This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add Jess Robinson to AUTHORS
[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
9fa196e1 5# that does fit that format, add it to re/re_tests, not here.
8d063cd8 6
84281c31
A
7use strict;
8use warnings;
9use 5.010;
10
84281c31
A
11sub run_tests;
12
9133bbab 13$| = 1;
3568d838 14
8d37f932 15
e4d48cc9
GS
16BEGIN {
17 chdir 't' if -d 't';
9d45b377 18 @INC = ('../lib','.');
6f4e0180 19 require './test.pl';
e4d48cc9 20}
84281c31 21
4fab19ce 22plan tests => 466; # Update this when adding/deleting tests.
b7a35066 23
9d45b377 24run_tests() unless caller;
b7a35066 25
84281c31
A
26#
27# Tests start here.
28#
29sub run_tests {
0ef3e39e 30
84281c31 31 {
84281c31 32 my $x = "abc\ndef\n";
5895685f 33 (my $x_pretty = $x) =~ s/\n/\\n/g;
fd291da9 34
5895685f
NC
35 ok $x =~ /^abc/, qq ["$x_pretty" =~ /^abc/];
36 ok $x !~ /^def/, qq ["$x_pretty" !~ /^def/];
fd291da9 37
84281c31 38 # used to be a test for $*
5895685f 39 ok $x =~ /^def/m, qq ["$x_pretty" =~ /^def/m];
fd291da9 40
b33825c4
NC
41 ok(!($x =~ /^xxx/), qq ["$x_pretty" =~ /^xxx/]);
42 ok(!($x !~ /^abc/), qq ["$x_pretty" !~ /^abc/]);
fd291da9 43
5895685f 44 ok $x =~ /def/, qq ["$x_pretty" =~ /def/];
b33825c4 45 ok(!($x !~ /def/), qq ["$x_pretty" !~ /def/]);
4765795a 46
5895685f 47 ok $x !~ /.def/, qq ["$x_pretty" !~ /.def/];
b33825c4 48 ok(!($x =~ /.def/), qq ["$x_pretty" =~ /.def/]);
4765795a 49
5895685f 50 ok $x =~ /\ndef/, qq ["$x_pretty" =~ /\\ndef/];
b33825c4 51 ok(!($x !~ /\ndef/), qq ["$x_pretty" !~ /\\ndef/]);
84281c31 52 }
4765795a 53
84281c31
A
54 {
55 $_ = '123';
56 ok /^([0-9][0-9]*)/, qq [\$_ = '$_'; /^([0-9][0-9]*)/];
57 }
f9969324 58
84281c31
A
59 {
60 $_ = 'aaabbbccc';
61 ok /(a*b*)(c*)/ && $1 eq 'aaabbb' && $2 eq 'ccc',
62 qq [\$_ = '$_'; /(a*b*)(c*)/];
63 ok /(a+b+c+)/ && $1 eq 'aaabbbccc', qq [\$_ = '$_'; /(a+b+c+)/];
b33825c4 64 unlike($_, qr/a+b?c+/, qq [\$_ = '$_'; /a+b?c+/]);
84281c31
A
65
66 $_ = 'aaabccc';
67 ok /a+b?c+/, qq [\$_ = '$_'; /a+b?c+/];
68 ok /a*b?c*/, qq [\$_ = '$_'; /a*b?c*/];
69
70 $_ = 'aaaccc';
71 ok /a*b?c*/, qq [\$_ = '$_'; /a*b?c*/];
b33825c4 72 unlike($_, qr/a*b+c*/, qq [\$_ = '$_'; /a*b+c*/]);
84281c31
A
73
74 $_ = 'abcdef';
75 ok /bcd|xyz/, qq [\$_ = '$_'; /bcd|xyz/];
76 ok /xyz|bcd/, qq [\$_ = '$_'; /xyz|bcd/];
77 ok m|bc/*d|, qq [\$_ = '$_'; m|bc/*d|];
78 ok /^$_$/, qq [\$_ = '$_'; /^\$_\$/];
4765795a 79 }
4765795a 80
84281c31
A
81 {
82 # used to be a test for $*
5895685f 83 ok "ab\ncd\n" =~ /^cd/m, q ["ab\ncd\n" =~ /^cd/m];
84281c31 84 }
4765795a 85
84281c31
A
86 {
87 our %XXX = map {($_ => $_)} 123, 234, 345;
88
89 our @XXX = ('ok 1','not ok 1', 'ok 2','not ok 2','not ok 3');
90 while ($_ = shift(@XXX)) {
b33825c4 91 my $e = index ($_, 'not') >= 0 ? '' : 1;
725a61d7 92 my $r = m?(.*)?;
b33825c4 93 is($r, $e, "?(.*)?");
84281c31
A
94 /not/ && reset;
95 if (/not ok 2/) {
96 if ($^O eq 'VMS') {
97 $_ = shift(@XXX);
98 }
99 else {
100 reset 'X';
101 }
102 }
103 }
4765795a 104
84281c31
A
105 SKIP: {
106 if ($^O eq 'VMS') {
107 skip "Reset 'X'", 1;
108 }
109 ok !keys %XXX, "%XXX is empty";
110 }
4765795a 111
84281c31 112 }
4765795a 113
84281c31 114 {
4f890a30 115 my $message = "Test empty pattern";
84281c31
A
116 my $xyz = 'xyz';
117 my $cde = 'cde';
118
119 $cde =~ /[^ab]*/;
120 $xyz =~ //;
4f890a30 121 is($&, $xyz, $message);
84281c31
A
122
123 my $foo = '[^ab]*';
124 $cde =~ /$foo/;
125 $xyz =~ //;
4f890a30 126 is($&, $xyz, $message);
84281c31
A
127
128 $cde =~ /$foo/;
129 my $null;
130 no warnings 'uninitialized';
131 $xyz =~ /$null/;
4f890a30 132 is($&, $xyz, $message);
84281c31
A
133
134 $null = "";
135 $xyz =~ /$null/;
4f890a30 136 is($&, $xyz, $message);
84281c31 137 }
4765795a 138
84281c31 139 {
4f890a30 140 my $message = q !Check $`, $&, $'!;
84281c31 141 $_ = 'abcdefghi';
0f289c68 142 /def/; # optimized up to cmd
4f890a30 143 is("$`:$&:$'", 'abc:def:ghi', $message);
4765795a 144
84281c31 145 no warnings 'void';
0f289c68 146 /cde/ + 0; # optimized only to spat
4f890a30 147 is("$`:$&:$'", 'ab:cde:fghi', $message);
4765795a 148
0f289c68 149 /[d][e][f]/; # not optimized
4f890a30 150 is("$`:$&:$'", 'abc:def:ghi', $message);
84281c31 151 }
4765795a 152
84281c31
A
153 {
154 $_ = 'now is the {time for all} good men to come to.';
2a53d331 155 / \{([^}]*)}/;
de26e0cc 156 is($1, 'time for all', "Match braces");
84281c31 157 }
4765795a 158
84281c31 159 {
4f890a30 160 my $message = "{N,M} quantifier";
84281c31 161 $_ = 'xxx {3,4} yyy zzz';
4f890a30
NC
162 ok(/( {3,4})/, $message);
163 is($1, ' ', $message);
164 unlike($_, qr/( {4,})/, $message);
165 ok(/( {2,3}.)/, $message);
166 is($1, ' y', $message);
167 ok(/(y{2,3}.)/, $message);
168 is($1, 'yyy ', $message);
169 unlike($_, qr/x {3,4}/, $message);
170 unlike($_, qr/^xxx {3,4}/, $message);
84281c31 171 }
4765795a 172
84281c31 173 {
4f890a30 174 my $message = "Test /g";
84281c31
A
175 local $" = ":";
176 $_ = "now is the time for all good men to come to.";
177 my @words = /(\w+)/g;
178 my $exp = "now:is:the:time:for:all:good:men:to:come:to";
4765795a 179
4f890a30 180 is("@words", $exp, $message);
4765795a 181
