202 /* querylocale() may conflate the C locale with something that
2203 * isn't exactly the same. See for example
2204 * https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269375
2205 * We know that if the locale object is the C one, we
2206 * are in the C locale, which may go by the name POSIX, as both, by
2207 * definition, are equivalent. But we consider any other name
2208 * spurious, so override with "C". As in the PL_CURLOCALES case
2209 * above, this hides those glitches, for the most part, from the
2210 * rest of our code. (The code is ordered this way so that if the
2211 * system distinugishes "C" from "POSIX", we do too.) */
2212 if (cur_obj == PL_C_locale_obj && ! isNAME_C_OR_POSIX(retval)) {
2214 retval = savepv("C");
2221 # undef QUERYLOCALE_LOCK
2222 # undef QUERYLOCALE_UNLOCK
2225 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2226 "querylocale_2008_i(%s) returning '%s'\n",
2227 category_names[index], retval));
2228 assert(strNE(retval, ""));
2232 /*---------------------------------------------------------------------------*/
2234 # define bool_setlocale_i(i, locale) \
2235 bool_setlocale_2008_i(i, locale, __LINE__)
2236 # define bool_setlocale_c(cat, locale) \
2237 bool_setlocale_i(cat##_INDEX_, locale)
2238 # define bool_setlocale_r(cat, locale) \
2239 bool_setlocale_i(get_category_index(cat), locale)
2241 /* If this doesn't exist on this platform, make it a no-op (to save #ifdefs) */
2242 # ifndef update_PL_curlocales_i
2243 # define update_PL_curlocales_i(index, new_locale, caller_line)
2247 S_bool_setlocale_2008_i(pTHX_
2249 /* Our internal index of the 'category' setlocale is called with */
2250 const locale_category_index index,
2251 const char * new_locale, /* The locale to set the category to */
2252 const line_t caller_line /* Called from this line number */
2255 PERL_ARGS_ASSERT_BOOL_SETLOCALE_2008_I;
2256 assert(index <= LC_ALL_INDEX_);
2258 /* This function effectively performs a setlocale() on just the current
2259 * thread; thus it is thread-safe. It does this by using the POSIX 2008
2260 * locale functions to emulate the behavior of setlocale(). Similar to
2261 * regular setlocale(), the return from this function points to memory that
2262 * can be overwritten by other system calls, so needs to be copied
2263 * immediately if you need to retain it. The difference here is that
2264 * system calls besides another setlocale() can overwrite it.
2266 * By doing this, most locale-sensitive functions become thread-safe. The
2267 * exceptions are mostly those that return a pointer to static memory.
2270 int mask = category_masks[index];
2271 const locale_t entry_obj = uselocale((locale_t) 0);
2272 const char * locale_on_entry = querylocale_i(index);
2274 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2275 "bool_setlocale_2008_i: input=%d (%s), mask=0x%x,"
2276 " new locale=\"%s\", current locale=\"%s\","
2277 " index=%d, entry object=%p;"
2278 " called from %" LINE_Tf "\n",
2279 categories[index], category_names[index], mask,
2280 ((new_locale == NULL) ? "(nil)" : new_locale),
2281 locale_on_entry, index, entry_obj, caller_line));
2283 /* Here, trying to change the locale, but it is a no-op if the new boss is
2284 * the same as the old boss. Except this routine is called when converting
2285 * from the global locale, so in that case we will create a per-thread
2286 * locale below (with the current values). It also seemed that newlocale()
2287 * could free up the basis locale memory if we called it with the new and
2288 * old being the same, but khw now thinks that this was due to some other
2289 * bug, since fixed, as there are other places where newlocale() gets
2290 * similarly called without problems. */
2291 if ( entry_obj != LC_GLOBAL_LOCALE
2293 && strEQ(new_locale, locale_on_entry))
2295 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2296 "bool_setlocale_2008_i: no-op to change to"
2297 " what it already was\n"));
2301 # ifndef USE_QUERYLOCALE
2303 /* Without a querylocale() mechanism, we have to figure out ourselves what
2304 * happens with setting a locale to "" */
2306 if (strEQ(new_locale, "")) {
2307 new_locale = find_locale_from_environment(index);
2315 # ifdef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
2317 const bool need_loop = false;
2321 bool need_loop = false;
2322 const char * new_locales[LC_ALL_INDEX_] = { NULL };
2324 /* If we're going to have to parse the LC_ALL string, might as well do it
2325 * now before we have made changes that we would have to back out of if the
2327 if (index == LC_ALL_INDEX_) {
2328 switch (parse_LC_ALL_string(new_locale,
2329 (const char **) &new_locales,
2330 override_if_ignored,
2331 false, /* Return only [0] if suffices */
2332 false, /* Don't panic on error */
2343 case only_element_0:
2344 SAVEFREEPV(new_locales[0]);
2345 new_locale = new_locales[0];
2356 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
2358 /* For this bug, if the LC_MESSAGES locale changes, we have to do an
2359 * expensive workaround. Save the current value so we can later determine
2361 const char * old_messages_locale = NULL;
2362 if ( (index == LC_MESSAGES_INDEX_ || index == LC_ALL_INDEX_)
2363 && LIKELY(PL_phase != PERL_PHASE_CONSTRUCT))
2365 old_messages_locale = querylocale_c(LC_MESSAGES);
2370 assert(PL_C_locale_obj);
2372 /* Now ready to switch to the input 'new_locale' */
2374 /* Switching locales generally entails freeing the current one's space (at
2375 * the C library's discretion), hence we can't be using that locale at the
2376 * time of the switch (this wasn't obvious to khw from the man pages). So
2377 * switch to a known locale object that we don't otherwise mess with. */
2378 if (! uselocale(PL_C_locale_obj)) {
2380 /* Not being able to change to the C locale is severe; don't keep
2382 setlocale_failure_panic_i(index, locale_on_entry, "C",
2383 __LINE__, caller_line);
2384 NOT_REACHED; /* NOTREACHED */
2387 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2388 "bool_setlocale_2008_i: now using C"
2389 " object=%p\n", PL_C_locale_obj));
2391 /* These two objects are special:
2392 * LC_GLOBAL_LOCALE because it is undefined behavior to call
2393 * newlocale() with it as a parameter.
2394 * PL_C_locale_obj because newlocale() generally destroys its locale
2395 * object parameter when it succeeds; and we don't
2396 * want that happening to this immutable object.
2397 * Copies will be made for them to use instead if we get so far as to call
2399 bool entry_obj_is_special = ( entry_obj == LC_GLOBAL_LOCALE
2400 || entry_obj == PL_C_locale_obj);
2403 /* PL_C_locale_obj is LC_ALL set to the C locale. If this call is to
2404 * switch to LC_ALL => C, simply use that object. But in fact, we already
2405 * have switched to it just above, in preparation for the general case.
2406 * Since we're already there, no need to do further switching. */
2407 if (mask == LC_ALL_MASK && isNAME_C_OR_POSIX(new_locale)) {
2408 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2409 "bool_setlocale_2008_i: will stay in C"
2411 new_obj = PL_C_locale_obj;
2413 /* 'entry_obj' is now dangling, of no further use to anyone (unless it
2414 * is one of the special ones). Free it to avoid a leak */
2415 if (! entry_obj_is_special) {
2416 freelocale(entry_obj);
2419 update_PL_curlocales_i(index, new_locale, caller_line);
2421 else { /* Here is the general case, not to LC_ALL => C */
2423 /* The newlocale() call(s) below take a basis object to build upon to
2424 * create the changed locale, trashing it iff successful.
2426 * For the objects that are not to be modified by this function, we
2427 * create a duplicate that gets trashed instead.
2429 * Also if we will have to loop doing multiple newlocale()s, there is a
2430 * chance we will succeed for the first few, and then fail, having to
2431 * back out. We need to duplicate 'entry_obj' in this case as well, so
2432 * it remains valid as something to back out to. */
2433 locale_t basis_obj = entry_obj;
2435 if (entry_obj_is_special || need_loop) {
2436 basis_obj = duplocale(basis_obj);
2438 locale_panic_via_("duplocale failed", __FILE__, caller_line);
2439 NOT_REACHED; /* NOTREACHED */
2442 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2443 "bool_setlocale_2008_i created %p by"
2444 " duping the input\n", basis_obj));
2447 # define DEBUG_NEW_OBJECT_CREATED(category, locale, new, old, caller_line) \
2448 DEBUG_Lv(PerlIO_printf(Perl_debug_log, \
2449 "bool_setlocale_2008_i(%s, %s): created %p" \
2450 " while freeing %p; called from %" LINE_Tf \
2451 " via %" LINE_Tf "\n", \
2452 category, locale, new, old, \
2453 caller_line, __LINE__))
2454 # define DEBUG_NEW_OBJECT_FAILED(category, locale, basis_obj) \
2455 DEBUG_L(PerlIO_printf(Perl_debug_log, \
2456 "bool_setlocale_2008_i: creating new object" \
2457 " for (%s '%s') from %p failed; called from %" \
2458 LINE_Tf " via %" LINE_Tf "\n", \
2459 category, locale, basis_obj, \
2460 caller_line, __LINE__));
2462 /* Ready to create a new locale by modification of the existing one.
2464 * NOTE: This code may incorrectly show up as a leak under the address
2465 * sanitizer. We do not free this object under normal teardown, however
2466 * you can set PERL_DESTRUCT_LEVEL=2 to cause it to be freed.
2469 # ifdef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
2471 /* Some platforms have a newlocale() that can handle disparate LC_ALL
2472 * input, so on these a single call to newlocale() always works */
2475 /* If a single call to newlocale() will do */
2481 new_obj = newlocale(mask,
2482 override_ignored_category(index, new_locale),
2485 DEBUG_NEW_OBJECT_FAILED(category_names[index], new_locale,
2488 /* Since the call failed, it didn't trash 'basis_obj', which is
2489 * a dup for these objects, and hence would leak if we don't
2490 * free it. XXX However, something is seriously wrong if we
2491 * can't switch to C or the global locale, so maybe should
2493 if (entry_obj_is_special) {
2494 freelocale(basis_obj);
2497 goto must_restore_state;
2500 DEBUG_NEW_OBJECT_CREATED(category_names[index], new_locale,
2501 new_obj, basis_obj, caller_line);
2503 update_PL_curlocales_i(index, new_locale, caller_line);
2506 # ifndef NEWLOCALE_HANDLES_DISPARATE_LC_ALL
2508 else { /* Need multiple newlocale() calls */
2510 /* Loop through the individual categories, setting the locale of
2511 * each to the corresponding name previously populated into
2512 * newlocales[]. Each iteration builds on the previous one, adding
2513 * its category to what's already been calculated, and taking as a
2514 * basis for what's been calculated 'basis_obj', which is updated
2515 * each iteration to be the result of the previous one. Upon
2516 * success, newlocale() trashes the 'basis_obj' parameter to it.
2517 * If any iteration fails, we immediately give up, restore the
2518 * locale to what it was at the time this function was called
2519 * (saved in 'entry_obj'), and return failure. */
2521 /* Loop, using the previous iteration's result as the basis for the
2522 * next one. (The first time we effectively use the locale in
2523 * force upon entry to this function.) */
2524 for_all_individual_category_indexes(i) {
2525 new_obj = newlocale(category_masks[i],
2529 DEBUG_NEW_OBJECT_CREATED(category_names[i],
2533 basis_obj = new_obj;
2537 /* Failed. Likely this is because the proposed new locale
2538 * isn't valid on this system. */
2540 DEBUG_NEW_OBJECT_FAILED(category_names[i],
2544 /* newlocale() didn't trash this, since the function call
2546 freelocale(basis_obj);
2548 for_all_individual_category_indexes(j) {
2549 Safefree(new_locales[j]);
2552 goto must_restore_state;
2555 /* Success for all categories. */
2556 for_all_individual_category_indexes(i) {
2557 update_PL_curlocales_i(i, new_locales[i], caller_line);
2558 Safefree(new_locales[i]);
2561 /* We dup'd entry_obj in case we had to fall back to it. The
2562 * newlocale() above destroyed the dup when it first succeeded, but
2563 * entry_obj itself is left dangling, so free it */
2564 if (! entry_obj_is_special) {
2565 freelocale(entry_obj);
2569 # endif /* End of newlocale can't handle disparate LC_ALL input */
2573 # undef DEBUG_NEW_OBJECT_CREATED
2574 # undef DEBUG_NEW_OBJECT_FAILED
2576 /* Here, successfully created an object representing the desired locale;
2577 * now switch into it */
2578 if (! uselocale(new_obj)) {
2579 freelocale(new_obj);
2580 locale_panic_(Perl_form(aTHX_ "(called from %" LINE_Tf "):"
2581 " bool_setlocale_2008_i: switching"
2582 " into new locale failed",
2586 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2587 "bool_setlocale_2008_i: now using %p\n", new_obj));
2589 # ifdef MULTIPLICITY /* Unlikely, but POSIX 2008 functions could be
2590 Configured to be used on unthreaded perls, in which
2591 case this object doesn't exist */
2593 if (DEBUG_Lv_TEST) {
2594 if (PL_cur_locale_obj != new_obj) {
2595 PerlIO_printf(Perl_debug_log,
2596 "bool_setlocale_2008_i: PL_cur_locale_obj"
2597 " was %p, now is %p\n",
2598 PL_cur_locale_obj, new_obj);
2602 /* Update the current object */
2603 PL_cur_locale_obj = new_obj;
2606 # ifdef HAS_GLIBC_LC_MESSAGES_BUG
2608 /* Invalidate the glibc cache of loaded translations if the locale has
2609 * changed, see [perl #134264] and
2610 * https://sourceware.org/bugzilla/show_bug.cgi?id=24936 */
2611 if (old_messages_locale) {
2612 if (strNE(old_messages_locale, querylocale_c(LC_MESSAGES))) {
2613 textdomain(textdomain(NULL));
2623 /* We earlier switched to the LC_ALL => C locale in anticipation of it
2624 * succeeding, Now have to switch back to the state upon entry. */
2625 if (! uselocale(entry_obj)) {
2626 setlocale_failure_panic_i(index, "switching back to",
2627 locale_on_entry, __LINE__, caller_line);
2633 /*---------------------------------------------------------------------------*/
2635 # define void_setlocale_i_with_caller(i, locale, file, line) \
2637 if (! bool_setlocale_i(i, locale)) \
2638 setlocale_failure_panic_via_i(i, NULL, locale, __LINE__, 0, \
2642 # define void_setlocale_r_with_caller(cat, locale, file, line) \
2643 void_setlocale_i_with_caller(get_category_index(cat), locale, \
2646 # define void_setlocale_c_with_caller(cat, locale, file, line) \
2647 void_setlocale_i_with_caller(cat##_INDEX_, locale, file, line)
2649 # define void_setlocale_i(i, locale) \
2650 void_setlocale_i_with_caller(i, locale, __FILE__, __LINE__)
2651 # define void_setlocale_c(cat, locale) \
2652 void_setlocale_i(cat##_INDEX_, locale)
2653 # define void_setlocale_r(cat, locale) \
2654 void_setlocale_i(get_category_index(cat), locale)
2656 /*===========================================================================*/
2659 # error Unexpected Configuration
2660 #endif /* End of the various implementations of the setlocale and
2661 querylocale macros used in the remainder of this program */
2663 /* query_nominal_locale_i() is used when the caller needs the locale that an
2664 * external caller would be expecting, and not what we're secretly using
2665 * behind the scenes. It deliberately doesn't handle LC_ALL; use
2666 * calculate_LC_ALL_string() for that. */
2667 #ifdef USE_LOCALE_NUMERIC
2668 # define query_nominal_locale_i(i) \
2669 (__ASSERT_(i != LC_ALL_INDEX_) \
2670 ((i == LC_NUMERIC_INDEX_) ? PL_numeric_name : querylocale_i(i)))
2672 # define query_nominal_locale_i(i) \
2673 (__ASSERT_(i != LC_ALL_INDEX_) querylocale_i(i))
2676 #ifdef USE_PL_CURLOCALES
2679 S_update_PL_curlocales_i(pTHX_
2680 const locale_category_index index,
2681 const char * new_locale,
2682 const line_t caller_line)
2684 /* Update PL_curlocales[], which is parallel to the other ones indexed by
2685 * our mapping of libc category number to our internal equivalents. */
2687 PERL_ARGS_ASSERT_UPDATE_PL_CURLOCALES_I;
2688 assert(index <= LC_ALL_INDEX_);
2690 if (index == LC_ALL_INDEX_) {
2692 /* For LC_ALL, we change all individual categories to correspond,
2693 * including the LC_ALL element */
2694 for (unsigned int i = 0; i <= LC_ALL_INDEX_; i++) {
2695 Safefree(PL_curlocales[i]);
2696 PL_curlocales[i] = NULL;
2699 switch (parse_LC_ALL_string(new_locale,
2700 (const char **) &PL_curlocales,
2701 check_that_overridden, /* things should
2705 true, /* Always fill array */
2706 true, /* Panic if fails, as to get here
2707 it earlier had to have succeeded
2713 case only_element_0:
2714 locale_panic_via_("Unexpected return from parse_LC_ALL_string",
2715 __FILE__, caller_line);
2718 /* parse_LC_ALL_string() has already filled PL_curlocales properly,
2719 * except for the LC_ALL element, which should be set to
2721 PL_curlocales[LC_ALL_INDEX_] = savepv(new_locale);
2724 else { /* Not LC_ALL */
2726 /* Update the single category's record */
2727 Safefree(PL_curlocales[index]);
2728 PL_curlocales[index] = savepv(new_locale);
2730 /* Invalidate LC_ALL */
2731 Safefree(PL_curlocales[LC_ALL_INDEX_]);
2732 PL_curlocales[LC_ALL_INDEX_] = NULL;
2736 # endif /* Need PL_curlocales[] */
2738 /*===========================================================================*/
2740 #if defined(USE_LOCALE)
2742 /* This paradigm is needed in several places in the function below. We have to
2743 * substitute the nominal locale for LC_NUMERIC when returning a value for
2744 * external consumption */
2745 # ifndef USE_LOCALE_NUMERIC
2746 # define ENTRY(i, array, format) array[i]
2748 # define ENTRY(i, array, format) \
2749 (UNLIKELY( format == EXTERNAL_FORMAT_FOR_QUERY \
2750 && i == LC_NUMERIC_INDEX_) \
2757 S_calculate_LC_ALL_string(pTHX_ const char ** category_locales_list,
2758 const calc_LC_ALL_format format,
2759 const calc_LC_ALL_return returning,
2760 const line_t caller_line)
2762 PERL_ARGS_ASSERT_CALCULATE_LC_ALL_STRING;
2764 /* NOTE: On Configurations that have PL_curlocales[], this function has the
2765 * side effect of updating the LC_ALL_INDEX_ element with its result.
2767 * This function calculates a string that defines the locale(s) LC_ALL is
2768 * set to, in either:
2769 * 1) Our internal format if 'format' is set to INTERNAL_FORMAT.
2770 * 2) The external format returned by Perl_setlocale() if 'format' is set
2771 * to EXTERNAL_FORMAT_FOR_QUERY or EXTERNAL_FORMAT_FOR_SET.
2773 * These two are distinguished by:
2774 * a) EXTERNAL_FORMAT_FOR_SET returns the actual locale currently in
2776 * b) EXTERNAL_FORMAT_FOR_QUERY returns the nominal locale.
2777 * Currently this can differ only from the actual locale in the
2778 * LC_NUMERIC category when it is set to a locale whose radix is
2779 * not a dot. (The actual locale is kept as a dot to accommodate
2780 * the large corpus of XS code that expects it to be that;
2781 * switched to a non-dot temporarily during certain operations
2782 * that require the actual radix.)
2784 * In both 1) and 2), LC_ALL's values are passed to this function by
2785 * 'category_locales_list' which is either:
2786 * 1) a pointer to an array of strings with up-to-date values of all the
2787 * individual categories; or
2788 * 2) NULL, to indicate to use querylocale_i() to get each individual
2791 * The caller sets 'returning' to
2792 * WANT_TEMP_PV the function returns the calculated string
2793 * as a mortalized temporary, so the caller
2794 * doesn't have to worry about it being
2795 * per-thread, nor needs to arrange for its
2797 * WANT_PL_setlocale_buf the function stores the calculated string
2798 * into the per-thread buffer PL_setlocale_buf
2799 * and returns a pointer to that. The buffer
2800 * is cleaned up automatically in process
2801 * destruction. This return method avoids
2802 * extra copies in some circumstances.
2803 * WANT_VOID NULL is returned. This is used when the
2804 * function is being called only for its side
2805 * effect of updating
2806 * PL_curlocales[LC_ALL_INDEX_]
2808 * querylocale(), on systems that have it, doesn't tend to work for LC_ALL.
2809 * So we have to construct the answer ourselves based on the passed in
2812 * If all individual categories are the same locale, we can just set LC_ALL
2813 * to that locale. But if not, we have to create an aggregation of all the
2814 * categories on the system. Platforms differ as to the syntax they use
2815 * for these non-uniform locales for LC_ALL. Some, like glibc and Windows,
2816 * use an unordered series of name=value pairs, like
2817 * LC_NUMERIC=C;LC_TIME=en_US.UTF-8;...
2818 * to specify LC_ALL; others, like *BSD, use a positional notation with a
2819 * delimitter, typically a single '/' character:
2822 * When the external format is desired, this function returns whatever the
2823 * system expects. The internal format is always name=value pairs.
2825 * For systems that have categories we don't know about, the algorithm
2826 * below won't know about those missing categories, leading to potential
2827 * bugs for code that looks at them. If there is an environment variable
2828 * that sets that category, we won't know to look for it, and so our use of
2829 * LANG or "C" improperly overrides it. On the other hand, if we don't do
2830 * what is done here, and there is no environment variable, the category's
2831 * locale should be set to LANG or "C". So there is no good solution. khw
2832 * thinks the best is to make sure we have a complete list of possible
2833 * categories, adding new ones as they show up on obscure platforms.
2836 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
2837 "Entering calculate_LC_ALL_string(%s);"
2838 " called from %" LINE_Tf "\n",
2839 ((format == EXTERNAL_FORMAT_FOR_QUERY)
2840 ? "EXTERNAL_FORMAT_FOR_QUERY"
2841 : ((format == EXTERNAL_FORMAT_FOR_SET)
2842 ? "EXTERNAL_FORMAT_FOR_SET"
2843 : "INTERNAL_FORMAT")),
2846 bool input_list_was_NULL = (category_locales_list == NULL);
2848 /* If there was no input category list, construct a temporary one
2850 const char * my_category_locales_list[LC_ALL_INDEX_];
2851 const char ** locales_list = category_locales_list;
2852 if (locales_list == NULL) {
2853 locales_list = my_category_locales_list;
2855 if (format == EXTERNAL_FORMAT_FOR_QUERY) {
2856 for_all_individual_category_indexes(i) {
2857 locales_list[i] = query_nominal_locale_i(i);
2861 for_all_individual_category_indexes(i) {
2862 locales_list[i] = querylocale_i(i);
2867 /* While we are calculating LC_ALL, we see if every category's locale is
2868 * the same as every other's or not. */
2869 # ifndef HAS_IGNORED_LOCALE_CATEGORIES_
2871 /* When we pay attention to all categories, we assume they are all the same
2872 * until proven different */
2873 bool disparate = false;
2877 /* But if there are ignored categories, those will be set to "C", so try an
2878 * arbitrary category, and if it isn't C, we know immediately that the
2879 * locales are disparate. (The #if conditionals are to handle the case
2880 * where LC_NUMERIC_INDEX_ is 0. We don't want to use LC_NUMERIC to
2881 * compare, as that may be different between external and internal forms.)
2883 # if ! defined(USE_LOCALE_NUMERIC)
2885 bool disparate = ! isNAME_C_OR_POSIX(locales_list[0]);
2887 # elif LC_NUMERIC_INDEX_ != 0
2889 bool disparate = ! isNAME_C_OR_POSIX(locales_list[0]);
2893 /* Would need revision to handle the very unlikely case where only a single
2894 * category, LC_NUMERIC, is defined */
2895 assert(LOCALE_CATEGORIES_COUNT_ > 0);
2897 bool disparate = ! isNAME_C_OR_POSIX(locales_list[1]);
2902 /* Calculate the needed size for the string listing the individual locales.
2903 * Initialize with values known at compile time. */
2905 const char *separator;
2907 # ifdef PERL_LC_ALL_USES_NAME_VALUE_PAIRS /* Positional formatted LC_ALL */
2908 PERL_UNUSED_ARG(format);
2911 if (format != INTERNAL_FORMAT) {
2913 /* Here, we will be using positional notation. it includes n-1
2915 total_len = ( LOCALE_CATEGORIES_COUNT_ - 1)
2916 * STRLENs(PERL_LC_ALL_SEPARATOR)
2917 + 1; /* And a trailing NUL */
2918 separator = PERL_LC_ALL_SEPARATOR;
2925 /* name=value output is always used in internal format, and when
2926 * positional isn't available on the platform. */
2927 total_len = lc_all_boiler_plate_length;
2931 /* The total length then is just the sum of the above boiler-plate plus the
2932 * total strlen()s of the locale name of each individual category. */
2933 for_all_individual_category_indexes(i) {
2934 const char * entry = ENTRY(i, locales_list, format);
2936 total_len += strlen(entry);
2937 if (! disparate && strNE(entry, locales_list[0])) {
2942 bool free_if_void_return = false;
2943 const char * retval;
2945 /* If all categories have the same locale, we already know the answer */
2947 if (returning == WANT_PL_setlocale_buf) {
2948 save_to_buffer(locales_list[0],
2950 &PL_setlocale_bufsize);
2951 retval = PL_setlocale_buf;
2955 retval = locales_list[0];
2957 /* If a temporary is wanted for the return, and we had to create
2958 * the input list ourselves, we created it into such a temporary,
2959 * so no further work is needed; but otherwise, make a mortal copy
2960 * of this passed-in list element */
2961 if (returning == WANT_TEMP_PV && ! input_list_was_NULL) {
2962 retval = savepv(retval);
2966 /* In all cases here, there's nothing we create that needs to be
2967 * freed, so leave 'free_if_void_return' set to the default
2971 else { /* Here, not all categories have the same locale */
2975 /* If returning to PL_setlocale_buf, set up to write directly to it,
2976 * being sure it is resized to be large enough */
2977 if (returning == WANT_PL_setlocale_buf) {
2978 set_save_buffer_min_size(total_len,
2980 &PL_setlocale_bufsize);
2981 constructed = PL_setlocale_buf;
2983 else { /* Otherwise we need new memory to hold the calculated value. */
2985 Newx(constructed, total_len, char);
2987 /* If returning the new memory, it must be set up to be freed
2988 * later; otherwise at the end of this function */
2989 if (returning == WANT_TEMP_PV) {
2990 SAVEFREEPV(constructed);
2993 free_if_void_return = true;
2997 constructed[0] = '\0';
2999 /* Loop through all the categories */
3000 for_all_individual_category_indexes(j) {
3002 /* Add a separator, except before the first one */
3004 my_strlcat(constructed, separator, total_len);
3011 # ifndef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
3013 if (UNLIKELY(format != INTERNAL_FORMAT)) {
3015 /* In positional notation 'j' means the position, and we have
3016 * to convert to the index 'i' */
3017 i = map_LC_ALL_position_to_index[j];
3019 entry = ENTRY(i, locales_list, format);
3020 needed_len = my_strlcat(constructed, entry, total_len);
3026 /* Below, we are to use name=value notation, either because
3027 * that's what the platform uses, or because this is the
3028 * internal format, which uses that notation regardless of the
3031 entry = ENTRY(i, locales_list, format);
3033 /* "name=locale;" */
3034 my_strlcat(constructed, category_names[i], total_len);
3035 my_strlcat(constructed, "=", total_len);
3036 needed_len = my_strlcat(constructed, entry, total_len);
3039 if (LIKELY(needed_len <= total_len)) {
3043 /* If would have overflowed, panic */
3044 locale_panic_via_(Perl_form(aTHX_
3045 "Internal length calculation wrong.\n"
3046 "\"%s\" was not entirely added to"
3047 " \"%.*s\"; needed=%zu, had=%zu",
3048 entry, (int) total_len,
3050 needed_len, total_len),
3053 } /* End of loop through the categories */
3055 retval = constructed;
3056 } /* End of the categories' locales are displarate */
3058 # if defined(USE_PL_CURLOCALES) && defined(LC_ALL)
3060 if (format == INTERNAL_FORMAT) {
3062 /* PL_curlocales[LC_ALL_INDEX_] is updated as a side-effect of this
3063 * function for internal format. */
3064 Safefree(PL_curlocales[LC_ALL_INDEX_]);
3065 PL_curlocales[LC_ALL_INDEX_] = savepv(retval);
3070 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3071 "calculate_LC_ALL_string calculated '%s'\n",
3074 if (returning == WANT_VOID) {
3075 if (free_if_void_return) {
3085 # if defined(WIN32) || ( defined(USE_POSIX_2008_LOCALE) \
3086 && ! defined(USE_QUERYLOCALE))
3089 S_find_locale_from_environment(pTHX_ const locale_category_index index)
3091 /* NB: This function may actually change the locale on Windows. It
3092 * currently is designed to be called only from setting the locale on
3093 * Windows, and POSIX 2008
3095 * This function returns the locale specified by the program's environment
3096 * for the category specified by our internal index number 'index'. It
3097 * therefore simulates:
3098 * setlocale(cat, "")
3099 * but, except for some cases in Windows, doesn't actually change the
3100 * locale; merely returns it.
3102 * The return need not be freed by the caller. This
3103 * promise relies on PerlEnv_getenv() returning a mortalized copy to us.
