# anyway later during the scanning process (and besides, some clueless
# vendor might have them capitalized erroneously anyway).
-sub _trylocale { # Adds the locale given by the first parameter to the list
- # given by the 2nd iff the platform supports the locale,
- # and it is not already on the list
+# Functions whose names begin with underscore are internal helper functions
+# for this file, and are not to be used by outside callers.
+
+eval { require POSIX; import POSIX 'locale_h'; };
+my $has_posix_locales = defined &POSIX::LC_CTYPE;
+
+sub _trylocale ($$$$) { # For use only by other functions in this file!
+
+ # Adds the locale given by the first parameter to the list given by the
+ # 3rd iff the platform supports the locale in each of the categories given
+ # by the 2nd parameter, which is either a single category or a reference
+ # to a list of categories The 4th parameter is true if to reject locales
+ # that aren't apparently fully compatible with Perl.
+
my $locale = shift;
+ my $categories = shift;
my $list = shift;
- return if grep { $locale eq $_ } @$list;
- return unless setlocale(&POSIX::LC_ALL, $locale);
- my $badutf8;
- {
- local $SIG{__WARN__} = sub {
- $badutf8 = $_[0] =~ /Malformed UTF-8/;
- };
+ my $only_plays_well = shift;
+
+ return if ! $locale || grep { $locale eq $_ } @$list;
+
+ $categories = [ $categories ] unless ref $categories;
+
+ my $badutf8 = 0;
+ my $plays_well = 1;
+
+ use warnings 'locale';
+
+ local $SIG{__WARN__} = sub {
+ $badutf8 = 1 if $_[0] =~ /Malformed UTF-8/;
+ $plays_well = 0 if $_[0] =~ /Locale .* may not work well/i
+ };
+
+ foreach my $category (@$categories) {
+ return unless setlocale($category, $locale);
+ return if $only_plays_well && ! $plays_well;
}
if ($badutf8) {
push @$list, $locale;
}
-sub _decode_encodings {
+sub _decode_encodings { # For use only by other functions in this file!
my @enc;
foreach (split(/ /, shift)) {
push @enc, "$_.65001"; # Windows UTF-8
push @enc, "$_.ACP"; # Windows ANSI code page
push @enc, "$_.OCP"; # Windows OEM code page
+ push @enc, "$_.1252"; # Windows
}
}
if ($^O eq 'os390') {
return @enc;
}
-sub find_locales { # Returns an array of all the locales we found on the
- # system
+# Initialize this hash so that it looks like e.g.,
+# 6 => 'CTYPE',
+# where 6 is the value of &POSIX::LC_CTYPE
+my %category_name;
+eval { require POSIX; import POSIX 'locale_h'; };
+unless ($@) {
+ my $number_for_missing_category = 0;
+ foreach my $name (qw(ALL COLLATE CTYPE MESSAGES MONETARY NUMERIC TIME)) {
+ my $number = eval "&POSIX::LC_$name";
+
+ # Use a negative number if the platform doesn't support this category,
+ # so we have an entry for all ones that might be specified in calls to
+ # us.
+ $number = --$number_for_missing_category if $@;
+
+ $category_name{$number} = "$name";
+ }
+}
+
+sub locales_enabled(;$) {
+ # Returns 0 if no locale handling is available on this platform; otherwise
+ # 1.
+ #
+ # The optional parameter is a reference to a list of individual POSIX
+ # locale categories. If present, this function also returns 0 if any of
+ # them are individually not available on this platform; otherwise 1.
+ # Actually, it is acceptable for the list to be just a simple scalar
+ # denoting a single category.
+ #
+ # If any of the individual categories specified by the optional parameter
+ # is all digits, it is taken to be the C enum for the category (e.g.,
+ # &POSIX::LC_CTYPE). Otherwise it should be a string name of the
+ # category, like 'LC_TIME'. The initial 'LC_' is optional. It is a fatal
+ # error to call this with something that isn't a known category
+
+ use Config;
+
+ return 0 unless $Config{d_setlocale}
+ # I (khw) cargo-culted the '?' in the pattern on the
+ # next line.
+ && $Config{ccflags} !~ /\bD?NO_LOCALE\b/
+ && $has_posix_locales;
+
+ # Done with the global possibilities. Now check if any passed in category
+ # is disabled.
+ my $categories_ref = shift;
+ if (defined $categories_ref) {
+ $categories_ref = [ $categories_ref ] if ! ref $categories_ref;
+ my @local_categories_copy = @$categories_ref;
+ for my $category (@local_categories_copy) {
+ if ($category =~ / ^ -? \d+ $ /x) {
+ die "Invalid locale category number '$category'"
+ unless grep { $category == $_ } keys %category_name;
+ $category = $category_name{$category};
+ }
+ else {
+ $category =~ s/ ^ LC_ //x;
+ die "Invalid locale category name '$category'"
+ unless grep { $category eq $_ } values %category_name;
+ }
+
+ return 0 if $Config{ccflags} =~ /\bD?NO_LOCALE_$category\b/;
+ }
+ }
+
+ return 1;
+}
+
- use Config;;
- my $have_setlocale = $Config{d_setlocale};
+sub find_locales ($;$) { # Returns an array of all the locales we found on the
+ # system. If the optional 2nd parameter is
+ # non-zero, the list is restricted to those locales
+ # that play well with Perl.
+ # The first parameter is either a single locale
+ # category or a reference to a list of categories to
+ # find valid locales for it (or in the case of
+ # multiple) for all of them.
+ my $categories = shift;
+ my $only_plays_well = shift // 0;
+
+ return unless locales_enabled($categories);
# Visual C's CRT goes silly on strings of the form "en_US.ISO8859-1"
# and mingw32 uses said silly CRT
# so re-enable the tests for Windows XP onwards.
my $winxp = ($^O eq 'MSWin32' && defined &Win32::GetOSVersion &&
join('.', (Win32::GetOSVersion())[1..2]) >= 5.1);
- $have_setlocale = 0 if ((($^O eq 'MSWin32' && !$winxp) || $^O eq 'NetWare') &&
- $Config{cc} =~ /^(cl|gcc|g\+\+|ici)/i);
+ return if ((($^O eq 'MSWin32' && !$winxp) || $^O eq 'NetWare')
+ && $Config{cc} =~ /^(cl|gcc|g\+\+|ici)/i);
# UWIN seems to loop after taint tests, just skip for now
- $have_setlocale = 0 if ($^O =~ /^uwin/);
-
- return unless $have_setlocale;
+ return if ($^O =~ /^uwin/);
# Done this way in case this is 'required' in the caller before seeing if
# this is miniperl.
- require POSIX; import POSIX 'locale_h';
+ return unless $has_posix_locales;
- _trylocale("C", \@Locale);
- _trylocale("POSIX", \@Locale);
+ _trylocale("C", $categories, \@Locale, $only_plays_well);
+ _trylocale("POSIX", $categories, \@Locale, $only_plays_well);
foreach (0..15) {
- _trylocale("ISO8859-$_", \@Locale);
- _trylocale("iso8859$_", \@Locale);
- _trylocale("iso8859-$_", \@Locale);
- _trylocale("iso_8859_$_", \@Locale);
- _trylocale("isolatin$_", \@Locale);
- _trylocale("isolatin-$_", \@Locale);
- _trylocale("iso_latin_$_", \@Locale);
+ _trylocale("ISO8859-$_", $categories, \@Locale, $only_plays_well);
+ _trylocale("iso8859$_", $categories, \@Locale, $only_plays_well);
+ _trylocale("iso8859-$_", $categories, \@Locale, $only_plays_well);
+ _trylocale("iso_8859_$_", $categories, \@Locale, $only_plays_well);
+ _trylocale("isolatin$_", $categories, \@Locale, $only_plays_well);
+ _trylocale("isolatin-$_", $categories, \@Locale, $only_plays_well);
+ _trylocale("iso_latin_$_", $categories, \@Locale, $only_plays_well);
}
# Sanitize the environment so that we can run the external 'locale'
# program without the taint mode getting grumpy.
# $ENV{PATH} is special in VMS.
- delete $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};
+ delete local $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};
# Other subversive stuff.
- delete @ENV{qw(IFS CDPATH ENV BASH_ENV)};
+ delete local @ENV{qw(IFS CDPATH ENV BASH_ENV)};
if (-x "/usr/bin/locale"
&& open(LOCALES, "/usr/bin/locale -a 2>/dev/null|"))
# locales will cause all IO hadles to default to (assume) utf8
next unless utf8::valid($_);
chomp;
- _trylocale($_, \@Locale);
+ _trylocale($_, $categories, \@Locale, $only_plays_well);
}
close(LOCALES);
} elsif ($^O eq 'VMS'
opendir(LOCALES, "SYS\$I18N_LOCALE:");
while ($_ = readdir(LOCALES)) {
chomp;
- _trylocale($_, \@Locale);
+ _trylocale($_, $categories, \@Locale, $only_plays_well);
}
close(LOCALES);
} elsif (($^O eq 'openbsd' || $^O eq 'bitrig' ) && -e '/usr/share/locale') {
opendir(LOCALES, '/usr/share/locale');
while ($_ = readdir(LOCALES)) {
chomp;
- _trylocale($_, \@Locale);
+ _trylocale($_, $categories, \@Locale, $only_plays_well);
}
close(LOCALES);