84281c31
A
182 @words = ();
183 while (/\w+/g) {
184 push (@words, $&);
185 }
4f890a30 186 is("@words", $exp, $message);
4765795a 187
84281c31
A
188 @words = ();
189 pos = 0;
190 while (/to/g) {
191 push(@words, $&);
192 }
4f890a30 193 is("@words", "to:to", $message);
4765795a 194
84281c31
A
195 pos $_ = 0;
196 @words = /to/g;
4f890a30 197 is("@words", "to:to", $message);
84281c31 198 }
4765795a 199
84281c31
A
200 {
201 $_ = "abcdefghi";
202
203 my $pat1 = 'def';
204 my $pat2 = '^def';
205 my $pat3 = '.def.';
206 my $pat4 = 'abc';
207 my $pat5 = '^abc';
208 my $pat6 = 'abc$';
209 my $pat7 = 'ghi';
210 my $pat8 = '\w*ghi';
211 my $pat9 = 'ghi$';
212
213 my $t1 = my $t2 = my $t3 = my $t4 = my $t5 =
214 my $t6 = my $t7 = my $t8 = my $t9 = 0;
215
216 for my $iter (1 .. 5) {
217 $t1++ if /$pat1/o;
218 $t2++ if /$pat2/o;
219 $t3++ if /$pat3/o;
220 $t4++ if /$pat4/o;
221 $t5++ if /$pat5/o;
222 $t6++ if /$pat6/o;
223 $t7++ if /$pat7/o;
224 $t8++ if /$pat8/o;
225 $t9++ if /$pat9/o;
226 }
227 my $x = "$t1$t2$t3$t4$t5$t6$t7$t8$t9";
de26e0cc 228 is($x, '505550555', "Test /o");
84281c31 229 }
4765795a 230
4f890a30 231 {
84281c31
A
232 my $xyz = 'xyz';
233 ok "abc" =~ /^abc$|$xyz/, "| after \$";
4765795a 234
84281c31 235 # perl 4.009 says "unmatched ()"
4f890a30 236 my $message = '$ inside ()';
4765795a 237
84281c31
A
238 my $result;
239 eval '"abc" =~ /a(bc$)|$xyz/; $result = "$&:$1"';
4f890a30
NC
240 is($@, "", $message);
241 is($result, "abc:bc", $message);
84281c31 242 }
4765795a 243
84281c31 244 {
4f890a30 245 my $message = "Scalar /g";
84281c31
A
246 $_ = "abcfooabcbar";
247
4f890a30
NC
248 ok( /abc/g && $` eq "", $message);
249 ok( /abc/g && $` eq "abcfoo", $message);
250 ok(!/abc/g, $message);
84281c31 251
4f890a30 252 $message = "Scalar /gi";
84281c31 253 pos = 0;
4f890a30
NC
254 ok( /ABC/gi && $` eq "", $message);
255 ok( /ABC/gi && $` eq "abcfoo", $message);
256 ok(!/ABC/gi, $message);
84281c31 257
4f890a30 258 $message = "Scalar /g";
84281c31 259 pos = 0;
4f890a30
NC
260 ok( /abc/g && $' eq "fooabcbar", $message);
261 ok( /abc/g && $' eq "bar", $message);
84281c31
A
262
263 $_ .= '';
264 my @x = /abc/g;
de26e0cc 265 is(@x, 2, "/g reset after assignment");
4765795a 266 }
4765795a 267
84281c31 268 {
4f890a30 269 my $message = '/g, \G and pos';
84281c31
A
270 $_ = "abdc";
271 pos $_ = 2;
272 /\Gc/gc;
4f890a30 273 is(pos $_, 2, $message);
84281c31 274 /\Gc/g;
4f890a30 275 is(pos $_, undef, $message);
84281c31 276 }
4765795a 277
84281c31 278 {
4f890a30 279 my $message = '(?{ })';
84281c31
A
280 our $out = 1;
281 'abc' =~ m'a(?{ $out = 2 })b';
4f890a30 282 is($out, 2, $message);
84281c31
A
283
284 $out = 1;
285 'abc' =~ m'a(?{ $out = 3 })c';
4f890a30 286 is($out, 1, $message);
84281c31 287 }
4765795a 288
84281c31
A
289 {
290 $_ = 'foobar1 bar2 foobar3 barfoobar5 foobar6';
291 my @out = /(?<!foo)bar./g;
de26e0cc 292 is("@out", 'bar2 barf', "Negative lookbehind");
84281c31 293 }
4765795a 294
84281c31 295 {
4f890a30 296 my $message = "REG_INFTY tests";
84281c31 297 # Tests which depend on REG_INFTY
19d6612d
NC
298
299 # Defaults assumed if this fails
300 eval { require Config; };
301 $::reg_infty = $Config::Config{reg_infty} // 32767;
84281c31
A
302 $::reg_infty_m = $::reg_infty - 1;
303 $::reg_infty_p = $::reg_infty + 1;
93f09d7b 304 $::reg_infty_m = $::reg_infty_m; # Suppress warning.
84281c31
A
305
306 # As well as failing if the pattern matches do unexpected things, the
307 # next three tests will fail if you should have picked up a lower-than-
308 # default value for $reg_infty from Config.pm, but have not.
309
14358a41
NC
310 is(eval q{('aaa' =~ /(a{1,$::reg_infty_m})/)[0]}, 'aaa', $message);
311 is($@, '', $message);
312 is(eval q{('a' x $::reg_infty_m) =~ /a{$::reg_infty_m}/}, 1, $message);
313 is($@, '', $message);
314 isnt(q{('a' x ($::reg_infty_m - 1)) !~ /a{$::reg_infty_m}/}, 1, $message);
315 is($@, '', $message);
316
84281c31 317 eval "'aaa' =~ /a{1,$::reg_infty}/";
224b2e7e 318 like($@, qr/^\QQuantifier in {,} bigger than/, $message);
84281c31 319 eval "'aaa' =~ /a{1,$::reg_infty_p}/";
4f890a30 320 like($@, qr/^\QQuantifier in {,} bigger than/, $message);
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
84281c31
A
332 for my $l (125, 140, 250, 270, 300000, 30) { # Ordered to free memory
333 my $a = 'a' x $l;
4f890a30
NC
334 my $message = "Long monster, length = $l";
335 like("ba$a=", qr/a$a=/, $message);
336 unlike("b$a=", qr/a$a=/, $message);
337 like("b$a=", qr/ba+=/, $message);
84281c31 338
224b2e7e 339 like("ba$a=", qr/b(?:a|b)+=/, $message);
84281c31
A
340 }
341 }
8269fa76 342
84281c31
A
343 {
344 # 20000 nodes, each taking 3 words per string, and 1 per branch
345 my $long_constant_len = join '|', 12120 .. 32645;
346 my $long_var_len = join '|', 8120 .. 28645;
347 my %ans = ( 'ax13876y25677lbc' => 1,
348 'ax13876y25677mcb' => 0, # not b.
349 'ax13876y35677nbc' => 0, # Num too big
350 'ax13876y25677y21378obc' => 1,
0f289c68 351 'ax13876y25677y21378zbc' => 0, # Not followed by [k-o]
84281c31
A
352 'ax13876y25677y21378y21378kbc' => 1,
353 'ax13876y25677y21378y21378kcb' => 0, # Not b.
