1PERLLOCALE(1) Perl Programmers Reference Guide PERLLOCALE(1)
2
3
4
6 perllocale - Perl locale handling (internationalization and
7 localization)
8
10 In the beginning there was ASCII, the "American Standard Code for
11 Information Interchange", which works quite well for Americans with
12 their English alphabet and dollar-denominated currency. But it doesn't
13 work so well even for other English speakers, who may use different
14 currencies, such as the pound sterling (as the symbol for that currency
15 is not in ASCII); and it's hopelessly inadequate for many of the
16 thousands of the world's other languages.
17
18 To address these deficiencies, the concept of locales was invented
19 (formally the ISO C, XPG4, POSIX 1.c "locale system"). And
20 applications were and are being written that use the locale mechanism.
21 The process of making such an application take account of its users'
22 preferences in these kinds of matters is called internationalization
23 (often abbreviated as i18n); telling such an application about a
24 particular set of preferences is known as localization (l10n).
25
26 Perl has been extended to support the locale system. This is
27 controlled per application by using one pragma, one function call, and
28 several environment variables.
29
30 Unfortunately, there are quite a few deficiencies with the design (and
31 often, the implementations) of locales. Unicode was invented (see
32 perlunitut for an introduction to that) in part to address these design
33 deficiencies, and nowadays, there is a series of "UTF-8 locales", based
34 on Unicode. These are locales whose character set is Unicode, encoded
35 in UTF-8. Starting in v5.20, Perl fully supports UTF-8 locales, except
36 for sorting and string comparisons like "lt" and "ge". Starting in
37 v5.26, Perl can handle these reasonably as well, depending on the
38 platform's implementation. However, for earlier releases or for better
39 control, use Unicode::Collate . Perl continues to support the old non
40 UTF-8 locales as well. There are currently no UTF-8 locales for EBCDIC
41 platforms.
42
43 (Unicode is also creating "CLDR", the "Common Locale Data Repository",
44 <http://cldr.unicode.org/> which includes more types of information
45 than are available in the POSIX locale system. At the time of this
46 writing, there was no CPAN module that provides access to this XML-
47 encoded data. However, it is possible to compute the POSIX locale data
48 from them, and earlier CLDR versions had these already extracted for
49 you as UTF-8 locales <http://unicode.org/Public/cldr/2.0.1/>.)
50
52 A locale is a set of data that describes various aspects of how various
53 communities in the world categorize their world. These categories are
54 broken down into the following types (some of which include a brief
55 note here):
56
57 Category "LC_NUMERIC": Numeric formatting
58 This indicates how numbers should be formatted for human
59 readability, for example the character used as the decimal point.
60
61 Category "LC_MONETARY": Formatting of monetary amounts
62
63
64 Category "LC_TIME": Date/Time formatting
65
66
67 Category "LC_MESSAGES": Error and other messages
68 This is used by Perl itself only for accessing operating system
69 error messages via $! and $^E.
70
71 Category "LC_COLLATE": Collation
72 This indicates the ordering of letters for comparison and sorting.
73 In Latin alphabets, for example, "b", generally follows "a".
74
75 Category "LC_CTYPE": Character Types
76 This indicates, for example if a character is an uppercase letter.
77
78 Other categories
79 Some platforms have other categories, dealing with such things as
80 measurement units and paper sizes. None of these are used directly
81 by Perl, but outside operations that Perl interacts with may use
82 these. See "Not within the scope of "use locale"" below.
83
84 More details on the categories used by Perl are given below in "LOCALE
85 CATEGORIES".
86
87 Together, these categories go a long way towards being able to
88 customize a single program to run in many different locations. But
89 there are deficiencies, so keep reading.
90
92 Perl itself (outside the POSIX module) will not use locales unless
93 specifically requested to (but again note that Perl may interact with
94 code that does use them). Even if there is such a request, all of the
95 following must be true for it to work properly:
96
97 · Your operating system must support the locale system. If it does,
98 you should find that the "setlocale()" function is a documented
99 part of its C library.
100
101 · Definitions for locales that you use must be installed. You, or
102 your system administrator, must make sure that this is the case.
103 The available locales, the location in which they are kept, and the
104 manner in which they are installed all vary from system to system.
105 Some systems provide only a few, hard-wired locales and do not
106 allow more to be added. Others allow you to add "canned" locales
107 provided by the system supplier. Still others allow you or the
108 system administrator to define and add arbitrary locales. (You may
109 have to ask your supplier to provide canned locales that are not
110 delivered with your operating system.) Read your system
111 documentation for further illumination.
112
113 · Perl must believe that the locale system is supported. If it does,
114 "perl -V:d_setlocale" will say that the value for "d_setlocale" is
115 "define".
116
117 If you want a Perl application to process and present your data
118 according to a particular locale, the application code should include
119 the "use locale" pragma (see "The "use locale" pragma") where
120 appropriate, and at least one of the following must be true:
121
122 1. The locale-determining environment variables (see "ENVIRONMENT")
123 must be correctly set up at the time the application is started,
124 either by yourself or by whomever set up your system account; or
125
126 2. The application must set its own locale using the method described
127 in "The setlocale function".
128
130 The "use locale" pragma
131 WARNING! Do NOT use this pragma in scripts that have multiple threads
132 active. The locale is not local to a single thread. Another thread
133 may change the locale at any time, which could cause at a minimum that
134 a given thread is operating in a locale it isn't expecting to be in.
135 On some platforms, segfaults can also occur. The locale change need
136 not be explicit; some operations cause perl to change the locale
137 itself. You are vulnerable simply by having done a "use locale".
138
139 By default, Perl itself (outside the POSIX module) ignores the current
140 locale. The "use locale" pragma tells Perl to use the current locale
141 for some operations. Starting in v5.16, there are optional parameters
142 to this pragma, described below, which restrict which operations are
143 affected by it.
144
145 The current locale is set at execution time by setlocale() described
146 below. If that function hasn't yet been called in the course of the
147 program's execution, the current locale is that which was determined by
148 the "ENVIRONMENT" in effect at the start of the program. If there is
149 no valid environment, the current locale is whatever the system default
150 has been set to. On POSIX systems, it is likely, but not necessarily,
151 the "C" locale. On Windows, the default is set via the computer's
152 "Control Panel->Regional and Language Options" (or its current
153 equivalent).
154
155 The operations that are affected by locale are:
156
157 Not within the scope of "use locale"
158 Only certain operations originating outside Perl should be
159 affected, as follows:
160
161 · The current locale is used when going outside of Perl with
162 operations like system() or qx//, if those operations are
163 locale-sensitive.
164
165 · Also Perl gives access to various C library functions through
166 the POSIX module. Some of those functions are always affected
167 by the current locale. For example, "POSIX::strftime()" uses
168 "LC_TIME"; "POSIX::strtod()" uses "LC_NUMERIC";
169 "POSIX::strcoll()" and "POSIX::strxfrm()" use "LC_COLLATE".
170 All such functions will behave according to the current
171 underlying locale, even if that locale isn't exposed to Perl
172 space.
173
174 · XS modules for all categories but "LC_NUMERIC" get the
175 underlying locale, and hence any C library functions they call
176 will use that underlying locale. For more discussion, see
177 "CAVEATS" in perlxs.
178
179 Note that all C programs (including the perl interpreter, which is
180 written in C) always have an underlying locale. That locale is the
181 "C" locale unless changed by a call to setlocale(). When Perl
182 starts up, it changes the underlying locale to the one which is
183 indicated by the "ENVIRONMENT". When using the POSIX module or
184 writing XS code, it is important to keep in mind that the
185 underlying locale may be something other than "C", even if the
186 program hasn't explicitly changed it.
187
188
189
190 Lingering effects of "use locale"
191 Certain Perl operations that are set-up within the scope of a "use
192 locale" retain that effect even outside the scope. These include:
193
194 · The output format of a write() is determined by an earlier
195 format declaration ("format" in perlfunc), so whether or not
196 the output is affected by locale is determined by if the
197 "format()" is within the scope of a "use locale", not whether
198 the "write()" is.
