1 # Common tools for test files files to find the locales which exist on the
2 # system. Caller should have defined ok() for the unlikely event that setup
3 # here fails, and should have verified that this isn't miniperl before calling
6 # Note that it's okay that some languages have their native names
7 # capitalized here even though that's not "right". They are lowercased
8 # anyway later during the scanning process (and besides, some clueless
9 # vendor might have them capitalized erroneously anyway).
11 # Functions whose names begin with underscore are internal helper functions
12 # for this file, and are not to be used by outside callers.
14 eval { require POSIX; import POSIX 'locale_h'; };
15 my $has_posix_locales = defined &POSIX::LC_CTYPE;
17 sub _trylocale ($$$$) { # For use only by other functions in this file!
19 # Adds the locale given by the first parameter to the list given by the
20 # 3rd iff the platform supports the locale in each of the categories given
21 # by the 2nd parameter, which is either a single category or a reference
22 # to a list of categories The 4th parameter is true if to reject locales
23 # that aren't apparently fully compatible with Perl.
26 my $categories = shift;
28 my $only_plays_well = shift;
30 return if ! $locale || grep { $locale eq $_ } @$list;
32 $categories = [ $categories ] unless ref $categories;
37 use warnings 'locale';
39 local $SIG{__WARN__} = sub {
40 $badutf8 = 1 if $_[0] =~ /Malformed UTF-8/;
41 $plays_well = 0 if $_[0] =~ /Locale .* may not work well/i
44 foreach my $category (@$categories) {
45 return unless setlocale($category, $locale);
46 return if $only_plays_well && ! $plays_well;
50 ok(0, "Verify locale name doesn't contain malformed utf8");
56 sub _decode_encodings { # For use only by other functions in this file!
59 foreach (split(/ /, shift)) {
61 push @enc, "ISO8859-$1";
62 push @enc, "iso8859$1"; # HP
64 push @enc, "roman8"; # HP
67 push @enc, "$_.UTF-8";
68 push @enc, "$_.65001"; # Windows UTF-8
69 push @enc, "$_.ACP"; # Windows ANSI code page
70 push @enc, "$_.OCP"; # Windows OEM code page
71 push @enc, "$_.1252"; # Windows
75 push @enc, qw(IBM-037 IBM-819 IBM-1047);
78 push @enc, "65001"; # Windows UTF-8
83 # Initialize this hash so that it looks like e.g.,
85 # where 6 is the value of &POSIX::LC_CTYPE
87 eval { require POSIX; import POSIX 'locale_h'; };
89 my $number_for_missing_category = 0;
90 foreach my $name (qw(ALL COLLATE CTYPE MESSAGES MONETARY NUMERIC TIME)) {
91 my $number = eval "&POSIX::LC_$name";
93 # Use a negative number if the platform doesn't support this category,
94 # so we have an entry for all ones that might be specified in calls to
96 $number = --$number_for_missing_category if $@;
98 $category_name{$number} = "$name";
102 sub locales_enabled(;$) {
103 # Returns 0 if no locale handling is available on this platform; otherwise
106 # The optional parameter is a reference to a list of individual POSIX
107 # locale categories. If present, this function also returns 0 if any of
108 # them are individually not available on this platform; otherwise 1.
109 # Actually, it is acceptable for the list to be just a simple scalar
110 # denoting a single category.
112 # If any of the individual categories specified by the optional parameter
113 # is all digits, it is taken to be the C enum for the category (e.g.,
114 # &POSIX::LC_CTYPE). Otherwise it should be a string name of the
115 # category, like 'LC_TIME'. The initial 'LC_' is optional. It is a fatal
116 # error to call this with something that isn't a known category
120 return 0 unless $Config{d_setlocale}
121 # I (khw) cargo-culted the '?' in the pattern on the
123 && $Config{ccflags} !~ /\bD?NO_LOCALE\b/
124 && $has_posix_locales;
126 # Done with the global possibilities. Now check if any passed in category
128 my $categories_ref = shift;
129 if (defined $categories_ref) {
130 $categories_ref = [ $categories_ref ] if ! ref $categories_ref;
131 my @local_categories_copy = @$categories_ref;
132 for my $category (@local_categories_copy) {
133 if ($category =~ / ^ -? \d+ $ /x) {
134 die "Invalid locale category number '$category'"
135 unless grep { $category == $_ } keys %category_name;
136 $category = $category_name{$category};
139 $category =~ s/ ^ LC_ //x;
140 die "Invalid locale category name '$category'"
141 unless grep { $category eq $_ } values %category_name;
144 return 0 if $Config{ccflags} =~ /\bD?NO_LOCALE_$category\b/;
152 sub find_locales ($;$) { # Returns an array of all the locales we found on the
153 # system. If the optional 2nd parameter is
154 # non-zero, the list is restricted to those locales
155 # that play well with Perl.
