[perl5.git] / t / loc_tools.pl

# 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
# the functions.

# Note that it's okay that some languages have their native names
# capitalized here even though that's not "right".  They are lowercased
# 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
    my $locale = 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/;
        };
    }

    if ($badutf8) {
        ok(0, "Verify locale name doesn't contain malformed utf8");
        return;
    }
    push @$list, $locale;
}

sub _decode_encodings {
    my @enc;

    foreach (split(/ /, shift)) {
	if (/^(\d+)$/) {
	    push @enc, "ISO8859-$1";
	    push @enc, "iso8859$1";	# HP
	    if ($1 eq '1') {
		 push @enc, "roman8";	# HP
	    }
	    push @enc, $_;
            push @enc, "$_.UTF-8";
            push @enc, "$_.65001"; # Windows UTF-8
            push @enc, "$_.ACP"; # Windows ANSI code page
            push @enc, "$_.OCP"; # Windows OEM code page
	}
    }
    if ($^O eq 'os390') {
	push @enc, qw(IBM-037 IBM-819 IBM-1047);
    }
    push @enc, "UTF-8";
    push @enc, "65001"; # Windows UTF-8

    return @enc;
}

sub find_locales {  # Returns an array of all the locales we found on the
                    # system

    use Config;;
    my $have_setlocale = $Config{d_setlocale};

    # Visual C's CRT goes silly on strings of the form "en_US.ISO8859-1"
    # and mingw32 uses said silly CRT
    # This doesn't seem to be an issue any more, at least on Windows XP,
    # 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);

    # 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.
    require POSIX; import POSIX 'locale_h';

    _trylocale("C", \@Locale);
    _trylocale("POSIX", \@Locale);
    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);
    }

    # 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 local $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};

    # Other subversive stuff.
    delete local @ENV{qw(IFS CDPATH ENV BASH_ENV)};

    if (-x "/usr/bin/locale"
        && open(LOCALES, "/usr/bin/locale -a 2>/dev/null|"))
    {
        while (<LOCALES>) {
            # It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
            # ain't great when we're running this testPERL_UNICODE= so that utf8
            # locales will cause all IO hadles to default to (assume) utf8
            next unless utf8::valid($_);
            chomp;
            _trylocale($_, \@Locale);
        }
        close(LOCALES);
    } elsif ($^O eq 'VMS'
             && defined($ENV{'SYS$I18N_LOCALE'})
             && -d 'SYS$I18N_LOCALE')
    {
    # The SYS$I18N_LOCALE logical name search list was not present on
    # VAX VMS V5.5-12, but was on AXP && VAX VMS V6.2 as well as later versions.
        opendir(LOCALES, "SYS\$I18N_LOCALE:");
        while ($_ = readdir(LOCALES)) {
            chomp;
            _trylocale($_, \@Locale);
        }
        close(LOCALES);
    } elsif (($^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.

        opendir(LOCALES, '/usr/share/locale');
        while ($_ = readdir(LOCALES)) {
            chomp;
            _trylocale($_, \@Locale);
        }
        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";
        }

        # The rest of the locales are in this file.
        push @Data, <DATA>;

        foreach my $line (@Data) {
            my ($locale_name, $language_codes, $country_codes, $encodings) =
                split /:/, $line;
            my @enc = _decode_encodings($encodings);
            foreach my $loc (split(/ /, $locale_name)) {
                _trylocale($loc, \@Locale);
                foreach my $enc (@enc) {
                    _trylocale("$loc.$enc", \@Locale);
                }
                $loc = lc $loc;
                foreach my $enc (@enc) {
                    _trylocale("$loc.$enc", \@Locale);
                }
            }
            foreach my $lang (split(/ /, $language_codes)) {
                _trylocale($lang, \@Locale);
                foreach my $country (split(/ /, $country_codes)) {
                    my $lc = "${lang}_${country}";
                    _trylocale($lc, \@Locale);
                    foreach my $enc (@enc) {
                        _trylocale("$lc.$enc", \@Locale);
                    }
                    my $lC = "${lang}_\U${country}";
                    _trylocale($lC, \@Locale);
                    foreach my $enc (@enc) {
                        _trylocale("$lC.$enc", \@Locale);
                    }
                }
            }
        }
    }

