3 # This is a home for regular expression tests that don't fit into
4 # the format supported by re/regexp.t. If you want to add a test
5 # that does fit that format, add it to re/re_tests, not here.
18 require Config; import Config;
19 require './test.pl'; require './charset_tools.pl';
20 require './loc_tools.pl';
21 set_up_inc('../lib', '.', '../ext/re');
23 skip_all('no re module') unless defined &DynaLoader::boot_DynaLoader;
24 skip_all_without_unicode_tables();
26 plan tests => 844; # Update this when adding/deleting tests.
28 run_tests() unless caller;
37 (my $x_pretty = $x) =~ s/\n/\\n/g;
39 ok $x =~ /^abc/, qq ["$x_pretty" =~ /^abc/];
40 ok $x !~ /^def/, qq ["$x_pretty" !~ /^def/];
42 # used to be a test for $*
43 ok $x =~ /^def/m, qq ["$x_pretty" =~ /^def/m];
45 ok(!($x =~ /^xxx/), qq ["$x_pretty" =~ /^xxx/]);
46 ok(!($x !~ /^abc/), qq ["$x_pretty" !~ /^abc/]);
48 ok $x =~ /def/, qq ["$x_pretty" =~ /def/];
49 ok(!($x !~ /def/), qq ["$x_pretty" !~ /def/]);
51 ok $x !~ /.def/, qq ["$x_pretty" !~ /.def/];
52 ok(!($x =~ /.def/), qq ["$x_pretty" =~ /.def/]);
54 ok $x =~ /\ndef/, qq ["$x_pretty" =~ /\\ndef/];
55 ok(!($x !~ /\ndef/), qq ["$x_pretty" !~ /\\ndef/]);
60 ok /^([0-9][0-9]*)/, qq [\$_ = '$_'; /^([0-9][0-9]*)/];
65 ok /(a*b*)(c*)/ && $1 eq 'aaabbb' && $2 eq 'ccc',
66 qq [\$_ = '$_'; /(a*b*)(c*)/];
67 ok /(a+b+c+)/ && $1 eq 'aaabbbccc', qq [\$_ = '$_'; /(a+b+c+)/];
68 unlike($_, qr/a+b?c+/, qq [\$_ = '$_'; /a+b?c+/]);
71 ok /a+b?c+/, qq [\$_ = '$_'; /a+b?c+/];
72 ok /a*b?c*/, qq [\$_ = '$_'; /a*b?c*/];
75 ok /a*b?c*/, qq [\$_ = '$_'; /a*b?c*/];
76 unlike($_, qr/a*b+c*/, qq [\$_ = '$_'; /a*b+c*/]);
79 ok /bcd|xyz/, qq [\$_ = '$_'; /bcd|xyz/];
80 ok /xyz|bcd/, qq [\$_ = '$_'; /xyz|bcd/];
81 ok m|bc/*d|, qq [\$_ = '$_'; m|bc/*d|];
82 ok /^$_$/, qq [\$_ = '$_'; /^\$_\$/];
86 # used to be a test for $*
87 ok "ab\ncd\n" =~ /^cd/m, q ["ab\ncd\n" =~ /^cd/m];
91 our %XXX = map {($_ => $_)} 123, 234, 345;
93 our @XXX = ('ok 1','not ok 1', 'ok 2','not ok 2','not ok 3');
94 while ($_ = shift(@XXX)) {
95 my $e = index ($_, 'not') >= 0 ? '' : 1;
113 ok !keys %XXX, "%XXX is empty";
119 my $message = "Test empty pattern";
125 is($&, $xyz, $message);
130 is($&, $xyz, $message);
134 no warnings 'uninitialized';
136 is($&, $xyz, $message);
140 is($&, $xyz, $message);
142 # each entry: regexp, match string, $&, //o match success
145 [ "", "xy", "x", 1 ],
146 [ "y", "yz", "y", !1 ],
148 for my $test (@tests) {
149 my ($re, $str, $matched, $omatch) = @$test;
151 ok($str =~ /$re/, "$str matches /$re/");
152 is($&, $matched, "on $matched");
154 is($str =~ /$re/o, $omatch, "$str matches /$re/o (or not)");
159 my $message = q !Check $`, $&, $'!;
161 /def/; # optimized up to cmd
162 is("$`:$&:$'", 'abc:def:ghi', $message);
165 /cde/ + 0; # optimized only to spat
166 is("$`:$&:$'", 'ab:cde:fghi', $message);
168 /[d][e][f]/; # not optimized
169 is("$`:$&:$'", 'abc:def:ghi', $message);
173 $_ = 'now is the {time for all} good men to come to.';
175 is($1, 'time for all', "Match braces");
179 my $message = "{N,M} quantifier";
180 $_ = 'xxx {3,4} yyy zzz';
181 ok(/( {3,4})/, $message);
182 is($1, ' ', $message);
183 unlike($_, qr/( {4,})/, $message);
184 ok(/( {2,3}.)/, $message);
185 is($1, ' y', $message);
186 ok(/(y{2,3}.)/, $message);
187 is($1, 'yyy ', $message);
188 unlike($_, qr/x {3,4}/, $message);
189 unlike($_, qr/^xxx {3,4}/, $message);
193 my $message = "Test /g";
195 $_ = "now is the time for all good men to come to.";
196 my @words = /(\w+)/g;
197 my $exp = "now:is:the:time:for:all:good:men:to:come:to";
199 is("@words", $exp, $message);
205 is("@words", $exp, $message);
212 is("@words", "to:to", $message);
216 is("@words", "to:to", $message);
232 my $t1 = my $t2 = my $t3 = my $t4 = my $t5 =
233 my $t6 = my $t7 = my $t8 = my $t9 = 0;
235 for my $iter (1 .. 5) {
246 my $x = "$t1$t2$t3$t4$t5$t6$t7$t8$t9";
247 is($x, '505550555', "Test /o");
252 ok "abc" =~ /^abc$|$xyz/, "| after \$";
254 # perl 4.009 says "unmatched ()"
255 my $message = '$ inside ()';
258 eval '"abc" =~ /a(bc$)|$xyz/; $result = "$&:$1"';
259 is($@, "", $message);
260 is($result, "abc:bc", $message);
264 my $message = "Scalar /g";
267 ok( /abc/g && $` eq "", $message);
268 ok( /abc/g && $` eq "abcfoo", $message);
269 ok(!/abc/g, $message);
271 $message = "Scalar /gi";
273 ok( /ABC/gi && $` eq "", $message);
274 ok( /ABC/gi && $` eq "abcfoo", $message);
275 ok(!/ABC/gi, $message);
277 $message = "Scalar /g";
279 ok( /abc/g && $' eq "fooabcbar", $message);
280 ok( /abc/g && $' eq "bar", $message);
284 is(@x, 2, "/g reset after assignment");
288 my $message = '/g, \G and pos';
292 is(pos $_, 2, $message);
294 is(pos $_, undef, $message);
298 my $message = '(?{ })';
300 'abc' =~ m'a(?{ $out = 2 })b';
301 is($out, 2, $message);
304 'abc' =~ m'a(?{ $out = 3 })c';
305 is($out, 1, $message);
309 $_ = 'foobar1 bar2 foobar3 barfoobar5 foobar6';
310 my @out = /(?<!foo)bar./g;
311 is("@out", 'bar2 barf', "Negative lookbehind");
315 my $message = "REG_INFTY tests";
316 # Tests which depend on REG_INFTY
318 # Defaults assumed if this fails
319 eval { require Config; };
320 $::reg_infty = $Config::Config{reg_infty} // 32767;
321 $::reg_infty_m = $::reg_infty - 1;
322 $::reg_infty_p = $::reg_infty + 1;
323 $::reg_infty_m = $::reg_infty_m; # Suppress warning.
325 # As well as failing if the pattern matches do unexpected things, the
326 # next three tests will fail if you should have picked up a lower-than-
327 # default value for $reg_infty from Config.pm, but have not.
