This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
lib/locale.t: Add comments; clarify some test names
[perl5.git] / lib / locale.t
CommitLineData
8ebc5c01
PP
1#!./perl -wT
2
66cbab2c
KW
3# This tests plain 'use locale' and adorned 'use locale ":not_characters"'
4# Because these pragmas are compile time, and I (khw) am trying to test
5# without using 'eval' as much as possible, which might cloud the issue, the
6# crucial parts of the code are duplicated in a block for each pragma.
7
6c2e653d
KW
8# To make a TODO test, add the string 'TODO' to its %test_names value
9
e3a2734b
KW
10binmode STDOUT, ':utf8';
11binmode STDERR, ':utf8';
12
8ebc5c01
PP
13BEGIN {
14 chdir 't' if -d 't';
20822f61 15 @INC = '../lib';
f9cbebe1 16 unshift @INC, '.';
b002077a 17 require Config; import Config;
97a0514d 18 if (!$Config{d_setlocale} || $Config{ccflags} =~ /\bD?NO_LOCALE\b/) {
b002077a
CS
19 print "1..0\n";
20 exit;
21 }
2de3dbcc 22 $| = 1;
8ebc5c01
PP
23}
24
25use strict;
26c1569f 26use feature 'fc';
8ebc5c01 27
108a305e 28my $debug = $ENV{PERL_DEBUG_FULL_TEST} // 0;
284102e8 29
6d5d702a
KW
30# Certain tests have been shown to be problematical for a few locales. Don't
31# fail them unless at least this percentage of the tested locales fail.
32my $acceptable_fold_failure_percentage = 5;
33
db4b7445
A
34use Dumpvalue;
35
36my $dumper = Dumpvalue->new(
37 tick => qq{"},
38 quoteHighBit => 0,
39 unctrl => "quote"
40 );
6be75cd7 41sub debug {
db4b7445
A
42 return unless $debug;
43 my($mess) = join "", @_;
44 chop $mess;
45 print $dumper->stringify($mess,1), "\n";
6be75cd7
JH
46}
47
48sub debugf {
49 printf @_ if $debug;
50}
51
8ebc5c01
PP
52my $have_setlocale = 0;
53eval {
54 require POSIX;
55 import POSIX ':locale_h';
56 $have_setlocale++;
57};
58
6dead956 59# Visual C's CRT goes silly on strings of the form "en_US.ISO8859-1"
f6c6487a 60# and mingw32 uses said silly CRT
3a2d1764
SH
61# This doesn't seem to be an issue any more, at least on Windows XP,
62# so re-enable the tests for Windows XP onwards.
63my $winxp = ($^O eq 'MSWin32' && defined &Win32::GetOSVersion &&
64 join('.', (Win32::GetOSVersion())[1..2]) >= 5.1);
65$have_setlocale = 0 if ((($^O eq 'MSWin32' && !$winxp) || $^O eq 'NetWare') &&
66 $Config{cc} =~ /^(cl|gcc)/i);
6dead956 67
36a42ae7 68# UWIN seems to loop after taint tests, just skip for now
cd19b65c
JH
69$have_setlocale = 0 if ($^O =~ /^uwin/);
70
9a66ea41 71sub LC_ALL ();
8ebc5c01 72
0e053d1e 73$a = 'abc %';
8ebc5c01 74
c213d471
KW
75my $test_num = 0;
76
8ebc5c01 77sub ok {
c213d471 78 my ($result, $message) = @_;
e3a2734b 79 $message = "" unless defined $message;
8ebc5c01
PP
80
81 print 'not ' unless ($result);
c213d471 82 print "ok " . ++$test_num;
e3a2734b
KW
83 print " $message";
84 print "\n";
8ebc5c01
PP
85}
86
87# First we'll do a lot of taint checking for locales.
88# This is the easiest to test, actually, as any locale,
89# even the default locale will taint under 'use locale'.
90
91sub is_tainted { # hello, camel two.
9f1b1f2d 92 no warnings 'uninitialized' ;
8ebc5c01 93 my $dummy;
ba74571d 94 local $@;
8ebc5c01
PP
95 not eval { $dummy = join("", @_), kill 0; 1 }
96}
97
a9b7c637
KW
98sub check_taint ($;$) {
99 my $message_tail = $_[1] // "";
100 $message_tail = ": $message_tail" if $message_tail;
101 ok is_tainted($_[0]), "verify that is tainted$message_tail";
8ebc5c01
PP
102}
103
a9b7c637
KW
104sub check_taint_not ($;$) {
105 my $message_tail = $_[1] // "";
106 $message_tail = ": $message_tail" if $message_tail;
107 ok((not is_tainted($_[0])), "verify that isn't tainted$message_tail");
8ebc5c01
PP
108}
109
bf3cd0e6
KW
110"\tb\t" =~ /^m?(\s)(.*)\1$/;
111check_taint_not $&, "not tainted outside 'use locale'";
112;
113
8ebc5c01
PP
114use locale; # engage locale and therefore locale taint.
115
36a42ae7 116check_taint_not $a;
8ebc5c01 117
36a42ae7
KW
118check_taint uc($a);
119check_taint "\U$a";
120check_taint ucfirst($a);
121check_taint "\u$a";
122check_taint lc($a);
26c1569f 123check_taint fc($a);
36a42ae7 124check_taint "\L$a";
26c1569f 125check_taint "\F$a";
36a42ae7
KW
126check_taint lcfirst($a);
127check_taint "\l$a";
8ebc5c01 128
36a42ae7
KW
129check_taint_not sprintf('%e', 123.456);
130check_taint_not sprintf('%f', 123.456);
131check_taint_not sprintf('%g', 123.456);
132check_taint_not sprintf('%d', 123.456);
133check_taint_not sprintf('%x', 123.456);
8ebc5c01
PP
134
135$_ = $a; # untaint $_
136
137$_ = uc($a); # taint $_
138
36a42ae7 139check_taint $_;
8ebc5c01
PP
140
141/(\w)/; # taint $&, $`, $', $+, $1.
36a42ae7
KW
142check_taint $&;
143check_taint $`;
144check_taint $';
145check_taint $+;
146check_taint $1;
147check_taint_not $2;
8ebc5c01
PP
148
149/(.)/; # untaint $&, $`, $', $+, $1.
36a42ae7
KW
150check_taint_not $&;
151check_taint_not $`;
152check_taint_not $';
153check_taint_not $+;
154check_taint_not $1;
155check_taint_not $2;
8ebc5c01
PP
156
157/(\W)/; # taint $&, $`, $', $+, $1.
36a42ae7
KW
158check_taint $&;
159check_taint $`;
160check_taint $';
161check_taint $+;
162check_taint $1;
163check_taint_not $2;
8ebc5c01
PP
164
165/(\s)/; # taint $&, $`, $', $+, $1.
36a42ae7
KW
166check_taint $&;
167check_taint $`;
168check_taint $';
169check_taint $+;
170check_taint $1;
171check_taint_not $2;
8ebc5c01
PP
172
173/(\S)/; # taint $&, $`, $', $+, $1.
36a42ae7
KW
174check_taint $&;
175check_taint $`;
176check_taint $';
177check_taint $+;
178check_taint $1;
179check_taint_not $2;
8ebc5c01
PP
180
181$_ = $a; # untaint $_
182
36a42ae7 183check_taint_not $_;
8ebc5c01
PP
184
185/(b)/; # this must not taint
36a42ae7
KW
186check_taint_not $&;
187check_taint_not $`;
188check_taint_not $';
189check_taint_not $+;
190check_taint_not $1;
191check_taint_not $2;
8ebc5c01
PP
192
193$_ = $a; # untaint $_
194
36a42ae7 195check_taint_not $_;
8ebc5c01
PP
196
197$b = uc($a); # taint $b
198s/(.+)/$b/; # this must taint only the $_
199
36a42ae7
KW
200check_taint $_;
201check_taint_not $&;
202check_taint_not $`;
203check_taint_not $';
204check_taint_not $+;
205check_taint_not $1;
206check_taint_not $2;
8ebc5c01
PP
207
208$_ = $a; # untaint $_
209
210s/(.+)/b/; # this must not taint
36a42ae7
KW
211check_taint_not $_;
212check_taint_not $&;
213check_taint_not $`;
214check_taint_not $';
215check_taint_not $+;
216check_taint_not $1;
217check_taint_not $2;
8ebc5c01
PP
218
219$b = $a; # untaint $b
220
221($b = $a) =~ s/\w/$&/;
36a42ae7
KW
222check_taint $b; # $b should be tainted.
223check_taint_not $a; # $a should be not.
