This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: Exceptions in IPC::Open2
[perl5.git] / pod / perllocale.pod
CommitLineData
5f05dabc
PP
1=head1 NAME
2
b0c42ed9 3perllocale - Perl locale handling (internationalization and localization)
5f05dabc
PP
4
5=head1 DESCRIPTION
6
7Perl supports language-specific notions of data such as "is this a
68dc0745 8letter", "what is the uppercase equivalent of this letter", and "which
14280422
DD
9of these letters comes first". These are important issues, especially
10for languages other than English - but also for English: it would be
11very naE<iuml>ve to think that C<A-Za-z> defines all the "letters". Perl
12is also aware that some character other than '.' may be preferred as a
13decimal point, and that output date representations may be
14language-specific. The process of making an application take account of
15its users' preferences in such matters is called B<internationalization>
16(often abbreviated as B<i18n>); telling such an application about a
17particular set of preferences is known as B<localization> (B<l10n>).
18
19Perl can understand language-specific data via the standardized (ISO C,
20XPG4, POSIX 1.c) method called "the locale system". The locale system is
b0c42ed9 21controlled per application using one pragma, one function call, and
14280422
DD
22several environment variables.
23
24B<NOTE>: This feature is new in Perl 5.004, and does not apply unless an
25application specifically requests it - see L<Backward compatibility>.
e38874e2
DD
26The one exception is that write() now B<always> uses the current locale
27- see L<"NOTES">.
5f05dabc
PP
28
29=head1 PREPARING TO USE LOCALES
30
14280422
DD
31If Perl applications are to be able to understand and present your data
32correctly according a locale of your choice, B<all> of the following
5f05dabc
PP
33must be true:
34
35=over 4
36
37=item *
38
39B<Your operating system must support the locale system>. If it does,
14280422 40you should find that the setlocale() function is a documented part of
5f05dabc
PP
41its C library.
42
43=item *
44
14280422
DD
45B<Definitions for the locales which you use must be installed>. You, or
46your system administrator, must make sure that this is the case. The
47available locales, the location in which they are kept, and the manner
48in which they are installed, vary from system to system. Some systems
4a6725af 49provide only a few, hard-wired, locales, and do not allow more to be
14280422
DD
50added; others allow you to add "canned" locales provided by the system
51supplier; still others allow you or the system administrator to define
52and add arbitrary locales. (You may have to ask your supplier to
53provide canned locales which are not delivered with your operating
54system.) Read your system documentation for further illumination.
5f05dabc
PP
55
56=item *
57
58B<Perl must believe that the locale system is supported>. If it does,
59C<perl -V:d_setlocale> will say that the value for C<d_setlocale> is
60C<define>.
61
62=back
63
64If you want a Perl application to process and present your data
65according to a particular locale, the application code should include
2ae324a7 66the S<C<use locale>> pragma (see L<The use locale pragma>) where
5f05dabc
PP
67appropriate, and B<at least one> of the following must be true:
68
69=over 4
70
71=item *
72
14280422
DD
73B<The locale-determining environment variables (see L<"ENVIRONMENT">)
74must be correctly set up>, either by yourself, or by the person who set
75up your system account, at the time the application is started.
5f05dabc
PP
76
77=item *
78
14280422
DD
79B<The application must set its own locale> using the method described in
80L<The setlocale function>.
5f05dabc
PP
81
82=back
83
84=head1 USING LOCALES
85
86=head2 The use locale pragma
87
14280422
DD
88By default, Perl ignores the current locale. The S<C<use locale>>
89pragma tells Perl to use the current locale for some operations:
5f05dabc
PP
90
91=over 4
92
93=item *
94
14280422
DD
95B<The comparison operators> (C<lt>, C<le>, C<cmp>, C<ge>, and C<gt>) and
96the POSIX string collation functions strcoll() and strxfrm() use
97C<LC_COLLATE>. sort() is also affected if it is used without an
98explicit comparison function because it uses C<cmp> by default.
99
100B<Note:> C<eq> and C<ne> are unaffected by the locale: they always
101perform a byte-by-byte comparison of their scalar operands. What's
102more, if C<cmp> finds that its operands are equal according to the
103collation sequence specified by the current locale, it goes on to
104perform a byte-by-byte comparison, and only returns I<0> (equal) if the
105operands are bit-for-bit identical. If you really want to know whether
106two strings - which C<eq> and C<cmp> may consider different - are equal
107as far as collation in the locale is concerned, see the discussion in
108L<Category LC_COLLATE: Collation>.
5f05dabc
PP
109
110=item *
111
14280422
DD
112B<Regular expressions and case-modification functions> (uc(), lc(),
113ucfirst(), and lcfirst()) use C<LC_CTYPE>
5f05dabc
PP
114
115=item *
116
14280422 117B<The formatting functions> (printf(), sprintf() and write()) use
5f05dabc
PP
118C<LC_NUMERIC>
119
120=item *
121
14280422 122B<The POSIX date formatting function> (strftime()) uses C<LC_TIME>.
