-# Common tools for test files files to find the locales which exist on the
+# Common tools for test files to find the locales which exist on the
# system. Caller should have verified that this isn't miniperl before calling
# the functions.
# 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;
+use warnings;
+use feature 'state';
eval { require POSIX; import POSIX 'locale_h'; };
my $has_locale_h = ! $@;
+my @known_categories = ( qw(LC_ALL LC_COLLATE LC_CTYPE LC_MESSAGES LC_MONETARY
+ LC_NUMERIC LC_TIME LC_ADDRESS LC_IDENTIFICATION
+ LC_MEASUREMENT LC_PAPER LC_TELEPHONE LC_SYNTAX
+ LC_TOD));
+my @platform_categories;
+
# LC_ALL can be -1 on some platforms. And, in fact the implementors could
# legally use any integer to represent any category. But it makes the most
# sense for them to have used small integers. Below, we create new locale
my %category_number;
if ($has_locale_h) {
my $number_for_missing_category = $max_bad_category_number;
- foreach my $name (qw(ALL COLLATE CTYPE MESSAGES MONETARY NUMERIC TIME)) {
- my $number = eval "&POSIX::LC_$name";
-
+ foreach my $name (@known_categories) {
+ my $number = eval "&POSIX::$name";
if ($@) {
# Use a negative number (smaller than any legitimate category
# number) if the platform doesn't support this category, so we
{
# We think this should be an int. And it has to be larger than
# any of our synthetic numbers.
- die "Unexpected locale category number '$number' for LC_$name"
+ die "Unexpected locale category number '$number' for $name"
+ }
+ else {
+ push @platform_categories, $name;
}
+ $name =~ s/LC_//;
$category_name{$number} = "$name";
$category_number{$name} = $number;
}
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;
+
+ # Certain platforms have a crippled locale system in which setlocale
+ # returns success for just about any possible locale name, but if anything
+ # actually happens as a result of the call, it is that the underlying
+ # locale is set to a system default, likely C or C.UTF-8. We can't test
+ # such systems fully, but we shouldn't disable the user from using
+ # locales, as it may work out for them (or not).
+ return if defined $Config{d_setlocale_accepts_any_locale_name}
+ && $locale !~ / ^ (?: C | POSIX | C\.UTF-8 ) $/ix;
+
$categories = [ $categories ] unless ref $categories;
my $badutf8 = 0;
local $SIG{__WARN__} = sub {
$badutf8 = 1 if grep { /Malformed UTF-8/ } @_;
- $plays_well = 0 if grep { /Locale .* may not work well/i } @_;
+ $plays_well = 0 if grep {
+ /Locale .* may not work well(?#
+ )|The Perl program will use the expected meanings/i
+ } @_;
};
# Incompatible locales aren't warned about unless using locales.
return @enc;
}
+sub valid_locale_categories() {
+ # Returns a list of the locale categories (expressed as strings, like
+ # "LC_ALL) known to this program that are available on this platform.
+
+ return @platform_categories;
+}
+
sub locales_enabled(;$) {
- # Returns 0 if no locale handling is available on this platform; otherwise
- # 1.
+ # If no parameter is specified, the function returns 1 if there is any
+ # "safe" locale handling available to the caller; otherwise 0. Safeness
+ # is defined here as the caller operating in the main thread of a program,
+ # or if threaded locales are safe on the platform and Configured to be
+ # used. This sub is used for testing purposes, and for those, this
+ # definition of safety is sufficient, and necessary to get some tests to
+ # run on certain configurations on certain platforms. But beware that the
+ # main thread can change the locale of any subthreads unless
+ # ${^SAFE_LOCALES} is non-zero.
#
- # 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 the platform.