3105 * The simulation is needed only on certain platforms; otherwise, libc is
3106 * called with "" to get the actual value(s). The simulation is needed
3109 * 1) On Windows systems, the concept of the POSIX ordering of
3110 * environment variables is missing. To increase portability of
3111 * programs across platforms, the POSIX ordering is emulated on
3114 * 2) On POSIX 2008 systems without querylocale(), it is problematic
3115 * getting the results of the POSIX 2008 equivalent of
3117 * setlocale(category, "")
3119 * To ensure that we know exactly what those values are, we do the
3120 * setting ourselves, using the documented algorithm specified by the
3121 * POSIX standard (assuming the platform follows the Standard) rather
3122 * than use "" as the locale. This will lead to results that differ
3123 * from native behavior if the native behavior differs from the
3124 * Standard's documented value, but khw believes it is better to know
3125 * what's going on, even if different from native, than to just guess.
3127 * glibc systems differ from this standard in having a LANGUAGE
3128 * environment variable used for just LC_MESSAGES. This function does
3131 * Another option for the POSIX 2008 case would be, in a critical
3132 * section, to save the global locale's current value, and do a
3133 * straight setlocale(LC_ALL, ""). That would return our desired
3134 * values, destroying the global locale's, which we would then
3135 * restore. But that could cause races with any other thread that is
3136 * using the global locale and isn't using the mutex. And, the only
3137 * reason someone would have done that is because they are calling a
3138 * library function, like in gtk, that calls setlocale(), and which
3139 * can't be changed to use the mutex. That wouldn't be a problem if
3140 * this were to be done before any threads had switched, say during
3141 * perl construction time. But this code would still be needed for
3144 * The Windows and POSIX 2008 differ in that the ultimate fallback is "C"
3145 * in POSIX, and is the system default locale in Windows. To get that
3146 * system default value, we actually have to call setlocale() on Windows.
3149 const char * const lc_all = PerlEnv_getenv("LC_ALL");
3150 const char * locale_names[LC_ALL_INDEX_] = { NULL };
3152 /* Use any "LC_ALL" environment variable, as it overrides everything else.
3154 if (lc_all && strNE(lc_all, "")) {
3158 /* Here, no usable LC_ALL environment variable. We have to handle each
3159 * category separately. If all categories are desired, we loop through
3160 * them all. If only an individual category is desired, to avoid
3161 * duplicating logic, we use the same loop, but set up the limits so it is
3162 * only executed once, for that particular category. */
3163 locale_category_index lower, upper, offset;
3164 if (index == LC_ALL_INDEX_) {
3165 lower = (locale_category_index) 0;
3166 upper = (locale_category_index) ((int) LC_ALL_INDEX_ - 1);
3167 offset = (locale_category_index) 0;
3173 /* 'offset' is used so that the result of the single loop iteration is
3174 * stored into output[0] */
3178 /* When no LC_ALL environment variable, LANG is used as a default, but
3179 * overridden for individual categories that have corresponding environment
3180 * variables. If no LANG exists, the default is "C" on POSIX 2008, or the
3181 * system default for the category on Windows. */
3182 const char * env_lang = NULL;
3184 /* For each desired category, use any corresponding environment variable;
3185 * or the default if none such exists. */
3186 bool is_disparate = false; /* Assume is uniform until proven otherwise */
3187 for (unsigned i = lower; i <= upper; i++) {
3188 const char * const env_override = PerlEnv_getenv(category_names[i]);
3189 unsigned int j = i - offset;
3191 if (env_override && strNE(env_override, "")) {
3192 locale_names[j] = env_override;
3194 else { /* Here, no corresponding environment variable, see if LANG
3195 exists and is usable. Done this way to avoid fetching LANG
3196 unless it is actually needed */
3197 if (env_lang == NULL) {
3198 env_lang = PerlEnv_getenv("LANG");
3200 /* If not usable, set it to a non-NULL illegal value so won't
3201 * try to use it below */
3202 if (env_lang == NULL || strEQ(env_lang, "")) {
3203 env_lang = (const char *) 1;
3207 /* If a usable LANG exists, use it. */
3208 if (env_lang != NULL && env_lang != (const char *) 1) {
3209 locale_names[j] = env_lang;
3214 /* If no LANG, use the system default on Windows. */
3215 locale_names[j] = wrap_wsetlocale(categories[i], ".ACP");
3216 if (locale_names[j]) {
3217 SAVEFREEPV(locale_names[j]);
3221 { /* If nothing was found or worked, use C */
3222 locale_names[j] = "C";
3227 if (j > 0 && ! is_disparate && strNE(locale_names[0], locale_names[j]))
3229 is_disparate = true;
3232 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3233 "find_locale_from_environment i=%u, j=%u, name=%s,"
3234 " locale=%s, locale of 0th category=%s, disparate=%d\n",
3235 i, j, category_names[i],
3236 locale_names[j], locale_names[0], is_disparate));
3239 if (! is_disparate) {
3240 return locale_names[0];
3243 return calculate_LC_ALL_string(locale_names, INTERNAL_FORMAT,
3249 # if defined(DEBUGGING) || defined(USE_PERL_SWITCH_LOCALE_CONTEXT)
3252 S_get_LC_ALL_display(pTHX)
3254 return calculate_LC_ALL_string(NULL, INTERNAL_FORMAT,
3262 S_setlocale_failure_panic_via_i(pTHX_
3263 const locale_category_index cat_index,
3264 const char * current,
3265 const char * failed,
3266 const line_t proxy_caller_line,
3267 const line_t immediate_caller_line,
3268 const char * const higher_caller_file,
3269 const line_t higher_caller_line)
3271 PERL_ARGS_ASSERT_SETLOCALE_FAILURE_PANIC_VIA_I;
3273 /* Called to panic when a setlocale form unexpectedly failed for the
3274 * category determined by 'cat_index', and the locale that was in effect
3275 * (and likely still is) is 'current'. 'current' may be NULL, which causes
3276 * this function to query what it is.
3278 * The extra caller information is used for when a function acts as a
3279 * stand-in for another function, which a typical reader would more likely
3280 * think would be the caller
3282 * If a line number is 0, its stack (sort-of) frame is omitted; same if
3283 * it's the same line number as the next higher caller. */
3285 const int cat = categories[cat_index];
3286 const char * name = category_names[cat_index];
3290 if (current == NULL) {
3291 current = querylocale_i(cat_index);
3294 const char * proxy_text = "";
3295 if (proxy_caller_line != 0 && proxy_caller_line != immediate_caller_line)
3297 proxy_text = Perl_form(aTHX_ "\nCalled via %s: %" LINE_Tf,
3298 __FILE__, proxy_caller_line);
3300 if ( strNE(__FILE__, higher_caller_file)
3301 || ( immediate_caller_line != 0
3302 && immediate_caller_line != higher_caller_line))
3304 proxy_text = Perl_form(aTHX_ "%s\nCalled via %s: %" LINE_Tf,
3305 proxy_text, __FILE__,
3306 immediate_caller_line);
3309 /* 'false' in the get_displayable_string() calls makes it not think the
3310 * locale is UTF-8, so just dumps bytes. Actually figuring it out can be
3311 * too complicated for a panic situation. */
3312 const char * msg = Perl_form(aTHX_
3313 "Can't change locale for %s (%d) from '%s' to '%s'"
3316 get_displayable_string(current,
3317 current + strlen(current),
3319 get_displayable_string(failed,
3320 failed + strlen(failed),
3325 Perl_locale_panic(msg, __LINE__, higher_caller_file, higher_caller_line);
3326 NOT_REACHED; /* NOTREACHED */
3329 /* Any of these will allow us to find the RADIX */
3330 # if defined(USE_LOCALE_NUMERIC) && ( defined(HAS_SOME_LANGINFO) \
3331 || defined(HAS_LOCALECONV) \
3332 || defined(HAS_SNPRINTF))
3333 # define CAN_CALCULATE_RADIX
3335 # ifdef USE_LOCALE_NUMERIC
3338 S_new_numeric(pTHX_ const char *newnum, bool force)
3340 PERL_ARGS_ASSERT_NEW_NUMERIC;
3342 /* Called after each libc setlocale() or uselocale() call affecting
3343 * LC_NUMERIC, to tell core Perl this and that 'newnum' is the name of the
3344 * new locale, and we are switched into it. It installs this locale as the
3345 * current underlying default, and then switches to the C locale, if
3346 * necessary, so that the code that has traditionally expected the radix
3347 * character to be a dot may continue to do so.
3349 * The default locale and the C locale can be toggled between by use of the
3350 * set_numeric_underlying() and set_numeric_standard() functions, which
3351 * should probably not be called directly, but only via macros like
3352 * SET_NUMERIC_STANDARD() in perl.h.
3354 * The toggling is necessary mainly so that a non-dot radix decimal point
3355 * character can be input and output, while allowing internal calculations
3358 * This sets several interpreter-level variables:
3359 * PL_numeric_name The underlying locale's name: a copy of 'newnum'
3360 * PL_numeric_underlying A boolean indicating if the toggled state is
3361 * such that the current locale is the program's
3363 * PL_numeric_standard An int indicating if the toggled state is such
3364 * that the current locale is the C locale or
3365 * indistinguishable from the C locale. If non-zero, it
3366 * is in C; if > 1, it means it may not be toggled away
3368 * PL_numeric_underlying_is_standard A bool kept by this function
3369 * indicating that the underlying locale and the standard
3370 * C locale are indistinguishable for the purposes of
3371 * LC_NUMERIC. This happens when both of the above two
3372 * variables are true at the same time. (Toggling is a
3373 * no-op under these circumstances.) This variable is
3374 * used to avoid having to recalculate.
3375 * PL_numeric_radix_sv Contains the string that code should use for the
3376 * decimal point. It is set to either a dot or the
3377 * program's underlying locale's radix character string,
3378 * depending on the situation.
3379 * PL_underlying_radix_sv Contains the program's underlying locale's
3380 * radix character string. This is copied into
3381 * PL_numeric_radix_sv when the situation warrants. It
3382 * exists to avoid having to recalculate it when toggling.
3385 DEBUG_L( PerlIO_printf(Perl_debug_log,
3386 "Called new_numeric with %s, PL_numeric_name=%s\n",
3387 newnum, PL_numeric_name));
3389 /* If not forcing this procedure, and there isn't actually a change from
3390 * our records, do nothing. (Our records can be wrong when sync'ing to the
3391 * locale set up by an external library, hence the 'force' parameter) */
3392 if (! force && strEQ(PL_numeric_name, newnum)) {
3396 Safefree(PL_numeric_name);
3397 PL_numeric_name = savepv(newnum);
3399 /* Handle the trivial case. Since this is called at process
3400 * initialization, be aware that this bit can't rely on much being
3402 if (isNAME_C_OR_POSIX(PL_numeric_name)) {
3403 PL_numeric_standard = TRUE;
3404 PL_numeric_underlying_is_standard = TRUE;
3405 PL_numeric_underlying = TRUE;
3406 sv_setpv(PL_numeric_radix_sv, C_decimal_point);
3407 SvUTF8_off(PL_numeric_radix_sv);
3408 sv_setpv(PL_underlying_radix_sv, C_decimal_point);
3409 SvUTF8_off(PL_underlying_radix_sv);
3413 /* We are in the underlying locale until changed at the end of this
3415 PL_numeric_underlying = TRUE;
3417 char * radix = NULL;
3418 utf8ness_t utf8ness = UTF8NESS_IMMATERIAL;
3420 /* Find and save this locale's radix character. */
3421 my_langinfo_c(RADIXCHAR, LC_NUMERIC, PL_numeric_name,
3422 &radix, NULL, &utf8ness);
3423 sv_setpv(PL_underlying_radix_sv, radix);
3425 if (utf8ness == UTF8NESS_YES) {
3426 SvUTF8_on(PL_underlying_radix_sv);
3429 SvUTF8_off(PL_underlying_radix_sv);
3432 DEBUG_L(PerlIO_printf(Perl_debug_log,
3433 "Locale radix is '%s', ?UTF-8=%d\n",
3434 SvPVX(PL_underlying_radix_sv),
3435 cBOOL(SvUTF8(PL_underlying_radix_sv))));
3437 /* This locale is indistinguishable from C (for numeric purposes) if both
3438 * the radix character and the thousands separator are the same as C's.
3439 * Start with the radix. */
3440 PL_numeric_underlying_is_standard = strEQ(C_decimal_point, radix);
3443 # ifndef TS_W32_BROKEN_LOCALECONV
3445 /* If the radix isn't the same as C's, we know it is distinguishable from
3446 * C; otherwise check the thousands separator too. Only if both are the
3447 * same as C's is the locale indistinguishable from C.
3449 * But on earlier Windows versions, there is a potential race. This code
3450 * knows that localeconv() (elsewhere in this file) will be used to extract
3451 * the needed value, and localeconv() was buggy for quite a while, and that
3452 * code in this file hence uses a workaround. And that workaround may have
3453 * an (unlikely) race. Gathering the radix uses a different workaround on
3454 * Windows that doesn't involve a race. It might be possible to do the
3455 * same for this (patches welcome).
3457 * Until then khw doesn't think it's worth even the small risk of a race to
3458 * get this value, which doesn't appear to be used in any of the Microsoft
3459 * library routines anyway. */
3461 if (PL_numeric_underlying_is_standard) {
3462 char * scratch_buffer = NULL;
3463 PL_numeric_underlying_is_standard = strEQ(C_thousands_sep,
3464 my_langinfo_c(THOUSEP, LC_NUMERIC,
3468 Safefree(scratch_buffer);
3473 PL_numeric_standard = PL_numeric_underlying_is_standard;
3475 /* Keep LC_NUMERIC so that it has the C locale radix and thousands
3476 * separator. This is for XS modules, so they don't have to worry about
3477 * the radix being a non-dot. (Core operations that need the underlying
3478 * locale change to it temporarily). */
3479 if (! PL_numeric_standard) {
3480 set_numeric_standard(__FILE__, __LINE__);
3487 Perl_set_numeric_standard(pTHX_ const char * const file, const line_t line)
3489 PERL_ARGS_ASSERT_SET_NUMERIC_STANDARD;
3490 PERL_UNUSED_ARG(file); /* Some Configurations ignore these */
3491 PERL_UNUSED_ARG(line);
3493 # ifdef USE_LOCALE_NUMERIC
3495 /* Unconditionally toggle the LC_NUMERIC locale to the C locale
3497 * Most code should use the macro SET_NUMERIC_STANDARD() in perl.h
3498 * instead of calling this directly. The macro avoids calling this routine
3499 * if toggling isn't necessary according to our records (which could be
3500 * wrong if some XS code has changed the locale behind our back) */
3502 DEBUG_L(PerlIO_printf(Perl_debug_log, "Setting LC_NUMERIC locale to"
3503 " standard C; called from %s: %"
3504 LINE_Tf "\n", file, line));
3506 void_setlocale_c_with_caller(LC_NUMERIC, "C", file, line);
3507 PL_numeric_standard = TRUE;
3508 sv_setpv(PL_numeric_radix_sv, C_decimal_point);
3509 SvUTF8_off(PL_numeric_radix_sv);
3511 PL_numeric_underlying = PL_numeric_underlying_is_standard;
3513 # endif /* USE_LOCALE_NUMERIC */
3518 Perl_set_numeric_underlying(pTHX_ const char * const file, const line_t line)
3520 PERL_ARGS_ASSERT_SET_NUMERIC_UNDERLYING;
3521 PERL_UNUSED_ARG(file); /* Some Configurations ignore these */
3522 PERL_UNUSED_ARG(line);
3524 # ifdef USE_LOCALE_NUMERIC
3526 /* Unconditionally toggle the LC_NUMERIC locale to the current underlying
3529 * Most code should use the macro SET_NUMERIC_UNDERLYING() in perl.h
3530 * instead of calling this directly. The macro avoids calling this routine
3531 * if toggling isn't necessary according to our records (which could be
3532 * wrong if some XS code has changed the locale behind our back) */
3534 DEBUG_L(PerlIO_printf(Perl_debug_log, "Setting LC_NUMERIC locale to %s;"
3535 " called from %s: %" LINE_Tf "\n",
3536 PL_numeric_name, file, line));
3537 /* Maybe not in init? assert(PL_locale_mutex_depth > 0);*/
3539 void_setlocale_c_with_caller(LC_NUMERIC, PL_numeric_name, file, line);
3540 PL_numeric_underlying = TRUE;
3541 sv_setsv_nomg(PL_numeric_radix_sv, PL_underlying_radix_sv);
3543 PL_numeric_standard = PL_numeric_underlying_is_standard;
3545 # endif /* USE_LOCALE_NUMERIC */
3549 # ifdef USE_LOCALE_CTYPE
3552 S_new_ctype(pTHX_ const char *newctype, bool force)
3554 PERL_ARGS_ASSERT_NEW_CTYPE;
3555 PERL_UNUSED_ARG(force);
3557 /* Called after each libc setlocale() call affecting LC_CTYPE, to tell
3558 * core Perl this and that 'newctype' is the name of the new locale.
3560 * This function sets up the folding arrays for all 256 bytes, assuming
3561 * that tofold() is tolc() since fold case is not a concept in POSIX,
3564 DEBUG_L(PerlIO_printf(Perl_debug_log, "Entering new_ctype(%s)\n",
3567 /* No change means no-op */
3568 if (strEQ(PL_ctype_name, newctype)) {
3572 /* We will replace any bad locale warning with
3573 * 1) nothing if the new one is ok; or
3574 * 2) a new warning for the bad new locale */
3575 if (PL_warn_locale) {
3576 SvREFCNT_dec_NN(PL_warn_locale);
3577 PL_warn_locale = NULL;
3581 Safefree(PL_ctype_name);
3584 PL_in_utf8_turkic_locale = FALSE;
3586 /* For the C locale, just use the standard folds, and we know there are no
3587 * glitches possible, so return early. Since this is called at process
3588 * initialization, be aware that this bit can't rely on much being
3590 if (isNAME_C_OR_POSIX(newctype)) {
3591 Copy(PL_fold, PL_fold_locale, 256, U8);
3592 PL_ctype_name = savepv(newctype);
3593 PL_in_utf8_CTYPE_locale = FALSE;
3597 /* The cache being cleared signals the called function to compute a new
3599 PL_in_utf8_CTYPE_locale = is_locale_utf8(newctype);
3601 PL_ctype_name = savepv(newctype);
3602 bool maybe_utf8_turkic = FALSE;
3604 /* Don't check for problems if we are suppressing the warnings */
3605 bool check_for_problems = ckWARN_d(WARN_LOCALE) || UNLIKELY(DEBUG_L_TEST);
3607 if (PL_in_utf8_CTYPE_locale) {
3609 /* A UTF-8 locale gets standard rules. But note that code still has to
3610 * handle this specially because of the three problematic code points
3612 Copy(PL_fold_latin1, PL_fold_locale, 256, U8);
3614 /* UTF-8 locales can have special handling for 'I' and 'i' if they are
3615 * Turkic. Make sure these two are the only anomalies. (We don't
3616 * require towupper and towlower because they aren't in C89.) */
3618 # if defined(HAS_TOWUPPER) && defined (HAS_TOWLOWER)
3620 if (towupper('i') == 0x130 && towlower('I') == 0x131)
3624 if (toU8_UPPER_LC('i') == 'i' && toU8_LOWER_LC('I') == 'I')
3629 /* This is how we determine it really is Turkic */
3630 check_for_problems = TRUE;
3631 maybe_utf8_turkic = TRUE;
3634 else { /* Not a canned locale we know the values for. Compute them */
3638 bool has_non_ascii_fold = FALSE;
3639 bool found_unexpected = FALSE;
3641 /* Under -DLv, see if there are any folds outside the ASCII range.
3642 * This factoid is used below */
3643 if (DEBUG_Lv_TEST) {
3644 for (unsigned i = 128; i < 256; i++) {
3645 int j = LATIN1_TO_NATIVE(i);
3646 if (toU8_LOWER_LC(j) != j || toU8_UPPER_LC(j) != j) {
3647 has_non_ascii_fold = TRUE;
3655 for (unsigned i = 0; i < 256; i++) {
3656 if (isU8_UPPER_LC(i))
3657 PL_fold_locale[i] = (U8) toU8_LOWER_LC(i);
3658 else if (isU8_LOWER_LC(i))
3659 PL_fold_locale[i] = (U8) toU8_UPPER_LC(i);
3661 PL_fold_locale[i] = (U8) i;
3665 /* Most locales these days are supersets of ASCII. When debugging
3666 * with -DLv, it is helpful to know what the exceptions to that are
3668 if (DEBUG_Lv_TEST) {
3669 bool unexpected = FALSE;
3671 if (isUPPER_L1(i)) {
3673 if (PL_fold_locale[i] != toLOWER_A(i)) {
3677 else if (has_non_ascii_fold) {
3678 if (PL_fold_locale[i] != toLOWER_L1(i)) {
3682 else if (PL_fold_locale[i] != i) {
3686 else if ( isLOWER_L1(i)
3687 && i != LATIN_SMALL_LETTER_SHARP_S
3691 if (PL_fold_locale[i] != toUPPER_A(i)) {
3695 else if (has_non_ascii_fold) {
3696 if (PL_fold_locale[i] != toUPPER_LATIN1_MOD(i)) {
3700 else if (PL_fold_locale[i] != i) {
3704 else if (PL_fold_locale[i] != i) {
3709 found_unexpected = TRUE;
3710 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3711 "For %s, fold of %02x is %02x\n",
3712 newctype, i, PL_fold_locale[i]));
3717 if (found_unexpected) {
3718 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3719 "All bytes not mentioned above either fold to"
3720 " themselves or are the expected ASCII or"
3724 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
3725 "No nonstandard folds were found\n"));
3733 /* We only handle single-byte locales (outside of UTF-8 ones); so if this
3734 * locale requires more than one byte, there are going to be BIG problems.
3737 const int mb_cur_max = MB_CUR_MAX;
3739 if (mb_cur_max > 1 && ! PL_in_utf8_CTYPE_locale
3741 /* Some platforms return MB_CUR_MAX > 1 for even the "C" locale.
3742 * Just assume that the implementation for them (plus for POSIX) is
3743 * correct and the > 1 value is spurious. (Since these are
3744 * specially handled to never be considered UTF-8 locales, as long
3745 * as this is the only problem, everything should work fine */
3746 && ! isNAME_C_OR_POSIX(newctype))
3748 DEBUG_L(PerlIO_printf(Perl_debug_log,
3749 "Unsupported, MB_CUR_MAX=%d\n", mb_cur_max));
3751 Perl_ck_warner_d(aTHX_ packWARN(WARN_LOCALE),
3752 "Locale '%s' is unsupported, and may crash the"
3759 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "check_for_problems=%d\n",
3760 check_for_problems));
3762 /* We don't populate the other lists if a UTF-8 locale, but do check that
3763 * everything works as expected, unless checking turned off */
3764 if (check_for_problems) {
3765 /* Assume enough space for every character being bad. 4 spaces each
3766 * for the 94 printable characters that are output like "'x' "; and 5
3767 * spaces each for "'\\' ", "'\t' ", and "'\n' "; plus a terminating
3769 char bad_chars_list[ (94 * 4) + (3 * 5) + 1 ] = { '\0' };
3770 unsigned int bad_count = 0; /* Count of bad characters */
3772 for (unsigned i = 0; i < 256; i++) {
3774 /* If checking for locale problems, see if the native ASCII-range
3775 * printables plus \n and \t are in their expected categories in
3776 * the new locale. If not, this could mean big trouble, upending
3777 * Perl's and most programs' assumptions, like having a
3778 * metacharacter with special meaning become a \w. Fortunately,
3779 * it's very rare to find locales that aren't supersets of ASCII
3780 * nowadays. It isn't a problem for most controls to be changed
3781 * into something else; we check only \n and \t, though perhaps \r
3782 * could be an issue as well. */
3783 if (isGRAPH_A(i) || isBLANK_A(i) || i == '\n') {
3784 bool is_bad = FALSE;
3785 char name[4] = { '\0' };
3787 /* Convert the name into a string */
3792 else if (i == '\n') {
3793 my_strlcpy(name, "\\n", sizeof(name));
3795 else if (i == '\t') {
3796 my_strlcpy(name, "\\t", sizeof(name));
3800 my_strlcpy(name, "' '", sizeof(name));
3803 /* Check each possibe class */
3804 if (UNLIKELY(cBOOL(isU8_ALPHANUMERIC_LC(i)) !=
3805 cBOOL(isALPHANUMERIC_A(i))))
3808 DEBUG_L(PerlIO_printf(Perl_debug_log,
3809 "isalnum('%s') unexpectedly is %x\n",
3810 name, cBOOL(isU8_ALPHANUMERIC_LC(i))));
3812 if (UNLIKELY(cBOOL(isU8_ALPHA_LC(i)) != cBOOL(isALPHA_A(i)))) {
3814 DEBUG_L(PerlIO_printf(Perl_debug_log,
3815 "isalpha('%s') unexpectedly is %x\n",
3816 name, cBOOL(isU8_ALPHA_LC(i))));
3818 if (UNLIKELY(cBOOL(isU8_DIGIT_LC(i)) != cBOOL(isDIGIT_A(i)))) {
3820 DEBUG_L(PerlIO_printf(Perl_debug_log,
3821 "isdigit('%s') unexpectedly is %x\n",
3822 name, cBOOL(isU8_DIGIT_LC(i))));
3824 if (UNLIKELY(cBOOL(isU8_GRAPH_LC(i)) != cBOOL(isGRAPH_A(i)))) {
3826 DEBUG_L(PerlIO_printf(Perl_debug_log,
3827 "isgraph('%s') unexpectedly is %x\n",
3828 name, cBOOL(isU8_GRAPH_LC(i))));
3830 if (UNLIKELY(cBOOL(isU8_LOWER_LC(i)) != cBOOL(isLOWER_A(i)))) {
3832 DEBUG_L(PerlIO_printf(Perl_debug_log,
3833 "islower('%s') unexpectedly is %x\n",
3834 name, cBOOL(isU8_LOWER_LC(i))));
3836 if (UNLIKELY(cBOOL(isU8_PRINT_LC(i)) != cBOOL(isPRINT_A(i)))) {
3838 DEBUG_L(PerlIO_printf(Perl_debug_log,
3839 "isprint('%s') unexpectedly is %x\n",
3840 name, cBOOL(isU8_PRINT_LC(i))));
3842 if (UNLIKELY(cBOOL(isU8_PUNCT_LC(i)) != cBOOL(isPUNCT_A(i)))) {
3844 DEBUG_L(PerlIO_printf(Perl_debug_log,
3845 "ispunct('%s') unexpectedly is %x\n",
3846 name, cBOOL(isU8_PUNCT_LC(i))));
3848 if (UNLIKELY(cBOOL(isU8_SPACE_LC(i)) != cBOOL(isSPACE_A(i)))) {
3850 DEBUG_L(PerlIO_printf(Perl_debug_log,
3851 "isspace('%s') unexpectedly is %x\n",
3852 name, cBOOL(isU8_SPACE_LC(i))));
3854 if (UNLIKELY(cBOOL(isU8_UPPER_LC(i)) != cBOOL(isUPPER_A(i)))) {
3856 DEBUG_L(PerlIO_printf(Perl_debug_log,
3857 "isupper('%s') unexpectedly is %x\n",
3858 name, cBOOL(isU8_UPPER_LC(i))));
3860 if (UNLIKELY(cBOOL(isU8_XDIGIT_LC(i))!= cBOOL(isXDIGIT_A(i)))) {
3862 DEBUG_L(PerlIO_printf(Perl_debug_log,
3863 "isxdigit('%s') unexpectedly is %x\n",
3864 name, cBOOL(isU8_XDIGIT_LC(i))));
3866 if (UNLIKELY(toU8_LOWER_LC(i) != (int) toLOWER_A(i))) {
3868 DEBUG_L(PerlIO_printf(Perl_debug_log,
3869 "tolower('%s')=0x%x instead of the expected 0x%x\n",
3870 name, toU8_LOWER_LC(i), (int) toLOWER_A(i)));
3872 if (UNLIKELY(toU8_UPPER_LC(i) != (int) toUPPER_A(i))) {
3874 DEBUG_L(PerlIO_printf(Perl_debug_log,
3875 "toupper('%s')=0x%x instead of the expected 0x%x\n",
3876 name, toU8_UPPER_LC(i), (int) toUPPER_A(i)));
3878 if (UNLIKELY((i == '\n' && ! isCNTRL_LC(i)))) {
3880 DEBUG_L(PerlIO_printf(Perl_debug_log,
3881 "'\\n' (=%02X) is not a control\n", (int) i));
3884 /* Add to the list; Separate multiple entries with a blank */
3887 my_strlcat(bad_chars_list, " ", sizeof(bad_chars_list));
3889 my_strlcat(bad_chars_list, name, sizeof(bad_chars_list));
3895 if (bad_count == 2 && maybe_utf8_turkic) {
3897 *bad_chars_list = '\0';
3899 /* The casts are because otherwise some compilers warn:
3900 gcc.gnu.org/bugzilla/show_bug.cgi?id=99950
3901 gcc.gnu.org/bugzilla/show_bug.cgi?id=94182
3903 PL_fold_locale[ (U8) 'I' ] = 'I';
3904 PL_fold_locale[ (U8) 'i' ] = 'i';
3905 PL_in_utf8_turkic_locale = TRUE;
3906 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s is turkic\n", newctype));
3909 /* If we found problems and we want them output, do so */
3910 if ( (UNLIKELY(bad_count))
3911 && (LIKELY(ckWARN_d(WARN_LOCALE)) || UNLIKELY(DEBUG_L_TEST)))
3913 /* WARNING. If you change the wording of these; be sure to update
3914 * t/loc_tools.pl correspondingly */
3916 if (PL_in_utf8_CTYPE_locale) {
3917 PL_warn_locale = Perl_newSVpvf(aTHX_
3918 "Locale '%s' contains (at least) the following characters"
3919 " which have\nunexpected meanings: %s\nThe Perl program"
3920 " will use the expected meanings",
3921 newctype, bad_chars_list);
3926 "\nThe following characters (and maybe"
3927 " others) may not have the same meaning as"
3928 " the Perl program expects: %s\n",
3933 # if defined(HAS_SOME_LANGINFO) || defined(WIN32)
3935 char * scratch_buffer = NULL;
3936 Perl_sv_catpvf(aTHX_ PL_warn_locale, "; codeset=%s",
3937 my_langinfo_c(CODESET, LC_CTYPE,
3939 &scratch_buffer, NULL,
3941 Safefree(scratch_buffer);
3945 Perl_sv_catpvf(aTHX_ PL_warn_locale, "\n");
3947 /* If we are actually in the scope of the locale or are debugging,
3948 * output the message now. If not in that scope, we save the
3949 * message to be output at the first operation using this locale,
3950 * if that actually happens. Most programs don't use locales, so
3951 * they are immune to bad ones. */
3952 if (IN_LC(LC_CTYPE) || UNLIKELY(DEBUG_L_TEST)) {
3954 /* The '0' below suppresses a bogus gcc compiler warning */
3955 Perl_warner(aTHX_ packWARN(WARN_LOCALE), SvPVX(PL_warn_locale),
3957 if (IN_LC(LC_CTYPE)) {
3958 SvREFCNT_dec_NN(PL_warn_locale);
3959 PL_warn_locale = NULL;
3967 Perl_warn_problematic_locale()
3971 /* Core-only function that outputs the message in PL_warn_locale,
3972 * and then NULLS it. Should be called only through the macro
3973 * CHECK_AND_WARN_PROBLEMATIC_LOCALE_ */
3975 if (PL_warn_locale) {
3976 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
3977 SvPVX(PL_warn_locale),
3978 0 /* dummy to avoid compiler warning */ );
3979 SvREFCNT_dec_NN(PL_warn_locale);
3980 PL_warn_locale = NULL;
3984 # endif /* USE_LOCALE_CTYPE */
3987 S_new_LC_ALL(pTHX_ const char *lc_all, bool force)
3989 PERL_ARGS_ASSERT_NEW_LC_ALL;
3991 /* new_LC_ALL() updates all the things we care about. Note that this is
3992 * called just after a change, so uses the actual underlying locale just
3993 * set, and not the nominal one (should they differ, as they may in
3996 const char * individ_locales[LC_ALL_INDEX_] = { NULL };
3998 switch (parse_LC_ALL_string(lc_all,
4000 override_if_ignored, /* Override any ignored
4002 true, /* Always fill array */
4003 true, /* Panic if fails, as to get here it
4004 earlier had to have succeeded */
4009 case only_element_0:
4010 locale_panic_("Unexpected return from parse_LC_ALL_string");
4016 for_all_individual_category_indexes(i) {
4017 if (update_functions[i]) {
4018 const char * this_locale = individ_locales[i];
4019 update_functions[i](aTHX_ this_locale, force);
4022 Safefree(individ_locales[i]);
4026 # ifdef USE_LOCALE_COLLATE
4029 S_new_collate(pTHX_ const char *newcoll, bool force)
4031 PERL_ARGS_ASSERT_NEW_COLLATE;
4032 PERL_UNUSED_ARG(force);
4034 /* Called after each libc setlocale() call affecting LC_COLLATE, to tell
4035 * core Perl this and that 'newcoll' is the name of the new locale.