    @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
    my $locale = shift;

    use 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_locale (;$) { # Return the name of locale that core Perl thinks
                            # is a UTF-8 locale.  Optional second 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) {
        my @locales = find_locales();
        $locales_ref = \@locales;
    }

    foreach my $locale (@$locales_ref) {
        return $locale if is_locale_utf8($locale);
    }

    return;
}

1

# Format of data is: locale_name, language_codes, country_codes, encodings
__DATA__
Afrikaans:af:za:1 15
Arabic:ar:dz eg sa:6 arabic8
Brezhoneg Breton:br:fr:1 15
Bulgarski Bulgarian:bg:bg:5
Chinese:zh:cn tw:cn.EUC eucCN eucTW euc.CN euc.TW Big5 GB2312 tw.EUC
Hrvatski Croatian:hr:hr:2
Cymraeg Welsh:cy:cy:1 14 15
Czech:cs:cz:2
Dansk Danish:da:dk:1 15
Nederlands Dutch:nl:be nl:1 15
English American British:en:au ca gb ie nz us uk zw:1 15 cp850
Esperanto:eo:eo:3
Eesti Estonian:et:ee:4 6 13
Suomi Finnish:fi:fi:1 15
Flamish::fl:1 15
Deutsch German:de:at be ch de lu:1 15
Euskaraz Basque:eu:es fr:1 15
Galego Galician:gl:es:1 15
Ellada Greek:el:gr:7 g8
Frysk:fy:nl:1 15
Greenlandic:kl:gl:4 6
Hebrew:iw:il:8 hebrew8
Hungarian:hu:hu:2
Indonesian:id:id:1 15
Gaeilge Irish:ga:IE:1 14 15
Italiano Italian:it:ch it:1 15
Nihongo Japanese:ja:jp:euc eucJP jp.EUC sjis
Korean:ko:kr:
Latine Latin:la:va:1 15
Latvian:lv:lv:4 6 13
Lithuanian:lt:lt:4 6 13
Macedonian:mk:mk:1 15
Maltese:mt:mt:3
Moldovan:mo:mo:2
Norsk Norwegian:no no\@nynorsk nb nn:no:1 15
Occitan:oc:es:1 15
Polski Polish:pl:pl:2
Rumanian:ro:ro:2
Russki Russian:ru:ru su ua:5 koi8 koi8r KOI8-R koi8u cp1251 cp866
Serbski Serbian:sr:yu:5
Slovak:sk:sk:2
Slovene Slovenian:sl:si:2
Sqhip Albanian:sq:sq:1 15
Svenska Swedish:sv:fi se:1 15
Thai:th:th:11 tis620
Turkish:tr:tr:9 turkish8
Yiddish:yi::1 15
Commit	Line	Data
73fc293b	1	# Common tools for test files files to find the locales which exist on the
abfe20b2 KW	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
	4	# the functions.
73fc293b KW	5
	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).
	10
73fc293b KW	11	sub _trylocale { # Adds the locale given by the first parameter to the list
	12	# given by the 2nd iff the platform supports the locale,
	13	# and it is not already on the list
	14	my $locale = shift;
	15	my $list = shift;
	16	return if grep { $locale eq $_ } @$list;
	17	return unless setlocale(&POSIX::LC_ALL, $locale);
	18	my $badutf8;
	19	{
	20	local $SIG{__WARN__} = sub {
	21	$badutf8 = $_[0] =~ /Malformed UTF-8/;
	22	};
	23	}
	24
	25	if ($badutf8) {
	26	ok(0, "Verify locale name doesn't contain malformed utf8");
	27	return;
	28	}
	29	push @$list, $locale;
	30	}
	31
	32	sub _decode_encodings {
	33	my @enc;
	34
	35	foreach (split(/ /, shift)) {
