This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Tests for bug 77414
[perl5.git] / t / re / pat.t
CommitLineData
8d063cd8 1#!./perl
8d37f932
DD
2#
3# This is a home for regular expression tests that don't fit into
67a2b8c6 4# the format supported by re/regexp.t. If you want to add a test
ff3f963a
KW
5# that does fit that format, add it to re/re_tests, not here. Tests for \N
6# should be added here because they are treated as single quoted strings
7# there, which means they avoid the lexer which otherwise would look at them.
8d063cd8 8
84281c31
A
9use strict;
10use warnings;
11use 5.010;
12
13
14sub run_tests;
15
9133bbab 16$| = 1;
3568d838 17
8d37f932 18
e4d48cc9
GS
19BEGIN {
20 chdir 't' if -d 't';
9d45b377
YO
21 @INC = ('../lib','.');
22 do "re/ReTest.pl" or die $@;
e4d48cc9 23}
84281c31 24
84281c31 25
c920e018 26plan tests => 366; # Update this when adding/deleting tests.
b7a35066 27
9d45b377 28run_tests() unless caller;
b7a35066 29
84281c31
A
30#
31# Tests start here.
32#
33sub run_tests {
0ef3e39e 34
84281c31 35 {
b485d051 36
84281c31 37 my $x = "abc\ndef\n";
fd291da9 38
84281c31
A
39 ok $x =~ /^abc/, qq ["$x" =~ /^abc/];
40 ok $x !~ /^def/, qq ["$x" !~ /^def/];
fd291da9 41
84281c31
A
42 # used to be a test for $*
43 ok $x =~ /^def/m, qq ["$x" =~ /^def/m];
fd291da9 44
84281c31
A
45 nok $x =~ /^xxx/, qq ["$x" =~ /^xxx/];
46 nok $x !~ /^abc/, qq ["$x" !~ /^abc/];
fd291da9 47
84281c31
A
48 ok $x =~ /def/, qq ["$x" =~ /def/];
49 nok $x !~ /def/, qq ["$x" !~ /def/];
4765795a 50
84281c31
A
51 ok $x !~ /.def/, qq ["$x" !~ /.def/];
52 nok $x =~ /.def/, qq ["$x" =~ /.def/];
4765795a 53
84281c31
A
54 ok $x =~ /\ndef/, qq ["$x" =~ /\ndef/];
55 nok $x !~ /\ndef/, qq ["$x" !~ /\ndef/];
56 }
4765795a 57
84281c31
A
58 {
59 $_ = '123';
60 ok /^([0-9][0-9]*)/, qq [\$_ = '$_'; /^([0-9][0-9]*)/];
61 }
f9969324 62
84281c31
A
63 {
64 $_ = 'aaabbbccc';
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 nok /a+b?c+/, qq [\$_ = '$_'; /a+b?c+/];
69
70 $_ = 'aaabccc';
71 ok /a+b?c+/, qq [\$_ = '$_'; /a+b?c+/];
72 ok /a*b?c*/, qq [\$_ = '$_'; /a*b?c*/];
73
74 $_ = 'aaaccc';
75 ok /a*b?c*/, qq [\$_ = '$_'; /a*b?c*/];
76 nok /a*b+c*/, qq [\$_ = '$_'; /a*b+c*/];
77
78 $_ = 'abcdef';
79 ok /bcd|xyz/, qq [\$_ = '$_'; /bcd|xyz/];
80 ok /xyz|bcd/, qq [\$_ = '$_'; /xyz|bcd/];
81 ok m|bc/*d|, qq [\$_ = '$_'; m|bc/*d|];
82 ok /^$_$/, qq [\$_ = '$_'; /^\$_\$/];
4765795a 83 }
4765795a 84
84281c31
A
85 {
86 # used to be a test for $*
87 ok "ab\ncd\n" =~ /^cd/m, qq ["ab\ncd\n" =~ /^cd/m];
88 }
4765795a 89
84281c31
A
90 {
91 our %XXX = map {($_ => $_)} 123, 234, 345;
92
93 our @XXX = ('ok 1','not ok 1', 'ok 2','not ok 2','not ok 3');
94 while ($_ = shift(@XXX)) {
95 my $f = index ($_, 'not') >= 0 ? \&nok : \&ok;
96 my $r = ?(.*)?;
97 &$f ($r, "?(.*)?");
98 /not/ && reset;
99 if (/not ok 2/) {
100 if ($^O eq 'VMS') {
101 $_ = shift(@XXX);
102 }
103 else {
104 reset 'X';
105 }
106 }
107 }
4765795a 108
84281c31
A
109 SKIP: {
110 if ($^O eq 'VMS') {
111 skip "Reset 'X'", 1;
112 }
113 ok !keys %XXX, "%XXX is empty";
114 }
4765795a 115
84281c31 116 }
4765795a 117
84281c31
A
118 {
119 local $Message = "Test empty pattern";
120 my $xyz = 'xyz';
121 my $cde = 'cde';
122
123 $cde =~ /[^ab]*/;
124 $xyz =~ //;
125 iseq $&, $xyz;
126
127 my $foo = '[^ab]*';
128 $cde =~ /$foo/;
129 $xyz =~ //;
130 iseq $&, $xyz;
131
132 $cde =~ /$foo/;
133 my $null;
134 no warnings 'uninitialized';
135 $xyz =~ /$null/;
136 iseq $&, $xyz;
137
138 $null = "";
139 $xyz =~ /$null/;
140 iseq $&, $xyz;
141 }
4765795a 142
84281c31
A
143 {
144 local $Message = q !Check $`, $&, $'!;
145 $_ = 'abcdefghi';
0f289c68 146 /def/; # optimized up to cmd
84281c31 147 iseq "$`:$&:$'", 'abc:def:ghi';
4765795a 148
84281c31 149 no warnings 'void';