+ # Use the optional parameter to discover if a particular category or
+ # categories are available on the system. 1 is returned if the global
+ # criteria described in the previous paragraph are true, AND if all the
+ # specified categories are available on the platform and Configured to be
+ # used. Otherwise 0 is returned. The parameter is either a single POSIX
+ # locale category or a reference to a list of them. Each category must be
+ # its name as a string, like 'LC_TIME' (the initial 'LC_' is optional), or
+ # the number this platform uses to signify the category (e.g.,
+ # 'locales_enabled(&POSIX::LC_CTYPE)'
#
- # 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.
+ # When the function returns 1 and a parameter was specified as a list
+ # reference, the reference will be altered on return to point to an
+ # equivalent list such that the categories are numeric instead of strings
+ # and sorted to meet the input 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.
-
- use Config;
+ # It is a fatal error to call this with something that isn't a known
+ # category to this file. If this happens, look first for a typo, and
+ # second if you are using a category unknown to Perl. In the latter case
+ # a bug report should be submitted.
+
+ # khw cargo-culted the '?' in the pattern on the next line.
+ return 0 if $Config{ccflags} =~ /\bD?NO_LOCALE\b/;
+
+ # If we can't load the POSIX XS module, we can't have locales even if they
+ # normally would be available
+ return 0 if ! defined &DynaLoader::boot_DynaLoader;
+
+ # Don't test locales where they aren't safe. On systems with unsafe
+ # threads, for the purposes of testing, we consider the main thread safe,
+ # and all other threads unsafe.
+ if (! ${^SAFE_LOCALES}) {
+ require threads;
+ return 0 if threads->tid() != 0;
+ }
- 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;
+ # If no setlocale, we need the POSIX 2008 alternatives
+ if (! $Config{d_setlocale}) {
+ return 0 if $Config{ccflags} =~ /\bD?NO_POSIX_2008_LOCALE\b/;
+ return 0 unless $Config{d_newlocale};
+ return 0 unless $Config{d_uselocale};
+ return 0 unless $Config{d_duplocale};
+ return 0 unless $Config{d_freelocale};
+ }
# Done with the global possibilities. Now check if any passed in category
# is disabled.
- my $categories_ref = shift;
+ my $categories_ref = $_[0];
my $return_categories_numbers = 0;
my @categories_numbers;
my $has_LC_ALL = 0;
if (defined $categories_ref) {
my @local_categories_copy;
- if (ref $categories_ref) {
- @local_categories_copy = @$$categories_ref;
+ my $reftype = ref $categories_ref;
+ if ($reftype eq 'ARRAY') {
+ @local_categories_copy = @$categories_ref;
$return_categories_numbers = 1;
}
+ elsif ($reftype ne "") {
+ die "Parameter to locales_enabled() must be an ARRAY;"
+ . " instead you used a $reftype";
+ }
else { # Single category passed in
@local_categories_copy = $categories_ref;
}
if ($has_LC_ALL) {
push @categories_numbers, $category_number{'ALL'};
}
- $$categories_ref = \@categories_numbers;
+
+ @$categories_ref = @categories_numbers;
}
return 1;
# 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 $input_categories = shift;
my $allow_incompatible = shift // 0;
- $categories = [ $categories ] unless ref $categories;
- return unless locales_enabled(\$categories);
+ my @categories = (ref $input_categories) ? $input_categories->@* : $input_categories;
+ return unless locales_enabled(\@categories);
# Note, the subroutine call above converts the $categories into a form
# suitable for _trylocale().
return if ($^O =~ /^uwin/);
my @Locale;
- _trylocale("C", $categories, \@Locale, $allow_incompatible);
- _trylocale("POSIX", $categories, \@Locale, $allow_incompatible);
- foreach (0..15) {
- _trylocale("ISO8859-$_", $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);
+ _trylocale("C", \@categories, \@Locale, $allow_incompatible);
+ _trylocale("POSIX", \@categories, \@Locale, $allow_incompatible);
+
+ if ($Config{d_has_C_UTF8} && $Config{d_has_C_UTF8} eq 'true') {
+ _trylocale("C.UTF-8", \@categories, \@Locale, $allow_incompatible);
+ }
+
+ # There's no point in looking at anything more if we know that setlocale
+ # will return success on any garbage or non-garbage name.