	36	if (/^(\d+)$/) {
	37	push @enc, "ISO8859-$1";
	38	push @enc, "iso8859$1"; # HP
	39	if ($1 eq '1') {
	40	push @enc, "roman8"; # HP
	41	}
	42	push @enc, $_;
	43	push @enc, "$_.UTF-8";
d646bffe KW	44	push @enc, "$_.65001"; # Windows UTF-8
	45	push @enc, "$_.ACP"; # Windows ANSI code page
	46	push @enc, "$_.OCP"; # Windows OEM code page
73fc293b KW	47	}
	48	}
	49	if ($^O eq 'os390') {
	50	push @enc, qw(IBM-037 IBM-819 IBM-1047);
	51	}
	52	push @enc, "UTF-8";
d646bffe	53	push @enc, "65001"; # Windows UTF-8
73fc293b KW	54
	55	return @enc;
	56	}
	57
	58	sub find_locales { # Returns an array of all the locales we found on the
	59	# system
	60
abfe20b2 KW	61	use Config;;
abfe20b2 KW	62	my $have_setlocale = $Config{d_setlocale};
d369fd5b KW	63
	64	# Visual C's CRT goes silly on strings of the form "en_US.ISO8859-1"
	65	# and mingw32 uses said silly CRT
	66	# This doesn't seem to be an issue any more, at least on Windows XP,
	67	# so re-enable the tests for Windows XP onwards.
	68	my $winxp = ($^O eq 'MSWin32' && defined &Win32::GetOSVersion &&
	69	join('.', (Win32::GetOSVersion())[1..2]) >= 5.1);
	70	$have_setlocale = 0 if ((($^O eq 'MSWin32' && !$winxp) \|\| $^O eq 'NetWare') &&
	71	$Config{cc} =~ /^(cl\|gcc\|g\+\+\|ici)/i);
	72
	73	# UWIN seems to loop after taint tests, just skip for now
	74	$have_setlocale = 0 if ($^O =~ /^uwin/);
	75
	76	return unless $have_setlocale;
	77
abfe20b2 KW	78	# Done this way in case this is 'required' in the caller before seeing if
	79	# this is miniperl.
	80	require POSIX; import POSIX 'locale_h';
	81
73fc293b KW	82	_trylocale("C", \@Locale);
	83	_trylocale("POSIX", \@Locale);
	84	foreach (0..15) {
	85	_trylocale("ISO8859-$_", \@Locale);
	86	_trylocale("iso8859$_", \@Locale);
	87	_trylocale("iso8859-$_", \@Locale);
	88	_trylocale("iso_8859_$_", \@Locale);
	89	_trylocale("isolatin$_", \@Locale);
	90	_trylocale("isolatin-$_", \@Locale);
	91	_trylocale("iso_latin_$_", \@Locale);
	92	}
	93
	94	# Sanitize the environment so that we can run the external 'locale'
	95	# program without the taint mode getting grumpy.
	96
	97	# $ENV{PATH} is special in VMS.
8b76dc1d	98	delete local $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};
73fc293b KW	99
73fc293b KW	100	# Other subversive stuff.
8b76dc1d	101	delete local @ENV{qw(IFS CDPATH ENV BASH_ENV)};
73fc293b KW	102
	103	if (-x "/usr/bin/locale"
	104	&& open(LOCALES, "/usr/bin/locale -a 2>/dev/null\|"))
	105	{
	106	while (<LOCALES>) {
	107	# It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
	108	# ain't great when we're running this testPERL_UNICODE= so that utf8
	109	# locales will cause all IO hadles to default to (assume) utf8
	110	next unless utf8::valid($_);
	111	chomp;
	112	_trylocale($_, \@Locale);
	113	}
	114	close(LOCALES);
	115	} elsif ($^O eq 'VMS'