0f289c68 150 /cde/ + 0; # optimized only to spat
84281c31 151 iseq "$`:$&:$'", 'ab:cde:fghi';
4765795a 152
0f289c68 153 /[d][e][f]/; # not optimized
84281c31
A
154 iseq "$`:$&:$'", 'abc:def:ghi';
155 }
4765795a 156
84281c31
A
157 {
158 $_ = 'now is the {time for all} good men to come to.';
159 / {([^}]*)}/;
160 iseq $1, 'time for all', "Match braces";
161 }
4765795a 162
84281c31
A
163 {
164 local $Message = "{N,M} quantifier";
165 $_ = 'xxx {3,4} yyy zzz';
166 ok /( {3,4})/;
167 iseq $1, ' ';
168 ok !/( {4,})/;
169 ok /( {2,3}.)/;
170 iseq $1, ' y';
171 ok /(y{2,3}.)/;
172 iseq $1, 'yyy ';
173 ok !/x {3,4}/;
174 ok !/^xxx {3,4}/;
175 }
4765795a 176
84281c31
A
177 {
178 local $Message = "Test /g";
179 local $" = ":";
180 $_ = "now is the time for all good men to come to.";
181 my @words = /(\w+)/g;
182 my $exp = "now:is:the:time:for:all:good:men:to:come:to";
4765795a 183
84281c31 184 iseq "@words", $exp;
4765795a 185
84281c31
A
186 @words = ();
187 while (/\w+/g) {
188 push (@words, $&);
189 }
190 iseq "@words", $exp;
4765795a 191
84281c31
A
192 @words = ();
193 pos = 0;
194 while (/to/g) {
195 push(@words, $&);
196 }
197 iseq "@words", "to:to";
4765795a 198
84281c31
A
199 pos $_ = 0;
200 @words = /to/g;
201 iseq "@words", "to:to";
202 }
4765795a 203
84281c31
A
204 {
205 $_ = "abcdefghi";
206
207 my $pat1 = 'def';
208 my $pat2 = '^def';
209 my $pat3 = '.def.';
210 my $pat4 = 'abc';
211 my $pat5 = '^abc';
212 my $pat6 = 'abc$';
213 my $pat7 = 'ghi';
214 my $pat8 = '\w*ghi';
215 my $pat9 = 'ghi$';
216
217 my $t1 = my $t2 = my $t3 = my $t4 = my $t5 =
218 my $t6 = my $t7 = my $t8 = my $t9 = 0;
219
220 for my $iter (1 .. 5) {
221 $t1++ if /$pat1/o;
222 $t2++ if /$pat2/o;
223 $t3++ if /$pat3/o;
224 $t4++ if /$pat4/o;
225 $t5++ if /$pat5/o;
226 $t6++ if /$pat6/o;
227 $t7++ if /$pat7/o;
228 $t8++ if /$pat8/o;
229 $t9++ if /$pat9/o;
230 }
231 my $x = "$t1$t2$t3$t4$t5$t6$t7$t8$t9";
232 iseq $x, '505550555', "Test /o";
233 }
4765795a 234
4765795a 235
84281c31
A
236 SKIP: {
237 my $xyz = 'xyz';
238 ok "abc" =~ /^abc$|$xyz/, "| after \$";
4765795a 239
84281c31
A
240 # perl 4.009 says "unmatched ()"
241 local $Message = '$ inside ()';
4765795a 242
84281c31
A
243 my $result;
244 eval '"abc" =~ /a(bc$)|$xyz/; $result = "$&:$1"';
245 iseq $@, "" or skip "eval failed", 1;
246 iseq $result, "abc:bc";
247 }
4765795a 248
4765795a 249
84281c31
A
250 {
251 local $Message = "Scalar /g";
252 $_ = "abcfooabcbar";
253
254 ok /abc/g && $` eq "";
255 ok /abc/g && $` eq "abcfoo";
256 ok !/abc/g;
257
258 local $Message = "Scalar /gi";
259 pos = 0;
260 ok /ABC/gi && $` eq "";
261 ok /ABC/gi && $` eq "abcfoo";
262 ok !/ABC/gi;
263
264 local $Message = "Scalar /g";
265 pos = 0;
266 ok /abc/g && $' eq "fooabcbar";
267 ok /abc/g && $' eq "bar";
268
269 $_ .= '';
270 my @x = /abc/g;
271 iseq @x, 2, "/g reset after assignment";
4765795a 272 }
4765795a 273
84281c31
A
274 {
275 local $Message = '/g, \G and pos';
276 $_ = "abdc";
277 pos $_ = 2;
278 /\Gc/gc;
279 iseq pos $_, 2;
280 /\Gc/g;
281 ok !defined pos $_;
282 }
4765795a 283
84281c31
A
284 {
285 local $Message = '(?{ })';
286 our $out = 1;
287 'abc' =~ m'a(?{ $out = 2 })b';
288 iseq $out, 2;
289
290 $out = 1;
291 'abc' =~ m'a(?{ $out = 3 })c';
292 iseq $out, 1;
293 }
4765795a 294
4765795a 295
84281c31
A
296 {
297 $_ = 'foobar1 bar2 foobar3 barfoobar5 foobar6';
298 my @out = /(?<!foo)bar./g;
299 iseq "@out", 'bar2 barf', "Negative lookbehind";
300 }
4765795a 301
84281c31
A
302 {
303 local $Message = "REG_INFTY tests";
304 # Tests which depend on REG_INFTY
305 $::reg_infty = $Config {reg_infty} // 32767;
306 $::reg_infty_m = $::reg_infty - 1;
307 $::reg_infty_p = $::reg_infty + 1;
308 $::reg_infty_m = $::reg_infty_m; # Surpress warning.