156 # The first parameter is either a single locale
157 # category or a reference to a list of categories to
158 # find valid locales for it (or in the case of
159 # multiple) for all of them.
160 my $categories = shift;
161 my $only_plays_well = shift // 0;
163 return unless locales_enabled($categories);
165 # Visual C's CRT goes silly on strings of the form "en_US.ISO8859-1"
166 # and mingw32 uses said silly CRT
167 # This doesn't seem to be an issue any more, at least on Windows XP,
168 # so re-enable the tests for Windows XP onwards.
169 my $winxp = ($^O eq 'MSWin32' && defined &Win32::GetOSVersion &&
170 join('.', (Win32::GetOSVersion())[1..2]) >= 5.1);
171 return if ((($^O eq 'MSWin32' && !$winxp) || $^O eq 'NetWare')
172 && $Config{cc} =~ /^(cl|gcc|g\+\+|ici)/i);
174 # UWIN seems to loop after taint tests, just skip for now
175 return if ($^O =~ /^uwin/);
177 # Done this way in case this is 'required' in the caller before seeing if
179 return unless $has_posix_locales;
181 _trylocale("C", $categories, \@Locale, $only_plays_well);
182 _trylocale("POSIX", $categories, \@Locale, $only_plays_well);
184 _trylocale("ISO8859-$_", $categories, \@Locale, $only_plays_well);
185 _trylocale("iso8859$_", $categories, \@Locale, $only_plays_well);
186 _trylocale("iso8859-$_", $categories, \@Locale, $only_plays_well);
187 _trylocale("iso_8859_$_", $categories, \@Locale, $only_plays_well);
188 _trylocale("isolatin$_", $categories, \@Locale, $only_plays_well);
189 _trylocale("isolatin-$_", $categories, \@Locale, $only_plays_well);
190 _trylocale("iso_latin_$_", $categories, \@Locale, $only_plays_well);
193 # Sanitize the environment so that we can run the external 'locale'
194 # program without the taint mode getting grumpy.
196 # $ENV{PATH} is special in VMS.
197 delete local $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};
199 # Other subversive stuff.
200 delete local @ENV{qw(IFS CDPATH ENV BASH_ENV)};
202 if (-x "/usr/bin/locale"
203 && open(LOCALES, "/usr/bin/locale -a 2>/dev/null|"))
206 # It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
207 # ain't great when we're running this testPERL_UNICODE= so that utf8
208 # locales will cause all IO hadles to default to (assume) utf8
209 next unless utf8::valid($_);
211 _trylocale($_, $categories, \@Locale, $only_plays_well);
214 } elsif ($^O eq 'VMS'
215 && defined($ENV{'SYS$I18N_LOCALE'})
216 && -d 'SYS$I18N_LOCALE')
218 # The SYS$I18N_LOCALE logical name search list was not present on
219 # VAX VMS V5.5-12, but was on AXP && VAX VMS V6.2 as well as later versions.
220 opendir(LOCALES, "SYS\$I18N_LOCALE:");
221 while ($_ = readdir(LOCALES)) {
223 _trylocale($_, $categories, \@Locale, $only_plays_well);
226 } elsif (($^O eq 'openbsd' || $^O eq 'bitrig' ) && -e '/usr/share/locale') {
228 # OpenBSD doesn't have a locale executable, so reading /usr/share/locale
229 # is much easier and faster than the last resort method.
231 opendir(LOCALES, '/usr/share/locale');
232 while ($_ = readdir(LOCALES)) {
234 _trylocale($_, $categories, \@Locale, $only_plays_well);
237 } else { # Final fallback. Try our list of locales hard-coded here
239 # This is going to be slow.
242 # Locales whose name differs if the utf8 bit is on are stored in these two
243 # files with appropriate encodings.
244 if ($^H & 0x08 || (${^OPEN} || "") =~ /:utf8/) {
245 @Data = do "lib/locale/utf8";
247 @Data = do "lib/locale/latin1";
250 # The rest of the locales are in this file.