5f05dabc
PP
123
124=back
125
14280422
DD
126C<LC_COLLATE>, C<LC_CTYPE>, and so on, are discussed further in L<LOCALE
127CATEGORIES>.
5f05dabc 128
b0c42ed9 129The default behavior returns with S<C<no locale>> or on reaching the
14280422 130end of the enclosing block.
5f05dabc 131
14280422
DD
132Note that the string result of any operation that uses locale
133information is tainted, as it is possible for a locale to be
134untrustworthy. See L<"SECURITY">.
5f05dabc
PP
135
136=head2 The setlocale function
137
14280422
DD
138You can switch locales as often as you wish at run time with the
139POSIX::setlocale() function:
5f05dabc
PP
140
141 # This functionality not usable prior to Perl 5.004
142 require 5.004;
143
144 # Import locale-handling tool set from POSIX module.
145 # This example uses: setlocale -- the function call
146 # LC_CTYPE -- explained below
147 use POSIX qw(locale_h);
148
14280422 149 # query and save the old locale
5f05dabc
PP
150 $old_locale = setlocale(LC_CTYPE);
151
152 setlocale(LC_CTYPE, "fr_CA.ISO8859-1");
153 # LC_CTYPE now in locale "French, Canada, codeset ISO 8859-1"
154
155 setlocale(LC_CTYPE, "");
156 # LC_CTYPE now reset to default defined by LC_ALL/LC_CTYPE/LANG
157 # environment variables. See below for documentation.
158
159 # restore the old locale
160 setlocale(LC_CTYPE, $old_locale);
161
14280422
DD
162The first argument of setlocale() gives the B<category>, the second the
163B<locale>. The category tells in what aspect of data processing you
164want to apply locale-specific rules. Category names are discussed in
165L<LOCALE CATEGORIES> and L<"ENVIRONMENT">. The locale is the name of a
166collection of customization information corresponding to a particular
167combination of language, country or territory, and codeset. Read on for
168hints on the naming of locales: not all systems name locales as in the
169example.
170
171If no second argument is provided, the function returns a string naming
172the current locale for the category. You can use this value as the
173second argument in a subsequent call to setlocale(). If a second
5f05dabc
PP
174argument is given and it corresponds to a valid locale, the locale for
175the category is set to that value, and the function returns the
176now-current locale value. You can use this in a subsequent call to
14280422 177setlocale(). (In some implementations, the return value may sometimes
5f05dabc
PP
178differ from the value you gave as the second argument - think of it as
179an alias for the value that you gave.)
180
181As the example shows, if the second argument is an empty string, the
182category's locale is returned to the default specified by the
183corresponding environment variables. Generally, this results in a
184return to the default which was in force when Perl started up: changes
54310121 185to the environment made by the application after startup may or may not
14280422 186be noticed, depending on the implementation of your system's C library.
5f05dabc 187
14280422
DD
188If the second argument does not correspond to a valid locale, the locale
189for the category is not changed, and the function returns I<undef>.
5f05dabc 190
14280422 191For further information about the categories, consult L<setlocale(3)>.
3e6e419a
JH
192
193=head2 Finding locales
194
14280422
DD
195For the locales available in your system, also consult L<setlocale(3)>
196and see whether it leads you to the list of the available locales
197(search for the I<SEE ALSO> section). If that fails, try the following
198command lines:
5f05dabc
PP
199
200 locale -a
201
202 nlsinfo
203
204 ls /usr/lib/nls/loc
205
206 ls /usr/lib/locale
207
208 ls /usr/lib/nls
209
210and see whether they list something resembling these
211
2bdf8add
JH
212 en_US.ISO8859-1 de_DE.ISO8859-1 ru_RU.ISO8859-5
213 en_US de_DE ru_RU
14280422 214 en de ru
2bdf8add
JH
215 english german russian
216 english.iso88591 german.iso88591 russian.iso88595
5f05dabc 217
14280422 218Sadly, even though the calling interface for setlocale() has been
3e6e419a
JH
219standardized, the names of the locales and the directories where the
220configuration is, have not. The basic form of the name is
221I<language_country/territory>B<.>I<codeset>, but the latter parts
222after the I<language> are not always present. The I<language> and the
223I<country> are usually from the standards B<ISO 3166> and B<ISO 639>,
224respectively, the two-letter abbreviations for the countries and the
225languages of the world. The I<codeset> part often mentions some B<ISO
2268859> character set, the Latin codesets. For example the C<ISO
2278859-1> is the so-called "Western codeset" that can be used to encode
228most of the Western European languages. Again, sadly, as you can see,
229there are several ways to write even the name of that one standard.
5f05dabc 230
14280422
DD
231Two special locales are worth particular mention: "C" and "POSIX".
232Currently these are effectively the same locale: the difference is
233mainly that the first one is defined by the C standard and the second by
234the POSIX standard. What they define is the B<default locale> in which
235every program starts in the absence of locale information in its
236environment. (The default default locale, if you will.) Its language
237is (American) English and its character codeset ASCII.
5f05dabc 238
14280422
DD
239B<NOTE>: Not all systems have the "POSIX" locale (not all systems are
240POSIX-conformant), so use "C" when you need explicitly to specify this
241default locale.