+ return sort @Locale if defined $Config{d_setlocale_accepts_any_locale_name};
+
+ foreach (1..16) {
+ _trylocale("ISO8859-$_", \@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, $allow_incompatible);
}
# Sanitize the environment so that we can run the external 'locale'
# locales will cause all IO hadles to default to (assume) utf8
next unless utf8::valid($_);
chomp;
- _trylocale($_, $categories, \@Locale, $allow_incompatible);
+ _trylocale($_, \@categories, \@Locale, $allow_incompatible);
}
close(LOCALES);
} elsif ($^O eq 'VMS'
opendir(LOCALES, "SYS\$I18N_LOCALE:");
while ($_ = readdir(LOCALES)) {
chomp;
- _trylocale($_, $categories, \@Locale, $allow_incompatible);
+ _trylocale($_, \@categories, \@Locale, $allow_incompatible);
}
close(LOCALES);
} elsif (($^O eq 'openbsd' || $^O eq 'bitrig' ) && -e '/usr/share/locale') {
opendir(LOCALES, '/usr/share/locale');
while ($_ = readdir(LOCALES)) {
chomp;
- _trylocale($_, $categories, \@Locale, $allow_incompatible);
+ _trylocale($_, \@categories, \@Locale, $allow_incompatible);
}
close(LOCALES);
} else { # Final fallback. Try our list of locales hard-coded here
# 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/)
- ? "./lib/locale/utf8"
- : "./lib/locale/latin1";
+ ? _source_location() . "/lib/locale/utf8"
+ : _source_location() . "/lib/locale/latin1";
if (-e $data_file) {
@Data = do $data_file;
}
}
# The rest of the locales are in this file.
- push @Data, <DATA>;
+ state @my_data = <DATA>; close DATA if fileno DATA;
+ push @Data, @my_data;
foreach my $line (@Data) {
+ chomp $line;
my ($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, $allow_incompatible);
+ _trylocale($loc, \@categories, \@Locale, $allow_incompatible);
foreach my $enc (@enc) {
- _trylocale("$loc.$enc", $categories, \@Locale,
+ _trylocale("$loc.$enc", \@categories, \@Locale,
$allow_incompatible);
}
$loc = lc $loc;
foreach my $enc (@enc) {
- _trylocale("$loc.$enc", $categories, \@Locale,
+ _trylocale("$loc.$enc", \@categories, \@Locale,
$allow_incompatible);
}
}
foreach my $lang (split(/ /, $language_codes)) {
- _trylocale($lang, $categories, \@Locale, $allow_incompatible);
+ _trylocale($lang, \@categories, \@Locale, $allow_incompatible);
foreach my $country (split(/ /, $country_codes)) {
my $lc = "${lang}_${country}";
- _trylocale($lc, $categories, \@Locale, $allow_incompatible);
+ _trylocale($lc, \@categories, \@Locale, $allow_incompatible);
foreach my $enc (@enc) {
- _trylocale("$lc.$enc", $categories, \@Locale,
+ _trylocale("$lc.$enc", \@categories, \@Locale,
$allow_incompatible);
}
my $lC = "${lang}_\U${country}";
- _trylocale($lC, $categories, \@Locale, $allow_incompatible);
+ _trylocale($lC, \@categories, \@Locale, $allow_incompatible);
foreach my $enc (@enc) {
- _trylocale("$lC.$enc", $categories, \@Locale,
+ _trylocale("$lC.$enc", \@categories, \@Locale,
$allow_incompatible);
}
}
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
return unless locales_enabled('LC_CTYPE');
my $locales_ref = shift;
+ my @return;
if (! defined $locales_ref) {
}
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;
+
+ return unless locales_enabled('LC_CTYPE');
+
+ 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