199
200 · Regular expression patterns can be compiled using qr// with
201 actual matching deferred to later. Again, it is whether or not
202 the compilation was done within the scope of "use locale" that
203 determines the match behavior, not if the matches are done
204 within such a scope or not.
205
206
207
208 Under ""use locale";"
209 · All the above operations
210
211 · Format declarations ("format" in perlfunc) and hence any
212 subsequent "write()"s use "LC_NUMERIC".
213
214 · stringification and output use "LC_NUMERIC". These include the
215 results of "print()", "printf()", "say()", and "sprintf()".
216
217 · The comparison operators ("lt", "le", "cmp", "ge", and "gt")
218 use "LC_COLLATE". "sort()" is also affected if used without an
219 explicit comparison function, because it uses "cmp" by default.
220
221 Note: "eq" and "ne" are unaffected by locale: they always
222 perform a char-by-char comparison of their scalar operands.
223 What's more, if "cmp" finds that its operands are equal
224 according to the collation sequence specified by the current
225 locale, it goes on to perform a char-by-char comparison, and
226 only returns 0 (equal) if the operands are char-for-char
227 identical. If you really want to know whether two
228 strings--which "eq" and "cmp" may consider different--are equal
229 as far as collation in the locale is concerned, see the
230 discussion in "Category "LC_COLLATE": Collation".
231
232 · Regular expressions and case-modification functions ("uc()",
233 "lc()", "ucfirst()", and "lcfirst()") use "LC_CTYPE"
234
235 · The variables $! (and its synonyms $ERRNO and $OS_ERROR) and
236 $^E (and its synonym $EXTENDED_OS_ERROR) when used as strings
237 use "LC_MESSAGES".
238
239 The default behavior is restored with the "no locale" pragma, or upon
240 reaching the end of the block enclosing "use locale". Note that "use
241 locale" calls may be nested, and that what is in effect within an inner
242 scope will revert to the outer scope's rules at the end of the inner
243 scope.
244
245 The string result of any operation that uses locale information is
246 tainted, as it is possible for a locale to be untrustworthy. See
247 "SECURITY".
248
249 Starting in Perl v5.16 in a very limited way, and more generally in
250 v5.22, you can restrict which category or categories are enabled by
251 this particular instance of the pragma by adding parameters to it. For
252 example,
253
254 use locale qw(:ctype :numeric);
255
256 enables locale awareness within its scope of only those operations
257 (listed above) that are affected by "LC_CTYPE" and "LC_NUMERIC".
258
259 The possible categories are: ":collate", ":ctype", ":messages",
260 ":monetary", ":numeric", ":time", and the pseudo category ":characters"
261 (described below).
262
263 Thus you can say
264
265 use locale ':messages';
266
267 and only $! and $^E will be locale aware. Everything else is
268 unaffected.
269
270 Since Perl doesn't currently do anything with the "LC_MONETARY"
271 category, specifying ":monetary" does effectively nothing. Some
272 systems have other categories, such as "LC_PAPER_SIZE", but Perl also
273 doesn't know anything about them, and there is no way to specify them
274 in this pragma's arguments.
275
276 You can also easily say to use all categories but one, by either, for
277 example,
278
279 use locale ':!ctype';
280 use locale ':not_ctype';
281
282 both of which mean to enable locale awarness of all categories but
283 "LC_CTYPE". Only one category argument may be specified in a
284 "use locale" if it is of the negated form.
285
286 Prior to v5.22 only one form of the pragma with arguments is available:
287
288 use locale ':not_characters';
289
290 (and you have to say "not_"; you can't use the bang "!" form). This
291 pseudo category is a shorthand for specifying both ":collate" and
292 ":ctype". Hence, in the negated form, it is nearly the same thing as
293 saying
294
295 use locale qw(:messages :monetary :numeric :time);
296
297 We use the term "nearly", because ":not_characters" also turns on
298 "use feature 'unicode_strings'" within its scope. This form is less
299 useful in v5.20 and later, and is described fully in "Unicode and
300 UTF-8", but briefly, it tells Perl to not use the character portions of
301 the locale definition, that is the "LC_CTYPE" and "LC_COLLATE"
302 categories. Instead it will use the native character set (extended by
303 Unicode). When using this parameter, you are responsible for getting
304 the external character set translated into the native/Unicode one
305 (which it already will be if it is one of the increasingly popular
306 UTF-8 locales). There are convenient ways of doing this, as described
307 in "Unicode and UTF-8".
308
309 The setlocale function
310 WARNING! Do NOT use this function in a thread. The locale will change
311 in all other threads at the same time, and should your thread get
312 paused by the operating system, and another started, that thread will
313 not have the locale it is expecting. On some platforms, there can be a
314 race leading to segfaults if two threads call this function nearly
315 simultaneously.
316
317 You can switch locales as often as you wish at run time with the
318 "POSIX::setlocale()" function:
319
320 # Import locale-handling tool set from POSIX module.
321 # This example uses: setlocale -- the function call
322 # LC_CTYPE -- explained below
323 # (Showing the testing for success/failure of operations is
324 # omitted in these examples to avoid distracting from the main
325 # point)
326
327 use POSIX qw(locale_h);
328 use locale;
329 my $old_locale;
330
331 # query and save the old locale
332 $old_locale = setlocale(LC_CTYPE);
333
334 setlocale(LC_CTYPE, "fr_CA.ISO8859-1");
335 # LC_CTYPE now in locale "French, Canada, codeset ISO 8859-1"
336
337 setlocale(LC_CTYPE, "");
338 # LC_CTYPE now reset to the default defined by the
339 # LC_ALL/LC_CTYPE/LANG environment variables, or to the system
340 # default. See below for documentation.
341
342 # restore the old locale
343 setlocale(LC_CTYPE, $old_locale);
344
345 The first argument of "setlocale()" gives the category, the second the
346 locale. The category tells in what aspect of data processing you want
347 to apply locale-specific rules. Category names are discussed in
348 "LOCALE CATEGORIES" and "ENVIRONMENT". The locale is the name of a
349 collection of customization information corresponding to a particular
350 combination of language, country or territory, and codeset. Read on
351 for hints on the naming of locales: not all systems name locales as in
352 the example.
353
354 If no second argument is provided and the category is something other
355 than "LC_ALL", the function returns a string naming the current locale
356 for the category. You can use this value as the second argument in a
357 subsequent call to "setlocale()", but on some platforms the string is
358 opaque, not something that most people would be able to decipher as to
359 what locale it means.
360
361 If no second argument is provided and the category is "LC_ALL", the
362 result is implementation-dependent. It may be a string of concatenated
363 locale names (separator also implementation-dependent) or a single
364 locale name. Please consult your setlocale(3) man page for details.
365
366 If a second argument is given and it corresponds to a valid locale, the
367 locale for the category is set to that value, and the function returns
368 the now-current locale value. You can then use this in yet another
369 call to "setlocale()". (In some implementations, the return value may
370 sometimes differ from the value you gave as the second argument--think
371 of it as an alias for the value you gave.)
372
373 As the example shows, if the second argument is an empty string, the
374 category's locale is returned to the default specified by the
375 corresponding environment variables. Generally, this results in a
376 return to the default that was in force when Perl started up: changes
377 to the environment made by the application after startup may or may not
378 be noticed, depending on your system's C library.
379
380 Note that when a form of "use locale" that doesn't include all
381 categories is specified, Perl ignores the excluded categories.
382
383 If "set_locale()" fails for some reason (for example, an attempt to set
384 to a locale unknown to the system), the locale for the category is not
385 changed, and the function returns "undef".