309
310 # As well as failing if the pattern matches do unexpected things, the
311 # next three tests will fail if you should have picked up a lower-than-
312 # default value for $reg_infty from Config.pm, but have not.
313
314 eval_ok q (('aaa' =~ /(a{1,$::reg_infty_m})/)[0] eq 'aaa');
315 eval_ok q (('a' x $::reg_infty_m) =~ /a{$::reg_infty_m}/);
316 eval_ok q (('a' x ($::reg_infty_m - 1)) !~ /a{$::reg_infty_m}/);
317 eval "'aaa' =~ /a{1,$::reg_infty}/";
318 ok $@ =~ /^\QQuantifier in {,} bigger than/;
319 eval "'aaa' =~ /a{1,$::reg_infty_p}/";
320 ok $@ =~ /^\QQuantifier in {,} bigger than/;
4765795a 321 }
8269fa76 322
84281c31
A
323 {
324 # Poke a couple more parse failures
325 my $context = 'x' x 256;
326 eval qq("${context}y" =~ /(?<=$context)y/);
327 ok $@ =~ /^\QLookbehind longer than 255 not/, "Lookbehind limit";
328 }
8269fa76 329
84281c31
A
330 {
331 # Long Monsters
332 local $Message = "Long monster";
333 for my $l (125, 140, 250, 270, 300000, 30) { # Ordered to free memory
334 my $a = 'a' x $l;
335 local $Error = "length = $l";
336 ok "ba$a=" =~ /a$a=/;
337 nok "b$a=" =~ /a$a=/;
338 ok "b$a=" =~ /ba+=/;
339
84281c31
A
340 ok "ba$a=" =~ /b(?:a|b)+=/;
341 }
342 }
8269fa76 343
b8ef571c 344
84281c31
A
345 {
346 # 20000 nodes, each taking 3 words per string, and 1 per branch
347 my $long_constant_len = join '|', 12120 .. 32645;
348 my $long_var_len = join '|', 8120 .. 28645;
349 my %ans = ( 'ax13876y25677lbc' => 1,
350 'ax13876y25677mcb' => 0, # not b.
351 'ax13876y35677nbc' => 0, # Num too big
352 'ax13876y25677y21378obc' => 1,
0f289c68 353 'ax13876y25677y21378zbc' => 0, # Not followed by [k-o]
84281c31
A
354 'ax13876y25677y21378y21378kbc' => 1,
355 'ax13876y25677y21378y21378kcb' => 0, # Not b.
356 'ax13876y25677y21378y21378y21378kbc' => 0, # 5 runs
357 );
358
359 local $Message = "20000 nodes";
360 for (keys %ans) {
361 local $Error = "const-len '$_'";
362 ok !($ans{$_} xor /a(?=([yx]($long_constant_len)){2,4}[k-o]).*b./o);
363
364 local $Error = "var-len '$_'";
365 ok !($ans{$_} xor /a(?=([yx]($long_var_len)){2,4}[k-o]).*b./o);
366 }
b8ef571c 367 }
209a9bc1 368
84281c31
A
369 {
370 local $Message = "Complicated backtracking";
371 $_ = " a (bla()) and x(y b((l)u((e))) and b(l(e)e)e";
372 my $expect = "(bla()) ((l)u((e))) (l(e)e)";
373
374 use vars '$c';
375 sub matchit {
376 m/
377 (
378 \(
0f289c68 379 (?{ $c = 1 }) # Initialize
84281c31
A
380 (?:
381 (?(?{ $c == 0 }) # PREVIOUS iteration was OK, stop the loop
382 (?!
0f289c68
YO
383 ) # Fail: will unwind one iteration back
384 )
84281c31 385 (?:
0f289c68 386 [^()]+ # Match a big chunk
84281c31
A
387 (?=
388 [()]
0f289c68 389 ) # Do not try to match subchunks
84281c31
A
390 |
391 \(
392 (?{ ++$c })
393 |
394 \)
395 (?{ --$c })
396 )
0f289c68 397 )+ # This may not match with different subblocks
84281c31
A
398 )
399 (?(?{ $c != 0 })
400 (?!