5f05dabc 242
3e6e419a
JH
243=head2 LOCALE PROBLEMS
244
245You may meet the following warning message at Perl startup:
246
247 perl: warning: Setting locale failed.
248 perl: warning: Please check that your locale settings:
249 LC_ALL = "En_US",
250 LANG = (unset)
251 are supported and installed on your system.
252 perl: warning: Falling back to the standard locale ("C").
253
254This means that your locale settings were that LC_ALL equals "En_US"
255and LANG exists but has no value. Perl tried to believe you but it
256could not. Instead Perl gave up and fell back to the "C" locale, the
257default locale that is supposed to work no matter what. This usually
258means either or both of the two problems: either your locale settings
259were wrong, they talk of locales your system has never heard of, or
260that the locale installation in your system has problems, for example
261some system files are broken or missing. For the problems there are
262quick and temporary fixes and more thorough and lasting fixes.
263
264=head2 Temporarily fixing locale problems
265
266The two quickest fixes are either to make Perl be silent about any
267locale inconsistencies or to run Perl under the default locale "C".
268
269Perl's moaning about locale problems can be silenced by setting the
270environment variable PERL_BADLANG to a non-zero value, for example
271"1". This method really just sweeps the problem under the carpet: you
272tell Perl to shut up even when Perl sees that something is wrong. Do
273not be surprised if later something locale-dependent works funny.
274
275Perl can be run under the "C" locale by setting the environment
276variable LC_ALL to "C". This method is perhaps a bit more civilised
277than the PERL_BADLANG one but please note that setting the LC_ALL (or
278the other locale variables) may affect also other programs, not just
279Perl. Especially external programs run from within Perl will see
280these changes. If you make the new settings permanent (read on), all
281the programs you run will see the changes. See L<ENVIRONMENT> for for
282the full list of all the environment variables and L<USING LOCALES>
283for their effects in Perl. The effects in other programs are quite
284easily deducible: for example the variable LC_COLLATE may well affect
285your "sort" program (or whatever the program that arranges `records'
286alphabetically in your system is called).
287
288You can first try out changing these variables temporarily and if the
289new settings seem to help then put the settings into the startup files
290of your environment. Please consult your local documentation for the
291exact details but very shortly for UNIXish systems: in Bourneish
292shells (sh, ksh, bash, zsh) for example
293
294 LC_ALL=en_US.ISO8859-1
295 export LC_ALL
296
297We assume here that we saw with the above discussed commands the
298locale "en_US.ISO8859-1" and decided to try that instead of the above
299faulty locale "En_US" -- and in Cshish shells (csh, tcsh)
300
301 setenv LC_ALL en_US.ISO8859-1
302
303If you do not know what shell you have, please consult your local
304helpdesk or the equivalent.
305
306=head2 Permanently fixing locale problems
307
308Then the slower but better fixes: the misconfiguration of your own
309environment variables you may be able to fix yourself; the
310mis(sing)configuration of the whole system's locales usually requires
311the help of your friendly system administrator.
312
313First, see earlier in this document about L<Finding locales>. That
314tells how you can find which locales really are supported and more
315importantly, installed, in your system. In our example error message
316the environment variables affecting the locale are listed in the order
317of decreasing importance and unset variables do not matter, therefore
318in the above error message the LC_ALL being "En_US" must have been the
319bad choice. Always try fixing first the locale settings listed first.
320
321Second, if you see with the listed commands something B<exactly> (for
322example prefix matches do not count and case usually matters) like
323"En_US" (without the quotes), then you should be okay because you are
324using a locale name that should be installed and available in your
325system. In this case skip forward to L<Fixing the system locale
326configuration>.
327
328=head2 Permantently fixing your locale configuration
329
330This is the case when for example you see
331
332 perl: warning: Please check that your locale settings:
333 LC_ALL = "En_US",
334 LANG = (unset)
335 are supported and installed on your system.
336
337but then cannot see that "En_US" listed by the above-mentioned
338commands. You may see things like "en_US.ISO8859-1" but that is not
339the same thing. In this case you might try running under a locale
340that you could list and somehow matches with what you tried. The
341rules for matching locale names are a bit vague because
342standardisation is weak in this area. See again the L<Finding
343locales> about the general rules.
344
345=head2 Permanently fixing the system locale configuration
346
347Please contact your system administrator and tell her the exact error
348message you get and ask her to read this same documentation you are
349now reading. She should be able to check whether there is something
350wrong with the locale configuration of the system. The L<Finding
351locales> section is unfortunately a bit vague about the exact commands
352and places because these things are not that standardised.
353
5f05dabc
PP
354=head2 The localeconv function
355
14280422
DD
356The POSIX::localeconv() function allows you to get particulars of the
357locale-dependent numeric formatting information specified by the current
358C<LC_NUMERIC> and C<LC_MONETARY> locales. (If you just want the name of
359the current locale for a particular category, use POSIX::setlocale()
360with a single parameter - see L<The setlocale function>.)