386
387 For further information about the categories, consult setlocale(3).
388
389 Finding locales
390 For locales available in your system, consult also setlocale(3) to see
391 whether it leads to the list of available locales (search for the SEE
392 ALSO section). If that fails, try the following command lines:
393
394 locale -a
395
396 nlsinfo
397
398 ls /usr/lib/nls/loc
399
400 ls /usr/lib/locale
401
402 ls /usr/lib/nls
403
404 ls /usr/share/locale
405
406 and see whether they list something resembling these
407
408 en_US.ISO8859-1 de_DE.ISO8859-1 ru_RU.ISO8859-5
409 en_US.iso88591 de_DE.iso88591 ru_RU.iso88595
410 en_US de_DE ru_RU
411 en de ru
412 english german russian
413 english.iso88591 german.iso88591 russian.iso88595
414 english.roman8 russian.koi8r
415
416 Sadly, even though the calling interface for "setlocale()" has been
417 standardized, names of locales and the directories where the
418 configuration resides have not been. The basic form of the name is
419 language_territory.codeset, but the latter parts after language are not
420 always present. The language and country are usually from the
421 standards ISO 3166 and ISO 639, the two-letter abbreviations for the
422 countries and the languages of the world, respectively. The codeset
423 part often mentions some ISO 8859 character set, the Latin codesets.
424 For example, "ISO 8859-1" is the so-called "Western European codeset"
425 that can be used to encode most Western European languages adequately.
426 Again, there are several ways to write even the name of that one
427 standard. Lamentably.
428
429 Two special locales are worth particular mention: "C" and "POSIX".
430 Currently these are effectively the same locale: the difference is
431 mainly that the first one is defined by the C standard, the second by
432 the POSIX standard. They define the default locale in which every
433 program starts in the absence of locale information in its environment.
434 (The default default locale, if you will.) Its language is (American)
435 English and its character codeset ASCII or, rarely, a superset thereof
436 (such as the "DEC Multinational Character Set (DEC-MCS)"). Warning.
437 The C locale delivered by some vendors may not actually exactly match
438 what the C standard calls for. So beware.
439
440 NOTE: Not all systems have the "POSIX" locale (not all systems are
441 POSIX-conformant), so use "C" when you need explicitly to specify this
442 default locale.
443
444 LOCALE PROBLEMS
445 You may encounter the following warning message at Perl startup:
446
447 perl: warning: Setting locale failed.
448 perl: warning: Please check that your locale settings:
449 LC_ALL = "En_US",
450 LANG = (unset)
451 are supported and installed on your system.
452 perl: warning: Falling back to the standard locale ("C").
453
454 This means that your locale settings had "LC_ALL" set to "En_US" and
455 LANG exists but has no value. Perl tried to believe you but could not.
456 Instead, Perl gave up and fell back to the "C" locale, the default
457 locale that is supposed to work no matter what. (On Windows, it first
458 tries falling back to the system default locale.) This usually means
459 your locale settings were wrong, they mention locales your system has
460 never heard of, or the locale installation in your system has problems
461 (for example, some system files are broken or missing). There are
462 quick and temporary fixes to these problems, as well as more thorough
463 and lasting fixes.
464
465 Testing for broken locales
466 If you are building Perl from source, the Perl test suite file
467 lib/locale.t can be used to test the locales on your system. Setting
468 the environment variable "PERL_DEBUG_FULL_TEST" to 1 will cause it to
469 output detailed results. For example, on Linux, you could say
470
471 PERL_DEBUG_FULL_TEST=1 ./perl -T -Ilib lib/locale.t > locale.log 2>&1
472
473 Besides many other tests, it will test every locale it finds on your
474 system to see if they conform to the POSIX standard. If any have
475 errors, it will include a summary near the end of the output of which
476 locales passed all its tests, and which failed, and why.
477
478 Temporarily fixing locale problems
479 The two quickest fixes are either to render Perl silent about any
480 locale inconsistencies or to run Perl under the default locale "C".
481
482 Perl's moaning about locale problems can be silenced by setting the
483 environment variable "PERL_BADLANG" to "0" or "". This method really
484 just sweeps the problem under the carpet: you tell Perl to shut up even
485 when Perl sees that something is wrong. Do not be surprised if later
486 something locale-dependent misbehaves.
487
488 Perl can be run under the "C" locale by setting the environment
489 variable "LC_ALL" to "C". This method is perhaps a bit more civilized
490 than the "PERL_BADLANG" approach, but setting "LC_ALL" (or other locale
491 variables) may affect other programs as well, not just Perl. In
492 particular, external programs run from within Perl will see these
493 changes. If you make the new settings permanent (read on), all
494 programs you run see the changes. See "ENVIRONMENT" for the full list
495 of relevant environment variables and "USING LOCALES" for their effects
496 in Perl. Effects in other programs are easily deducible. For example,
497 the variable "LC_COLLATE" may well affect your sort program (or
498 whatever the program that arranges "records" alphabetically in your
499 system is called).
500
501 You can test out changing these variables temporarily, and if the new
502 settings seem to help, put those settings into your shell startup
503 files. Consult your local documentation for the exact details. For
504 Bourne-like shells (sh, ksh, bash, zsh):
505
506 LC_ALL=en_US.ISO8859-1
507 export LC_ALL
508
509 This assumes that we saw the locale "en_US.ISO8859-1" using the
510 commands discussed above. We decided to try that instead of the above
511 faulty locale "En_US"--and in Cshish shells (csh, tcsh)
512
513 setenv LC_ALL en_US.ISO8859-1
514
515 or if you have the "env" application you can do (in any shell)
516
517 env LC_ALL=en_US.ISO8859-1 perl ...
518
519 If you do not know what shell you have, consult your local helpdesk or
520 the equivalent.
521
522 Permanently fixing locale problems
523 The slower but superior fixes are when you may be able to yourself fix
524 the misconfiguration of your own environment variables. The
525 mis(sing)configuration of the whole system's locales usually requires
526 the help of your friendly system administrator.
527
528 First, see earlier in this document about "Finding locales". That
529 tells how to find which locales are really supported--and more
530 importantly, installed--on your system. In our example error message,
531 environment variables affecting the locale are listed in the order of
532 decreasing importance (and unset variables do not matter). Therefore,
533 having LC_ALL set to "En_US" must have been the bad choice, as shown by
534 the error message. First try fixing locale settings listed first.
535
536 Second, if using the listed commands you see something exactly (prefix
537 matches do not count and case usually counts) like "En_US" without the
538 quotes, then you should be okay because you are using a locale name
539 that should be installed and available in your system. In this case,
540 see "Permanently fixing your system's locale configuration".
541
542 Permanently fixing your system's locale configuration
543 This is when you see something like:
544
545 perl: warning: Please check that your locale settings:
546 LC_ALL = "En_US",
547 LANG = (unset)
548 are supported and installed on your system.
549
550 but then cannot see that "En_US" listed by the above-mentioned
551 commands. You may see things like "en_US.ISO8859-1", but that isn't
552 the same. In this case, try running under a locale that you can list
553 and which somehow matches what you tried. The rules for matching
554 locale names are a bit vague because standardization is weak in this
555 area. See again the "Finding locales" about general rules.
556
557 Fixing system locale configuration
558 Contact a system administrator (preferably your own) and report the
559 exact error message you get, and ask them to read this same
560 documentation you are now reading. They should be able to check
561 whether there is something wrong with the locale configuration of the
562 system. The "Finding locales" section is unfortunately a bit vague
563 about the exact commands and places because these things are not that
564 standardized.
565
566 The localeconv function
567 The "POSIX::localeconv()" function allows you to get particulars of the
568 locale-dependent numeric formatting information specified by the
569 current underlying "LC_NUMERIC" and "LC_MONETARY" locales (regardless
570 of whether called from within the scope of "use locale" or not). (If
571 you just want the name of the current locale for a particular category,
572 use "POSIX::setlocale()" with a single parameter--see "The setlocale
573 function".)
574
575 use POSIX qw(locale_h);
576
577 # Get a reference to a hash of locale-dependent info
578 $locale_values = localeconv();
579
580 # Output sorted list of the values
581 for (sort keys %$locale_values) {
582 printf "%-20s = %s\n", $_, $locale_values->{$_}
583 }
584
585 "localeconv()" takes no arguments, and returns a reference to a hash.
586 The keys of this hash are variable names for formatting, such as
587 "decimal_point" and "thousands_sep". The values are the corresponding,
588 er, values. See "localeconv" in POSIX for a longer example listing the
589 categories an implementation might be expected to provide; some provide
590 more and others fewer. You don't need an explicit "use locale",
591 because "localeconv()" always observes the current locale.