8ebc5c01
PP
224
225$_ = $a; # untaint $_
226
227s/(\w)/\l$1/; # this must taint
36a42ae7
KW
228check_taint $_;
229check_taint $&;
230check_taint $`;
231check_taint $';
232check_taint $+;
233check_taint $1;
234check_taint_not $2;
8ebc5c01
PP
235
236$_ = $a; # untaint $_
237
238s/(\w)/\L$1/; # this must taint
36a42ae7
KW
239check_taint $_;
240check_taint $&;
241check_taint $`;
242check_taint $';
243check_taint $+;
244check_taint $1;
245check_taint_not $2;
8ebc5c01
PP
246
247$_ = $a; # untaint $_
248
249s/(\w)/\u$1/; # this must taint
36a42ae7
KW
250check_taint $_;
251check_taint $&;
252check_taint $`;
253check_taint $';
254check_taint $+;
255check_taint $1;
256check_taint_not $2;
8ebc5c01
PP
257
258$_ = $a; # untaint $_
259
260s/(\w)/\U$1/; # this must taint
36a42ae7
KW
261check_taint $_;
262check_taint $&;
263check_taint $`;
264check_taint $';
265check_taint $+;
266check_taint $1;
267check_taint_not $2;
8ebc5c01
PP
268
269# After all this tainting $a should be cool.
270
36a42ae7 271check_taint_not $a;
8ebc5c01 272
66cbab2c
KW
273{ # This is just the previous tests copied here with a different
274 # compile-time pragma.
275
276 use locale ':not_characters'; # engage restricted locale with different
277 # tainting rules
278
279 check_taint_not $a;
280
281 check_taint_not uc($a);
282 check_taint_not "\U$a";
283 check_taint_not ucfirst($a);
284 check_taint_not "\u$a";
285 check_taint_not lc($a);
26c1569f 286 check_taint_not fc($a);
66cbab2c 287 check_taint_not "\L$a";
26c1569f 288 check_taint_not "\F$a";
66cbab2c
KW
289 check_taint_not lcfirst($a);
290 check_taint_not "\l$a";
291
292 check_taint_not sprintf('%e', 123.456);
293 check_taint_not sprintf('%f', 123.456);
294 check_taint_not sprintf('%g', 123.456);
295 check_taint_not sprintf('%d', 123.456);
296 check_taint_not sprintf('%x', 123.456);
297
298 $_ = $a; # untaint $_
299
300 $_ = uc($a); # taint $_
301
302 check_taint_not $_;
303
304 /(\w)/; # taint $&, $`, $', $+, $1.
305 check_taint_not $&;
306 check_taint_not $`;
307 check_taint_not $';
308 check_taint_not $+;
309 check_taint_not $1;
310 check_taint_not $2;
311
312 /(.)/; # untaint $&, $`, $', $+, $1.
313 check_taint_not $&;
314 check_taint_not $`;
315 check_taint_not $';
316 check_taint_not $+;
317 check_taint_not $1;
318 check_taint_not $2;
319
320 /(\W)/; # taint $&, $`, $', $+, $1.
321 check_taint_not $&;
322 check_taint_not $`;
323 check_taint_not $';
324 check_taint_not $+;
325 check_taint_not $1;
326 check_taint_not $2;
327
328 /(\s)/; # taint $&, $`, $', $+, $1.
329 check_taint_not $&;
330 check_taint_not $`;
331 check_taint_not $';
332 check_taint_not $+;
333 check_taint_not $1;
334 check_taint_not $2;
335
336 /(\S)/; # taint $&, $`, $', $+, $1.
337 check_taint_not $&;
338 check_taint_not $`;
339 check_taint_not $';
340 check_taint_not $+;
341 check_taint_not $1;
342 check_taint_not $2;
343
344 $_ = $a; # untaint $_
345
346 check_taint_not $_;
347
348 /(b)/; # this must not taint
349 check_taint_not $&;
350 check_taint_not $`;
351 check_taint_not $';
352 check_taint_not $+;
353 check_taint_not $1;
354 check_taint_not $2;
355
356 $_ = $a; # untaint $_
357
358 check_taint_not $_;
359
360 $b = uc($a); # taint $b
361 s/(.+)/$b/; # this must taint only the $_
362
363 check_taint_not $_;
364 check_taint_not $&;
365 check_taint_not $`;
366 check_taint_not $';
367 check_taint_not $+;
368 check_taint_not $1;
369 check_taint_not $2;
370
371 $_ = $a; # untaint $_
372
373 s/(.+)/b/; # this must not taint
374 check_taint_not $_;
375 check_taint_not $&;
376 check_taint_not $`;
377 check_taint_not $';
378 check_taint_not $+;
379 check_taint_not $1;
380 check_taint_not $2;
381
382 $b = $a; # untaint $b
383
384 ($b = $a) =~ s/\w/$&/;
385 check_taint_not $b; # $b should be tainted.
386 check_taint_not $a; # $a should be not.
387
388 $_ = $a; # untaint $_
389
390 s/(\w)/\l$1/; # this must taint
391 check_taint_not $_;
392 check_taint_not $&;
393 check_taint_not $`;
394 check_taint_not $';
395 check_taint_not $+;
396 check_taint_not $1;
397 check_taint_not $2;
398
399 $_ = $a; # untaint $_
400
401 s/(\w)/\L$1/; # this must taint
402 check_taint_not $_;
403 check_taint_not $&;
404 check_taint_not $`;
405 check_taint_not $';
406 check_taint_not $+;
407 check_taint_not $1;
408 check_taint_not $2;
409
410 $_ = $a; # untaint $_
411
412 s/(\w)/\u$1/; # this must taint
413 check_taint_not $_;
414 check_taint_not $&;
415 check_taint_not $`;
416 check_taint_not $';
417 check_taint_not $+;
418 check_taint_not $1;
419 check_taint_not $2;
420
421 $_ = $a; # untaint $_
422
423 s/(\w)/\U$1/; # this must taint
424 check_taint_not $_;
425 check_taint_not $&;
426 check_taint_not $`;
427 check_taint_not $';
428 check_taint_not $+;
429 check_taint_not $1;
430 check_taint_not $2;
431
432 # After all this tainting $a should be cool.
433
434 check_taint_not $a;
435}
436
437# Here are in scope of 'use locale'
438
8ebc5c01
PP
439# I think we've seen quite enough of taint.
440# Let us do some *real* locale work now,
284102e8 441# unless setlocale() is missing (i.e. minitest).
8ebc5c01 442
fdf053ee
KW
443unless ($have_setlocale) {
444 print "1..$test_num\n";
445 exit;
446}
8ebc5c01 447
6cf0b567 448# The test number before our first setlocale()
66330f13 449my $final_without_setlocale = $test_num;
6cf0b567 450
284102e8
JH
451# Find locales.
452
6be75cd7
JH
453debug "# Scanning for locales...\n";
454
455# Note that it's okay that some languages have their native names
456# capitalized here even though that's not "right". They are lowercased
457# anyway later during the scanning process (and besides, some clueless
98dc9551 458# vendor might have them capitalized erroneously anyway).
6be75cd7 459
284102e8 460my $locales = <<EOF;
6be75cd7 461Afrikaans:af:za:1 15
284102e8 462Arabic:ar:dz eg sa:6 arabic8
6be75cd7
JH
463Brezhoneg Breton:br:fr:1 15
464Bulgarski Bulgarian:bg:bg:5
dd8482fc 465Chinese:zh:cn tw:cn.EUC eucCN eucTW euc.CN euc.TW Big5 GB2312 tw.EUC
6be75cd7
JH
466Hrvatski Croatian:hr:hr:2
467Cymraeg Welsh:cy:cy:1 14 15
284102e8 468Czech:cs:cz:2
df8a53a3 469Dansk Danish:da:dk:1 15
6be75cd7 470Nederlands Dutch:nl:be nl:1 15
dd8482fc 471English American British:en:au ca gb ie nz us uk zw:1 15 cp850
6be75cd7
JH
472Esperanto:eo:eo:3
473Eesti Estonian:et:ee:4 6 13
474Suomi Finnish:fi:fi:1 15
475Flamish::fl:1 15
6be75cd7
JH
476Deutsch German:de:at be ch de lu:1 15
477Euskaraz Basque:eu:es fr:1 15
6be75cd7
JH
478Galego Galician:gl:es:1 15
479Ellada Greek:el:gr:7 g8
6be75cd7
JH
480Frysk:fy:nl:1 15
481Greenlandic:kl:gl:4 6
284102e8
JH
482Hebrew:iw:il:8 hebrew8
483Hungarian:hu:hu:2
df8a53a3 484Indonesian:id:id:1 15
6be75cd7
JH
485Gaeilge Irish:ga:IE:1 14 15
486Italiano Italian:it:ch it:1 15
487Nihongo Japanese:ja:jp:euc eucJP jp.EUC sjis
284102e8 488Korean:ko:kr:
6be75cd7
JH
489Latine Latin:la:va:1 15
490Latvian:lv:lv:4 6 13
491Lithuanian:lt:lt:4 6 13
492Macedonian:mk:mk:1 15
493Maltese:mt:mt:3
dd8482fc 494Moldovan:mo:mo:2
df8a53a3 495Norsk Norwegian:no no\@nynorsk nb nn:no:1 15
6be75cd7
JH
496Occitan:oc:es:1 15
497Polski Polish:pl:pl:2
284102e8 498Rumanian:ro:ro:2
a528dad0 499Russki Russian:ru:ru su ua:5 koi8 koi8r KOI8-R koi8u cp1251 cp866
6be75cd7 500Serbski Serbian:sr:yu:5
284102e8 501Slovak:sk:sk:2
6be75cd7 502Slovene Slovenian:sl:si:2
d43ce814
JH
503Sqhip Albanian:sq:sq:1 15
504Svenska Swedish:sv:fi se:1 15
6be75cd7 505Thai:th:th:11 tis620
284102e8 506Turkish:tr:tr:9 turkish8
dd8482fc 507Yiddish:yi::1 15
284102e8
JH
508EOF
509
ee50adbe 510if ($^O eq 'os390') {
dd8482fc 511 # These cause heartburn. Broken locales?