253 foreach my $line (@Data) {
254 my ($locale_name, $language_codes, $country_codes, $encodings) =
256 my @enc = _decode_encodings($encodings);
257 foreach my $loc (split(/ /, $locale_name)) {
258 _trylocale($loc, $categories, \@Locale, $only_plays_well);
259 foreach my $enc (@enc) {
260 _trylocale("$loc.$enc", $categories, \@Locale,
264 foreach my $enc (@enc) {
265 _trylocale("$loc.$enc", $categories, \@Locale,
269 foreach my $lang (split(/ /, $language_codes)) {
270 _trylocale($lang, $categories, \@Locale, $only_plays_well);
271 foreach my $country (split(/ /, $country_codes)) {
272 my $lc = "${lang}_${country}";
273 _trylocale($lc, $categories, \@Locale, $only_plays_well);
274 foreach my $enc (@enc) {
275 _trylocale("$lc.$enc", $categories, \@Locale,
278 my $lC = "${lang}_\U${country}";
279 _trylocale($lC, $categories, \@Locale, $only_plays_well);
280 foreach my $enc (@enc) {
281 _trylocale("$lC.$enc", $categories, \@Locale,
289 @Locale = sort @Locale;
294 sub is_locale_utf8 ($) { # Return a boolean as to if core Perl thinks the input
297 # On z/OS, even locales marked as UTF-8 aren't.
298 return 0 if ord "A" != 65;
300 return 0 if ! $has_posix_locales;
301 return 0 if ! locales_enabled('LC_CTYPE');
306 no warnings 'locale'; # We may be trying out a weird locale
308 my $save_locale = setlocale(&POSIX::LC_CTYPE());
309 if (! $save_locale) {
310 ok(0, "Verify could save previous locale");
314 if (! setlocale(&POSIX::LC_CTYPE(), $locale)) {
315 ok(0, "Verify could setlocale to $locale");
321 # Use an op that gives different results for UTF-8 than any other locale.
322 # If a platform has UTF-8 locales, there should be at least one locale on
323 # most platforms with UTF-8 in its name, so if there is a bug in the op
324 # giving a false negative, we should get a failure for those locales as we
325 # go through testing all the locales on the platform.
326 if (CORE::fc(chr utf8::unicode_to_native(0xdf)) ne "ss") {
327 if ($locale =~ /UTF-?8/i) {
328 ok (0, "Verify $locale with UTF-8 in name is a UTF-8 locale");
335 die "Couldn't restore locale '$save_locale'"
336 unless setlocale(&POSIX::LC_CTYPE(), $save_locale);
341 sub find_utf8_ctype_locale (;$) { # Return the name of a locale that core Perl
342 # thinks is a UTF-8 LC_CTYPE locale.
343 # Optional parameter is a reference to a
344 # list of locales to try; if omitted, this
345 # tries all locales it can find on the
347 my $locales_ref = shift;
349 if (! defined $locales_ref) {
350 return if ! $has_posix_locales;
352 my @locales = find_locales(&POSIX::LC_CTYPE(),
353 1 # Reject iffy locales.
355 $locales_ref = \@locales;
358 foreach my $locale (@$locales_ref) {
359 return $locale if is_locale_utf8($locale);
367 # Format of data is: locale_name, language_codes, country_codes, encodings
370 Arabic:ar:dz eg sa:6 arabic8
371 Brezhoneg Breton:br:fr:1 15
372 Bulgarski Bulgarian:bg:bg:5
373 Chinese:zh:cn tw:cn.EUC eucCN eucTW euc.CN euc.TW Big5 GB2312 tw.EUC
374 Hrvatski Croatian:hr:hr:2
375 Cymraeg Welsh:cy:cy:1 14 15
377 Dansk Danish:da:dk:1 15
378 Nederlands Dutch:nl:be nl:1 15
379 English American British:en:au ca gb ie nz us uk zw:1 15 cp850
381 Eesti Estonian:et:ee:4 6 13
382 Suomi Finnish:fi:fi:1 15
384 Deutsch German:de:at be ch de lu:1 15
385 Euskaraz Basque:eu:es fr:1 15
386 Galego Galician:gl:es:1 15
387 Ellada Greek:el:gr:7 g8
389 Greenlandic:kl:gl:4 6
390 Hebrew:iw:il:8 hebrew8
392 Indonesian:id:id:1 15
393 Gaeilge Irish:ga:IE:1 14 15
394 Italiano Italian:it:ch it:1 15
395 Nihongo Japanese:ja:jp:euc eucJP jp.EUC sjis
397 Latine Latin:la:va:1 15
399 Lithuanian:lt:lt:4 6 13
400 Macedonian:mk:mk:1 15
403 Norsk Norwegian:no no\@nynorsk nb nn:no:1 15
405 Polski Polish:pl:pl:2
407 Russki Russian:ru:ru su ua:5 koi8 koi8r KOI8-R koi8u cp1251 cp866
408 Serbski Serbian:sr:yu:5
410 Slovene Slovenian:sl:si:2
411 Sqhip Albanian:sq:sq:1 15
412 Svenska Swedish:sv:fi se:1 15
414 Turkish:tr:tr:9 turkish8