592
593 Here's a simple-minded example program that rewrites its command-line
594 parameters as integers correctly formatted in the current locale:
595
596 use POSIX qw(locale_h);
597
598 # Get some of locale's numeric formatting parameters
599 my ($thousands_sep, $grouping) =
600 @{localeconv()}{'thousands_sep', 'grouping'};
601
602 # Apply defaults if values are missing
603 $thousands_sep = ',' unless $thousands_sep;
604
605 # grouping and mon_grouping are packed lists
606 # of small integers (characters) telling the
607 # grouping (thousand_seps and mon_thousand_seps
608 # being the group dividers) of numbers and
609 # monetary quantities. The integers' meanings:
610 # 255 means no more grouping, 0 means repeat
611 # the previous grouping, 1-254 means use that
612 # as the current grouping. Grouping goes from
613 # right to left (low to high digits). In the
614 # below we cheat slightly by never using anything
615 # else than the first grouping (whatever that is).
616 if ($grouping) {
617 @grouping = unpack("C*", $grouping);
618 } else {
619 @grouping = (3);
620 }
621
622 # Format command line params for current locale
623 for (@ARGV) {
624 $_ = int; # Chop non-integer part
625 1 while
626 s/(\d)(\d{$grouping[0]}($|$thousands_sep))/$1$thousands_sep$2/;
627 print "$_";
628 }
629 print "\n";
630
631 Note that if the platform doesn't have "LC_NUMERIC" and/or
632 "LC_MONETARY" available or enabled, the corresponding elements of the
633 hash will be missing.
634
635 I18N::Langinfo
636 Another interface for querying locale-dependent information is the
637 "I18N::Langinfo::langinfo()" function, available at least in Unix-like
638 systems and VMS.
639
640 The following example will import the "langinfo()" function itself and
641 three constants to be used as arguments to "langinfo()": a constant for
642 the abbreviated first day of the week (the numbering starts from Sunday
643 = 1) and two more constants for the affirmative and negative answers
644 for a yes/no question in the current locale.
645
646 use I18N::Langinfo qw(langinfo ABDAY_1 YESSTR NOSTR);
647
648 my ($abday_1, $yesstr, $nostr)
649 = map { langinfo } qw(ABDAY_1 YESSTR NOSTR);
650
651 print "$abday_1? [$yesstr/$nostr] ";
652
653 In other words, in the "C" (or English) locale the above will probably
654 print something like:
655
656 Sun? [yes/no]
657
658 See I18N::Langinfo for more information.
659
661 The following subsections describe basic locale categories. Beyond
662 these, some combination categories allow manipulation of more than one
663 basic category at a time. See "ENVIRONMENT" for a discussion of these.
664
665 Category "LC_COLLATE": Collation: Text Comparisons and Sorting
666 In the scope of a "use locale" form that includes collation, Perl looks
667 to the "LC_COLLATE" environment variable to determine the application's
668 notions on collation (ordering) of characters. For example, "b"
669 follows "a" in Latin alphabets, but where do "a" and "aa" belong? And
670 while "color" follows "chocolate" in English, what about in traditional
671 Spanish?
672
673 The following collations all make sense and you may meet any of them if
674 you "use locale".
675
676 A B C D E a b c d e
677 A a B b C c D d E e
678 a A b B c C d D e E
679 a b c d e A B C D E
680
681 Here is a code snippet to tell what "word" characters are in the
682 current locale, in that locale's order:
683
684 use locale;
685 print +(sort grep /\w/, map { chr } 0..255), "\n";
686
687 Compare this with the characters that you see and their order if you
688 state explicitly that the locale should be ignored:
689
690 no locale;
691 print +(sort grep /\w/, map { chr } 0..255), "\n";
692
693 This machine-native collation (which is what you get unless
694 "use locale" has appeared earlier in the same block) must be used for
695 sorting raw binary data, whereas the locale-dependent collation of the
696 first example is useful for natural text.
697
698 As noted in "USING LOCALES", "cmp" compares according to the current
699 collation locale when "use locale" is in effect, but falls back to a
700 char-by-char comparison for strings that the locale says are equal. You
701 can use "POSIX::strcoll()" if you don't want this fall-back:
702
703 use POSIX qw(strcoll);
704 $equal_in_locale =
705 !strcoll("space and case ignored", "SpaceAndCaseIgnored");
706
707 $equal_in_locale will be true if the collation locale specifies a
708 dictionary-like ordering that ignores space characters completely and
709 which folds case.
710
711 Perl uses the platform's C library collation functions "strcoll()" and
712 "strxfrm()". That means you get whatever they give. On some
713 platforms, these functions work well on UTF-8 locales, giving a
714 reasonable default collation for the code points that are important in
715 that locale. (And if they aren't working well, the problem may only be
716 that the locale definition is deficient, so can be fixed by using a
717 better definition file. Unicode's definitions (see "Freely available
718 locale definitions") provide reasonable UTF-8 locale collation
719 definitions.) Starting in Perl v5.26, Perl's use of these functions
720 has been made more seamless. This may be sufficient for your needs.
721 For more control, and to make sure strings containing any code point
722 (not just the ones important in the locale) collate properly, the
723 Unicode::Collate module is suggested.
724
725 In non-UTF-8 locales (hence single byte), code points above 0xFF are
726 technically invalid. But if present, again starting in v5.26, they
727 will collate to the same position as the highest valid code point does.
728 This generally gives good results, but the collation order may be
729 skewed if the valid code point gets special treatment when it forms
730 particular sequences with other characters as defined by the locale.
731 When two strings collate identically, the code point order is used as a
732 tie breaker.
733
734 If Perl detects that there are problems with the locale collation
735 order, it reverts to using non-locale collation rules for that locale.
736
737 If Perl detects that there are problems with the locale collation
738 order, it reverts to using non-locale collation rules for that locale.
739
740 If you have a single string that you want to check for "equality in
741 locale" against several others, you might think you could gain a little
742 efficiency by using "POSIX::strxfrm()" in conjunction with "eq":
743
744 use POSIX qw(strxfrm);
745 $xfrm_string = strxfrm("Mixed-case string");
746 print "locale collation ignores spaces\n"
747 if $xfrm_string eq strxfrm("Mixed-casestring");
748 print "locale collation ignores hyphens\n"
749 if $xfrm_string eq strxfrm("Mixedcase string");
750 print "locale collation ignores case\n"
751 if $xfrm_string eq strxfrm("mixed-case string");
752
753 "strxfrm()" takes a string and maps it into a transformed string for
754 use in char-by-char comparisons against other transformed strings
755 during collation. "Under the hood", locale-affected Perl comparison
756 operators call "strxfrm()" for both operands, then do a char-by-char
757 comparison of the transformed strings. By calling "strxfrm()"
758 explicitly and using a non locale-affected comparison, the example
759 attempts to save a couple of transformations. But in fact, it doesn't
760 save anything: Perl magic (see "Magic Variables" in perlguts) creates
761 the transformed version of a string the first time it's needed in a
762 comparison, then keeps this version around in case it's needed again.
763 An example rewritten the easy way with "cmp" runs just about as fast.
764 It also copes with null characters embedded in strings; if you call
765 "strxfrm()" directly, it treats the first null it finds as a
766 terminator. don't expect the transformed strings it produces to be
767 portable across systems--or even from one revision of your operating
768 system to the next. In short, don't call "strxfrm()" directly: let
769 Perl do it for you.
770
771 Note: "use locale" isn't shown in some of these examples because it
772 isn't needed: "strcoll()" and "strxfrm()" are POSIX functions which use
773 the standard system-supplied "libc" functions that always obey the
774 current "LC_COLLATE" locale.
775
776 Category "LC_CTYPE": Character Types
777 In the scope of a "use locale" form that includes "LC_CTYPE", Perl
778 obeys the "LC_CTYPE" locale setting. This controls the application's
779 notion of which characters are alphabetic, numeric, punctuation, etc.