ee50adbe
PP
512 $locales =~ s/Svenska Swedish:sv:fi se:1 15\n//;
513 $locales =~ s/Thai:th:th:11 tis620\n//;
514}
515
ef4a39e5 516sub in_utf8 () { $^H & 0x08 || (${^OPEN} || "") =~ /:utf8/ }
f9cbebe1
JH
517
518if (in_utf8) {
8a6cb2cb 519 require "lib/locale/utf8";
f9cbebe1 520} else {
8a6cb2cb 521 require "lib/locale/latin1";
f9cbebe1
JH
522}
523
284102e8
JH
524my @Locale;
525my $Locale;
526my @Alnum_;
527
284102e8
JH
528sub trylocale {
529 my $locale = shift;
0b9f254b 530 return if grep { $locale eq $_ } @Locale;
e439cacb
KW
531 return unless setlocale(LC_ALL, $locale);
532 my $badutf8;
533 {
534 local $SIG{__WARN__} = sub {
535 $badutf8 = $_[0] =~ /Malformed UTF-8/;
536 };
537 $Locale =~ /UTF-?8/i;
284102e8 538 }
e439cacb
KW
539
540 if ($badutf8) {
541 ok(0, "Locale name contains malformed utf8");
542 return;
543 }
544 push @Locale, $locale;
284102e8 545}
8ebc5c01 546
284102e8
JH
547sub decode_encodings {
548 my @enc;
8ebc5c01 549
284102e8
JH
550 foreach (split(/ /, shift)) {
551 if (/^(\d+)$/) {
552 push @enc, "ISO8859-$1";
553 push @enc, "iso8859$1"; # HP
554 if ($1 eq '1') {
555 push @enc, "roman8"; # HP
556 }
557 } else {
558 push @enc, $_;
dd8482fc 559 push @enc, "$_.UTF-8";
8ebc5c01
PP
560 }
561 }
ee50adbe
PP
562 if ($^O eq 'os390') {
563 push @enc, qw(IBM-037 IBM-819 IBM-1047);
564 }
8ebc5c01 565
284102e8 566 return @enc;
8ebc5c01
PP
567}
568
284102e8
JH
569trylocale("C");
570trylocale("POSIX");
571foreach (0..15) {
572 trylocale("ISO8859-$_");
284102e8 573 trylocale("iso8859$_");
097ee67d
JH
574 trylocale("iso8859-$_");
575 trylocale("iso_8859_$_");
576 trylocale("isolatin$_");
577 trylocale("isolatin-$_");
578 trylocale("iso_latin_$_");
8ebc5c01
PP
579}
580
645e49ed
JH
581# Sanitize the environment so that we can run the external 'locale'
582# program without the taint mode getting grumpy.
cce5967e
JH
583
584# $ENV{PATH} is special in VMS.
585delete $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};
586
587# Other subversive stuff.
588delete @ENV{qw(IFS CDPATH ENV BASH_ENV)};
dd8482fc 589
21477fb4 590if (-x "/usr/bin/locale" && open(LOCALES, "/usr/bin/locale -a 2>/dev/null|")) {
dd8482fc 591 while (<LOCALES>) {
d281a6ac
NC
592 # It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
593 # ain't great when we're running this testPERL_UNICODE= so that utf8
594 # locales will cause all IO hadles to default to (assume) utf8
595 next unless utf8::valid($_);
dd8482fc
JH
596 chomp;
597 trylocale($_);
284102e8 598 }
dd8482fc 599 close(LOCALES);
a6259068 600} elsif ($^O eq 'VMS' && defined($ENV{'SYS$I18N_LOCALE'}) && -d 'SYS$I18N_LOCALE') {
71e5cbb3 601# The SYS$I18N_LOCALE logical name search list was not present on
a6259068
PP
602# VAX VMS V5.5-12, but was on AXP && VAX VMS V6.2 as well as later versions.
603 opendir(LOCALES, "SYS\$I18N_LOCALE:");
604 while ($_ = readdir(LOCALES)) {
605 chomp;
606 trylocale($_);
607 }
608 close(LOCALES);
87e33296
SP
609} elsif ($^O eq 'openbsd' && -e '/usr/share/locale') {
610
611 # OpenBSD doesn't have a locale executable, so reading /usr/share/locale
612 # is much easier and faster than the last resort method.
613
614 opendir(LOCALES, '/usr/share/locale');
615 while ($_ = readdir(LOCALES)) {
616 chomp;
617 trylocale($_);
618 }
619 close(LOCALES);
dd8482fc
JH
620} else {
621
622 # This is going to be slow.
623
624 foreach my $locale (split(/\n/, $locales)) {
625 my ($locale_name, $language_codes, $country_codes, $encodings) =
626 split(/:/, $locale);
627 my @enc = decode_encodings($encodings);
628 foreach my $loc (split(/ /, $locale_name)) {
629 trylocale($loc);
284102e8 630 foreach my $enc (@enc) {
dd8482fc 631 trylocale("$loc.$enc");
284102e8 632 }
dd8482fc 633 $loc = lc $loc;
284102e8 634 foreach my $enc (@enc) {
dd8482fc
JH
635 trylocale("$loc.$enc");
636 }
637 }
638 foreach my $lang (split(/ /, $language_codes)) {
639 trylocale($lang);
640 foreach my $country (split(/ /, $country_codes)) {
641 my $lc = "${lang}_${country}";
642 trylocale($lc);
643 foreach my $enc (@enc) {
644 trylocale("$lc.$enc");
645 }
646 my $lC = "${lang}_\U${country}";
647 trylocale($lC);
648 foreach my $enc (@enc) {
649 trylocale("$lC.$enc");
650 }
284102e8
JH
651 }
652 }
653 }
654}
4599a1de 655
d43ce814
JH
656setlocale(LC_ALL, "C");
657
86f50d7d 658if ($^O eq 'darwin') {
4373e181 659 # Darwin 8/Mac OS X 10.4 and 10.5 have bad Basque locales: perl bug #35895,
86f50d7d 660 # Apple bug ID# 4139653. It also has a problem in Byelorussian.
4373e181
RGS
661 (my $v) = $Config{osvers} =~ /^(\d+)/;
662 if ($v >= 8 and $v < 10) {
86f50d7d 663 debug "# Skipping eu_ES, be_BY locales -- buggy in Darwin\n";
a5ec937f 664 @Locale = grep ! m/^(eu_ES(?:\..*)?|be_BY\.CP1131)$/, @Locale;
dfa5c78f 665 } elsif ($v < 12) {
a44d0896
NC
666 debug "# Skipping be_BY locales -- buggy in Darwin\n";
667 @Locale = grep ! m/^be_BY\.CP1131$/, @Locale;
a5ec937f 668 }
86f50d7d
DD
669}
670
4599a1de
JH
671@Locale = sort @Locale;
672
887ef7ed
PP
673debug "# Locales =\n";
674for ( @Locale ) {
675 debug "# $_\n";
676}
8ebc5c01 677
284102e8 678my %Problem;
2a680da6
JH
679my %Okay;
680my %Testing;
9445c837 681my @Neoalpha; # Alnums that aren't in the C locale.
c08acc4c 682my %test_names;
284102e8 683
2a680da6 684sub tryneoalpha {
15bbd6a2
KW
685 my ($Locale, $i, $test, $message) = @_;
686 $message //= "";
687 $message = " ($message)" if $message;
2a680da6
JH
688 unless ($test) {
689 $Problem{$i}{$Locale} = 1;
15bbd6a2 690 debug "# failed $i with locale '$Locale'$message\n";
2a680da6
JH
691 } else {
692 push @{$Okay{$i}}, $Locale;
693 }
694}
695
c4093d7d
KW
696my $first_locales_test_number = $final_without_setlocale + 1;
697my $locales_test_number;
698my $not_necessarily_a_problem_test_number;
6d5d702a
KW
699my $first_casing_test_number;
700my $final_casing_test_number;
c4093d7d
KW
701my %setlocale_failed; # List of locales that setlocale() didn't work on
702
284102e8 703foreach $Locale (@Locale) {
c4093d7d 704 $locales_test_number = $first_locales_test_number - 1;
284102e8 705 debug "# Locale = $Locale\n";
284102e8
JH
706
707 unless (setlocale(LC_ALL, $Locale)) {
c4093d7d 708 $setlocale_failed{$Locale} = $Locale;
284102e8 709 next;
8ebc5c01 710 }
8ebc5c01 711
66cbab2c
KW
712 # We test UTF-8 locales only under ':not_characters'; otherwise they have
713 # documented deficiencies. Non- UTF-8 locales are tested only under plain
714 # 'use locale', as otherwise we would have to convert everything in them
715 # to Unicode.