354 'ax13876y25677y21378y21378y21378kbc' => 0, # 5 runs
355 );
356
84281c31 357 for (keys %ans) {
4f890a30
NC
358 my $message = "20000 nodes, const-len '$_'";
359 ok !($ans{$_} xor /a(?=([yx]($long_constant_len)){2,4}[k-o]).*b./o), $message;
84281c31 360
4f890a30
NC
361 $message = "20000 nodes, var-len '$_'";
362 ok !($ans{$_} xor /a(?=([yx]($long_var_len)){2,4}[k-o]).*b./o,), $message;
84281c31 363 }
b8ef571c 364 }
209a9bc1 365
84281c31 366 {
4f890a30 367 my $message = "Complicated backtracking";
84281c31
A
368 $_ = " a (bla()) and x(y b((l)u((e))) and b(l(e)e)e";
369 my $expect = "(bla()) ((l)u((e))) (l(e)e)";
370
371 use vars '$c';
372 sub matchit {
373 m/
374 (
375 \(
0f289c68 376 (?{ $c = 1 }) # Initialize
84281c31
A
377 (?:
378 (?(?{ $c == 0 }) # PREVIOUS iteration was OK, stop the loop
379 (?!
0f289c68
YO
380 ) # Fail: will unwind one iteration back
381 )
84281c31 382 (?:
0f289c68 383 [^()]+ # Match a big chunk
84281c31
A
384 (?=
385 [()]
0f289c68 386 ) # Do not try to match subchunks
84281c31
A
387 |
388 \(
389 (?{ ++$c })
390 |
391 \)
392 (?{ --$c })
393 )
0f289c68 394 )+ # This may not match with different subblocks
84281c31
A
395 )
396 (?(?{ $c != 0 })
397 (?!
0f289c68
YO
398 ) # Fail
399 ) # Otherwise the chunk 1 may succeed with $c>0
84281c31
A
400 /xg;
401 }
3568d838 402
84281c31
A
403 my @ans = ();
404 my $res;
405 push @ans, $res while $res = matchit;
4f890a30 406 is("@ans", "1 1 1", $message);
3568d838 407
84281c31 408 @ans = matchit;
4f890a30 409 is("@ans", $expect, $message);
3568d838 410
4f890a30 411 $message = "Recursion with (??{ })";
84281c31
A
412 our $matched;
413 $matched = qr/\((?:(?>[^()]+)|(??{$matched}))*\)/;
3568d838 414
84281c31
A
415 @ans = my @ans1 = ();
416 push (@ans, $res), push (@ans1, $&) while $res = m/$matched/g;
3568d838 417
4f890a30
NC
418 is("@ans", "1 1 1", $message);
419 is("@ans1", $expect, $message);
3568d838 420
84281c31 421 @ans = m/$matched/g;
4f890a30 422 is("@ans", $expect, $message);
3568d838 423
84281c31 424 }
3568d838 425
84281c31
A
426 {
427 ok "abc" =~ /^(??{"a"})b/, '"abc" =~ /^(??{"a"})b/';
428 }
3568d838 429
84281c31 430 {
0f289c68 431 my @ans = ('a/b' =~ m%(.*/)?(.*)%); # Stack may be bad
de26e0cc 432 is("@ans", 'a/ b', "Stack may be bad");
84281c31 433 }
3568d838 434
84281c31 435 {
4f890a30 436 my $message = "Eval-group not allowed at runtime";
84281c31
A
437 my $code = '{$blah = 45}';
438 our $blah = 12;
439 eval { /(?$code)/ };
4f890a30 440 ok($@ && $@ =~ /not allowed at runtime/ && $blah == 12, $message);
84281c31 441
3044771b
NC
442 $blah = 12;
443 my $res = eval { "xx" =~ /(?$code)/o };
444 {
445 no warnings 'uninitialized';
5895685f 446 chomp $@; my $message = "$message '$@', '$res', '$blah'";
4f890a30 447 ok($@ && $@ =~ /not allowed at runtime/ && $blah == 12, $message);
3044771b
NC
448 }
449
450 $code = '=xx';
451 $blah = 12;
452 $res = eval { "xx" =~ /(?$code)/o };
453 {
454 no warnings 'uninitialized';
4f890a30
NC
455 my $message = "$message '$@', '$res', '$blah'";
456 ok(!$@ && $res, $message);
3044771b 457 }
3568d838 458
84281c31
A
459 $code = '{$blah = 45}';
460 $blah = 12;
461 eval "/(?$code)/";
4f890a30 462 is($blah, 45, $message);
3568d838 463
84281c31
A
464 $blah = 12;
465 /(?{$blah = 45})/;
4f890a30 466 is($blah, 45, $message);
84281c31 467 }
3568d838 468
84281c31 469 {
4f890a30 470 my $message = "Pos checks";
84281c31
A
471 my $x = 'banana';
472 $x =~ /.a/g;
4f890a30 473 is(pos $x, 2, $message);
3568d838 474
84281c31 475 $x =~ /.z/gc;
4f890a30 476 is(pos $x, 2, $message);
3568d838 477
84281c31
A
478 sub f {
479 my $p = $_[0];
480 return $p;
481 }
3568d838 482
84281c31 483 $x =~ /.a/g;
4f890a30 484 is(f (pos $x), 4, $message);
84281c31 485 }
3568d838 486
84281c31 487 {
4f890a30 488 my $message = 'Checking $^R';
84281c31
A
489 our $x = $^R = 67;
490 'foot' =~ /foo(?{$x = 12; 75})[t]/;
4f890a30 491 is($^R, 75, $message);
84281c31
A
492
493 $x = $^R = 67;
494 'foot' =~ /foo(?{$x = 12; 75})[xy]/;
4f890a30 495 ok($^R eq '67' && $x eq '12', $message);
84281c31
A
496
497 $x = $^R = 67;
498 'foot' =~ /foo(?{ $^R + 12 })((?{ $x = 12; $^R + 17 })[xy])?/;
4f890a30 499 ok($^R eq '79' && $x eq '12', $message);
84281c31 500 }
3568d838 501
84281c31 502 {
de26e0cc
NC
503 is(qr/\b\v$/i, '(?^i:\b\v$)', 'qr/\b\v$/i');
504 is(qr/\b\v$/s, '(?^s:\b\v$)', 'qr/\b\v$/s');
505 is(qr/\b\v$/m, '(?^m:\b\v$)', 'qr/\b\v$/m');
506 is(qr/\b\v$/x, '(?^x:\b\v$)', 'qr/\b\v$/x');
507 is(qr/\b\v$/xism, '(?^msix:\b\v$)', 'qr/\b\v$/xism');
508 is(qr/\b\v$/, '(?^:\b\v$)', 'qr/\b\v$/');
84281c31 509 }
3568d838 510
9de15fec 511 { # Test that charset modifier work, and are interpolated
de26e0cc
NC
512 is(qr/\b\v$/, '(?^:\b\v$)', 'Verify no locale, no unicode_strings gives default modifier');
513 is(qr/(?l:\b\v$)/, '(?^:(?l:\b\v$))', 'Verify infix l modifier compiles');
514 is(qr/(?u:\b\v$)/, '(?^:(?u:\b\v$))', 'Verify infix u modifier compiles');
515 is(qr/(?l)\b\v$/, '(?^:(?l)\b\v$)', 'Verify (?l) compiles');