} else { # Final fallback. Try our list of locales hard-coded here
# This is going to be slow.
my @Data;
-
# Locales whose name differs if the utf8 bit is on are stored in these two
# files with appropriate encodings.
if ($^H & 0x08 || (${^OPEN} || "") =~ /:utf8/) {
split /:/, $line;
my @enc = _decode_encodings($encodings);
foreach my $loc (split(/ /, $locale_name)) {
- _trylocale($loc, \@Locale);
+ _trylocale($loc, $categories, \@Locale, $only_plays_well);
foreach my $enc (@enc) {
- _trylocale("$loc.$enc", \@Locale);
+ _trylocale("$loc.$enc", $categories, \@Locale,
+ $only_plays_well);
}
$loc = lc $loc;
foreach my $enc (@enc) {
- _trylocale("$loc.$enc", \@Locale);
+ _trylocale("$loc.$enc", $categories, \@Locale,
+ $only_plays_well);
}
}
foreach my $lang (split(/ /, $language_codes)) {
- _trylocale($lang, \@Locale);
+ _trylocale($lang, $categories, \@Locale, $only_plays_well);
foreach my $country (split(/ /, $country_codes)) {
my $lc = "${lang}_${country}";
- _trylocale($lc, \@Locale);
+ _trylocale($lc, $categories, \@Locale, $only_plays_well);
foreach my $enc (@enc) {
- _trylocale("$lc.$enc", \@Locale);
+ _trylocale("$lc.$enc", $categories, \@Locale,
+ $only_plays_well);
}
my $lC = "${lang}_\U${country}";
- _trylocale($lC, \@Locale);
+ _trylocale($lC, $categories, \@Locale, $only_plays_well);
foreach my $enc (@enc) {
- _trylocale("$lC.$enc", \@Locale);
+ _trylocale("$lC.$enc", $categories, \@Locale,
+ $only_plays_well);
}
}
}
@Locale = sort @Locale;
return @Locale;
+}
+
+sub is_locale_utf8 ($) { # Return a boolean as to if core Perl thinks the input
+ # is a UTF-8 locale
+
+ # On z/OS, even locales marked as UTF-8 aren't.
+ return 0 if ord "A" != 65;
+
+ return 0 if ! $has_posix_locales;
+ return 0 if ! locales_enabled('LC_CTYPE');
+
+ my $locale = shift;
+
+ use locale;
+ no warnings 'locale'; # We may be trying out a weird locale
+
+ my $save_locale = setlocale(&POSIX::LC_CTYPE());
+ if (! $save_locale) {
+ ok(0, "Verify could save previous locale");
+ return 0;
+ }
+
+ if (! setlocale(&POSIX::LC_CTYPE(), $locale)) {
+ ok(0, "Verify could setlocale to $locale");
+ return 0;
+ }
+ my $ret = 0;
+
+ # Use an op that gives different results for UTF-8 than any other locale.
+ # If a platform has UTF-8 locales, there should be at least one locale on
+ # most platforms with UTF-8 in its name, so if there is a bug in the op
+ # giving a false negative, we should get a failure for those locales as we
+ # go through testing all the locales on the platform.
+ if (CORE::fc(chr utf8::unicode_to_native(0xdf)) ne "ss") {
+ if ($locale =~ /UTF-?8/i) {
+ ok (0, "Verify $locale with UTF-8 in name is a UTF-8 locale");
+ }
+ }
+ else {
+ $ret = 1;
+ }
+
+ die "Couldn't restore locale '$save_locale'"
+ unless setlocale(&POSIX::LC_CTYPE(), $save_locale);
+
+ return $ret;
+}
+
+sub find_utf8_ctype_locale (;$) { # Return the name of a locale that core Perl
+ # thinks is a UTF-8 LC_CTYPE locale.
+ # Optional parameter is a reference to a
+ # list of locales to try; if omitted, this
+ # tries all locales it can find on the
+ # platform
+ my $locales_ref = shift;
+
+ if (! defined $locales_ref) {
+ return if ! $has_posix_locales;
+
+ my @locales = find_locales(&POSIX::LC_CTYPE(),
+ 1 # Reject iffy locales.
+ );
+ $locales_ref = \@locales;
+ }
+
+ foreach my $locale (@$locales_ref) {
+ return $locale if is_locale_utf8($locale);
+ }
+ return;
}
1