329 is(eval q{('aaa' =~ /(a{1,$::reg_infty_m})/)[0]}, 'aaa', $message);
330 is($@, '', $message);
331 is(eval q{('a' x $::reg_infty_m) =~ /a{$::reg_infty_m}/}, 1, $message);
332 is($@, '', $message);
333 isnt(q{('a' x ($::reg_infty_m - 1)) !~ /a{$::reg_infty_m}/}, 1, $message);
334 is($@, '', $message);
336 eval "'aaa' =~ /a{1,$::reg_infty}/";
337 like($@, qr/^\QQuantifier in {,} bigger than/, $message);
338 eval "'aaa' =~ /a{1,$::reg_infty_p}/";
339 like($@, qr/^\QQuantifier in {,} bigger than/, $message);
343 # Poke a couple more parse failures
344 my $context = 'x' x 256;
345 eval qq("${context}y" =~ /(?<=$context)y/);
346 ok $@ =~ /^\QLookbehind longer than 255 not/, "Lookbehind limit";
351 for my $l (125, 140, 250, 270, 300000, 30) { # Ordered to free memory
353 my $message = "Long monster, length = $l";
354 like("ba$a=", qr/a$a=/, $message);
355 unlike("b$a=", qr/a$a=/, $message);
356 like("b$a=", qr/ba+=/, $message);
358 like("ba$a=", qr/b(?:a|b)+=/, $message);
363 # 20000 nodes, each taking 3 words per string, and 1 per branch
364 my $long_constant_len = join '|', 12120 .. 32645;
365 my $long_var_len = join '|', 8120 .. 28645;
366 my %ans = ( 'ax13876y25677lbc' => 1,
367 'ax13876y25677mcb' => 0, # not b.
368 'ax13876y35677nbc' => 0, # Num too big
369 'ax13876y25677y21378obc' => 1,
370 'ax13876y25677y21378zbc' => 0, # Not followed by [k-o]
371 'ax13876y25677y21378y21378kbc' => 1,
372 'ax13876y25677y21378y21378kcb' => 0, # Not b.
373 'ax13876y25677y21378y21378y21378kbc' => 0, # 5 runs
377 my $message = "20000 nodes, const-len '$_'";
378 ok !($ans{$_} xor /a(?=([yx]($long_constant_len)){2,4}[k-o]).*b./o), $message;
380 $message = "20000 nodes, var-len '$_'";
381 ok !($ans{$_} xor /a(?=([yx]($long_var_len)){2,4}[k-o]).*b./o,), $message;
386 my $message = "Complicated backtracking";
387 $_ = " a (bla()) and x(y b((l)u((e))) and b(l(e)e)e";
388 my $expect = "(bla()) ((l)u((e))) (l(e)e)";
395 (?{ $c = 1 }) # Initialize
397 (?(?{ $c == 0 }) # PREVIOUS iteration was OK, stop the loop
399 ) # Fail: will unwind one iteration back
402 [^()]+ # Match a big chunk
405 ) # Do not try to match subchunks
413 )+ # This may not match with different subblocks
418 ) # Otherwise the chunk 1 may succeed with $c>0
424 push @ans, $res while $res = matchit;
425 is("@ans", "1 1 1", $message);
428 is("@ans", $expect, $message);
430 $message = "Recursion with (??{ })";
432 $matched = qr/\((?:(?>[^()]+)|(??{$matched}))*\)/;
434 @ans = my @ans1 = ();
435 push (@ans, $res), push (@ans1, $&) while $res = m/$matched/g;
437 is("@ans", "1 1 1", $message);
438 is("@ans1", $expect, $message);
441 is("@ans", $expect, $message);
446 ok "abc" =~ /^(??{"a"})b/, '"abc" =~ /^(??{"a"})b/';
450 my @ans = ('a/b' =~ m%(.*/)?(.*)%); # Stack may be bad
451 is("@ans", 'a/ b', "Stack may be bad");
455 my $message = "Eval-group not allowed at runtime";
456 my $code = '{$blah = 45}';
459 ok($@ && $@ =~ /not allowed at runtime/ && $blah == 12, $message);
462 my $res = eval { "xx" =~ /(?$code)/o };
464 no warnings 'uninitialized';
465 chomp $@; my $message = "$message '$@', '$res', '$blah'";
466 ok($@ && $@ =~ /not allowed at runtime/ && $blah == 12, $message);
471 $res = eval { "xx" =~ /(?$code)/o };
473 no warnings 'uninitialized';
474 my $message = "$message '$@', '$res', '$blah'";
475 ok(!$@ && $res, $message);
478 $code = '{$blah = 45}';
481 is($blah, 45, $message);
485 is($blah, 45, $message);
489 my $message = "Pos checks";
492 is(pos $x, 2, $message);
495 is(pos $x, 2, $message);
503 is(f (pos $x), 4, $message);
507 my $message = 'Checking $^R';
509 'foot' =~ /foo(?{$x = 12; 75})[t]/;
510 is($^R, 75, $message);
513 'foot' =~ /foo(?{$x = 12; 75})[xy]/;
514 ok($^R eq '67' && $x eq '12', $message);
517 'foot' =~ /foo(?{ $^R + 12 })((?{ $x = 12; $^R + 17 })[xy])?/;
518 ok($^R eq '79' && $x eq '12', $message);
522 is(qr/\b\v$/i, '(?^i:\b\v$)', 'qr/\b\v$/i');
523 is(qr/\b\v$/s, '(?^s:\b\v$)', 'qr/\b\v$/s');
524 is(qr/\b\v$/m, '(?^m:\b\v$)', 'qr/\b\v$/m');
525 is(qr/\b\v$/x, '(?^x:\b\v$)', 'qr/\b\v$/x');
526 is(qr/\b\v$/xism, '(?^msix:\b\v$)', 'qr/\b\v$/xism');
527 is(qr/\b\v$/, '(?^:\b\v$)', 'qr/\b\v$/');
530 { # Test that charset modifier work, and are interpolated
531 is(qr/\b\v$/, '(?^:\b\v$)', 'Verify no locale, no unicode_strings gives default modifier');
532 is(qr/(?l:\b\v$)/, '(?^:(?l:\b\v$))', 'Verify infix l modifier compiles');
533 is(qr/(?u:\b\v$)/, '(?^:(?u:\b\v$))', 'Verify infix u modifier compiles');
534 is(qr/(?l)\b\v$/, '(?^:(?l)\b\v$)', 'Verify (?l) compiles');
535 is(qr/(?u)\b\v$/, '(?^:(?u)\b\v$)', 'Verify (?u) compiles');
537 my $dual = qr/\b\v$/;
541 skip 'Locales not available', 1 unless locales_enabled('LC_CTYPE');
545 is($locale, '(?^l:\b\v$)', 'Verify has l modifier when compiled under use locale');
549 use feature 'unicode_strings';
550 my $unicode = qr/\b\v$/;
551 is($unicode, '(?^u:\b\v$)', 'Verify has u modifier when compiled under unicode_strings');
552 is(qr/abc$dual/, '(?^u:abc(?^:\b\v$))', 'Verify retains d meaning when interpolated under locale');
555 skip 'Locales not available', 1 unless locales_enabled('LC_CTYPE');
557 is(qr/abc$locale/, '(?^u:abc(?^l:\b\v$))', 'Verify retains l when interpolated under unicode_strings');
560 no feature 'unicode_strings';
562 skip 'Locales not available', 1 unless locales_enabled('LC_CTYPE');
563 is(qr/abc$locale/, '(?^:abc(?^l:\b\v$))', 'Verify retains l when interpolated outside locale and unicode strings');
566 is(qr/def$unicode/, '(?^:def(?^u:\b\v$))', 'Verify retains u when interpolated outside locale and unicode strings');
569 skip 'Locales not available', 2 unless locales_enabled('LC_CTYPE');
572 is(qr/abc$dual/, '(?^l:abc(?^:\b\v$))', 'Verify retains d meaning when interpolated under locale');
573 is(qr/abc$unicode/, '(?^l:abc(?^u:\b\v$))', 'Verify retains u when interpolated under locale');
578 my $message = "Look around";
580 foreach my $ans ('', 'c') {
581 ok(/(?<=(?=a)..)((?=c)|.)/g, $message);
582 is($1, $ans, $message);
587 my $message = "Empty clause";
589 foreach my $ans ('', 'a', '') {
590 ok(/^|a|$/g, $message);
591 is($&, $ans, $message);