716 my $is_utf8_locale = $Locale =~ /UTF-?8/i;
717
95eaa1bf
KW
718 my %UPPER = (); # All alpha X for which uc(X) == X and lc(X) != X
719 my %lower = (); # All alpha X for which lc(X) == X and uc(X) != X
720 my %BoThCaSe = (); # All alpha X for which uc(X) == lc(X) == X
66cbab2c
KW
721
722 if (! $is_utf8_locale) {
723 use locale;
71e5cbb3 724 @Alnum_ = sort grep /\w/, map { chr } 0..255;
f07538a5 725
71e5cbb3 726 debug "# w = ", join("",@Alnum_), "\n";
e5272a46 727
71e5cbb3
KW
728 # Sieve the uppercase and the lowercase.
729
730 for (@Alnum_) {
731 if (/[^\d_]/) { # skip digits and the _
732 if (uc($_) eq $_) {
733 $UPPER{$_} = $_;
734 }
735 if (lc($_) eq $_) {
736 $lower{$_} = $_;
737 }
738 }
739 }
66cbab2c
KW
740 }
741 else {
742 use locale ':not_characters';
743 @Alnum_ = sort grep /\w/, map { chr } 0..255;
744 debug "# w = ", join("",@Alnum_), "\n";
745 for (@Alnum_) {
746 if (/[^\d_]/) { # skip digits and the _
747 if (uc($_) eq $_) {
748 $UPPER{$_} = $_;
749 }
750 if (lc($_) eq $_) {
751 $lower{$_} = $_;
752 }
753 }
754 }
755 }
284102e8 756 foreach (keys %UPPER) {
097ee67d 757 $BoThCaSe{$_}++ if exists $lower{$_};
284102e8
JH
758 }
759 foreach (keys %lower) {
097ee67d 760 $BoThCaSe{$_}++ if exists $UPPER{$_};
284102e8 761 }
097ee67d 762 foreach (keys %BoThCaSe) {
284102e8
JH
763 delete $UPPER{$_};
764 delete $lower{$_};
765 }
766
db4b7445
A
767 debug "# UPPER = ", join("", sort keys %UPPER ), "\n";
768 debug "# lower = ", join("", sort keys %lower ), "\n";
769 debug "# BoThCaSe = ", join("", sort keys %BoThCaSe), "\n";
284102e8 770
baa71cfd 771 my @failures;
3da38613 772 my @fold_failures;
baa71cfd
KW
773 foreach my $x (sort keys %UPPER) {
774 my $ok;
3da38613 775 my $fold_ok;
baa71cfd
KW
776 if ($is_utf8_locale) {
777 use locale ':not_characters';
778 $ok = $x =~ /[[:upper:]]/;
3da38613 779 $fold_ok = $x =~ /[[:lower:]]/i;
baa71cfd
KW
780 }
781 else {
782 use locale;
783 $ok = $x =~ /[[:upper:]]/;
3da38613 784 $fold_ok = $x =~ /[[:lower:]]/i;
baa71cfd
KW
785 }
786 push @failures, $x unless $ok;
3da38613 787 push @fold_failures, $x unless $fold_ok;
baa71cfd
KW
788 }
789 my $message = "";
790 $locales_test_number++;
6d5d702a 791 $first_casing_test_number = $locales_test_number;
95eaa1bf 792 $test_names{$locales_test_number} = 'Verify that /[[:upper:]]/ matches all alpha X for which uc(X) == X and lc(X) != X';
baa71cfd
KW
793 $message = 'Failed for ' . join ", ", @failures if @failures;
794 tryneoalpha($Locale, $locales_test_number, scalar @failures == 0, $message);
6d5d702a 795
3da38613
KW
796 $message = "";
797 $locales_test_number++;
6d5d702a 798
95eaa1bf 799 $test_names{$locales_test_number} = 'Verify that /[[:lower:]]/i matches all alpha X for which uc(X) == X and lc(X) != X';
3da38613
KW
800 $message = 'Failed for ' . join ", ", @fold_failures if @fold_failures;
801 tryneoalpha($Locale, $locales_test_number, scalar @fold_failures == 0, $message);
baa71cfd
KW
802
803 $message = "";
804 undef @failures;
3da38613 805 undef @fold_failures;
baa71cfd
KW
806
807 foreach my $x (sort keys %lower) {
808 my $ok;
3da38613 809 my $fold_ok;
baa71cfd
KW
810 if ($is_utf8_locale) {
811 use locale ':not_characters';
812 $ok = $x =~ /[[:lower:]]/;
3da38613 813 $fold_ok = $x =~ /[[:upper:]]/i;
baa71cfd
KW
814 }
815 else {
816 use locale;
817 $ok = $x =~ /[[:lower:]]/;
3da38613 818 $fold_ok = $x =~ /[[:upper:]]/i;
baa71cfd
KW
819 }
820 push @failures, $x unless $ok;
3da38613 821 push @fold_failures, $x unless $fold_ok;
baa71cfd
KW
822 }
823
824 $locales_test_number++;
95eaa1bf 825 $test_names{$locales_test_number} = 'Verify that /[[:lower:]]/ matches all alpha X for which lc(X) == X and uc(X) != X';
baa71cfd
KW
826 $message = 'Failed for ' . join ", ", @failures if @failures;
827 tryneoalpha($Locale, $locales_test_number, scalar @failures == 0, $message);
828 $message = "";
3da38613 829 $locales_test_number++;
6d5d702a 830 $final_casing_test_number = $locales_test_number;
95eaa1bf 831 $test_names{$locales_test_number} = 'Verify that /[[:upper:]]/i matches all alpha X for which lc(X) == X and uc(X) != X';
3dd9aa8b 832 $message = 'Failed for ' . join ", ", @fold_failures if @fold_failures;
3da38613 833 tryneoalpha($Locale, $locales_test_number, scalar @fold_failures == 0, $message);
baa71cfd 834
9445c837
KW
835 { # Find the alphabetic characters that are not considered alphabetics
836 # in the default (C) locale.
8ebc5c01 837
284102e8 838 no locale;
71e5cbb3 839
284102e8
JH
840 @Neoalpha = ();
841 for (keys %UPPER, keys %lower) {
842 push(@Neoalpha, $_) if (/\W/);
843 }
8ebc5c01 844 }
8ebc5c01 845
284102e8 846 @Neoalpha = sort @Neoalpha;
8ebc5c01 847
db4b7445 848 debug "# Neoalpha = ", join("",@Neoalpha), "\n";
8ebc5c01 849
871d8451
KW
850 my $first_Neoalpha_test_number = $locales_test_number + 1;
851 my $final_Neoalpha_test_number = $first_Neoalpha_test_number + 3;
284102e8
JH
852 if (@Neoalpha == 0) {
853 # If we have no Neoalphas the remaining tests are no-ops.
871d8451 854 debug "# no Neoalpha, skipping tests $first_Neoalpha_test_number..$final_Neoalpha_test_number for locale '$Locale'\n";
c4093d7d 855 foreach ($locales_test_number+1..$final_Neoalpha_test_number) {
a88c3d7c 856 push @{$Okay{$_}}, $Locale;
c4093d7d 857 $locales_test_number++;
a88c3d7c 858 }
6be75cd7 859 } else {
8ebc5c01 860
6be75cd7 861 # Test \w.
71e5cbb3 862
ef4a39e5 863 my $word = join('', @Neoalpha);
8ebc5c01 864
95eaa1bf
KW
865 # This test is likely pointless, as everything
866 # matched \w in the first place.
c4093d7d 867 ++$locales_test_number;
95eaa1bf 868 $test_names{$locales_test_number} = 'Verify that alphas outside the C locale match \w';
66cbab2c
KW
869 my $ok;
870 if ($is_utf8_locale) {
871 use locale ':not_characters';
872 $ok = $word =~ /^(\w+)$/;
873 }
874 else {
875 # Already in 'use locale'; this tests that exiting scopes works
876 $ok = $word =~ /^(\w+)$/;
877 }
878 tryneoalpha($Locale, $locales_test_number, $ok);
ef4a39e5 879
2a680da6 880 # Cross-check the whole 8-bit character set.
8ebc5c01 881
c4093d7d 882 ++$locales_test_number;
c08acc4c 883 $test_names{$locales_test_number} = 'Verify that \w and \W are mutually exclusive, as are \d, \D; \s, \S';
6be75cd7 884 for (map { chr } 0..255) {
66cbab2c
KW
885 if ($is_utf8_locale) {
886 use locale ':not_characters';
887 $ok = (/\w/ xor /\W/) ||
888 (/\d/ xor /\D/) ||
889 (/\s/ xor /\S/);
890 }
891 else {
892 $ok = (/\w/ xor /\W/) ||
2a680da6 893 (/\d/ xor /\D/) ||
66cbab2c
KW
894 (/\s/ xor /\S/);
895 }
896 tryneoalpha($Locale, $locales_test_number, $ok);
284102e8 897 }
8ebc5c01 898
6be75cd7 899 # Test for read-only scalars' locale vs non-locale comparisons.
284102e8 900
284102e8 901 {
6be75cd7
JH
902 no locale;
903 $a = "qwerty";
66cbab2c
KW
904 if ($is_utf8_locale) {
905 use locale ':not_characters';
906 $ok = ($a cmp "qwerty") == 0;
907 }
908 else {
909 use locale;
910 $ok = ($a cmp "qwerty") == 0;
911 }
912 tryneoalpha($Locale, ++$locales_test_number, $ok);
913 $test_names{$locales_test_number} = 'Verify that cmp works with a read-only scalar; no- vs locale';
8ebc5c01 914 }
8ebc5c01 915
6be75cd7
JH
916 {
917 my ($from, $to, $lesser, $greater,
918 @test, %test, $test, $yes, $no, $sign);
919
c4093d7d 920 ++$locales_test_number;
c08acc4c 921 $test_names{$locales_test_number} = 'Verify that "le", "ne", etc work';
c4093d7d 922 $not_necessarily_a_problem_test_number = $locales_test_number;
6be75cd7
JH
923 for (0..9) {
924 # Select a slice.
925 $from = int(($_*@Alnum_)/10);
926 $to = $from + int(@Alnum_/10);
927 $to = $#Alnum_ if ($to > $#Alnum_);
928 $lesser = join('', @Alnum_[$from..$to]);
929 # Select a slice one character on.
930 $from++; $to++;
931 $to = $#Alnum_ if ($to > $#Alnum_);
932 $greater = join('', @Alnum_[$from..$to]);
66cbab2c
KW
933 if ($is_utf8_locale) {
934 use locale ':not_characters';
935 ($yes, $no, $sign) = ($lesser lt $greater
936 ? (" ", "not ", 1)
937 : ("not ", " ", -1));
938 }
939 else {
940 use locale;
71e5cbb3 941 ($yes, $no, $sign) = ($lesser lt $greater
6be75cd7
JH
942 ? (" ", "not ", 1)
943 : ("not ", " ", -1));
66cbab2c 944 }
6be75cd7
JH
945 # all these tests should FAIL (return 0).
946 # Exact lt or gt cannot be tested because
947 # in some locales, say, eacute and E may test equal.
71e5cbb3 948 @test =
6be75cd7
JH
949 (
950 $no.' ($lesser le $greater)', # 1
951 'not ($lesser ne $greater)', # 2
952 ' ($lesser eq $greater)', # 3
953 $yes.' ($lesser ge $greater)', # 4
954 $yes.' ($lesser ge $greater)', # 5
955 $yes.' ($greater le $lesser )', # 7
956 'not ($greater ne $lesser )', # 8
957 ' ($greater eq $lesser )', # 9
958 $no.' ($greater ge $lesser )', # 10
0e053d1e 959 'not (($lesser cmp $greater) == -($sign))' # 11
6be75cd7
JH
960 );
961 @test{@test} = 0 x @test;
962 $test = 0;
284102e8 963 for my $ti (@test) {
66cbab2c
KW
964 if ($is_utf8_locale) {
965 use locale ':not_characters';
966 $test{$ti} = eval $ti;
967 }
968 else {
969 # Already in 'use locale';
71e5cbb3 970 $test{$ti} = eval $ti;
66cbab2c 971 }
6be75cd7 972 $test ||= $test{$ti}
284102e8 973 }
c4093d7d 974 tryneoalpha($Locale, $locales_test_number, $test == 0);
6be75cd7 975 if ($test) {
6be75cd7
JH
976 debug "# lesser = '$lesser'\n";
977 debug "# greater = '$greater'\n";
978 debug "# lesser cmp greater = ",
979 $lesser cmp $greater, "\n";
980 debug "# greater cmp lesser = ",
981 $greater cmp $lesser, "\n";
982 debug "# (greater) from = $from, to = $to\n";
983 for my $ti (@test) {
984 debugf("# %-40s %-4s", $ti,
985 $test{$ti} ? 'FAIL' : 'ok');
986 if ($ti =~ /\(\.*(\$.+ +cmp +\$[^\)]+)\.*\)/) {
987 debugf("(%s == %4d)", $1, eval $1);
988 }
989 debug "\n#";
990 }
284102e8 991
6be75cd7
JH
992 last;
993 }
284102e8 994 }
8ebc5c01
PP
995 }
996 }
6be75cd7 997
c4093d7d
KW
998 if ($locales_test_number != $final_Neoalpha_test_number) {
999 die("The delta for \$final_Neoalpha needs to be updated from "
1000 . ($final_Neoalpha_test_number - $first_Neoalpha_test_number)
1001 . " to "
1002 . ($locales_test_number - $first_Neoalpha_test_number)
1003 );
1004 }
1005
66cbab2c
KW
1006 my $ok1;
1007 my $ok2;
1008 my $ok3;
1009 my $ok4;
1010 my $ok5;
1011 my $ok6;
1012 my $ok7;
1013 my $ok8;
1014 my $ok9;
1015 my $ok10;
1016 my $ok11;
1017 my $ok12;
1018 my $ok13;
1500bd91 1019 my $ok14;
28acfe03
KW
1020 my $ok15;
1021 my $ok16;
66cbab2c
KW
1022
1023 my $c;
1024 my $d;
1025 my $e;
1026 my $f;
1027 my $g;
1028
1029 if (! $is_utf8_locale) {
71e5cbb3 1030 use locale;
6be75cd7 1031
71e5cbb3 1032 my ($x, $y) = (1.23, 1.23);
6be75cd7 1033
71e5cbb3
KW
1034 $a = "$x";
1035 printf ''; # printf used to reset locale to "C"
1036 $b = "$y";
1037 $ok1 = $a eq $b;
6be75cd7 1038
71e5cbb3
KW
1039 $c = "$x";
1040 my $z = sprintf ''; # sprintf used to reset locale to "C"
1041 $d = "$y";
1042 $ok2 = $c eq $d;
1043 {
66cbab2c 1044
71e5cbb3
KW
1045 use warnings;
1046 my $w = 0;
1047 local $SIG{__WARN__} =
1048 sub {
1049 print "# @_\n";
1050 $w++;
1051 };
6be75cd7 1052
71e5cbb3
KW
1053 # The == (among other ops) used to warn for locales
1054 # that had something else than "." as the radix character.
6be75cd7 1055
71e5cbb3
KW
1056 $ok3 = $c == 1.23;
1057 $ok4 = $c == $x;
1058 $ok5 = $c == $d;
1059 {
1060 no locale;
66cbab2c 1061
b79536ea 1062 $e = "$x";
71e5cbb3
KW
1063
1064 $ok6 = $e == 1.23;
1065 $ok7 = $e == $x;
1066 $ok8 = $e == $c;
1067 }
66cbab2c 1068
71e5cbb3
KW
1069 $f = "1.23";
1070 $g = 2.34;
66cbab2c 1071
71e5cbb3
KW
1072 $ok9 = $f == 1.23;
1073 $ok10 = $f == $x;
1074 $ok11 = $f == $c;
1075 $ok12 = abs(($f + $g) - 3.57) < 0.01;
1076 $ok13 = $w == 0;
28acfe03 1077 $ok14 = $ok15 = $ok16 = 1; # Skip for non-utf8 locales
71e5cbb3 1078 }
66cbab2c
KW
1079 }
1080 else {
1081 use locale ':not_characters';
1082
1083 my ($x, $y) = (1.23, 1.23);
1084 $a = "$x";
1085 printf ''; # printf used to reset locale to "C"
1086 $b = "$y";
1087 $ok1 = $a eq $b;
1088
1089 $c = "$x";
1090 my $z = sprintf ''; # sprintf used to reset locale to "C"
1091 $d = "$y";
1092 $ok2 = $c eq $d;
1093 {
1094 use warnings;
1095 my $w = 0;
1096 local $SIG{__WARN__} =
1097 sub {
1098 print "# @_\n";
1099 $w++;
1100 };
1101 $ok3 = $c == 1.23;
1102 $ok4 = $c == $x;
1103 $ok5 = $c == $d;
1104 {
1105 no locale;
b79536ea 1106 $e = "$x";
66cbab2c
KW
1107
1108 $ok6 = $e == 1.23;
1109 $ok7 = $e == $x;
1110 $ok8 = $e == $c;
1111 }
1112
1113 $f = "1.23";
1114 $g = 2.34;
1115
1116 $ok9 = $f == 1.23;
1117 $ok10 = $f == $x;
1118 $ok11 = $f == $c;
1119 $ok12 = abs(($f + $g) - 3.57) < 0.01;
1120 $ok13 = $w == 0;
1500bd91
KW
1121
1122 # Look for non-ASCII error messages, and verify that the first
1123 # such is in UTF-8 (the others almost certainly will be like the
1124 # first).