4037 * The design of locale collation is that every locale change is given an
4038 * index 'PL_collation_ix'. The first time a string participates in an
4039 * operation that requires collation while locale collation is active, it
4040 * is given PERL_MAGIC_collxfrm magic (via sv_collxfrm_flags()). That
4041 * magic includes the collation index, and the transformation of the string
4042 * by strxfrm(), q.v. That transformation is used when doing comparisons,
4043 * instead of the string itself. If a string changes, the magic is
4044 * cleared. The next time the locale changes, the index is incremented,
4045 * and so we know during a comparison that the transformation is not
4046 * necessarily still valid, and so is recomputed. Note that if the locale
4047 * changes enough times, the index could wrap, and it is possible that a
4048 * transformation would improperly be considered valid, leading to an
4049 * unlikely bug. The value is declared to the widest possible type on this
4052 /* Return if the locale isn't changing */
4053 if (strEQ(PL_collation_name, newcoll)) {
4057 Safefree(PL_collation_name);
4058 PL_collation_name = savepv(newcoll);
4061 /* Set the new one up if trivial. Since this is called at process
4062 * initialization, be aware that this bit can't rely on much being
4064 PL_collation_standard = isNAME_C_OR_POSIX(newcoll);
4065 if (PL_collation_standard) {
4066 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4067 "Setting PL_collation name='%s'\n",
4068 PL_collation_name));
4069 PL_collxfrm_base = 0;
4070 PL_collxfrm_mult = 2;
4071 PL_in_utf8_COLLATE_locale = FALSE;
4072 PL_strxfrm_NUL_replacement = '\0';
4073 PL_strxfrm_max_cp = 0;
4077 /* Flag that the remainder of the set up is being deferred until first
4079 PL_collxfrm_mult = 0;
4080 PL_collxfrm_base = 0;
4084 # endif /* USE_LOCALE_COLLATE */
4089 S_Win_byte_string_to_wstring(const UINT code_page, const char * byte_string)
4091 /* Caller must arrange to free the returned string */
4093 int req_size = MultiByteToWideChar(code_page, 0, byte_string, -1, NULL, 0);
4100 Newx(wstring, req_size, wchar_t);
4102 if (! MultiByteToWideChar(code_page, 0, byte_string, -1, wstring, req_size))
4112 # define Win_utf8_string_to_wstring(s) \
4113 Win_byte_string_to_wstring(CP_UTF8, (s))
4116 S_Win_wstring_to_byte_string(const UINT code_page, const wchar_t * wstring)
4118 /* Caller must arrange to free the returned string */
4121 WideCharToMultiByte(code_page, 0, wstring, -1, NULL, 0, NULL, NULL);
4124 Newx(byte_string, req_size, char);
4126 if (! WideCharToMultiByte(code_page, 0, wstring, -1, byte_string,
4127 req_size, NULL, NULL))
4129 Safefree(byte_string);
4137 # define Win_wstring_to_utf8_string(ws) \
4138 Win_wstring_to_byte_string(CP_UTF8, (ws))
4141 S_wrap_wsetlocale(pTHX_ const int category, const char *locale)
4143 PERL_ARGS_ASSERT_WRAP_WSETLOCALE;
4145 /* Calls _wsetlocale(), converting the parameters/return to/from
4146 * Perl-expected forms as if plain setlocale() were being called instead.
4148 * Caller must arrange for the returned PV to be freed.
4151 const wchar_t * wlocale = NULL;
4154 wlocale = Win_utf8_string_to_wstring(locale);
4161 const wchar_t * wresult = _wsetlocale(category, wlocale);
4169 const char * result = Win_wstring_to_utf8_string(wresult);
4177 S_win32_setlocale(pTHX_ int category, const char* locale)
4179 /* This, for Windows, emulates POSIX setlocale() behavior. There is no
4180 * difference between the two unless the input locale is "", which normally
4181 * means on Windows to get the machine default, which is set via the
4182 * computer's "Regional and Language Options" (or its current equivalent).
4183 * In POSIX, it instead means to find the locale from the user's
4184 * environment. This routine changes the Windows behavior to try the POSIX
4185 * behavior first. Further details are in the called function
4186 * find_locale_from_environment().
4189 if (locale != NULL && strEQ(locale, "")) {
4190 /* Note this function may change the locale, but that's ok because we
4191 * are about to change it anyway */
4192 locale = find_locale_from_environment(get_category_index(category));
4193 if (locale == NULL) {
4199 const char * result = wrap_wsetlocale(category, locale);
4200 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n",
4201 setlocale_debug_string_r(category, locale, result)));
4208 save_to_buffer(result, &PL_setlocale_buf, &PL_setlocale_bufsize);
4210 # ifndef USE_PL_CUR_LC_ALL
4216 /* Here, we need to keep track of LC_ALL, so store the new value. but if
4217 * the input locale is NULL, we were just querying, so the original value
4219 if (locale == NULL) {
4224 /* If we set LC_ALL directly above, we already know its new value; but
4225 * if we changed just an individual category, find the new LC_ALL */
4226 if (category != LC_ALL) {
4228 result = wrap_wsetlocale(LC_ALL, NULL);
4231 Safefree(PL_cur_LC_ALL);
4232 PL_cur_LC_ALL = result;
4235 DEBUG_L(PerlIO_printf(Perl_debug_log, "new PL_cur_LC_ALL=%s\n",
4239 return PL_setlocale_buf;
4245 S_native_querylocale_i(pTHX_ const locale_category_index cat_index)
4247 /* Determine the current locale and return it in the form the platform's
4248 * native locale handling understands. This is different only from our
4249 * internal form for the LC_ALL category, as platforms differ in how they
4252 * This is only called from Perl_setlocale(). As such it returns in
4253 * PL_setlocale_buf */
4255 # ifdef USE_LOCALE_NUMERIC
4257 /* We have the LC_NUMERIC name saved, because we are normally switched into
4258 * the C locale (or equivalent) for it. */
4259 if (cat_index == LC_NUMERIC_INDEX_) {
4261 /* We don't have to copy this return value, as it is a per-thread
4262 * variable, and won't change until a future setlocale */
4263 return PL_numeric_name;
4269 if (cat_index != LC_ALL_INDEX_)
4274 /* Here, not LC_ALL, and not LC_NUMERIC: the actual and native values
4277 # ifdef setlocale_i /* Can shortcut if this is defined */
4279 return setlocale_i(cat_index, NULL);
4283 return save_to_buffer(querylocale_i(cat_index),
4284 &PL_setlocale_buf, &PL_setlocale_bufsize);
4289 /* Below, querying LC_ALL */
4292 # ifdef USE_PL_CURLOCALES
4293 # define LC_ALL_ARG PL_curlocales
4295 # define LC_ALL_ARG NULL /* Causes calculate_LC_ALL_string() to find the
4296 locale using a querylocale function */
4299 return calculate_LC_ALL_string(LC_ALL_ARG, EXTERNAL_FORMAT_FOR_QUERY,
4300 WANT_PL_setlocale_buf,
4303 # endif /* has LC_ALL */
4307 #endif /* USE_LOCALE */
4310 =for apidoc Perl_setlocale
4312 This is an (almost) drop-in replacement for the system L<C<setlocale(3)>>,
4313 taking the same parameters, and returning the same information, except that it
4314 returns the correct underlying C<LC_NUMERIC> locale. Regular C<setlocale> will
4315 instead return C<C> if the underlying locale has a non-dot decimal point
4316 character, or a non-empty thousands separator for displaying floating point
4317 numbers. This is because perl keeps that locale category such that it has a
4318 dot and empty separator, changing the locale briefly during the operations
4319 where the underlying one is required. C<Perl_setlocale> knows about this, and
4320 compensates; regular C<setlocale> doesn't.
4322 Another reason it isn't completely a drop-in replacement is that it is
4323 declared to return S<C<const char *>>, whereas the system setlocale omits the
4324 C<const> (presumably because its API was specified long ago, and can't be
4325 updated; it is illegal to change the information C<setlocale> returns; doing
4326 so leads to segfaults.)
4328 Finally, C<Perl_setlocale> works under all circumstances, whereas plain
4329 C<setlocale> can be completely ineffective on some platforms under some
4332 Changing the locale is not a good idea when more than one thread is running,
4333 except on systems where the predefined variable C<${^SAFE_LOCALES}> is
4334 non-zero. This is because on such systems the locale is global to the whole
4335 process and not local to just the thread calling the function. So changing it
4336 in one thread instantaneously changes it in all. On some such systems, the
4337 system C<setlocale()> is ineffective, returning the wrong information, and
4338 failing to actually change the locale. z/OS refuses to try to change the
4339 locale once a second thread is created. C<Perl_setlocale>, should give you
4340 accurate results of what actually happened on these problematic platforms,
4341 returning NULL if the system forbade the locale change.
4343 The return points to a per-thread static buffer, which is overwritten the next
4344 time C<Perl_setlocale> is called from the same thread.
4351 Perl_setlocale(const int category, const char * locale)
4353 /* This wraps POSIX::setlocale() */
4357 PERL_UNUSED_ARG(category);
4358 PERL_UNUSED_ARG(locale);
4366 DEBUG_L(PerlIO_printf(Perl_debug_log,
4367 "Entering Perl_setlocale(%d, \"%s\")\n",
4370 bool valid_category;
4371 locale_category_index cat_index = get_category_index_helper(category,
4374 if (! valid_category) {
4375 if (ckWARN(WARN_LOCALE)) {
4376 const char * conditional_warn_text;
4377 if (locale == NULL) {
4378 conditional_warn_text = "";
4382 conditional_warn_text = "; can't set it to ";
4385 /* diag_listed_as: Unknown locale category %d; can't set it to %s */
4387 packWARN(WARN_LOCALE),
4388 "Unknown locale category %d%s%s",
4389 category, conditional_warn_text, locale);
4398 /* setlocale_i() gets defined only on Configurations that use setlocale()
4399 * in a simple manner that adequately handles all cases. If this category
4400 * doesn't have any perl complications, just do that. */
4401 if (! update_functions[cat_index]) {
4402 return setlocale_i(cat_index, locale);
4407 /* Get current locale */
4408 const char * current_locale = native_querylocale_i(cat_index);
4410 /* A NULL locale means only query what the current one is. */
4411 if (locale == NULL) {
4412 return current_locale;
4415 if (strEQ(current_locale, locale)) {
4416 DEBUG_L(PerlIO_printf(Perl_debug_log,
4417 "Already in requested locale: no action taken\n"));
4418 return current_locale;
4421 /* Here, an actual change is being requested. Do it */
4422 if (! bool_setlocale_i(cat_index, locale)) {
4423 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n",
4424 setlocale_debug_string_i(cat_index, locale, "NULL")));
4428 /* At this point, the locale has been changed based on the requested value,
4429 * and the querylocale_i() will return the actual new value that the system
4430 * has for the category. That may not be the same as the input, as libc
4431 * may have returned a synonymous locale name instead of the input one; or,
4432 * if there are locale categories that we are compiled to ignore, any
4433 * attempt to change them away from "C" is overruled */
4434 current_locale = querylocale_i(cat_index);
4436 /* But certain categories need further work. For example we may need to
4437 * calculate new folding or collation rules. And for LC_NUMERIC, we have
4438 * to switch into a locale that has a dot radix. */
4439 if (update_functions[cat_index]) {
4440 update_functions[cat_index](aTHX_ current_locale,
4441 /* No need to force recalculation, as
4442 * aren't coming from a situation
4443 * where Perl hasn't been controlling
4444 * the locale, so has accurate
4449 /* Make sure the result is in a stable buffer for the caller's use, and is
4450 * in the expected format */
4451 current_locale = native_querylocale_i(cat_index);
4453 DEBUG_L(PerlIO_printf(Perl_debug_log, "returning '%s'\n", current_locale));
4455 return current_locale;
4461 #if defined(USE_LOCALE) || defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV)
4464 S_get_locale_string_utf8ness_i(pTHX_ const char * string,
4465 const locale_utf8ness_t known_utf8,
4466 const char * locale,
4467 const locale_category_index cat_index)
4469 PERL_ARGS_ASSERT_GET_LOCALE_STRING_UTF8NESS_I;
4474 PERL_UNUSED_ARG(string);
4475 PERL_UNUSED_ARG(known_utf8);
4476 PERL_UNUSED_ARG(locale);
4477 PERL_UNUSED_ARG(cat_index);
4481 assert(cat_index <= LC_ALL_INDEX_);
4483 /* Return to indicate if 'string' in the locale given by the input
4484 * arguments should be considered UTF-8 or not.
4486 * If the input 'locale' is not NULL, use that for the locale; otherwise
4487 * use the current locale for the category specified by 'cat_index'.
4490 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4491 "Entering get_locale_string_utf8ness_i; locale=%s,"
4492 " index=%u(%s), string=%s, known_utf8=%d\n",
4493 locale, cat_index, category_names[cat_index],
4495 ? _byte_dump_string((U8 *) string,
4500 if (string == NULL) {
4501 return UTF8NESS_IMMATERIAL;
4504 if (IN_BYTES) { /* respect 'use bytes' */
4508 Size_t len = strlen(string);
4510 /* UTF8ness is immaterial if the representation doesn't vary */
4511 const U8 * first_variant = NULL;
4512 if (is_utf8_invariant_string_loc((U8 *) string, len, &first_variant)) {
4513 return UTF8NESS_IMMATERIAL;
4516 /* Can't be UTF-8 if invalid */
4517 if (! is_utf8_string((U8 *) first_variant,
4518 len - ((char *) first_variant - string)))
4523 /* Here and below, we know the string is legal UTF-8, containing at least
4524 * one character requiring a sequence of two or more bytes. It is quite
4525 * likely to be UTF-8. But it pays to be paranoid and do further checking.
4527 * If we already know the UTF-8ness of the locale, then we immediately know
4528 * what the string is */
4529 if (UNLIKELY(known_utf8 != LOCALE_UTF8NESS_UNKNOWN)) {
4530 return (known_utf8 == LOCALE_IS_UTF8) ? UTF8NESS_YES : UTF8NESS_NO;
4533 if (locale == NULL) {
4534 locale = querylocale_i(cat_index);
4537 /* If the locale is UTF-8, the string is UTF-8; otherwise it was
4538 * coincidental that the string is legal UTF-8
4540 * However, if the perl is compiled to not pay attention to the category
4541 * being passed in, you might think that that locale is essentially always
4542 * the C locale, so it would make sense to say it isn't UTF-8. But to get
4543 * here, the string has to contain characters unknown in the C locale. And
4544 * in fact, Windows boxes are compiled without LC_MESSAGES, as their
4545 * message catalog isn't really a part of the locale system. But those
4546 * messages really could be UTF-8, and given that the odds are rather small
4547 * of something not being UTF-8 but being syntactically valid UTF-8, khw
4548 * has decided to call such strings as UTF-8. */
4549 return (is_locale_utf8(locale)) ? UTF8NESS_YES : UTF8NESS_NO;
4556 S_is_locale_utf8(pTHX_ const char * locale)
4558 PERL_ARGS_ASSERT_IS_LOCALE_UTF8;
4560 /* Returns TRUE if the locale 'locale' is UTF-8; FALSE otherwise. */
4562 # if ! defined(USE_LOCALE) \
4563 || ! defined(USE_LOCALE_CTYPE) \
4564 || defined(EBCDIC) /* There aren't any real UTF-8 locales at this time */
4566 PERL_UNUSED_ARG(locale);
4570 /* Definitively, can't be UTF-8 */
4571 # define HAS_DEFINITIVE_UTF8NESS_DETERMINATION
4574 /* If the input happens to be the same locale as we are currently setup
4575 * for, the answer has already been cached. */
4576 if (strEQ(locale, PL_ctype_name)) {
4577 return PL_in_utf8_CTYPE_locale;
4580 if (isNAME_C_OR_POSIX(locale)) {
4584 # if ! defined(HAS_SOME_LANGINFO) && ! defined(WIN32)
4586 /* On non-Windows without nl_langinfo(), we have to do some digging to get
4587 * the answer. First, toggle to the desired locale so can query its state
4589 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
4591 # define TEARDOWN_FOR_IS_LOCALE_UTF8 \
4592 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale)
4596 /* If there are fewer bytes available in this locale than are required
4597 * to represent the largest legal UTF-8 code point, this isn't a UTF-8
4599 const int mb_cur_max = MB_CUR_MAX;
4600 if (mb_cur_max < (int) UNISKIP(PERL_UNICODE_MAX)) {
4601 TEARDOWN_FOR_IS_LOCALE_UTF8;
4606 # if defined(HAS_MBTOWC) || defined(HAS_MBRTOWC)
4608 /* With these functions, we can definitively determine a locale's
4610 # define HAS_DEFINITIVE_UTF8NESS_DETERMINATION
4612 /* If libc mbtowc() evaluates the bytes that form the REPLACEMENT CHARACTER
4613 * as that Unicode code point, this has to be a UTF-8 locale; otherwise it
4616 (void) Perl_mbtowc_(aTHX_ NULL, NULL, 0);/* Reset shift state */
4617 int mbtowc_ret = Perl_mbtowc_(aTHX_ &wc,
4618 STR_WITH_LEN(REPLACEMENT_CHARACTER_UTF8));
4619 TEARDOWN_FOR_IS_LOCALE_UTF8;
4620 return ( mbtowc_ret == STRLENs(REPLACEMENT_CHARACTER_UTF8)
4621 && wc == UNICODE_REPLACEMENT);
4625 /* If the above two C99 functions aren't working, you could try some
4626 * different methods. It seems likely that the obvious choices,
4627 * wctomb() and wcrtomb(), wouldn't be working either. But you could
4628 * choose one of the dozen-ish Unicode titlecase triples and verify
4629 * that towupper/towlower work as expected.
4631 * But, our emulation of nl_langinfo() works quite well, so avoid the
4632 * extra code until forced to by some weird non-conforming platform. */
4633 # define USE_LANGINFO_FOR_UTF8NESS
4634 # undef HAS_DEFINITIVE_UTF8NESS_DETERMINATION
4638 /* On Windows or on platforms with nl_langinfo(), there is a direct way to
4639 * get the locale's codeset, which will be some form of 'UTF-8' for a
4640 * UTF-8 locale. my_langinfo_i() handles this, and we will call that
4642 # define HAS_DEFINITIVE_UTF8NESS_DETERMINATION
4643 # define USE_LANGINFO_FOR_UTF8NESS
4644 # define TEARDOWN_FOR_IS_LOCALE_UTF8
4645 # endif /* USE_LANGINFO_FOR_UTF8NESS */
4647 /* If the above compiled into code, it found the locale's UTF-8ness,
4648 * nothing more to do; if it didn't get compiled,
4649 * USE_LANGINFO_FOR_UTF8NESS is defined. There are two possible reasons:
4650 * 1) it is the preferred method because it knows directly for sure
4651 * what the codeset is because the platform has libc functions that
4653 * 2) the functions the above code section would compile to use don't
4654 * exist or are unreliable on this platform; we are less sure of the
4655 * my_langinfo() result, though it is very unlikely to be wrong
4656 * about if it is UTF-8 or not */
4657 # ifdef USE_LANGINFO_FOR_UTF8NESS
4659 char * scratch_buffer = NULL;
4660 const char * codeset = my_langinfo_c(CODESET, LC_CTYPE, locale,
4661 &scratch_buffer, NULL, NULL);
4662 bool retval = is_codeset_name_UTF8(codeset);
4664 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4665 "found codeset=%s, is_utf8=%d\n", codeset, retval));
4667 Safefree(scratch_buffer);
4669 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "is_locale_utf8(%s) returning %d\n",
4671 TEARDOWN_FOR_IS_LOCALE_UTF8;
4675 # endif /* End of the #else clause, for the non-trivial case */
4683 S_set_save_buffer_min_size(pTHX_ Size_t min_len,
4685 Size_t * buf_cursize)
4687 /* Make sure the buffer pointed to by *buf is at least as large 'min_len';
4688 * *buf_cursize is the size of 'buf' upon entry; it will be updated to the
4689 * new size on exit. 'buf_cursize' being NULL is to be used when this is a
4690 * single use buffer, which will shortly be freed by the caller. */
4692 if (buf_cursize == NULL) {
4693 Newx(*buf, min_len, char);
4695 else if (*buf_cursize == 0) {
4696 Newx(*buf, min_len, char);
4697 *buf_cursize = min_len;
4699 else if (min_len > *buf_cursize) {
4700 Renew(*buf, min_len, char);
4701 *buf_cursize = min_len;
4706 S_save_to_buffer(pTHX_ const char * string, char **buf, Size_t *buf_size)
4708 PERL_ARGS_ASSERT_SAVE_TO_BUFFER;
4710 /* Copy the NUL-terminated 'string' to a buffer whose address before this
4711 * call began at *buf, and whose available length before this call was
4714 * If the length of 'string' is greater than the space available, the
4715 * buffer is grown accordingly, which may mean that it gets relocated.
4716 * *buf and *buf_size will be updated to reflect this.
4718 * Regardless, the function returns a pointer to where 'string' is now
4721 * 'string' may be NULL, which means no action gets taken, and NULL is
4724 * 'buf_size' being NULL is to be used when this is a single use buffer,
4725 * which will shortly be freed by the caller.
4727 * If *buf or 'buf_size' are NULL or *buf_size is 0, the buffer is assumed
4728 * empty, and memory is malloc'd.
4735 /* No-op to copy over oneself */
4736 if (string == *buf) {
4740 Size_t string_size = strlen(string) + 1;
4741 set_save_buffer_min_size(string_size, buf, buf_size);
4745 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
4746 "Copying '%s' to %p\n",
4747 ((is_utf8_string((U8 *) string, 0))
4749 :_byte_dump_string((U8 *) string, strlen(string), 0)),
4752 /* Catch glitches. Usually this is because LC_CTYPE needs to be the same
4753 * locale as whatever is being worked on */
4754 if (UNLIKELY(instr(string, REPLACEMENT_CHARACTER_UTF8))) {
4755 locale_panic_(Perl_form(aTHX_
4756 "Unexpected REPLACEMENT_CHARACTER in '%s'\n%s",
4757 string, get_LC_ALL_display()));
4762 Copy(string, *buf, string_size, char);
4769 Perl_get_win32_message_utf8ness(pTHX_ const char * string)
4771 /* This is because Windows doesn't have LC_MESSAGES. */
4773 # ifdef USE_LOCALE_CTYPE
4775 /* We don't know the locale utf8ness here, and not even the locale itself.
4776 * Since Windows uses a different mechanism to specify message language
4777 * output than the locale system, it is going to be problematic deciding
4778 * if we are to store it as UTF-8 or not. By specifying LOCALE_IS_UTF8, we
4779 * are telling the called function to return true iff the string has
4780 * non-ASCII characters in it that are all syntactically UTF-8. We are
4781 * thus relying on the fact that a string that is syntactically valid UTF-8
4782 * is likely to be UTF-8. Should this ever cause problems, this function
4783 * could be replaced by something more Windows-specific */
4784 return get_locale_string_utf8ness_i(string, LOCALE_IS_UTF8,
4785 NULL, LC_CTYPE_INDEX_);
4788 PERL_UNUSED_ARG(string);
4796 #endif /* USE_LOCALE */
4799 Perl_mbtowc_(pTHX_ const wchar_t * pwc, const char * s, const Size_t len)
4802 #if ! defined(HAS_MBRTOWC) && ! defined(HAS_MBTOWC)
4804 PERL_UNUSED_ARG(pwc);
4806 PERL_UNUSED_ARG(len);
4809 #else /* Below we have some form of mbtowc() */
4810 # if defined(HAS_MBRTOWC) \
4811 && (defined(USE_LOCALE_THREADS) || ! defined(HAS_MBTOWC))
4812 # define USE_MBRTOWC
4819 if (s == NULL) { /* Initialize the shift state to all zeros in
4822 # if defined(USE_MBRTOWC)
4824 memzero(&PL_mbrtowc_ps, sizeof(PL_mbrtowc_ps));
4831 retval = mbtowc(NULL, NULL, 0);
4839 # if defined(USE_MBRTOWC)
4843 retval = (SSize_t) mbrtowc((wchar_t *) pwc, s, len, &PL_mbrtowc_ps);
4848 /* Locking prevents races, but locales can be switched out without locking,
4849 * so this isn't a cure all */
4852 retval = mbtowc((wchar_t *) pwc, s, len);
4864 =for apidoc Perl_localeconv
4866 This is a thread-safe version of the libc L<localeconv(3)>. It is the same as
4867 L<POSIX::localeconv|POSIX/localeconv> (returning a hash of the C<localeconv()>
4868 fields), but directly callable from XS code.
4874 Perl_localeconv(pTHX)
4877 #if ! defined(HAS_LOCALECONV)
4883 return my_localeconv(0);
4889 #if defined(HAS_LOCALECONV)
4892 S_my_localeconv(pTHX_ const int item)
4894 PERL_ARGS_ASSERT_MY_LOCALECONV;
4896 /* This returns a mortalized hash containing all or certain elements
4897 * returned by localeconv(). */
4898 HV * hv = newHV(); /* The returned hash, initially empty */
4899 sv_2mortal((SV*)hv);
4901 /* The function is used by Perl_localeconv() and POSIX::localeconv(), or
4902 * internally from this file, and is thread-safe.
4904 * localeconv() returns items from two different locale categories,
4905 * LC_MONETARY and LC_NUMERIC. Various data structures in this function
4906 * are arrays with two elements, one for each category, and these indexes
4907 * indicate which array element applies to which category */
4908 #define NUMERIC_OFFSET 0
4909 #define MONETARY_OFFSET 1
4911 /* Some operations apply to one or the other category, or both. A mask
4912 * is used to specify all the possibilities. This macro converts from the
4913 * category offset to its bit position in the mask. */
4914 #define OFFSET_TO_BIT(i) (1 << (i))
4916 /* There are two use cases for this function:
4917 * 1) Called as Perl_localeconv(), or from POSIX::locale_conv(). This
4918 * returns the lconv structure copied to a hash, based on the current
4919 * underlying locales for LC_NUMERIC and LC_MONETARY. An input item==0
4920 * signifies this case, or on many platforms it is the only use case
4922 * 2) Certain items that nl_langinfo() provides are also derivable from
4923 * the return of localeconv(). Windows notably doesn't have
4924 * nl_langinfo(), so on that, and actually any platform lacking it,
4925 * my_localeconv() is used also to emulate it for those particular
4926 * items. The code to do this is compiled only on such platforms.
4927 * Rather than going to the expense of creating a full hash when only
4928 * one item is needed, the returned hash has just the desired item in
4931 * To access all the localeconv() struct lconv fields, there is a data
4932 * structure that contains every commonly documented field in it. (Maybe
4933 * some minority platforms have extra fields. Those could be added here
4934 * without harm; they would just be ignored on platforms lacking them.)