597 my $message = "Prefixify";
599 my ($v, $a, $b, $res) = @_;
600 ok($v =~ s/\Q$a\E/$b/, $message);
601 is($v, $res, $message);
605 prefixify ('/a/b/lib/arch', "/a/b/lib", 'X/lib', 'X/lib/arch');
606 prefixify ('/a/b/man/arch', "/a/b/man", 'X/man', 'X/man/arch');
612 ok $1 && /$1/, "Capture a quote";
616 no warnings 'closure';
617 my $message = '(?{ $var } refers to package vars';
621 '' =~ /(?{ $c = 4 })/;
622 main::is($c, 4, $message);
623 main::is($::c, 3, $message);
627 is(eval 'q(a:[b]:) =~ /[x[:foo:]]/', undef);
628 like ($@, qr/POSIX class \[:[^:]+:\] unknown in regex/,
629 'POSIX class [: :] must have valid name');
631 for my $d (qw [= .]) {
632 is(eval "/[[${d}foo${d}]]/", undef);
633 like ($@, qr/\QPOSIX syntax [$d $d] is reserved for future extensions/,
634 "POSIX syntax [[$d $d]] is an error");
639 # test if failure of patterns returns empty list
640 my $message = "Failed pattern returns empty list";
643 is("@_", "", $message);
646 is("@_", "", $message);
649 is("@_", "", $message);
652 is("@_", "", $message);
656 my $message = '@- and @+ and @{^CAPTURE} tests';
660 is($#{^CAPTURE}, -1, $message);
661 is($#+, 0, $message);
662 is($#-, 0, $message);
663 is($+ [0], 2, $message);
664 is($- [0], 1, $message);
665 ok(!defined $+ [1] && !defined $- [1] &&
666 !defined $+ [2] && !defined $- [2], $message);
669 is($#{^CAPTURE}, 1, $message); # one less than $#-
670 is($#+, 2, $message);
671 is($#-, 2, $message);
672 is($+ [0], 3, $message);
673 is($- [0], 0, $message);
674 is(${^CAPTURE}[0], "c", $message);
675 is($+ [1], 2, $message);
676 is($- [1], 1, $message);
677 is(${^CAPTURE}[1], "e", $message);
678 is($+ [2], 3, $message);
679 is($- [2], 2, $message);
680 ok(!defined $+ [3] && !defined $- [3] &&
681 !defined ${^CAPTURE}[2] && !defined ${^CAPTURE}[3] &&
682 !defined $+ [4] && !defined $- [4], $message);
684 # Exists has a special check for @-/@+ - bug 45147
685 ok(exists $-[0], $message);
686 ok(exists $+[0], $message);
687 ok(exists ${^CAPTURE}[0], $message);
688 ok(exists ${^CAPTURE}[1], $message);
689 ok(exists $-[2], $message);
690 ok(exists $+[2], $message);
691 ok(!exists ${^CAPTURE}[2], $message);
692 ok(!exists $-[3], $message);
693 ok(!exists $+[3], $message);
694 ok(exists ${^CAPTURE}[-1], $message);
695 ok(exists ${^CAPTURE}[-2], $message);
696 ok(exists $-[-1], $message);
697 ok(exists $+[-1], $message);
698 ok(exists $-[-3], $message);
699 ok(exists $+[-3], $message);
700 ok(!exists $-[-4], $message);
701 ok(!exists $+[-4], $message);
702 ok(!exists ${^CAPTURE}[-3], $message);
706 is($#{^CAPTURE}, 2, $message); # one less than $#-
707 is($#+, 3, $message);
708 is($#-, 3, $message);
709 is(${^CAPTURE}[0], "c", $message);
710 is(${^CAPTURE}[2], "e", $message . "[$1 $3]");
711 is($+ [1], 2, $message);
712 is($- [1], 1, $message);
713 is($+ [3], 3, $message);
714 is($- [3], 2, $message);
715 ok(!defined $+ [2] && !defined $- [2] &&
716 !defined $+ [4] && !defined $- [4] &&
717 !defined ${^CAPTURE}[1], $message);
720 is($#{^CAPTURE}, 0, $message); # one less than $#-
721 is($#+, 1, $message);
722 is($#-, 1, $message);
723 is(${^CAPTURE}[0], "c", $message);
724 is($+ [0], 2, $message);
725 is($- [0], 0, $message);
726 is($+ [1], 2, $message);
727 is($- [1], 1, $message);
728 ok(!defined $+ [2] && !defined $- [2] &&
729 !defined $+ [3] && !defined $- [3] &&
730 !defined ${^CAPTURE}[1], $message);
733 is($#{^CAPTURE}, 0, $message); # one less than $#-
734 is($#+, 2, $message);
735 is($#-, 1, $message);
737 # Check that values don’t stick
738 " "=~/()()()(.)(..)/;
739 my($m,$p,$q) = (\$-[5], \$+[5], \${^CAPTURE}[4]);
740 () = "$$_" for $m, $p, $q; # FETCH (or eqv.)
742 is $$m, undef, 'values do not stick to @- elements';
743 is $$p, undef, 'values do not stick to @+ elements';
744 is $$q, undef, 'values do not stick to @{^CAPTURE} elements';
747 foreach ('$+[0] = 13', '$-[0] = 13', '@+ = (7, 6, 5)',
748 '${^CAPTURE}[0] = 13',
749 '@- = qw (foo bar)', '$^N = 42') {
751 like($@, qr/^Modification of a read-only value attempted/,
752 '$^N, @- and @+ are read-only');
756 my $message = '\G testing';
760 is("@a", "a a", $message);
764 unlike($str, qr/^\G/, $message);
765 unlike($str, qr/^.\G/, $message);
766 like($str, qr/^..\G/, $message);
767 unlike($str, qr/^...\G/, $message);
768 ok($str =~ /\G../ && $& eq 'cd', $message);
769 ok($str =~ /.\G./ && $& eq 'bc', $message);
774 my $message = '\G and intuit and anchoring';
777 ok($_ =~ /\Gabc/, $message);
778 ok($_ =~ /^\Gabc/, $message);
781 ok($_ =~ /\Gdef/, $message);
783 ok($_ =~ /\Gdef$/, $message);
785 ok($_ =~ /abc\Gdef$/, $message);
787 ok($_ =~ /^abc\Gdef$/, $message);
789 ok($_ =~ /c\Gd/, $message);
791 ok($_ =~ /..\GX?def/, $message);
797 my @a = $s =~ /(\d)\G/g; # this infinitely looped up till 5.19.1
798 is("@a", "1", '\G looping');
803 my $message = 'pos inside (?{ })';
806 like($str, qr/b(?{$foo = $_; $bar = pos})c/, $message);
807 is($foo, $str, $message);
808 is($bar, 2, $message);
809 is(pos $str, undef, $message);
814 ok($str =~ /b(?{$foo = $_; $bar = pos})c/g, $message);
815 is($foo, $str, $message);
816 is($bar, 2, $message);
817 is(pos $str, 3, $message);
822 like($_, qr/b(?{$foo = $_; $bar = pos})c/, $message);
823 is($foo, $str, $message);
824 is($bar, 2, $message);
828 ok(/b(?{$foo = $_; $bar = pos})c/g, $message);
829 is($foo, $str, $message);
830 is($bar, 2, $message);
831 is(pos, 3, $message);
836 1 while /b(?{$foo = $_; $bar = pos})c/g;
837 is($foo, $str, $message);
838 is($bar, 2, $message);
839 is(pos, undef, $message);
844 ok(s/b(?{$foo = $_; $bar = pos})c/x/g, $message);
845 is($foo, 'abcde|abcde', $message);
846 is($bar, 8, $message);
847 is($_, 'axde|axde', $message);
852 () = /([ace]).(?{push @res, $1,$2})([ce])(?{push @res, $1,$2})/g;
853 @res = map {defined $_ ? "'$_'" : 'undef'} @res;
854 is("@res", "'a' undef 'a' 'c' 'e' undef 'a' undef 'a' 'c'", $message);
857 () = /([ace]).(?{push @res, $`,$&,$'})([ce])(?{push @res, $`,$&,$'})/g;
858 @res = map {defined $_ ? "'$_'" : 'undef'} @res;
859 is("@res", "'' 'ab' 'cde|abcde' " .
860 "'' 'abc' 'de|abcde' " .
861 "'abcd' 'e|' 'abcde' " .
862 "'abcde|' 'ab' 'cde' " .
863 "'abcde|' 'abc' 'de'", $message);
867 my $message = '\G anchor checks';
868 my $foo = 'aabbccddeeffgg';
871 ok($foo =~ /.\G(..)/g, $message);
872 is($1, 'ab', $message);
875 ok($foo =~ /.\G(..)/g, $message);
876 is($1, 'cc', $message);
879 ok($foo =~ /.\G(..)/g, $message);
880 is($1, 'de', $message);
882 ok($foo =~ /\Gef/g, $message);
885 ok($foo =~ /\G(..)/g, $message);
886 is($1, 'aa', $message);
888 ok($foo =~ /\G(..)/g, $message);
889 is($1, 'bb', $message);
892 ok($foo =~ /\G(..)/g, $message);
893 is($1, 'cd', $message);
897 my $message = 'basic \G floating checks';
898 my $foo = 'aabbccddeeffgg';
901 ok($foo =~ /a+\G(..)/g, "$message: a+\\G");
902 is($1, 'ab', "$message: ab");
905 ok($foo =~ /b+\G(..)/g, "$message: b+\\G");
906 is($1, 'cc', "$message: cc");
909 ok($foo =~ /d+\G(..)/g, "$message: d+\\G");
910 is($1, 'de', "$message: de");
912 ok($foo =~ /\Gef/g, "$message: \\Gef");
916 ok($foo =~ /(?=a+\G)(..)/g, "$message: (?a+\\G)");
917 is($1, 'aa', "$message: aa");
921 ok($foo =~ /a(?=a+\G)(..)/g, "$message: a(?=a+\\G)");
922 is($1, 'ab', "$message: ab");
928 my @res = /(\d*|x)/g;
930 is("@res", "123||x|123|", "0 match in alternation");
934 my $message = "Match against temporaries (created via pp_helem())" .
936 ok({foo => "bar\n" . $^X} -> {foo} =~ /^(.*)\n/g, $message);
937 is($1, "bar", $message);
941 my $message = 'package $i inside (?{ }), ' .
942 'saved substrings and changing $_';
943 our @a = qw [foo bar];
945 s/(\w)(?{push @b, $1})/,$1,/g for @a;
946 is("@b", "f o o b a r", $message);
947 is("@a", ",f,,o,,o, ,b,,a,,r,", $message);
949 $message = 'lexical $i inside (?{ }), ' .
950 'saved substrings and changing $_';
951 no warnings 'closure';
952 my @c = qw [foo bar];
954 s/(\w)(?{push @d, $1})/,$1,/g for @c;
955 is("@d", "f o o b a r", $message);
956 is("@c", ",f,,o,,o, ,b,,a,,r,", $message);
960 my $message = 'Brackets';
963 { (?> [^{}]+ | (??{ $brackets }) )* }
966 ok("{{}" =~ $brackets, $message);
967 is($&, "{}", $message);
968 ok("something { long { and } hairy" =~ $brackets, $message);
969 is($&, "{ and }", $message);
970 ok("something { long { and } hairy" =~ m/((??{ $brackets }))/, $message);
971 is($&, "{ and }", $message);
977 ok(!m/^-.*bb/mg, '$_ = "a-a\nxbb"; m/^-.*bb/mg');
981 my $message = '\G anchor checks';
982 my $text = "aaXbXcc";
984 ok($text !~ /\GXb*X/g, $message);
989 unlike($_, qr/^\s*A/m, '$_ = "xA\n" x 500; /^\s*A/m"');
991 my $text = "abc dbf";
992 my @res = ($text =~ /.*?(b).*?\b/g);
993 is("@res", "b b", '\b is not special');
997 my $message = '\S, [\S], \s, [\s]';
998 my @a = map chr, 0 .. 255;
999 my @b = grep m/\S/, @a;
1000 my @c = grep m/[^\s]/, @a;
1001 is("@b", "@c", $message);
1004 @c = grep /[\S]/, @a;
1005 is("@b", "@c", $message);
1008 @c = grep /[^\S]/, @a;
1009 is("@b", "@c", $message);
1012 @c = grep /[\s]/, @a;
1013 is("@b", "@c", $message);
1015 # Test an inverted posix class with a char also in the class.
1016 my $nbsp = chr utf8::unicode_to_native(0xA0);
1017 my $non_s = chr utf8::unicode_to_native(0xA1);
1018 my $pat_string = "[^\\S ]";
1019 unlike(" ", qr/$pat_string/, "Verify ' ' !~ /$pat_string/");
1020 like("\t", qr/$pat_string/, "Verify '\\t =~ /$pat_string/");
1021 unlike($nbsp, qr/$pat_string/, "Verify non-utf8-NBSP !~ /$pat_string/");
1022 utf8::upgrade($nbsp);
1023 like($nbsp, qr/$pat_string/, "Verify utf8-NBSP =~ /$pat_string/");
1024 unlike($non_s, qr/$pat_string/, "Verify non-utf8-inverted-bang !~ /$pat_string/");
1025 utf8::upgrade($non_s);
1026 unlike($non_s, qr/$pat_string/, "Verify utf8-inverted-bang !~ /$pat_string/");
1029 my $message = '\D, [\D], \d, [\d]';
1030 my @a = map chr, 0 .. 255;
1031 my @b = grep /\D/, @a;
1032 my @c = grep /[^\d]/, @a;
1033 is("@b", "@c", $message);
1036 @c = grep /[\D]/, @a;
1037 is("@b", "@c", $message);
1040 @c = grep /[^\D]/, @a;
1041 is("@b", "@c", $message);
1044 @c = grep /[\d]/, @a;
1045 is("@b", "@c", $message);
1048 my $message = '\W, [\W], \w, [\w]';
1049 my @a = map chr, 0 .. 255;
1050 my @b = grep /\W/, @a;
1051 my @c = grep /[^\w]/, @a;
1052 is("@b", "@c", $message);
1055 @c = grep /[\W]/, @a;
1056 is("@b", "@c", $message);
1059 @c = grep /[^\W]/, @a;
1060 is("@b", "@c", $message);
1063 @c = grep /[\w]/, @a;
1064 is("@b", "@c", $message);
1068 # see if backtracking optimization works correctly
1069 my $message = 'Backtrack optimization';
1070 like("\n\n", qr/\n $ \n/x, $message);
1071 like("\n\n", qr/\n* $ \n/x, $message);
1072 like("\n\n", qr/\n+ $ \n/x, $message);
1073 like("\n\n", qr/\n? $ \n/x, $message);
1074 like("\n\n", qr/\n*? $ \n/x, $message);
1075 like("\n\n", qr/\n+? $ \n/x, $message);
1076 like("\n\n", qr/\n?? $ \n/x, $message);
1077 unlike("\n\n", qr/\n*+ $ \n/x, $message);
1078 unlike("\n\n", qr/\n++ $ \n/x, $message);
1079 like("\n\n", qr/\n?+ $ \n/x, $message);
1084 use overload '""' => sub {'Object S'};
1087 my $message = "Ref stringification";
1088 ::ok(do { \my $v} =~ /^SCALAR/, "Scalar ref stringification") or diag($message);
1089 ::ok(do {\\my $v} =~ /^REF/, "Ref ref stringification") or diag($message);
1090 ::ok([] =~ /^ARRAY/, "Array ref stringification") or diag($message);
1091 ::ok({} =~ /^HASH/, "Hash ref stringification") or diag($message);
1092 ::ok('S' -> new =~ /^Object S/, "Object stringification") or diag($message);
1096 my $message = "Test result of match used as match";
1097 ok('a1b' =~ ('xyz' =~ /y/), $message);
1098 is($`, 'a', $message);
1099 ok('a1b' =~ ('xyz' =~ /t/), $message);
1100 is($`, 'a', $message);
1104 my $message = '"1" is not \s';
1105 warning_is(sub {unlike("1\n" x 102, qr/^\s*\n/m, $message)},
1106 undef, "$message (did not warn)");
1110 my $message = '\s, [[:space:]] and [[:blank:]]';
1111 my %space = (spc => " ",
1116 # There's no \v but the vertical tabulator seems miraculously
1117 # be 11 both in ASCII and EBCDIC.
1121 my @space0 = sort grep {$space {$_} =~ /\s/ } keys %space;
1122 my @space1 = sort grep {$space {$_} =~ /[[:space:]]/} keys %space;
1123 my @space2 = sort grep {$space {$_} =~ /[[:blank:]]/} keys %space;
1125 is("@space0", "cr ff lf spc tab vt", $message);
1126 is("@space1", "cr ff lf spc tab vt", $message);
1127 is("@space2", "spc tab", $message);
1132 # this must be a high number and go from 0 to N, as the bug we are looking for doesn't
1133 # seem to be predictable. Slight changes to the test make it fail earlier or later.
1134 foreach my $i (0 .. $n)
1137 ok $str=~/.*\z/, "implicit MBOL check string disable does not break things length=$i";
1141 # we are actually testing that we dont die when executing these patterns
1144 ok(utf8::is_utf8($e),"got a unicode string - rt75680");
1146 ok($e !~ m/.*?[x]$/, "unicode string against /.*?[x]\$/ - rt75680");
1147 ok($e !~ m/.*?\p{Space}$/i, "unicode string against /.*?\\p{space}\$/i - rt75680");
1148 ok($e !~ m/.*?[xyz]$/, "unicode string against /.*?[xyz]\$/ - rt75680");
1149 ok($e !~ m/(.*?)[,\p{isSpace}]+((?:\p{isAlpha}[\p{isSpace}\.]{1,2})+)\p{isSpace}*$/, "unicode string against big pattern - rt75680");
1152 # we are actually testing that we dont die when executing these patterns
1153 my $e = "B" . uni_to_native("\x{f6}") . "ck";
1154 ok(!utf8::is_utf8($e), "got a latin string - rt75680");
1156 ok($e !~ m/.*?[x]$/, "latin string against /.*?[x]\$/ - rt75680");
1157 ok($e !~ m/.*?\p{Space}$/i, "latin string against /.*?\\p{space}\$/i - rt75680");
1158 ok($e !~ m/.*?[xyz]$/,"latin string against /.*?[xyz]\$/ - rt75680");
1159 ok($e !~ m/(.*?)[,\p{isSpace}]+((?:\p{isAlpha}[\p{isSpace}\.]{1,2})+)\p{isSpace}*$/,"latin string against big pattern - rt75680");
1164 # Tests for bug 77414.
1167 my $message = '\p property after empty * match';
1169 like("1", qr/\s*\pN/, $message);
1170 like("-", qr/\s*\p{Dash}/, $message);
1171 like(" ", qr/\w*\p{Blank}/, $message);
1174 like("1", qr/\s*\pN+/, $message);
1175 like("-", qr/\s*\p{Dash}{1}/, $message);
1176 like(" ", qr/\w*\p{Blank}{1,4}/, $message);
1180 { # Some constructs with Latin1 characters cause a utf8 string not
1181 # to match itself in non-utf8
1182 my $c = uni_to_native("\xc0");
1183 my $pattern = my $utf8_pattern = qr/(($c)+,?)/;
1184 utf8::upgrade($utf8_pattern);
1185 ok $c =~ $pattern, "\\xc0 =~ $pattern; Neither pattern nor target utf8";
1186 ok $c =~ /$pattern/i, "\\xc0 =~ /$pattern/i; Neither pattern nor target utf8";
1187 ok $c =~ $utf8_pattern, "\\xc0 =~ $pattern; pattern utf8, target not";
1188 ok $c =~ /$utf8_pattern/i, "\\xc0 =~ /$pattern/i; pattern utf8, target not";
1190 ok $c =~ $pattern, "\\xc0 =~ $pattern; target utf8, pattern not";
1191 ok $c =~ /$pattern/i, "\\xc0 =~ /$pattern/i; target utf8, pattern not";
1192 ok $c =~ $utf8_pattern, "\\xc0 =~ $pattern; Both target and pattern utf8";
1193 ok $c =~ /$utf8_pattern/i, "\\xc0 =~ /$pattern/i; Both target and pattern utf8";
1196 { # Make sure can override the formatting
1197 use feature 'unicode_strings';
1198 ok uni_to_native("\xc0") =~ /\w/, 'Under unicode_strings: "\xc0" =~ /\w/';
1199 ok uni_to_native("\xc0") !~ /(?d:\w)/, 'Under unicode_strings: "\xc0" !~ /(?d:\w)/';
1206 is("$qr", "(?^:)", "Empty pattern qr// stringifies to (?^:) with unicode flag enabled - Bug #80212");
1209 is("$qr", "(?^:)", "Empty pattern qr// stringifies to (?^:) with unicode flag disabled - Bug #80212");
1214 local $::TODO = "[perl #38133]";
1216 "A" =~ /(((?:A))?)+/;
1222 is($first, $second);
1226 # RT #3516: \G in a m//g expression causes problems
1228 while ("abc" =~ m/(\G[ac])?/g) {
1229 last if $count++ > 10;
1231 ok($count < 10, 'RT #3516 A');
1234 while ("abc" =~ m/(\G|.)[ac]/g) {
1235 last if $count++ > 10;
1237 ok($count < 10, 'RT #3516 B');
1240 while ("abc" =~ m/(\G?[ac])?/g) {
1241 last if $count++ > 10;
1243 ok($count < 10, 'RT #3516 C');
1246 # RT #84294: Is this a bug in the simple Perl regex?
1247 # : Nested buffers and (?{...}) dont play nicely on partial matches
1249 ok("ab" =~ /((\w+)(?{ push @got, $2 })){2}/,"RT #84294: Pattern should match");
1250 my $want= "'ab', 'a', 'b'";
1251 my $got= join(", ", map { defined($_) ? "'$_'" : "undef" } @got);
1252 is($got,$want,'RT #84294: check that "ab" =~ /((\w+)(?{ push @got, $2 })){2}/ leaves @got in the correct state');
1256 # Suppress warnings, as the non-unicode one comes out even if turn off
1257 # warnings here (because the execution is done in another scope).
1258 local $SIG{__WARN__} = sub {};
1259 my $str = "\x{110000}";
1261 unlike($str, qr/\p{ASCII_Hex_Digit=True}/, "Non-Unicode doesn't match \\p{AHEX=True}");
1262 like($str, qr/\p{ASCII_Hex_Digit=False}/, "Non-Unicode matches \\p{AHEX=False}");
1263 like($str, qr/\P{ASCII_Hex_Digit=True}/, "Non-Unicode matches \\P{AHEX=True}");
1264 unlike($str, qr/\P{ASCII_Hex_Digit=False}/, "Non-Unicode matches \\P{AHEX=FALSE}");
1268 # Test that IDstart works, but because the author (khw) knows
1269 # regexes much better than the rest of the core, it is being done here
1270 # in the context of a regex which relies on buffer names beginng with
1274 like($str, qr/(?<a>abc)/, "'a' is legal IDStart");
1275 like($str, qr/(?<_>abc)/, "'_' is legal IDStart");
1276 like($str, qr/(?<ß>abc)/, "U+00DF is legal IDStart");
1277 like($str, qr/(?<â„•>abc)/, "U+2115' is legal IDStart");
1279 # This test works on Unicode 6.0 in which U+2118 and U+212E are legal
1280 # IDStarts there, but are not Word characters, and therefore Perl
1281 # doesn't allow them to be IDStarts. But there is no guarantee that
1282 # Unicode won't change things around in the future so that at some
1283 # future Unicode revision these tests would need to be revised.
1284 foreach my $char ("%", "×", chr(0x2118), chr(0x212E)) {
1287 "abc" =~ qr/(?<$char>abc)/;
1289 utf8::encode($prog);
1290 fresh_perl_like($prog, qr!Group name must start with a non-digit word character!, {},
1291 sprintf("'U+%04X not legal IDFirst'", ord($char)));
1297 utf8::upgrade($pat);
1298 like("\xffb", qr/$pat/i, "/i: utf8 pattern, non-utf8 string, latin1-char preceding matching char in string");
1301 { # Crash with @a =~ // warning
1302 local $SIG{__WARN__} = sub {
1303 pass 'no crash for @a =~ // warning'
1305 eval ' sub { my @a =~ // } ';
1308 { # Concat overloading and qr// thingies
1314 '""' => sub { ${$_[0]} },
1316 push @refs, ref $_[1] if ref $_[1];
1317 bless $_[2] ? \"$_[1]${$_[0]}" : \"${$_[0]}$_[1]"
1322 my $o = bless \$s, Cat::;
1324 is "@refs", "Regexp", '/$o$qr/ passes qr ref to cat overload meth';
1330 $count++ while $str=~/.*/g;
1331 is $count, 2, 'test that ANCH_MBOL works properly. We should get 2 from $count++ while "\n"=~/.*/g';
1333 $class_count++ while $str=~/[^\n]*/g;
1334 is $class_count, $count, 'while "\n"=~/.*/g and while "\n"=~/[^\n]*/g should behave the same';
1336 $anch_count++ while $str=~/^.*/mg;
1337 is $anch_count, 1, 'while "\n"=~/^.*/mg should match only once';
1342 my $A_grave = uni_to_native("\xc0");
1343 like uni_to_native("\xe0"), qr/(?i:$A_grave)/, "(?i: shouldn't lose the passed in /u";
1345 unlike "\x{100}", qr/(?i:\w)/, "(?i: shouldn't lose the passed in /a";
1347 unlike 'k', qr/(?i:\N{KELVIN SIGN})/, "(?i: shouldn't lose the passed in /aa";
1351 # the test for whether the pattern should be re-compiled should
1352 # consider the UTF8ness of the previous and current pattern
1353 # string, as well as the physical bytes of the pattern string
1355 for my $s (byte_utf8a_to_utf8n("\xc4\x80"), "\x{100}") {
1356 ok($s =~ /^$s$/, "re-compile check is UTF8-aware");
1360 # #113682 more overloading and qr//
1361 # when doing /foo$overloaded/, if $overloaded returns
1362 # a qr/(?{})/ via qr or "" overloading, then 'use re 'eval'
1363 # shouldn't be required. Via '.', it still is.
1366 use overload 'qr' => sub { qr/(??{50})/ };
1369 use overload '""' => sub { qr/(??{51})/ };
1372 use overload '.' => sub { $_[1] . qr/(??{52})/ };
1375 use overload '""' => sub { qr/(??{7})/ },
1376 '.' => sub { $_[1] . qr/(??{53})/ };
1378 package Qr_indirect;
1379 use overload '""' => sub { $_[0][0] };
1384 my $o = bless [], "Qr$i";
1385 if ((0,0,1,1)[$i]) {
1386 eval { "A5$i" =~ /^A$o$/ };
1387 like($@, qr/Eval-group not allowed/, "Qr$i");
1388 eval { "5$i" =~ /$o/ };
1389 like($@, ($i == 3 ? qr/^$/ : qr/no method found,/),
1393 ok("A5$i" =~ /^A$o$/, "Qr$i - with use re eval");
1394 eval { "5$i" =~ /$o/ };
1395 like($@, ($i == 3 ? qr/^$/ : qr/no method found,/),
1396 "Qr$i bare - with use re eval");
1400 ok("A5$i" =~ /^A$o$/, "Qr$i");
1401 ok("5$i" =~ /$o/, "Qr$i bare");
1405 my $o = bless [ bless [], "Qr1" ], 'Qr_indirect';
1406 ok("A51" =~ /^A$o/, "Qr_indirect");
1407 ok("51" =~ /$o/, "Qr_indirect bare");
1410 { # Various flags weren't being set when a [] is optimized into an
1414 my $sharp_s = uni_to_native("\xdf");
1415 ok("\x{017F}\x{017F}" =~ qr/^[$sharp_s]?$/i, "[] to EXACTish optimization");
1419 for my $char (":", uni_to_native("\x{f7}"), "\x{2010}") {
1420 my $utf8_char = $char;
1421 utf8::upgrade($utf8_char);
1422 my $display = $char;
1423 $display = display($display);
1424 my $utf8_display = "utf8::upgrade(\"$display\")";
1426 like($char, qr/^$char?$/, "\"$display\" =~ /^$display?\$/");
1427 like($char, qr/^$utf8_char?$/, "my \$p = \"$display\"; utf8::upgrade(\$p); \"$display\" =~ /^\$p?\$/");
1428 like($utf8_char, qr/^$char?$/, "my \$c = \"$display\"; utf8::upgrade(\$c); \"\$c\" =~ /^$display?\$/");
1429 like($utf8_char, qr/^$utf8_char?$/, "my \$c = \"$display\"; utf8::upgrade(\$c); my \$p = \"$display\"; utf8::upgrade(\$p); \"\$c\" =~ /^\$p?\$/");
1434 # #116148: Pattern utf8ness sticks around globally
1435 # the utf8 in the first match was sticking around for the second
1438 use feature 'unicode_strings';
1444 ok("Perl" =~ /P.*$/i, '#116148');
1447 { # 118297: Mixing up- and down-graded strings in regex
1448 utf8::upgrade(my $u = "\x{e5}");
1449 utf8::downgrade(my $d = "\x{e5}");
1451 local $SIG{__WARN__} = sub { $warned++ if $_[0] =~ /\AMalformed UTF-8/ };
1453 ok(!$warned, "no warnings when interpolating mixed up-/downgraded strings in pattern");
1454 my $c = "\x{e5}\x{e5}";
1455 utf8::downgrade($c);
1456 like($c, $re, "mixed up-/downgraded pattern matches downgraded string");
1458 like($c, $re, "mixed up-/downgraded pattern matches upgraded string");
1462 # if we have 87 capture buffers defined then \87 should refer to the 87th.
1463 # test that this is true for 1..100
1464 # Note that this test causes the engine to recurse at runtime, and
1465 # hence use a lot of C stack.
1466 for my $i (1..100) {
1468 $capture= "($capture)" for 1 .. $i;
1469 for my $mid ("","b") {
1470 my $str= "a${mid}a";
1471 my $backref= "\\$i";
1473 ok($str=~/$capture$mid$backref/,"\\$i works with $i buffers '$str'=~/...$mid$backref/");
1476 is("$@","","\\$i works with $i buffers works with $i buffers '$str'=~/...$mid$backref/");
1482 # this mixture of readonly (not COWable) and COWable strings
1483 # messed up the capture buffers under COW. The actual test results
1484 # are incidental; the issue is was an AddressSanitizer failure
1488 for ($c, 'C', $c, 'DE') {
1489 ok(/(.)/, "COWable match");
1498 # /[#$x]/x didn't interpolate the var $x.
1501 $s =~ s/[a#$b%]/X/g;
1502 is ($s, 'XbXX$XX&', 'RT #45667 without /x');
1504 $s =~ s/[a#$b%]/X/gx;
1505 is ($s, 'XbXX$XX&', 'RT #45667 with /x');
1509 no warnings "uninitialized";
1513 pass('no crash with /@a/ when array has nonexistent elems');
1517 is runperl(prog => 'delete $::{qq-\cR-}; //; print qq-ok\n-'),
1519 'deleting *^R does not result in crashes';
1521 *^R = *caretRglobwithnoscalar;
1523 is $^R, 42, 'assigning to *^R does not result in a crash';
1527 .' q-..- =~ /(??{undef *^R;q--})(?{42})/; '
1532 'undefining *^R within (??{}) does not result in a crash';
1535 SKIP: { # Test literal range end point special handling
1536 unless ($::IS_EBCDIC) {
1537 skip "Valid only for EBCDIC", 24;
1540 like("\x89", qr/[i-j]/, '"\x89" should match [i-j]');
1541 unlike("\x8A", qr/[i-j]/, '"\x8A" shouldnt match [i-j]');
1542 unlike("\x90", qr/[i-j]/, '"\x90" shouldnt match [i-j]');
1543 like("\x91", qr/[i-j]/, '"\x91" should match [i-j]');
1545 like("\x89", qr/[i-\N{LATIN SMALL LETTER J}]/, '"\x89" should match [i-\N{LATIN SMALL LETTER J}]');
1546 unlike("\x8A", qr/[i-\N{LATIN SMALL LETTER J}]/, '"\x8A" shouldnt match [i-\N{LATIN SMALL LETTER J}]');
1547 unlike("\x90", qr/[i-\N{LATIN SMALL LETTER J}]/, '"\x90" shouldnt match [i-\N{LATIN SMALL LETTER J}]');
1548 like("\x91", qr/[i-\N{LATIN SMALL LETTER J}]/, '"\x91" should match [i-\N{LATIN SMALL LETTER J}]');
1550 like("\x89", qr/[i-\N{U+6A}]/, '"\x89" should match [i-\N{U+6A}]');
1551 unlike("\x8A", qr/[i-\N{U+6A}]/, '"\x8A" shouldnt match [i-\N{U+6A}]');
1552 unlike("\x90", qr/[i-\N{U+6A}]/, '"\x90" shouldnt match [i-\N{U+6A}]');
1553 like("\x91", qr/[i-\N{U+6A}]/, '"\x91" should match [i-\N{U+6A}]');
1555 like("\x89", qr/[\N{U+69}-\N{U+6A}]/, '"\x89" should match [\N{U+69}-\N{U+6A}]');
1556 unlike("\x8A", qr/[\N{U+69}-\N{U+6A}]/, '"\x8A" shouldnt match [\N{U+69}-\N{U+6A}]');
1557 unlike("\x90", qr/[\N{U+69}-\N{U+6A}]/, '"\x90" shouldnt match [\N{U+69}-\N{U+6A}]');
1558 like("\x91", qr/[\N{U+69}-\N{U+6A}]/, '"\x91" should match [\N{U+69}-\N{U+6A}]');
1560 like("\x89", qr/[i-\x{91}]/, '"\x89" should match [i-\x{91}]');
1561 like("\x8A", qr/[i-\x{91}]/, '"\x8A" should match [i-\x{91}]');
1562 like("\x90", qr/[i-\x{91}]/, '"\x90" should match [i-\x{91}]');
1563 like("\x91", qr/[i-\x{91}]/, '"\x91" should match [i-\x{91}]');
1565 # Need to use eval, because tries to compile on ASCII platforms even
1566 # though the tests are skipped, and fails because 0x89-j is an illegal
1568 like("\x89", eval 'qr/[\x{89}-j]/', '"\x89" should match [\x{89}-j]');
1569 like("\x8A", eval 'qr/[\x{89}-j]/', '"\x8A" should match [\x{89}-j]');
1570 like("\x90", eval 'qr/[\x{89}-j]/', '"\x90" should match [\x{89}-j]');
1571 like("\x91", eval 'qr/[\x{89}-j]/', '"\x91" should match [\x{89}-j]');
1574 # These are based on looking at the code in regcomp.c
1575 # We don't look for specific code, just the existence of an SSC
1576 foreach my $re (qw( qr/a?c/
1586 skip "no re-debug under miniperl" if is_miniperl;
1588 use re qw(Debug COMPILE);
1591 fresh_perl_like($prog, qr/synthetic stclass/, { stderr=>1 }, "$re generates a synthetic start class");
1596 like "\x{AA}", qr/a?[\W_]/d, "\\W with /d synthetic start class works";
1600 skip("Tests are ASCII-centric, some would fail on EBCDIC", 12) if $::IS_EBCDIC;
1602 # Verify that the very last Latin-1 U+00FF
1603 # (LATIN SMALL LETTER Y WITH DIAERESIS)
1604 # and its UPPER counterpart (U+0178 which is pure Unicode),
1605 # and likewise for the very first pure Unicode
1606 # (LATIN CAPITAL LETTER A WITH MACRON) fold-match properly,
1607 # and there are no off-by-one logic errors in the transition zone.
1609 ok("\xFF" =~ /\xFF/i, "Y WITH DIAERESIS l =~ l");
1610 ok("\xFF" =~ /\x{178}/i, "Y WITH DIAERESIS l =~ u");
1611 ok("\x{178}" =~ /\xFF/i, "Y WITH DIAERESIS u =~ l");
1612 ok("\x{178}" =~ /\x{178}/i, "Y WITH DIAERESIS u =~ u");
1614 # U+00FF with U+05D0 (non-casing Hebrew letter).
1615 ok("\xFF\x{5D0}" =~ /\xFF\x{5D0}/i, "Y WITH DIAERESIS l =~ l");
1616 ok("\xFF\x{5D0}" =~ /\x{178}\x{5D0}/i, "Y WITH DIAERESIS l =~ u");
1617 ok("\x{178}\x{5D0}" =~ /\xFF\x{5D0}/i, "Y WITH DIAERESIS u =~ l");
1618 ok("\x{178}\x{5D0}" =~ /\x{178}\x{5D0}/i, "Y WITH DIAERESIS u =~ u");
1621 ok("\x{100}" =~ /\x{100}/i, "A WITH MACRON u =~ u");
1622 ok("\x{100}" =~ /\x{101}/i, "A WITH MACRON u =~ l");
1623 ok("\x{101}" =~ /\x{100}/i, "A WITH MACRON l =~ u");
1624 ok("\x{101}" =~ /\x{101}/i, "A WITH MACRON l =~ l");
1629 ok("abc" =~ /a
\85b
\85c/x, "NEL is white-space under /x");
1633 ok('a(b)c' =~ qr(a\(b\)c), "'\\(' is a literal in qr(...)");
1634 ok('a[b]c' =~ qr[a\[b\]c], "'\\[' is a literal in qr[...]");
1635 ok('a{3}c' =~ qr{a\{3\}c}, # Only failed when { could be a meta
1636 "'\\{' is a literal in qr{...}, where it could be a quantifier");
1638 # This one is for completeness
1639 ok('a<b>c' =~ qr<a\<b\>c>, "'\\<' is a literal in qr<...>)");
1642 { # Was getting optimized into EXACT (non-folding node)
1645 like("X", qr/$x/, "UTF-8 of /[x]/i matches upper case");
1648 { # make sure we get an error when \p{} cannot load Unicode tables
1649 fresh_perl_like(<<' prog that cannot load uni tables',
1652 require utf8; require 'utf8_heavy.pl';
1656 if ($name =~ /(\p{IsUpper}) (\p{IsUpper})/){
1657 print "It's good! >$1< >$2<\n";
1659 print "It's not good...\n";
1661 prog that cannot load uni tables
1662 qr/^Can't locate unicore\/Heavy\.pl(?x:
1663 )|^Can't find Unicode property definition/,
1665 '\p{} should not fail silently when uni tables evanesce');
1668 { # Special handling of literal-ended ranges in [...] was breaking this
1670 like("ÿ", qr/[ÿ-ÿ]/, "\"ÿ\" should match [ÿ-ÿ]");
1674 like("TffffffffffffTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT5TTTTTTTTTTTTTTTTTTTTTTTTT3TTgTTTTTTTTTTTTTTTTTTTTT2TTTTTTTTTTTTTTTTTTTTTTTHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHiHHHHHHHfffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff&ffff", qr/TffffffffffffTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT5TTTTTTTTTTTTTTTTTTTTTTTTT3TTgTTTTTTTTTTTTTTTTTTTTT2TTTTTTTTTTTTTTTTTTTTTTTHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHiHHHHHHHfffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff&ffff/il, "");
1675 like("TffffffffffffT\x{100}TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT5TTTTTTTTTTTTTTTTTTTTTTTTT3TTgTTTTTTTTTTTTTTTTTTTTT2TTTTTTTTTTTTTTTTTTTTTTTHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHiHHHHHHHfffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff&ffff", qr/TffffffffffffT\x{100}TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT5TTTTTTTTTTTTTTTTTTTTTTTTT3TTgTTTTTTTTTTTTTTTTTTTTT2TTTTTTTTTTTTTTTTTTTTTTTHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHiHHHHHHHfffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff&ffff/il, "");
1679 my($s, $x, @x) = ('abc', 'a', 'd');
1680 my $long = 'b' x 2000;
1681 my $eval = q{$s =~ m{$x[bbb]c} ? 1 : 0};
1682 $eval =~ s{bbb}{$long};
1683 my $match = eval $eval;
1684 ok(1, "did not crash");
1685 ok($match, "[bbb...] resolved as character class, not subscript");
1689 for my $pat ('(??', '(?P', '(?i-') {
1690 eval qq{ qr/$pat/ };
1691 ok(1, "qr/$pat/ did not crash");
1692 eval qq{ qr/${pat}\x{123}/ };
1694 like($e, qr{\x{123}},
1695 "qr/${pat}x/ shows x in error even if it's a wide character");
1700 # Expect one of these sizes to cause overflow and wrap to negative
1701 for my $bits (32, 64) {
1702 my $wrapneg = 2 ** ($bits - 2) * 3;
1703 for my $sign ('', '-') {
1704 my $pat = sprintf "qr/(?%s%u)/", $sign, $wrapneg;
1706 ok(1, "big backref $pat did not crash");
1711 # Test that we handle qr/\8888888/ and variants without an infinite loop,
1712 # we use a test within a test so we can todo it, and make sure we don't
1713 # infinite loop our tests.
1714 # NOTE - Do not put quotes in the code!
1715 # NOTE - We have to triple escape the backref in the pattern below.
1717 BEGIN{require q(./test.pl);}
1719 for my $len (1 .. 20) {
1720 my $eights= q(8) x $len;
1721 eval qq{ qr/\\\\$eights/ };
1723 print q(No infinite loop here!);
1725 fresh_perl_is($code, "No infinite loop here!", {},
1726 "test that we handle things like m/\\888888888/ without infinite loops" );
1729 { # Test that we handle some malformed UTF-8 without looping [perl
1733 BEGIN{require q(./test.pl);}
1734 use Encode qw(_utf8_on);
1735 # \x80 and \x41 are continuation bytes in their respective
1737 my $malformed = (ord("A") == 65) ? "a\x80\n" : "a\x41\n";
1738 utf8::downgrade($malformed);
1739 _utf8_on($malformed);
1741 $malformed =~ /(\n\r|\r)$/;
1742 print q(No infinite loop here!);
1744 fresh_perl_like($code, qr/Malformed UTF-8 character/, {},
1745 "test that we handle some UTF-8 malformations without looping" );
1749 # [perl #123843] hits SEGV trying to compile this pattern
1751 eval q{ ($match) = ("xxyxxyxy" =~ m{(x+(y(?1))*)}) };
1752 ok(1, "compiled GOSUB in CURLYM ok");
1753 is($match, 'xxyxxyx', "matched GOSUB in CURLYM");
1757 # [perl #123852] doesn't avoid all the capture-related work with
1758 # //n, leading to possible memory corruption
1759 eval q{ qr{()(?1)}n };
1761 ok(1, "qr{()(?1)}n didn't crash");
1762 like($error, qr{Reference to nonexistent group},
1763 'gave appropriate error for qr{()(?1)}n');
1767 # [perl #126406] panic with unmatchable quantifier
1769 no warnings "regexp";
1770 "" =~ m/(.0\N{6,0}0\N{6,0}000000000000000000000000000000000)/;
1772 fresh_perl_is($code, "", {},
1773 "perl [#126406] panic");
1776 my $bug="[perl #126182]"; # test for infinite pattern recursion
1778 [ 'q(a)=~/(.(?2))((?<=(?=(?1)).))/', "died", "look ahead left recursion fails fast" ],
1779 [ 'q(aa)=~/(?R)a/', "died", "left-recursion fails fast", ],
1780 [ 'q(bbaa)=~/(?&x)(?(DEFINE)(?<x>(?&y)*a)(?<y>(?&x)*b))/',
1781 "died", "inter-cyclic optional left recursion dies" ],
1782 [ 'q(abc) =~ /a((?1)?)c/', "died", "optional left recursion dies" ],
1783 [ 'q(abc) =~ /a((?1)??)c/', "died", "min mod left recursion dies" ],
1784 [ 'q(abc) =~ /a((?1)*)c/', "died", "* left recursion dies" ],
1785 [ 'q(abc) =~ /a((?1)+)c/', "died", "+ left recursion dies" ],
1786 [ 'q(abc) =~ /a((?1){0,3})c/', "died", "{0,3} left recursion fails fast" ],
1788 [ 'q(aaabbb)=~/a(?R)?b/', "matched", "optional self recursion works" ],
1789 [ '"((5maa-maa)(maa-3maa))" =~ /(\\\\((?:[^()]++|(?0))*+\\\\))/', "matched",
1790 "recursion and possessive captures", "((5maa-maa)(maa-3maa))"],
1791 [ '"((5maa-maa)(maa-3maa))" =~ /(\\\\((?:[^()]++|(?1))*+\\\\))/', "matched",
1792 "recursion and possessive captures", "((5maa-maa)(maa-3maa))"],
1793 [ '"((5maa-maa)(maa-3maa))" =~ /(\\\\((?:[^()]+|(?0))*\\\\))/', "matched",
1794 "recursion and possessive captures", "((5maa-maa)(maa-3maa))"],
1795 [ '"((5maa-maa)(maa-3maa))" =~ /(\\\\((?:[^()]+|(?1))*\\\\))/', "matched",
1796 "recursion and possessive captures", "((5maa-maa)(maa-3maa))"],
1798 my ($expr, $expect, $test_name, $cap1)= @$tuple;
1799 # avoid quotes in this code!
1801 BEGIN{require q(./test.pl);}
1803 my $status= eval(q{ !(' . $expr . ') ? q(failed) : ' .
1804 ($cap1 ? '($1 ne q['.$cap1.']) ? qq(badmatch:$1) : ' : '') .
1806 || ( ( $@ =~ /Infinite recursion/ ) ? qq(died) : q(strange-death) );
1809 fresh_perl_is($code, $expect, {}, "$bug - $test_name" );
1814 BEGIN{require q(test.pl);}
1816 $SIG{ALRM} = sub {print "Timeout\n"; exit(1)};
1818 $_ = "a" x 1000 . "b" x 1000 . "c" x 1000;
1820 ',"Timeout",{},"Test Perl 73464")
1823 { # [perl #128686], crashed the the interpreter
1824 my $AE = chr utf8::unicode_to_native(0xC6);
1825 my $ae = chr utf8::unicode_to_native(0xE6);
1826 my $re = qr/[$ae\s]/i;
1827 ok($AE !~ $re, '/[\xE6\s]/i doesn\'t match \xC6 when not in UTF-8');
1829 ok($AE =~ $re, '/[\xE6\s]/i matches \xC6 when in UTF-8');
1833 is(0+("\n" =~ m'\n'), 1, q|m'\n' should interpolate escapes|);
1838 ok( $str =~ m{^(a|a\x{b6})$}, "fix [perl #129950] - latin1 case" );
1839 utf8::upgrade($str);
1840 ok( $str =~ m{^(a|a\x{b6})$}, "fix [perl #129950] - utf8 case" );
1843 my $got= run_perl( switches => [ '-l' ], prog => <<'EOF_CODE' );
1847 my $p=qr/(?{ s!!x! })/;
1859 $error = $died ? ($@ || qq(Zombie)) : qq(none);
1860 print $died ? qq(died) : qq(lived);
1861 print qq(Error: $@);
1863 my @got= split /\n/, $got;
1864 is($got[0],"ab","empty pattern in regex codeblock: got expected start string");
1866 "empty pattern in regex codeblock: first subst with no last-match worked right");
1867 is($got[2],"xxb","empty pattern in regex codeblock: second subst worked right");
1868 is($got[3],"xxx","empty pattern in regex codeblock: third subst worked right");
1869 is($got[4],"died","empty pattern in regex codeblock: died as expected");
1870 like($got[5],qr/Error: Infinite recursion via empty pattern/,
1871 "empty pattern in regex codeblock: produced the right exception message" );
1874 # [perl #130495] /x comment skipping stopped a byte short, leading
1875 # to assertion failure or 'malformed utf-8 character" warning
1877 "use utf8; m{a#\x{124}}x", '', {wide_chars => 1},
1878 '[perl #130495] utf-8 character at end of /x comment should not misparse',
1882 # [perl #130522] causes out-of-bounds read detected by clang with
1883 # address=sanitized when length of the STCLASS string is greater than
1884 # length of target string.
1885 my $re = qr{(?=\0z)\0?z?$}i;
1886 my($yes, $no) = (1, "");
1888 [ $no, undef, '<undef>' ],
1890 [ $no, "\0", '\0' ],
1891 [ $yes, "\0z", '\0z' ],
1892 [ $no, "\0z\0", '\0z\0' ],
1893 [ $yes, "\0z\n", '\0z\n' ],
1895 my($result, $target, $disp) = @$test;
1896 no warnings qw/uninitialized/;
1897 is($target =~ $re, $result, "[perl #130522] with target '$disp'");
1901 # [perl #129377] backref to an unmatched capture should not cause
1902 # reading before start of string.
1904 skip "no re-debug under miniperl" if is_miniperl;
1906 use re qw(Debug EXECUTE);
1907 "x" =~ m{ () y | () \1 }x;
1909 fresh_perl_like($prog, qr{
1911 }msx, { stderr => 1 }, "Offsets in debug output are not negative");
1916 fresh_perl_is("BEGIN{\$^H=0x200000}\ns/[(?{//xx",
1917 "Unmatched [ in regex; marked by <-- HERE in m/[ <-- HERE (?{/ at (eval 1) line 1.\n",
1918 {}, "buffer overflow for regexp component");
1921 # [perl #129281] buffer write overflow, detected by ASAN, valgrind
1922 fresh_perl_is('/0(?0)|^*0(?0)|^*(^*())0|/', '', {}, "don't bump whilem_c too much");
1926 # RT #131575 intuit skipping back from the end to find the highest
1927 # possible start point, was potentially hopping back beyond pos()
1928 # and crashing by calling fbm_instr with a negative length
1930 my $text = "=t=\x{5000}";
1932 ok(scalar($text !~ m{(~*=[a-z]=)}g), "RT #131575");
1935 fresh_perl_is('"AA" =~ m/AA{1,0}/','',{},"handle OPFAIL insert properly");
1938 } # End of sub run_tests
1943 # ex: set ts=8 sts=4 sw=4 et: