print "1..0\n";
exit;
}
+ require './loc_tools.pl';
$| = 1;
}
use strict;
use feature 'fc';
+# =1 adds debugging output; =2 increases the verbosity somewhat
my $debug = $ENV{PERL_DEBUG_FULL_TEST} // 0;
# Certain tests have been shown to be problematical for a few locales. Don't
# fail them unless at least this percentage of the tested locales fail.
-# Some Windows machines are defective in every in every locale but the C,
-# calling \t printable; superscripts to be digits, etc. See
-# http://markmail.org/message/5jwam4xsx4amsdnv
+# Some Windows machines are defective in every locale but the C, calling \t
+# printable; superscripts to be digits, etc. See
+# http://markmail.org/message/5jwam4xsx4amsdnv. Also on AIX machines, many
+# locales call a no-break space a graphic.
# (There aren't 1000 locales currently in existence, so 99.9 works)
-my $acceptable_fold_failure_percentage = $^O eq 'MSWin32' ? 99.9 : 5;
+my $acceptable_fold_failure_percentage = ($^O =~ / ^ ( MSWin32 | AIX ) $ /ix)
+ ? 99.9
+ : 5;
+
+# The list of test numbers of the problematic tests.
+my @problematical_tests;
+
use Dumpvalue;
print $dumper->stringify($mess,1), "\n";
}
+sub debug_more {
+ return unless $debug > 1;
+ return debug(@_);
+}
+
sub debugf {
printf @_ if $debug;
}
-my $have_setlocale = 0;
-eval {
- require POSIX;
- import POSIX ':locale_h';
- $have_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') &&
- $Config{cc} =~ /^(cl|gcc)/i);
-
-# UWIN seems to loop after taint tests, just skip for now
-$have_setlocale = 0 if ($^O =~ /^uwin/);
-
$a = 'abc %';
my $test_num = 0;
use locale; # engage locale and therefore locale taint.
-check_taint_not $a;
+check_taint_not $a, "\t\$a";
check_taint uc($a);
check_taint "\U$a";
$_ = uc($a); # taint $_
-check_taint $_;
+check_taint $_, "\t\$_";
/(\w)/; # taint $&, $`, $', $+, $1.
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $&, "\t/(\\w)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
/(.)/; # untaint $&, $`, $', $+, $1.
-check_taint_not $&;
-check_taint_not $`;
-check_taint_not $';
-check_taint_not $+;
-check_taint_not $1;
-check_taint_not $2;
+check_taint_not $&, "\t/(.)/ \$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+check_taint_not $2, "\t\$2";
/(\W)/; # taint $&, $`, $', $+, $1.
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $&, "\t/(\\W)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/(.)/ \$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+check_taint_not $2, "\t\$2";
/(\s)/; # taint $&, $`, $', $+, $1.
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $&, "\t/(\\s)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/(.)/ \$&";
/(\S)/; # taint $&, $`, $', $+, $1.
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $&, "\t/(\\S)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/(.)/ \$&";
+
+"a" =~ /(a)|(\w)/; # taint $&, $`, $', $+, $1.
+check_taint $&, "\t/(a)|(\\w)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+ok($1 eq 'a', ("\t" x 4) . "\$1 is 'a'");
+ok(! defined $2, ("\t" x 4) . "\$2 is undefined");
+check_taint_not $2, "\t\$2";
+check_taint_not $3, "\t\$3";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/(.)/ \$&";
+
+"\N{CYRILLIC SMALL LETTER A}" =~ /(\N{CYRILLIC CAPITAL LETTER A})/i; # no tainting because no locale dependence
+check_taint_not $&, "\t/(\\N{CYRILLIC CAPITAL LETTER A})/i \$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+ok($1 eq "\N{CYRILLIC SMALL LETTER A}", ("\t" x 4) . "\$1 is 'small cyrillic a'");
+check_taint_not $2, "\t\$2";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/./ \$&";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/(.)/ \$&";
+
+"a:" =~ /(.)\b(.)/; # taint $&, $`, $', $+, $1.
+check_taint $&, "\t/(.)\\b(.)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint $2, "\t\$2";
+check_taint_not $3, "\t\$3";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/./ \$&";
+
+"aa" =~ /(.)\B(.)/; # taint $&, $`, $', $+, $1.