4936 * Our structure is compiled to make looping through the fields easier by
4937 * pointing each name to its value's offset within lconv, e.g.,
4938 { "thousands_sep", STRUCT_OFFSET(struct lconv, thousands_sep) }
4940 # define LCONV_ENTRY(name) \
4941 {STRINGIFY(name), STRUCT_OFFSET(struct lconv, name)}
4943 /* These synonyms are just for clarity, and to make it easier in case
4944 * something needs to change in the future */
4945 # define LCONV_NUMERIC_ENTRY(name) LCONV_ENTRY(name)
4946 # define LCONV_MONETARY_ENTRY(name) LCONV_ENTRY(name)
4948 /* There are just a few fields for NUMERIC strings */
4949 const lconv_offset_t lconv_numeric_strings[] = {
4950 # ifndef NO_LOCALECONV_GROUPING
4951 LCONV_NUMERIC_ENTRY(grouping),
4953 LCONV_NUMERIC_ENTRY(thousands_sep),
4954 # define THOUSANDS_SEP_LITERAL "thousands_sep"
4955 LCONV_NUMERIC_ENTRY(decimal_point),
4956 # define DECIMAL_POINT_LITERAL "decimal_point"
4960 /* When used to implement nl_langinfo(), we save time by only populating
4961 * the hash with the field(s) needed. Thus we would need a data structure
4963 * LCONV_NUMERIC_ENTRY(decimal_point),
4966 * By placing the decimal_point field last in the full structure, we can
4967 * use just the tail for this bit of it, saving space. This macro yields
4968 * the address of the sub structure. */
4969 # define DECIMAL_POINT_ADDRESS \
4970 &lconv_numeric_strings[(C_ARRAY_LENGTH(lconv_numeric_strings) - 2)]
4972 /* And the MONETARY string fields */
4973 const lconv_offset_t lconv_monetary_strings[] = {
4974 LCONV_MONETARY_ENTRY(int_curr_symbol),
4975 LCONV_MONETARY_ENTRY(mon_decimal_point),
4976 # ifndef NO_LOCALECONV_MON_THOUSANDS_SEP
4977 LCONV_MONETARY_ENTRY(mon_thousands_sep),
4979 # ifndef NO_LOCALECONV_MON_GROUPING
4980 LCONV_MONETARY_ENTRY(mon_grouping),
4982 LCONV_MONETARY_ENTRY(positive_sign),
4983 LCONV_MONETARY_ENTRY(negative_sign),
4984 LCONV_MONETARY_ENTRY(currency_symbol),
4985 # define CURRENCY_SYMBOL_LITERAL "currency_symbol"
4989 /* Like above, this field being last can be used as a sub structure */
4990 # define CURRENCY_SYMBOL_ADDRESS \
4991 &lconv_monetary_strings[(C_ARRAY_LENGTH(lconv_monetary_strings) - 2)]
4993 /* Finally there are integer fields, all are for monetary purposes */
4994 const lconv_offset_t lconv_integers[] = {
4995 LCONV_ENTRY(int_frac_digits),
4996 LCONV_ENTRY(frac_digits),
4997 LCONV_ENTRY(p_sep_by_space),
4998 LCONV_ENTRY(n_cs_precedes),
4999 LCONV_ENTRY(n_sep_by_space),
5000 LCONV_ENTRY(p_sign_posn),
5001 LCONV_ENTRY(n_sign_posn),
5002 # ifdef HAS_LC_MONETARY_2008
5003 LCONV_ENTRY(int_p_cs_precedes),
5004 LCONV_ENTRY(int_p_sep_by_space),
5005 LCONV_ENTRY(int_n_cs_precedes),
5006 LCONV_ENTRY(int_n_sep_by_space),
5007 LCONV_ENTRY(int_p_sign_posn),
5008 LCONV_ENTRY(int_n_sign_posn),
5010 # define P_CS_PRECEDES_LITERAL "p_cs_precedes"
5011 LCONV_ENTRY(p_cs_precedes),
5015 /* Like above, this field being last can be used as a sub structure */
5016 # define P_CS_PRECEDES_ADDRESS \
5017 &lconv_integers[(C_ARRAY_LENGTH(lconv_integers) - 2)]
5019 /* The actual populating of the hash is done by two sub functions that get
5020 * passed an array of length two containing the data structure they are
5021 * supposed to use to get the key names to fill the hash with. One element
5022 * is always for the NUMERIC strings (or NULL if none to use), and the
5023 * other element similarly for the MONETARY ones. */
5024 const lconv_offset_t * strings[2] = { lconv_numeric_strings,
5025 lconv_monetary_strings
5028 /* The LC_MONETARY category also has some integer-valued fields, whose
5029 * information is kept in a separate parallel array to 'strings' */
5030 const lconv_offset_t * integers[2] = {
5035 # if ! defined(USE_LOCALE_NUMERIC) && ! defined(USE_LOCALE_MONETARY)
5037 /* If both NUMERIC and MONETARY must be the "C" locale, simply populate the
5038 * hash using the function that works on just that locale. */
5039 populate_hash_from_C_localeconv(hv,
5041 ( OFFSET_TO_BIT(NUMERIC_OFFSET)
5042 | OFFSET_TO_BIT(MONETARY_OFFSET)),
5045 /* We shouldn't get to here for the case of an individual item, as
5046 * preprocessor directives elsewhere in this file should have filled in the
5047 * correct values at a higher level */
5049 PERL_UNUSED_ARG(item);
5055 /* From here to the end of this function, at least one of NUMERIC or
5056 * MONETARY can be non-C */
5058 /* This is a mask, with one bit to tell the populate functions to populate
5059 * the NUMERIC items; another bit for the MONETARY ones. This way they can
5060 * choose which (or both) to populate from */
5063 /* Some platforms, for correct non-mojibake results, require LC_CTYPE's
5064 * locale to match LC_NUMERIC's for the numeric fields, and LC_MONETARY's
5065 * for the monetary ones. What happens if LC_NUMERIC and LC_MONETARY
5066 * aren't compatible? Wrong results. To avoid that, we call localeconv()
5067 * twice, once for each locale, setting LC_CTYPE to match the category.
5068 * But if the locales of both categories are the same, there is no need for
5069 * a second call. Assume this is the case unless overridden below */
5070 bool requires_2nd_localeconv = false;
5072 /* The actual hash populating is done by one of the two populate functions.
5073 * Which one is appropriate for either the MONETARY_OFFSET or the
5074 * NUMERIC_OFFSET is calculated and then stored in this table */
5075 void (*populate[2]) (pTHX_
5079 const lconv_offset_t **,
5080 const lconv_offset_t **);
5082 /* This gives the locale to use for the corresponding OFFSET, like the
5083 * 'populate' array above */
5084 const char * locales[2];
5086 # ifdef HAS_SOME_LANGINFO
5088 /* If the only use-case for this is the full localeconv(), the 'item'
5089 * parameter is ignored. */
5090 PERL_UNUSED_ARG(item);
5092 # else /* This only gets compiled for the use-case of using localeconv()
5093 to emulate nl_langinfo() when missing from the platform. */
5095 # ifdef USE_LOCALE_NUMERIC
5097 /* We need this substructure to only return this field for the THOUSEP
5098 * item. The other items also need substructures, but they were handled
5099 * above by placing the substructure's item at the end of the full one, so
5100 * the data structure could do double duty. However, both this and
5101 * RADIXCHAR would need to be in the final position of the same full
5102 * structure; an impossibility. So make this into a separate structure */
5103 const lconv_offset_t thousands_sep_string[] = {
5104 LCONV_NUMERIC_ENTRY(thousands_sep),
5110 /* End of all the initialization of data structures. Now for actual code.
5112 * Without nl_langinfo(), the call to my_localeconv() could be for all of
5113 * the localeconv() items or for just one of the following 3 items to
5114 * emulate nl_langinfo().
5116 * This is compiled only when using perl_langinfo.h, which we control, and
5117 * it has been constructed so that no item is numbered 0.
5119 * For each individual item, either return the known value if the current
5120 * locale is "C", or set up the appropriate parameters for the call below
5121 * to the populate function */
5127 locale_panic_(Perl_form(aTHX_
5128 "Unexpected item passed to my_localeconv: %d", item));
5131 # ifdef USE_LOCALE_NUMERIC
5134 if (isNAME_C_OR_POSIX(PL_numeric_name)) {
5135 (void) hv_stores(hv, DECIMAL_POINT_LITERAL, newSVpvs("."));
5139 strings[NUMERIC_OFFSET] = DECIMAL_POINT_ADDRESS;
5140 goto numeric_common;
5143 if (isNAME_C_OR_POSIX(PL_numeric_name)) {
5144 (void) hv_stores(hv, THOUSANDS_SEP_LITERAL, newSVpvs(""));
5148 strings[NUMERIC_OFFSET] = thousands_sep_string;
5151 index_bits = OFFSET_TO_BIT(NUMERIC_OFFSET);
5152 locale = PL_numeric_name;
5156 # ifdef USE_LOCALE_MONETARY
5158 case CRNCYSTR: /* This item needs the values for both the currency
5159 symbol, and another one used to construct the
5160 nl_langino()-compatible return. */
5162 locale = querylocale_c(LC_MONETARY);
5163 if (isNAME_C_OR_POSIX(locale)) {
5164 (void) hv_stores(hv, CURRENCY_SYMBOL_LITERAL, newSVpvs(""));
5165 (void) hv_stores(hv, P_CS_PRECEDES_LITERAL, newSViv(-1));
5169 strings[MONETARY_OFFSET] = CURRENCY_SYMBOL_ADDRESS;
5170 integers[MONETARY_OFFSET] = P_CS_PRECEDES_ADDRESS;
5172 index_bits = OFFSET_TO_BIT(MONETARY_OFFSET);
5177 } /* End of switch() */
5179 /* There's only one item, so only one of each of these will get used,
5180 * but cheap to initialize both */
5181 populate[MONETARY_OFFSET] =
5182 populate[NUMERIC_OFFSET] = S_populate_hash_from_localeconv;
5183 locales[MONETARY_OFFSET] = locales[NUMERIC_OFFSET] = locale;
5185 else /* End of for just one item to emulate nl_langinfo() */
5190 /* Here, the call is for all of localeconv(). It has a bunch of
5191 * items. The first function call always gets the MONETARY values */
5192 index_bits = OFFSET_TO_BIT(MONETARY_OFFSET);
5194 # ifdef USE_LOCALE_MONETARY
5196 locales[MONETARY_OFFSET] = querylocale_c(LC_MONETARY);
5197 populate[MONETARY_OFFSET] =
5198 (isNAME_C_OR_POSIX(locales[MONETARY_OFFSET]))
5199 ? S_populate_hash_from_C_localeconv
5200 : S_populate_hash_from_localeconv;
5204 locales[MONETARY_OFFSET] = "C";
5205 populate[MONETARY_OFFSET] = S_populate_hash_from_C_localeconv;
5208 # ifdef USE_LOCALE_NUMERIC
5210 /* And if the locales for the two categories are the same, we can also
5211 * do the NUMERIC values in the same call */
5212 if (strEQ(PL_numeric_name, locales[MONETARY_OFFSET])) {
5213 index_bits |= OFFSET_TO_BIT(NUMERIC_OFFSET);
5214 locales[NUMERIC_OFFSET] = locales[MONETARY_OFFSET];
5215 populate[NUMERIC_OFFSET] = populate[MONETARY_OFFSET];
5218 requires_2nd_localeconv = true;
5219 locales[NUMERIC_OFFSET] = PL_numeric_name;
5220 populate[NUMERIC_OFFSET] = (isNAME_C_OR_POSIX(PL_numeric_name))
5221 ? S_populate_hash_from_C_localeconv
5222 : S_populate_hash_from_localeconv;
5227 /* When LC_NUMERIC is confined to "C", the two locales are the same
5228 iff LC_MONETARY in this case is also "C". We set up the function
5229 for that case above, so fastest to test just its address */
5230 locales[NUMERIC_OFFSET] = "C";
5231 if (populate[MONETARY_OFFSET] == S_populate_hash_from_C_localeconv) {
5232 index_bits |= OFFSET_TO_BIT(NUMERIC_OFFSET);
5233 populate[NUMERIC_OFFSET] = populate[MONETARY_OFFSET];
5236 requires_2nd_localeconv = true;
5237 populate[NUMERIC_OFFSET] = S_populate_hash_from_C_localeconv;
5242 } /* End of call is for localeconv() */
5244 /* Call the proper populate function (which may call localeconv()) and copy
5245 * its results into the hash. All the parameters have been initialized
5247 (*populate[MONETARY_OFFSET])(aTHX_
5248 hv, locales[MONETARY_OFFSET],
5249 index_bits, strings, integers);
5251 # ifndef HAS_SOME_LANGINFO /* Could be using this function to emulate
5254 /* We are done when called with an individual item. There are no integer
5255 * items to adjust, and it's best for the caller to determine if this
5256 * string item is UTF-8 or not. This is because the locale's UTF-8ness is
5257 * calculated below, and in some Configurations, that can lead to a
5258 * recursive call to here, which could recurse infinitely. */
5265 /* The above call may have done all the hash fields, but not always, as
5266 * already explained. If we need a second call it is always for the
5268 if (requires_2nd_localeconv) {
5269 (*populate[NUMERIC_OFFSET])(aTHX_
5271 locales[NUMERIC_OFFSET],
5272 OFFSET_TO_BIT(NUMERIC_OFFSET),
5276 /* Here, the hash has been completely populated.
5278 * Now go through all the items and:
5279 * a) For string items, see if they should be marked as UTF-8 or not.
5280 * This would have been more convenient and faster to do while
5281 * populating the hash in the first place, but that operation has to be
5282 * done within a critical section, keeping other threads from
5283 * executing, so only the minimal amount of work necessary is done at
5285 * b) For integer items, convert the C CHAR_MAX value into -1. Again,
5286 * this could have been done in the critical section, but was deferred
5287 * to here to keep to the bare minimum amount the time spent owning the
5288 * processor. CHAR_MAX is a C concept for an 8-bit character type.
5289 * Perl has no such type; the closest fit is a -1.
5291 * XXX On unthreaded perls, this code could be #ifdef'd out, and the
5292 * corrections determined at hash population time, at an extra maintenance
5293 * cost which khw doesn't think is worth it
5296 for (unsigned int i = 0; i < 2; i++) { /* Try both types of strings */
5298 /* The return from this function is already adjusted */
5299 if (populate[i] == S_populate_hash_from_C_localeconv) {
5303 /* Examine each string */
5304 for (const lconv_offset_t *strp = strings[i]; strp->name; strp++) {
5305 const char * name = strp->name;
5307 /* 'value' will contain the string that may need to be marked as
5309 SV ** value = hv_fetch(hv, name, strlen(name), true);
5310 if (value == NULL) {
5314 /* Determine if the string should be marked as UTF-8. */
5315 if (UTF8NESS_YES == (get_locale_string_utf8ness_i(SvPVX(*value),
5316 LOCALE_UTF8NESS_UNKNOWN,
5318 LC_ALL_INDEX_ /* OOB */)))
5324 if (integers[i] == NULL) {
5328 /* And each integer */
5329 for (const lconv_offset_t *intp = integers[i]; intp->name; intp++) {
5330 const char * name = intp->name;
5332 if (! name) { /* Reached the end */
5336 SV ** value = hv_fetch(hv, name, strlen(name), true);
5341 /* Change CHAR_MAX to -1 */
5342 if (SvIV(*value) == CHAR_MAX) {
5343 sv_setiv(*value, -1);
5350 # endif /* End of must have one or both USE_MONETARY, USE_NUMERIC */
5355 S_populate_hash_from_C_localeconv(pTHX_ HV * hv,
5356 const char * locale, /* Unused */
5358 /* bit mask of which categories to
5360 const U32 which_mask,
5362 /* The string type values to return;
5363 * one element for numeric; the other
5365 const lconv_offset_t * strings[2],
5367 /* And the integer fields */
5368 const lconv_offset_t * integers[2])
5370 PERL_ARGS_ASSERT_POPULATE_HASH_FROM_C_LOCALECONV;
5371 PERL_UNUSED_ARG(locale);
5372 assert(isNAME_C_OR_POSIX(locale));
5374 /* Fill hv with the values that localeconv() is supposed to return for
5377 U32 working_mask = which_mask;
5378 while (working_mask) {
5380 /* Get the bit position of the next lowest set bit. That is the
5381 * index into the 'strings' array of the category we use in this loop
5382 * iteration. Turn the bit off so we don't work on this category
5383 * again in this function call. */
5384 const PERL_UINT_FAST8_T i = lsbit_pos(working_mask);
5385 working_mask &= ~ (1 << i);
5387 /* This category's string fields */
5388 const lconv_offset_t * category_strings = strings[i];
5390 # ifndef HAS_SOME_LANGINFO /* This doesn't work properly if called on a single
5391 item, which could only happen when there isn't
5392 nl_langinfo on the platform */
5393 assert(category_strings[1].name != NULL);
5396 /* All string fields are empty except for one NUMERIC one. That one
5397 * has been initialized to be the final one in the NUMERIC strings, so
5398 * stop the loop early in that case. Otherwise, we would store an
5399 * empty string to the hash, and immediately overwrite it with the
5401 const unsigned int stop_early = (i == NUMERIC_OFFSET) ? 1 : 0;
5403 /* A NULL element terminates the list */
5404 while ((category_strings + stop_early)->name) {
5406 category_strings->name,
5407 strlen(category_strings->name),
5414 /* And fill in the NUMERIC exception */
5415 if (i == NUMERIC_OFFSET) {
5416 (void) hv_stores(hv, "decimal_point", newSVpvs("."));
5420 /* Add any int fields. In the C locale, all are -1 */
5422 const lconv_offset_t * current = integers[i];
5423 while (current->name) {
5425 current->name, strlen(current->name),
5434 # if defined(USE_LOCALE_NUMERIC) || defined(USE_LOCALE_MONETARY)
5437 S_populate_hash_from_localeconv(pTHX_ HV * hv,
5439 /* Switch to this locale to run
5440 * localeconv() from */
5441 const char * locale,
5443 /* bit mask of which categories to
5445 const U32 which_mask,
5447 /* The string type values to return; one
5448 * element for numeric; the other for
5450 const lconv_offset_t * strings[2],
5452 /* And similarly the integer fields */
5453 const lconv_offset_t * integers[2])
5455 PERL_ARGS_ASSERT_POPULATE_HASH_FROM_LOCALECONV;
5456 PERL_UNUSED_ARG(which_mask); /* Some configurations don't use this;
5457 complicated to figure out which */
5459 /* Run localeconv() and copy some or all of its results to the input 'hv'
5460 * hash. Most localeconv() implementations return the values in a global
5461 * static buffer, so the operation must be performed in a critical section,
5462 * ending only after the copy is completed. There are so many locks
5463 * because localeconv() deals with two categories, and returns in a single
5464 * global static buffer. Some locks might be no-ops on this platform, but
5465 * not others. We need to lock if any one isn't a no-op. */
5467 # ifdef WE_MUST_DEAL_WITH_MISMATCHED_CTYPE
5469 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
5472 # ifdef USE_LOCALE_NUMERIC
5474 /* We need to toggle to the underlying NUMERIC locale if we are getting
5475 * NUMERIC strings */
5476 const char * orig_NUMERIC_locale = NULL;
5477 if (which_mask & OFFSET_TO_BIT(NUMERIC_OFFSET)) {
5482 /* There is a bug in Windows in which setting LC_CTYPE after the others
5483 * doesn't actually take effect for localeconv(). See commit
5484 * 418efacd1950763f74ed3cc22f8cf9206661b892 for details. Thus we have
5485 * to make sure that the locale we want is set after LC_CTYPE. We
5486 * unconditionally toggle away from and back to the current locale
5487 * prior to calling localeconv(). */
5488 orig_NUMERIC_locale = toggle_locale_c(LC_NUMERIC, "C");
5489 toggle_locale_c(LC_NUMERIC, locale);
5493 /* No need for the extra toggle when not on Windows */
5494 orig_NUMERIC_locale = toggle_locale_c(LC_NUMERIC, locale);
5501 # if defined(USE_LOCALE_MONETARY) && defined(WIN32)
5503 /* Same Windows bug as described just above for NUMERIC. Otherwise, no
5504 * need to toggle LC_MONETARY, as it is kept in the underlying locale */
5505 const char * orig_MONETARY_locale = NULL;
5506 if (which_mask & OFFSET_TO_BIT(MONETARY_OFFSET)) {
5507 orig_MONETARY_locale = toggle_locale_c(LC_MONETARY, "C");
5508 toggle_locale_c(LC_MONETARY, locale);
5513 /* Finally ready to do the actual localeconv(). Lock to prevent other
5514 * accesses until we have made a copy of its returned static buffer */
5517 # if defined(TS_W32_BROKEN_LOCALECONV) && defined(USE_THREAD_SAFE_LOCALE)
5519 /* This is a workaround for another bug in Windows. localeconv() was
5520 * broken with thread-safe locales prior to VS 15. It looks at the global
5521 * locale instead of the thread one. As a work-around, we toggle to the
5522 * global locale; populate the return; then toggle back. We have to use
5523 * LC_ALL instead of the individual categories because of yet another bug
5524 * in Windows. And this all has to be done in a critical section.
5526 * This introduces a potential race with any other thread that has also
5527 * converted to use the global locale, and doesn't protect its locale calls
5528 * with mutexes. khw can't think of any reason for a thread to do so on
5529 * Windows, as the locale API is the same regardless of thread-safety,
5530 * except if the code is ported from working on another platform where
5531 * there might be some reason to do this. But this is typically due to
5532 * some alien-to-Perl library that thinks it owns locale setting. Such a
5533 * library isn't likely to exist on Windows, so such an application is
5534 * unlikely to be run on Windows
5536 bool restore_per_thread = FALSE;
5538 /* Save the per-thread locale state */
5539 const char * save_thread = querylocale_c(LC_ALL);
5541 /* Change to the global locale, and note if we already were there */
5542 int config_return = _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
5543 if (config_return != _DISABLE_PER_THREAD_LOCALE) {
5544 if (config_return == -1) {
5545 locale_panic_("_configthreadlocale returned an error");
5548 restore_per_thread = TRUE;
5551 /* Save the state of the global locale; then convert to our desired
5553 const char * save_global = querylocale_c(LC_ALL);
5554 void_setlocale_c(LC_ALL, save_thread);
5556 # endif /* TS_W32_BROKEN_LOCALECONV */
5558 /* Finally, do the actual localeconv */
5559 const char *lcbuf_as_string = (const char *) localeconv();
5561 /* Copy its results for each desired category as determined by
5563 U32 working_mask = which_mask;
5564 while (working_mask) {
5566 /* Get the bit position of the next lowest set bit. That is the
5567 * index into the 'strings' array of the category we use in this loop
5568 * iteration. Turn the bit off so we don't work on this category
5569 * again in this function call. */
5570 const PERL_UINT_FAST8_T i = lsbit_pos32(working_mask);
5571 working_mask &= ~ (1 << i);
5573 /* Point to the string field list for the given category ... */
5574 const lconv_offset_t * category_strings = strings[i];
5575 while (category_strings->name) {
5577 /* We have set things up so that we know where in the returned
5578 * structure, when viewed as a string, the corresponding value is.
5580 const char *value = *((const char **)( lcbuf_as_string
5581 + category_strings->offset));
5582 if (value) { /* Copy to the hash */
5584 category_strings->name,
5585 strlen(category_strings->name),
5586 newSVpv(value, strlen(value)),
5593 /* Add any int fields to the HV*. */
5595 const lconv_offset_t * current = integers[i];
5596 while (current->name) {
5597 const char value = *((const char *)( lcbuf_as_string
5598 + current->offset));
5600 current->name, strlen(current->name),
5606 } /* End of loop through the fields */
5608 /* Done with copying to the hash. Can unwind the critical section locks */
5610 # if defined(TS_W32_BROKEN_LOCALECONV) && defined(USE_THREAD_SAFE_LOCALE)
5612 /* Restore the global locale's prior state */
5613 void_setlocale_c(LC_ALL, save_global);
5615 /* And back to per-thread locales */
5616 if (restore_per_thread) {
5617 if (_configthreadlocale(_ENABLE_PER_THREAD_LOCALE) == -1) {
5618 locale_panic_("_configthreadlocale returned an error");
5622 /* Restore the per-thread locale state */
5623 void_setlocale_c(LC_ALL, save_thread);
5625 # endif /* TS_W32_BROKEN_LOCALECONV */
5627 gwLOCALE_UNLOCK; /* Finished with the critical section of a
5628 globally-accessible buffer */
5630 # if defined(USE_LOCALE_MONETARY) && defined(WIN32)
5632 restore_toggled_locale_c(LC_MONETARY, orig_MONETARY_locale);
5635 # ifdef USE_LOCALE_NUMERIC
5637 restore_toggled_locale_c(LC_NUMERIC, orig_NUMERIC_locale);
5638 if (which_mask & OFFSET_TO_BIT(NUMERIC_OFFSET)) {
5643 # ifdef WE_MUST_DEAL_WITH_MISMATCHED_CTYPE
5645 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
5651 # endif /* defined(USE_LOCALE_NUMERIC) || defined(USE_LOCALE_MONETARY) */
5652 #endif /* defined(HAS_LOCALECONV) */
5653 #ifndef HAS_SOME_LANGINFO
5655 typedef int nl_item; /* Substitute 'int' for emulated nl_langinfo() */
5661 =for apidoc Perl_langinfo
5662 =for apidoc_item Perl_langinfo8
5664 C<Perl_langinfo> is an (almost) drop-in replacement for the system
5665 C<L<nl_langinfo(3)>>, taking the same C<item> parameter values, and returning
5666 the same information. But it is more thread-safe than regular
5667 C<nl_langinfo()>, and hides the quirks of Perl's locale handling from your
5668 code, and can be used on systems that lack a native C<nl_langinfo>.
5670 However, you should instead use the improved version of this:
5671 L</Perl_langinfo8>, which behaves identically except for an additional
5672 parameter, a pointer to a variable declared as L</C<utf8ness_t>>, into which it
5673 returns to you how you should treat the returned string with regards to it
5674 being encoded in UTF-8 or not.
5676 Concerning the differences between these and plain C<nl_langinfo()>:
5682 C<Perl_langinfo8> has an extra parameter, described above. Besides this, the
5683 other reason they aren't quite a drop-in replacement is actually an advantage.
5684 The C<const>ness of the return allows the compiler to catch attempts to write
5685 into the returned buffer, which is illegal and could cause run-time crashes.
5689 They deliver the correct results for the C<RADIXCHAR> and C<THOUSEP> items,
5690 without you having to write extra code. The reason for the extra code would be
5691 because these are from the C<LC_NUMERIC> locale category, which is normally
5692 kept set by Perl so that the radix is a dot, and the separator is the empty
5693 string, no matter what the underlying locale is supposed to be, and so to get
5694 the expected results, you have to temporarily toggle into the underlying
5695 locale, and later toggle back. (You could use plain C<nl_langinfo> and
5696 C<L</STORE_LC_NUMERIC_FORCE_TO_UNDERLYING>> for this but then you wouldn't get
5697 the other advantages of C<Perl_langinfo()>; not keeping C<LC_NUMERIC> in the C
5698 (or equivalent) locale would break a lot of CPAN, which is expecting the radix
5699 (decimal point) character to be a dot.)
5703 The system function they replace can have its static return buffer trashed,
5704 not only by a subsequent call to that function, but by a C<freelocale>,
5705 C<setlocale>, or other locale change. The returned buffer of these functions
5706 is not changed until the next call to one or the other, so the buffer is never
5711 The return buffer is per-thread, so it also is never overwritten by a call to
5712 these functions from another thread; unlike the function it replaces.
5716 But most importantly, they work on systems that don't have C<nl_langinfo>, such
5717 as Windows, hence making your code more portable. Of the fifty-some possible
5718 items specified by the POSIX 2008 standard,
5719 L<http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/langinfo.h.html>,
5720 only one is completely unimplemented, though on non-Windows platforms, another
5721 significant one is not fully implemented). They use various techniques to
5722 recover the other items, including calling C<L<localeconv(3)>>, and
5723 C<L<strftime(3)>>, both of which are specified in C89, so should be always be
5724 available. Later C<strftime()> versions have additional capabilities.
5725 If an item is not available on your system, this returns either the value
5726 associated with the C locale, or simply C<"">, whichever is more appropriate.