	116	&& defined($ENV{'SYS$I18N_LOCALE'})
	117	&& -d 'SYS$I18N_LOCALE')
	118	{
	119	# The SYS$I18N_LOCALE logical name search list was not present on
	120	# VAX VMS V5.5-12, but was on AXP && VAX VMS V6.2 as well as later versions.
	121	opendir(LOCALES, "SYS\$I18N_LOCALE:");
	122	while ($_ = readdir(LOCALES)) {
	123	chomp;
	124	_trylocale($_, \@Locale);
	125	}
	126	close(LOCALES);
	127	} elsif (($^O eq 'openbsd' \|\| $^O eq 'bitrig' ) && -e '/usr/share/locale') {
	128
	129	# OpenBSD doesn't have a locale executable, so reading /usr/share/locale
	130	# is much easier and faster than the last resort method.
	131
	132	opendir(LOCALES, '/usr/share/locale');
	133	while ($_ = readdir(LOCALES)) {
	134	chomp;
	135	_trylocale($_, \@Locale);
	136	}
	137	close(LOCALES);
	138	} else { # Final fallback. Try our list of locales hard-coded here
	139
	140	# This is going to be slow.
	141	my @Data;
	142
	143
	144	# Locales whose name differs if the utf8 bit is on are stored in these two
	145	# files with appropriate encodings.
	146	if ($^H & 0x08 \|\| (${^OPEN} \|\| "") =~ /:utf8/) {
	147	@Data = do "lib/locale/utf8";
	148	} else {
	149	@Data = do "lib/locale/latin1";
	150	}
	151
	152	# The rest of the locales are in this file.
	153	push @Data, <DATA>;
	154
	155	foreach my $line (@Data) {
	156	my ($locale_name, $language_codes, $country_codes, $encodings) =
	157	split /:/, $line;
	158	my @enc = _decode_encodings($encodings);
	159	foreach my $loc (split(/ /, $locale_name)) {
	160	_trylocale($loc, \@Locale);
	161	foreach my $enc (@enc) {
	162	_trylocale("$loc.$enc", \@Locale);
	163	}
	164	$loc = lc $loc;
	165	foreach my $enc (@enc) {
166	_trylocale("$loc.$enc", \@Locale);
167	}
168	}
169	foreach my $lang (split(/ /, $language_codes)) {
170	_trylocale($lang, \@Locale);
171	foreach my $country (split(/ /, $country_codes)) {
172	my $lc = "${lang}_${country}";
173	_trylocale($lc, \@Locale);
174	foreach my $enc (@enc) {
175	_trylocale("$lc.$enc", \@Locale);
176	}
177	my $lC = "${lang}_\U${country}";
178	_trylocale($lC, \@Locale);
179	foreach my $enc (@enc) {
180	_trylocale("$lC.$enc", \@Locale);
181	}
182	}
183	}
184	}
185	}
186
187	@Locale = sort @Locale;
188
189	return @Locale;
190
191
192	}
193
ab8b8bcc KW	194	sub is_locale_utf8 ($) { # Return a boolean as to if core Perl thinks the input
	195	# is a UTF-8 locale
	196	my $locale = shift;
	197
31f05a37 KW	198	use locale;
	199
	200	my $save_locale = setlocale(&POSIX::LC_CTYPE());
	201	if (! $save_locale) {
	202	ok(0, "Verify could save previous locale");
	203	return 0;
	204	}
	205
	206	if (! setlocale(&POSIX::LC_CTYPE(), $locale)) {
	207	ok(0, "Verify could setlocale to $locale");
	208	return 0;
	209	}
	210
	211	my $ret = 0;
	212
	213	# Use an op that gives different results for UTF-8 than any other locale.