1125 $ok14 = 1;
1126 foreach my $err (keys %!) {
1127 use Errno;
1128 $! = eval "&Errno::$err"; # Convert to strerror() output
1129 my $strerror = "$!";
1130 if ("$strerror" =~ /\P{ASCII}/) {
1131 my $utf8_strerror = $strerror;
1132 utf8::upgrade($utf8_strerror);
1133
1134 # If $! was already in UTF-8, the upgrade was a no-op;
1135 # otherwise they will be different byte strings.
1136 use bytes;
1137 $ok14 = $utf8_strerror eq $strerror;
1138 last;
1139 }
1140 }
28acfe03
KW
1141
1142 # Similarly, we verify that a non-ASCII radix is in UTF-8. This
1143 # also catches if there is a disparity between sprintf and
1144 # stringification.
1145
1146 my $string_g = "$g";
1147
1148 my $utf8_string_g = "$g";
1149 utf8::upgrade($utf8_string_g);
1150
1151 my $utf8_sprintf_g = sprintf("%g", $g);
1152 utf8::upgrade($utf8_sprintf_g);
1153 use bytes;
1154 $ok15 = $utf8_string_g eq $string_g;
1155 $ok16 = $utf8_sprintf_g eq $string_g;
66cbab2c
KW
1156 }
1157 }
1158
1159 tryneoalpha($Locale, ++$locales_test_number, $ok1);
1160 $test_names{$locales_test_number} = 'Verify that an intervening printf doesn\'t change assignment results';
1161 my $first_a_test = $locales_test_number;
1162
1163 debug "# $first_a_test..$locales_test_number: \$a = $a, \$b = $b, Locale = $Locale\n";
1164
1165 tryneoalpha($Locale, ++$locales_test_number, $ok2);
1166 $test_names{$locales_test_number} = 'Verify that an intervening sprintf doesn\'t change assignment results';
1167
1168 my $first_c_test = $locales_test_number;
1169
71e5cbb3
KW
1170 tryneoalpha($Locale, ++$locales_test_number, $ok3);
1171 $test_names{$locales_test_number} = 'Verify that a different locale radix works when doing "==" with a constant';
6be75cd7 1172
71e5cbb3
KW
1173 tryneoalpha($Locale, ++$locales_test_number, $ok4);
1174 $test_names{$locales_test_number} = 'Verify that a different locale radix works when doing "==" with a scalar';
66cbab2c 1175
71e5cbb3
KW
1176 tryneoalpha($Locale, ++$locales_test_number, $ok5);
1177 $test_names{$locales_test_number} = 'Verify that a different locale radix works when doing "==" with a scalar and an intervening sprintf';
66cbab2c 1178
71e5cbb3 1179 debug "# $first_c_test..$locales_test_number: \$c = $c, \$d = $d, Locale = $Locale\n";
66cbab2c 1180
71e5cbb3 1181 tryneoalpha($Locale, ++$locales_test_number, $ok6);
b79536ea 1182 $test_names{$locales_test_number} = 'Verify that can assign stringified under inner no-locale block';
71e5cbb3 1183 my $first_e_test = $locales_test_number;
6be75cd7 1184
71e5cbb3
KW
1185 tryneoalpha($Locale, ++$locales_test_number, $ok7);
1186 $test_names{$locales_test_number} = 'Verify that "==" with a scalar still works in inner no locale';
66cbab2c 1187
71e5cbb3
KW
1188 tryneoalpha($Locale, ++$locales_test_number, $ok8);
1189 $test_names{$locales_test_number} = 'Verify that "==" with a scalar and an intervening sprintf still works in inner no locale';
c4093d7d 1190
71e5cbb3 1191 debug "# $first_e_test..$locales_test_number: \$e = $e, no locale\n";
2a680da6 1192
71e5cbb3
KW
1193 tryneoalpha($Locale, ++$locales_test_number, $ok9);
1194 $test_names{$locales_test_number} = 'Verify that after a no-locale block, a different locale radix still works when doing "==" with a constant';
1195 my $first_f_test = $locales_test_number;
6be75cd7 1196
71e5cbb3
KW
1197 tryneoalpha($Locale, ++$locales_test_number, $ok10);
1198 $test_names{$locales_test_number} = 'Verify that after a no-locale block, a different locale radix still works when doing "==" with a scalar';
66cbab2c 1199
71e5cbb3
KW
1200 tryneoalpha($Locale, ++$locales_test_number, $ok11);
1201 $test_names{$locales_test_number} = 'Verify that after a no-locale block, a different locale radix still works when doing "==" with a scalar and an intervening sprintf';
906f284f 1202
71e5cbb3
KW
1203 tryneoalpha($Locale, ++$locales_test_number, $ok12);
1204 $test_names{$locales_test_number} = 'Verify that after a no-locale block, a different locale radix can participate in an addition and function call as numeric';
c4093d7d 1205
71e5cbb3
KW
1206 tryneoalpha($Locale, ++$locales_test_number, $ok13);
1207 $test_names{$locales_test_number} = 'Verify that don\'t get warning under "==" even if radix is not a dot';
c4093d7d 1208
1500bd91
KW
1209 tryneoalpha($Locale, ++$locales_test_number, $ok14);
1210 $test_names{$locales_test_number} = 'Verify that non-ASCII UTF-8 error messages are in UTF-8';
1211
28acfe03
KW
1212 tryneoalpha($Locale, ++$locales_test_number, $ok15);
1213 $test_names{$locales_test_number} = 'Verify that a number with a UTF-8 radix has a UTF-8 stringification';
1214
1215 tryneoalpha($Locale, ++$locales_test_number, $ok16);
1216 $test_names{$locales_test_number} = 'Verify that a sprintf of a number with a UTF-8 radix yields UTF-8';
1217
71e5cbb3 1218 debug "# $first_f_test..$locales_test_number: \$f = $f, \$g = $g, back to locale = $Locale\n";
906f284f 1219
26d80d95
LC
1220 # Does taking lc separately differ from taking
1221 # the lc "in-line"? (This was the bug 19990704.002, change #3568.)
1222 # The bug was in the caching of the 'o'-magic.
66cbab2c 1223 if (! $is_utf8_locale) {
2a680da6 1224 use locale;
6be75cd7 1225
2a680da6
JH
1226 sub lcA {
1227 my $lc0 = lc $_[0];
1228 my $lc1 = lc $_[1];
1229 return $lc0 cmp $lc1;
1230 }
6be75cd7 1231
2a680da6
JH
1232 sub lcB {
1233 return lc($_[0]) cmp lc($_[1]);
1234 }
6be75cd7 1235
2a680da6
JH
1236 my $x = "ab";
1237 my $y = "aa";
1238 my $z = "AB";
6be75cd7 1239
c4093d7d 1240 tryneoalpha($Locale, ++$locales_test_number,
2a680da6
JH
1241 lcA($x, $y) == 1 && lcB($x, $y) == 1 ||
1242 lcA($x, $z) == 0 && lcB($x, $z) == 0);
6be75cd7 1243 }
66cbab2c
KW
1244 else {
1245 use locale ':not_characters';
1246
1247 sub lcC {
1248 my $lc0 = lc $_[0];
1249 my $lc1 = lc $_[1];
1250 return $lc0 cmp $lc1;
1251 }
1252
1253 sub lcD {
1254 return lc($_[0]) cmp lc($_[1]);
1255 }
1256
1257 my $x = "ab";
1258 my $y = "aa";
1259 my $z = "AB";
1260
1261 tryneoalpha($Locale, ++$locales_test_number,
1262 lcC($x, $y) == 1 && lcD($x, $y) == 1 ||
1263 lcC($x, $z) == 0 && lcD($x, $z) == 0);
1264 }
1265 $test_names{$locales_test_number} = 'Verify "lc(foo) cmp lc(bar)" is the same as using intermediaries for the cmp';
d8093b23 1266
26d80d95
LC
1267 # Does lc of an UPPER (if different from the UPPER) match
1268 # case-insensitively the UPPER, and does the UPPER match
1269 # case-insensitively the lc of the UPPER. And vice versa.