0f289c68
YO
401 ) # Fail
402 ) # Otherwise the chunk 1 may succeed with $c>0
84281c31
A
403 /xg;
404 }
3568d838 405
84281c31
A
406 my @ans = ();
407 my $res;
408 push @ans, $res while $res = matchit;
409 iseq "@ans", "1 1 1";
3568d838 410
84281c31
A
411 @ans = matchit;
412 iseq "@ans", $expect;
3568d838 413
84281c31
A
414 local $Message = "Recursion with (??{ })";
415 our $matched;
416 $matched = qr/\((?:(?>[^()]+)|(??{$matched}))*\)/;
3568d838 417
84281c31
A
418 @ans = my @ans1 = ();
419 push (@ans, $res), push (@ans1, $&) while $res = m/$matched/g;
3568d838 420
84281c31
A
421 iseq "@ans", "1 1 1";
422 iseq "@ans1", $expect;
3568d838 423
84281c31
A
424 @ans = m/$matched/g;
425 iseq "@ans", $expect;
3568d838 426
84281c31 427 }
3568d838 428
84281c31
A
429 {
430 ok "abc" =~ /^(??{"a"})b/, '"abc" =~ /^(??{"a"})b/';
431 }
3568d838 432
84281c31 433 {
0f289c68 434 my @ans = ('a/b' =~ m%(.*/)?(.*)%); # Stack may be bad
84281c31
A
435 iseq "@ans", 'a/ b', "Stack may be bad";
436 }
3568d838 437
84281c31
A
438 {
439 local $Message = "Eval-group not allowed at runtime";
440 my $code = '{$blah = 45}';
441 our $blah = 12;
442 eval { /(?$code)/ };
443 ok $@ && $@ =~ /not allowed at runtime/ && $blah == 12;
444
445 for $code ('{$blah = 45}','=xx') {
446 $blah = 12;
447 my $res = eval { "xx" =~ /(?$code)/o };
448 no warnings 'uninitialized';
449 local $Error = "'$@', '$res', '$blah'";
450 if ($code eq '=xx') {
451 ok !$@ && $res;
452 }
453 else {
454 ok $@ && $@ =~ /not allowed at runtime/ && $blah == 12;
455 }
456 }
3568d838 457
84281c31
A
458 $code = '{$blah = 45}';
459 $blah = 12;
460 eval "/(?$code)/";
461 iseq $blah, 45;
3568d838 462
84281c31
A
463 $blah = 12;
464 /(?{$blah = 45})/;
465 iseq $blah, 45;
466 }
3568d838 467
84281c31
A
468 {
469 local $Message = "Pos checks";
470 my $x = 'banana';
471 $x =~ /.a/g;
472 iseq pos ($x), 2;
3568d838 473
84281c31
A
474 $x =~ /.z/gc;
475 iseq pos ($x), 2;
3568d838 476
84281c31
A
477 sub f {
478 my $p = $_[0];
479 return $p;
480 }
3568d838 481
84281c31
A
482 $x =~ /.a/g;
483 iseq f (pos ($x)), 4;
484 }
3568d838 485
84281c31
A
486 {
487 local $Message = 'Checking $^R';
488 our $x = $^R = 67;
489 'foot' =~ /foo(?{$x = 12; 75})[t]/;
490 iseq $^R, 75;
491
492 $x = $^R = 67;
493 'foot' =~ /foo(?{$x = 12; 75})[xy]/;
494 ok $^R eq '67' && $x eq '12';
495
496 $x = $^R = 67;
497 'foot' =~ /foo(?{ $^R + 12 })((?{ $x = 12; $^R + 17 })[xy])?/;
498 ok $^R eq '79' && $x eq '12';
499 }
3568d838 500
84281c31
A
501 {
502 iseq qr/\b\v$/i, '(?i-xsm:\b\v$)', 'qr/\b\v$/i';
503 iseq qr/\b\v$/s, '(?s-xim:\b\v$)', 'qr/\b\v$/s';
504 iseq qr/\b\v$/m, '(?m-xis:\b\v$)', 'qr/\b\v$/m';
505 iseq qr/\b\v$/x, '(?x-ism:\b\v$)', 'qr/\b\v$/x';
506 iseq qr/\b\v$/xism, '(?msix:\b\v$)', 'qr/\b\v$/xism';
507 iseq qr/\b\v$/, '(?-xism:\b\v$)', 'qr/\b\v$/';
508 }
3568d838 509
3568d838 510
84281c31
A
511 {
512 local $Message = "Look around";
513 $_ = 'xabcx';
514 SKIP:
515 foreach my $ans ('', 'c') {
516 ok /(?<=(?=a)..)((?=c)|.)/g or skip "Match failed", 1;
517 iseq $1, $ans;
518 }
519 }
3568d838 520
84281c31
A
521 {
522 local $Message = "Empty clause";
523 $_ = 'a';
524 foreach my $ans ('', 'a', '') {
525 ok /^|a|$/g or skip "Match failed", 1;
526 iseq $&, $ans;
527 }
528 }
3568d838 529
84281c31
A
530 {
531 local $Message = "Prefixify";
532 sub prefixify {
533 SKIP: {
534 my ($v, $a, $b, $res) = @_;
535 ok $v =~ s/\Q$a\E/$b/ or skip "Match failed", 1;
536 iseq $v, $res;
537 }
538 }
3568d838 539
84281c31
A
540 prefixify ('/a/b/lib/arch', "/a/b/lib", 'X/lib', 'X/lib/arch');
541 prefixify ('/a/b/man/arch', "/a/b/man", 'X/man', 'X/man/arch');
542 }
3568d838 543
84281c31
A
544 {
545 $_ = 'var="foo"';
546 /(\")/;
547 ok $1 && /$1/, "Capture a quote";
548 }
3568d838 549
84281c31 550 {
84281c31
A
551 no warnings 'closure';
552 local $Message = '(?{ $var } refers to package vars';
553 package aa;
554 our $c = 2;
555 $::c = 3;
556 '' =~ /(?{ $c = 4 })/;
557 main::iseq $c, 4;
558 main::iseq $::c, 3;
559 }
3568d838 560
84281c31
A
561 {
562 must_die 'q(a:[b]:) =~ /[x[:foo:]]/',
563 'POSIX class \[:[^:]+:\] unknown in regex',
564 'POSIX class [: :] must have valid name';
565
566 for my $d (qw [= .]) {
567 must_die "/[[${d}foo${d}]]/",
568 "\QPOSIX syntax [$d $d] is reserved for future extensions",
569 "POSIX syntax [[$d $d]] is an error";
570 }
571 }
3568d838 572
3568d838 573
84281c31
A
574 {
575 # test if failure of patterns returns empty list
576 local $Message = "Failed pattern returns empty list";
577 $_ = 'aaa';
578 @_ = /bbb/;
579 iseq "@_", "";
3568d838 580
84281c31
A
581 @_ = /bbb/g;
582 iseq "@_", "";
a72deede 583
84281c31
A
584 @_ = /(bbb)/;
585 iseq "@_", "";
a72deede 586
84281c31
A
587 @_ = /(bbb)/g;
588 iseq "@_", "";
589 }
a72deede 590
0f289c68 591
84281c31
A
592 {
593 local $Message = '@- and @+ tests';
594
595 /a(?=.$)/;
596 iseq $#+, 0;
597 iseq $#-, 0;
598 iseq $+ [0], 2;
599 iseq $- [0], 1;
600 ok !defined $+ [1] && !defined $- [1] &&
601 !defined $+ [2] && !defined $- [2];
602
603 /a(a)(a)/;
604 iseq $#+, 2;
605 iseq $#-, 2;
606 iseq $+ [0], 3;
607 iseq $- [0], 0;
608 iseq $+ [1], 2;
609 iseq $- [1], 1;
610 iseq $+ [2], 3;
611 iseq $- [2], 2;
612 ok !defined $+ [3] && !defined $- [3] &&
613 !defined $+ [4] && !defined $- [4];
614
615
616 /.(a)(b)?(a)/;
617 iseq $#+, 3;
618 iseq $#-, 3;
619 iseq $+ [1], 2;
620 iseq $- [1], 1;
621 iseq $+ [3], 3;
622 iseq $- [3], 2;
623 ok !defined $+ [2] && !defined $- [2] &&
624 !defined $+ [4] && !defined $- [4];
625
626
627 /.(a)/;
628 iseq $#+, 1;
629 iseq $#-, 1;
630 iseq $+ [0], 2;
631 iseq $- [0], 0;
632 iseq $+ [1], 2;
633 iseq $- [1], 1;
634 ok !defined $+ [2] && !defined $- [2] &&
635 !defined $+ [3] && !defined $- [3];
636
637 /.(a)(ba*)?/;
638 iseq $#+, 2;
639 iseq $#-, 1;
640 }
a72deede 641
a72deede 642
84281c31
A
643 {
644 local $DiePattern = '^Modification of a read-only value attempted';
645 local $Message = 'Elements of @- and @+ are read-only';
646 must_die '$+[0] = 13';
647 must_die '$-[0] = 13';
648 must_die '@+ = (7, 6, 5)';
649 must_die '@- = qw (foo bar)';
650 }
a72deede 651
a72deede 652
84281c31
A
653 {
654 local $Message = '\G testing';
655 $_ = 'aaa';
656 pos = 1;
657 my @a = /\Ga/g;
658 iseq "@a", "a a";
659
660 my $str = 'abcde';
661 pos $str = 2;
662 ok $str !~ /^\G/;
663 ok $str !~ /^.\G/;
664 ok $str =~ /^..\G/;
665 ok $str !~ /^...\G/;
666 ok $str =~ /\G../ && $& eq 'cd';
667
668 local $TODO = $running_as_thread;
669 ok $str =~ /.\G./ && $& eq 'bc';
670 }
a72deede 671
569b5e07 672
84281c31
A
673 {
674 local $Message = 'pos inside (?{ })';
675 my $str = 'abcde';
676 our ($foo, $bar);
677 ok $str =~ /b(?{$foo = $_; $bar = pos})c/;
678 iseq $foo, $str;
679 iseq $bar, 2;
680 ok !defined pos ($str);
681
682 undef $foo;
683 undef $bar;
684 pos $str = undef;
685 ok $str =~ /b(?{$foo = $_; $bar = pos})c/g;
686 iseq $foo, $str;
687 iseq $bar, 2;
688 iseq pos ($str), 3;
689
690 $_ = $str;
691 undef $foo;
692 undef $bar;
693 ok /b(?{$foo = $_; $bar = pos})c/;
694 iseq $foo, $str;
695 iseq $bar, 2;
696
697 undef $foo;
698 undef $bar;
699 ok /b(?{$foo = $_; $bar = pos})c/g;
700 iseq $foo, $str;
701 iseq $bar, 2;
702 iseq pos, 3;
703
704 undef $foo;
705 undef $bar;
706 pos = undef;
707 1 while /b(?{$foo = $_; $bar = pos})c/g;
708 iseq $foo, $str;
709 iseq $bar, 2;
710 ok !defined pos;
711
712 undef $foo;
713 undef $bar;
714 $_ = 'abcde|abcde';
715 ok s/b(?{$foo = $_; $bar = pos})c/x/g;
716 iseq $foo, 'abcde|abcde';
717 iseq $bar, 8;
718 iseq $_, 'axde|axde';
719
720 # List context:
721 $_ = 'abcde|abcde';
722 our @res;
723 () = /([ace]).(?{push @res, $1,$2})([ce])(?{push @res, $1,$2})/g;
724 @res = map {defined $_ ? "'$_'" : 'undef'} @res;
725 iseq "@res", "'a' undef 'a' 'c' 'e' undef 'a' undef 'a' 'c'";
726
727 @res = ();
728 () = /([ace]).(?{push @res, $`,$&,$'})([ce])(?{push @res, $`,$&,$'})/g;
729 @res = map {defined $_ ? "'$_'" : 'undef'} @res;
730 iseq "@res", "'' 'ab' 'cde|abcde' " .
731 "'' 'abc' 'de|abcde' " .
732 "'abcd' 'e|' 'abcde' " .
733 "'abcde|' 'ab' 'cde' " .
734 "'abcde|' 'abc' 'de'" ;
735 }
f33976b4 736
cce850e4 737
84281c31
A
738 {
739 local $Message = '\G anchor checks';
740 my $foo = 'aabbccddeeffgg';
741 pos ($foo) = 1;
742 {
743 local $TODO = $running_as_thread;
744 no warnings 'uninitialized';
745 ok $foo =~ /.\G(..)/g;
746 iseq $1, 'ab';
cce850e4 747
84281c31
A
748 pos ($foo) += 1;
749 ok $foo =~ /.\G(..)/g;
750 iseq $1, 'cc';
cce850e4 751
84281c31
A
752 pos ($foo) += 1;
753 ok $foo =~ /.\G(..)/g;
754 iseq $1, 'de';
cce850e4 755
84281c31
A
756 ok $foo =~ /\Gef/g;
757 }
cce850e4 758
84281c31
A
759 undef pos $foo;
760 ok $foo =~ /\G(..)/g;
761 iseq $1, 'aa';
cce850e4 762
84281c31
A
763 ok $foo =~ /\G(..)/g;
764 iseq $1, 'bb';
cce850e4 765
84281c31
A
766 pos ($foo) = 5;
767 ok $foo =~ /\G(..)/g;
768 iseq $1, 'cd';
769 }
cce850e4 770
cce850e4 771
84281c31
A
772 {
773 $_ = '123x123';
774 my @res = /(\d*|x)/g;
775 local $" = '|';
776 iseq "@res", "123||x|123|", "0 match in alternation";
777 }
cce850e4 778
d9f424b2 779
84281c31
A
780 {
781 local $Message = "Match against temporaries (created via pp_helem())" .
782 " is safe";
783 ok {foo => "bar\n" . $^X} -> {foo} =~ /^(.*)\n/g;
784 iseq $1, "bar";
785 }
75685a94 786
d9f424b2 787
84281c31
A
788 {
789 local $Message = 'package $i inside (?{ }), ' .
790 'saved substrings and changing $_';
791 our @a = qw [foo bar];
792 our @b = ();
793 s/(\w)(?{push @b, $1})/,$1,/g for @a;
794 iseq "@b", "f o o b a r";
795 iseq "@a", ",f,,o,,o, ,b,,a,,r,";
796
797 local $Message = 'lexical $i inside (?{ }), ' .