	214	# If a platform has UTF-8 locales, there should be at least one locale on
	215	# most platforms with UTF-8 in its name, so if there is a bug in the op
	216	# giving a false negative, we should get a failure for those locales as we
	217	# go through testing all the locales on the platform.
	218	if (CORE::fc(chr utf8::unicode_to_native(0xdf)) ne "ss") {
	219	if ($locale =~ /UTF-?8/i) {
	220	ok (0, "Verify $locale with UTF-8 in name is a UTF-8 locale");
	221	}
	222	}
	223	else {
	224	$ret = 1;
	225	}
	226
	227	die "Couldn't restore locale '$save_locale'"
	228	unless setlocale(&POSIX::LC_CTYPE(), $save_locale);
	229
	230	return $ret;
ab8b8bcc KW	231	}
	232
	233	sub find_utf8_locale (;$) { # Return the name of locale that core Perl thinks
	234	# is a UTF-8 locale. Optional second parameter is
	235	# a reference to a list of locales to try; if
	236	# omitted, this tries all locales it can find on
	237	# the platform
	238	my $locales_ref = shift;
	239	if (! defined $locales_ref) {
	240	my @locales = find_locales();
	241	$locales_ref = \@locales;
	242	}
	243
	244	foreach my $locale (@$locales_ref) {
	245	return $locale if is_locale_utf8($locale);
	246	}
	247
	248	return;
	249	}
	250
73fc293b KW	251	1
	252
	253	# Format of data is: locale_name, language_codes, country_codes, encodings
	254	__DATA__
	255	Afrikaans:af:za:1 15
	256	Arabic:ar:dz eg sa:6 arabic8
	257	Brezhoneg Breton:br:fr:1 15
	258	Bulgarski Bulgarian:bg:bg:5
	259	Chinese:zh:cn tw:cn.EUC eucCN eucTW euc.CN euc.TW Big5 GB2312 tw.EUC
	260	Hrvatski Croatian:hr:hr:2
	261	Cymraeg Welsh:cy:cy:1 14 15
	262	Czech:cs:cz:2
	263	Dansk Danish:da:dk:1 15
	264	Nederlands Dutch:nl:be nl:1 15
	265	English American British:en:au ca gb ie nz us uk zw:1 15 cp850
	266	Esperanto:eo:eo:3
	267	Eesti Estonian:et:ee:4 6 13
	268	Suomi Finnish:fi:fi:1 15
	269	Flamish::fl:1 15
	270	Deutsch German:de:at be ch de lu:1 15
	271	Euskaraz Basque:eu:es fr:1 15
	272	Galego Galician:gl:es:1 15
	273	Ellada Greek:el:gr:7 g8
	274	Frysk:fy:nl:1 15
	275	Greenlandic:kl:gl:4 6
	276	Hebrew:iw:il:8 hebrew8
	277	Hungarian:hu:hu:2
	278	Indonesian:id:id:1 15
	279	Gaeilge Irish:ga:IE:1 14 15
	280	Italiano Italian:it:ch it:1 15
	281	Nihongo Japanese:ja:jp:euc eucJP jp.EUC sjis
	282	Korean:ko:kr:
	283	Latine Latin:la:va:1 15
	284	Latvian:lv:lv:4 6 13
	285	Lithuanian:lt:lt:4 6 13
	286	Macedonian:mk:mk:1 15
	287	Maltese:mt:mt:3
	288	Moldovan:mo:mo:2
	289	Norsk Norwegian:no no\@nynorsk nb nn:no:1 15
	290	Occitan:oc:es:1 15
	291	Polski Polish:pl:pl:2
	292	Rumanian:ro:ro:2
	293	Russki Russian:ru:ru su ua:5 koi8 koi8r KOI8-R koi8u cp1251 cp866
	294	Serbski Serbian:sr:yu:5
	295	Slovak:sk:sk:2
	296	Slovene Slovenian:sl:si:2
	297	Sqhip Albanian:sq:sq:1 15
	298	Svenska Swedish:sv:fi se:1 15
	299	Thai:th:th:11 tis620
	300	Turkish:tr:tr:9 turkish8
	301	Yiddish:yi::1 15