3ba0e062 1270 {
ef4a39e5
JH
1271 use locale;
1272 no utf8;
1273 my $re = qr/[\[\(\{\*\+\?\|\^\$\\]/;
1274
1275 my @f = ();
c4093d7d 1276 ++$locales_test_number;
c08acc4c 1277 $test_names{$locales_test_number} = 'Verify case insensitive matching works';
f78d9f29 1278 foreach my $x (sort keys %UPPER) {
66cbab2c 1279 if (! $is_utf8_locale) {
71e5cbb3
KW
1280 my $y = lc $x;
1281 next unless uc $y eq $x;
1282 print "# UPPER $x lc $y ",
faf0c248
KW
1283 $x =~ /$y/i ? 1 : 0, " ",
1284 $y =~ /$x/i ? 1 : 0, "\n" if 0;
71e5cbb3
KW
1285 #
1286 # If $x and $y contain regular expression characters
1287 # AND THEY lowercase (/i) to regular expression characters,
1288 # regcomp() will be mightily confused. No, the \Q doesn't
1289 # help here (maybe regex engine internal lowercasing
1290 # is done after the \Q?) An example of this happening is
1291 # the bg_BG (Bulgarian) locale under EBCDIC (OS/390 USS):
1292 # the chr(173) (the "[") is the lowercase of the chr(235).
1293 #
1294 # Similarly losing EBCDIC locales include cs_cz, cs_CZ,
1295 # el_gr, el_GR, en_us.IBM-037 (!), en_US.IBM-037 (!),
1296 # et_ee, et_EE, hr_hr, hr_HR, hu_hu, hu_HU, lt_LT,
1297 # mk_mk, mk_MK, nl_nl.IBM-037, nl_NL.IBM-037,
1298 # pl_pl, pl_PL, ro_ro, ro_RO, ru_ru, ru_RU,
1299 # sk_sk, sk_SK, sl_si, sl_SI, tr_tr, tr_TR.
1300 #
1301 # Similar things can happen even under (bastardised)
1302 # non-EBCDIC locales: in many European countries before the
1303 # advent of ISO 8859-x nationally customised versions of
1304 # ISO 646 were devised, reusing certain punctuation
1305 # characters for modified characters needed by the
1306 # country/language. For example, the "|" might have
1307 # stood for U+00F6 or LATIN SMALL LETTER O WITH DIAERESIS.
1308 #
1309 if ($x =~ $re || $y =~ $re) {
1310 print "# Regex characters in '$x' or '$y', skipping test $locales_test_number for locale '$Locale'\n";
1311 next;
1312 }
1313 # With utf8 both will fail since the locale concept
1314 # of upper/lower does not work well in Unicode.
1315 push @f, $x unless $x =~ /$y/i == $y =~ /$x/i;
26c1569f
KW
1316
1317 # fc is not a locale concept, so Perl uses lc for it.
1318 push @f, $x unless lc $x eq fc $x;
66cbab2c
KW
1319 }
1320 else {
1321 use locale ':not_characters';
1322 my $y = lc $x;
1323 next unless uc $y eq $x;
1324 print "# UPPER $x lc $y ",
faf0c248
KW
1325 $x =~ /$y/i ? 1 : 0, " ",
1326 $y =~ /$x/i ? 1 : 0, "\n" if 0;
66cbab2c
KW
1327
1328 # Here, we can fully test things, unlike plain 'use locale',
1329 # because this form does work well with Unicode
1330 push @f, $x unless $x =~ /$y/i && $y =~ /$x/i;
26c1569f
KW
1331
1332 # The places where Unicode's lc is different from fc are
1333 # skipped here by virtue of the 'next unless uc...' line above
1334 push @f, $x unless lc $x eq fc $x;
66cbab2c 1335 }
c00ff1c7 1336 }
ef4a39e5 1337
f78d9f29 1338 foreach my $x (sort keys %lower) {
66cbab2c 1339 if (! $is_utf8_locale) {
71e5cbb3
KW
1340 my $y = uc $x;
1341 next unless lc $y eq $x;
1342 print "# lower $x uc $y ",
faf0c248
KW
1343 $x =~ /$y/i ? 1 : 0, " ",
1344 $y =~ /$x/i ? 1 : 0, "\n" if 0;
71e5cbb3
KW
1345 if ($x =~ $re || $y =~ $re) { # See above.
1346 print "# Regex characters in '$x' or '$y', skipping test $locales_test_number for locale '$Locale'\n";
1347 next;
1348 }
1349 # With utf8 both will fail since the locale concept
1350 # of upper/lower does not work well in Unicode.
1351 push @f, $x unless $x =~ /$y/i == $y =~ /$x/i;
26c1569f
KW
1352
1353 push @f, $x unless lc $x eq fc $x;
66cbab2c
KW
1354 }
1355 else {
1356 use locale ':not_characters';
1357 my $y = uc $x;
1358 next unless lc $y eq $x;
1359 print "# lower $x uc $y ",
faf0c248
KW
1360 $x =~ /$y/i ? 1 : 0, " ",
1361 $y =~ /$x/i ? 1 : 0, "\n" if 0;
66cbab2c 1362 push @f, $x unless $x =~ /$y/i && $y =~ /$x/i;
26c1569f
KW
1363
1364 push @f, $x unless lc $x eq fc $x;
66cbab2c 1365 }
c00ff1c7 1366 }
c4093d7d 1367 tryneoalpha($Locale, $locales_test_number, @f == 0);
c00ff1c7 1368 if (@f) {
c4093d7d 1369 print "# failed $locales_test_number locale '$Locale' characters @f\n"
c00ff1c7 1370 }
d8093b23 1371 }
78787052
JL
1372
1373 # [perl #109318]
1374 {
1375 my @f = ();
1376 ++$locales_test_number;
1377 $test_names{$locales_test_number} = 'Verify atof with locale radix and negative exponent';
1378
1379 my $radix = POSIX::localeconv()->{decimal_point};
1380 my @nums = (
1381 "3.14e+9", "3${radix}14e+9", "3.14e-9", "3${radix}14e-9",
1382 "-3.14e+9", "-3${radix}14e+9", "-3.14e-9", "-3${radix}14e-9",
1383 );
1384
1385 if (! $is_utf8_locale) {
1386 use locale;
1387 for my $num (@nums) {
1388 push @f, $num
1389 unless sprintf("%g", $num) =~ /3.+14/;
1390 }
1391 }
1392 else {
1393 use locale ':not_characters';
1394 for my $num (@nums) {
1395 push @f, $num
1396 unless sprintf("%g", $num) =~ /3.+14/;
1397 }
1398 }
1399
1400 tryneoalpha($Locale, $locales_test_number, @f == 0);
1401 if (@f) {
1402 print "# failed $locales_test_number locale '$Locale' numbers @f\n"
1403 }
1404 }
8ebc5c01 1405}
284102e8 1406
c4093d7d 1407my $final_locales_test_number = $locales_test_number;
6cf0b567 1408
2a680da6
JH
1409# Recount the errors.
1410
c4093d7d
KW
1411foreach ($first_locales_test_number..$final_locales_test_number) {
1412 if (%setlocale_failed) {
1413 print "not ";
1414 }
1415 elsif ($Problem{$_} || !defined $Okay{$_} || !@{$Okay{$_}}) {
1416 if (defined $not_necessarily_a_problem_test_number
1417 && $_ == $not_necessarily_a_problem_test_number)
1418 {
1419 print "# The failure of test $not_necessarily_a_problem_test_number is not necessarily fatal.\n";
b4e009be 1420 print "# It usually indicates a problem in the environment,\n";
284102e8
JH
1421 print "# not in Perl itself.\n";
1422 }
6d5d702a
KW
1423 if ($Okay{$_} && ($_ >= $first_casing_test_number
1424 && $_ <= $final_casing_test_number))
1425 {
1426 my $percent_fail = int(.5 + (100 * scalar(keys $Problem{$_})
1427 / scalar(@{$Okay{$_}})));
1428 if ($percent_fail < $acceptable_fold_failure_percentage) {
1429 $test_names{$_} .= 'TODO';
1430 print "# ", 100 - $percent_fail, "% of locales pass the following test, so it is likely that the failures\n";
1431 print "# are errors in the locale definitions. The test is marked TODO, as the\n";
1432 print "# problem is not likely to be Perl's\n";
1433 }
1434 }
108a305e
KW
1435 unless ($debug) {
1436 print "#\nFor more details, rerun, with environment variable PERL_DEBUG_FULL_TEST=1\n";
1437 }
284102e8 1438 print "not ";
8ebc5c01 1439 }
c4093d7d 1440 print "ok $_";
6c2e653d
KW
1441 if (defined $test_names{$_}) {
1442 # If TODO is in the test name, make it thus
1443 my $todo = $test_names{$_} =~ s/TODO\s*//;
1444 print " $test_names{$_}";
1445 print " # TODO" if $todo;
1446 }
c4093d7d 1447 print "\n";
8ebc5c01 1448}
fb73857a 1449
2a680da6
JH
1450# Give final advice.
1451
284102e8
JH
1452my $didwarn = 0;
1453
c4093d7d 1454foreach ($first_locales_test_number..$final_locales_test_number) {
284102e8
JH
1455 if ($Problem{$_}) {
1456 my @f = sort keys %{ $Problem{$_} };
1457 my $f = join(" ", @f);
1458 $f =~ s/(.{50,60}) /$1\n#\t/g;
2a680da6
JH
1459 print
1460 "#\n",
1461 "# The locale ", (@f == 1 ? "definition" : "definitions"), "\n#\n",
284102e8
JH
1462 "#\t", $f, "\n#\n",
1463 "# on your system may have errors because the locale test $_\n",
1464 "# failed in ", (@f == 1 ? "that locale" : "those locales"),
1465 ".\n";
2a680da6 1466 print <<EOW;
284102e8
JH
1467#
1468# If your users are not using these locales you are safe for the moment,
1469# but please report this failure first to perlbug\@perl.com using the
1470# perlbug script (as described in the INSTALL file) so that the exact
1471# details of the failures can be sorted out first and then your operating
1472# system supplier can be alerted about these anomalies.
1473#
1474EOW
1475 $didwarn = 1;
fb73857a
PP
1476 }
1477}
774d564b 1478
26d80d95 1479# Tell which locales were okay and which were not.
2a680da6 1480
284102e8 1481if ($didwarn) {
26d80d95 1482 my (@s, @F);
71e5cbb3 1483
284102e8
JH
1484 foreach my $l (@Locale) {
1485 my $p = 0;
c4093d7d
KW
1486 if ($setlocale_failed{$l}) {
1487 $p++;
1488 }
1489 else {
1f5852c9
KW
1490 foreach my $t
1491 ($first_locales_test_number..$final_locales_test_number)
1492 {
1493 $p++ if $Problem{$t}{$l};
1494 }
c4093d7d 1495 }
284102e8 1496 push @s, $l if $p == 0;
9445c837 1497 push @F, $l unless $p == 0;
8ebc5c01 1498 }
71e5cbb3 1499
68d47915
CK
1500 if (@s) {
1501 my $s = join(" ", @s);
1502 $s =~ s/(.{50,60}) /$1\n#\t/g;
1503
1504 warn
1505 "# The following locales\n#\n",
1506 "#\t", $s, "\n#\n",
1507 "# tested okay.\n#\n",
1508 } else {
26d80d95
LC
1509 warn "# None of your locales were fully okay.\n";
1510 }
1511
1512 if (@F) {
1513 my $F = join(" ", @F);
1514 $F =~ s/(.{50,60}) /$1\n#\t/g;
1515
1516 warn
1517 "# The following locales\n#\n",
0e053d1e 1518 "#\t", $F, "\n#\n",
26d80d95
LC
1519 "# had problems.\n#\n",
1520 } else {
1521 warn "# None of your locales were broken.\n";
68d47915 1522 }
8ebc5c01 1523}
90248788 1524
c4093d7d 1525$test_num = $final_locales_test_number;
c213d471 1526
fbd840df
KW
1527{ # perl #115808
1528 use warnings;
1529 my $warned = 0;
1530 local $SIG{__WARN__} = sub {
1531 $warned = $_[0] =~ /uninitialized/;
1532 };
1533 my $z = "y" . setlocale(&POSIX::LC_ALL, "xyzzy");
1534 ok($warned, "variable set to setlocale(BAD LOCALE) is considered uninitialized");
1535}
1536
094a2f8c 1537# Test that tainting and case changing works on utf8 strings. These tests are
1f5852c9
KW
1538# placed last to avoid disturbing the hard-coded test numbers that existed at
1539# the time these were added above this in this file.
0099bb8d
KW
1540# This also tests that locale overrides unicode_strings in the same scope for
1541# non-utf8 strings.
094a2f8c
KW
1542setlocale(LC_ALL, "C");
1543{
1544 use locale;
0099bb8d 1545 use feature 'unicode_strings';
094a2f8c 1546
26c1569f 1547 foreach my $function ("uc", "ucfirst", "lc", "lcfirst", "fc") {
094a2f8c
KW
1548 my @list; # List of code points to test for $function
1549
1550 # Used to calculate the changed case for ASCII characters by using the
1551 # ord, instead of using one of the functions under test.
1552 my $ascii_case_change_delta;
1553 my $above_latin1_case_change_delta; # Same for the specific ords > 255
1554 # that we use
1555
1556 # We test an ASCII character, which should change case and be tainted;
1557 # a Latin1 character, which shouldn't change case under this C locale,
1558 # and is tainted.
1559 # an above-Latin1 character that when the case is changed would cross
1560 # the 255/256 boundary, so doesn't change case and isn't tainted
1561 # (the \x{149} is one of these, but changes into 2 characters, the
1562 # first one of which doesn't cross the boundary.
1563 # the final one in each list is an above-Latin1 character whose case
1564 # does change, and shouldn't be tainted. The code below uses its
1565 # position in its list as a marker to indicate that it, unlike the
1566 # other code points above ASCII, has a successful case change
1567 if ($function =~ /^u/) {
094a2f8c
KW
1568 @list = ("", "a", "\xe0", "\xff", "\x{fb00}", "\x{149}", "\x{101}");
1569 $ascii_case_change_delta = -32;
1570 $above_latin1_case_change_delta = -1;
1571 }
1572 else {
1ca267a5 1573 @list = ("", "A", "\xC0", "\x{17F}", "\x{100}");
094a2f8c
KW
1574 $ascii_case_change_delta = +32;
1575 $above_latin1_case_change_delta = +1;
1576 }
66cbab2c 1577 foreach my $is_utf8_locale (0 .. 1) {
71e5cbb3
KW
1578 foreach my $j (0 .. $#list) {
1579 my $char = $list[$j];
0099bb8d
KW
1580
1581 for my $encoded_in_utf8 (0 .. 1) {
faf0c248
KW
1582 my $should_be;
1583 my $changed;
1584 if (! $is_utf8_locale) {
1585 $should_be = ($j == $#list)
1586 ? chr(ord($char) + $above_latin1_case_change_delta)
1587 : (length $char == 0 || ord($char) > 127)
1588 ? $char
1589 : chr(ord($char) + $ascii_case_change_delta);
1590
1591 # This monstrosity is in order to avoid using an eval,
1592 # which might perturb the results
1593 $changed = ($function eq "uc")
1594 ? uc($char)
1595 : ($function eq "ucfirst")
1596 ? ucfirst($char)
1597 : ($function eq "lc")
1598 ? lc($char)
1599 : ($function eq "lcfirst")
1600 ? lcfirst($char)
26c1569f
KW
1601 : ($function eq "fc")
1602 ? fc($char)
faf0c248
KW
1603 : die("Unexpected function \"$function\"");
1604 }
1605 else {
1606 {
1607 no locale;
71e5cbb3 1608
faf0c248
KW
1609 # For utf8-locales the case changing functions
1610 # should work just like they do outside of locale.
1611 # Can use eval here because not testing it when
1612 # not in locale.
1613 $should_be = eval "$function('$char')";
1614 die "Unexpected eval error $@ from 'eval \"$function('$char')\"'" if $@;
71e5cbb3 1615
faf0c248
KW
1616 }
1617 use locale ':not_characters';
1618 $changed = ($function eq "uc")
1619 ? uc($char)
1620 : ($function eq "ucfirst")
1621 ? ucfirst($char)
1622 : ($function eq "lc")
1623 ? lc($char)
1624 : ($function eq "lcfirst")
1625 ? lcfirst($char)
26c1569f
KW
1626 : ($function eq "fc")
1627 ? fc($char)
faf0c248 1628 : die("Unexpected function \"$function\"");
71e5cbb3 1629 }
faf0c248
KW
1630 ok($changed eq $should_be,
1631 "$function(\"$char\") in C locale "
1632 . (($is_utf8_locale)
1633 ? "(use locale ':not_characters'"
1634 : "(use locale")
1635 . (($encoded_in_utf8)
1636 ? "; encoded in utf8)"
1637 : "; not encoded in utf8)")
1638 . " should be \"$should_be\", got \"$changed\"");
1639
1640 # Tainting shouldn't happen for utf8 locales, empty
1641 # strings, or those characters above 255.
1642 (! $is_utf8_locale && length($char) > 0 && ord($char) < 256)
1643 ? check_taint($changed)
1644 : check_taint_not($changed);
1645
1646 # Use UTF-8 next time through the loop
1647 utf8::upgrade($char);
0099bb8d 1648 }
66cbab2c 1649 }
094a2f8c
KW
1650 }
1651 }
1652}
1653
fdf053ee 1654print "1..$test_num\n";
906f284f 1655
90248788 1656# eof