780 This affects Perl's "\w" regular expression metanotation, which stands
781 for alphanumeric characters--that is, alphabetic, numeric, and the
782 platform's native underscore. (Consult perlre for more information
783 about regular expressions.) Thanks to "LC_CTYPE", depending on your
784 locale setting, characters like "ae", "d`", "ss", and "o" may be
785 understood as "\w" characters. It also affects things like "\s", "\D",
786 and the POSIX character classes, like "[[:graph:]]". (See
787 perlrecharclass for more information on all these.)
788
789 The "LC_CTYPE" locale also provides the map used in transliterating
790 characters between lower and uppercase. This affects the case-mapping
791 functions--"fc()", "lc()", "lcfirst()", "uc()", and "ucfirst()"; case-
792 mapping interpolation with "\F", "\l", "\L", "\u", or "\U" in double-
793 quoted strings and "s///" substitutions; and case-independent regular
794 expression pattern matching using the "i" modifier.
795
796 Starting in v5.20, Perl supports UTF-8 locales for "LC_CTYPE", but
797 otherwise Perl only supports single-byte locales, such as the ISO 8859
798 series. This means that wide character locales, for example for Asian
799 languages, are not well-supported. Use of these locales may cause core
800 dumps. If the platform has the capability for Perl to detect such a
801 locale, starting in Perl v5.22, Perl will warn, default enabled, using
802 the "locale" warning category, whenever such a locale is switched into.
803 The UTF-8 locale support is actually a superset of POSIX locales,
804 because it is really full Unicode behavior as if no "LC_CTYPE" locale
805 were in effect at all (except for tainting; see "SECURITY"). POSIX
806 locales, even UTF-8 ones, are lacking certain concepts in Unicode, such
807 as the idea that changing the case of a character could expand to be
808 more than one character. Perl in a UTF-8 locale, will give you that
809 expansion. Prior to v5.20, Perl treated a UTF-8 locale on some
810 platforms like an ISO 8859-1 one, with some restrictions, and on other
811 platforms more like the "C" locale. For releases v5.16 and v5.18,
812 "use locale 'not_characters" could be used as a workaround for this
813 (see "Unicode and UTF-8").
814
815 Note that there are quite a few things that are unaffected by the
816 current locale. Any literal character is the native character for the
817 given platform. Hence 'A' means the character at code point 65 on
818 ASCII platforms, and 193 on EBCDIC. That may or may not be an 'A' in
819 the current locale, if that locale even has an 'A'. Similarly, all the
820 escape sequences for particular characters, "\n" for example, always
821 mean the platform's native one. This means, for example, that "\N" in
822 regular expressions (every character but new-line) works on the
823 platform character set.
824
825 Starting in v5.22, Perl will by default warn when switching into a
826 locale that redefines any ASCII printable character (plus "\t" and
827 "\n") into a different class than expected. This is likely to happen
828 on modern locales only on EBCDIC platforms, where, for example, a CCSID
829 0037 locale on a CCSID 1047 machine moves "[", but it can happen on
830 ASCII platforms with the ISO 646 and other 7-bit locales that are
831 essentially obsolete. Things may still work, depending on what
832 features of Perl are used by the program. For example, in the example
833 from above where "|" becomes a "\w", and there are no regular
834 expressions where this matters, the program may still work properly.
835 The warning lists all the characters that it can determine could be
836 adversely affected.
837
838 Note: A broken or malicious "LC_CTYPE" locale definition may result in
839 clearly ineligible characters being considered to be alphanumeric by
840 your application. For strict matching of (mundane) ASCII letters and
841 digits--for example, in command strings--locale-aware applications
842 should use "\w" with the "/a" regular expression modifier. See
843 "SECURITY".
844
845 Category "LC_NUMERIC": Numeric Formatting
846 After a proper "POSIX::setlocale()" call, and within the scope of of a
847 "use locale" form that includes numerics, Perl obeys the "LC_NUMERIC"
848 locale information, which controls an application's idea of how numbers
849 should be formatted for human readability. In most implementations the
850 only effect is to change the character used for the decimal
851 point--perhaps from "." to ",". The functions aren't aware of such
852 niceties as thousands separation and so on. (See "The localeconv
853 function" if you care about these things.)
854
855 use POSIX qw(strtod setlocale LC_NUMERIC);
856 use locale;
857
858 setlocale LC_NUMERIC, "";
859
860 $n = 5/2; # Assign numeric 2.5 to $n
861
862 $a = " $n"; # Locale-dependent conversion to string
863
864 print "half five is $n\n"; # Locale-dependent output
865
866 printf "half five is %g\n", $n; # Locale-dependent output
867
868 print "DECIMAL POINT IS COMMA\n"
869 if $n == (strtod("2,5"))[0]; # Locale-dependent conversion
870
871 See also I18N::Langinfo and "RADIXCHAR".
872
873 Category "LC_MONETARY": Formatting of monetary amounts
874 The C standard defines the "LC_MONETARY" category, but not a function
875 that is affected by its contents. (Those with experience of standards
876 committees will recognize that the working group decided to punt on the
877 issue.) Consequently, Perl essentially takes no notice of it. If you
878 really want to use "LC_MONETARY", you can query its contents--see "The
879 localeconv function"--and use the information that it returns in your
880 application's own formatting of currency amounts. However, you may
881 well find that the information, voluminous and complex though it may
882 be, still does not quite meet your requirements: currency formatting is
883 a hard nut to crack.
884
885 See also I18N::Langinfo and "CRNCYSTR".
886
887 Category "LC_TIME": Respresentation of time
888 Output produced by "POSIX::strftime()", which builds a formatted human-
889 readable date/time string, is affected by the current "LC_TIME" locale.
890 Thus, in a French locale, the output produced by the %B format element
891 (full month name) for the first month of the year would be "janvier".
892 Here's how to get a list of long month names in the current locale:
893
894 use POSIX qw(strftime);
895 for (0..11) {
896 $long_month_name[$_] =
897 strftime("%B", 0, 0, 0, 1, $_, 96);
898 }
899
900 Note: "use locale" isn't needed in this example: "strftime()" is a
901 POSIX function which uses the standard system-supplied "libc" function
902 that always obeys the current "LC_TIME" locale.
903
904 See also I18N::Langinfo and "ABDAY_1".."ABDAY_7", "DAY_1".."DAY_7",
905 "ABMON_1".."ABMON_12", and "ABMON_1".."ABMON_12".
906
907 Other categories
908 The remaining locale categories are not currently used by Perl itself.
909 But again note that things Perl interacts with may use these, including
910 extensions outside the standard Perl distribution, and by the operating
911 system and its utilities. Note especially that the string value of $!
912 and the error messages given by external utilities may be changed by
913 "LC_MESSAGES". If you want to have portable error codes, use "%!".
914 See Errno.
915
917 Although the main discussion of Perl security issues can be found in
918 perlsec, a discussion of Perl's locale handling would be incomplete if
919 it did not draw your attention to locale-dependent security issues.
920 Locales--particularly on systems that allow unprivileged users to build
921 their own locales--are untrustworthy. A malicious (or just plain
922 broken) locale can make a locale-aware application give unexpected
923 results. Here are a few possibilities:
924
925 · Regular expression checks for safe file names or mail addresses
926 using "\w" may be spoofed by an "LC_CTYPE" locale that claims that
927 characters such as ">" and "|" are alphanumeric.
928
929 · String interpolation with case-mapping, as in, say, "$dest =
930 "C:\U$name.$ext"", may produce dangerous results if a bogus
931 "LC_CTYPE" case-mapping table is in effect.
932
933 · A sneaky "LC_COLLATE" locale could result in the names of students
934 with "D" grades appearing ahead of those with "A"s.
935
936 · An application that takes the trouble to use information in
937 "LC_MONETARY" may format debits as if they were credits and vice
938 versa if that locale has been subverted. Or it might make payments
939 in US dollars instead of Hong Kong dollars.
940
941 · The date and day names in dates formatted by "strftime()" could be
942 manipulated to advantage by a malicious user able to subvert the
943 "LC_DATE" locale. ("Look--it says I wasn't in the building on
944 Sunday.")
945
946 Such dangers are not peculiar to the locale system: any aspect of an
947 application's environment which may be modified maliciously presents
948 similar challenges. Similarly, they are not specific to Perl: any
949 programming language that allows you to write programs that take
950 account of their environment exposes you to these issues.
951
952 Perl cannot protect you from all possibilities shown in the
953 examples--there is no substitute for your own vigilance--but, when "use
954 locale" is in effect, Perl uses the tainting mechanism (see perlsec) to
955 mark string results that become locale-dependent, and which may be
956 untrustworthy in consequence. Here is a summary of the tainting
957 behavior of operators and functions that may be affected by the locale:
958
959 · Comparison operators ("lt", "le", "ge", "gt" and "cmp"):
960
961 Scalar true/false (or less/equal/greater) result is never tainted.
962
963 · Case-mapping interpolation (with "\l", "\L", "\u", "\U", or "\F")
964
965 The result string containing interpolated material is tainted if a
966 "use locale" form that includes "LC_CTYPE" is in effect.
967
968 · Matching operator ("m//"):
969
970 Scalar true/false result never tainted.
971
972 All subpatterns, either delivered as a list-context result or as $1
973 etc., are tainted if a "use locale" form that includes "LC_CTYPE"
974 is in effect, and the subpattern regular expression contains a
975 locale-dependent construct. These constructs include "\w" (to
976 match an alphanumeric character), "\W" (non-alphanumeric
977 character), "\b" and "\B" (word-boundary and non-boundardy, which
978 depend on what "\w" and "\W" match), "\s" (whitespace character),
979 "\S" (non whitespace character), "\d" and "\D" (digits and non-
980 digits), and the POSIX character classes, such as "[:alpha:]" (see
981 "POSIX Character Classes" in perlrecharclass).
982
983 Tainting is also likely if the pattern is to be matched case-
984 insensitively (via "/i"). The exception is if all the code points
985 to be matched this way are above 255 and do not have folds under
986 Unicode rules to below 256. Tainting is not done for these because
987 Perl only uses Unicode rules for such code points, and those rules
988 are the same no matter what the current locale.
989
990 The matched-pattern variables, $&, "$`" (pre-match), "$'" (post-
991 match), and $+ (last match) also are tainted.
992
993 · Substitution operator ("s///"):
994
995 Has the same behavior as the match operator. Also, the left
996 operand of "=~" becomes tainted when a "use locale" form that
997 includes "LC_CTYPE" is in effect, if modified as a result of a
998 substitution based on a regular expression match involving any of
999 the things mentioned in the previous item, or of case-mapping, such
1000 as "\l", "\L","\u", "\U", or "\F".
1001
1002 · Output formatting functions ("printf()" and "write()"):
1003
1004 Results are never tainted because otherwise even output from print,
1005 for example "print(1/7)", should be tainted if "use locale" is in
1006 effect.
1007
1008 · Case-mapping functions ("lc()", "lcfirst()", "uc()", "ucfirst()"):
1009
1010 Results are tainted if a "use locale" form that includes "LC_CTYPE"
1011 is in effect.
1012
1013 · POSIX locale-dependent functions ("localeconv()", "strcoll()",
1014 "strftime()", "strxfrm()"):
1015
1016 Results are never tainted.
1017
1018 Three examples illustrate locale-dependent tainting. The first
1019 program, which ignores its locale, won't run: a value taken directly
1020 from the command line may not be used to name an output file when taint
1021 checks are enabled.
1022
1023 #/usr/local/bin/perl -T
1024 # Run with taint checking
1025
1026 # Command line sanity check omitted...
1027 $tainted_output_file = shift;
1028
1029 open(F, ">$tainted_output_file")
1030 or warn "Open of $tainted_output_file failed: $!\n";
1031
1032 The program can be made to run by "laundering" the tainted value
1033 through a regular expression: the second example--which still ignores
1034 locale information--runs, creating the file named on its command line
1035 if it can.
1036
1037 #/usr/local/bin/perl -T
1038
1039 $tainted_output_file = shift;
1040 $tainted_output_file =~ m%[\w/]+%;
1041 $untainted_output_file = $&;
1042
1043 open(F, ">$untainted_output_file")
1044 or warn "Open of $untainted_output_file failed: $!\n";
1045
1046 Compare this with a similar but locale-aware program:
1047
1048 #/usr/local/bin/perl -T
1049
1050 $tainted_output_file = shift;
1051 use locale;
1052 $tainted_output_file =~ m%[\w/]+%;
1053 $localized_output_file = $&;
1054
1055 open(F, ">$localized_output_file")
1056 or warn "Open of $localized_output_file failed: $!\n";
1057
1058 This third program fails to run because $& is tainted: it is the result
1059 of a match involving "\w" while "use locale" is in effect.
1060
1062 PERL_SKIP_LOCALE_INIT
1063 This environment variable, available starting in Perl
1064 v5.20, if set (to any value), tells Perl to not use the
1065 rest of the environment variables to initialize with.
1066 Instead, Perl uses whatever the current locale settings
1067 are. This is particularly useful in embedded environments,
1068 see "Using embedded Perl with POSIX locales" in perlembed.
1069
1070 PERL_BADLANG
1071 A string that can suppress Perl's warning about failed
1072 locale settings at startup. Failure can occur if the
1073 locale support in the operating system is lacking (broken)
1074 in some way--or if you mistyped the name of a locale when
1075 you set up your environment. If this environment variable
1076 is absent, or has a value other than "0" or "", Perl will
1077 complain about locale setting failures.
1078
1079 NOTE: "PERL_BADLANG" only gives you a way to hide the
1080 warning message. The message tells about some problem in
1081 your system's locale support, and you should investigate
1082 what the problem is.
1083
1084 The following environment variables are not specific to Perl: They are
1085 part of the standardized (ISO C, XPG4, POSIX 1.c) "setlocale()" method
1086 for controlling an application's opinion on data. Windows is non-
1087 POSIX, but Perl arranges for the following to work as described anyway.
1088 If the locale given by an environment variable is not valid, Perl tries
1089 the next lower one in priority. If none are valid, on Windows, the
1090 system default locale is then tried. If all else fails, the "C" locale
1091 is used. If even that doesn't work, something is badly broken, but
1092 Perl tries to forge ahead with whatever the locale settings might be.
1093
1094 "LC_ALL" "LC_ALL" is the "override-all" locale environment variable.
1095 If set, it overrides all the rest of the locale environment
1096 variables.
1097
1098 "LANGUAGE" NOTE: "LANGUAGE" is a GNU extension, it affects you only if
1099 you are using the GNU libc. This is the case if you are
1100 using e.g. Linux. If you are using "commercial" Unixes you
1101 are most probably not using GNU libc and you can ignore
1102 "LANGUAGE".
1103
1104 However, in the case you are using "LANGUAGE": it affects
1105 the language of informational, warning, and error messages
1106 output by commands (in other words, it's like
1107 "LC_MESSAGES") but it has higher priority than "LC_ALL".
1108 Moreover, it's not a single value but instead a "path"
1109 (":"-separated list) of languages (not locales). See the
1110 GNU "gettext" library documentation for more information.
1111
1112 "LC_CTYPE" In the absence of "LC_ALL", "LC_CTYPE" chooses the
1113 character type locale. In the absence of both "LC_ALL" and
1114 "LC_CTYPE", "LANG" chooses the character type locale.
1115
1116 "LC_COLLATE"
1117 In the absence of "LC_ALL", "LC_COLLATE" chooses the
1118 collation (sorting) locale. In the absence of both
1119 "LC_ALL" and "LC_COLLATE", "LANG" chooses the collation
1120 locale.
1121
1122 "LC_MONETARY"
1123 In the absence of "LC_ALL", "LC_MONETARY" chooses the
1124 monetary formatting locale. In the absence of both
1125 "LC_ALL" and "LC_MONETARY", "LANG" chooses the monetary
1126 formatting locale.
1127
1128 "LC_NUMERIC"
1129 In the absence of "LC_ALL", "LC_NUMERIC" chooses the
1130 numeric format locale. In the absence of both "LC_ALL" and
1131 "LC_NUMERIC", "LANG" chooses the numeric format.
1132
1133 "LC_TIME" In the absence of "LC_ALL", "LC_TIME" chooses the date and
1134 time formatting locale. In the absence of both "LC_ALL"
1135 and "LC_TIME", "LANG" chooses the date and time formatting
1136 locale.