+check_taint $&, "\t/(.)\\B(.)/ \$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint $2, "\t\$2";
+check_taint_not $3, "\t\$3";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/./ \$&";
+
+"aaa" =~ /(.).(\1)/i; # notaint because not locale dependent
+check_taint_not $&, "\t/(.).(\\1)/ \$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+check_taint_not $2, "\t\$2";
+check_taint_not $3, "\t\$3";
+
+/(.)/; # untaint $&, $`, $', $+, $1.
+check_taint_not $&, "\t/./ \$&";
$_ = $a; # untaint $_
-check_taint_not $_;
+check_taint_not $_, "\t\$_";
/(b)/; # this must not taint
-check_taint_not $&;
-check_taint_not $`;
-check_taint_not $';
-check_taint_not $+;
-check_taint_not $1;
-check_taint_not $2;
+check_taint_not $&, "\t/(b)/ \$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+check_taint_not $2, "\t\$2";
$_ = $a; # untaint $_
-check_taint_not $_;
+check_taint_not $_, "\t\$_";
$b = uc($a); # taint $b
s/(.+)/$b/; # this must taint only the $_
-check_taint $_;
-check_taint_not $&;
-check_taint_not $`;
-check_taint_not $';
-check_taint_not $+;
-check_taint_not $1;
-check_taint_not $2;
+check_taint $_, "\t\$_";
+check_taint_not $&, "\t\$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+check_taint_not $2, "\t\$2";
$_ = $a; # untaint $_
s/(.+)/b/; # this must not taint
-check_taint_not $_;
-check_taint_not $&;
-check_taint_not $`;
-check_taint_not $';
-check_taint_not $+;
-check_taint_not $1;
-check_taint_not $2;
+check_taint_not $_, "\t\$_";
+check_taint_not $&, "\t\$&";
+check_taint_not $`, "\t\$`";
+check_taint_not $', "\t\$'";
+check_taint_not $+, "\t\$+";
+check_taint_not $1, "\t\$1";
+check_taint_not $2, "\t\$2";
$b = $a; # untaint $b
($b = $a) =~ s/\w/$&/;
-check_taint $b; # $b should be tainted.
-check_taint_not $a; # $a should be not.
+check_taint $b, "\t\$b"; # $b should be tainted.
+check_taint_not $a, "\t\$a"; # $a should be not.
$_ = $a; # untaint $_
s/(\w)/\l$1/; # this must taint
-check_taint $_;
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $_, "\t\$_";
+check_taint $&, "\t\$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
$_ = $a; # untaint $_
s/(\w)/\L$1/; # this must taint
-check_taint $_;
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $_, "\t\$_";
+check_taint $&, "\t\$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
$_ = $a; # untaint $_
s/(\w)/\u$1/; # this must taint
-check_taint $_;
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $_, "\t\$_";
+check_taint $&, "\t\$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
$_ = $a; # untaint $_
s/(\w)/\U$1/; # this must taint
-check_taint $_;
-check_taint $&;
-check_taint $`;
-check_taint $';
-check_taint $+;
-check_taint $1;
-check_taint_not $2;
+check_taint $_, "\t\$_";
+check_taint $&, "\t\$&";
+check_taint $`, "\t\$`";
+check_taint $', "\t\$'";
+check_taint $+, "\t\$+";
+check_taint $1, "\t\$1";
+check_taint_not $2, "\t\$2";
# After all this tainting $a should be cool.
-check_taint_not $a;
+check_taint_not $a, "\t\$a";
+
+"a" =~ /([a-z])/;
+check_taint_not $1, '"a" =~ /([a-z])/';
+"foo.bar_baz" =~ /^(.*)[._](.*?)$/; # Bug 120675
+check_taint_not $1, '"foo.bar_baz" =~ /^(.*)[._](.*?)$/';
+
+# BE SURE TO COPY ANYTHING YOU ADD to the block below
{ # This is just the previous tests copied here with a different
# compile-time pragma.
# After all this tainting $a should be cool.
check_taint_not $a;
+
+ "a" =~ /([a-z])/;
+ check_taint_not $1, '"a" =~ /([a-z])/';
+ "foo.bar_baz" =~ /^(.*)[._](.*?)$/; # Bug 120675
+ check_taint_not $1, '"foo.bar_baz" =~ /^(.*)[._](.*?)$/';
}
# Here are in scope of 'use locale'
# Let us do some *real* locale work now,
# unless setlocale() is missing (i.e. minitest).
-unless ($have_setlocale) {
- print "1..$test_num\n";
- exit;
-}
-
# The test number before our first setlocale()
my $final_without_setlocale = $test_num;
debug "# Scanning for locales...\n";
-# 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).