5f05dabc
PP
361
362 use POSIX qw(locale_h);
5f05dabc
PP
363
364 # Get a reference to a hash of locale-dependent info
365 $locale_values = localeconv();
366
367 # Output sorted list of the values
368 for (sort keys %$locale_values) {
14280422 369 printf "%-20s = %s\n", $_, $locale_values->{$_}
5f05dabc
PP
370 }
371
14280422
DD
372localeconv() takes no arguments, and returns B<a reference to> a hash.
373The keys of this hash are formatting variable names such as
374C<decimal_point> and C<thousands_sep>; the values are the corresponding
375values. See L<POSIX (3)/localeconv> for a longer example, which lists
376all the categories an implementation might be expected to provide; some
377provide more and others fewer, however. Note that you don't need C<use
378locale>: as a function with the job of querying the locale, localeconv()
379always observes the current locale.
5f05dabc
PP
380
381Here's a simple-minded example program which rewrites its command line
382parameters as integers formatted correctly in the current locale:
383
384 # See comments in previous example
385 require 5.004;
386 use POSIX qw(locale_h);
5f05dabc
PP
387
388 # Get some of locale's numeric formatting parameters
389 my ($thousands_sep, $grouping) =
14280422 390 @{localeconv()}{'thousands_sep', 'grouping'};
5f05dabc
PP
391
392 # Apply defaults if values are missing
393 $thousands_sep = ',' unless $thousands_sep;
394 $grouping = 3 unless $grouping;
395
396 # Format command line params for current locale
14280422
DD
397 for (@ARGV) {
398 $_ = int; # Chop non-integer part
5f05dabc 399 1 while
14280422
DD
400 s/(\d)(\d{$grouping}($|$thousands_sep))/$1$thousands_sep$2/;
401 print "$_";
5f05dabc
PP
402 }
403 print "\n";
404
5f05dabc
PP
405=head1 LOCALE CATEGORIES
406
14280422 407The subsections which follow describe basic locale categories. As well
5f05dabc 408as these, there are some combination categories which allow the
14280422
DD
409manipulation of more than one basic category at a time. See
410L<"ENVIRONMENT"> for a discussion of these.
5f05dabc
PP
411
412=head2 Category LC_COLLATE: Collation
413
14280422 414When in the scope of S<C<use locale>>, Perl looks to the C<LC_COLLATE>
5f05dabc 415environment variable to determine the application's notions on the
14280422
DD
416collation (ordering) of characters. ('b' follows 'a' in Latin
417alphabets, but where do 'E<aacute>' and 'E<aring>' belong?)
5f05dabc
PP
418
419Here is a code snippet that will tell you what are the alphanumeric
420characters in the current locale, in the locale order:
421
422 use locale;
423 print +(sort grep /\w/, map { chr() } 0..255), "\n";
424
14280422
DD
425Compare this with the characters that you see and their order if you
426state explicitly that the locale should be ignored:
5f05dabc
PP
427
428 no locale;
429 print +(sort grep /\w/, map { chr() } 0..255), "\n";
430
431This machine-native collation (which is what you get unless S<C<use
432locale>> has appeared earlier in the same block) must be used for
433sorting raw binary data, whereas the locale-dependent collation of the
b0c42ed9 434first example is useful for natural text.
5f05dabc 435
14280422
DD
436As noted in L<USING LOCALES>, C<cmp> compares according to the current
437collation locale when C<use locale> is in effect, but falls back to a
438byte-by-byte comparison for strings which the locale says are equal. You
439can use POSIX::strcoll() if you don't want this fall-back:
440
441 use POSIX qw(strcoll);
442 $equal_in_locale =
443 !strcoll("space and case ignored", "SpaceAndCaseIgnored");
444
445$equal_in_locale will be true if the collation locale specifies a
446dictionary-like ordering which ignores space characters completely, and
9e3a2af8 447which folds case.
14280422
DD
448
449If you have a single string which you want to check for "equality in
450locale" against several others, you might think you could gain a little
451efficiency by using POSIX::strxfrm() in conjunction with C<eq>:
452
453 use POSIX qw(strxfrm);
454 $xfrm_string = strxfrm("Mixed-case string");
455 print "locale collation ignores spaces\n"
456 if $xfrm_string eq strxfrm("Mixed-casestring");
457 print "locale collation ignores hyphens\n"
458 if $xfrm_string eq strxfrm("Mixedcase string");
459 print "locale collation ignores case\n"
460 if $xfrm_string eq strxfrm("mixed-case string");
461
462strxfrm() takes a string and maps it into a transformed string for use
463in byte-by-byte comparisons against other transformed strings during
464collation. "Under the hood", locale-affected Perl comparison operators
465call strxfrm() for both their operands, then do a byte-by-byte
466comparison of the transformed strings. By calling strxfrm() explicitly,
467and using a non locale-affected comparison, the example attempts to save
468a couple of transformations. In fact, it doesn't save anything: Perl
2ae324a7 469magic (see L<perlguts/Magic Variables>) creates the transformed version of a
14280422
DD
470string the first time it's needed in a comparison, then keeps it around
471in case it's needed again. An example rewritten the easy way with
e38874e2 472C<cmp> runs just about as fast. It also copes with null characters
14280422 473embedded in strings; if you call strxfrm() directly, it treats the first
e38874e2
DD
474null it finds as a terminator. And don't expect the transformed strings
475it produces to be portable across systems - or even from one revision
476of your operating system to the next. In short, don't call strxfrm()
477directly: let Perl do it for you.
14280422
DD
478
479Note: C<use locale> isn't shown in some of these examples, as it isn't
480needed: strcoll() and strxfrm() exist only to generate locale-dependent
481results, and so always obey the current C<LC_COLLATE> locale.
5f05dabc
PP
482
483=head2 Category LC_CTYPE: Character Types
484
485When in the scope of S<C<use locale>>, Perl obeys the C<LC_CTYPE> locale
14280422
DD
486setting. This controls the application's notion of which characters are
487alphabetic. This affects Perl's C<\w> regular expression metanotation,
488which stands for alphanumeric characters - that is, alphabetic and
489numeric characters. (Consult L<perlre> for more information about
490regular expressions.) Thanks to C<LC_CTYPE>, depending on your locale
491setting, characters like 'E<aelig>', 'E<eth>', 'E<szlig>', and
492'E<oslash>' may be understood as C<\w> characters.
5f05dabc 493
2c268ad5 494The C<LC_CTYPE> locale also provides the map used in transliterating
68dc0745 495characters between lower and uppercase. This affects the case-mapping
e38874e2 496functions - lc(), lcfirst, uc() and ucfirst(); case-mapping
7b8d334a 497interpolation with C<\l>, C<\L>, C<\u> or C<\U> in double-quoted strings
e38874e2
DD
498and in C<s///> substitutions; and case-independent regular expression
499pattern matching using the C<i> modifier.
500
501Finally, C<LC_CTYPE> affects the POSIX character-class test functions -
14280422
DD
502isalpha(), islower() and so on. For example, if you move from the "C"
503locale to a 7-bit Scandinavian one, you may find - possibly to your
504surprise - that "|" moves from the ispunct() class to isalpha().
5f05dabc 505
14280422
DD
506B<Note:> A broken or malicious C<LC_CTYPE> locale definition may result
507in clearly ineligible characters being considered to be alphanumeric by
508your application. For strict matching of (unaccented) letters and
509digits - for example, in command strings - locale-aware applications
510should use C<\w> inside a C<no locale> block. See L<"SECURITY">.
5f05dabc
PP
511
512=head2 Category LC_NUMERIC: Numeric Formatting
513
514When in the scope of S<C<use locale>>, Perl obeys the C<LC_NUMERIC>
14280422
DD
515locale information, which controls application's idea of how numbers
516should be formatted for human readability by the printf(), sprintf(),
517and write() functions. String to numeric conversion by the
518POSIX::strtod() function is also affected. In most implementations the
519only effect is to change the character used for the decimal point -
520perhaps from '.' to ',': these functions aren't aware of such niceties
521as thousands separation and so on. (See L<The localeconv function> if
522you care about these things.)
523
524Note that output produced by print() is B<never> affected by the
5f05dabc 525current locale: it is independent of whether C<use locale> or C<no
14280422 526locale> is in effect, and corresponds to what you'd get from printf()
5f05dabc
PP
527in the "C" locale. The same is true for Perl's internal conversions
528between numeric and string formats:
529
530 use POSIX qw(strtod);
531 use locale;
14280422 532
5f05dabc
PP
533 $n = 5/2; # Assign numeric 2.5 to $n
534
535 $a = " $n"; # Locale-independent conversion to string
536
537 print "half five is $n\n"; # Locale-independent output
538
539 printf "half five is %g\n", $n; # Locale-dependent output
540
14280422
DD
541 print "DECIMAL POINT IS COMMA\n"
542 if $n == (strtod("2,5"))[0]; # Locale-dependent conversion
5f05dabc
PP
543
544=head2 Category LC_MONETARY: Formatting of monetary amounts
545
14280422
DD
546The C standard defines the C<LC_MONETARY> category, but no function that
547is affected by its contents. (Those with experience of standards
b0c42ed9 548committees will recognize that the working group decided to punt on the
14280422
DD
549issue.) Consequently, Perl takes no notice of it. If you really want
550to use C<LC_MONETARY>, you can query its contents - see L<The localeconv
551function> - and use the information that it returns in your
b0c42ed9 552application's own formatting of currency amounts. However, you may well
14280422
DD
553find that the information, though voluminous and complex, does not quite
554meet your requirements: currency formatting is a hard nut to crack.
5f05dabc
PP
555
556=head2 LC_TIME
557
14280422 558The output produced by POSIX::strftime(), which builds a formatted
5f05dabc
PP
559human-readable date/time string, is affected by the current C<LC_TIME>
560locale. Thus, in a French locale, the output produced by the C<%B>
561format element (full month name) for the first month of the year would
562be "janvier". Here's how to get a list of the long month names in the
563current locale:
564
565 use POSIX qw(strftime);
14280422
DD
566 for (0..11) {
567 $long_month_name[$_] =
568 strftime("%B", 0, 0, 0, 1, $_, 96);
5f05dabc
PP
569 }
570
14280422
DD
571Note: C<use locale> isn't needed in this example: as a function which
572exists only to generate locale-dependent results, strftime() always
573obeys the current C<LC_TIME> locale.