516 is(qr/(?u)\b\v$/, '(?^:(?u)\b\v$)', 'Verify (?u) compiles');
9de15fec
KW
517
518 my $dual = qr/\b\v$/;
519 use locale;
520 my $locale = qr/\b\v$/;
de26e0cc 521 is($locale, '(?^l:\b\v$)', 'Verify has l modifier when compiled under use locale');
9de15fec
KW
522 no locale;
523
524 use feature 'unicode_strings';
525 my $unicode = qr/\b\v$/;
de26e0cc
NC
526 is($unicode, '(?^u:\b\v$)', 'Verify has u modifier when compiled under unicode_strings');
527 is(qr/abc$dual/, '(?^u:abc(?^:\b\v$))', 'Verify retains d meaning when interpolated under locale');
528 is(qr/abc$locale/, '(?^u:abc(?^l:\b\v$))', 'Verify retains l when interpolated under unicode_strings');
9de15fec
KW
529
530 no feature 'unicode_strings';
de26e0cc
NC
531 is(qr/abc$locale/, '(?^:abc(?^l:\b\v$))', 'Verify retains l when interpolated outside locale and unicode strings');
532 is(qr/def$unicode/, '(?^:def(?^u:\b\v$))', 'Verify retains u when interpolated outside locale and unicode strings');
9de15fec
KW
533
534 use locale;
de26e0cc
NC
535 is(qr/abc$dual/, '(?^l:abc(?^:\b\v$))', 'Verify retains d meaning when interpolated under locale');
536 is(qr/abc$unicode/, '(?^l:abc(?^u:\b\v$))', 'Verify retains u when interpolated under locale');
9de15fec
KW
537 }
538
84281c31 539 {
4f890a30 540 my $message = "Look around";
84281c31 541 $_ = 'xabcx';
84281c31 542 foreach my $ans ('', 'c') {
4f890a30
NC
543 ok(/(?<=(?=a)..)((?=c)|.)/g, $message);
544 is($1, $ans, $message);
84281c31
A
545 }
546 }
3568d838 547
84281c31 548 {
4f890a30 549 my $message = "Empty clause";
84281c31
A
550 $_ = 'a';
551 foreach my $ans ('', 'a', '') {
4f890a30
NC
552 ok(/^|a|$/g, $message);
553 is($&, $ans, $message);
84281c31
A
554 }
555 }
3568d838 556
84281c31 557 {
84281c31 558 sub prefixify {
4f890a30
NC
559 my $message = "Prefixify";
560 {
84281c31 561 my ($v, $a, $b, $res) = @_;
4f890a30
NC
562 ok($v =~ s/\Q$a\E/$b/, $message);
563 is($v, $res, $message);
84281c31
A
564 }
565 }
3568d838 566
84281c31
A
567 prefixify ('/a/b/lib/arch', "/a/b/lib", 'X/lib', 'X/lib/arch');
568 prefixify ('/a/b/man/arch', "/a/b/man", 'X/man', 'X/man/arch');
569 }
3568d838 570
84281c31
A
571 {
572 $_ = 'var="foo"';
573 /(\")/;
574 ok $1 && /$1/, "Capture a quote";
575 }
3568d838 576
84281c31 577 {
84281c31 578 no warnings 'closure';
4f890a30 579 my $message = '(?{ $var } refers to package vars';
84281c31
A
580 package aa;
581 our $c = 2;
582 $::c = 3;
583 '' =~ /(?{ $c = 4 })/;
4f890a30
NC
584 main::is($c, 4, $message);
585 main::is($::c, 3, $message);
84281c31 586 }
3568d838 587
84281c31 588 {
cb124425
NC
589 is(eval 'q(a:[b]:) =~ /[x[:foo:]]/', undef);
590 like ($@, qr/POSIX class \[:[^:]+:\] unknown in regex/,
591 'POSIX class [: :] must have valid name');
84281c31
A
592
593 for my $d (qw [= .]) {
cb124425
NC
594 is(eval "/[[${d}foo${d}]]/", undef);
595 like ($@, qr/\QPOSIX syntax [$d $d] is reserved for future extensions/,
596 "POSIX syntax [[$d $d]] is an error");
84281c31
A
597 }
598 }
3568d838 599
84281c31
A
600 {
601 # test if failure of patterns returns empty list
4f890a30 602 my $message = "Failed pattern returns empty list";
84281c31
A
603 $_ = 'aaa';
604 @_ = /bbb/;
4f890a30 605 is("@_", "", $message);
3568d838 606
84281c31 607 @_ = /bbb/g;
4f890a30 608 is("@_", "", $message);
a72deede 609
84281c31 610 @_ = /(bbb)/;
4f890a30 611 is("@_", "", $message);
a72deede 612
84281c31 613 @_ = /(bbb)/g;
4f890a30 614 is("@_", "", $message);
84281c31 615 }
a72deede 616
84281c31 617 {
4f890a30 618 my $message = '@- and @+ tests';
84281c31
A
619
620 /a(?=.$)/;
4f890a30
NC
621 is($#+, 0, $message);
622 is($#-, 0, $message);
623 is($+ [0], 2, $message);
624 is($- [0], 1, $message);
625 ok(!defined $+ [1] && !defined $- [1] &&
626 !defined $+ [2] && !defined $- [2], $message);
84281c31
A
627
628 /a(a)(a)/;
4f890a30
NC
629 is($#+, 2, $message);
630 is($#-, 2, $message);
631 is($+ [0], 3, $message);
632 is($- [0], 0, $message);
633 is($+ [1], 2, $message);
634 is($- [1], 1, $message);
635 is($+ [2], 3, $message);
636 is($- [2], 2, $message);
637 ok(!defined $+ [3] && !defined $- [3] &&
638 !defined $+ [4] && !defined $- [4], $message);
84281c31 639
54a4274e 640 # Exists has a special check for @-/@+ - bug 45147
4f890a30
NC
641 ok(exists $-[0], $message);
642 ok(exists $+[0], $message);
643 ok(exists $-[2], $message);
644 ok(exists $+[2], $message);
645 ok(!exists $-[3], $message);
646 ok(!exists $+[3], $message);
647 ok(exists $-[-1], $message);
648 ok(exists $+[-1], $message);
649 ok(exists $-[-3], $message);
650 ok(exists $+[-3], $message);
651 ok(!exists $-[-4], $message);
652 ok(!exists $+[-4], $message);
84281c31
A
653
654 /.(a)(b)?(a)/;
4f890a30
NC
655 is($#+, 3, $message);
656 is($#-, 3, $message);
657 is($+ [1], 2, $message);
658 is($- [1], 1, $message);
659 is($+ [3], 3, $message);
660 is($- [3], 2, $message);
661 ok(!defined $+ [2] && !defined $- [2] &&
662 !defined $+ [4] && !defined $- [4], $message);
84281c31 663
84281c31 664 /.(a)/;
4f890a30
NC
665 is($#+, 1, $message);
666 is($#-, 1, $message);