-
-my $locales = <<EOF;
-Afrikaans:af:za:1 15
-Arabic:ar:dz eg sa:6 arabic8
-Brezhoneg Breton:br:fr:1 15
-Bulgarski Bulgarian:bg:bg:5
-Chinese:zh:cn tw:cn.EUC eucCN eucTW euc.CN euc.TW Big5 GB2312 tw.EUC
-Hrvatski Croatian:hr:hr:2
-Cymraeg Welsh:cy:cy:1 14 15
-Czech:cs:cz:2
-Dansk Danish:da:dk:1 15
-Nederlands Dutch:nl:be nl:1 15
-English American British:en:au ca gb ie nz us uk zw:1 15 cp850
-Esperanto:eo:eo:3
-Eesti Estonian:et:ee:4 6 13
-Suomi Finnish:fi:fi:1 15
-Flamish::fl:1 15
-Deutsch German:de:at be ch de lu:1 15
-Euskaraz Basque:eu:es fr:1 15
-Galego Galician:gl:es:1 15
-Ellada Greek:el:gr:7 g8
-Frysk:fy:nl:1 15
-Greenlandic:kl:gl:4 6
-Hebrew:iw:il:8 hebrew8
-Hungarian:hu:hu:2
-Indonesian:id:id:1 15
-Gaeilge Irish:ga:IE:1 14 15
-Italiano Italian:it:ch it:1 15
-Nihongo Japanese:ja:jp:euc eucJP jp.EUC sjis
-Korean:ko:kr:
-Latine Latin:la:va:1 15
-Latvian:lv:lv:4 6 13
-Lithuanian:lt:lt:4 6 13
-Macedonian:mk:mk:1 15
-Maltese:mt:mt:3
-Moldovan:mo:mo:2
-Norsk Norwegian:no no\@nynorsk nb nn:no:1 15
-Occitan:oc:es:1 15
-Polski Polish:pl:pl:2
-Rumanian:ro:ro:2
-Russki Russian:ru:ru su ua:5 koi8 koi8r KOI8-R koi8u cp1251 cp866
-Serbski Serbian:sr:yu:5
-Slovak:sk:sk:2
-Slovene Slovenian:sl:si:2
-Sqhip Albanian:sq:sq:1 15
-Svenska Swedish:sv:fi se:1 15
-Thai:th:th:11 tis620
-Turkish:tr:tr:9 turkish8
-Yiddish:yi::1 15
-EOF
-
-if ($^O eq 'os390') {
- # These cause heartburn. Broken locales?
- $locales =~ s/Svenska Swedish:sv:fi se:1 15\n//;
- $locales =~ s/Thai:th:th:11 tis620\n//;
-}
-
-sub in_utf8 () { $^H & 0x08 || (${^OPEN} || "") =~ /:utf8/ }
-
-if (in_utf8) {
- require "lib/locale/utf8";
-} else {
- require "lib/locale/latin1";
-}
-
-my @Locale;
-my $Locale;
-my %posixes;
-
-sub trylocale {
- my $locale = shift;
- return if grep { $locale eq $_ } @Locale;
- return unless setlocale(&POSIX::LC_ALL, $locale);
- my $badutf8;
- {
- local $SIG{__WARN__} = sub {
- $badutf8 = $_[0] =~ /Malformed UTF-8/;
- };
- $Locale =~ /UTF-?8/i;
- }
-
- if ($badutf8) {
- ok(0, "Locale name contains malformed utf8");
- return;
- }
- push @Locale, $locale;
-}
+require POSIX; import POSIX ':locale_h';
-sub decode_encodings {
- my @enc;
-
- foreach (split(/ /, shift)) {
- if (/^(\d+)$/) {
- push @enc, "ISO8859-$1";
- push @enc, "iso8859$1"; # HP
- if ($1 eq '1') {
- push @enc, "roman8"; # HP
- }
- } else {
- push @enc, $_;
- push @enc, "$_.UTF-8";
- }
- }
- if ($^O eq 'os390') {
- push @enc, qw(IBM-037 IBM-819 IBM-1047);
- }
+my @Locale = find_locales([ &POSIX::LC_CTYPE, &POSIX::LC_ALL ]);
- return @enc;
+debug "# Locales =\n";
+for ( @Locale ) {
+ debug "# $_\n";
}
-trylocale("C");
-trylocale("POSIX");
-foreach (0..15) {
- trylocale("ISO8859-$_");
- trylocale("iso8859$_");
- trylocale("iso8859-$_");
- trylocale("iso_8859_$_");
- trylocale("isolatin$_");
- trylocale("isolatin-$_");
- trylocale("iso_latin_$_");
+unless (@Locale) {
+ print "1..$test_num\n";
+ exit;
}
-# 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 $Config{d_setenv};
-
-# Other subversive stuff.