798 'saved substrings and changing $_';
799 no warnings 'closure';
800 my @c = qw [foo bar];
801 my @d = ();
802 s/(\w)(?{push @d, $1})/,$1,/g for @c;
803 iseq "@d", "f o o b a r";
804 iseq "@c", ",f,,o,,o, ,b,,a,,r,";
d9f424b2
JH
805 }
806
d9f424b2 807
84281c31
A
808 {
809 local $Message = 'Brackets';
810 our $brackets;
811 $brackets = qr {
812 { (?> [^{}]+ | (??{ $brackets }) )* }
813 }x;
814
815 ok "{{}" =~ $brackets;
816 iseq $&, "{}";
817 ok "something { long { and } hairy" =~ $brackets;
818 iseq $&, "{ and }";
819 ok "something { long { and } hairy" =~ m/((??{ $brackets }))/;
820 iseq $&, "{ and }";
821 }
a4c04bdc 822
e2d8ce26 823
84281c31
A
824 {
825 $_ = "a-a\nxbb";
826 pos = 1;
827 nok m/^-.*bb/mg, '$_ = "a-a\nxbb"; m/^-.*bb/mg';
828 }
a4c04bdc 829
a4c04bdc 830
84281c31
A
831 {
832 local $Message = '\G anchor checks';
833 my $text = "aaXbXcc";
834 pos ($text) = 0;
835 ok $text !~ /\GXb*X/g;
836 }
a4c04bdc 837
a4c04bdc 838
84281c31
A
839 {
840 $_ = "xA\n" x 500;
841 nok /^\s*A/m, '$_ = "xA\n" x 500; /^\s*A/m"';
a4c04bdc 842
84281c31
A
843 my $text = "abc dbf";
844 my @res = ($text =~ /.*?(b).*?\b/g);
845 iseq "@res", "b b", '\b is not special';
987aaf07 846 }
a4c04bdc 847
a4c04bdc 848
84281c31
A
849 {
850 local $Message = '\S, [\S], \s, [\s]';
851 my @a = map chr, 0 .. 255;
9d45b377
YO
852 my @b = grep m/\S/, @a;
853 my @c = grep m/[^\s]/, @a;
84281c31
A
854 iseq "@b", "@c";
855
856 @b = grep /\S/, @a;
857 @c = grep /[\S]/, @a;
858 iseq "@b", "@c";
859
860 @b = grep /\s/, @a;
861 @c = grep /[^\S]/, @a;
862 iseq "@b", "@c";
863
864 @b = grep /\s/, @a;
865 @c = grep /[\s]/, @a;
866 iseq "@b", "@c";
867 }
868 {
869 local $Message = '\D, [\D], \d, [\d]';
870 my @a = map chr, 0 .. 255;
871 my @b = grep /\D/, @a;
872 my @c = grep /[^\d]/, @a;
873 iseq "@b", "@c";
874
875 @b = grep /\D/, @a;
876 @c = grep /[\D]/, @a;
877 iseq "@b", "@c";
878
879 @b = grep /\d/, @a;
880 @c = grep /[^\D]/, @a;
881 iseq "@b", "@c";
882
883 @b = grep /\d/, @a;
884 @c = grep /[\d]/, @a;
885 iseq "@b", "@c";
886 }
887 {
888 local $Message = '\W, [\W], \w, [\w]';
889 my @a = map chr, 0 .. 255;
890 my @b = grep /\W/, @a;
891 my @c = grep /[^\w]/, @a;
892 iseq "@b", "@c";
893
894 @b = grep /\W/, @a;
895 @c = grep /[\W]/, @a;
896 iseq "@b", "@c";
897
898 @b = grep /\w/, @a;
899 @c = grep /[^\W]/, @a;
900 iseq "@b", "@c";
901
902 @b = grep /\w/, @a;
903 @c = grep /[\w]/, @a;
904 iseq "@b", "@c";
905 }
a4c04bdc 906
a4c04bdc 907
84281c31
A
908 {
909 # see if backtracking optimization works correctly
910 local $Message = 'Backtrack optimization';
911 ok "\n\n" =~ /\n $ \n/x;
912 ok "\n\n" =~ /\n* $ \n/x;
913 ok "\n\n" =~ /\n+ $ \n/x;
914 ok "\n\n" =~ /\n? $ \n/x;
915 ok "\n\n" =~ /\n*? $ \n/x;
916 ok "\n\n" =~ /\n+? $ \n/x;
917 ok "\n\n" =~ /\n?? $ \n/x;
918 ok "\n\n" !~ /\n*+ $ \n/x;
919 ok "\n\n" !~ /\n++ $ \n/x;
920 ok "\n\n" =~ /\n?+ $ \n/x;
921 }
a4c04bdc 922
a4c04bdc 923
84281c31
A
924 {
925 package S;
926 use overload '""' => sub {'Object S'};
927 sub new {bless []}
0f289c68 928
9d45b377 929 local $::Message = "Ref stringification";
84281c31
A
930 ::ok do { \my $v} =~ /^SCALAR/, "Scalar ref stringification";
931 ::ok do {\\my $v} =~ /^REF/, "Ref ref stringification";
932 ::ok [] =~ /^ARRAY/, "Array ref stringification";
933 ::ok {} =~ /^HASH/, "Hash ref stringification";
934 ::ok 'S' -> new =~ /^Object S/, "Object stringification";
935 }
a4c04bdc 936
a4c04bdc 937
84281c31
A
938 {
939 local $Message = "Test result of match used as match";
940 ok 'a1b' =~ ('xyz' =~ /y/);
941 iseq $`, 'a';
942 ok 'a1b' =~ ('xyz' =~ /t/);
943 iseq $`, 'a';
944 }
a4c04bdc 945
a4c04bdc 946
84281c31
A
947 {
948 local $Message = '"1" is not \s';
949 may_not_warn sub {ok ("1\n" x 102) !~ /^\s*\n/m};
950 }
a4c04bdc 951
a4c04bdc 952
84281c31
A
953 {
954 local $Message = '\s, [[:space:]] and [[:blank:]]';
955 my %space = (spc => " ",
956 tab => "\t",
957 cr => "\r",
958 lf => "\n",
959 ff => "\f",
960 # There's no \v but the vertical tabulator seems miraculously
961 # be 11 both in ASCII and EBCDIC.
962 vt => chr(11),
963 false => "space");
964
965 my @space0 = sort grep {$space {$_} =~ /\s/ } keys %space;
966 my @space1 = sort grep {$space {$_} =~ /[[:space:]]/} keys %space;
967 my @space2 = sort grep {$space {$_} =~ /[[:blank:]]/} keys %space;
968
969 iseq "@space0", "cr ff lf spc tab";
970 iseq "@space1", "cr ff lf spc tab vt";
971 iseq "@space2", "spc tab";
972 }
a4c04bdc 973
ff3f963a
KW
974 {
975 use charnames ":full";
976 local $Message = 'Delayed interpolation of \N';
977 my $r1 = qr/\N{THAI CHARACTER SARA I}/;
978 my $s1 = "\x{E34}\x{E34}\x{E34}\x{E34}";
979
980 # Bug #56444
981 ok $s1 =~ /$r1+/, 'my $r1 = qr/\N{THAI CHARACTER SARA I}/; my $s1 = "\x{E34}\x{E34}\x{E34}\x{E34}; $s1 =~ /$r1+/';
982
983 # Bug #62056
984 ok "${s1}A" =~ m/$s1\N{LATIN CAPITAL LETTER A}/, '"${s1}A" =~ m/$s1\N{LATIN CAPITAL LETTER A}/';
985
986 ok "abbbbc" =~ m/\N{1}/ && $& eq "a", '"abbbbc" =~ m/\N{1}/ && $& eq "a"';
987 ok "abbbbc" =~ m/\N{3,4}/ && $& eq "abbb", '"abbbbc" =~ m/\N{3,4}/ && $& eq "abbb"';
988 }
989
37820adc
KW
990 {
991 use charnames ":full";
992 local $Message = '[perl #74982] Period coming after \N{}';
993 ok "\x{ff08}." =~ m/\N{FULLWIDTH LEFT PARENTHESIS}./ && $& eq "\x{ff08}.";
994 ok "\x{ff08}." =~ m/[\N{FULLWIDTH LEFT PARENTHESIS}]./ && $& eq "\x{ff08}.";
995 }
c9415951
YO
996 {
997 my $n= 50;
998 # this must be a high number and go from 0 to N, as the bug we are looking for doesnt
999 # seem to be predictable. Slight changes to the test make it fail earlier or later.
1000 foreach my $i (0 .. $n)
1001 {
1002 my $str= "\n" x $i;
1003 ok $str=~/.*\z/, "implict MBOL check string disable does not break things length=$i";
1004 }
1005 }
92f3d482
YO
1006 {
1007 # we are actually testing that we dont die when executing these patterns
1008 use utf8;
1009 my $e = "Böck";
1010 ok(utf8::is_utf8($e),"got a unicode string - rt75680");
1011
1012 ok($e !~ m/.*?[x]$/, "unicode string against /.*?[x]\$/ - rt75680");
1013 ok($e !~ m/.*?\p{Space}$/i, "unicode string against /.*?\\p{space}\$/i - rt75680");
1014 ok($e !~ m/.*?[xyz]$/, "unicode string against /.*?[xyz]\$/ - rt75680");
1015 ok($e !~ m/(.*?)[,\p{isSpace}]+((?:\p{isAlpha}[\p{isSpace}\.]{1,2})+)\p{isSpace}*$/, "unicode string against big pattern - rt75680");
1016 }
1017 {
1018 # we are actually testing that we dont die when executing these patterns
1019 my $e = "B\x{f6}ck";
1020 ok(!utf8::is_utf8($e), "got a latin string - rt75680");
1021
1022 ok($e !~ m/.*?[x]$/, "latin string against /.*?[x]\$/ - rt75680");
1023 ok($e !~ m/.*?\p{Space}$/i, "latin string against /.*?\\p{space}\$/i - rt75680");
1024 ok($e !~ m/.*?[xyz]$/,"latin string against /.*?[xyz]\$/ - rt75680");
1025 ok($e !~ m/(.*?)[,\p{isSpace}]+((?:\p{isAlpha}[\p{isSpace}\.]{1,2})+)\p{isSpace}*$/,"latin string against big pattern - rt75680");
1026 }
c920e018
A
1027
1028 {
1029 #
1030 # Tests for bug 77414.
1031 #
1032
1033 local $Message = '\p property after empty * match';
1034 {
1035 local $TODO = "Bug 77414";
1036 ok "1" =~ /\s*\pN/;
1037 ok "-" =~ /\s*\p{Dash}/;
1038 ok " " =~ /\w*\p{Blank}/;
1039 }
1040
1041 ok "1" =~ /\s*\pN+/;
1042 ok "-" =~ /\s*\p{Dash}{1}/;
1043 ok " " =~ /\w*\p{Blank}{1,4}/;
1044
1045 }
1046
84281c31
A
1047} # End of sub run_tests
1048
10491;