5f05dabc
PP
574
575=head2 Other categories
576
577The remaining locale category, C<LC_MESSAGES> (possibly supplemented by
578others in particular implementations) is not currently used by Perl -
b0c42ed9 579except possibly to affect the behavior of library functions called by
14280422
DD
580extensions which are not part of the standard Perl distribution.
581
582=head1 SECURITY
583
584While the main discussion of Perl security issues can be found in
585L<perlsec>, a discussion of Perl's locale handling would be incomplete
586if it did not draw your attention to locale-dependent security issues.
587Locales - particularly on systems which allow unprivileged users to
588build their own locales - are untrustworthy. A malicious (or just plain
589broken) locale can make a locale-aware application give unexpected
590results. Here are a few possibilities:
591
592=over 4
593
594=item *
595
596Regular expression checks for safe file names or mail addresses using
597C<\w> may be spoofed by an C<LC_CTYPE> locale which claims that
598characters such as "E<gt>" and "|" are alphanumeric.
599
600=item *
601
e38874e2
DD
602String interpolation with case-mapping, as in, say, C<$dest =
603"C:\U$name.$ext">, may produce dangerous results if a bogus LC_CTYPE
604case-mapping table is in effect.
605
606=item *
607
14280422
DD
608If the decimal point character in the C<LC_NUMERIC> locale is
609surreptitiously changed from a dot to a comma, C<sprintf("%g",
6100.123456e3)> produces a string result of "123,456". Many people would
611interpret this as one hundred and twenty-three thousand, four hundred
612and fifty-six.
613
614=item *
615
616A sneaky C<LC_COLLATE> locale could result in the names of students with
617"D" grades appearing ahead of those with "A"s.
618
619=item *
620
621An application which takes the trouble to use the information in
622C<LC_MONETARY> may format debits as if they were credits and vice versa
623if that locale has been subverted. Or it make may make payments in US
624dollars instead of Hong Kong dollars.
625
626=item *
627
628The date and day names in dates formatted by strftime() could be
629manipulated to advantage by a malicious user able to subvert the
630C<LC_DATE> locale. ("Look - it says I wasn't in the building on
631Sunday.")
632
633=back
634
635Such dangers are not peculiar to the locale system: any aspect of an
636application's environment which may maliciously be modified presents
637similar challenges. Similarly, they are not specific to Perl: any
638programming language which allows you to write programs which take
639account of their environment exposes you to these issues.
640
641Perl cannot protect you from all of the possibilities shown in the
642examples - there is no substitute for your own vigilance - but, when
643C<use locale> is in effect, Perl uses the tainting mechanism (see
644L<perlsec>) to mark string results which become locale-dependent, and
645which may be untrustworthy in consequence. Here is a summary of the
b0c42ed9 646tainting behavior of operators and functions which may be affected by
14280422
DD
647the locale:
648
649=over 4
650
651=item B<Comparison operators> (C<lt>, C<le>, C<ge>, C<gt> and C<cmp>):
652
653Scalar true/false (or less/equal/greater) result is never tainted.
654
7b8d334a 655=item B<Case-mapping interpolation> (with C<\l>, C<\L>, C<\u> or C<\U>)
e38874e2
DD
656
657Result string containing interpolated material is tainted if
658C<use locale> is in effect.
659
14280422
DD
660=item B<Matching operator> (C<m//>):
661
662Scalar true/false result never tainted.
663
664Subpatterns, either delivered as an array-context result, or as $1 etc.
665are tainted if C<use locale> is in effect, and the subpattern regular
e38874e2
DD
666expression contains C<\w> (to match an alphanumeric character), C<\W>
667(non-alphanumeric character), C<\s> (white-space character), or C<\S>
668(non white-space character). The matched pattern variable, $&, $`
669(pre-match), $' (post-match), and $+ (last match) are also tainted if
670C<use locale> is in effect and the regular expression contains C<\w>,
671C<\W>, C<\s>, or C<\S>.
14280422
DD
672
673=item B<Substitution operator> (C<s///>):
674
e38874e2
DD
675Has the same behavior as the match operator. Also, the left
676operand of C<=~> becomes tainted when C<use locale> in effect,
677if it is modified as a result of a substitution based on a regular
678expression match involving C<\w>, C<\W>, C<\s>, or C<\S>; or of
7b8d334a 679case-mapping with C<\l>, C<\L>,C<\u> or C<\U>.
14280422
DD
680
681=item B<In-memory formatting function> (sprintf()):
682
683Result is tainted if "use locale" is in effect.
684
685=item B<Output formatting functions> (printf() and write()):
686
687Success/failure result is never tainted.
688
689=item B<Case-mapping functions> (lc(), lcfirst(), uc(), ucfirst()):
690
691Results are tainted if C<use locale> is in effect.
692
693=item B<POSIX locale-dependent functions> (localeconv(), strcoll(),
694strftime(), strxfrm()):
695
696Results are never tainted.
697
698=item B<POSIX character class tests> (isalnum(), isalpha(), isdigit(),
699isgraph(), islower(), isprint(), ispunct(), isspace(), isupper(),
700isxdigit()):
701
702True/false results are never tainted.
703
704=back
705
706Three examples illustrate locale-dependent tainting.
707The first program, which ignores its locale, won't run: a value taken
54310121 708directly from the command line may not be used to name an output file
14280422
DD
709when taint checks are enabled.
710
711 #/usr/local/bin/perl -T
712 # Run with taint checking
713
54310121 714 # Command line sanity check omitted...
14280422
DD
715 $tainted_output_file = shift;
716
717 open(F, ">$tainted_output_file")
718 or warn "Open of $untainted_output_file failed: $!\n";
719
720The program can be made to run by "laundering" the tainted value through
721a regular expression: the second example - which still ignores locale
54310121 722information - runs, creating the file named on its command line
14280422
DD
723if it can.
724
725 #/usr/local/bin/perl -T
726
727 $tainted_output_file = shift;
728 $tainted_output_file =~ m%[\w/]+%;
729 $untainted_output_file = $&;
730
731 open(F, ">$untainted_output_file")
732 or warn "Open of $untainted_output_file failed: $!\n";
733
734Compare this with a very similar program which is locale-aware:
735
736 #/usr/local/bin/perl -T
737
738 $tainted_output_file = shift;
739 use locale;
740 $tainted_output_file =~ m%[\w/]+%;
741 $localized_output_file = $&;
742
743 open(F, ">$localized_output_file")
744 or warn "Open of $localized_output_file failed: $!\n";
745
746This third program fails to run because $& is tainted: it is the result
747of a match involving C<\w> when C<use locale> is in effect.
5f05dabc
PP
748
749=head1 ENVIRONMENT
750
751=over 12
752
753=item PERL_BADLANG
754
14280422 755A string that can suppress Perl's warning about failed locale settings
54310121 756at startup. Failure can occur if the locale support in the operating
7b8d334a 757system is lacking (broken) in some way - or if you mistyped the name of
14280422
DD
758a locale when you set up your environment. If this environment variable
759is absent, or has a value which does not evaluate to integer zero - that
760is "0" or "" - Perl will complain about locale setting failures.
5f05dabc 761
14280422
DD
762B<NOTE>: PERL_BADLANG only gives you a way to hide the warning message.
763The message tells about some problem in your system's locale support,
764and you should investigate what the problem is.
5f05dabc
PP
765
766=back
767
768The following environment variables are not specific to Perl: They are
14280422
DD
769part of the standardized (ISO C, XPG4, POSIX 1.c) setlocale() method
770for controlling an application's opinion on data.
5f05dabc
PP
771
772=over 12
773
774=item LC_ALL
775
776C<LC_ALL> is the "override-all" locale environment variable. If it is
777set, it overrides all the rest of the locale environment variables.
778
779=item LC_CTYPE
780
781In the absence of C<LC_ALL>, C<LC_CTYPE> chooses the character type
782locale. In the absence of both C<LC_ALL> and C<LC_CTYPE>, C<LANG>
783chooses the character type locale.
784
785=item LC_COLLATE
786
14280422
DD
787In the absence of C<LC_ALL>, C<LC_COLLATE> chooses the collation
788(sorting) locale. In the absence of both C<LC_ALL> and C<LC_COLLATE>,
789C<LANG> chooses the collation locale.
5f05dabc
PP
790
791=item LC_MONETARY
792
14280422
DD
793In the absence of C<LC_ALL>, C<LC_MONETARY> chooses the monetary
794formatting locale. In the absence of both C<LC_ALL> and C<LC_MONETARY>,
795C<LANG> chooses the monetary formatting locale.
5f05dabc
PP
796
797=item LC_NUMERIC
798
799In the absence of C<LC_ALL>, C<LC_NUMERIC> chooses the numeric format
800locale. In the absence of both C<LC_ALL> and C<LC_NUMERIC>, C<LANG>
801chooses the numeric format.
802
803=item LC_TIME
804
14280422
DD
805In the absence of C<LC_ALL>, C<LC_TIME> chooses the date and time
806formatting locale. In the absence of both C<LC_ALL> and C<LC_TIME>,
807C<LANG> chooses the date and time formatting locale.
5f05dabc
PP
808
809=item LANG
810
14280422
DD
811C<LANG> is the "catch-all" locale environment variable. If it is set, it
812is used as the last resort after the overall C<LC_ALL> and the
5f05dabc
PP
813category-specific C<LC_...>.
814
815=back
816
817=head1 NOTES
818
819=head2 Backward compatibility
820
b0c42ed9
JH
821Versions of Perl prior to 5.004 B<mostly> ignored locale information,
822generally behaving as if something similar to the C<"C"> locale (see
823L<The setlocale function>) was always in force, even if the program
5f05dabc
PP
824environment suggested otherwise. By default, Perl still behaves this
825way so as to maintain backward compatibility. If you want a Perl
b0c42ed9 826application to pay attention to locale information, you B<must> use
2ae324a7 827the S<C<use locale>> pragma (see L<The use locale Pragma>) to
b0c42ed9
JH
828instruct it to do so.