-delete @ENV{qw(IFS CDPATH ENV BASH_ENV)};
-
-if (-x "/usr/bin/locale" && open(LOCALES, "/usr/bin/locale -a 2>/dev/null|")) {
- while (<LOCALES>) {
- # It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
- # ain't great when we're running this testPERL_UNICODE= so that utf8
- # locales will cause all IO hadles to default to (assume) utf8
- next unless utf8::valid($_);
- chomp;
- trylocale($_);
- }
- close(LOCALES);
-} elsif ($^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 ($_ = readdir(LOCALES)) {
- chomp;
- trylocale($_);
- }
- 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($_);
- }
- close(LOCALES);
-} else {
-
- # This is going to be slow.
-
- foreach my $locale (split(/\n/, $locales)) {
- my ($locale_name, $language_codes, $country_codes, $encodings) =
- split(/:/, $locale);
- my @enc = decode_encodings($encodings);
- foreach my $loc (split(/ /, $locale_name)) {
- trylocale($loc);
- foreach my $enc (@enc) {
- trylocale("$loc.$enc");
- }
- $loc = lc $loc;
- foreach my $enc (@enc) {
- trylocale("$loc.$enc");
- }
- }
- foreach my $lang (split(/ /, $language_codes)) {
- trylocale($lang);
- foreach my $country (split(/ /, $country_codes)) {
- my $lc = "${lang}_${country}";
- trylocale($lc);
- foreach my $enc (@enc) {
- trylocale("$lc.$enc");
- }
- my $lC = "${lang}_\U${country}";
- trylocale($lC);
- foreach my $enc (@enc) {
- trylocale("$lC.$enc");
- }
- }
- }
- }
-}
setlocale(&POSIX::LC_ALL, "C");
-if ($^O eq 'darwin') {
- # Darwin 8/Mac OS X 10.4 and 10.5 have bad Basque locales: perl bug #35895,
- # Apple bug ID# 4139653. It also has a problem in Byelorussian.
- (my $v) = $Config{osvers} =~ /^(\d+)/;
- if ($v >= 8 and $v < 10) {
- debug "# Skipping eu_ES, be_BY locales -- buggy in Darwin\n";
- @Locale = grep ! m/^(eu_ES(?:\..*)?|be_BY\.CP1131)$/, @Locale;
- } elsif ($v < 12) {
- debug "# Skipping be_BY locales -- buggy in Darwin\n";
- @Locale = grep ! m/^be_BY\.CP1131$/, @Locale;
- }
-}
-
-@Locale = sort @Locale;
-
-debug "# Locales =\n";
-for ( @Locale ) {
- debug "# $_\n";
-}
+my %posixes;
my %Problem;
my %Okay;
my $locales_test_number;
my $not_necessarily_a_problem_test_number;
my $first_casing_test_number;
-my $final_casing_test_number;
my %setlocale_failed; # List of locales that setlocale() didn't work on
-foreach $Locale (@Locale) {
+foreach my $Locale (@Locale) {
$locales_test_number = $first_locales_test_number - 1;
debug "#\n";
debug "# Locale = $Locale\n";
next;
}
- # We test UTF-8 locales only under ':not_characters'; otherwise they have
- # documented deficiencies. Non- UTF-8 locales are tested only under plain
- # 'use locale', as otherwise we would have to convert everything in them
- # to Unicode.
- # The locale name doesn't necessarily have to have "utf8" in it to be a
- # UTF-8 locale, but it works mostly.