5728 It is important to note that, when called with an item that is recovered by
5729 using C<localeconv>, the buffer from any previous explicit call to
5730 C<L<localeconv(3)>> will be overwritten. But you shouldn't be using
5731 C<localeconv> anyway because it is is very much not thread-safe, and suffers
5732 from the same problems outlined in item 'b.' above for the fields it returns
5733 that are controlled by the LC_NUMERIC locale category. Instead, avoid all of
5734 those problems by calling L</Perl_localeconv>, which is thread-safe; or by
5735 using the methods given in L<perlcall> to call
5736 L<C<POSIX::localeconv()>|POSIX/localeconv>, which is also thread-safe.
5740 The details for those items which may deviate from what this emulation returns
5741 and what a native C<nl_langinfo()> would return are specified in
5744 When using C<Perl_langinfo8> (or plain C<Perl_langinfo>) on systems that don't
5745 have a native C<nl_langinfo()>, you must
5747 #include "perl_langinfo.h"
5749 before the C<perl.h> C<#include>. You can replace your F<langinfo.h>
5750 C<#include> with this one. (Doing it this way keeps out the symbols that plain
5751 F<langinfo.h> would try to import into the namespace for code that doesn't need
5759 Perl_langinfo(const nl_item item)
5761 return Perl_langinfo8(item, NULL);
5765 Perl_langinfo8(const nl_item item, utf8ness_t * utf8ness)
5768 locale_category_index cat_index;
5770 PERL_ARGS_ASSERT_PERL_LANGINFO8;
5772 if (utf8ness) { /* Assume for now */
5773 *utf8ness = UTF8NESS_IMMATERIAL;
5776 /* Find the locale category that controls the input 'item'. If we are not
5777 * paying attention to that category, instead return a default value. Also
5778 * return the default value if there is no way for us to figure out the
5779 * correct value. If we have some form of nl_langinfo(), we can always
5780 * figure it out, but lacking that, there may be alternative methods that
5781 * can be used to recover most of the possible items. Some of those
5782 * methods need libc functions, which may or may not be available. If
5783 * unavailable, we can't compute the correct value, so must here return the
5789 #ifdef USE_LOCALE_CTYPE
5791 cat_index = LC_CTYPE_INDEX_;
5797 #if defined(USE_LOCALE_MESSAGES) && defined(HAS_SOME_LANGINFO)
5799 case YESEXPR: case YESSTR: case NOEXPR: case NOSTR:
5800 cat_index = LC_MESSAGES_INDEX_;
5803 case YESEXPR: return "^[+1yY]";
5804 case YESSTR: return "yes";
5805 case NOEXPR: return "^[-0nN]";
5806 case NOSTR: return "no";
5811 #if defined(USE_LOCALE_MONETARY) \
5812 && (defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV))
5814 cat_index = LC_MONETARY_INDEX_;
5822 #ifdef CAN_CALCULATE_RADIX
5824 cat_index = LC_NUMERIC_INDEX_;
5827 return C_decimal_point;
5832 #if defined(USE_LOCALE_NUMERIC) \
5833 && (defined(HAS_SOME_LANGINFO) || defined(HAS_LOCALECONV))
5835 cat_index = LC_NUMERIC_INDEX_;
5838 return C_thousands_sep;
5841 /* The other possible items are all in LC_TIME. */
5842 #ifdef USE_LOCALE_TIME
5845 cat_index = LC_TIME_INDEX_;
5849 #if ! defined(USE_LOCALE_TIME) || ! defined(HAS_SOME_LANGINFO)
5851 /* If not using LC_TIME, hard code the rest. Or, if there is no
5852 * nl_langinfo(), we use strftime() as an alternative, and it is missing
5853 * functionality to get every single one, so hard-code those */
5855 case ERA: return ""; /* Unimplemented; for use with strftime() %E
5858 /* These formats are defined by C89, so we assume that strftime supports
5859 * them, and so are returned unconditionally; they may not be what the
5860 * locale actually says, but should give good enough results for someone
5861 * using them as formats (as opposed to trying to parse them to figure
5862 * out what the locale says). The other format items are actually tested
5863 * to verify they work on the platform */
5864 case D_FMT: return "%x";
5865 case T_FMT: return "%X";
5866 case D_T_FMT: return "%c";
5868 # if defined(WIN32) || ! defined(USE_LOCALE_TIME)
5870 /* strftime() on Windows doesn't have the POSIX (beyond C89) extensions
5871 * that would allow it to recover these */
5872 case ERA_D_FMT: return "%x";
5873 case ERA_T_FMT: return "%X";
5874 case ERA_D_T_FMT: return "%c";
5875 case ALT_DIGITS: return "0";
5878 # ifndef USE_LOCALE_TIME
5880 case T_FMT_AMPM: return "%r";
5881 case ABDAY_1: return "Sun";
5882 case ABDAY_2: return "Mon";
5883 case ABDAY_3: return "Tue";
5884 case ABDAY_4: return "Wed";
5885 case ABDAY_5: return "Thu";
5886 case ABDAY_6: return "Fri";
5887 case ABDAY_7: return "Sat";
5888 case AM_STR: return "AM";
5889 case PM_STR: return "PM";
5890 case ABMON_1: return "Jan";
5891 case ABMON_2: return "Feb";
5892 case ABMON_3: return "Mar";
5893 case ABMON_4: return "Apr";
5894 case ABMON_5: return "May";
5895 case ABMON_6: return "Jun";
5896 case ABMON_7: return "Jul";
5897 case ABMON_8: return "Aug";
5898 case ABMON_9: return "Sep";
5899 case ABMON_10: return "Oct";
5900 case ABMON_11: return "Nov";
5901 case ABMON_12: return "Dec";
5902 case DAY_1: return "Sunday";
5903 case DAY_2: return "Monday";
5904 case DAY_3: return "Tuesday";
5905 case DAY_4: return "Wednesday";
5906 case DAY_5: return "Thursday";
5907 case DAY_6: return "Friday";
5908 case DAY_7: return "Saturday";
5909 case MON_1: return "January";
5910 case MON_2: return "February";
5911 case MON_3: return "March";
5912 case MON_4: return "April";
5913 case MON_5: return "May";
5914 case MON_6: return "June";
5915 case MON_7: return "July";
5916 case MON_8: return "August";
5917 case MON_9: return "September";
5918 case MON_10: return "October";
5919 case MON_11: return "November";
5920 case MON_12: return "December";
5925 } /* End of switch on item */
5929 Perl_croak_nocontext("panic: Unexpected nl_langinfo() item %d", item);
5930 NOT_REACHED; /* NOTREACHED */
5931 PERL_UNUSED_VAR(cat_index);
5935 return my_langinfo_i(item, cat_index, query_nominal_locale_i(cat_index),
5936 &PL_langinfo_buf, &PL_langinfo_bufsize, utf8ness);
5942 Perl_my_strftime(pTHX_ const char *fmt, int sec, int min, int hour,
5943 int mday, int mon, int year, int wday, int yday,
5945 { /* Documented above */
5946 PERL_ARGS_ASSERT_MY_STRFTIME;
5949 ints_to_tm(&mytm, sec, min, hour, mday, mon, year, wday, yday, isdst);
5950 char * ret = strftime_tm(fmt, &mytm);
5955 Perl_sv_strftime_tm(pTHX_ SV * fmt, const struct tm * mytm)
5956 { /* Documented above */
5957 PERL_ARGS_ASSERT_SV_STRFTIME_TM;
5959 utf8ness_t fmt_utf8ness = (SvUTF8(fmt) && LIKELY(! IN_BYTES))
5963 utf8ness_t result_utf8ness;
5964 char * retval = strftime8(SvPV_nolen(fmt),
5968 true /* calling from sv_strftime */
5972 sv = newSV_type(SVt_PV);
5973 sv_usepvn_flags(sv, retval, strlen(retval), SV_HAS_TRAILING_NUL);
5975 if (result_utf8ness == UTF8NESS_YES) {
5984 Perl_sv_strftime_ints(pTHX_ SV * fmt, int sec, int min, int hour,
5985 int mday, int mon, int year, int wday,
5986 int yday, int isdst)
5987 { /* Documented above */
5988 PERL_ARGS_ASSERT_SV_STRFTIME_INTS;
5991 ints_to_tm(&mytm, sec, min, hour, mday, mon, year, wday, yday, isdst);
5992 SV * ret = sv_strftime_tm(fmt, &mytm);
5997 # ifndef HAS_DEFINITIVE_UTF8NESS_DETERMINATION
5999 /* Forward declaration of function that we don't put into embed.fnc so as to
6000 * make its removal easier, as there may not be any extant platforms that need
6001 * it; and the function is located after S_my_langinfo_i() because it's easier
6002 * to understand when placed in the context of that code */
6003 STATIC const char * S_override_codeset_if_utf8_found(pTHX_
6004 const char *codeset,
6005 const char *locale);
6008 /* There are several implementations of my_langinfo, depending on the
6009 * Configuration. They all share the same beginning of the function */
6011 S_my_langinfo_i(pTHX_
6012 const nl_item item, /* The item to look up */
6013 const locale_category_index cat_index, /* The locale category
6015 /* The locale to look up 'item' in. */
6016 const char * locale,
6018 /* Where to store the result, and where the size of that buffer
6019 * is stored, updated on exit. retbuf_sizep may be NULL for an
6020 * empty-on-entry, single use buffer whose size we don't need
6021 * to keep track of */
6023 Size_t * retbuf_sizep,
6025 /* If not NULL, the location to store the UTF8-ness of 'item's
6026 * value, as documented */
6027 utf8ness_t * utf8ness)
6029 const char * retval = NULL;
6031 PERL_ARGS_ASSERT_MY_LANGINFO_I;
6032 assert(cat_index < LC_ALL_INDEX_);
6034 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
6035 "Entering my_langinfo item=%ld, using locale %s\n",
6036 (long) item, locale));
6037 /*--------------------------------------------------------------------------*/
6038 /* Above is the common beginning to all the implementations of my_langinfo().
6039 * Below are the various completions.
6041 * Some platforms don't deal well with non-ASCII strings in locale X when
6042 * LC_CTYPE is not in X. (Actually it is probably when X is UTF-8 and LC_CTYPE
6043 * isn't, or vice versa). There is explicit code to bring the categories into
6044 * sync. This doesn't seem to be a problem with nl_langinfo(), so that
6045 * implementation doesn't currently worry about it. But it is a problem on
6046 * Windows boxes, which don't have nl_langinfo(). */
6048 /*--------------------------------------------------------------------------*/
6049 # if defined(HAS_NL_LANGINFO) /* nl_langinfo() is available. */
6050 # ifdef WE_MUST_DEAL_WITH_MISMATCHED_CTYPE
6052 /* This function sorts out if things actually have to be switched or not,
6053 * for both save and restore. */
6054 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
6058 const char * orig_switched_locale = toggle_locale_i(cat_index, locale);
6061 retval = save_to_buffer(nl_langinfo(item), retbufp, retbuf_sizep);
6065 *utf8ness = get_locale_string_utf8ness_i(retval,
6066 LOCALE_UTF8NESS_UNKNOWN,
6070 restore_toggled_locale_i(cat_index, orig_switched_locale);
6072 # ifdef WE_MUST_DEAL_WITH_MISMATCHED_CTYPE
6074 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
6079 /*--------------------------------------------------------------------------*/
6080 # else /* Below, emulate nl_langinfo as best we can */
6082 /* The other completion is where we have to emulate nl_langinfo(). There
6083 * are various possibilities depending on the Configuration. The major
6084 * platform lacking nl_langinfo is Windows. It does have GetLocaleInfoEx()
6085 * that could be used to get most of the items, but it (and other similar
6086 * Windows API functions) use what MS calls "locale names", whereas the C
6087 * functions use what MS calls "locale strings". The locale string
6088 * "English_United_States.1252" is equivalent to the locale name "en_US".
6089 * There are tables inside Windows that translate between the two forms,
6090 * but they are not exposed. Also calling setlocale(), then calling
6091 * GetThreadLocale() doesn't work, as the former doesn't change the
6092 * latter's return. Therefore we are stuck using the mechanisms below. */
6093 /* Almost all the items will have ASCII return values. Set that here, and
6094 * override if necessary */
6095 utf8ness_t is_utf8 = UTF8NESS_IMMATERIAL;
6099 assert(item < 0); /* Make sure using perl_langinfo.h */
6105 # if defined(HAS_SNPRINTF) \
6106 && (! defined(HAS_LOCALECONV) || defined(TS_W32_BROKEN_LOCALECONV))
6109 /* snprintf() can be used to find the radix character by outputting
6110 * a known simple floating point number to a buffer, and parsing
6111 * it, inferring the radix as the bytes separating the integer and
6112 * fractional parts. But localeconv() is more direct, not
6113 * requiring inference, so use it instead of the code just below,
6114 * if (likely) it is available and works ok */
6116 char * floatbuf = NULL;
6117 const Size_t initial_size = 10;
6119 Newx(floatbuf, initial_size, char);
6121 # if defined(WE_MUST_DEAL_WITH_MISMATCHED_CTYPE)
6122 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
6125 const char * orig_NUMERIC_locale = toggle_locale_c(LC_NUMERIC,
6127 /* 1.5 is exactly representable on binary computers */
6128 Size_t needed_size = snprintf(floatbuf, initial_size, "%.1f", 1.5);
6130 /* If our guess wasn't big enough, increase and try again, based on
6131 * the real number that snprintf() is supposed to return */
6132 if (UNLIKELY(needed_size >= initial_size)) {
6133 needed_size++; /* insurance */
6134 Renew(floatbuf, needed_size, char);
6135 Size_t new_needed = snprintf(floatbuf, needed_size, "%.1f", 1.5);
6136 assert(new_needed <= needed_size);
6137 needed_size = new_needed;
6140 restore_toggled_locale_c(LC_NUMERIC, orig_NUMERIC_locale);
6142 # if defined(WE_MUST_DEAL_WITH_MISMATCHED_CTYPE)
6143 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
6146 char * s = floatbuf;
6147 char * e = floatbuf + needed_size;
6150 while (s < e && *s != '1') {
6154 if (LIKELY(s < e)) {
6159 char * item_start = s;
6160 while (s < e && *s != '5') {
6164 /* Everything in between is the radix string */
6165 if (LIKELY(s < e)) {
6167 retval = save_to_buffer(item_start, retbufp, retbuf_sizep);
6171 is_utf8 = get_locale_string_utf8ness_i(retval,
6172 LOCALE_UTF8NESS_UNKNOWN,
6182 # ifdef HAS_LOCALECONV /* snprintf() failed; drop down to use
6187 # else /* snprintf() failed and no localeconv() */
6189 retval = C_decimal_point;
6194 # ifdef HAS_LOCALECONV
6196 /* These items are available from localeconv(). */
6198 /* case RADIXCHAR: // May drop down to here in some configurations */
6203 /* The hash gets populated with just the field(s) related to 'item'. */
6204 HV * result_hv = my_localeconv(item);
6207 if (item != CRNCYSTR) {
6209 /* These items have been populated with just one key => value */
6210 (void) hv_iterinit(result_hv);
6211 HE * entry = hv_iternext(result_hv);
6212 string = hv_iterval(result_hv, entry);
6216 /* But CRNCYSTR localeconv() returns a slightly different value
6217 * than the nl_langinfo() API calls for, so have to modify this one
6218 * to conform. We need another value from localeconv() to know
6219 * what to change it to. my_localeconv() has populated the hash
6220 * with exactly both fields. Delete this one, leaving just the
6221 * CRNCYSTR one in the hash */
6222 SV* precedes = hv_delete(result_hv,
6223 P_CS_PRECEDES_LITERAL,
6224 STRLENs(P_CS_PRECEDES_LITERAL),
6227 locale_panic_("my_localeconv() unexpectedly didn't return"
6228 " a value for " P_CS_PRECEDES_LITERAL);
6231 /* The modification is to prefix the localeconv() return with a
6232 * single byte, calculated as follows: */
6233 char prefix = (LIKELY(SvIV(precedes) != -1))
6234 ? ((precedes != 0) ? '-' : '+')
6236 /* khw couldn't find any documentation that
6237 * CHAR_MAX (which we modify to -1) is the signal,
6238 * but cygwin uses it thusly, and it makes sense
6239 * given that CHAR_MAX indicates the value isn't
6240 * used, so it neither precedes nor succeeds */
6243 /* Now get CRNCYSTR */
6244 (void) hv_iterinit(result_hv);
6245 HE * entry = hv_iternext(result_hv);
6246 string = hv_iterval(result_hv, entry);
6248 /* And perform the modification */
6249 Perl_sv_setpvf(aTHX_ string, "%c%s", prefix, SvPV_nolen(string));
6252 /* Here, 'string' contains the value we want to return */
6253 retval = save_to_buffer(SvPV_nolen(string), retbufp, retbuf_sizep);
6256 is_utf8 = get_locale_string_utf8ness_i(retval,
6257 LOCALE_UTF8NESS_UNKNOWN,
6266 # endif /* Some form of localeconv */
6267 # ifdef HAS_STRFTIME
6269 /* These formats are only available in later strftime's */
6270 case ERA_D_FMT: case ERA_T_FMT: case ERA_D_T_FMT: case T_FMT_AMPM:
6272 /* The rest can be gotten from most versions of strftime(). */
6273 case ABDAY_1: case ABDAY_2: case ABDAY_3:
6274 case ABDAY_4: case ABDAY_5: case ABDAY_6: case ABDAY_7:
6276 case AM_STR: case PM_STR:
6277 case ABMON_1: case ABMON_2: case ABMON_3: case ABMON_4:
6278 case ABMON_5: case ABMON_6: case ABMON_7: case ABMON_8:
6279 case ABMON_9: case ABMON_10: case ABMON_11: case ABMON_12:
6280 case DAY_1: case DAY_2: case DAY_3: case DAY_4:
6281 case DAY_5: case DAY_6: case DAY_7:
6282 case MON_1: case MON_2: case MON_3: case MON_4:
6283 case MON_5: case MON_6: case MON_7: case MON_8:
6284 case MON_9: case MON_10: case MON_11: case MON_12:
6286 const char * format;
6287 bool return_format = FALSE;
6292 GCC_DIAG_IGNORE_STMT(-Wimplicit-fallthrough);
6296 locale_panic_(Perl_form(aTHX_ "switch case: %d problem", item));
6297 NOT_REACHED; /* NOTREACHED */
6299 case PM_STR: hour = 18;
6303 case ABDAY_7: mday++;
6304 case ABDAY_6: mday++;
6305 case ABDAY_5: mday++;
6306 case ABDAY_4: mday++;
6307 case ABDAY_3: mday++;
6308 case ABDAY_2: mday++;
6321 case ABMON_12: mon++;
6322 case ABMON_11: mon++;
6323 case ABMON_10: mon++;
6324 case ABMON_9: mon++;
6325 case ABMON_8: mon++;
6326 case ABMON_7: mon++;
6327 case ABMON_6: mon++;
6328 case ABMON_5: mon++;
6329 case ABMON_4: mon++;
6330 case ABMON_3: mon++;
6331 case ABMON_2: mon++;
6351 return_format = TRUE;
6355 return_format = TRUE;
6359 return_format = TRUE;
6363 return_format = TRUE;
6366 format = "%Ow"; /* Find the alternate digit for 0 */
6370 GCC_DIAG_RESTORE_STMT;
6372 /* The year was deliberately chosen so that January 1 is on the
6373 * first day of the week. Since we're only getting one thing at a
6374 * time, it all works */
6377 const char * orig_TIME_locale = toggle_locale_c(LC_TIME, locale);
6379 ints_to_tm(&mytm, 30, 30, hour, mday, mon, 2011, 0, 0, 0);
6382 temp = strftime8(format,
6384 UTF8NESS_IMMATERIAL, /* All possible formats
6388 false /* not calling from sv_strftime */
6392 temp = strftime_tm(format, &mytm);
6395 restore_toggled_locale_c(LC_TIME, orig_TIME_locale);
6397 retval = save_to_buffer(temp, retbufp, retbuf_sizep);
6400 /* If the item is 'ALT_DIGITS', '*retbuf' contains the alternate
6401 * format for wday 0. If the value is the same as the normal 0,
6402 * there isn't an alternate, so clear the buffer.
6404 * (wday was chosen because its range is all a single digit.
6405 * Things like tm_sec have two digits as the minimum: '00'.) */
6406 if (item == ALT_DIGITS && strEQ(*retbufp, "0")) {
6411 /* ALT_DIGITS is problematic. Experiments on it showed that
6412 * strftime() did not always work properly when going from alt-9 to
6413 * alt-10. Only a few locales have this item defined, and in all
6414 * of them on Linux that khw was able to find, nl_langinfo() merely
6415 * returned the alt-0 character, possibly doubled. Most Unicode
6416 * digits are in blocks of 10 consecutive code points, so that is
6417 * sufficient information for such scripts, as we can infer alt-1,
6418 * alt-2, .... But for a Japanese locale, a CJK ideographic 0 is
6419 * returned, and the CJK digits are not in code point order, so you
6420 * can't really infer anything. The localedef for this locale did
6421 * specify the succeeding digits, so that strftime() works properly
6422 * on them, without needing to infer anything. But the
6423 * nl_langinfo() return did not give sufficient information for the
6424 * caller to understand what's going on. So until there is
6425 * evidence that it should work differently, this returns the alt-0
6426 * string for ALT_DIGITS. */
6428 if (return_format) {
6430 /* If to return the format, not the value, overwrite the buffer
6431 * with it. But some strftime()s will keep the original format
6432 * if illegal, so change those to "" */
6433 if (strEQ(*retbufp, format)) {
6440 /* A format is always in ASCII */
6441 is_utf8 = UTF8NESS_IMMATERIAL;
6448 # ifdef USE_LOCALE_CTYPE
6452 /* The trivial case */
6453 if (isNAME_C_OR_POSIX(locale)) {
6458 /* If this happens to match our cached value */
6459 if (PL_in_utf8_CTYPE_locale && strEQ(locale, PL_ctype_name)) {
6466 const char * orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, locale);
6468 # ifndef WIN32_USE_FAKE_OLD_MINGW_LOCALES
6470 /* This function retrieves the code page. It is subject to change, but
6471 * is documented and has been stable for many releases */
6472 retval = save_to_buffer(Perl_form(aTHX_ "%d", ___lc_codepage_func()),
6473 retbufp, retbuf_sizep);
6476 retval = save_to_buffer(nl_langinfo(CODESET),
6477 retbufp, retbuf_sizep);
6480 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
6482 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "locale='%s' cp=%s\n",
6488 /* The codeset is important, but khw did not figure out a way for it to
6489 * be retrieved on non-Windows boxes without nl_langinfo(). But even
6490 * if we can't get it directly, we can usually determine if it is a
6491 * UTF-8 locale or not. If it is UTF-8, we (correctly) use that for
6494 # ifdef HAS_DEFINITIVE_UTF8NESS_DETERMINATION
6496 if (is_locale_utf8(locale)) {
6503 /* Here, the code set has not been found. The only other option khw
6504 * could think of is to see if the codeset is part of the locale name.
6505 * This is very less than ideal; often there is no code set in the
6506 * name; and at other times they even lie.
6508 * But there is an XPG standard syntax, which many locales follow:
6510 * language[_territory[.codeset]][@modifier]
6512 * So we take the part between the dot and any '@' */
6513 retval = strchr(locale, '.');
6515 retval = ""; /* Alas, no dot */
6519 /* Don't include the dot */
6522 /* And stop before any '@' */
6523 const char * modifier = strchr(retval, '@');
6525 char * code_set_name;
6526 const Size_t name_len = modifier - retval;
6527 Newx(code_set_name, name_len + 1, char); /* +1 for NUL */
6528 my_strlcpy(code_set_name, retval, name_len + 1);
6529 SAVEFREEPV(code_set_name);
6530 retval = code_set_name;
6533 /* The code set name is considered to be everything between the dot
6535 retval = save_to_buffer(retval, retbufp, retbuf_sizep);
6538 # ifndef HAS_DEFINITIVE_UTF8NESS_DETERMINATION
6540 /* Here, 'retval' contains any codeset name derived from the locale
6541 * name. That derived name may be empty or not necessarily indicative
6542 * of the real codeset. But we can often determine if it should be
6543 * UTF-8, regardless of what the name is. On most platforms, that
6544 * determination is definitive, and was already done. But for this
6545 * code to be compiled, this platform is not one of them. However,
6546 * there are typically tools available to make a very good guess, and
6547 * knowing the derived codeset name improves the quality of that guess.
6548 * The following function overrides the derived codeset name when it
6549 * guesses that it actually should be UTF-8. It could be inlined here,
6550 * but was moved out of this switch() so as to make the switch()
6551 * control flow easier to follow */
6552 retval = S_override_codeset_if_utf8_found(aTHX_ retval, locale);
6556 break; /* All done */
6558 # endif /* ! WIN32 */
6559 # endif /* USE_LOCALE_CTYPE */
6561 } /* Giant switch() of nl_langinfo() items */
6564 *utf8ness = is_utf8;
6569 # endif /* All the implementations of my_langinfo() */
6571 /*--------------------------------------------------------------------------*/
6573 } /* my_langinfo() */
6575 # ifndef HAS_DEFINITIVE_UTF8NESS_DETERMINATION
6578 S_override_codeset_if_utf8_found(pTHX_ const char * codeset,
6579 const char * locale)
6581 # define NAME_INDICATES_UTF8 0x1
6582 # define MB_CUR_MAX_SUGGESTS_UTF8 0x2
6584 /* Override 'codeset' with UTF-8 if this routine guesses that it should be.
6585 * Conversely (but rarely), "UTF-8" in the locale name might be wrong. We
6586 * return "" as the code set name if we find that to be the case. */
6588 unsigned int lean_towards_being_utf8 = 0;
6589 if (is_codeset_name_UTF8(codeset)) {
6590 lean_towards_being_utf8 |= NAME_INDICATES_UTF8;
6593 /* For this portion of the file to compile, some C99 functions aren't
6594 * available to us, even though we now require C99. So, something must be
6595 * wrong with them. The code here should be good enough to work around
6596 * this issue, but should the need arise, comments in S_is_locale_utf8()
6597 * list some alternative C99 functions that could be tried.
6599 * But MB_CUR_MAX is a C89 construct that helps a lot, is simple for a
6600 * vendor to implement, and our experience with it is that it works well on
6601 * a variety of platforms. We have found that it returns a too-large
6602 * number on some platforms for the C locale, but for no others. That
6603 * locale was already ruled out in the code that called this function. (If
6604 * MB_CUR_MAX returned too small a number, that would break a lot of
6605 * things, and likely would be quickly corrected by the vendor.) khw has
6606 * some confidence that it doesn't return >1 when 1 is meant, as that would
6607 * trigger a Perl warning, and we've had no reports of invalid occurrences
6611 /* If there are fewer bytes available in this locale than are required to
6612 * represent the largest legal UTF-8 code point, this definitely isn't a
6613 * UTF-8 locale, even if the locale name says it is. */
6614 const int mb_cur_max = MB_CUR_MAX;
6615 if (mb_cur_max < (int) UNISKIP(PERL_UNICODE_MAX)) {
6616 if (lean_towards_being_utf8 & NAME_INDICATES_UTF8) {
6617 return ""; /* The name is wrong; override */
6623 /* But if the locale could be UTF-8, and also the name corroborates this,
6624 * assume it is so */
6625 if (lean_towards_being_utf8 & NAME_INDICATES_UTF8) {
6629 /* Here, the name doesn't indicate UTF-8, but MB_CUR_MAX indicates it could
6630 * be. khw knows of only two other locales in the world, EUC-TW and GB
6631 * 18030, that legitimately require this many bytes (4). In both, the
6632 * single byte characters are the same as ASCII. No multi-byte character
6633 * in EUC-TW is legal UTF-8 (since the first byte of each is a
6634 * continuation). GB 18030 has no three byte sequences, and none of the
6635 * four byte ones is legal UTF-8 (as the second byte for these is a
6636 * non-continuation). But every legal UTF-8 two byte sequence is also
6637 * legal in GB 18030, though none have the same meaning, and no Han code
6638 * point expressed in UTF-8 is two byte. So the further tests below which
6639 * look for native expressions of currency and time will not return two
6640 * byte sequences, hence they will reliably rule out this locale as being
6641 * UTF-8. So, if we get this far, the result is almost certainly UTF-8.
6642 * But to be really sure, also check that there is no illegal UTF-8. */
6643 lean_towards_being_utf8 |= MB_CUR_MAX_SUGGESTS_UTF8;
6645 # endif /* has MB_CUR_MAX */
6647 /* Here, MB_CUR_MAX is not available, or was inconclusive. What we do is
6648 * to look at various strings associated with the locale:
6649 * 1) If any are illegal UTF-8, the locale can't be UTF-8.
6650 * 2) If all are legal UTF-8, and some non-ASCII characters are present,
6651 * it is likely to be UTF-8, because of the strictness of UTF-8
6652 * syntax. So assume it is UTF-8
6653 * 3) If all are ASCII and the locale name and/or MB_CUR_MAX indicate
6654 * UTF-8, assume the locale is UTF-8.
6655 * 4) Otherwise, assume the locale isn't UTF-8
6657 * To save cycles, if the locale name indicates it is a UTF-8 locale, we
6658 * stop looking at the first instance with legal non-ASCII UTF-8. It is
6659 * very unlikely this combination is coincidental. */
6661 utf8ness_t strings_utf8ness = UTF8NESS_UNKNOWN;
6662 char * scratch_buf = NULL;
6663 Size_t scratch_buf_size = 0;
6665 /* List of strings to look at */
6666 const int trials[] = {
6668 # if defined(USE_LOCALE_MONETARY) && defined(HAS_LOCALECONV)
6670 /* The first string tried is the locale currency name. Often that will
6671 * be in the native script.
6673 * But this is usable only if localeconv() is available, as that's the
6674 * way we find out the currency symbol. */
6679 # ifdef USE_LOCALE_TIME
6681 /* We can also try various strings associated with LC_TIME, like the names
6682 * of months or days of the week */
6684 DAY_1, DAY_2, DAY_3, DAY_4, DAY_5, DAY_6, DAY_7,
6685 MON_1, MON_2, MON_3, MON_4, MON_5, MON_6, MON_7, MON_8,
6686 MON_9, MON_10, MON_11, MON_12,
6687 ALT_DIGITS, AM_STR, PM_STR,
6688 ABDAY_1, ABDAY_2, ABDAY_3, ABDAY_4, ABDAY_5, ABDAY_6, ABDAY_7,
6689 ABMON_1, ABMON_2, ABMON_3, ABMON_4, ABMON_5, ABMON_6,
6690 ABMON_7, ABMON_8, ABMON_9, ABMON_10, ABMON_11, ABMON_12
6696 # ifdef USE_LOCALE_TIME
6698 /* The code in the recursive call below can handle switching the locales,
6699 * but by doing it now here, that code will check and discover that there
6700 * is no need to switch then restore, avoiding those each loop iteration */
6701 const char * orig_TIME_locale = toggle_locale_c(LC_TIME, locale);
6705 /* The trials array may consist of strings from two different locale
6706 * categories. The call to my_langinfo_i() below needs to pass the proper
6707 * category for each string. There is a max of 1 trial for LC_MONETARY;
6708 * the rest are LC_TIME. So the array is arranged so the LC_MONETARY item
6709 * (if any) is first, and all subsequent iterations will use LC_TIME.