829
830Versions of Perl from 5.002 to 5.003 did use the C<LC_CTYPE>
831information if that was available, that is, C<\w> did understand what
832are the letters according to the locale environment variables.
833The problem was that the user had no control over the feature:
834if the C library supported locales, Perl used them.
835
836=head2 I18N:Collate obsolete
837
838In versions of Perl prior to 5.004 per-locale collation was possible
839using the C<I18N::Collate> library module. This module is now mildly
840obsolete and should be avoided in new applications. The C<LC_COLLATE>
841functionality is now integrated into the Perl core language: One can
842use locale-specific scalar data completely normally with C<use locale>,
843so there is no longer any need to juggle with the scalar references of
844C<I18N::Collate>.
5f05dabc 845
14280422 846=head2 Sort speed and memory use impacts
5f05dabc
PP
847
848Comparing and sorting by locale is usually slower than the default
14280422
DD
849sorting; slow-downs of two to four times have been observed. It will
850also consume more memory: once a Perl scalar variable has participated
851in any string comparison or sorting operation obeying the locale
852collation rules, it will take 3-15 times more memory than before. (The
853exact multiplier depends on the string's contents, the operating system
854and the locale.) These downsides are dictated more by the operating
855system's implementation of the locale system than by Perl.
5f05dabc 856
e38874e2
DD
857=head2 write() and LC_NUMERIC
858
859Formats are the only part of Perl which unconditionally use information
860from a program's locale; if a program's environment specifies an
861LC_NUMERIC locale, it is always used to specify the decimal point
862character in formatted output. Formatted output cannot be controlled by
863C<use locale> because the pragma is tied to the block structure of the
864program, and, for historical reasons, formats exist outside that block
865structure.
866
5f05dabc
PP
867=head2 Freely available locale definitions
868
869There is a large collection of locale definitions at
14280422
DD
870C<ftp://dkuug.dk/i18n/WG15-collection>. You should be aware that it is
871unsupported, and is not claimed to be fit for any purpose. If your
872system allows the installation of arbitrary locales, you may find the
873definitions useful as they are, or as a basis for the development of
874your own locales.
5f05dabc 875
14280422 876=head2 I18n and l10n
5f05dabc 877
b0c42ed9
JH
878"Internationalization" is often abbreviated as B<i18n> because its first
879and last letters are separated by eighteen others. (You may guess why
880the internalin ... internaliti ... i18n tends to get abbreviated.) In
881the same way, "localization" is often abbreviated to B<l10n>.
14280422
DD
882
883=head2 An imperfect standard
884
885Internationalization, as defined in the C and POSIX standards, can be
886criticized as incomplete, ungainly, and having too large a granularity.
887(Locales apply to a whole process, when it would arguably be more useful
888to have them apply to a single thread, window group, or whatever.) They
889also have a tendency, like standards groups, to divide the world into
890nations, when we all know that the world can equally well be divided
891into bankers, bikers, gamers, and so on. But, for now, it's the only
892standard we've got. This may be construed as a bug.
5f05dabc
PP
893
894=head1 BUGS
895
896=head2 Broken systems
897
2bdf8add
JH
898In certain system environments the operating system's locale support
899is broken and cannot be fixed or used by Perl. Such deficiencies can
900and will result in mysterious hangs and/or Perl core dumps when the
901C<use locale> is in effect. When confronted with such a system,
9607fc9c 902please report in excruciating detail to <F<perlbug@perl.com>>, and
2bdf8add
JH
903complain to your vendor: maybe some bug fixes exist for these problems
904in your operating system. Sometimes such bug fixes are called an
905operating system upgrade.
5f05dabc
PP
906
907=head1 SEE ALSO
908
7b8d334a
GS
909L<POSIX (3)/isalnum>
910
911L<POSIX (3)/isalpha>
912
913L<POSIX (3)/isdigit>
914
915L<POSIX (3)/isgraph>
916
917L<POSIX (3)/islower>
918
919L<POSIX (3)/isprint>,
920
921L<POSIX (3)/ispunct>
922
923L<POSIX (3)/isspace>
924
925L<POSIX (3)/isupper>,
926
927L<POSIX (3)/isxdigit>
928
929L<POSIX (3)/localeconv>
930
931L<POSIX (3)/setlocale>,
932
933L<POSIX (3)/strcoll>
934
935L<POSIX (3)/strftime>
936
937L<POSIX (3)/strtod>,
938
14280422 939L<POSIX (3)/strxfrm>
5f05dabc
PP
940
941=head1 HISTORY
942
b0c42ed9 943Jarkko Hietaniemi's original F<perli18n.pod> heavily hacked by Dominic
14280422 944Dunlop, assisted by the perl5-porters.
5f05dabc 945
3e6e419a 946Last update: Mon Nov 17 22:48:48 EET 1997