- my $is_utf8_locale = $Locale =~ /UTF-?8/i;
+ # We test UTF-8 locales only under ':not_characters'; It is easier to
+ # test them in other test files than here. Non- UTF-8 locales are tested
+ # only under plain 'use locale', as otherwise we would have to convert
+ # everything in them to Unicode.
my %UPPER = (); # All alpha X for which uc(X) == X and lc(X) != X
my %lower = (); # All alpha X for which lc(X) == X and uc(X) != X
my %BoThCaSe = (); # All alpha X for which uc(X) == lc(X) == X
+ my $is_utf8_locale = is_locale_utf8($Locale);
+
+ debug "# is utf8 locale? = $is_utf8_locale\n";
+
+ my $radix = localeconv()->{decimal_point};
+ if ($radix !~ / ^ [[:ascii:]] + $/x) {
+ use bytes;
+ $radix = disp_chars(split "", $radix);
+ }
+ debug "# radix = $radix\n";
+
if (! $is_utf8_locale) {
use locale;
@{$posixes{'word'}} = grep /\w/, map { chr } 0..255;
}
}
- debug "# UPPER = ", disp_chars(keys %UPPER), "\n";
- debug "# lower = ", disp_chars(keys %lower), "\n";
- debug "# BoThCaSe = ", disp_chars(keys %BoThCaSe), "\n";
+ debug "# UPPER = ", disp_chars(sort { ord $a <=> ord $b } keys %UPPER), "\n";
+ debug "# lower = ", disp_chars(sort { ord $a <=> ord $b } keys %lower), "\n";
+ debug "# BoThCaSe = ", disp_chars(sort { ord $a <=> ord $b } keys %BoThCaSe), "\n";
debug "# Unassigned = ", disp_chars(sort { ord $a <=> ord $b } keys %Unassigned), "\n";
my @failures;
my @fold_failures;
- foreach my $x (sort keys %UPPER) {
+ foreach my $x (sort { ord $a <=> ord $b } keys %UPPER) {
my $ok;
my $fold_ok;
if ($is_utf8_locale) {
undef @failures;
undef @fold_failures;
- foreach my $x (sort keys %lower) {
+ foreach my $x (sort { ord $a <=> ord $b } keys %lower) {
my $ok;
my $fold_ok;
if ($is_utf8_locale) {
}
}
- @Added_alpha = sort @Added_alpha;
+ @Added_alpha = sort { ord $a <=> ord $b } @Added_alpha;
debug "# Added_alpha = ", disp_chars(@Added_alpha), "\n";
}
report_multi_result($Locale, $locales_test_number, \@f);
- $final_casing_test_number = $locales_test_number;
+ foreach ($first_casing_test_number..$locales_test_number) {
+ push @problematical_tests, $_;
+ }
+
# Test for read-only scalars' locale vs non-locale comparisons.
my @f = ();
++$locales_test_number;
$test_names{$locales_test_number} = 'Verify case insensitive matching works';
- foreach my $x (sort keys %UPPER) {
+ foreach my $x (sort { ord $a <=> ord $b } keys %UPPER) {
if (! $is_utf8_locale) {
my $y = lc $x;
next unless uc $y eq $x;
- print "# UPPER $x lc $y ",
- $x =~ /$y/i ? 1 : 0, " ",
- $y =~ /$x/i ? 1 : 0, "\n" if 0;
+ debug_more( "# UPPER=", disp_chars(($x)),
+ "; lc=", disp_chars(($y)), "; ",
+ "; fc=", disp_chars((fc $x)), "; ",
+ disp_chars(($x)), "=~/", disp_chars(($y)), "/i=",
+ $x =~ /$y/i ? 1 : 0,
+ "; ",
+ disp_chars(($y)), "=~/", disp_chars(($x)), "/i=",
+ $y =~ /$x/i ? 1 : 0,
+ "\n");
#
# If $x and $y contain regular expression characters
# AND THEY lowercase (/i) to regular expression characters,
print "# Regex characters in '$x' or '$y', skipping test $locales_test_number for locale '$Locale'\n";
next;
}
- # With utf8 both will fail since the locale concept
- # of upper/lower does not work well in Unicode.