1137
1138 "LANG" "LANG" is the "catch-all" locale environment variable. If
1139 it is set, it is used as the last resort after the overall
1140 "LC_ALL" and the category-specific "LC_foo".
1141
1142 Examples
1143 The "LC_NUMERIC" controls the numeric output:
1144
1145 use locale;
1146 use POSIX qw(locale_h); # Imports setlocale() and the LC_ constants.
1147 setlocale(LC_NUMERIC, "fr_FR") or die "Pardon";
1148 printf "%g\n", 1.23; # If the "fr_FR" succeeded, probably shows 1,23.
1149
1150 and also how strings are parsed by "POSIX::strtod()" as numbers:
1151
1152 use locale;
1153 use POSIX qw(locale_h strtod);
1154 setlocale(LC_NUMERIC, "de_DE") or die "Entschuldigung";
1155 my $x = strtod("2,34") + 5;
1156 print $x, "\n"; # Probably shows 7,34.
1157
1159 String "eval" and "LC_NUMERIC"
1160 A string eval parses its expression as standard Perl. It is therefore
1161 expecting the decimal point to be a dot. If "LC_NUMERIC" is set to
1162 have this be a comma instead, the parsing will be confused, perhaps
1163 silently.
1164
1165 use locale;
1166 use POSIX qw(locale_h);
1167 setlocale(LC_NUMERIC, "fr_FR") or die "Pardon";
1168 my $a = 1.2;
1169 print eval "$a + 1.5";
1170 print "\n";
1171
1172 prints "13,5". This is because in that locale, the comma is the
1173 decimal point character. The "eval" thus expands to:
1174
1175 eval "1,2 + 1.5"
1176
1177 and the result is not what you likely expected. No warnings are
1178 generated. If you do string "eval"'s within the scope of "use locale",
1179 you should instead change the "eval" line to do something like:
1180
1181 print eval "no locale; $a + 1.5";
1182
1183 This prints 2.7.
1184
1185 You could also exclude "LC_NUMERIC", if you don't need it, by
1186
1187 use locale ':!numeric';
1188
1189 Backward compatibility
1190 Versions of Perl prior to 5.004 mostly ignored locale information,
1191 generally behaving as if something similar to the "C" locale were
1192 always in force, even if the program environment suggested otherwise
1193 (see "The setlocale function"). By default, Perl still behaves this
1194 way for backward compatibility. If you want a Perl application to pay
1195 attention to locale information, you must use the "use locale" pragma
1196 (see "The "use locale" pragma") or, in the unlikely event that you want
1197 to do so for just pattern matching, the "/l" regular expression
1198 modifier (see "Character set modifiers" in perlre) to instruct it to do
1199 so.
1200
1201 Versions of Perl from 5.002 to 5.003 did use the "LC_CTYPE" information
1202 if available; that is, "\w" did understand what were the letters
1203 according to the locale environment variables. The problem was that
1204 the user had no control over the feature: if the C library supported
1205 locales, Perl used them.
1206
1207 I18N:Collate obsolete
1208 In versions of Perl prior to 5.004, per-locale collation was possible
1209 using the "I18N::Collate" library module. This module is now mildly
1210 obsolete and should be avoided in new applications. The "LC_COLLATE"
1211 functionality is now integrated into the Perl core language: One can
1212 use locale-specific scalar data completely normally with "use locale",
1213 so there is no longer any need to juggle with the scalar references of
1214 "I18N::Collate".
1215
1216 Sort speed and memory use impacts
1217 Comparing and sorting by locale is usually slower than the default
1218 sorting; slow-downs of two to four times have been observed. It will
1219 also consume more memory: once a Perl scalar variable has participated
1220 in any string comparison or sorting operation obeying the locale
1221 collation rules, it will take 3-15 times more memory than before. (The
1222 exact multiplier depends on the string's contents, the operating system
1223 and the locale.) These downsides are dictated more by the operating
1224 system's implementation of the locale system than by Perl.
1225
1226 Freely available locale definitions
1227 The Unicode CLDR project extracts the POSIX portion of many of its
1228 locales, available at
1229
1230 http://unicode.org/Public/cldr/2.0.1/
1231
1232 (Newer versions of CLDR require you to compute the POSIX data yourself.
1233 See <http://unicode.org/Public/cldr/latest/>.)
1234
1235 There is a large collection of locale definitions at:
1236
1237 http://std.dkuug.dk/i18n/WG15-collection/locales/
1238
1239 You should be aware that it is unsupported, and is not claimed to be
1240 fit for any purpose. If your system allows installation of arbitrary
1241 locales, you may find the definitions useful as they are, or as a basis
1242 for the development of your own locales.
1243
1244 I18n and l10n
1245 "Internationalization" is often abbreviated as i18n because its first
1246 and last letters are separated by eighteen others. (You may guess why
1247 the internalin ... internaliti ... i18n tends to get abbreviated.) In
1248 the same way, "localization" is often abbreviated to l10n.
1249
1250 An imperfect standard
1251 Internationalization, as defined in the C and POSIX standards, can be
1252 criticized as incomplete, ungainly, and having too large a granularity.
1253 (Locales apply to a whole process, when it would arguably be more
1254 useful to have them apply to a single thread, window group, or
1255 whatever.) They also have a tendency, like standards groups, to divide
1256 the world into nations, when we all know that the world can equally
1257 well be divided into bankers, bikers, gamers, and so on.
1258
1260 The support of Unicode is new starting from Perl version v5.6, and more
1261 fully implemented in versions v5.8 and later. See perluniintro.
1262
1263 Starting in Perl v5.20, UTF-8 locales are supported in Perl, except
1264 "LC_COLLATE" is only partially supported; collation support is improved
1265 in Perl v5.26 to a level that may be sufficient for your needs (see
1266 "Category "LC_COLLATE": Collation: Text Comparisons and Sorting").
1267
1268 If you have Perl v5.16 or v5.18 and can't upgrade, you can use
1269
1270 use locale ':not_characters';
1271
1272 When this form of the pragma is used, only the non-character portions
1273 of locales are used by Perl, for example "LC_NUMERIC". Perl assumes
1274 that you have translated all the characters it is to operate on into
1275 Unicode (actually the platform's native character set (ASCII or EBCDIC)
1276 plus Unicode). For data in files, this can conveniently be done by
1277 also specifying
1278
1279 use open ':locale';
1280
1281 This pragma arranges for all inputs from files to be translated into
1282 Unicode from the current locale as specified in the environment (see
1283 "ENVIRONMENT"), and all outputs to files to be translated back into the
1284 locale. (See open). On a per-filehandle basis, you can instead use
1285 the PerlIO::locale module, or the Encode::Locale module, both available
1286 from CPAN. The latter module also has methods to ease the handling of
1287 "ARGV" and environment variables, and can be used on individual
1288 strings. If you know that all your locales will be UTF-8, as many are
1289 these days, you can use the -C command line switch.
1290
1291 This form of the pragma allows essentially seamless handling of locales
1292 with Unicode. The collation order will be by Unicode code point order.
1293 Unicode::Collate can be used to get Unicode rules collation.
1294
1295 All the modules and switches just described can be used in v5.20 with
1296 just plain "use locale", and, should the input locales not be UTF-8,
1297 you'll get the less than ideal behavior, described below, that you get
1298 with pre-v5.16 Perls, or when you use the locale pragma without the
1299 ":not_characters" parameter in v5.16 and v5.18. If you are using
1300 exclusively UTF-8 locales in v5.20 and higher, the rest of this section
1301 does not apply to you.