667 is($+ [0], 2, $message);
668 is($- [0], 0, $message);
669 is($+ [1], 2, $message);
670 is($- [1], 1, $message);
671 ok(!defined $+ [2] && !defined $- [2] &&
672 !defined $+ [3] && !defined $- [3], $message);
84281c31
A
673
674 /.(a)(ba*)?/;
4f890a30
NC
675 is($#+, 2, $message);
676 is($#-, 1, $message);
84281c31 677 }
a72deede 678
d9bad346
FC
679 foreach ('$+[0] = 13', '$-[0] = 13', '@+ = (7, 6, 5)',
680 '@- = qw (foo bar)', '$^N = 42') {
cb124425
NC
681 is(eval $_, undef);
682 like($@, qr/^Modification of a read-only value attempted/,
d9bad346 683 '$^N, @- and @+ are read-only');
84281c31 684 }
a72deede 685
84281c31 686 {
4f890a30 687 my $message = '\G testing';
84281c31
A
688 $_ = 'aaa';
689 pos = 1;
690 my @a = /\Ga/g;
4f890a30 691 is("@a", "a a", $message);
84281c31
A
692
693 my $str = 'abcde';
694 pos $str = 2;
4f890a30
NC
695 unlike($str, qr/^\G/, $message);
696 unlike($str, qr/^.\G/, $message);
697 like($str, qr/^..\G/, $message);
698 unlike($str, qr/^...\G/, $message);
699 ok($str =~ /\G../ && $& eq 'cd', $message);
84281c31 700
04934b6d 701 local $::TODO = $::running_as_thread;
4f890a30 702 ok($str =~ /.\G./ && $& eq 'bc', $message);
84281c31 703 }
a72deede 704
84281c31 705 {
4f890a30 706 my $message = 'pos inside (?{ })';
84281c31
A
707 my $str = 'abcde';
708 our ($foo, $bar);
4f890a30
NC
709 like($str, qr/b(?{$foo = $_; $bar = pos})c/, $message);
710 is($foo, $str, $message);
711 is($bar, 2, $message);
712 is(pos $str, undef, $message);
84281c31
A
713
714 undef $foo;
715 undef $bar;
716 pos $str = undef;
4f890a30
NC
717 ok($str =~ /b(?{$foo = $_; $bar = pos})c/g, $message);
718 is($foo, $str, $message);
719 is($bar, 2, $message);
720 is(pos $str, 3, $message);
84281c31
A
721
722 $_ = $str;
723 undef $foo;
724 undef $bar;
4f890a30
NC
725 like($_, qr/b(?{$foo = $_; $bar = pos})c/, $message);
726 is($foo, $str, $message);
727 is($bar, 2, $message);
84281c31
A
728
729 undef $foo;
730 undef $bar;
4f890a30
NC
731 ok(/b(?{$foo = $_; $bar = pos})c/g, $message);
732 is($foo, $str, $message);
733 is($bar, 2, $message);
734 is(pos, 3, $message);
84281c31
A
735
736 undef $foo;
737 undef $bar;
738 pos = undef;
739 1 while /b(?{$foo = $_; $bar = pos})c/g;
4f890a30
NC
740 is($foo, $str, $message);
741 is($bar, 2, $message);
742 is(pos, undef, $message);
84281c31
A
743
744 undef $foo;
745 undef $bar;
746 $_ = 'abcde|abcde';
4f890a30
NC
747 ok(s/b(?{$foo = $_; $bar = pos})c/x/g, $message);
748 is($foo, 'abcde|abcde', $message);
749 is($bar, 8, $message);
750 is($_, 'axde|axde', $message);
84281c31
A
751
752 # List context:
753 $_ = 'abcde|abcde';
754 our @res;
755 () = /([ace]).(?{push @res, $1,$2})([ce])(?{push @res, $1,$2})/g;
756 @res = map {defined $_ ? "'$_'" : 'undef'} @res;
4f890a30 757 is("@res", "'a' undef 'a' 'c' 'e' undef 'a' undef 'a' 'c'", $message);
84281c31
A
758
759 @res = ();
760 () = /([ace]).(?{push @res, $`,$&,$'})([ce])(?{push @res, $`,$&,$'})/g;
761 @res = map {defined $_ ? "'$_'" : 'undef'} @res;
4f890a30 762 is("@res", "'' 'ab' 'cde|abcde' " .
84281c31
A
763 "'' 'abc' 'de|abcde' " .
764 "'abcd' 'e|' 'abcde' " .
765 "'abcde|' 'ab' 'cde' " .
4f890a30 766 "'abcde|' 'abc' 'de'", $message);
84281c31 767 }
f33976b4 768
84281c31 769 {
4f890a30 770 my $message = '\G anchor checks';
84281c31
A
771 my $foo = 'aabbccddeeffgg';
772 pos ($foo) = 1;
773 {
04934b6d 774 local $::TODO = $::running_as_thread;
84281c31 775 no warnings 'uninitialized';
4f890a30
NC
776 ok($foo =~ /.\G(..)/g, $message);
777 is($1, 'ab', $message);
cce850e4 778
84281c31 779 pos ($foo) += 1;
4f890a30
NC
780 ok($foo =~ /.\G(..)/g, $message);
781 is($1, 'cc', $message);
cce850e4 782
84281c31 783 pos ($foo) += 1;
4f890a30
NC
784 ok($foo =~ /.\G(..)/g, $message);
785 is($1, 'de', $message);
cce850e4 786
4f890a30 787 ok($foo =~ /\Gef/g, $message);
84281c31 788 }
cce850e4 789
84281c31 790 undef pos $foo;
4f890a30
NC
791 ok($foo =~ /\G(..)/g, $message);
792 is($1, 'aa', $message);
cce850e4 793
4f890a30
NC
794 ok($foo =~ /\G(..)/g, $message);
795 is($1, 'bb', $message);
cce850e4 796
84281c31 797 pos ($foo) = 5;
4f890a30
NC
798 ok($foo =~ /\G(..)/g, $message);
799 is($1, 'cd', $message);
84281c31 800 }
cce850e4 801
84281c31
A
802 {
803 $_ = '123x123';
804 my @res = /(\d*|x)/g;
805 local $" = '|';
de26e0cc 806 is("@res", "123||x|123|", "0 match in alternation");
84281c31 807 }
cce850e4 808
84281c31 809 {
4f890a30 810 my $message = "Match against temporaries (created via pp_helem())" .
84281c31 811 " is safe";
4f890a30
NC
812 ok({foo => "bar\n" . $^X} -> {foo} =~ /^(.*)\n/g, $message);
813 is($1, "bar", $message);
84281c31 814 }
75685a94 815
84281c31 816 {
4f890a30 817 my $message = 'package $i inside (?{ }), ' .
84281c31
A
818 'saved substrings and changing $_';
819 our @a = qw [foo bar];
820 our @b = ();
821 s/(\w)(?{push @b, $1})/,$1,/g for @a;
4f890a30
NC
822 is("@b", "f o o b a r", $message);
823 is("@a", ",f,,o,,o, ,b,,a,,r,", $message);
84281c31 824
4f890a30 825 $message = 'lexical $i inside (?{ }), ' .
84281c31
A
826 'saved substrings and changing $_';
827 no warnings 'closure';
828 my @c = qw [foo bar];
829 my @d = ();
830 s/(\w)(?{push @d, $1})/,$1,/g for @c;
4f890a30
NC
831 is("@d", "f o o b a r", $message);
832 is("@c", ",f,,o,,o, ,b,,a,,r,", $message);
d9f424b2
JH
833 }
834
84281c31 835 {
4f890a30 836 my $message = 'Brackets';
84281c31
A
837 our $brackets;
838 $brackets = qr {
839 { (?> [^{}]+ | (??{ $brackets }) )* }
840 }x;
841
4f890a30
NC
842 ok("{{}" =~ $brackets, $message);
843 is($&, "{}", $message);
844 ok("something { long { and } hairy" =~ $brackets, $message);
845 is($&, "{ and }", $message);
846 ok("something { long { and } hairy" =~ m/((??{ $brackets }))/, $message);
847 is($&, "{ and }", $message);
84281c31 848 }
a4c04bdc 849
84281c31
A
850 {
851 $_ = "a-a\nxbb";
852 pos = 1;
b33825c4 853 ok(!m/^-.*bb/mg, '$_ = "a-a\nxbb"; m/^-.*bb/mg');
84281c31 854 }
a4c04bdc 855
84281c31 856 {
4f890a30 857 my $message = '\G anchor checks';
84281c31
A
858 my $text = "aaXbXcc";
859 pos ($text) = 0;
4f890a30 860 ok($text !~ /\GXb*X/g, $message);
84281c31 861 }
a4c04bdc 862
84281c31
A
863 {
864 $_ = "xA\n" x 500;
b33825c4 865 unlike($_, qr/^\s*A/m, '$_ = "xA\n" x 500; /^\s*A/m"');
a4c04bdc 866
84281c31
A
867 my $text = "abc dbf";
868 my @res = ($text =~ /.*?(b).*?\b/g);
de26e0cc 869 is("@res", "b b", '\b is not special');
987aaf07 870 }
a4c04bdc 871
84281c31 872 {
4f890a30 873 my $message = '\S, [\S], \s, [\s]';
84281c31 874 my @a = map chr, 0 .. 255;
9d45b377
YO
875 my @b = grep m/\S/, @a;
876 my @c = grep m/[^\s]/, @a;
4f890a30 877 is("@b", "@c", $message);
84281c31
A
878
879 @b = grep /\S/, @a;
880 @c = grep /[\S]/, @a;
4f890a30 881 is("@b", "@c", $message);
84281c31
A
882
883 @b = grep /\s/, @a;
884 @c = grep /[^\S]/, @a;
4f890a30 885 is("@b", "@c", $message);
84281c31
A
886
887 @b = grep /\s/, @a;
888 @c = grep /[\s]/, @a;
4f890a30 889 is("@b", "@c", $message);
84281c31
A
890 }
891 {
4f890a30 892 my $message = '\D, [\D], \d, [\d]';
84281c31
A
893 my @a = map chr, 0 .. 255;
894 my @b = grep /\D/, @a;
895 my @c = grep /[^\d]/, @a;
4f890a30 896 is("@b", "@c", $message);
84281c31
A
897
898 @b = grep /\D/, @a;
899 @c = grep /[\D]/, @a;
4f890a30 900 is("@b", "@c", $message);
84281c31
A
901
902 @b = grep /\d/, @a;
903 @c = grep /[^\D]/, @a;
4f890a30 904 is("@b", "@c", $message);
84281c31
A
905
906 @b = grep /\d/, @a;
907 @c = grep /[\d]/, @a;
4f890a30 908 is("@b", "@c", $message);
84281c31
A
909 }
910 {
4f890a30 911 my $message = '\W, [\W], \w, [\w]';
84281c31
A
912 my @a = map chr, 0 .. 255;
913 my @b = grep /\W/, @a;
914 my @c = grep /[^\w]/, @a;
4f890a30 915 is("@b", "@c", $message);
84281c31
A
916
917 @b = grep /\W/, @a;
918 @c = grep /[\W]/, @a;
4f890a30 919 is("@b", "@c", $message);
84281c31
A
920
921 @b = grep /\w/, @a;
922 @c = grep /[^\W]/, @a;
4f890a30 923 is("@b", "@c", $message);
84281c31
A
924
925 @b = grep /\w/, @a;
926 @c = grep /[\w]/, @a;
4f890a30 927 is("@b", "@c", $message);
84281c31 928 }
a4c04bdc 929
84281c31
A
930 {
931 # see if backtracking optimization works correctly
4f890a30
NC
932 my $message = 'Backtrack optimization';
933 like("\n\n", qr/\n $ \n/x, $message);
934 like("\n\n", qr/\n* $ \n/x, $message);
935 like("\n\n", qr/\n+ $ \n/x, $message);
936 like("\n\n", qr/\n? $ \n/x, $message);
937 like("\n\n", qr/\n*? $ \n/x, $message);
938 like("\n\n", qr/\n+? $ \n/x, $message);
939 like("\n\n", qr/\n?? $ \n/x, $message);
940 unlike("\n\n", qr/\n*+ $ \n/x, $message);
941 unlike("\n\n", qr/\n++ $ \n/x, $message);
942 like("\n\n", qr/\n?+ $ \n/x, $message);
84281c31 943 }
a4c04bdc 944
84281c31
A
945 {
946 package S;
947 use overload '""' => sub {'Object S'};
948 sub new {bless []}
0f289c68 949
4f890a30 950 my $message = "Ref stringification";
5895685f
NC
951 ::ok(do { \my $v} =~ /^SCALAR/, "Scalar ref stringification") or diag($message);
952 ::ok(do {\\my $v} =~ /^REF/, "Ref ref stringification") or diag($message);
953 ::ok([] =~ /^ARRAY/, "Array ref stringification") or diag($message);
954 ::ok({} =~ /^HASH/, "Hash ref stringification") or diag($message);
955 ::ok('S' -> new =~ /^Object S/, "Object stringification") or diag($message);
84281c31 956 }
a4c04bdc 957
84281c31 958 {
4f890a30
NC
959 my $message = "Test result of match used as match";
960 ok('a1b' =~ ('xyz' =~ /y/), $message);
961 is($`, 'a', $message);
962 ok('a1b' =~ ('xyz' =~ /t/), $message);
963 is($`, 'a', $message);
84281c31 964 }
a4c04bdc 965
84281c31 966 {
d728c370 967 my $message = '"1" is not \s';
c11a8df3
NC
968 warning_is(sub {unlike("1\n" x 102, qr/^\s*\n/m, $message)},
969 undef, "$message (did not warn)");
84281c31 970 }
a4c04bdc 971
84281c31 972 {
4f890a30 973 my $message = '\s, [[:space:]] and [[:blank:]]';
84281c31
A
974 my %space = (spc => " ",
975 tab => "\t",
976 cr => "\r",
977 lf => "\n",
978 ff => "\f",
979 # There's no \v but the vertical tabulator seems miraculously
980 # be 11 both in ASCII and EBCDIC.
981 vt => chr(11),
982 false => "space");
983
984 my @space0 = sort grep {$space {$_} =~ /\s/ } keys %space;
985 my @space1 = sort grep {$space {$_} =~ /[[:space:]]/} keys %space;
986 my @space2 = sort grep {$space {$_} =~ /[[:blank:]]/} keys %space;
987
075b9d7d 988 is("@space0", "cr ff lf spc tab vt", $message);
4f890a30
NC
989 is("@space1", "cr ff lf spc tab vt", $message);
990 is("@space2", "spc tab", $message);
84281c31 991 }
a4c04bdc 992
ff3f963a 993 {
c9415951 994 my $n= 50;
93f09d7b 995 # this must be a high number and go from 0 to N, as the bug we are looking for doesn't
c9415951
YO
996 # seem to be predictable. Slight changes to the test make it fail earlier or later.
997 foreach my $i (0 .. $n)
998 {
999 my $str= "\n" x $i;
93f09d7b 1000 ok $str=~/.*\z/, "implicit MBOL check string disable does not break things length=$i";
c9415951
YO
1001 }
1002 }
92f3d482
YO
1003 {
1004 # we are actually testing that we dont die when executing these patterns
1005 use utf8;
1006 my $e = "Böck";
1007 ok(utf8::is_utf8($e),"got a unicode string - rt75680");
1008
1009 ok($e !~ m/.*?[x]$/, "unicode string against /.*?[x]\$/ - rt75680");
1010 ok($e !~ m/.*?\p{Space}$/i, "unicode string against /.*?\\p{space}\$/i - rt75680");
1011 ok($e !~ m/.*?[xyz]$/, "unicode string against /.*?[xyz]\$/ - rt75680");
1012 ok($e !~ m/(.*?)[,\p{isSpace}]+((?:\p{isAlpha}[\p{isSpace}\.]{1,2})+)\p{isSpace}*$/, "unicode string against big pattern - rt75680");
1013 }
1014 {
1015 # we are actually testing that we dont die when executing these patterns
1016 my $e = "B\x{f6}ck";
1017 ok(!utf8::is_utf8($e), "got a latin string - rt75680");
1018
1019 ok($e !~ m/.*?[x]$/, "latin string against /.*?[x]\$/ - rt75680");
1020 ok($e !~ m/.*?\p{Space}$/i, "latin string against /.*?\\p{space}\$/i - rt75680");
1021 ok($e !~ m/.*?[xyz]$/,"latin string against /.*?[xyz]\$/ - rt75680");
1022 ok($e !~ m/(.*?)[,\p{isSpace}]+((?:\p{isAlpha}[\p{isSpace}\.]{1,2})+)\p{isSpace}*$/,"latin string against big pattern - rt75680");
1023 }
c920e018
A
1024
1025 {
1026 #
1027 # Tests for bug 77414.
1028 #
1029
4f890a30 1030 my $message = '\p property after empty * match';
c920e018 1031 {
4f890a30
NC
1032 like("1", qr/\s*\pN/, $message);
1033 like("-", qr/\s*\p{Dash}/, $message);
1034 like(" ", qr/\w*\p{Blank}/, $message);
c920e018
A
1035 }
1036
4f890a30
NC
1037 like("1", qr/\s*\pN+/, $message);
1038 like("-", qr/\s*\p{Dash}{1}/, $message);
1039 like(" ", qr/\w*\p{Blank}{1,4}/, $message);
c920e018
A
1040
1041 }
1042
7c17ea2f
KW
1043 SKIP: { # Some constructs with Latin1 characters cause a utf8 string not
1044 # to match itself in non-utf8
ef237063 1045 if ($::IS_EBCDIC) {
7c17ea2f
KW
1046 skip "Needs to be customized to run on EBCDIC", 6;
1047 }
634c83a2
KW
1048 my $c = "\xc0";
1049 my $pattern = my $utf8_pattern = qr/((\xc0)+,?)/;
1050 utf8::upgrade($utf8_pattern);
1051 ok $c =~ $pattern, "\\xc0 =~ $pattern; Neither pattern nor target utf8";
d4e0b827 1052 ok $c =~ /$pattern/i, "\\xc0 =~ /$pattern/i; Neither pattern nor target utf8";
634c83a2 1053 ok $c =~ $utf8_pattern, "\\xc0 =~ $pattern; pattern utf8, target not";
d4e0b827 1054 ok $c =~ /$utf8_pattern/i, "\\xc0 =~ /$pattern/i; pattern utf8, target not";
634c83a2
KW
1055 utf8::upgrade($c);
1056 ok $c =~ $pattern, "\\xc0 =~ $pattern; target utf8, pattern not";
d4e0b827 1057 ok $c =~ /$pattern/i, "\\xc0 =~ /$pattern/i; target utf8, pattern not";
634c83a2 1058 ok $c =~ $utf8_pattern, "\\xc0 =~ $pattern; Both target and pattern utf8";
d4e0b827 1059 ok $c =~ /$utf8_pattern/i, "\\xc0 =~ /$pattern/i; Both target and pattern utf8";
634c83a2
KW
1060 }
1061
8cc86590 1062 SKIP: { # Make sure can override the formatting
ef237063 1063 if ($::IS_EBCDIC) {
8cc86590
KW
1064 skip "Needs to be customized to run on EBCDIC", 2;
1065 }
1066 use feature 'unicode_strings';
1067 ok "\xc0" =~ /\w/, 'Under unicode_strings: "\xc0" =~ /\w/';
1068 ok "\xc0" !~ /(?d:\w)/, 'Under unicode_strings: "\xc0" !~ /(?d:\w)/';
1069 }
1070
704f71be 1071 {
5b6010b3
YO
1072 my $str= "\x{100}";
1073 chop $str;
1074 my $qr= qr/$str/;
de26e0cc 1075 is("$qr", "(?^:)", "Empty pattern qr// stringifies to (?^:) with unicode flag enabled - Bug #80212");
5b6010b3
YO
1076 $str= "";
1077 $qr= qr/$str/;
de26e0cc 1078 is("$qr", "(?^:)", "Empty pattern qr// stringifies to (?^:) with unicode flag disabled - Bug #80212");
5b6010b3
YO
1079
1080 }
1081
72aa120d 1082 {
04934b6d 1083 local $::TODO = "[perl #38133]";
72aa120d
KW
1084
1085 "A" =~ /(((?:A))?)+/;
1086 my $first = $2;
1087
1088 "A" =~ /(((A))?)+/;
1089 my $second = $2;
1090
de26e0cc 1091 is($first, $second);
72aa120d
KW
1092 }
1093
99ca48e1
DM
1094 {
1095 # RT #3516: \G in a m//g expression causes problems
1096 my $count = 0;
1097 while ("abc" =~ m/(\G[ac])?/g) {
1098 last if $count++ > 10;
1099 }
1100 ok($count < 10, 'RT #3516 A');
1101
1102 $count = 0;
1103 while ("abc" =~ m/(\G|.)[ac]/g) {
1104 last if $count++ > 10;
1105 }
1106 ok($count < 10, 'RT #3516 B');
1107
1108 $count = 0;
1109 while ("abc" =~ m/(\G?[ac])?/g) {
1110 last if $count++ > 10;
1111 }
1112 ok($count < 10, 'RT #3516 C');
1113 }
d774cd11
YO
1114 {
1115 # RT #84294: Is this a bug in the simple Perl regex?
1116 # : Nested buffers and (?{...}) dont play nicely on partial matches
1117 our @got= ();
1118 ok("ab" =~ /((\w+)(?{ push @got, $2 })){2}/,"RT #84294: Pattern should match");
d774cd11
YO
1119 my $want= "'ab', 'a', 'b'";
1120 my $got= join(", ", map { defined($_) ? "'$_'" : "undef" } @got);
1121 is($got,$want,'RT #84294: check that "ab" =~ /((\w+)(?{ push @got, $2 })){2}/ leaves @got in the correct state');
1122 }
1123
0bda3001
KW
1124 {
1125 # Suppress warnings, as the non-unicode one comes out even if turn off
1126 # warnings here (because the execution is done in another scope).
1127 local $SIG{__WARN__} = sub {};
1128 my $str = "\x{110000}";
1129
1130 # No non-unicode code points match any Unicode property, even inverse
1131 # ones
1132 unlike($str, qr/\p{ASCII_Hex_Digit=True}/, "Non-Unicode doesn't match \\p{}");
1133 unlike($str, qr/\p{ASCII_Hex_Digit=False}/, "Non-Unicode doesn't match \\p{}");
1134 like($str, qr/\P{ASCII_Hex_Digit=True}/, "Non-Unicode matches \\P{}");
1135 like($str, qr/\P{ASCII_Hex_Digit=False}/, "Non-Unicode matches \\P{}");
1136 }
1137
b289b443 1138 {
c256547e 1139 # Test that IDstart works, but because the author (khw) knows
b289b443
KW
1140 # regexes much better than the rest of the core, it is being done here
1141 # in the context of a regex which relies on buffer names beginng with
1142 # IDStarts.
1143 use utf8;
1144 my $str = "abc";
1145 like($str, qr/(?<a>abc)/, "'a' is legal IDStart");
1146 like($str, qr/(?<_>abc)/, "'_' is legal IDStart");
1147 like($str, qr/(?<ß>abc)/, "U+00DF is legal IDStart");
1148 like($str, qr/(?<ℕ>abc)/, "U+2115' is legal IDStart");
1149
1150 # This test works on Unicode 6.0 in which U+2118 and U+212E are legal
1151 # IDStarts there, but are not Word characters, and therefore Perl
1152 # doesn't allow them to be IDStarts. But there is no guarantee that
1153 # Unicode won't change things around in the future so that at some
1154 # future Unicode revision these tests would need to be revised.
1155 foreach my $char ("%", "×", chr(0x2118), chr(0x212E)) {
1156 my $prog = <<"EOP";
1157use utf8;;
1158"abc" =~ qr/(?<$char>abc)/;
1159EOP
1160 utf8::encode($prog);
1f4f6bf1 1161 fresh_perl_like($prog, qr!Group name must start with a non-digit word character!, "",
b289b443
KW
1162 sprintf("'U+%04X not legal IDFirst'", ord($char)));
1163 }
1164 }
bbdd8bad
KW
1165
1166 { # [perl #101710]
1167 my $pat = "b";
1168 utf8::upgrade($pat);
1169 like("\xffb", qr/$pat/i, "/i: utf8 pattern, non-utf8 string, latin1-char preceding matching char in string");
1170 }
1171
ba510004
FC
1172 { # Crash with @a =~ // warning
1173 local $SIG{__WARN__} = sub {
1174 pass 'no crash for @a =~ // warning'
1175 };
1176 eval ' sub { my @a =~ // } ';
1177 }
1178
cc88c9aa
FC
1179 { # Concat overloading and qr// thingies
1180 my @refs;
1181 my $qr = qr//;
3bbc3d30
YO
1182 package Cat {
1183 require overload;
1184 overload->import(
cc88c9aa
FC
1185 '""' => sub { ${$_[0]} },
1186 '.' => sub {
1187 push @refs, ref $_[1] if ref $_[1];
1188 bless $_[2] ? \"$_[1]${$_[0]}" : \"${$_[0]}$_[1]"
1189 }
3bbc3d30 1190 );
cc88c9aa
FC
1191 }
1192 my $s = "foo";
1193 my $o = bless \$s, Cat::;
1194 /$o$qr/;
1195 is "@refs", "Regexp", '/$o$qr/ passes qr ref to cat overload meth';
1196 }
1197
21eede78
YO
1198 {
1199 my $count=0;
1200 my $str="\n";
1201 $count++ while $str=~/.*/g;
1202 is $count, 2, 'test that ANCH_MBOL works properly. We should get 2 from $count++ while "\n"=~/.*/g';
1203 my $class_count= 0;
1204 $class_count++ while $str=~/[^\n]*/g;
1205 is $class_count, $count, 'while "\n"=~/.*/g and while "\n"=~/[^\n]*/g should behave the same';
1206 my $anch_count= 0;
1207 $anch_count++ while $str=~/^.*/mg;
1208 is $anch_count, 1, 'while "\n"=~/^.*/mg should match only once';
1209 }
96f54887
KW
1210
1211 { # [perl #111174]
1212 use re '/u';
1213 like "\xe0", qr/(?i:\xc0)/, "(?i: shouldn't lose the passed in /u";
1214 use re '/a';
1215 unlike "\x{100}", qr/(?i:\w)/, "(?i: shouldn't lose the passed in /a";
1216 use re '/aa';
1217 unlike 'k', qr/(?i:\N{KELVIN SIGN})/, "(?i: shouldn't lose the passed in /aa";
1218 }
6ae44cd2
DM
1219
1220 {
1221 # the test for whether the pattern should be re-compiled should
1222 # consider the UTF8ness of the previous and current pattern
1223 # string, as well as the physical bytes of the pattern string
1224
1225 for my $s ("\xc4\x80", "\x{100}") {
1226 ok($s =~ /^$s$/, "re-compile check is UTF8-aware");
1227 }
1228 }
1229
e03b874a
DM
1230 # #113682 more overloading and qr//
1231 # when doing /foo$overloaded/, if $overloaded returns
1232 # a qr/(?{})/ via qr or "" overloading, then 'use re 'eval'
1233 # shouldn't be required. Via '.', it still is.
1234 {
1235 package Qr0;
1236 use overload 'qr' => sub { qr/(??{50})/ };
1237
1238 package Qr1;
1239 use overload '""' => sub { qr/(??{51})/ };
1240
1241 package Qr2;
1242 use overload '.' => sub { $_[1] . qr/(??{52})/ };
1243
1244 package Qr3;
1245 use overload '""' => sub { qr/(??{7})/ },
1246 '.' => sub { $_[1] . qr/(??{53})/ };
1247
1248 package Qr_indirect;
1249 use overload '""' => sub { $_[0][0] };
1250
1251 package main;
1252
1253 for my $i (0..3) {
1254 my $o = bless [], "Qr$i";
1255 if ((0,0,1,1)[$i]) {
1256 eval { "A5$i" =~ /^A$o$/ };
1257 like($@, qr/Eval-group not allowed/, "Qr$i");
1258 eval { "5$i" =~ /$o/ };
1259 like($@, ($i == 3 ? qr/^$/ : qr/no method found,/),
1260 "Qr$i bare");
1261 {
1262 use re 'eval';
1263 ok("A5$i" =~ /^A$o$/, "Qr$i - with use re eval");
1264 eval { "5$i" =~ /$o/ };
1265 like($@, ($i == 3 ? qr/^$/ : qr/no method found,/),
1266 "Qr$i bare - with use re eval");
1267 }
1268 }
1269 else {
1270 ok("A5$i" =~ /^A$o$/, "Qr$i");
1271 ok("5$i" =~ /$o/, "Qr$i bare");
1272 }
1273 }
1274
1275 my $o = bless [ bless [], "Qr1" ], 'Qr_indirect';
1276 ok("A51" =~ /^A$o/, "Qr_indirect");
1277 ok("51" =~ /$o/, "Qr_indirect bare");
1278 }
1279
34b39fc9
KW
1280 { # Various flags weren't being set when a [] is optimized into an
1281 # EXACTish node
1282 ;
1283 ;
1284 ok("\x{017F}\x{017F}" =~ qr/^[\x{00DF}]?$/i, "[] to EXACTish optimization");
1285 }
1286
5e4a1da1
KW
1287 {
1288 for my $char (":", "\x{f7}", "\x{2010}") {
1289 my $utf8_char = $char;
1290 utf8::upgrade($utf8_char);
1291 my $display = $char;
1292 $display = display($display);
1293 my $utf8_display = "utf8::upgrade(\"$display\")";
1294
1295 like($char, qr/^$char?$/, "\"$display\" =~ /^$display?\$/");
1296 like($char, qr/^$utf8_char?$/, "my \$p = \"$display\"; utf8::upgrade(\$p); \"$display\" =~ /^\$p?\$/");
1297 like($utf8_char, qr/^$char?$/, "my \$c = \"$display\"; utf8::upgrade(\$c); \"\$c\" =~ /^$display?\$/");
1298 like($utf8_char, qr/^$utf8_char?$/, "my \$c = \"$display\"; utf8::upgrade(\$c); my \$p = \"$display\"; utf8::upgrade(\$p); \"\$c\" =~ /^\$p?\$/");
1299 }
1300 }
1301
4fab19ce
DM
1302 {
1303 # #116148: Pattern utf8ness sticks around globally
1304 # the utf8 in the first match was sticking around for the second
1305 # match
1306
1307 use feature 'unicode_strings';
1308
1309 my $x = "\x{263a}";
1310 $x =~ /$x/;
1311
1312 my $text = "Perl";
1313 ok("Perl" =~ /P.*$/i, '#116148');
1314 }
1315
1316
84281c31
A
1317} # End of sub run_tests
1318
13191;