- push @f, $x unless $x =~ /$y/i == $y =~ /$x/i;
+ push @f, $x unless $x =~ /$y/i && $y =~ /$x/i;
# fc is not a locale concept, so Perl uses lc for it.
push @f, $x unless lc $x eq fc $x;
use locale ':not_characters';
my $y = lc $x;
next unless uc $y eq $x;
- print "# UPPER $x lc $y ",
- $x =~ /$y/i ? 1 : 0, " ",
- $y =~ /$x/i ? 1 : 0, "\n" if 0;
+ debug_more( "# UPPER=", disp_chars(($x)),
+ "; lc=", disp_chars(($y)), "; ",
+ "; fc=", disp_chars((fc $x)), "; ",
+ disp_chars(($x)), "=~/", disp_chars(($y)), "/i=",
+ $x =~ /$y/i ? 1 : 0,
+ "; ",
+ disp_chars(($y)), "=~/", disp_chars(($x)), "/i=",
+ $y =~ /$x/i ? 1 : 0,
+ "\n");
- # Here, we can fully test things, unlike plain 'use locale',
- # because this form does work well with Unicode
push @f, $x unless $x =~ /$y/i && $y =~ /$x/i;
# The places where Unicode's lc is different from fc are
}
}
- foreach my $x (sort keys %lower) {
+ foreach my $x (sort { ord $a <=> ord $b } keys %lower) {
if (! $is_utf8_locale) {
my $y = uc $x;
next unless lc $y eq $x;
- print "# lower $x uc $y ",
- $x =~ /$y/i ? 1 : 0, " ",
- $y =~ /$x/i ? 1 : 0, "\n" if 0;
+ debug_more( "# lower=", disp_chars(($x)),
+ "; uc=", disp_chars(($y)), "; ",
+ "; fc=", disp_chars((fc $x)), "; ",
+ disp_chars(($x)), "=~/", disp_chars(($y)), "/i=",
+ $x =~ /$y/i ? 1 : 0,
+ "; ",
+ disp_chars(($y)), "=~/", disp_chars(($x)), "/i=",
+ $y =~ /$x/i ? 1 : 0,
+ "\n");
if ($x =~ $re || $y =~ $re) { # See above.
print "# Regex characters in '$x' or '$y', skipping test $locales_test_number for locale '$Locale'\n";
next;
}
- # With utf8 both will fail since the locale concept
- # of upper/lower does not work well in Unicode.
- push @f, $x unless $x =~ /$y/i == $y =~ /$x/i;
+ push @f, $x unless $x =~ /$y/i && $y =~ /$x/i;
push @f, $x unless lc $x eq fc $x;
}
use locale ':not_characters';
my $y = uc $x;
next unless lc $y eq $x;
- print "# lower $x uc $y ",
- $x =~ /$y/i ? 1 : 0, " ",
- $y =~ /$x/i ? 1 : 0, "\n" if 0;
+ debug_more( "# lower=", disp_chars(($x)),
+ "; uc=", disp_chars(($y)), "; ",
+ "; fc=", disp_chars((fc $x)), "; ",
+ disp_chars(($x)), "=~/", disp_chars(($y)), "/i=",
+ $x =~ /$y/i ? 1 : 0,
+ "; ",
+ disp_chars(($y)), "=~/", disp_chars(($x)), "/i=",
+ $y =~ /$x/i ? 1 : 0,
+ "\n");
push @f, $x unless $x =~ /$y/i && $y =~ /$x/i;
push @f, $x unless lc $x eq fc $x;
}
}
report_multi_result($Locale, $locales_test_number, \@f);
+ push @problematical_tests, $locales_test_number;