6710 * These #ifdefs set up the values for all possible combinations. */
6711 # if defined(USE_LOCALE_MONETARY) && defined(HAS_LOCALECONV)
6713 locale_category_index cat_index = LC_MONETARY_INDEX_;
6715 # ifdef USE_LOCALE_TIME
6717 const locale_category_index follow_on_cat_index = LC_TIME_INDEX_;
6718 assert(trials[1] == DAY_1); /* Make sure only a single non-time entry */
6722 /* Effectively out-of-bounds, as there is only the monetary entry */
6723 const locale_category_index follow_on_cat_index = LC_ALL_INDEX_;
6726 # elif defined(USE_LOCALE_TIME)
6728 locale_category_index cat_index = LC_TIME_INDEX_;
6729 const locale_category_index follow_on_cat_index = LC_TIME_INDEX_;
6733 /* Effectively out-of-bounds, as here there are no trial entries at all.
6734 * This allows this code to compile, but there are no strings to test, and
6735 * so the answer will always be non-UTF-8. */
6736 locale_category_index cat_index = LC_ALL_INDEX_;
6737 const locale_category_index follow_on_cat_index = LC_ALL_INDEX_;
6741 /* Everything set up; look through all the strings */
6742 for (PERL_UINT_FAST8_T i = 0; i < C_ARRAY_LENGTH(trials); i++) {
6743 (void) my_langinfo_i(trials[i], cat_index, locale,
6744 &scratch_buf, &scratch_buf_size, NULL);
6745 cat_index = follow_on_cat_index;
6747 /* To prevent infinite recursive calls, we don't ask for the UTF-8ness
6748 * of the string (in 'trials[i]') above. Instead we examine the
6749 * returned string here */
6750 const Size_t len = strlen(scratch_buf);
6751 const U8 * first_variant;
6753 /* If the string is identical whether or not it is encoded as UTF-8, it
6754 * isn't helpful in determining UTF8ness. */
6755 if (is_utf8_invariant_string_loc((U8 *) scratch_buf, len,
6761 /* Here, has non-ASCII. If not legal UTF-8, isn't a UTF-8 locale */
6762 if (! is_utf8_string(first_variant,
6763 len - (first_variant - (U8 *) scratch_buf)))
6765 strings_utf8ness = UTF8NESS_NO;
6769 /* Here, is a legal non-ASCII UTF-8 string; tentatively set the return
6770 * to YES; possibly overridden by later iterations */
6771 strings_utf8ness = UTF8NESS_YES;
6773 /* But if this corroborates our expectation, quit now */
6774 if (lean_towards_being_utf8 & NAME_INDICATES_UTF8) {
6779 # ifdef USE_LOCALE_TIME
6781 restore_toggled_locale_c(LC_TIME, orig_TIME_locale);
6785 Safefree(scratch_buf);
6788 if (strings_utf8ness == UTF8NESS_NO) {
6789 return codeset; /* No override */
6792 /* Here all tested strings are legal UTF-8.
6794 * Above we set UTF8NESS_YES if any string wasn't ASCII. But even if they
6795 * are all ascii, and the locale name indicates it is a UTF-8 locale,
6796 * assume the locale is UTF-8. */
6797 if (lean_towards_being_utf8) {
6798 strings_utf8ness = UTF8NESS_YES;
6801 if (strings_utf8ness == UTF8NESS_YES) {
6805 /* Here, nothing examined indicates that the codeset is or isn't UTF-8.
6806 * But what is it? The other locale categories are not likely to be of
6809 * LC_NUMERIC Only a few locales in the world have a non-ASCII radix or
6811 * LC_CTYPE This code wouldn't be compiled if mbtowc() existed and was
6812 * reliable. This is unlikely in C99. There are other
6813 * functions that could be used instead, but are they going to
6814 * exist, and be able to distinguish between UTF-8 and 8859-1?
6815 * Deal with this only if it becomes necessary.
6816 * LC_MESSAGES The strings returned from strerror() would seem likely
6817 * candidates, but experience has shown that many systems
6818 * don't actually have translations installed for them. They
6819 * are instead always in English, so everything in them is
6820 * ASCII, which is of no help to us. A Configure probe could
6821 * possibly be written to see if this platform has non-ASCII
6822 * error messages. But again, wait until it turns out to be
6823 * an actual problem.
6825 * Things like YESSTR, NOSTR, might not be in ASCII, but need
6826 * nl_langinfo() to access, which we don't have.
6829 /* Otherwise, assume the locale isn't UTF-8. This can be wrong if we don't
6830 * have MB_CUR_MAX, and the locale is English without UTF-8 in its name,
6831 * and with a dollar currency symbol. */
6832 return codeset; /* No override */
6835 # endif /* ! HAS_DEFINITIVE_UTF8NESS_DETERMINATION */
6836 #endif /* USE_LOCALE */
6839 =for apidoc_section $time
6840 =for apidoc sv_strftime_tm
6841 =for apidoc_item sv_strftime_ints
6842 =for apidoc_item my_strftime
6844 These implement the libc strftime(), but with a different API so that the return
6845 value is a pointer to the formatted result (which MUST be arranged to be FREED
6846 BY THE CALLER). This allows these functions to increase the buffer size as
6847 needed, so that the caller doesn't have to worry about that.
6849 On failure they return NULL, and set errno to C<EINVAL>.
6851 C<sv_strftime_tm> and C<sv_strftime_ints> are preferred, as they transparently
6852 handle the UTF-8ness of the current locale, the input C<fmt>, and the returned
6853 result. Only if the current C<LC_TIME> locale is a UTF-8 one (and S<C<use
6854 bytes>> is not in effect) will the result be marked as UTF-8. These differ
6855 only in the form of their inputs. C<sv_strftime_tm> takes a filled-in
6856 S<C<struct tm>> parameter. C<sv_strftime_ints> takes a bunch of integer
6857 parameters that together completely define a given time.
6859 C<my_strftime> is kept for backwards compatibility. Knowing if its result
6860 should be considered UTF-8 or not requires significant extra logic.
6862 Note that C<yday> and C<wday> effectively are ignored by C<sv_strftime_ints>
6863 and C<my_strftime>, as mini_mktime() overwrites them
6865 Also note that all three functions are always executed in the underlying
6866 C<LC_TIME> locale of the program, giving results based on that locale.
6872 S_ints_to_tm(pTHX_ struct tm * mytm,
6873 int sec, int min, int hour, int mday, int mon, int year,
6874 int wday, int yday, int isdst)
6876 /* Create a struct tm structure from the input time-related integer
6879 /* Override with the passed-in values */
6880 Zero(mytm, 1, struct tm);
6883 mytm->tm_hour = hour;
6884 mytm->tm_mday = mday;
6886 mytm->tm_year = year;
6887 mytm->tm_wday = wday;
6888 mytm->tm_yday = yday;
6889 mytm->tm_isdst = isdst;
6892 /* use libc to get the values for tm_gmtoff and tm_zone on platforms that
6893 * have them [perl #18238] */
6894 #if defined(HAS_MKTIME) \
6895 && (defined(HAS_TM_TM_GMTOFF) || defined(HAS_TM_TM_ZONE))
6896 struct tm mytm2 = *mytm;
6900 # ifdef HAS_TM_TM_GMTOFF
6901 mytm->tm_gmtoff = mytm2.tm_gmtoff;
6903 # ifdef HAS_TM_TM_ZONE
6904 mytm->tm_zone = mytm2.tm_zone;
6912 S_strftime_tm(pTHX_ const char *fmt, const struct tm *mytm)
6914 PERL_ARGS_ASSERT_STRFTIME_TM;
6916 /* Execute strftime() based on the input struct tm */
6918 /* An empty format yields an empty result */
6919 const int fmtlen = strlen(fmt);
6922 Newxz (ret, 1, char);
6926 #ifndef HAS_STRFTIME
6927 Perl_croak(aTHX_ "panic: no strftime");
6929 # if defined(WE_MUST_DEAL_WITH_MISMATCHED_CTYPE) && defined(USE_LOCALE_TIME)
6931 const char * orig_CTYPE_LOCALE = toggle_locale_c(LC_CTYPE,
6932 querylocale_c(LC_TIME));
6935 /* Guess an initial size for the returned string based on an expansion
6936 * factor of the input format, but with a minimum that should handle most
6937 * common cases. If this guess is too small, we will try again with a
6939 int bufsize = MAX(fmtlen * 2, 64);
6941 char *buf = NULL; /* Makes Renew() act as Newx() on the first iteration */
6943 Renew(buf, bufsize, char);
6945 /* allowing user-supplied (rather than literal) formats is normally
6946 * frowned upon as a potential security risk; but this is part of the
6947 * API so we have to allow it (and the available formats have a much
6948 * lower chance of doing something bad than the ones for printf etc. */
6949 GCC_DIAG_IGNORE_STMT(-Wformat-nonliteral);
6952 int len = strftime(buf, bufsize, fmt, mytm);
6955 GCC_DIAG_RESTORE_STMT;
6957 /* A non-zero return indicates success. But to make sure we're not
6958 * dealing with some rogue strftime that returns how much space it
6959 * needs instead of 0 when there isn't enough, check that the return
6960 * indicates we have at least one byte of spare space (which will be
6961 * used for the terminating NUL). */
6962 if (inRANGE(len, 1, bufsize - 1)) {
6963 goto strftime_return;
6966 /* There are several possible reasons for a 0 return code for a
6967 * non-empty format, and they are not trivial to tease apart. This
6968 * issue is a known bug in the strftime() API. What we do to cope is
6969 * to assume that the reason is not enough space in the buffer, so
6970 * increase it and try again. */
6973 /* But don't just keep increasing the size indefinitely. Stop when it
6974 * becomes obvious that the reason for failure is something besides not
6975 * enough space. The most likely largest expanding format is %c. On
6976 * khw's Linux box, the maximum result of this is 67 characters, in the
6977 * km_KH locale. If a new script comes along that uses 4 UTF-8 bytes
6978 * per character, and with a similar expansion factor, that would be a
6979 * 268:2 byte ratio, or a bit more than 128:1 = 2**7:1. Some strftime
6980 * implementations allow you to say %1000c to pad to 1000 bytes. This
6981 * shows that it is impossible to implement this without a heuristic
6982 * (which can fail). But it indicates we need to be generous in the
6983 * upper limit before failing. The previous heuristic used was too
6984 * stingy. Since the size doubles per iteration, it doesn't take many
6985 * to reach the limit */
6986 } while (bufsize < ((1 << 11) + 1) * fmtlen);
6988 /* Here, strftime() returned 0, and it likely wasn't for lack of space.
6989 * There are two possible reasons:
6991 * First is that the result is legitimately 0 length. This can happen
6992 * when the format is precisely "%p". That is the only documented format
6993 * that can have an empty result. */
6994 if (strEQ(fmt, "%p")) {
6995 Renew(buf, 1, char);
6997 goto strftime_return;
7000 /* The other reason is that the format string is malformed. Probably it is
7001 * that the string is syntactically invalid for the locale. On some
7002 * platforms an invalid conversion specifier '%?' (for all illegal '?') is
7003 * treated as a literal, but others may fail when '?' is illegal */
7010 # if defined(WE_MUST_DEAL_WITH_MISMATCHED_CTYPE) && defined(USE_LOCALE_TIME)
7012 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_LOCALE);
7023 S_strftime8(pTHX_ const char * fmt,
7024 const struct tm * mytm,
7025 const utf8ness_t fmt_utf8ness,
7026 utf8ness_t * result_utf8ness,
7027 const bool came_from_sv)
7029 PERL_ARGS_ASSERT_STRFTIME8;
7031 /* Wrap strftime_tm, taking into account the input and output UTF-8ness */
7033 #ifdef USE_LOCALE_TIME
7034 # define INDEX_TO_USE LC_TIME_INDEX_
7036 const char * locale = querylocale_c(LC_TIME);
7037 locale_utf8ness_t locale_utf8ness = LOCALE_UTF8NESS_UNKNOWN;
7040 # define INDEX_TO_USE LC_ALL_INDEX_ /* Effectively out of bounds */
7042 const char * locale = "C";
7043 locale_utf8ness_t locale_utf8ness = LOCALE_NOT_UTF8;
7047 switch (fmt_utf8ness) {
7048 case UTF8NESS_IMMATERIAL:
7051 case UTF8NESS_NO: /* Known not to be UTF-8; must not be UTF-8 locale */
7052 if (is_locale_utf8(locale)) {
7057 locale_utf8ness = LOCALE_NOT_UTF8;
7060 case UTF8NESS_YES: /* Known to be UTF-8; must be UTF-8 locale if can't
7062 if (! is_locale_utf8(locale)) {
7063 locale_utf8ness = LOCALE_NOT_UTF8;
7065 bool is_utf8 = true;
7066 Size_t fmt_len = strlen(fmt);
7067 fmt = (char *) bytes_from_utf8((U8 *) fmt, &fmt_len, &is_utf8);
7076 locale_utf8ness = LOCALE_IS_UTF8;
7081 case UTF8NESS_UNKNOWN:
7082 if (! is_locale_utf8(locale)) {
7083 locale_utf8ness = LOCALE_NOT_UTF8;
7086 locale_utf8ness = LOCALE_IS_UTF8;
7089 /* Upgrade 'fmt' to UTF-8 for a UTF-8 locale. Otherwise the
7090 * locale would find any UTF-8 variant characters to be
7092 Size_t fmt_len = strlen(fmt);
7093 fmt = (char *) bytes_to_utf8((U8 *) fmt, &fmt_len);
7101 char * retval = strftime_tm(fmt, mytm);
7102 *result_utf8ness = get_locale_string_utf8ness_i(retval,
7106 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
7107 "fmt=%s, retval=%s; utf8ness=%d",
7109 ((is_utf8_string((U8 *) retval, 0))
7111 :_byte_dump_string((U8 *) retval, strlen(retval),0)),
7122 S_give_perl_locale_control(pTHX_
7124 const char * lc_all_string,
7126 const char ** locales,
7128 const line_t caller_line)
7130 PERL_UNUSED_ARG(caller_line);
7132 /* This is called when the program is in the global locale and are
7133 * switching to per-thread (if available). And it is called at
7134 * initialization time to do the same.
7137 # if defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE)
7139 /* On Windows, convert to per-thread behavior. This isn't necessary in
7140 * POSIX 2008, as the conversion gets done automatically in the
7141 * void_setlocale_i() calls below. */
7142 if (_configthreadlocale(_ENABLE_PER_THREAD_LOCALE) == -1) {
7143 locale_panic_("_configthreadlocale returned an error");
7147 # if ! defined(USE_THREAD_SAFE_LOCALE) \
7148 && ! defined(USE_POSIX_2008_LOCALE)
7149 # if defined(LC_ALL)
7150 PERL_UNUSED_ARG(lc_all_string);
7152 PERL_UNUSED_ARG(locales);
7156 /* This platform has per-thread locale handling. Do the conversion. */
7158 # if defined(LC_ALL)
7160 void_setlocale_c_with_caller(LC_ALL, lc_all_string, __FILE__, caller_line);
7164 for_all_individual_category_indexes(i) {
7165 void_setlocale_i_with_caller(i, locales[i], __FILE__, caller_line);
7171 /* Finally, update our remaining records. 'true' => force recalculation.
7172 * This is needed because we don't know what's happened while Perl hasn't
7173 * had control, so we need to figure out the current state */
7175 # if defined(LC_ALL)
7177 new_LC_ALL(lc_all_string, true);
7181 new_LC_ALL(calculate_LC_ALL_string(locales,
7191 S_output_check_environment_warning(pTHX_ const char * const language,
7192 const char * const lc_all,
7193 const char * const lang)
7195 PerlIO_printf(Perl_error_log,
7196 "perl: warning: Please check that your locale settings:\n");
7200 PerlIO_printf(Perl_error_log, "\tLANGUAGE = %c%s%c,\n",
7201 language ? '"' : '(',
7202 language ? language : "unset",
7203 language ? '"' : ')');
7205 PERL_UNUSED_ARG(language);
7208 PerlIO_printf(Perl_error_log, "\tLC_ALL = %c%s%c,\n",
7210 lc_all ? lc_all : "unset",
7211 lc_all ? '"' : ')');
7213 for_all_individual_category_indexes(i) {
7214 const char * value = PerlEnv_getenv(category_names[i]);
7215 PerlIO_printf(Perl_error_log,
7219 value ? value : "unset",
7223 PerlIO_printf(Perl_error_log, "\tLANG = %c%s%c\n",
7225 lang ? lang : "unset",
7227 PerlIO_printf(Perl_error_log,
7228 " are supported and installed on your system.\n");
7233 /* A helper macro for the next function. Needed because would be called in two
7234 * places. Knows about the internal workings of the function */
7235 #define GET_DESCRIPTION(trial, name) \
7236 ((isNAME_C_OR_POSIX(name)) \
7237 ? "the standard locale" \
7238 : ((trial == (system_default_trial) \
7239 ? "the system default locale" \
7240 : "a fallback locale")))
7243 * Initialize locale awareness.
7246 Perl_init_i18nl10n(pTHX_ int printwarn)
7249 * 0 if not to output warning when setup locale is bad
7250 * 1 if to output warning based on value of PERL_BADLANG
7251 * >1 if to output regardless of PERL_BADLANG
7254 * 1 = set ok or not applicable,
7255 * 0 = fallback to a locale of lower priority
7256 * -1 = fallback to all locales failed, not even to the C locale
7258 * Under -DDEBUGGING, if the environment variable PERL_DEBUG_LOCALE_INIT is
7259 * set, debugging information is output.
7261 * This routine effectively does the following in most cases:
7263 * basic initialization;
7264 * asserts that the compiled tables are consistent;
7265 * initialize data structures;
7266 * make sure we are in the global locale;
7267 * setlocale(LC_ALL, "");
7268 * switch to per-thread locale if applicable;
7270 * The "" causes the locale to be set to what the environment variables at
7271 * the time say it should be.
7273 * To handle possible failures, the setlocale is expanded to be like:
7275 * trial_locale = pre-first-trial;
7276 * while (has_another_trial()) {
7277 * trial_locale = next_trial();
7278 * if setlocale(LC_ALL, trial_locale) {
7283 * had_failure = true;
7287 * if (had_failure) {
7289 * if (! ok) warn_still_more();
7292 * The first trial is either:
7293 * "" to examine the environment variables for the locale
7294 * NULL to use the values already set for the locale by the program
7295 * embedding this perl instantiation.
7297 * Something is wrong if this trial fails, but there is a sequence of
7298 * fallbacks to try should that happen. They are given in the enum below.
7300 * If there is no LC_ALL defined on the system, the setlocale() above is
7301 * replaced by a loop setting each individual category separately.
7303 * In a non-embeded environment, this code is executed exactly once. It
7304 * sets up the global locale environment. At the end, if some sort of
7305 * thread-safety is in effect, it will turn thread 0 into using that, with
7306 * the same locale as the global initially. thread 0 can then change its
7307 * locale at will without affecting the global one.
7309 * At destruction time, thread 0 will revert to the global locale as the
7310 * other threads die.
7312 * Care must be taken in an embedded environment. This code will be
7313 * executed for each instantiation. Since it changes the global locale, it
7314 * could clash with another running instantiation that isn't using
7315 * per-thread locales. perlembed suggests having the controlling program
7316 * set each instantiation's locale and set PERL_SKIP_LOCALE_INIT so this
7317 * code uses that without actually changing anything. Then the onus is on
7318 * the controlling program to prevent any races. The code below does
7319 * enough locking so as to prevent system calls from overwriting data
7320 * before it is safely copied here, but that isn't a general solution.
7325 PERL_UNUSED_ARG(printwarn);
7328 #else /* USE_LOCALE to near the end of the routine */
7334 const char * const language = PerlEnv_getenv("LANGUAGE");
7337 const char * const language = NULL; /* Unused placeholder */
7340 /* A later getenv() could zap this, so only use here */
7341 const char * const bad_lang_use_once = PerlEnv_getenv("PERL_BADLANG");
7343 const bool locwarn = (printwarn > 1
7345 && ( ! bad_lang_use_once
7347 /* disallow with "" or "0" */
7349 && strNE("0", bad_lang_use_once)))));
7352 # define DEBUG_LOCALE_INIT(a,b,c)
7355 DEBUG_INITIALIZATION_set(cBOOL(PerlEnv_getenv("PERL_DEBUG_LOCALE_INIT")));
7357 # define DEBUG_LOCALE_INIT(cat_index, locale, result) \
7358 DEBUG_L(PerlIO_printf(Perl_debug_log, "%s\n", \
7359 setlocale_debug_string_i(cat_index, locale, result)));
7362 assert(categories[LC_ALL_INDEX_] == LC_ALL);
7363 assert(strEQ(category_names[LC_ALL_INDEX_], "LC_ALL"));
7364 # ifdef USE_POSIX_2008_LOCALE
7365 assert(category_masks[LC_ALL_INDEX_] == LC_ALL_MASK);
7369 for_all_individual_category_indexes(i) {
7370 assert(category_name_lengths[i] == strlen(category_names[i]));
7373 # endif /* DEBUGGING */
7375 /* Initialize the per-thread mbrFOO() state variables. See POSIX.xs for
7376 * why these particular incantations are used. */
7378 memzero(&PL_mbrlen_ps, sizeof(PL_mbrlen_ps));
7381 memzero(&PL_mbrtowc_ps, sizeof(PL_mbrtowc_ps));
7384 wcrtomb(NULL, L'\0', &PL_wcrtomb_ps);
7386 # ifdef USE_PL_CURLOCALES
7388 for (unsigned int i = 0; i <= LC_ALL_INDEX_; i++) {
7389 PL_curlocales[i] = savepv("C");
7393 # ifdef USE_PL_CUR_LC_ALL
7395 PL_cur_LC_ALL = savepv("C");
7398 # if ! defined(PERL_LC_ALL_USES_NAME_VALUE_PAIRS) && defined(LC_ALL)
7402 /* If we haven't done so already, translate the LC_ALL positions of
7403 * categories into our internal indices. */
7404 if (map_LC_ALL_position_to_index[0] == LC_ALL_INDEX_) {
7406 /* Use this array, initialized by a config.h constant */
7407 int lc_all_category_positions[] = PERL_LC_ALL_CATEGORY_POSITIONS_INIT;
7408 STATIC_ASSERT_STMT( C_ARRAY_LENGTH(lc_all_category_positions)
7411 for (unsigned int i = 0;
7412 i < C_ARRAY_LENGTH(lc_all_category_positions);
7415 map_LC_ALL_position_to_index[i] =
7416 get_category_index(lc_all_category_positions[i]);
7423 # ifdef USE_POSIX_2008_LOCALE
7425 /* This is a global, so be sure to keep another instance from zapping it */
7427 if (PL_C_locale_obj) {
7431 PL_C_locale_obj = newlocale(LC_ALL_MASK, "C", (locale_t) 0);
7432 if (! PL_C_locale_obj) {
7434 locale_panic_(Perl_form(aTHX_
7435 "Cannot create POSIX 2008 C locale object"));
7439 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "created C object %p\n",
7443 /* Switch to using the POSIX 2008 interface now. This would happen below
7444 * anyway, but deferring it can lead to leaks of memory that would also get
7445 * malloc'd in the interim. We arbitrarily switch to the C locale,
7446 * overridden below */
7447 if (! uselocale(PL_C_locale_obj)) {
7448 locale_panic_(Perl_form(aTHX_
7449 "Can't uselocale(%p), LC_ALL supposed to"
7454 # ifdef MULTIPLICITY
7456 PL_cur_locale_obj = PL_C_locale_obj;
7461 /* Now initialize some data structures. This is entirely so that
7462 * later-executed code doesn't have to concern itself with things not being
7463 * initialized. Arbitrarily use the C locale (which we know has to exist
7464 * on the system). */
7466 # ifdef USE_LOCALE_NUMERIC
7468 PL_numeric_radix_sv = newSV(1);
7469 PL_underlying_radix_sv = newSV(1);
7470 Newxz(PL_numeric_name, 1, char); /* Single NUL character */
7473 # ifdef USE_LOCALE_COLLATE
7475 Newxz(PL_collation_name, 1, char);
7478 # ifdef USE_LOCALE_CTYPE
7480 Newxz(PL_ctype_name, 1, char);
7484 new_LC_ALL("C", true /* Don't shortcut */);
7486 /*===========================================================================*/
7488 /* Now ready to override the initialization with the values that the user
7489 * wants. This is done in the global locale as explained in the
7490 * introductory comments to this function */
7491 switch_to_global_locale();
7493 const char * const lc_all = PerlEnv_getenv("LC_ALL");
7494 const char * const lang = PerlEnv_getenv("LANG");
7496 /* We try each locale in the enum, in order, until we get one that works,
7497 * or exhaust the list. Normally the loop is executed just once.
7499 * Each enum value is +1 from the previous */
7502 environment_trial = 0, /* "" or NULL; code below assumes value
7503 0 is the first real trial */
7504 LC_ALL_trial, /* ENV{LC_ALL} */
7505 LANG_trial, /* ENV{LANG} */
7506 system_default_trial, /* Windows .ACP */
7507 C_trial, /* C locale */
7512 unsigned int already_checked = 0;
7513 const char * checked[C_trial];
7516 const char * lc_all_string;
7518 const char * curlocales[LC_ALL_INDEX_];
7521 /* Loop through the initial setting and all the possible fallbacks,
7522 * breaking out of the loop on success */
7523 trial = dummy_trial;
7524 while (trial != beyond_final_trial) {
7526 /* Each time through compute the next trial to use based on the one in
7527 * the previous iteration and switch to the new one. This enforces the
7528 * order in which the fallbacks are applied */
7530 trial = (trials) ((int) trial + 1); /* Casts are needed for g++ */
7532 const char * locale = NULL;
7534 /* Set up the parameters for this trial */
7537 locale_panic_("Unexpectedly got 'dummy_trial");
7540 case environment_trial:
7541 /* This is either "" to get the values from the environment, or
7542 * NULL if the calling program has initialized the values already.
7544 locale = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
7550 if (! lc_all || strEQ(lc_all, "")) {
7551 continue; /* No-op */
7558 if (! lang || strEQ(lang, "")) {
7559 continue; /* No-op */
7565 case system_default_trial:
7567 # if ! defined(WIN32) || ! defined(LC_ALL)
7569 continue; /* No-op */
7572 /* For Windows, we also try the system default locale before "C".