1302
1303 There are two cases, multi-byte and single-byte locales. First multi-
1304 byte:
1305
1306 The only multi-byte (or wide character) locale that Perl is ever likely
1307 to support is UTF-8. This is due to the difficulty of implementation,
1308 the fact that high quality UTF-8 locales are now published for every
1309 area of the world (<http://unicode.org/Public/cldr/2.0.1/> for ones
1310 that are already set-up, but from an earlier version;
1311 <http://unicode.org/Public/cldr/latest/> for the most up-to-date, but
1312 you have to extract the POSIX information yourself), and that failing
1313 all that you can use the Encode module to translate to/from your
1314 locale. So, you'll have to do one of those things if you're using one
1315 of these locales, such as Big5 or Shift JIS. For UTF-8 locales, in
1316 Perls (pre v5.20) that don't have full UTF-8 locale support, they may
1317 work reasonably well (depending on your C library implementation)
1318 simply because both they and Perl store characters that take up
1319 multiple bytes the same way. However, some, if not most, C library
1320 implementations may not process the characters in the upper half of the
1321 Latin-1 range (128 - 255) properly under "LC_CTYPE". To see if a
1322 character is a particular type under a locale, Perl uses the functions
1323 like "isalnum()". Your C library may not work for UTF-8 locales with
1324 those functions, instead only working under the newer wide library
1325 functions like "iswalnum()", which Perl does not use. These multi-byte
1326 locales are treated like single-byte locales, and will have the
1327 restrictions described below. Starting in Perl v5.22 a warning message
1328 is raised when Perl detects a multi-byte locale that it doesn't fully
1329 support.
1330
1331 For single-byte locales, Perl generally takes the tack to use locale
1332 rules on code points that can fit in a single byte, and Unicode rules
1333 for those that can't (though this isn't uniformly applied, see the note
1334 at the end of this section). This prevents many problems in locales
1335 that aren't UTF-8. Suppose the locale is ISO8859-7, Greek. The
1336 character at 0xD7 there is a capital Chi. But in the ISO8859-1 locale,
1337 Latin1, it is a multiplication sign. The POSIX regular expression
1338 character class "[[:alpha:]]" will magically match 0xD7 in the Greek
1339 locale but not in the Latin one.
1340
1341 However, there are places where this breaks down. Certain Perl
1342 constructs are for Unicode only, such as "\p{Alpha}". They assume that
1343 0xD7 always has its Unicode meaning (or the equivalent on EBCDIC
1344 platforms). Since Latin1 is a subset of Unicode and 0xD7 is the
1345 multiplication sign in both Latin1 and Unicode, "\p{Alpha}" will never
1346 match it, regardless of locale. A similar issue occurs with "\N{...}".
1347 Prior to v5.20, It is therefore a bad idea to use "\p{}" or "\N{}"
1348 under plain "use locale"--unless you can guarantee that the locale will
1349 be ISO8859-1. Use POSIX character classes instead.
1350
1351 Another problem with this approach is that operations that cross the
1352 single byte/multiple byte boundary are not well-defined, and so are
1353 disallowed. (This boundary is between the codepoints at 255/256.) For
1354 example, lower casing LATIN CAPITAL LETTER Y WITH DIAERESIS (U+0178)
1355 should return LATIN SMALL LETTER Y WITH DIAERESIS (U+00FF). But in the
1356 Greek locale, for example, there is no character at 0xFF, and Perl has
1357 no way of knowing what the character at 0xFF is really supposed to
1358 represent. Thus it disallows the operation. In this mode, the
1359 lowercase of U+0178 is itself.
1360
1361 The same problems ensue if you enable automatic UTF-8-ification of your
1362 standard file handles, default "open()" layer, and @ARGV on
1363 non-ISO8859-1, non-UTF-8 locales (by using either the -C command line
1364 switch or the "PERL_UNICODE" environment variable; see perlrun).
1365 Things are read in as UTF-8, which would normally imply a Unicode
1366 interpretation, but the presence of a locale causes them to be
1367 interpreted in that locale instead. For example, a 0xD7 code point in
1368 the Unicode input, which should mean the multiplication sign, won't be
1369 interpreted by Perl that way under the Greek locale. This is not a
1370 problem provided you make certain that all locales will always and only
1371 be either an ISO8859-1, or, if you don't have a deficient C library, a
1372 UTF-8 locale.
1373
1374 Still another problem is that this approach can lead to two code points
1375 meaning the same character. Thus in a Greek locale, both U+03A7 and
1376 U+00D7 are GREEK CAPITAL LETTER CHI.
1377
1378 Because of all these problems, starting in v5.22, Perl will raise a
1379 warning if a multi-byte (hence Unicode) code point is used when a
1380 single-byte locale is in effect. (Although it doesn't check for this
1381 if doing so would unreasonably slow execution down.)
1382
1383 Vendor locales are notoriously buggy, and it is difficult for Perl to
1384 test its locale-handling code because this interacts with code that
1385 Perl has no control over; therefore the locale-handling code in Perl
1386 may be buggy as well. (However, the Unicode-supplied locales should be
1387 better, and there is a feed back mechanism to correct any problems.
1388 See "Freely available locale definitions".)
1389
1390 If you have Perl v5.16, the problems mentioned above go away if you use
1391 the ":not_characters" parameter to the locale pragma (except for vendor
1392 bugs in the non-character portions). If you don't have v5.16, and you
1393 do have locales that work, using them may be worthwhile for certain
1394 specific purposes, as long as you keep in mind the gotchas already
1395 mentioned. For example, if the collation for your locales works, it
1396 runs faster under locales than under Unicode::Collate; and you gain
1397 access to such things as the local currency symbol and the names of the
1398 months and days of the week. (But to hammer home the point, in v5.16,
1399 you get this access without the downsides of locales by using the
1400 ":not_characters" form of the pragma.)
1401
1402 Note: The policy of using locale rules for code points that can fit in
1403 a byte, and Unicode rules for those that can't is not uniformly
1404 applied. Pre-v5.12, it was somewhat haphazard; in v5.12 it was applied
1405 fairly consistently to regular expression matching except for bracketed
1406 character classes; in v5.14 it was extended to all regex matches; and
1407 in v5.16 to the casing operations such as "\L" and "uc()". For
1408 collation, in all releases so far, the system's "strxfrm()" function is
1409 called, and whatever it does is what you get. Starting in v5.26,
1410 various bugs are fixed with the way perl uses this function.
1411
1413 Collation of strings containing embedded "NUL" characters
1414 "NUL" characters will sort the same as the lowest collating control
1415 character does, or to "\001" in the unlikely event that there are no
1416 control characters at all in the locale. In cases where the strings
1417 don't contain this non-"NUL" control, the results will be correct, and
1418 in many locales, this control, whatever it might be, will rarely be
1419 encountered. But there are cases where a "NUL" should sort before this
1420 control, but doesn't. If two strings do collate identically, the one
1421 containing the "NUL" will sort to earlier.
1422
1423 Broken systems
1424 In certain systems, the operating system's locale support is broken and
1425 cannot be fixed or used by Perl. Such deficiencies can and will result
1426 in mysterious hangs and/or Perl core dumps when "use locale" is in
1427 effect. When confronted with such a system, please report in
1428 excruciating detail to <perlbug@perl.org>, and also contact your
1429 vendor: bug fixes may exist for these problems in your operating
1430 system. Sometimes such bug fixes are called an operating system
1431 upgrade. If you have the source for Perl, include in the perlbug email
1432 the output of the test described above in "Testing for broken locales".
1433
1435 I18N::Langinfo, perluniintro, perlunicode, open, "isalnum" in POSIX,
1436 "isalpha" in POSIX, "isdigit" in POSIX, "isgraph" in POSIX, "islower"
1437 in POSIX, "isprint" in POSIX, "ispunct" in POSIX, "isspace" in POSIX,
1438 "isupper" in POSIX, "isxdigit" in POSIX, "localeconv" in POSIX,
1439 "setlocale" in POSIX, "strcoll" in POSIX, "strftime" in POSIX, "strtod"
1440 in POSIX, "strxfrm" in POSIX.
1441
1442 For special considerations when Perl is embedded in a C program, see
1443 "Using embedded Perl with POSIX locales" in perlembed.
1444
1446 Jarkko Hietaniemi's original perli18n.pod heavily hacked by Dominic
1447 Dunlop, assisted by the perl5-porters. Prose worked over a bit by Tom
1448 Christiansen, and updated by Perl 5 porters.
1449
1450
1451
1452perl v5.26.3 2018-03-23 PERLLOCALE(1)