}
# [perl #109318]
# Recount the errors.
-foreach ($first_locales_test_number..$final_locales_test_number) {
+foreach $test_num ($first_locales_test_number..$final_locales_test_number) {
if (%setlocale_failed) {
print "not ";
}
- elsif ($Problem{$_} || !defined $Okay{$_} || !@{$Okay{$_}}) {
+ elsif ($Problem{$test_num} || !defined $Okay{$test_num} || !@{$Okay{$test_num}}) {
if (defined $not_necessarily_a_problem_test_number
- && $_ == $not_necessarily_a_problem_test_number)
+ && $test_num == $not_necessarily_a_problem_test_number)
{
print "# The failure of test $not_necessarily_a_problem_test_number is not necessarily fatal.\n";
print "# It usually indicates a problem in the environment,\n";
print "# not in Perl itself.\n";
}
- if ($Okay{$_} && ($_ >= $first_casing_test_number
- && $_ <= $final_casing_test_number))
- {
+ if ($Okay{$test_num} && grep { $_ == $test_num } @problematical_tests) {
+ no warnings 'experimental::autoderef';
# Round to nearest .1%
- my $percent_fail = (int(.5 + (1000 * scalar(keys $Problem{$_})
+ my $percent_fail = (int(.5 + (1000 * scalar(keys $Problem{$test_num})
/ scalar(@Locale))))
/ 10;
if (! $debug && $percent_fail < $acceptable_fold_failure_percentage)
{
- $test_names{$_} .= 'TODO';
+ $test_names{$test_num} .= 'TODO';
print "# ", 100 - $percent_fail, "% of locales pass the following test, so it is likely that the failures\n";
print "# are errors in the locale definitions. The test is marked TODO, as the\n";
print "# problem is not likely to be Perl's\n";
print "#\n";
if ($debug) {
print "# The code points that had this failure are given above. Look for lines\n";
- print "# that match 'failed $_'\n";
+ print "# that match 'failed $test_num'\n";
}
else {
print "# For more details, rerun, with environment variable PERL_DEBUG_FULL_TEST=1.\n";
- print "# Then look at that output for lines that match 'failed $_'\n";
+ print "# Then look at that output for lines that match 'failed $test_num'\n";
}
print "not ";
}
- print "ok $_";
- if (defined $test_names{$_}) {
+ print "ok $test_num";
+ if (defined $test_names{$test_num}) {
# If TODO is in the test name, make it thus
- my $todo = $test_names{$_} =~ s/TODO\s*//;
- print " $test_names{$_}";
+ my $todo = $test_names{$test_num} =~ s/TODO\s*//;
+ print " $test_names{$test_num}";
print " # TODO" if $todo;
}
print "\n";
$test_num = $final_locales_test_number;
-unless ( $^O eq 'dragonfly' ) {
+unless ( $^O =~ m!^(dragonfly|openbsd|bitrig|mirbsd)$! ) {
# perl #115808
use warnings;
my $warned = 0;
my $above_latin1_case_change_delta; # Same for the specific ords > 255
# that we use
- # We test an ASCII character, which should change case and be tainted;
+ # We test an ASCII character, which should change case;
# a Latin1 character, which shouldn't change case under this C locale,
- # and is tainted.
# an above-Latin1 character that when the case is changed would cross
- # the 255/256 boundary, so doesn't change case and isn't tainted
- # (the \x{149} is one of these, but changes into 2 characters, the
+ # the 255/256 boundary, so doesn't change case
+ # (the \x{149} is one of these, but changes into 2 characters, the
# first one of which doesn't cross the boundary.
# the final one in each list is an above-Latin1 character whose case
- # does change, and shouldn't be tainted. The code below uses its
- # position in its list as a marker to indicate that it, unlike the
- # other code points above ASCII, has a successful case change
+ # does change. The code below uses its position in its list as a
+ # marker to indicate that it, unlike the other code points above
+ # ASCII, has a successful case change
+ #
+ # All casing operations under locale (but not :not_characters) should
+ # taint
if ($function =~ /^u/) {
@list = ("", "a", "\xe0", "\xff", "\x{fb00}", "\x{149}", "\x{101}");
$ascii_case_change_delta = -32;
: "; not encoded in utf8)")
. " should be \"$should_be\", got \"$changed\"");
- # Tainting shouldn't happen for utf8 locales, empty
- # strings, or those characters above 255.
- (! $is_utf8_locale && length($char) > 0 && ord($char) < 256)
+ # Tainting shouldn't happen for use locale :not_character
+ # (a utf8 locale)
+ (! $is_utf8_locale)
? check_taint($changed)
: check_taint_not($changed);
my $F = join(" ", @F);
$F =~ s/(.{50,60}) /$1\n#\t/g;
+ my $details = "";
+ unless ($debug) {
+ $details = "# For more details, rerun, with environment variable PERL_DEBUG_FULL_TEST=1.\n";
+ }
+ elsif ($debug == 1) {
+ $details = "# For even more details, rerun, with environment variable PERL_DEBUG_FULL_TEST=2.\n";
+ }
+
warn
"# The following locales\n#\n",
"#\t", $F, "\n#\n",
"# had problems.\n#\n",
- "# For more details, rerun, with environment variable PERL_DEBUG_FULL_TEST=1.\n";
+ $details;
} else {
warn "# None of your locales were broken.\n";
}