7573 * (If there exists a Windows without LC_ALL we skip this because
7574 * it gets too complicated. For those, "C" is the next fallback
7584 case beyond_final_trial:
7585 continue; /* No-op, causes loop to exit */
7588 /* If the locale is a substantive name, don't try the same locale
7590 if (locale && strNE(locale, "")) {
7591 for (unsigned int i = 0; i < already_checked; i++) {
7592 if (strEQ(checked[i], locale)) {
7597 /* And, for future iterations, indicate we've tried this locale */
7598 assert(already_checked < C_ARRAY_LENGTH(checked));
7599 checked[already_checked] = savepv(locale);
7600 SAVEFREEPV(checked[already_checked]);
7606 STDIZED_SETLOCALE_LOCK;
7607 lc_all_string = savepv(stdized_setlocale(LC_ALL, locale));
7608 STDIZED_SETLOCALE_UNLOCK;
7610 DEBUG_LOCALE_INIT(LC_ALL_INDEX_, locale, lc_all_string);
7612 if (LIKELY(lc_all_string)) { /* Succeeded */
7617 if (trial == 0 && locwarn) {
7618 PerlIO_printf(Perl_error_log,
7619 "perl: warning: Setting locale failed.\n");
7620 output_check_environment_warning(language, lc_all, lang);
7623 # else /* Below is ! LC_ALL */
7625 bool setlocale_failure = FALSE; /* This trial hasn't failed so far */
7626 bool dowarn = trial == 0 && locwarn;
7628 for_all_individual_category_indexes(j) {
7629 STDIZED_SETLOCALE_LOCK;
7630 curlocales[j] = savepv(stdized_setlocale(categories[j], locale));
7631 STDIZED_SETLOCALE_UNLOCK;
7633 DEBUG_LOCALE_INIT(j, locale, curlocales[j]);
7635 if (UNLIKELY(! curlocales[j])) {
7636 setlocale_failure = TRUE;
7638 /* If are going to warn below, continue to loop so all failures
7639 * are included in the message */
7646 if (LIKELY(! setlocale_failure)) { /* All succeeded */
7648 break; /* Exit trial_locales loop */
7651 /* Here, this trial failed */
7654 PerlIO_printf(Perl_error_log,
7655 "perl: warning: Setting locale failed for the categories:\n");
7657 for_all_individual_category_indexes(j) {
7658 if (! curlocales[j]) {
7659 PerlIO_printf(Perl_error_log, "\t%s\n", category_names[j]);
7663 output_check_environment_warning(language, lc_all, lang);
7664 } /* end of warning on first failure */
7666 # endif /* LC_ALL */
7668 } /* end of looping through the trial locales */
7670 /* If we had to do more than the first trial, it means that one failed, and
7671 * we may need to output a warning, and, if none worked, do more */
7672 if (UNLIKELY(trial != 0)) {
7674 const char * description = "a fallback locale";
7675 const char * name = NULL;;
7677 /* If we didn't find a good fallback, list all we tried */
7678 if (! ok && already_checked > 0) {
7679 PerlIO_printf(Perl_error_log, "perl: warning: Failed to fall"
7681 if (already_checked > 1) { /* more than one was tried */
7682 PerlIO_printf(Perl_error_log, "any of:\n");
7685 while (already_checked > 0) {
7686 name = checked[--already_checked];
7687 description = GET_DESCRIPTION(trial, name);
7688 PerlIO_printf(Perl_error_log, "%s (\"%s\")\n",
7695 /* Here, a fallback worked. So we have saved its name, and the
7696 * trial that succeeded is still valid */
7698 const char * individ_locales[LC_ALL_INDEX_] = { NULL };
7700 /* Even though we know the valid string for LC_ALL that worked,
7701 * translate it into our internal format, which is the
7702 * name=value pairs notation. This is easier for a human to
7703 * decipher than the positional notation. Some platforms
7704 * can return "C C C C C C" for LC_ALL. This code also
7705 * standardizes that result into plain "C". */
7706 switch (parse_LC_ALL_string(lc_all_string,
7707 (const char **) &individ_locales,
7709 false, /* Return only [0] if
7711 false, /* Don't panic on error */
7716 /* Here, the parse failed, which shouldn't happen, but if
7717 * it does, we have an easy fallback that allows us to keep
7719 name = lc_all_string;
7722 case no_array: /* The original is a single locale */
7723 name = lc_all_string;
7726 case only_element_0: /* element[0] is a single locale valid
7727 for all categories */
7728 SAVEFREEPV(individ_locales[0]);
7729 name = individ_locales[0];
7733 name = calculate_LC_ALL_string(individ_locales,
7737 for_all_individual_category_indexes(j) {
7738 Safefree(individ_locales[j]);
7742 name = calculate_LC_ALL_string(curlocales,
7747 description = GET_DESCRIPTION(trial, name);
7751 /* Nothing seems to be working, yet we want to continue
7752 * executing. It may well be that locales are mostly
7753 * irrelevant to this particular program, and there must be
7754 * some locale underlying the program. Figure it out as best
7755 * we can, by querying the system's current locale */
7759 STDIZED_SETLOCALE_LOCK;
7760 name = stdized_setlocale(LC_ALL, NULL);
7761 STDIZED_SETLOCALE_UNLOCK;
7763 if (UNLIKELY(! name)) {
7764 name = "locale name not determinable";
7767 # else /* Below is ! LC_ALL */
7769 const char * system_locales[LC_ALL_INDEX_] = { NULL };
7771 for_all_individual_category_indexes(j) {
7772 STDIZED_SETLOCALE_LOCK;
7773 system_locales[j] = savepv(stdized_setlocale(categories[j],
7775 STDIZED_SETLOCALE_UNLOCK;
7777 if (UNLIKELY(! system_locales[j])) {
7778 system_locales[j] = "not determinable";
7782 /* We use the name=value form for the string, as that is more
7783 * human readable than the positional notation */
7784 name = calculate_LC_ALL_string(system_locales,
7788 description = "what the system says";
7790 for_all_individual_category_indexes(j) {
7791 Safefree(system_locales[j]);
7796 PerlIO_printf(Perl_error_log,
7797 "perl: warning: Falling back to %s (\"%s\").\n",
7800 /* Here, ok being true indicates that the first attempt failed, but
7801 * a fallback succeeded; false => nothing working. Translate to
7802 * API return values. */
7809 give_perl_locale_control(lc_all_string, __LINE__);
7810 Safefree(lc_all_string);
7814 give_perl_locale_control((const char **) &curlocales, __LINE__);
7816 for_all_individual_category_indexes(j) {
7817 Safefree(curlocales[j]);
7821 # if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
7823 /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
7824 * locale is UTF-8. give_perl_locale_control() just above has already
7825 * calculated the latter value and saved it in PL_in_utf8_CTYPE_locale. If
7826 * both PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
7827 * are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer on
7828 * STDIN, STDOUT, STDERR, _and_ the default open discipline. */
7829 PL_utf8locale = PL_in_utf8_CTYPE_locale;
7831 /* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
7832 This is an alternative to using the -C command line switch
7833 (the -C if present will override this). */
7835 const char *p = PerlEnv_getenv("PERL_UNICODE");
7836 PL_unicode = p ? parse_unicode_opts(&p) : 0;
7837 if (PL_unicode & PERL_UNICODE_UTF8CACHEASSERT_FLAG)
7842 # if defined(USE_POSIX_2008_LOCALE) && defined(MULTIPLICITY)
7843 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
7844 "finished Perl_init_i18nl10n; actual obj=%p,"
7845 " expected obj=%p, initial=%s\n",
7846 uselocale(0), PL_cur_locale_obj,
7847 get_LC_ALL_display()));
7850 /* So won't continue to output stuff */
7851 DEBUG_INITIALIZATION_set(FALSE);
7853 #endif /* USE_LOCALE */
7858 #undef GET_DESCRIPTION
7859 #ifdef USE_LOCALE_COLLATE
7862 S_compute_collxfrm_coefficients(pTHX)
7865 /* A locale collation definition includes primary, secondary, tertiary,
7866 * etc. weights for each character. To sort, the primary weights are used,
7867 * and only if they compare equal, then the secondary weights are used, and
7868 * only if they compare equal, then the tertiary, etc.
7870 * strxfrm() works by taking the input string, say ABC, and creating an
7871 * output transformed string consisting of first the primary weights,
7872 * A¹B¹C¹ followed by the secondary ones, A²B²C²; and then the tertiary,
7873 * etc, yielding A¹B¹C¹ A²B²C² A³B³C³ .... Some characters may not have
7874 * weights at every level. In our example, let's say B doesn't have a
7875 * tertiary weight, and A doesn't have a secondary weight. The constructed
7876 * string is then going to be
7877 * A¹B¹C¹ B²C² A³C³ ....
7878 * This has the desired effect that strcmp() will look at the secondary or
7879 * tertiary weights only if the strings compare equal at all higher
7880 * priority weights. The spaces shown here, like in
7882 * are not just for readability. In the general case, these must actually
7883 * be bytes, which we will call here 'separator weights'; and they must be
7884 * smaller than any other weight value, but since these are C strings, only
7885 * the terminating one can be a NUL (some implementations may include a
7886 * non-NUL separator weight just before the NUL). Implementations tend to
7887 * reserve 01 for the separator weights. They are needed so that a shorter
7888 * string's secondary weights won't be misconstrued as primary weights of a
7889 * longer string, etc. By making them smaller than any other weight, the
7890 * shorter string will sort first. (Actually, if all secondary weights are
7891 * smaller than all primary ones, there is no need for a separator weight
7892 * between those two levels, etc.)
7894 * The length of the transformed string is roughly a linear function of the
7895 * input string. It's not exactly linear because some characters don't
7896 * have weights at all levels. When we call strxfrm() we have to allocate
7897 * some memory to hold the transformed string. The calculations below try
7898 * to find coefficients 'm' and 'b' for this locale so that m*x + b equals
7899 * how much space we need, given the size of the input string in 'x'. If
7900 * we calculate too small, we increase the size as needed, and call
7901 * strxfrm() again, but it is better to get it right the first time to
7902 * avoid wasted expensive string transformations.
7904 * We use the string below to find how long the transformation of it is.
7905 * Almost all locales are supersets of ASCII, or at least the ASCII
7906 * letters. We use all of them, half upper half lower, because if we used
7907 * fewer, we might hit just the ones that are outliers in a particular
7908 * locale. Most of the strings being collated will contain a preponderance
7909 * of letters, and even if they are above-ASCII, they are likely to have
7910 * the same number of weight levels as the ASCII ones. It turns out that
7911 * digits tend to have fewer levels, and some punctuation has more, but
7912 * those are relatively sparse in text, and khw believes this gives a
7913 * reasonable result, but it could be changed if experience so dictates. */
7914 const char longer[] = "ABCDEFGHIJKLMnopqrstuvwxyz";
7915 char * x_longer; /* Transformed 'longer' */
7916 Size_t x_len_longer; /* Length of 'x_longer' */
7918 char * x_shorter; /* We also transform a substring of 'longer' */
7919 Size_t x_len_shorter;
7921 PL_in_utf8_COLLATE_locale = (PL_collation_standard)
7923 : is_locale_utf8(PL_collation_name);
7924 PL_strxfrm_NUL_replacement = '\0';
7925 PL_strxfrm_max_cp = 0;
7927 /* mem_collxfrm_() is used get the transformation (though here we are
7928 * interested only in its length). It is used because it has the
7929 * intelligence to handle all cases, but to work, it needs some values of
7930 * 'm' and 'b' to get it started. For the purposes of this calculation we
7931 * use a very conservative estimate of 'm' and 'b'. This assumes a weight
7932 * can be multiple bytes, enough to hold any UV on the platform, and there
7933 * are 5 levels, 4 weight bytes, and a trailing NUL. */
7934 PL_collxfrm_base = 5;
7935 PL_collxfrm_mult = 5 * sizeof(UV);
7937 /* Find out how long the transformation really is */
7938 x_longer = mem_collxfrm_(longer,
7942 /* We avoid converting to UTF-8 in the called
7943 * function by telling it the string is in UTF-8
7944 * if the locale is a UTF-8 one. Since the string
7945 * passed here is invariant under UTF-8, we can
7946 * claim it's UTF-8 even if it isn't. */
7947 PL_in_utf8_COLLATE_locale);
7950 /* Find out how long the transformation of a substring of 'longer' is.
7951 * Together the lengths of these transformations are sufficient to
7952 * calculate 'm' and 'b'. The substring is all of 'longer' except the
7953 * first character. This minimizes the chances of being swayed by outliers
7955 x_shorter = mem_collxfrm_(longer + 1,
7958 PL_in_utf8_COLLATE_locale);
7959 Safefree(x_shorter);
7961 /* If the results are nonsensical for this simple test, the whole locale
7962 * definition is suspect. Mark it so that locale collation is not active
7963 * at all for it. XXX Should we warn? */
7964 if ( x_len_shorter == 0
7965 || x_len_longer == 0
7966 || x_len_shorter >= x_len_longer)
7968 PL_collxfrm_mult = 0;
7969 PL_collxfrm_base = 1;
7970 DEBUG_L(PerlIO_printf(Perl_debug_log,
7971 "Disabling locale collation for LC_COLLATE='%s';"
7972 " length for shorter sample=%zu; longer=%zu\n",
7973 PL_collation_name, x_len_shorter, x_len_longer));
7976 SSize_t base; /* Temporary */
7978 /* We have both: m * strlen(longer) + b = x_len_longer
7979 * m * strlen(shorter) + b = x_len_shorter;
7980 * subtracting yields:
7981 * m * (strlen(longer) - strlen(shorter))
7982 * = x_len_longer - x_len_shorter
7983 * But we have set things up so that 'shorter' is 1 byte smaller than
7985 * m = x_len_longer - x_len_shorter
7987 * But if something went wrong, make sure the multiplier is at least 1.
7989 if (x_len_longer > x_len_shorter) {
7990 PL_collxfrm_mult = (STRLEN) x_len_longer - x_len_shorter;
7993 PL_collxfrm_mult = 1;
7998 * but in case something has gone wrong, make sure it is non-negative
8000 base = x_len_longer - PL_collxfrm_mult * (sizeof(longer) - 1);
8005 /* Add 1 for the trailing NUL */
8006 PL_collxfrm_base = base + 1;
8009 DEBUG_L(PerlIO_printf(Perl_debug_log,
8010 "?UTF-8 locale=%d; x_len_shorter=%zu, "
8012 " collate multipler=%zu, collate base=%zu\n",
8013 PL_in_utf8_COLLATE_locale,
8014 x_len_shorter, x_len_longer,
8015 PL_collxfrm_mult, PL_collxfrm_base));
8019 Perl_mem_collxfrm_(pTHX_ const char *input_string,
8020 STRLEN len, /* Length of 'input_string' */
8021 STRLEN *xlen, /* Set to length of returned string
8022 (not including the collation index
8024 bool utf8 /* Is the input in UTF-8? */
8027 /* mem_collxfrm_() is like strxfrm() but with two important differences.
8028 * First, it handles embedded NULs. Second, it allocates a bit more memory
8029 * than needed for the transformed data itself. The real transformed data
8030 * begins at offset COLLXFRM_HDR_LEN. *xlen is set to the length of that,
8031 * and doesn't include the collation index size.
8033 * It is the caller's responsibility to eventually free the memory returned
8036 * Please see sv_collxfrm() to see how this is used. */
8038 # define COLLXFRM_HDR_LEN sizeof(PL_collation_ix)
8040 char * s = (char *) input_string;
8041 STRLEN s_strlen = strlen(input_string);
8043 STRLEN xAlloc; /* xalloc is a reserved word in VC */
8044 STRLEN length_in_chars;
8045 bool first_time = TRUE; /* Cleared after first loop iteration */
8047 # ifdef USE_LOCALE_CTYPE
8048 const char * orig_CTYPE_locale = NULL;
8051 # if defined(USE_POSIX_2008_LOCALE) && defined HAS_STRXFRM_L
8052 locale_t constructed_locale = (locale_t) 0;
8055 PERL_ARGS_ASSERT_MEM_COLLXFRM_;
8057 /* Must be NUL-terminated */
8058 assert(*(input_string + len) == '\0');
8060 if (PL_collxfrm_mult == 0) { /* unknown or bad */
8061 if (PL_collxfrm_base != 0) { /* bad collation => skip */
8062 DEBUG_L(PerlIO_printf(Perl_debug_log,
8063 "mem_collxfrm_: locale's collation is defective\n"));
8067 /* (mult, base) == (0,0) means we need to calculate mult and base
8068 * before proceeding */
8069 S_compute_collxfrm_coefficients(aTHX);
8072 /* Replace any embedded NULs with the control that sorts before any others.
8073 * This will give as good as possible results on strings that don't
8074 * otherwise contain that character, but otherwise there may be
8075 * less-than-perfect results with that character and NUL. This is
8076 * unavoidable unless we replace strxfrm with our own implementation. */
8077 if (UNLIKELY(s_strlen < len)) { /* Only execute if there is an embedded
8081 STRLEN sans_nuls_len;
8082 int try_non_controls;
8083 char this_replacement_char[] = "?\0"; /* Room for a two-byte string,
8084 making sure 2nd byte is NUL.
8086 STRLEN this_replacement_len;
8088 /* If we don't know what non-NUL control character sorts lowest for
8089 * this locale, find it */
8090 if (PL_strxfrm_NUL_replacement == '\0') {
8092 char * cur_min_x = NULL; /* The min_char's xfrm, (except it also
8093 includes the collation index
8096 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "Looking to replace NUL\n"));
8098 /* Unlikely, but it may be that no control will work to replace
8099 * NUL, in which case we instead look for any character. Controls
8100 * are preferred because collation order is, in general, context
8101 * sensitive, with adjoining characters affecting the order, and
8102 * controls are less likely to have such interactions, allowing the
8103 * NUL-replacement to stand on its own. (Another way to look at it
8104 * is to imagine what would happen if the NUL were replaced by a
8105 * combining character; it wouldn't work out all that well.) */
8106 for (try_non_controls = 0;
8107 try_non_controls < 2;
8111 # ifdef USE_LOCALE_CTYPE
8113 /* In this case we use isCNTRL_LC() below, which relies on
8114 * LC_CTYPE, so that must be switched to correspond with the
8115 * LC_COLLATE locale */
8116 if (! try_non_controls && ! PL_in_utf8_COLLATE_locale) {
8117 orig_CTYPE_locale = toggle_locale_c(LC_CTYPE,
8121 /* Look through all legal code points (NUL isn't) */
8122 for (j = 1; j < 256; j++) {
8123 char * x; /* j's xfrm plus collation index */
8124 STRLEN x_len; /* length of 'x' */
8125 STRLEN trial_len = 1;
8126 char cur_source[] = { '\0', '\0' };
8128 /* Skip non-controls the first time through the loop. The
8129 * controls in a UTF-8 locale are the L1 ones */
8130 if (! try_non_controls && (PL_in_utf8_COLLATE_locale)
8137 /* Create a 1-char string of the current code point */
8138 cur_source[0] = (char) j;
8140 /* Then transform it */
8141 x = mem_collxfrm_(cur_source, trial_len, &x_len,
8142 0 /* The string is not in UTF-8 */);
8144 /* Ignore any character that didn't successfully transform.
8150 /* If this character's transformation is lower than
8151 * the current lowest, this one becomes the lowest */
8152 if ( cur_min_x == NULL
8153 || strLT(x + COLLXFRM_HDR_LEN,
8154 cur_min_x + COLLXFRM_HDR_LEN))
8156 PL_strxfrm_NUL_replacement = j;
8157 Safefree(cur_min_x);
8163 } /* end of loop through all 255 characters */
8165 # ifdef USE_LOCALE_CTYPE
8166 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
8169 /* Stop looking if found */
8174 /* Unlikely, but possible, if there aren't any controls that
8175 * work in the locale, repeat the loop, looking for any
8176 * character that works */
8177 DEBUG_L(PerlIO_printf(Perl_debug_log,
8178 "mem_collxfrm_: No control worked. Trying non-controls\n"));
8179 } /* End of loop to try first the controls, then any char */
8182 DEBUG_L(PerlIO_printf(Perl_debug_log,
8183 "mem_collxfrm_: Couldn't find any character to replace"
8184 " embedded NULs in locale %s with", PL_collation_name));
8188 DEBUG_L(PerlIO_printf(Perl_debug_log,
8189 "mem_collxfrm_: Replacing embedded NULs in locale %s with "
8190 "0x%02X\n", PL_collation_name, PL_strxfrm_NUL_replacement));
8192 Safefree(cur_min_x);
8193 } /* End of determining the character that is to replace NULs */
8195 /* If the replacement is variant under UTF-8, it must match the
8196 * UTF8-ness of the original */
8197 if ( ! UVCHR_IS_INVARIANT(PL_strxfrm_NUL_replacement) && utf8) {
8198 this_replacement_char[0] =
8199 UTF8_EIGHT_BIT_HI(PL_strxfrm_NUL_replacement);
8200 this_replacement_char[1] =
8201 UTF8_EIGHT_BIT_LO(PL_strxfrm_NUL_replacement);
8202 this_replacement_len = 2;
8205 this_replacement_char[0] = PL_strxfrm_NUL_replacement;
8206 /* this_replacement_char[1] = '\0' was done at initialization */
8207 this_replacement_len = 1;
8210 /* The worst case length for the replaced string would be if every
8211 * character in it is NUL. Multiply that by the length of each
8212 * replacement, and allow for a trailing NUL */
8213 sans_nuls_len = (len * this_replacement_len) + 1;
8214 Newx(sans_nuls, sans_nuls_len, char);
8217 /* Replace each NUL with the lowest collating control. Loop until have
8218 * exhausted all the NULs */
8219 while (s + s_strlen < e) {
8220 my_strlcat(sans_nuls, s, sans_nuls_len);
8222 /* Do the actual replacement */
8223 my_strlcat(sans_nuls, this_replacement_char, sans_nuls_len);
8225 /* Move past the input NUL */
8227 s_strlen = strlen(s);
8230 /* And add anything that trails the final NUL */
8231 my_strlcat(sans_nuls, s, sans_nuls_len);
8233 /* Switch so below we transform this modified string */
8236 } /* End of replacing NULs */
8238 /* Make sure the UTF8ness of the string and locale match */
8239 if (utf8 != PL_in_utf8_COLLATE_locale) {
8240 /* XXX convert above Unicode to 10FFFF? */
8241 const char * const t = s; /* Temporary so we can later find where the
8244 /* Here they don't match. Change the string's to be what the locale is
8247 if (! utf8) { /* locale is UTF-8, but input isn't; upgrade the input */
8248 s = (char *) bytes_to_utf8((const U8 *) s, &len);
8251 else { /* locale is not UTF-8; but input is; downgrade the input */
8253 s = (char *) bytes_from_utf8((const U8 *) s, &len, &utf8);
8255 /* If the downgrade was successful we are done, but if the input
8256 * contains things that require UTF-8 to represent, have to do
8257 * damage control ... */
8258 if (UNLIKELY(utf8)) {
8260 /* What we do is construct a non-UTF-8 string with
8261 * 1) the characters representable by a single byte converted
8262 * to be so (if necessary);
8263 * 2) and the rest converted to collate the same as the
8264 * highest collating representable character. That makes
8265 * them collate at the end. This is similar to how we
8266 * handle embedded NULs, but we use the highest collating
8267 * code point instead of the smallest. Like the NUL case,
8268 * this isn't perfect, but is the best we can reasonably
8269 * do. Every above-255 code point will sort the same as
8270 * the highest-sorting 0-255 code point. If that code
8271 * point can combine in a sequence with some other code
8272 * points for weight calculations, us changing something to
8273 * be it can adversely affect the results. But in most
8274 * cases, it should work reasonably. And note that this is
8275 * really an illegal situation: using code points above 255
8276 * on a locale where only 0-255 are valid. If two strings
8277 * sort entirely equal, then the sort order for the
8278 * above-255 code points will be in code point order. */
8282 /* If we haven't calculated the code point with the maximum
8283 * collating order for this locale, do so now */
8284 if (! PL_strxfrm_max_cp) {
8287 /* The current transformed string that collates the
8288 * highest (except it also includes the prefixed collation
8290 char * cur_max_x = NULL;
8292 /* Look through all legal code points (NUL isn't) */
8293 for (j = 1; j < 256; j++) {
8296 char cur_source[] = { '\0', '\0' };
8298 /* Create a 1-char string of the current code point */
8299 cur_source[0] = (char) j;
8301 /* Then transform it */
8302 x = mem_collxfrm_(cur_source, 1, &x_len, FALSE);
8304 /* If something went wrong (which it shouldn't), just
8305 * ignore this code point */
8310 /* If this character's transformation is higher than
8311 * the current highest, this one becomes the highest */
8312 if ( cur_max_x == NULL
8313 || strGT(x + COLLXFRM_HDR_LEN,
8314 cur_max_x + COLLXFRM_HDR_LEN))
8316 PL_strxfrm_max_cp = j;
8317 Safefree(cur_max_x);
8326 DEBUG_L(PerlIO_printf(Perl_debug_log,
8327 "mem_collxfrm_: Couldn't find any character to"
8328 " replace above-Latin1 chars in locale %s with",
8329 PL_collation_name));
8333 DEBUG_L(PerlIO_printf(Perl_debug_log,
8334 "mem_collxfrm_: highest 1-byte collating character"
8335 " in locale %s is 0x%02X\n",
8337 PL_strxfrm_max_cp));
8339 Safefree(cur_max_x);
8342 /* Here we know which legal code point collates the highest.
8343 * We are ready to construct the non-UTF-8 string. The length
8344 * will be at least 1 byte smaller than the input string
8345 * (because we changed at least one 2-byte character into a
8346 * single byte), but that is eaten up by the trailing NUL */
8352 char * e = (char *) t + len;
8354 for (i = 0; i < len; i+= UTF8SKIP(t + i)) {
8356 if (UTF8_IS_INVARIANT(cur_char)) {
8359 else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(t + i, e)) {
8360 s[d++] = EIGHT_BIT_UTF8_TO_NATIVE(cur_char, t[i+1]);
8362 else { /* Replace illegal cp with highest collating
8364 s[d++] = PL_strxfrm_max_cp;
8368 Renew(s, d, char); /* Free up unused space */
8373 /* Here, we have constructed a modified version of the input. It could
8374 * be that we already had a modified copy before we did this version.
8375 * If so, that copy is no longer needed */
8376 if (t != input_string) {
8381 length_in_chars = (utf8)
8382 ? utf8_length((U8 *) s, (U8 *) s + len)
8385 /* The first element in the output is the collation id, used by
8386 * sv_collxfrm(); then comes the space for the transformed string. The
8387 * equation should give us a good estimate as to how much is needed */
8388 xAlloc = COLLXFRM_HDR_LEN
8390 + (PL_collxfrm_mult * length_in_chars);
8391 Newx(xbuf, xAlloc, char);
8392 if (UNLIKELY(! xbuf)) {
8393 DEBUG_L(PerlIO_printf(Perl_debug_log,
8394 "mem_collxfrm_: Couldn't malloc %zu bytes\n", xAlloc));
8398 /* Store the collation id */
8399 *(PERL_UINTMAX_T *)xbuf = PL_collation_ix;
8401 # if defined(USE_POSIX_2008_LOCALE) && defined HAS_STRXFRM_L
8402 # ifdef USE_LOCALE_CTYPE
8404 constructed_locale = newlocale(LC_CTYPE_MASK, PL_collation_name,
8405 duplocale(use_curlocale_scratch()));
8408 constructed_locale = duplocale(use_curlocale_scratch());
8411 # define my_strxfrm(dest, src, n) strxfrm_l(dest, src, n, \
8413 # define CLEANUP_STRXFRM \
8415 if (constructed_locale != (locale_t) 0) \
8416 freelocale(constructed_locale); \
8419 # define my_strxfrm(dest, src, n) strxfrm(dest, src, n)
8420 # ifdef USE_LOCALE_CTYPE
8422 orig_CTYPE_locale = toggle_locale_c(LC_CTYPE, PL_collation_name);
8424 # define CLEANUP_STRXFRM \
8425 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale)
8427 # define CLEANUP_STRXFRM NOOP
8431 /* Then the transformation of the input. We loop until successful, or we
8436 *xlen = my_strxfrm(xbuf + COLLXFRM_HDR_LEN,
8438 xAlloc - COLLXFRM_HDR_LEN);
8441 /* If the transformed string occupies less space than we told strxfrm()
8442 * was available, it means it transformed the whole string. */
8443 if (*xlen < xAlloc - COLLXFRM_HDR_LEN) {
8445 /* But there still could have been a problem */
8447 DEBUG_L(PerlIO_printf(Perl_debug_log,
8448 "strxfrm failed for LC_COLLATE=%s; errno=%d, input=%s\n",
8449 PL_collation_name, errno,
8450 _byte_dump_string((U8 *) s, len, 0)));
8454 /* Here, the transformation was successful. Some systems include a
8455 * trailing NUL in the returned length. Ignore it, using a loop in
8456 * case multiple trailing NULs are returned. */
8458 && *(xbuf + COLLXFRM_HDR_LEN + (*xlen) - 1) == '\0')
8463 /* If the first try didn't get it, it means our prediction was low.
8464 * Modify the coefficients so that we predict a larger value in any
8465 * future transformations */
8467 STRLEN needed = *xlen + 1; /* +1 For trailing NUL */
8468 STRLEN computed_guess = PL_collxfrm_base
8469 + (PL_collxfrm_mult * length_in_chars);
8471 /* On zero-length input, just keep current slope instead of
8473 const STRLEN new_m = (length_in_chars != 0)
8474 ? needed / length_in_chars
8477 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8478 "initial size of %zu bytes for a length "
8479 "%zu string was insufficient, %zu needed\n",
8480 computed_guess, length_in_chars, needed));
8482 /* If slope increased, use it, but discard this result for
8483 * length 1 strings, as we can't be sure that it's a real slope
8485 if (length_in_chars > 1 && new_m > PL_collxfrm_mult) {
8489 STRLEN old_m = PL_collxfrm_mult;
8490 STRLEN old_b = PL_collxfrm_base;
8494 PL_collxfrm_mult = new_m;
8495 PL_collxfrm_base = 1; /* +1 For trailing NUL */
8496 computed_guess = PL_collxfrm_base
8497 + (PL_collxfrm_mult * length_in_chars);
8498 if (computed_guess < needed) {
8499 PL_collxfrm_base += needed - computed_guess;
8502 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8503 "slope is now %zu; was %zu, base "
8504 "is now %zu; was %zu\n",
8505 PL_collxfrm_mult, old_m,
8506 PL_collxfrm_base, old_b));
8508 else { /* Slope didn't change, but 'b' did */
8509 const STRLEN new_b = needed
8512 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8513 "base is now %zu; was %zu\n", new_b, PL_collxfrm_base));
8514 PL_collxfrm_base = new_b;
8521 if (UNLIKELY(*xlen >= PERL_INT_MAX)) {
8522 DEBUG_L(PerlIO_printf(Perl_debug_log,
8523 "mem_collxfrm_: Needed %zu bytes, max permissible is %u\n",
8524 *xlen, PERL_INT_MAX));
8528 /* A well-behaved strxfrm() returns exactly how much space it needs
8529 * (usually not including the trailing NUL) when it fails due to not
8530 * enough space being provided. Assume that this is the case unless
8531 * it's been proven otherwise */
8532 if (LIKELY(PL_strxfrm_is_behaved) && first_time) {
8533 xAlloc = *xlen + COLLXFRM_HDR_LEN + 1;
8535 else { /* Here, either:
8536 * 1) The strxfrm() has previously shown bad behavior; or
8537 * 2) It isn't the first time through the loop, which means
8538 * that the strxfrm() is now showing bad behavior, because
8539 * we gave it what it said was needed in the previous
8540 * iteration, and it came back saying it needed still more.
8541 * (Many versions of cygwin fit this. When the buffer size
8542 * isn't sufficient, they return the input size instead of
8543 * how much is needed.)
8544 * Increase the buffer size by a fixed percentage and try again.
8546 xAlloc += (xAlloc / 4) + 1;
8547 PL_strxfrm_is_behaved = FALSE;
8549 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8550 "mem_collxfrm_ required more space than previously"
8551 " calculated for locale %s, trying again with new"
8553 PL_collation_name, COLLXFRM_HDR_LEN,
8554 xAlloc - COLLXFRM_HDR_LEN));
8557 Renew(xbuf, xAlloc, char);
8558 if (UNLIKELY(! xbuf)) {
8559 DEBUG_L(PerlIO_printf(Perl_debug_log,
8560 "mem_collxfrm_: Couldn't realloc %zu bytes\n", xAlloc));
8569 DEBUG_L(print_collxfrm_input_and_return(s, s + len, xbuf, *xlen, utf8));
8571 /* Free up unneeded space; retain enough for trailing NUL */
8572 Renew(xbuf, COLLXFRM_HDR_LEN + *xlen + 1, char);
8574 if (s != input_string) {
8583 DEBUG_L(print_collxfrm_input_and_return(s, s + len, NULL, 0, utf8));
8586 if (s != input_string) {
8597 S_print_collxfrm_input_and_return(pTHX_
8605 PERL_ARGS_ASSERT_PRINT_COLLXFRM_INPUT_AND_RETURN;
8607 PerlIO_printf(Perl_debug_log,
8608 "mem_collxfrm_[ix %" UVuf "] for locale '%s':\n"
8609 " input=%s\n return=%s\n return len=%zu\n",
8610 (UV) PL_collation_ix, PL_collation_name,
8611 get_displayable_string(s, e, is_utf8),
8616 : _byte_dump_string((U8 *) xbuf + COLLXFRM_HDR_LEN,
8621 # endif /* DEBUGGING */
8624 Perl_strxfrm(pTHX_ SV * src)
8626 PERL_ARGS_ASSERT_STRXFRM;
8628 /* For use by POSIX::strxfrm(). If they differ, toggle LC_CTYPE to
8629 * LC_COLLATE to avoid potential mojibake.
8631 * If we can't calculate a collation, 'src' is instead returned, so that
8632 * future comparisons will be by code point order */
8634 # ifdef USE_LOCALE_CTYPE
8636 const char * orig_ctype = toggle_locale_c(LC_CTYPE,
8637 querylocale_c(LC_COLLATE));
8643 const char *p = SvPV_const(src, srclen);
8644 const U32 utf8_flag = SvUTF8(src);
8645 char *d = mem_collxfrm_(p, srclen, &dstlen, cBOOL(utf8_flag));
8647 assert(utf8_flag == 0 || utf8_flag == SVf_UTF8);
8651 dst =newSVpvn_flags(d + COLLXFRM_HDR_LEN,
8652 dstlen, SVs_TEMP|utf8_flag);
8656 # ifdef USE_LOCALE_CTYPE
8658 restore_toggled_locale_c(LC_CTYPE, orig_ctype);
8665 #endif /* USE_LOCALE_COLLATE */
8669 S_toggle_locale_i(pTHX_ const locale_category_index cat_index,
8670 const char * new_locale,
8671 const line_t caller_line)
8673 PERL_ARGS_ASSERT_TOGGLE_LOCALE_I;
8674 assert(cat_index <= LC_ALL_INDEX_);
8676 /* Changes the locale for the category specified by 'index' to 'new_locale,
8677 * if they aren't already the same.
8679 * Returns a copy of the name of the original locale for 'cat_index'
8680 * so can be switched back to with the companion function
8681 * restore_toggled_locale_i(), (NULL if no restoral is necessary.) */
8683 /* Find the original locale of the category we may need to change, so that
8684 * it can be restored to later */
8685 const char * locale_to_restore_to = querylocale_i(cat_index);
8687 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8688 "Entering toggle_locale_i: index=%d(%s)," \
8689 " wanted=%s, actual=%s; called from %" LINE_Tf \
8690 "\n", cat_index, category_names[cat_index],
8691 new_locale, locale_to_restore_to, caller_line));
8693 if (! locale_to_restore_to) {
8694 locale_panic_via_(Perl_form(aTHX_
8695 "Could not find current %s locale",
8696 category_names[cat_index]),
8697 __FILE__, caller_line);
8700 /* If the locales are the same, there's nothing to do */
8701 if (strEQ(locale_to_restore_to, new_locale)) {
8702 DEBUG_Lv(PerlIO_printf(Perl_debug_log, "%s locale unchanged as %s\n",
8703 category_names[cat_index],
8709 /* Finally, change the locale to the new one */
8710 void_setlocale_i_with_caller(cat_index, new_locale, __FILE__, caller_line);
8712 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8713 "%s locale switched to %s\n",
8714 category_names[cat_index], new_locale));
8716 return locale_to_restore_to;
8719 PERL_UNUSED_ARG(caller_line);
8725 S_restore_toggled_locale_i(pTHX_ const locale_category_index cat_index,
8726 const char * restore_locale,
8727 const line_t caller_line)
8729 /* Restores the locale for LC_category corresponding to cat_index to
8730 * 'restore_locale' (which is a copy that will be freed by this function),
8731 * or do nothing if the latter parameter is NULL */
8733 PERL_ARGS_ASSERT_RESTORE_TOGGLED_LOCALE_I;
8734 assert(cat_index <= LC_ALL_INDEX_);
8736 if (restore_locale == NULL) {
8737 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8738 "restore_toggled_locale_i: No need to" \
8739 " restore %s; called from %" LINE_Tf "\n", \
8740 category_names[cat_index], caller_line));
8744 DEBUG_Lv(PerlIO_printf(Perl_debug_log,
8745 "restore_toggled_locale_i: restoring locale for" \
8746 " %s to %s; called from %" LINE_Tf "\n", \
8747 category_names[cat_index], restore_locale,
8750 void_setlocale_i_with_caller(cat_index, restore_locale,
8751 __FILE__, caller_line);
8754 PERL_UNUSED_ARG(caller_line);
8759 # ifdef USE_LOCALE_CTYPE
8762 S_is_codeset_name_UTF8(const char * name)
8764 /* Return a boolean as to if the passed-in name indicates it is a UTF-8
8765 * code set. Several variants are possible */
8766 const Size_t len = strlen(name);
8768 PERL_ARGS_ASSERT_IS_CODESET_NAME_UTF8;
8772 /* http://msdn.microsoft.com/en-us/library/windows/desktop/dd317756.aspx */
8773 if (memENDs(name, len, "65001")) {
8778 /* 'UTF8' or 'UTF-8' */
8779 return ( inRANGE(len, 4, 5)
8780 && name[len-1] == '8'
8781 && ( memBEGINs(name, len, "UTF")
8782 || memBEGINs(name, len, "utf"))
8783 && (len == 4 || name[3] == '-'));
8787 #endif /* USE_LOCALE */
8790 Perl__is_in_locale_category(pTHX_ const bool compiling, const int category)
8792 /* Internal function which returns if we are in the scope of a pragma that
8793 * enables the locale category 'category'. 'compiling' should indicate if
8794 * this is during the compilation phase (TRUE) or not (FALSE). */
8796 const COP * const cop = (compiling) ? &PL_compiling : PL_curcop;
8798 SV *these_categories = cop_hints_fetch_pvs(cop, "locale", 0);
8799 if (! these_categories || these_categories == &PL_sv_placeholder) {
8803 /* The pseudo-category 'not_characters' is -1, so just add 1 to each to get
8804 * a valid unsigned */
8805 assert(category >= -1);
8806 return cBOOL(SvUV(these_categories) & (1U << (category + 1)));
8809 /* my_strerror() returns a mortalized copy of the text of the error message
8810 * associated with 'errnum'.
8812 * If not called from within the scope of 'use locale', it uses the text from
8813 * the C locale. If Perl is compiled to not pay attention to LC_CTYPE nor
8814 * LC_MESSAGES, it uses whatever strerror() returns. Otherwise the text is
8815 * derived from the locale, LC_MESSAGES if we have that; LC_CTYPE if not.
8817 * It returns in *utf8ness the result's UTF-8ness
8819 * The function just calls strerror(), but temporarily switches locales, if
8820 * needed. Many platforms require LC_CTYPE and LC_MESSAGES to be in the same
8821 * CODESET in order for the return from strerror() to not contain '?' symbols,
8822 * or worse, mojibaked. It's cheaper to just use the stricter criteria of
8823 * being in the same locale. So the code below uses a common locale for both
8824 * categories. Again, that is C if not within 'use locale' scope; or the
8825 * LC_MESSAGES locale if in scope and we have that category; and LC_CTYPE if we
8826 * don't have LC_MESSAGES; and whatever strerror returns if we don't have
8829 * There are two sets of implementations. The first below is if we have
8830 * strerror_l(). This is the simpler. We just use the already-built C locale
8831 * object if not in locale scope, or build up a custom one otherwise.
8833 * When strerror_l() is not available, we may have to swap locales temporarily
8834 * to bring the two categories into sync with each other, and possibly to the C
8837 * Because the prepropessing directives to conditionally compile this function
8838 * would greatly obscure the logic of the various implementations, the whole
8839 * function is repeated for each configuration, with some common macros. */
8841 /* Used to shorten the definitions of the following implementations of
8843 #define DEBUG_STRERROR_ENTER(errnum, in_locale) \
8844 DEBUG_Lv(PerlIO_printf(Perl_debug_log, \
8845 "my_strerror called with errnum %d;" \
8846 " Within locale scope=%d\n", \
8849 #define DEBUG_STRERROR_RETURN(errstr, utf8ness) \
8850 DEBUG_Lv(PerlIO_printf(Perl_debug_log, \
8851 "Strerror returned; saving a copy: '%s';" \
8853 get_displayable_string(errstr, \
8854 errstr + strlen(errstr), \
8858 /* On platforms that have precisely one of these categories (Windows
8859 * qualifies), these yield the correct one */
8860 #if defined(USE_LOCALE_CTYPE)
8861 # define WHICH_LC_INDEX LC_CTYPE_INDEX_
8862 #elif defined(USE_LOCALE_MESSAGES)
8863 # define WHICH_LC_INDEX LC_MESSAGES_INDEX_
8866 /*===========================================================================*/
8867 /* First set of implementations, when have strerror_l() */
8869 #if defined(USE_POSIX_2008_LOCALE) && defined(HAS_STRERROR_L)
8871 # if ! defined(USE_LOCALE_CTYPE) && ! defined(USE_LOCALE_MESSAGES)
8873 /* Here, neither category is defined: use the C locale */
8875 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8877 PERL_ARGS_ASSERT_MY_STRERROR;
8879 DEBUG_STRERROR_ENTER(errnum, 0);
8881 const char *errstr = savepv(strerror_l(errnum, PL_C_locale_obj));
8882 *utf8ness = UTF8NESS_IMMATERIAL;
8884 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8890 # elif ! defined(USE_LOCALE_CTYPE) || ! defined(USE_LOCALE_MESSAGES)
8892 /*--------------------------------------------------------------------------*/
8894 /* Here one or the other of CTYPE or MESSAGES is defined, but not both. If we
8895 * are not within 'use locale' scope of the only one defined, we use the C
8896 * locale; otherwise use the current locale object */
8899 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8901 PERL_ARGS_ASSERT_MY_STRERROR;
8903 DEBUG_STRERROR_ENTER(errnum, IN_LC(categories[WHICH_LC_INDEX]));
8905 /* Use C if not within locale scope; Otherwise, use current locale */
8906 const locale_t which_obj = (IN_LC(categories[WHICH_LC_INDEX]))
8908 : use_curlocale_scratch();
8910 const char *errstr = savepv(strerror_l(errnum, which_obj));
8911 *utf8ness = get_locale_string_utf8ness_i(errstr, LOCALE_UTF8NESS_UNKNOWN,
8912 NULL, WHICH_LC_INDEX);
8913 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8919 /*--------------------------------------------------------------------------*/
8920 # else /* Are using both categories. Place them in the same CODESET,
8921 * either C or the LC_MESSAGES locale */
8924 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8926 PERL_ARGS_ASSERT_MY_STRERROR;
8928 DEBUG_STRERROR_ENTER(errnum, IN_LC(LC_MESSAGES));
8931 if (! IN_LC(LC_MESSAGES)) { /* Use C if not within locale scope */
8932 errstr = savepv(strerror_l(errnum, PL_C_locale_obj));
8933 *utf8ness = UTF8NESS_IMMATERIAL;
8935 else { /* Otherwise, use the LC_MESSAGES locale, making sure LC_CTYPE
8937 locale_t cur = duplocale(use_curlocale_scratch());
8939 cur = newlocale(LC_CTYPE_MASK, querylocale_c(LC_MESSAGES), cur);
8940 errstr = savepv(strerror_l(errnum, cur));
8941 *utf8ness = get_locale_string_utf8ness_i(errstr,
8942 LOCALE_UTF8NESS_UNKNOWN,
8943 NULL, LC_MESSAGES_INDEX_);
8947 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8952 # endif /* Above is using strerror_l */
8953 /*===========================================================================*/
8954 #else /* Below is not using strerror_l */
8955 # if ! defined(USE_LOCALE_CTYPE) && ! defined(USE_LOCALE_MESSAGES)
8957 /* If not using using either of the categories, return plain, unadorned
8961 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8963 PERL_ARGS_ASSERT_MY_STRERROR;
8965 DEBUG_STRERROR_ENTER(errnum, 0);
8967 const char *errstr = savepv(Strerror(errnum));
8968 *utf8ness = UTF8NESS_IMMATERIAL;
8970 DEBUG_STRERROR_RETURN(errstr, utf8ness);
8976 /*--------------------------------------------------------------------------*/
8977 # elif ! defined(USE_LOCALE_CTYPE) || ! defined(USE_LOCALE_MESSAGES)
8979 /* Here one or the other of CTYPE or MESSAGES is defined, but not both. If we
8980 * are not within 'use locale' scope of the only one defined, we use the C
8981 * locale; otherwise use the current locale */
8984 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
8986 PERL_ARGS_ASSERT_MY_STRERROR;
8988 DEBUG_STRERROR_ENTER(errnum, IN_LC(categories[WHICH_LC_INDEX]));
8991 if (IN_LC(categories[WHICH_LC_INDEX])) {
8992 errstr = savepv(Strerror(errnum));
8993 *utf8ness = get_locale_string_utf8ness_i(errstr,
8994 LOCALE_UTF8NESS_UNKNOWN,
8995 NULL, WHICH_LC_INDEX);
9001 const char * orig_locale = toggle_locale_i(WHICH_LC_INDEX, "C");
9003 errstr = savepv(Strerror(errnum));
9005 restore_toggled_locale_i(WHICH_LC_INDEX, orig_locale);
9009 *utf8ness = UTF8NESS_IMMATERIAL;
9012 DEBUG_STRERROR_RETURN(errstr, utf8ness);
9018 /*--------------------------------------------------------------------------*/
9021 /* Below, have both LC_CTYPE and LC_MESSAGES. Place them in the same CODESET,
9022 * either C or the LC_MESSAGES locale */
9025 Perl_my_strerror(pTHX_ const int errnum, utf8ness_t * utf8ness)
9027 PERL_ARGS_ASSERT_MY_STRERROR;
9029 DEBUG_STRERROR_ENTER(errnum, IN_LC(LC_MESSAGES));
9031 const char * desired_locale = (IN_LC(LC_MESSAGES))
9032 ? querylocale_c(LC_MESSAGES)
9034 /* XXX Can fail on z/OS */
9038 const char* orig_CTYPE_locale = toggle_locale_c(LC_CTYPE,
9040 const char* orig_MESSAGES_locale = toggle_locale_c(LC_MESSAGES,
9042 const char *errstr = savepv(Strerror(errnum));
9044 restore_toggled_locale_c(LC_MESSAGES, orig_MESSAGES_locale);
9045 restore_toggled_locale_c(LC_CTYPE, orig_CTYPE_locale);
9049 *utf8ness = get_locale_string_utf8ness_i(errstr, LOCALE_UTF8NESS_UNKNOWN,
9050 NULL, LC_MESSAGES_INDEX_);
9051 DEBUG_STRERROR_RETURN(errstr, utf8ness);
9057 /*--------------------------------------------------------------------------*/
9058 # endif /* end of not using strerror_l() */
9059 #endif /* end of all the my_strerror() implementations */
9063 =for apidoc switch_to_global_locale
9065 This function copies the locale state of the calling thread into the program's
9066 global locale, and converts the thread to use that global locale.
9068 It is intended so that Perl can safely be used with C libraries that access the
9069 global locale and which can't be converted to not access it. Effectively, this
9070 means libraries that call C<L<setlocale(3)>> on non-Windows systems. (For
9071 portability, it is a good idea to use it on Windows as well.)
9073 A downside of using it is that it disables the services that Perl provides to
9074 hide locale gotchas from your code. The service you most likely will miss
9075 regards the radix character (decimal point) in floating point numbers. Code
9076 executed after this function is called can no longer just assume that this
9077 character is correct for the current circumstances.
9079 To return to Perl control, and restart the gotcha prevention services, call
9080 C<L</sync_locale>>. Behavior is undefined for any pure Perl code that executes
9081 while the switch is in effect.
9083 The global locale and the per-thread locales are independent. As long as just
9084 one thread converts to the global locale, everything works smoothly. But if
9085 more than one does, they can easily interfere with each other, and races are
9086 likely. On Windows systems prior to Visual Studio 15 (at which point Microsoft
9087 fixed a bug), races can occur (even if only one thread has been converted to
9088 the global locale), but only if you use the following operations:
9092 =item L<POSIX::localeconv|POSIX/localeconv>
9094 =item L<I18N::Langinfo>, items C<CRNCYSTR> and C<THOUSEP>
9096 =item L<perlapi/Perl_langinfo>, items C<CRNCYSTR> and C<THOUSEP>
9100 The first item is not fixable (except by upgrading to a later Visual Studio
9101 release), but it would be possible to work around the latter two items by
9102 having Perl change its algorithm for calculating these to use Windows API
9103 functions (likely C<GetNumberFormat> and C<GetCurrencyFormat>); patches
9106 XS code should never call plain C<setlocale>, but should instead be converted
9107 to either call L<C<Perl_setlocale>|perlapi/Perl_setlocale> (which is a drop-in
9108 for the system C<setlocale>) or use the methods given in L<perlcall> to call
9109 L<C<POSIX::setlocale>|POSIX/setlocale>. Either one will transparently properly
9110 handle all cases of single- vs multi-thread, POSIX 2008-supported or not.
9115 #if defined(WIN32) && defined(USE_THREAD_SAFE_LOCALE)
9116 # define CHANGE_SYSTEM_LOCALE_TO_GLOBAL \
9118 if (_configthreadlocale(_DISABLE_PER_THREAD_LOCALE) == -1) { \
9119 locale_panic_("_configthreadlocale returned an error"); \
9122 #elif defined(USE_POSIX_2008_LOCALE)
9123 # define CHANGE_SYSTEM_LOCALE_TO_GLOBAL \
9125 locale_t old_locale = uselocale(LC_GLOBAL_LOCALE); \
9126 if (! old_locale) { \
9127 locale_panic_("Could not change to global locale"); \
9130 /* Free the per-thread memory */ \
9131 if ( old_locale != LC_GLOBAL_LOCALE \
9132 && old_locale != PL_C_locale_obj) \
9134 freelocale(old_locale); \
9138 # define CHANGE_SYSTEM_LOCALE_TO_GLOBAL
9142 Perl_switch_to_global_locale(pTHX)
9147 DEBUG_L(PerlIO_printf(Perl_debug_log, "Entering switch_to_global; %s\n",
9148 get_LC_ALL_display()));
9150 /* In these cases, we use the system state to determine if we are in the
9151 * global locale or not. */
9152 # ifdef USE_POSIX_2008_LOCALE
9154 const bool perl_controls = (LC_GLOBAL_LOCALE != uselocale((locale_t) 0));
9156 # elif defined(USE_THREAD_SAFE_LOCALE) && defined(WIN32)
9158 int config_return = _configthreadlocale(0);
9159 if (config_return == -1) {
9160 locale_panic_("_configthreadlocale returned an error");
9162 const bool perl_controls = (config_return == _ENABLE_PER_THREAD_LOCALE);
9166 const bool perl_controls = false;
9170 /* No-op if already in global */
9171 if (! perl_controls) {
9177 const char * thread_locale = calculate_LC_ALL_string(NULL,
9178 EXTERNAL_FORMAT_FOR_SET,
9181 CHANGE_SYSTEM_LOCALE_TO_GLOBAL;
9182 posix_setlocale(LC_ALL, thread_locale);
9184 # else /* Must be USE_POSIX_2008_LOCALE) */
9186 const char * cur_thread_locales[LC_ALL_INDEX_];
9188 /* Save each category's current per-thread state */
9189 for_all_individual_category_indexes(i) {
9190 cur_thread_locales[i] = querylocale_i(i);
9193 CHANGE_SYSTEM_LOCALE_TO_GLOBAL;
9195 /* Set the global to what was our per-thread state */
9196 POSIX_SETLOCALE_LOCK;
9197 for_all_individual_category_indexes(i) {
9198 posix_setlocale(categories[i], cur_thread_locales[i]);
9200 POSIX_SETLOCALE_UNLOCK;
9203 # ifdef USE_LOCALE_NUMERIC
9205 /* Switch to the underlying C numeric locale; the application is on its
9207 POSIX_SETLOCALE_LOCK;
9208 posix_setlocale(LC_NUMERIC, PL_numeric_name);
9209 POSIX_SETLOCALE_UNLOCK;
9218 =for apidoc sync_locale
9220 This function copies the state of the program global locale into the calling
9221 thread, and converts that thread to using per-thread locales, if it wasn't
9222 already, and the platform supports them. The LC_NUMERIC locale is toggled into
9223 the standard state (using the C locale's conventions), if not within the
9224 lexical scope of S<C<use locale>>.
9226 Perl will now consider itself to have control of the locale.
9228 Since unthreaded perls have only a global locale, this function is a no-op
9231 This function is intended for use with C libraries that do locale manipulation.
9232 It allows Perl to accommodate the use of them. Call this function before
9233 transferring back to Perl space so that it knows what state the C code has left
9236 XS code should not manipulate the locale on its own. Instead,
9237 L<C<Perl_setlocale>|perlapi/Perl_setlocale> can be used at any time to query or
9238 change the locale (though changing the locale is antisocial and dangerous on
9239 multi-threaded systems that don't have multi-thread safe locale operations.
9240 (See L<perllocale/Multi-threaded operation>).
9242 Using the libc L<C<setlocale(3)>> function should be avoided. Nevertheless,
9243 certain non-Perl libraries called from XS, do call it, and their behavior may
9244 not be able to be changed. This function, along with
9245 C<L</switch_to_global_locale>>, can be used to get seamless behavior in these
9246 circumstances, as long as only one thread is involved.
9248 If the library has an option to turn off its locale manipulation, doing that is
9249 preferable to using this mechanism. C<Gtk> is such a library.
9251 The return value is a boolean: TRUE if the global locale at the time of call
9252 was in effect for the caller; and FALSE if a per-thread locale was in effect.
9258 Perl_sync_locale(pTHX)
9267 bool was_in_global = TRUE;
9269 # ifdef USE_THREAD_SAFE_LOCALE
9272 int config_return = _configthreadlocale(_DISABLE_PER_THREAD_LOCALE);
9273 if (config_return == -1) {
9274 locale_panic_("_configthreadlocale returned an error");
9276 was_in_global = (config_return == _DISABLE_PER_THREAD_LOCALE);
9278 # elif defined(USE_POSIX_2008_LOCALE)
9280 was_in_global = (LC_GLOBAL_LOCALE == uselocale(LC_GLOBAL_LOCALE));
9283 # error Unexpected Configuration
9285 # endif /* USE_THREAD_SAFE_LOCALE */
9287 /* Here, we are in the global locale. Get and save the values for each
9288 * category, and convert the current thread to use them */
9292 STDIZED_SETLOCALE_LOCK;
9293 const char * lc_all_string = savepv(stdized_setlocale(LC_ALL, NULL));
9294 STDIZED_SETLOCALE_UNLOCK;
9296 give_perl_locale_control(lc_all_string, __LINE__);
9297 Safefree(lc_all_string);
9301 const char * current_globals[LC_ALL_INDEX_];
9302 for_all_individual_category_indexes(i) {
9303 STDIZED_SETLOCALE_LOCK;
9304 current_globals[i] = savepv(stdized_setlocale(categories[i], NULL));
9305 STDIZED_SETLOCALE_UNLOCK;
9308 give_perl_locale_control((const char **) ¤t_globals, __LINE__);
9310 for_all_individual_category_indexes(i) {
9311 Safefree(current_globals[i]);
9316 return was_in_global;
9322 #if defined(DEBUGGING) && defined(USE_LOCALE)
9325 S_my_setlocale_debug_string_i(pTHX_
9326 const locale_category_index cat_index,
9327 const char* locale, /* Optional locale name */
9329 /* return value from setlocale() when attempting
9330 * to set 'category' to 'locale' */
9335 /* Returns a pointer to a NUL-terminated string in static storage with
9336 * added text about the info passed in. This is not thread safe and will
9337 * be overwritten by the next call, so this should be used just to
9338 * formulate a string to immediately print or savepv() on. */
9340 const char * locale_quote;
9341 const char * retval_quote;
9343 assert(cat_index <= LC_ALL_INDEX_);
9345 if (locale == NULL) {
9350 locale_quote = "\"";
9353 if (retval == NULL) {
9358 retval_quote = "\"";
9361 # ifdef USE_LOCALE_THREADS
9362 # define THREAD_FORMAT "%p:"
9363 # define THREAD_ARGUMENT aTHX_
9365 # define THREAD_FORMAT
9366 # define THREAD_ARGUMENT
9369 return Perl_form(aTHX_
9370 "%s:%" LINE_Tf ": " THREAD_FORMAT
9371 " setlocale(%s[%d], %s%s%s) returned %s%s%s\n",
9373 __FILE__, line, THREAD_ARGUMENT
9374 category_names[cat_index], categories[cat_index],
9375 locale_quote, locale, locale_quote,
9376 retval_quote, retval, retval_quote);
9380 #ifdef USE_PERL_SWITCH_LOCALE_CONTEXT
9383 Perl_switch_locale_context(pTHX)
9385 /* libc keeps per-thread locale status information in some configurations.
9386 * So, we can't just switch out aTHX to switch to a new thread. libc has
9387 * to follow along. This routine does that based on per-interpreter
9388 * variables we keep just for this purpose.
9390 * There are two implementations where this is an issue. For the other
9391 * implementations, it doesn't matter because libc is using global values
9392 * that all threads know about.
9394 * The two implementations are where libc keeps thread-specific information
9395 * on its own. These are
9397 * POSIX 2008: The current locale is kept by libc as an object. We save
9398 * a copy of that in the per-thread PL_cur_locale_obj, and so
9399 * this routine uses that copy to tell the thread it should be
9400 * operating with that object
9401 * Windows thread-safe locales: A given thread in Windows can be being run
9402 * with per-thread locales, or not. When the thread context
9403 * changes, libc doesn't automatically know if the thread is
9404 * using per-thread locales, nor does it know what the new
9405 * thread's locale is. We keep that information in the
9406 * per-thread variables:
9407 * PL_controls_locale indicates if this thread is using
9408 * per-thread locales or not
9409 * PL_cur_LC_ALL indicates what the locale should be
9410 * if it is a per-thread locale.
9413 if (UNLIKELY( PL_veto_switch_non_tTHX_context
9414 || PL_phase == PERL_PHASE_CONSTRUCT))
9419 # ifdef USE_POSIX_2008_LOCALE
9421 if (! uselocale(PL_cur_locale_obj)) {
9422 locale_panic_(Perl_form(aTHX_
9423 "Can't uselocale(%p), LC_ALL supposed to"
9425 PL_cur_locale_obj, get_LC_ALL_display()));
9428 # elif defined(WIN32)
9430 if (! bool_setlocale_c(LC_ALL, PL_cur_LC_ALL)) {
9431 locale_panic_(Perl_form(aTHX_ "Can't setlocale(%s)", PL_cur_LC_ALL));
9441 Perl_thread_locale_init(pTHX)
9444 #ifdef USE_THREAD_SAFE_LOCALE
9445 # ifdef USE_POSIX_2008_LOCALE
9447 /* Called from a thread on startup.
9449 * The operations here have to be done from within the calling thread, as
9450 * they affect libc's knowledge of the thread; libc has no knowledge of
9453 DEBUG_L(PerlIO_printf(Perl_debug_log,
9454 "new thread, initial locale is %s;"
9455 " calling setlocale(LC_ALL, \"C\")\n",
9456 get_LC_ALL_display()));
9458 if (! uselocale(PL_C_locale_obj)) {
9460 /* Not being able to change to the C locale is severe; don't keep
9462 locale_panic_(Perl_form(aTHX_
9463 "Can't uselocale(%p), 'C'", PL_C_locale_obj));
9464 NOT_REACHED; /* NOTREACHED */
9467 # ifdef MULTIPLICITY
9469 PL_cur_locale_obj = PL_C_locale_obj;
9472 # elif defined(WIN32)
9474 /* On Windows, make sure new thread has per-thread locales enabled */
9475 if (_configthreadlocale(_ENABLE_PER_THREAD_LOCALE) == -1) {
9476 locale_panic_("_configthreadlocale returned an error");
9478 void_setlocale_c(LC_ALL, "C");
9486 Perl_thread_locale_term(pTHX)
9488 /* Called from a thread as it gets ready to terminate.
9490 * The operations here have to be done from within the calling thread, as
9491 * they affect libc's knowledge of the thread; libc has no knowledge of
9494 #if defined(USE_POSIX_2008_LOCALE) && defined(USE_THREADS)
9496 /* Switch to the global locale, so can free up the per-thread object */
9497 locale_t actual_obj = uselocale(LC_GLOBAL_LOCALE);
9498 if (actual_obj != LC_GLOBAL_LOCALE && actual_obj != PL_C_locale_obj) {
9499 freelocale(actual_obj);
9502 /* Prevent leaks even if something has gone wrong */
9503 locale_t expected_obj = PL_cur_locale_obj;
9504 if (UNLIKELY( expected_obj != actual_obj
9505 && expected_obj != LC_GLOBAL_LOCALE
9506 && expected_obj != PL_C_locale_obj))
9508 freelocale(expected_obj);
9511 PL_cur_locale_obj = LC_GLOBAL_LOCALE;
9514 #ifdef WIN32_USE_FAKE_OLD_MINGW_LOCALES
9516 /* When faking the mingw implementation, we coerce this function into doing
9517 * something completely different from its intent -- namely to free up our
9518 * static buffer to avoid a leak. This function gets called for each
9519 * thread that is terminating, so will give us a chance to free the buffer
9520 * from the appropriate pool. On unthreaded systems, it gets called by the
9521 * mutex termination code. */
9523 # ifdef MULTIPLICITY
9525 if (aTHX != wsetlocale_buf_aTHX) {
9531 if (wsetlocale_buf_size > 0) {
9532 Safefree(wsetlocale_buf);
9533 wsetlocale_buf_size = 0;
9541 * ex: set ts=8 sts=4 sw=4 et: