# Common tools for test files files to find the locales which exist on the
-# system. Caller should have defined ok() for the unlikely event that setup
-# here fails, and should have verified that this isn't miniperl before calling
+# system. Caller should have verified that this isn't miniperl before calling
# the functions.
# Note that it's okay that some languages have their native names
# 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 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.
+# Functions whose names begin with underscore are internal helper functions
+# for this file, and are not to be used by outside callers.
+
+use Config;
+use strict;
+
+eval { require POSIX; import POSIX 'locale_h'; };
+my $has_locale_h = ! $@;
+
+my @known_categories = ( qw(LC_ALL LC_COLLATE LC_CTYPE LC_MESSAGES LC_MONETARY
+ LC_NUMERIC LC_TIME LC_ADDRESS LC_IDENTIFICATION
+ LC_MEASUREMENT LC_PAPER LC_TELEPHONE));
+my @platform_categories;
+
+# LC_ALL can be -1 on some platforms. And, in fact the implementors could
+# legally use any integer to represent any category. But it makes the most
+# sense for them to have used small integers. Below, we create new locale
+# numbers for ones missing from this machine. We make them very negative,
+# hopefully more negative than anything likely to be a valid category on the
+# platform, but also below is a check to be sure that our guess is valid.
+my $max_bad_category_number = -1000000;
+
+# Initialize this hash so that it looks like e.g.,
+# 6 => 'CTYPE',
+# where 6 is the value of &POSIX::LC_CTYPE
+my %category_name;
+my %category_number;
+if ($has_locale_h) {
+ my $number_for_missing_category = $max_bad_category_number;
+ foreach my $name (@known_categories) {
+ my $number = eval "&POSIX::$name";
+ if ($@) {
+ # Use a negative number (smaller than any legitimate category
+ # number) if the platform doesn't support this category, so we
+ # have an entry for all the ones that might be specified in calls
+ # to us.
+ $number = $number_for_missing_category-- if $@;
+ }
+ elsif ( $number !~ / ^ -? \d+ $ /x
+ || $number <= $max_bad_category_number)
+ {
+ # We think this should be an int. And it has to be larger than
+ # any of our synthetic numbers.
+ die "Unexpected locale category number '$number' for $name"
+ }
+ else {
+ push @platform_categories, $name;
+ }
+
+ $name =~ s/LC_//;
+ $category_name{$number} = "$name";
+ $category_number{$name} = $number;
+ }
+}
+
+sub _my_diag($) {
+ my $message = shift;
+ if (defined &main::diag) {
+ diag($message);
+ }
+ else {
+ local($\, $", $,) = (undef, ' ', '');
+ print STDERR $message, "\n";
+ }
+}
+
+sub _my_fail($) {
+ my $message = shift;
+ if (defined &main::fail) {
+ fail($message);
+ }
+ else {
+ local($\, $", $,) = (undef, ' ', '');
+ print "not ok 0 $message\n";
+ }
+}
+
+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 category numbers
+ # given by the 2nd parameter, which is either a single category or a
+ # reference to a list of categories. The list MUST be sorted so that
+ # CTYPE is first, COLLATE is last unless ALL is present, in which case
+ # that comes after COLLATE. This is because locale.c detects bad locales
+ # only with CTYPE, and COLLATE on some platforms can core dump if it is a
+ # bad locale.
+ #
+ # The 4th parameter is true if to accept locales that aren't apparently
+ # fully compatible with Perl.
+
my $locale = shift;
my $categories = shift;
my $list = shift;
- my $only_plays_well = shift;
+ my $allow_incompatible = shift;
return if ! $locale || grep { $locale eq $_ } @$list;
+ # This is a toy (pig latin) locale that is not fully implemented on some
+ # systems
+ return if $locale =~ / ^ pig $ /ix;
+
+ # As of 6.3, this platform's locale handling is basically broken. khw
+ # filed a bug report (no ticket number was returned), and it is supposedly
+ # going to change in a future release, so the statements here below sunset
+ # for any larger version, at which point this may start failing and have
+ # to be revisited.
+ #
+ # Given a legal individual category, basically whatever you set the locale
+ # to, the return from setlocale() indicates that it has taken effect, even
+ # if it hasn't. However, the return from querying LC_ALL won't reflect
+ # this.
+ if ($Config{osname} =~ /openbsd/i && $locale !~ / ^ (?: C | POSIX ) $/ix) {
+ my ($major, $minor) = $Config{osvers} =~ / ^ ( \d+ ) \. ( \d+ ) /ax;
+ return if ! defined $major || ! defined $minor
+ || $major < 6 || ($major == 6 && $minor <= 3);
+ }
+
$categories = [ $categories ] unless ref $categories;
my $badutf8 = 0;
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
+ $badutf8 = 1 if grep { /Malformed UTF-8/ } @_;
+ $plays_well = 0 if grep {
+ /Locale .* may not work well(?#
+ )|The Perl program will use the expected meanings/i
+ } @_;
};
+ # Incompatible locales aren't warned about unless using locales.
+ use locale;
+
foreach my $category (@$categories) {
+ die "category '$category' must instead be a number"
+ unless $category =~ / ^ -? \d+ $ /x;
+
return unless setlocale($category, $locale);
- return if $only_plays_well && ! $plays_well;
+ last if $badutf8 || ! $plays_well;
}
if ($badutf8) {
- ok(0, "Verify locale name doesn't contain malformed utf8");
+ _my_fail("Verify locale name doesn't contain malformed utf8");
return;
}
- push @$list, $locale;
+ push @$list, $locale if $plays_well || $allow_incompatible;
}
-sub _decode_encodings {
+sub _decode_encodings { # For use only by other functions in this file!
my @enc;
foreach (split(/ /, shift)) {
return @enc;
}
-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.
+sub valid_locale_categories() {
+ # Returns a list of the locale categories (expressed as strings, like
+ # "LC_ALL) known to this program that are available on this platform.
+
+ return @platform_categories;
+}
+
+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 any of the individual categories specified by the
+ # optional parameter is all digits (and an optional leading minus), 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 to this file.
+ #
+ # This optional parameter denotes which POSIX locale categories must be
+ # available on the platform. If any aren't available, this function
+ # returns 0; otherwise it returns 1 and changes the list for the caller so
+ # that any category names are converted into their equivalent numbers, and
+ # sorts it to match the expectations of _trylocale.
+ #
+ # It is acceptable for the second parameter to be just a simple scalar
+ # denoting a single category (either name or number). No conversion into
+ # a number is done in this case.
+
+ 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_locale_h;
+
+ # Done with the global possibilities. Now check if any passed in category
+ # is disabled.
+
+ my $categories_ref = shift;
+ my $return_categories_numbers = 0;
+ my @categories_numbers;
+ my $has_LC_ALL = 0;
+ my $has_LC_COLLATE = 0;
+
+ if (defined $categories_ref) {
+ my @local_categories_copy;
+
+ if (ref $categories_ref) {
+ @local_categories_copy = @$$categories_ref;
+ $return_categories_numbers = 1;
+ }
+ else { # Single category passed in
+ @local_categories_copy = $categories_ref;
+ }
+
+ for my $category_name_or_number (@local_categories_copy) {
+ my $name;
+ my $number;
+ if ($category_name_or_number =~ / ^ -? \d+ $ /x) {
+ $number = $category_name_or_number;
+ die "Invalid locale category number '$number'"
+ unless grep { $number == $_ } keys %category_name;
+ $name = $category_name{$number};
+ }
+ else {
+ $name = $category_name_or_number;
+ $name =~ s/ ^ LC_ //x;
+ foreach my $trial (keys %category_name) {
+ if ($category_name{$trial} eq $name) {
+ $number = $trial;
+ last;
+ }
+ }
+ die "Invalid locale category name '$name'"
+ unless defined $number;
+ }
+
+ return 0 if $number <= $max_bad_category_number
+ || $Config{ccflags} =~ /\bD?NO_LOCALE_$name\b/;
+
+ eval "defined &POSIX::LC_$name";
+ return 0 if $@;
+
+ if ($return_categories_numbers) {
+ if ($name eq 'CTYPE') {
+ unshift @categories_numbers, $number; # Always first
+ }
+ elsif ($name eq 'ALL') {
+ $has_LC_ALL = 1;
+ }
+ elsif ($name eq 'COLLATE') {
+ $has_LC_COLLATE = 1;
+ }
+ else {
+ push @categories_numbers, $number;
+ }
+ }
+ }
+ }
+
+ if ($return_categories_numbers) {
+
+ # COLLATE comes after all other locales except ALL, which comes last
+ if ($has_LC_COLLATE) {
+ push @categories_numbers, $category_number{'COLLATE'};
+ }
+ if ($has_LC_ALL) {
+ push @categories_numbers, $category_number{'ALL'};
+ }
+ $$categories_ref = \@categories_numbers;
+ }
+
+ return 1;
+}
+
+
+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 includes all found locales;
+ # otherwise it is restricted to those locales that play well with Perl, as
+ # far as we can easily determine.
+ #
+ # 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. Each category can be a name (like 'LC_ALL'
+ # or simply 'ALL') or the C enum value for the category.
+
my $categories = shift;
- my $only_plays_well = shift // 0;
+ my $allow_incompatible = shift // 0;
+
+ $categories = [ $categories ] unless ref $categories;
+ return unless locales_enabled(\$categories);
- use Config;;
- my $have_setlocale = $Config{d_setlocale};
+ # Note, the subroutine call above converts the $categories into a form
+ # suitable for _trylocale().
# 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;
-
- # Done this way in case this is 'required' in the caller before seeing if
- # this is miniperl.
- eval { require POSIX; import POSIX 'locale_h'; };
- unless (defined &POSIX::LC_CTYPE) {
- return;
- }
-
- _trylocale("C", $categories, \@Locale, $only_plays_well);
- _trylocale("POSIX", $categories, \@Locale, $only_plays_well);
- foreach (0..15) {
- _trylocale("ISO8859-$_", $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, $only_plays_well);
+ return if ($^O =~ /^uwin/);
+
+ my @Locale;
+ _trylocale("C", $categories, \@Locale, $allow_incompatible);
+ _trylocale("POSIX", $categories, \@Locale, $allow_incompatible);
+ _trylocale("C.UTF-8", $categories, \@Locale, $allow_incompatible);
+ foreach (1..16) {
+ _trylocale("ISO8859-$_", $categories, \@Locale, $allow_incompatible);
+ _trylocale("iso8859$_", $categories, \@Locale, $allow_incompatible);
+ _trylocale("iso8859-$_", $categories, \@Locale, $allow_incompatible);
+ _trylocale("iso_8859_$_", $categories, \@Locale, $allow_incompatible);
+ _trylocale("isolatin$_", $categories, \@Locale, $allow_incompatible);
+ _trylocale("isolatin-$_", $categories, \@Locale, $allow_incompatible);
+ _trylocale("iso_latin_$_", $categories, \@Locale, $allow_incompatible);
}
# Sanitize the environment so that we can run the external 'locale'
delete local @ENV{qw(IFS CDPATH ENV BASH_ENV)};
if (-x "/usr/bin/locale"
- && open(LOCALES, "/usr/bin/locale -a 2>/dev/null|"))
+ && open(LOCALES, '-|', "/usr/bin/locale -a 2>/dev/null"))
{
while (<LOCALES>) {
# It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
# locales will cause all IO hadles to default to (assume) utf8
next unless utf8::valid($_);
chomp;
- _trylocale($_, $categories, \@Locale, $only_plays_well);
+ _trylocale($_, $categories, \@Locale, $allow_incompatible);
}
close(LOCALES);
} elsif ($^O eq 'VMS'
opendir(LOCALES, "SYS\$I18N_LOCALE:");
while ($_ = readdir(LOCALES)) {
chomp;
- _trylocale($_, $categories, \@Locale, $only_plays_well);
+ _trylocale($_, $categories, \@Locale, $allow_incompatible);
}
close(LOCALES);
} elsif (($^O eq 'openbsd' || $^O eq 'bitrig' ) && -e '/usr/share/locale') {
- # OpenBSD doesn't have a locale executable, so reading /usr/share/locale
- # is much easier and faster than the last resort method.
+ # OpenBSD doesn't have a locale executable, so reading
+ # /usr/share/locale is much easier and faster than the last resort
+ # method.
opendir(LOCALES, '/usr/share/locale');
while ($_ = readdir(LOCALES)) {
chomp;
- _trylocale($_, $categories, \@Locale, $only_plays_well);
+ _trylocale($_, $categories, \@Locale, $allow_incompatible);
}
close(LOCALES);
} else { # Final fallback. Try our list of locales hard-coded here
# This is going to be slow.
my @Data;
- # Locales whose name differs if the utf8 bit is on are stored in these two
- # files with appropriate encodings.
- if ($^H & 0x08 || (${^OPEN} || "") =~ /:utf8/) {
- @Data = do "lib/locale/utf8";
- } else {
- @Data = do "lib/locale/latin1";
+ # Locales whose name differs if the utf8 bit is on are stored in these
+ # two files with appropriate encodings.
+ my $data_file = ($^H & 0x08 || (${^OPEN} || "") =~ /:utf8/)
+ ? _source_location() . "/lib/locale/utf8"
+ : _source_location() . "/lib/locale/latin1";
+ if (-e $data_file) {
+ @Data = do $data_file;
+ }
+ else {
+ _my_diag(__FILE__ . ":" . __LINE__ . ": '$data_file' doesn't exist");
}
# The rest of the locales are in this file.
foreach my $line (@Data) {
my ($locale_name, $language_codes, $country_codes, $encodings) =
split /:/, $line;
+ _my_diag(__FILE__ . ":" . __LINE__ . ": Unexpected syntax in '$line'")
+ unless defined $locale_name;
my @enc = _decode_encodings($encodings);
foreach my $loc (split(/ /, $locale_name)) {
- _trylocale($loc, $categories, \@Locale, $only_plays_well);
+ _trylocale($loc, $categories, \@Locale, $allow_incompatible);
foreach my $enc (@enc) {
_trylocale("$loc.$enc", $categories, \@Locale,
- $only_plays_well);
+ $allow_incompatible);
}
$loc = lc $loc;
foreach my $enc (@enc) {
_trylocale("$loc.$enc", $categories, \@Locale,
- $only_plays_well);
+ $allow_incompatible);
}
}
foreach my $lang (split(/ /, $language_codes)) {
- _trylocale($lang, $categories, \@Locale, $only_plays_well);
+ _trylocale($lang, $categories, \@Locale, $allow_incompatible);
foreach my $country (split(/ /, $country_codes)) {
my $lc = "${lang}_${country}";
- _trylocale($lc, $categories, \@Locale, $only_plays_well);
+ _trylocale($lc, $categories, \@Locale, $allow_incompatible);
foreach my $enc (@enc) {
_trylocale("$lc.$enc", $categories, \@Locale,
- $only_plays_well);
+ $allow_incompatible);
}
my $lC = "${lang}_\U${country}";
- _trylocale($lC, $categories, \@Locale, $only_plays_well);
+ _trylocale($lC, $categories, \@Locale, $allow_incompatible);
foreach my $enc (@enc) {
_trylocale("$lC.$enc", $categories, \@Locale,
- $only_plays_well);
+ $allow_incompatible);
}
}
}
# On z/OS, even locales marked as UTF-8 aren't.
return 0 if ord "A" != 65;
- eval { require POSIX; import POSIX 'locale_h'; };
- return 0 if ! defined &POSIX::LC_CTYPE;
+ return 0 unless locales_enabled('LC_CTYPE');
my $locale = shift;
return $ret;
}
-sub find_utf8_ctype_locale (;$) { # Return the name of a locale that core Perl
- # thinks is a UTF-8 LC_CTYPE locale.
+sub find_utf8_ctype_locales (;$) { # Return the names of the locales that core
+ # Perl thinks are UTF-8 LC_CTYPE locales.
# Optional parameter is a reference to a
# list of locales to try; if omitted, this
# tries all locales it can find on the
# platform
+ return unless locales_enabled('LC_CTYPE');
+
my $locales_ref = shift;
+ my @return;
if (! defined $locales_ref) {
- eval { require POSIX; import POSIX 'locale_h'; };
- return if ! defined &POSIX::LC_CTYPE;
- my @locales = find_locales(&POSIX::LC_CTYPE(),
- 1 # Reject iffy locales.
- );
+ my @locales = find_locales(&POSIX::LC_CTYPE());
$locales_ref = \@locales;
}
foreach my $locale (@$locales_ref) {
- return $locale if is_locale_utf8($locale);
+ push @return, $locale if is_locale_utf8($locale);
+ }
+
+ return @return;
+}
+
+
+sub find_utf8_ctype_locale (;$) { # Return the name of a locale that core Perl
+ # thinks is a UTF-8 LC_CTYPE non-turkic
+ # 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 $try_locales_ref = shift;
+
+ my @utf8_locales = find_utf8_ctype_locales($try_locales_ref);
+ my @turkic_locales = find_utf8_turkic_locales($try_locales_ref);
+
+ my %seen_turkic;
+
+ # Create undef elements in the hash for turkic locales
+ @seen_turkic{@turkic_locales} = ();
+
+ foreach my $locale (@utf8_locales) {
+ return $locale unless exists $seen_turkic{$locale};
}
return;
}
+sub find_utf8_turkic_locales (;$) {
+
+ # Return the name of all the locales that core Perl thinks are UTF-8
+ # Turkic LC_CTYPE. 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 @return;
+
+ my $save_locale = setlocale(&POSIX::LC_CTYPE());
+ foreach my $locale (find_utf8_ctype_locales(shift)) {
+ use locale;
+ setlocale(&POSIX::LC_CTYPE(), $locale);
+ push @return, $locale if uc('i') eq "\x{130}";
+ }
+ setlocale(&POSIX::LC_CTYPE(), $save_locale);
+
+ return @return;
+}
+
+sub find_utf8_turkic_locale (;$) {
+ my @turkics = find_utf8_turkic_locales(shift);
+
+ return unless @turkics;
+ return $turkics[0]
+}
+
+
+# returns full path to the directory containing the current source
+# file, inspired by mauke's Dir::Self
+sub _source_location {
+ require File::Spec;
+
+ my $caller_filename = (caller)[1];
+
+ my $loc = File::Spec->rel2abs(
+ File::Spec->catpath(
+ (File::Spec->splitpath($caller_filename))[0, 1], ''
+ )
+ );
+
+ return ($loc =~ /^(.*)$/)[0]; # untaint
+}
+
1
# Format of data is: locale_name, language_codes, country_codes, encodings
https://perl5.git.perl.org / perl5.git / blobdiff