This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
t/op/pat.t
[perl5.git] / t / op / pat.t
CommitLineData
8d063cd8 1#!./perl
8d37f932
DD
2#
3# This is a home for regular expression tests that don't fit into
4# the format supported by op/regexp.t. If you want to add a test
5# that does fit that format, add it to op/re_tests, not here.
8d063cd8 6
9133bbab 7$| = 1;
3568d838 8
fc8cd66c 9# Test counter output is generated by a BEGIN block at bottom of file
8d37f932 10
e4d48cc9
GS
11BEGIN {
12 chdir 't' if -d 't';
20822f61 13 @INC = '../lib';
e4d48cc9 14}
de734bd5 15our $Message = "Noname test";
ffbc6a93 16
8d37f932 17eval 'use Config'; # Defaults assumed if this fails
8d063cd8 18
e3faa678
NC
19run_tests() unless caller;
20
21sub run_tests {
22
8d063cd8
LW
23$x = "abc\ndef\n";
24
25if ($x =~ /^abc/) {print "ok 1\n";} else {print "not ok 1\n";}
26if ($x !~ /^def/) {print "ok 2\n";} else {print "not ok 2\n";}
27
f02c194e
RGS
28# used to be a test for $*
29if ($x =~ /^def/m) {print "ok 3\n";} else {print "not ok 3\n";}
8d063cd8
LW
30
31$_ = '123';
32if (/^([0-9][0-9]*)/) {print "ok 4\n";} else {print "not ok 4\n";}
33
34if ($x =~ /^xxx/) {print "not ok 5\n";} else {print "ok 5\n";}
35if ($x !~ /^abc/) {print "not ok 6\n";} else {print "ok 6\n";}
36
37if ($x =~ /def/) {print "ok 7\n";} else {print "not ok 7\n";}
38if ($x !~ /def/) {print "not ok 8\n";} else {print "ok 8\n";}
39
40if ($x !~ /.def/) {print "ok 9\n";} else {print "not ok 9\n";}
41if ($x =~ /.def/) {print "not ok 10\n";} else {print "ok 10\n";}
42
43if ($x =~ /\ndef/) {print "ok 11\n";} else {print "not ok 11\n";}
44if ($x !~ /\ndef/) {print "not ok 12\n";} else {print "ok 12\n";}
45
46$_ = 'aaabbbccc';
47if (/(a*b*)(c*)/ && $1 eq 'aaabbb' && $2 eq 'ccc') {
48 print "ok 13\n";
49} else {
50 print "not ok 13\n";
51}
52if (/(a+b+c+)/ && $1 eq 'aaabbbccc') {
53 print "ok 14\n";
54} else {
55 print "not ok 14\n";
56}
57
58if (/a+b?c+/) {print "not ok 15\n";} else {print "ok 15\n";}
59
60$_ = 'aaabccc';
61if (/a+b?c+/) {print "ok 16\n";} else {print "not ok 16\n";}
62if (/a*b+c*/) {print "ok 17\n";} else {print "not ok 17\n";}
63
64$_ = 'aaaccc';
65if (/a*b?c*/) {print "ok 18\n";} else {print "not ok 18\n";}
66if (/a*b+c*/) {print "not ok 19\n";} else {print "ok 19\n";}
67
68$_ = 'abcdef';
69if (/bcd|xyz/) {print "ok 20\n";} else {print "not ok 20\n";}
70if (/xyz|bcd/) {print "ok 21\n";} else {print "not ok 21\n";}
71
72if (m|bc/*d|) {print "ok 22\n";} else {print "not ok 22\n";}
378cc40b
LW
73
74if (/^$_$/) {print "ok 23\n";} else {print "not ok 23\n";}
75
f02c194e
RGS
76# used to be a test for $*
77if ("ab\ncd\n" =~ /^cd/m) {print "ok 24\n";} else {print "not ok 24\n";}
378cc40b 78
cb55de95
JH
79$XXX{123} = 123;
80$XXX{234} = 234;
81$XXX{345} = 345;
82
83@XXX = ('ok 25','not ok 25', 'ok 26','not ok 26','not ok 27');
84while ($_ = shift(@XXX)) {
85 ?(.*)? && (print $1,"\n");
86 /not/ && reset;
1606b0c7
JM
87 if (/not ok 26/) {
88 if ($^O eq 'VMS') {
89 $_ = shift(@XXX);
90 }
91 else {
92 reset 'X';
93 }
94 }
cb55de95
JH
95}
96
1606b0c7
JM
97if ($^O ne 'VMS') {
98 while (($key,$val) = each(%XXX)) {
cb55de95
JH
99 print "not ok 27\n";
100 exit;
1606b0c7 101 }
cb55de95
JH
102}
103
104print "ok 27\n";
378cc40b 105
378cc40b
LW
106'cde' =~ /[^ab]*/;
107'xyz' =~ //;
108if ($& eq 'xyz') {print "ok 28\n";} else {print "not ok 28\n";}
109
110$foo = '[^ab]*';
111'cde' =~ /$foo/;
112'xyz' =~ //;
113if ($& eq 'xyz') {print "ok 29\n";} else {print "not ok 29\n";}
114
115$foo = '[^ab]*';
116'cde' =~ /$foo/;
117'xyz' =~ /$null/;
118if ($& eq 'xyz') {print "ok 30\n";} else {print "not ok 30\n";}
a687059c
LW
119
120$_ = 'abcdefghi';
121/def/; # optimized up to cmd
122if ("$`:$&:$'" eq 'abc:def:ghi') {print "ok 31\n";} else {print "not ok 31\n";}
123
124/cde/ + 0; # optimized only to spat
125if ("$`:$&:$'" eq 'ab:cde:fghi') {print "ok 32\n";} else {print "not ok 32\n";}
126
127/[d][e][f]/; # not optimized
128if ("$`:$&:$'" eq 'abc:def:ghi') {print "ok 33\n";} else {print "not ok 33\n";}
129
130$_ = 'now is the {time for all} good men to come to.';
131/ {([^}]*)}/;
132if ($1 eq 'time for all') {print "ok 34\n";} else {print "not ok 34 $1\n";}
133
134$_ = 'xxx {3,4} yyy zzz';
135print /( {3,4})/ ? "ok 35\n" : "not ok 35\n";
136print $1 eq ' ' ? "ok 36\n" : "not ok 36\n";
137print /( {4,})/ ? "not ok 37\n" : "ok 37\n";
138print /( {2,3}.)/ ? "ok 38\n" : "not ok 38\n";
139print $1 eq ' y' ? "ok 39\n" : "not ok 39\n";
140print /(y{2,3}.)/ ? "ok 40\n" : "not ok 40\n";
141print $1 eq 'yyy ' ? "ok 41\n" : "not ok 41\n";
142print /x {3,4}/ ? "not ok 42\n" : "ok 42\n";
143print /^xxx {3,4}/ ? "not ok 43\n" : "ok 43\n";
352d5a3a
LW
144
145$_ = "now is the time for all good men to come to.";
146@words = /(\w+)/g;
147print join(':',@words) eq "now:is:the:time:for:all:good:men:to:come:to"
148 ? "ok 44\n"
149 : "not ok 44\n";
150
151@words = ();
152while (/\w+/g) {
153 push(@words, $&);
154}
155print join(':',@words) eq "now:is:the:time:for:all:good:men:to:come:to"
156 ? "ok 45\n"
157 : "not ok 45\n";
158
159@words = ();
71be2cbc 160pos = 0;
352d5a3a
LW
161while (/to/g) {
162 push(@words, $&);
163}
164print join(':',@words) eq "to:to"
165 ? "ok 46\n"
71be2cbc 166 : "not ok 46 `@words'\n";
352d5a3a 167
71be2cbc 168pos $_ = 0;
352d5a3a
LW
169@words = /to/g;
170print join(':',@words) eq "to:to"
171 ? "ok 47\n"
71be2cbc 172 : "not ok 47 `@words'\n";
352d5a3a
LW
173
174$_ = "abcdefghi";
175
176$pat1 = 'def';
177$pat2 = '^def';
178$pat3 = '.def.';
179$pat4 = 'abc';
180$pat5 = '^abc';
181$pat6 = 'abc$';
182$pat7 = 'ghi';
183$pat8 = '\w*ghi';
184$pat9 = 'ghi$';
185
186$t1=$t2=$t3=$t4=$t5=$t6=$t7=$t8=$t9=0;
187
188for $iter (1..5) {
189 $t1++ if /$pat1/o;
190 $t2++ if /$pat2/o;
191 $t3++ if /$pat3/o;
192 $t4++ if /$pat4/o;
193 $t5++ if /$pat5/o;
194 $t6++ if /$pat6/o;
195 $t7++ if /$pat7/o;
196 $t8++ if /$pat8/o;
197 $t9++ if /$pat9/o;
198}
199
200$x = "$t1$t2$t3$t4$t5$t6$t7$t8$t9";
201print $x eq '505550555' ? "ok 48\n" : "not ok 48 $x\n";
1462b684
LW
202
203$xyz = 'xyz';
204print "abc" =~ /^abc$|$xyz/ ? "ok 49\n" : "not ok 49\n";
205
206# perl 4.009 says "unmatched ()"
207eval '"abc" =~ /a(bc$)|$xyz/; $result = "$&:$1"';
208print $@ eq "" ? "ok 50\n" : "not ok 50\n";
209print $result eq "abc:bc" ? "ok 51\n" : "not ok 51\n";
a0d0e21e
LW
210
211
212$_="abcfooabcbar";
213$x=/abc/g;
214print $` eq "" ? "ok 52\n" : "not ok 52\n" if $x;
215$x=/abc/g;
216print $` eq "abcfoo" ? "ok 53\n" : "not ok 53\n" if $x;
217$x=/abc/g;
218print $x == 0 ? "ok 54\n" : "not ok 54\n";
71be2cbc 219pos = 0;
a0d0e21e
LW
220$x=/ABC/gi;
221print $` eq "" ? "ok 55\n" : "not ok 55\n" if $x;
222$x=/ABC/gi;
223print $` eq "abcfoo" ? "ok 56\n" : "not ok 56\n" if $x;
224$x=/ABC/gi;
225print $x == 0 ? "ok 57\n" : "not ok 57\n";
71be2cbc 226pos = 0;
a0d0e21e
LW
227$x=/abc/g;
228print $' eq "fooabcbar" ? "ok 58\n" : "not ok 58\n" if $x;
229$x=/abc/g;
230print $' eq "bar" ? "ok 59\n" : "not ok 59\n" if $x;
231$_ .= '';
232@x=/abc/g;
233print scalar @x == 2 ? "ok 60\n" : "not ok 60\n";
71be2cbc
PP
234
235$_ = "abdc";
236pos $_ = 2;
c90c0ff4 237/\Gc/gc;
71be2cbc
PP
238print "not " if (pos $_) != 2;
239print "ok 61\n";
c90c0ff4
PP
240/\Gc/g;
241print "not " if defined pos $_;
242print "ok 62\n";
c277df42
IZ
243
244$out = 1;
245'abc' =~ m'a(?{ $out = 2 })b';
246print "not " if $out != 2;
247print "ok 63\n";
248
249$out = 1;
250'abc' =~ m'a(?{ $out = 3 })c';
251print "not " if $out != 1;
459f542a 252print "ok 64\n";
c277df42
IZ
253
254$_ = 'foobar1 bar2 foobar3 barfoobar5 foobar6';
255@out = /(?<!foo)bar./g;
256print "not " if "@out" ne 'bar2 barf';
257print "ok 65\n";
258
8d37f932
DD
259# Tests which depend on REG_INFTY
260$reg_infty = defined $Config{reg_infty} ? $Config{reg_infty} : 32767;
261$reg_infty_m = $reg_infty - 1; $reg_infty_p = $reg_infty + 1;
262
263# As well as failing if the pattern matches do unexpected things, the
264# next three tests will fail if you should have picked up a lower-than-
265# default value for $reg_infty from Config.pm, but have not.
266
267undef $@;
268print "not " if eval q(('aaa' =~ /(a{1,$reg_infty_m})/)[0] ne 'aaa') || $@;
269print "ok 66\n";
270
271undef $@;
272print "not " if eval q(('a' x $reg_infty_m) !~ /a{$reg_infty_m}/) || $@;
273print "ok 67\n";
274
275undef $@;
276print "not " if eval q(('a' x ($reg_infty_m - 1)) =~ /a{$reg_infty_m}/) || $@;
277print "ok 68\n";
278
279undef $@;
280eval "'aaa' =~ /a{1,$reg_infty}/";
9baa0206 281print "not " if $@ !~ m%^\QQuantifier in {,} bigger than%;
8d37f932
DD
282print "ok 69\n";
283
284eval "'aaa' =~ /a{1,$reg_infty_p}/";
285print "not "
9baa0206 286 if $@ !~ m%^\QQuantifier in {,} bigger than%;
8d37f932
DD
287print "ok 70\n";
288undef $@;
289
290# Poke a couple more parse failures
291
292$context = 'x' x 256;
293eval qq("${context}y" =~ /(?<=$context)y/);
9baa0206 294print "not " if $@ !~ m%^\QLookbehind longer than 255 not%;
8d37f932
DD
295print "ok 71\n";
296
b8c5462f 297# removed test
8d37f932
DD
298print "ok 72\n";
299
c277df42 300# Long Monsters
8d37f932 301$test = 73;
c277df42
IZ
302for $l (125, 140, 250, 270, 300000, 30) { # Ordered to free memory
303 $a = 'a' x $l;
304 print "# length=$l\nnot " unless "ba$a=" =~ /a$a=/;
305 print "ok $test\n";
306 $test++;
73d6d589 307
c277df42
IZ
308 print "not " if "b$a=" =~ /a$a=/;
309 print "ok $test\n";
310 $test++;
311}
312
313# 20000 nodes, each taking 3 words per string, and 1 per branch
314$long_constant_len = join '|', 12120 .. 32645;
315$long_var_len = join '|', 8120 .. 28645;
316%ans = ( 'ax13876y25677lbc' => 1,
317 'ax13876y25677mcb' => 0, # not b.
318 'ax13876y35677nbc' => 0, # Num too big
319 'ax13876y25677y21378obc' => 1,
320 'ax13876y25677y21378zbc' => 0, # Not followed by [k-o]
321 'ax13876y25677y21378y21378kbc' => 1,
322 'ax13876y25677y21378y21378kcb' => 0, # Not b.
323 'ax13876y25677y21378y21378y21378kbc' => 0, # 5 runs
324 );
325
326for ( keys %ans ) {
73d6d589 327 print "# const-len `$_' not => $ans{$_}\nnot "
c277df42
IZ
328 if $ans{$_} xor /a(?=([yx]($long_constant_len)){2,4}[k-o]).*b./o;
329 print "ok $test\n";
330 $test++;
73d6d589 331 print "# var-len `$_' not => $ans{$_}\nnot "
c277df42
IZ
332 if $ans{$_} xor /a(?=([yx]($long_var_len)){2,4}[k-o]).*b./o;
333 print "ok $test\n";
334 $test++;
335}
336
337$_ = " a (bla()) and x(y b((l)u((e))) and b(l(e)e)e";
338$expect = "(bla()) ((l)u((e))) (l(e)e)";
339
73d6d589 340sub matchit {
cc6b7395 341 m/
c277df42 342 (
73d6d589 343 \(
c277df42
IZ
344 (?{ $c = 1 }) # Initialize
345 (?:
346 (?(?{ $c == 0 }) # PREVIOUS iteration was OK, stop the loop
347 (?!
348 ) # Fail: will unwind one iteration back
73d6d589 349 )
c277df42
IZ
350 (?:
351 [^()]+ # Match a big chunk
352 (?=
353 [()]
354 ) # Do not try to match subchunks
355 |
73d6d589 356 \(
c277df42
IZ
357 (?{ ++$c })
358 |
73d6d589 359 \)
c277df42
IZ
360 (?{ --$c })
361 )
362 )+ # This may not match with different subblocks
363 )
364 (?(?{ $c != 0 })
365 (?!
366 ) # Fail
367 ) # Otherwise the chunk 1 may succeed with $c>0
cc6b7395 368 /xg;
c277df42
IZ
369}
370
0f5d15d6 371@ans = ();
c277df42
IZ
372push @ans, $res while $res = matchit;
373
374print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne "1 1 1";
375print "ok $test\n";
376$test++;
377
378@ans = matchit;
379
380print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne $expect;
381print "ok $test\n";
382$test++;
383
96776eda
GS
384print "not " unless "abc" =~ /^(??{"a"})b/;
385print "ok $test\n";
386$test++;
387
0f5d15d6 388my $matched;
14455d6c 389$matched = qr/\((?:(?>[^()]+)|(??{$matched}))*\)/;
0f5d15d6
IZ
390
391@ans = @ans1 = ();
392push(@ans, $res), push(@ans1, $&) while $res = m/$matched/g;
393
394print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne "1 1 1";
395print "ok $test\n";
396$test++;
397
398print "# ans1='@ans1'\n# expect='$expect'\nnot " if "@ans1" ne $expect;
399print "ok $test\n";
400$test++;
401
402@ans = m/$matched/g;
403
404print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne $expect;
405print "ok $test\n";
406$test++;
407
c277df42
IZ
408@ans = ('a/b' =~ m%(.*/)?(.*)%); # Stack may be bad
409print "not " if "@ans" ne 'a/ b';
410print "ok $test\n";
411$test++;
412
cc6b7395 413$code = '{$blah = 45}';
c277df42 414$blah = 12;
2cd61cdb
IZ
415eval { /(?$code)/ };
416print "not " unless $@ and $@ =~ /not allowed at runtime/ and $blah == 12;
e4d48cc9
GS
417print "ok $test\n";
418$test++;
419
2cd61cdb
IZ
420for $code ('{$blah = 45}','=xx') {
421 $blah = 12;
422 $res = eval { "xx" =~ /(?$code)/o };
423 if ($code eq '=xx') {
424 print "#'$@','$res','$blah'\nnot " unless not $@ and $res;
425 } else {
73d6d589 426 print "#'$@','$res','$blah'\nnot " unless $@ and $@ =~ /not allowed at runtime/ and $blah == 12;
2cd61cdb
IZ
427 }
428 print "ok $test\n";
429 $test++;
430}
431
e4d48cc9
GS
432$code = '{$blah = 45}';
433$blah = 12;
434eval "/(?$code)/";
cc6b7395
IZ
435print "not " if $blah != 45;
436print "ok $test\n";
437$test++;
438
439$blah = 12;
440/(?{$blah = 45})/;
c277df42
IZ
441print "not " if $blah != 45;
442print "ok $test\n";
443$test++;
444
74d6a13a
MB
445$x = 'banana';
446$x =~ /.a/g;
447print "not " unless pos($x) == 2;
448print "ok $test\n";
449$test++;
450
451$x =~ /.z/gc;
452print "not " unless pos($x) == 2;
453print "ok $test\n";
454$test++;
455
456sub f {
457 my $p = $_[0];
458 return $p;
459}
460
461$x =~ /.a/g;
462print "not " unless f(pos($x)) == 4;
463print "ok $test\n";
464$test++;
4599a1de 465
ce862d02
IZ
466$x = $^R = 67;
467'foot' =~ /foo(?{$x = 12; 75})[t]/;
468print "not " unless $^R eq '75';
469print "ok $test\n";
470$test++;
471
472$x = $^R = 67;
473'foot' =~ /foo(?{$x = 12; 75})[xy]/;
474print "not " unless $^R eq '67' and $x eq '12';
475print "ok $test\n";
476$test++;
477
478$x = $^R = 67;
479'foot' =~ /foo(?{ $^R + 12 })((?{ $x = 12; $^R + 17 })[xy])?/;
480print "not " unless $^R eq '79' and $x eq '12';
481print "ok $test\n";
482$test++;
483
cc74c5bd 484print "not " unless qr/\b\v$/i eq '(?i-xsm:\b\v$)';
8782bef2
GB
485print "ok $test\n";
486$test++;
487
cc74c5bd 488print "not " unless qr/\b\v$/s eq '(?s-xim:\b\v$)';
8782bef2
GB
489print "ok $test\n";
490$test++;
491
cc74c5bd 492print "not " unless qr/\b\v$/m eq '(?m-xis:\b\v$)';
8782bef2
GB
493print "ok $test\n";
494$test++;
495
cc74c5bd 496print "not " unless qr/\b\v$/x eq '(?x-ism:\b\v$)';
8782bef2
GB
497print "ok $test\n";
498$test++;
499
cc74c5bd 500print "not " unless qr/\b\v$/xism eq '(?msix:\b\v$)';
8782bef2
GB
501print "ok $test\n";
502$test++;
503
cc74c5bd 504print "not " unless qr/\b\v$/ eq '(?-xism:\b\v$)';
97197631
IZ
505print "ok $test\n";
506$test++;
507
7e5428c5
IZ
508$_ = 'xabcx';
509foreach $ans ('', 'c') {
510 /(?<=(?=a)..)((?=c)|.)/g;
02db2b7b 511 print "# \$1 ='$1'\n# \$ans='$ans'\nnot " unless $1 eq $ans;
7e5428c5
IZ
512 print "ok $test\n";
513 $test++;
514}
515
516$_ = 'a';
517foreach $ans ('', 'a', '') {
518 /^|a|$/g;
02db2b7b 519 print "# \$& ='$&'\n# \$ans='$ans'\nnot " unless $& eq $ans;
7e5428c5
IZ
520 print "ok $test\n";
521 $test++;
522}
523
09f25ae4 524sub prefixify {
73d6d589
NIS
525 my($v,$a,$b,$res) = @_;
526 $v =~ s/\Q$a\E/$b/;
527 print "not " unless $res eq $v;
09f25ae4
IZ
528 print "ok $test\n";
529 $test++;
530}
531prefixify('/a/b/lib/arch', "/a/b/lib", 'X/lib', 'X/lib/arch');
532prefixify('/a/b/man/arch', "/a/b/man", 'X/man', 'X/man/arch');
533
534$_ = 'var="foo"';
535/(\")/;
536print "not " unless $1 and /$1/;
537print "ok $test\n";
538$test++;
539
73d6d589 540$a=qr/(?{++$b})/;
2cd61cdb 541$b = 7;
73d6d589
NIS
542/$a$a/;
543print "not " unless $b eq '9';
2cd61cdb
IZ
544print "ok $test\n";
545$test++;
546
e3faa678 547{
e3faa678
NC
548 $c="$a";
549 /$a$a/;
550 iseq($b, '11');
551}
2cd61cdb
IZ
552
553{
73d6d589
NIS
554 use re "eval";
555 /$a$c$a/;
f9077f1b 556 iseq($b, '14');
2cd61cdb 557
160cb429
JH
558 local $lex_a = 2;
559 my $lex_a = 43;
560 my $lex_b = 17;
561 my $lex_c = 27;
562 my $lex_res = ($lex_b =~ qr/$lex_b(?{ $lex_c = $lex_a++ })/);
563 print "not " unless $lex_res eq '1';
564 print "ok $test\n";
565 $test++;
566 print "not " unless $lex_a eq '44';
567 print "ok $test\n";
568 $test++;
569 print "not " unless $lex_c eq '43';
570 print "ok $test\n";
571 $test++;
572
573
73d6d589 574 no re "eval";
2cd61cdb 575 $match = eval { /$a$c$a/ };
e3faa678
NC
576 # FIXME - split this one. That would require removing a lot of hard coded
577 # test numbers.
e3faa678 578 ok($b eq '14' and $@ =~ /Eval-group not allowed/ and not $match);
2cd61cdb 579}
cbce877f
IZ
580
581{
160cb429
JH
582 local $lex_a = 2;
583 my $lex_a = 43;
584 my $lex_b = 17;
585 my $lex_c = 27;
586 my $lex_res = ($lex_b =~ qr/17(?{ $lex_c = $lex_a++ })/);
587 print "not " unless $lex_res eq '1';
588 print "ok $test\n";
589 $test++;
590 print "not " unless $lex_a eq '44';
591 print "ok $test\n";
592 $test++;
593 print "not " unless $lex_c eq '43';
594 print "ok $test\n";
595 $test++;
596}
597
598{
cbce877f
IZ
599 package aa;
600 $c = 2;
601 $::c = 3;
602 '' =~ /(?{ $c = 4 })/;
603 print "not " unless $c == 4;
604}
605print "ok $test\n";
606$test++;
607print "not " unless $c == 3;
608print "ok $test\n";
73d6d589
NIS
609$test++;
610
4599a1de
JH
611sub must_warn_pat {
612 my $warn_pat = shift;
613 return sub { print "not " unless $_[0] =~ /$warn_pat/ }
614}
615
616sub must_warn {
617 my ($warn_pat, $code) = @_;
9f1b1f2d
GS
618 local %SIG;
619 eval 'BEGIN { use warnings; $SIG{__WARN__} = $warn_pat };' . $code;
4599a1de
JH
620 print "ok $test\n";
621 $test++;
622}
623
624
625sub make_must_warn {
626 my $warn_pat = shift;
627 return sub { must_warn(must_warn_pat($warn_pat)) }
628}
629
630my $for_future = make_must_warn('reserved for future extensions');
631
632&$for_future('q(a:[b]:) =~ /[x[:foo:]]/');
9baa0206
HS
633
634#&$for_future('q(a=[b]=) =~ /[x[=foo=]]/');
635print "ok $test\n"; $test++; # now a fatal croak
636
637#&$for_future('q(a.[b].) =~ /[x[.foo.]]/');
638print "ok $test\n"; $test++; # now a fatal croak
f7e33566
GS
639
640# test if failure of patterns returns empty list
641$_ = 'aaa';
642@_ = /bbb/;
643print "not " if @_;
644print "ok $test\n";
645$test++;
646
647@_ = /bbb/g;
648print "not " if @_;
649print "ok $test\n";
650$test++;
651
652@_ = /(bbb)/;
653print "not " if @_;
654print "ok $test\n";
655$test++;
656
657@_ = /(bbb)/g;
658print "not " if @_;
659print "ok $test\n";
660$test++;
661
6cef1e77
IZ
662/a(?=.$)/;
663print "not " if $#+ != 0 or $#- != 0;
664print "ok $test\n";
665$test++;
666
667print "not " if $+[0] != 2 or $-[0] != 1;
668print "ok $test\n";
669$test++;
670
73d6d589 671print "not "
6cef1e77
IZ
672 if defined $+[1] or defined $-[1] or defined $+[2] or defined $-[2];
673print "ok $test\n";
674$test++;
675
676/a(a)(a)/;
677print "not " if $#+ != 2 or $#- != 2;
678print "ok $test\n";
679$test++;
680
681print "not " if $+[0] != 3 or $-[0] != 0;
682print "ok $test\n";
683$test++;
684
685print "not " if $+[1] != 2 or $-[1] != 1;
686print "ok $test\n";
687$test++;
688
689print "not " if $+[2] != 3 or $-[2] != 2;
690print "ok $test\n";
691$test++;
692
73d6d589 693print "not "
6cef1e77
IZ
694 if defined $+[3] or defined $-[3] or defined $+[4] or defined $-[4];
695print "ok $test\n";
696$test++;
697
698/.(a)(b)?(a)/;
699print "not " if $#+ != 3 or $#- != 3;
700print "ok $test\n";
701$test++;
702
703print "not " if $+[0] != 3 or $-[0] != 0;
704print "ok $test\n";
705$test++;
706
707print "not " if $+[1] != 2 or $-[1] != 1;
708print "ok $test\n";
709$test++;
710
711print "not " if $+[3] != 3 or $-[3] != 2;
712print "ok $test\n";
713$test++;
714
73d6d589 715print "not "
6cef1e77
IZ
716 if defined $+[2] or defined $-[2] or defined $+[4] or defined $-[4];
717print "ok $test\n";
718$test++;
719
720/.(a)/;
721print "not " if $#+ != 1 or $#- != 1;
722print "ok $test\n";
723$test++;
724
725print "not " if $+[0] != 2 or $-[0] != 0;
726print "ok $test\n";
727$test++;
728
729print "not " if $+[1] != 2 or $-[1] != 1;
730print "ok $test\n";
731$test++;
732
73d6d589 733print "not "
6cef1e77
IZ
734 if defined $+[2] or defined $-[2] or defined $+[3] or defined $-[3];
735print "ok $test\n";
736$test++;
737
03a27ae7 738eval { $+[0] = 13; };
73d6d589 739print "not "
03a27ae7
MG
740 if $@ !~ /^Modification of a read-only value attempted/;
741print "ok $test\n";
742$test++;
743
744eval { $-[0] = 13; };
73d6d589 745print "not "
03a27ae7
MG
746 if $@ !~ /^Modification of a read-only value attempted/;
747print "ok $test\n";
748$test++;
749
750eval { @+ = (7, 6, 5); };
73d6d589 751print "not "
03a27ae7
MG
752 if $@ !~ /^Modification of a read-only value attempted/;
753print "ok $test\n";
754$test++;
755
756eval { @- = qw(foo bar); };
73d6d589 757print "not "
03a27ae7
MG
758 if $@ !~ /^Modification of a read-only value attempted/;
759print "ok $test\n";
760$test++;
761
8f580fb8
IZ
762/.(a)(ba*)?/;
763print "#$#-..$#+\nnot " if $#+ != 2 or $#- != 1;
764print "ok $test\n";
765$test++;
766
ad94a511
IZ
767$_ = 'aaa';
768pos = 1;
769@a = /\Ga/g;
770print "not " unless "@a" eq "a a";
771print "ok $test\n";
772$test++;
773
22e551b9
IZ
774$str = 'abcde';
775pos $str = 2;
776
777print "not " if $str =~ /^\G/;
778print "ok $test\n";
779$test++;
780
781print "not " if $str =~ /^.\G/;
782print "ok $test\n";
783$test++;
784
785print "not " unless $str =~ /^..\G/;
786print "ok $test\n";
787$test++;
788
789print "not " if $str =~ /^...\G/;
790print "ok $test\n";
791$test++;
792
e3faa678
NC
793{
794 local $TODO = $::running_as_thread;
795 ok($str =~ /.\G./ and $& eq 'bc');
796}
22e551b9
IZ
797
798print "not " unless $str =~ /\G../ and $& eq 'cd';
799print "ok $test\n";
800$test++;
801
9661b544
IZ
802undef $foo; undef $bar;
803print "#'$str','$foo','$bar'\nnot "
73d6d589 804 unless $str =~ /b(?{$foo = $_; $bar = pos})c/
9661b544
IZ
805 and $foo eq 'abcde' and $bar eq 2;
806print "ok $test\n";
807$test++;
808
809undef $foo; undef $bar;
810pos $str = undef;
811print "#'$str','$foo','$bar'\nnot "
73d6d589 812 unless $str =~ /b(?{$foo = $_; $bar = pos})c/g
9661b544
IZ
813 and $foo eq 'abcde' and $bar eq 2 and pos $str eq 3;
814print "ok $test\n";
815$test++;
816
817$_ = $str;
818
819undef $foo; undef $bar;
820print "#'$str','$foo','$bar'\nnot "
73d6d589 821 unless /b(?{$foo = $_; $bar = pos})c/
9661b544
IZ
822 and $foo eq 'abcde' and $bar eq 2;
823print "ok $test\n";
824$test++;
825
826undef $foo; undef $bar;
827print "#'$str','$foo','$bar'\nnot "
73d6d589 828 unless /b(?{$foo = $_; $bar = pos})c/g
9661b544
IZ
829 and $foo eq 'abcde' and $bar eq 2 and pos eq 3;
830print "ok $test\n";
831$test++;
832
833undef $foo; undef $bar;
834pos = undef;
8351 while /b(?{$foo = $_; $bar = pos})c/g;
836print "#'$str','$foo','$bar'\nnot "
837 unless $foo eq 'abcde' and $bar eq 2 and not defined pos;
838print "ok $test\n";
839$test++;
840
841undef $foo; undef $bar;
842$_ = 'abcde|abcde';
843print "#'$str','$foo','$bar','$_'\nnot "
73d6d589 844 unless s/b(?{$foo = $_; $bar = pos})c/x/g and $foo eq 'abcde|abcde'
9661b544
IZ
845 and $bar eq 8 and $_ eq 'axde|axde';
846print "ok $test\n";
847$test++;
848
5c5e4c24
IZ
849@res = ();
850# List context:
851$_ = 'abcde|abcde';
852@dummy = /([ace]).(?{push @res, $1,$2})([ce])(?{push @res, $1,$2})/g;
853@res = map {defined $_ ? "'$_'" : 'undef'} @res;
854$res = "@res";
855print "#'@res' '$_'\nnot "
856 unless "@res" eq "'a' undef 'a' 'c' 'e' undef 'a' undef 'a' 'c'";
857print "ok $test\n";
858$test++;
859
860@res = ();
861@dummy = /([ace]).(?{push @res, $`,$&,$'})([ce])(?{push @res, $`,$&,$'})/g;
862@res = map {defined $_ ? "'$_'" : 'undef'} @res;
863$res = "@res";
864print "#'@res' '$_'\nnot "
865 unless "@res" eq
866 "'' 'ab' 'cde|abcde' " .
867 "'' 'abc' 'de|abcde' " .
868 "'abcd' 'e|' 'abcde' " .
869 "'abcde|' 'ab' 'cde' " .
870 "'abcde|' 'abc' 'de'" ;
871print "ok $test\n";
872$test++;
873
b7a35066
IZ
874#Some more \G anchor checks
875$foo='aabbccddeeffgg';
876
877pos($foo)=1;
878
879$foo=~/.\G(..)/g;
e3faa678
NC
880{
881 local $TODO = $::running_as_thread;
882 iseq($1,'ab');
883}
b7a35066
IZ
884
885pos($foo) += 1;
886$foo=~/.\G(..)/g;
e3faa678
NC
887{
888 local $TODO = $::running_as_thread;
889 iseq($1, 'cc');
890}
b7a35066
IZ
891
892pos($foo) += 1;
893$foo=~/.\G(..)/g;
e3faa678
NC
894{
895 local $TODO = $::running_as_thread;
896 iseq($1, 'de');
897}
b7a35066 898
e3faa678
NC
899{
900 local $TODO = $::running_as_thread;
901 ok($foo =~ /\Gef/g);
902}
0ef3e39e 903
b7a35066
IZ
904undef pos $foo;
905
906$foo=~/\G(..)/g;
907print "not " unless($1 eq 'aa');
908print "ok $test\n";
909$test++;
910
911$foo=~/\G(..)/g;
912print "not " unless($1 eq 'bb');
913print "ok $test\n";
914$test++;
915
916pos($foo)=5;
917$foo=~/\G(..)/g;
918print "not " unless($1 eq 'cd');
919print "ok $test\n";
920$test++;
921
73d6d589 922$_='123x123';
e60df1fa
IZ
923@res = /(\d*|x)/g;
924print "not " unless('123||x|123|' eq join '|', @res);
925print "ok $test\n";
926$test++;
927
9d080a66
GS
928# see if matching against temporaries (created via pp_helem()) is safe
929{ foo => "ok $test\n".$^X }->{foo} =~ /^(.*)\n/g;
930print "$1\n";
931$test++;
932
cf93c79d
IZ
933# See if $i work inside (?{}) in the presense of saved substrings and
934# changing $_
935@a = qw(foo bar);
936@b = ();
937s/(\w)(?{push @b, $1})/,$1,/g for @a;
938
939print "# \@b='@b', expect 'f o o b a r'\nnot " unless("@b" eq "f o o b a r");
940print "ok $test\n";
941$test++;
942
943print "not " unless("@a" eq ",f,,o,,o, ,b,,a,,r,");
944print "ok $test\n";
945$test++;
946
2c914db6 947$brackets = qr{
14455d6c 948 { (?> [^{}]+ | (??{ $brackets }) )* }
2c914db6
IZ
949 }x;
950
951"{{}" =~ $brackets;
952print "ok $test\n"; # Did we survive?
953$test++;
954
955"something { long { and } hairy" =~ $brackets;
956print "ok $test\n"; # Did we survive?
957$test++;
958
14455d6c 959"something { long { and } hairy" =~ m/((??{ $brackets }))/;
2c914db6
IZ
960print "not " unless $1 eq "{ and }";
961print "ok $test\n";
962$test++;
963
30944b6d
IZ
964$_ = "a-a\nxbb";
965pos=1;
966m/^-.*bb/mg and print "not ";
967print "ok $test\n";
968$test++;
30382c73
IZ
969
970$text = "aaXbXcc";
971pos($text)=0;
972$text =~ /\GXb*X/g and print 'not ';
973print "ok $test\n";
974$test++;
3cf5c195
IZ
975
976$text = "xA\n" x 500;
977$text =~ /^\s*A/m and print 'not ';
978print "ok $test\n";
979$test++;
d506a20d
IZ
980
981$text = "abc dbf";
982@res = ($text =~ /.*?(b).*?\b/g);
983"@res" eq 'b b' or print 'not ';
984print "ok $test\n";
985$test++;
986
9442cb0e 987@a = map chr,0..255;
aeaf5620
GS
988
989@b = grep(/\S/,@a);
990@c = grep(/[^\s]/,@a);
991print "not " if "@b" ne "@c";
9442cb0e
GS
992print "ok $test\n";
993$test++;
994
aeaf5620
GS
995@b = grep(/\S/,@a);
996@c = grep(/[\S]/,@a);
997print "not " if "@b" ne "@c";
9442cb0e
GS
998print "ok $test\n";
999$test++;
1000
aeaf5620
GS
1001@b = grep(/\s/,@a);
1002@c = grep(/[^\S]/,@a);
1003print "not " if "@b" ne "@c";
9442cb0e
GS
1004print "ok $test\n";
1005$test++;
1006
aeaf5620
GS
1007@b = grep(/\s/,@a);
1008@c = grep(/[\s]/,@a);
1009print "not " if "@b" ne "@c";
9442cb0e
GS
1010print "ok $test\n";
1011$test++;
1012
aeaf5620
GS
1013@b = grep(/\D/,@a);
1014@c = grep(/[^\d]/,@a);
1015print "not " if "@b" ne "@c";
9442cb0e
GS
1016print "ok $test\n";
1017$test++;
1018
aeaf5620
GS
1019@b = grep(/\D/,@a);
1020@c = grep(/[\D]/,@a);
1021print "not " if "@b" ne "@c";
9442cb0e
GS
1022print "ok $test\n";
1023$test++;
1024
aeaf5620
GS
1025@b = grep(/\d/,@a);
1026@c = grep(/[^\D]/,@a);
1027print "not " if "@b" ne "@c";
9442cb0e
GS
1028print "ok $test\n";
1029$test++;
1030
aeaf5620
GS
1031@b = grep(/\d/,@a);
1032@c = grep(/[\d]/,@a);
1033print "not " if "@b" ne "@c";
9442cb0e
GS
1034print "ok $test\n";
1035$test++;
1036
aeaf5620
GS
1037@b = grep(/\W/,@a);
1038@c = grep(/[^\w]/,@a);
1039print "not " if "@b" ne "@c";
9442cb0e
GS
1040print "ok $test\n";
1041$test++;
1042
aeaf5620
GS
1043@b = grep(/\W/,@a);
1044@c = grep(/[\W]/,@a);
1045print "not " if "@b" ne "@c";
9442cb0e
GS
1046print "ok $test\n";
1047$test++;
1048
aeaf5620
GS
1049@b = grep(/\w/,@a);
1050@c = grep(/[^\W]/,@a);
1051print "not " if "@b" ne "@c";
9442cb0e
GS
1052print "ok $test\n";
1053$test++;
1054
aeaf5620
GS
1055@b = grep(/\w/,@a);
1056@c = grep(/[\w]/,@a);
de734bd5 1057iseq("@b","@c");
1aeab75a
GS
1058
1059# see if backtracking optimization works correctly
1060"\n\n" =~ /\n $ \n/x or print "not ";
1061print "ok $test\n";
1062$test++;
1063
1064"\n\n" =~ /\n* $ \n/x or print "not ";
1065print "ok $test\n";
1066$test++;
1067
1068"\n\n" =~ /\n+ $ \n/x or print "not ";
1069print "ok $test\n";
1070$test++;
05b4157f
GS
1071
1072[] =~ /^ARRAY/ or print "# [] \nnot ";
1073print "ok $test\n";
1074$test++;
1075
1076eval << 'EOE';
1077{
1078 package S;
1079 use overload '""' => sub { 'Object S' };
1080 sub new { bless [] }
1081}
1082$a = 'S'->new;
1083EOE
1084
1085$a and $a =~ /^Object\sS/ or print "# '$a' \nnot ";
1086print "ok $test\n";
1087$test++;
815d35b9
G
1088
1089# test result of match used as match (!)
1090'a1b' =~ ('xyz' =~ /y/) and $` eq 'a' or print "not ";
1091print "ok $test\n";
1092$test++;
1093
1094'a1b' =~ ('xyz' =~ /t/) and $` eq 'a' or print "not ";
1095print "ok $test\n";
1096$test++;
5e39e1e5
HS
1097
1098$w = 0;
1099{
1100 local $SIG{__WARN__} = sub { $w = 1 };
1101 local $^W = 1;
1102 $w = 1 if ("1\n" x 102) =~ /^\s*\n/m;
1103}
1104print $w ? "not " : "", "ok $test\n";
1105$test++;
aaa51d5e
JF
1106
1107my %space = ( spc => " ",
1108 tab => "\t",
1109 cr => "\r",
1110 lf => "\n",
1111 ff => "\f",
75369ccb
JH
1112# There's no \v but the vertical tabulator seems miraculously
1113# be 11 both in ASCII and EBCDIC.
aaa51d5e
JF
1114 vt => chr(11),
1115 false => "space" );
1116
1117my @space0 = sort grep { $space{$_} =~ /\s/ } keys %space;
1118my @space1 = sort grep { $space{$_} =~ /[[:space:]]/ } keys %space;
1119my @space2 = sort grep { $space{$_} =~ /[[:blank:]]/ } keys %space;
1120
1121print "not " unless "@space0" eq "cr ff lf spc tab";
3bec3564 1122print "ok $test # @space0\n";
aaa51d5e
JF
1123$test++;
1124
1125print "not " unless "@space1" eq "cr ff lf spc tab vt";
3bec3564 1126print "ok $test # @space1\n";
aaa51d5e
JF
1127$test++;
1128
1129print "not " unless "@space2" eq "spc tab";
3bec3564 1130print "ok $test # @space2\n";
aaa51d5e 1131$test++;
73d6d589 1132
a1933d95
HS
1133# bugid 20001021.005 - this caused a SEGV
1134print "not " unless undef =~ /^([^\/]*)(.*)$/;
1135print "ok $test\n";
1136$test++;
b91bb191
JH
1137
1138# bugid 20000731.001
1139
1140print "not " unless "A \x{263a} B z C" =~ /A . B (??{ "z" }) C/;
1141print "ok $test\n";
1142$test++;
1143
5ae032e5
JH
1144my $ordA = ord('A');
1145
3baa4c62
JH
1146$_ = "a\x{100}b";
1147if (/(.)(\C)(\C)(.)/) {
1148 print "ok 232\n";
1149 if ($1 eq "a") {
1150 print "ok 233\n";
1151 } else {
1152 print "not ok 233\n";
1153 }
5ae032e5
JH
1154 if ($ordA == 65) { # ASCII (or equivalent), should be UTF-8
1155 if ($2 eq "\xC4") {
1156 print "ok 234\n";
1157 } else {
1158 print "not ok 234\n";
1159 }
1160 if ($3 eq "\x80") {
1161 print "ok 235\n";
1162 } else {
1163 print "not ok 235\n";
1164 }
1165 } elsif ($ordA == 193) { # EBCDIC (or equivalent), should be UTF-EBCDIC
1166 if ($2 eq "\x8C") {
1167 print "ok 234\n";
1168 } else {
1169 print "not ok 234\n";
1170 }
1171 if ($3 eq "\x41") {
1172 print "ok 235\n";
1173 } else {
1174 print "not ok 235\n";
1175 }
3baa4c62 1176 } else {
5ae032e5
JH
1177 for (234..235) {
1178 print "not ok $_ # ord('A') == $ordA\n";
1179 }
3baa4c62
JH
1180 }
1181 if ($4 eq "b") {
1182 print "ok 236\n";
1183 } else {
1184 print "not ok 236\n";
1185 }
1186} else {
1187 for (232..236) {
1188 print "not ok $_\n";
1189 }
1190}
1191$_ = "\x{100}";
1192if (/(\C)/g) {
1193 print "ok 237\n";
73d6d589 1194 # currently \C are still tagged as UTF-8
5ae032e5
JH
1195 if ($ordA == 65) {
1196 if ($1 eq "\xC4") {
1197 print "ok 238\n";
1198 } else {
1199 print "not ok 238\n";
1200 }
1201 } elsif ($ordA == 193) {
1202 if ($1 eq "\x8C") {
1203 print "ok 238\n";
1204 } else {
1205 print "not ok 238\n";
1206 }
3baa4c62 1207 } else {
5ae032e5 1208 print "not ok 238 # ord('A') == $ordA\n";
3baa4c62
JH
1209 }
1210} else {
1211 for (237..238) {
1212 print "not ok $_\n";
1213 }
1214}
1215if (/(\C)/g) {
1216 print "ok 239\n";
73d6d589 1217 # currently \C are still tagged as UTF-8
5ae032e5
JH
1218 if ($ordA == 65) {
1219 if ($1 eq "\x80") {
1220 print "ok 240\n";
1221 } else {
1222 print "not ok 240\n";
1223 }
1224 } elsif ($ordA == 193) {
1225 if ($1 eq "\x41") {
1226 print "ok 240\n";
1227 } else {
1228 print "not ok 240\n";
1229 }
3baa4c62 1230 } else {
5ae032e5 1231 print "not ok 240 # ord('A') == $ordA\n";
3baa4c62
JH
1232 }
1233} else {
1234 for (239..240) {
1235 print "not ok $_\n";
1236 }
1237}
b485d051 1238
db615365
JP
1239{
1240 # japhy -- added 03/03/2001
1241 () = (my $str = "abc") =~ /(...)/;
1242 $str = "def";
1243 print "not " if $1 ne "abc";
fd291da9
JH
1244 print "ok 241\n";
1245}
1246
1247# The 242 and 243 go with the 244 and 245.
1248# The trick is that in EBCDIC the explicit numeric range should match
1249# (as also in non-EBCDIC) but the explicit alphabetic range should not match.
1250
1251if ("\x8e" =~ /[\x89-\x91]/) {
1252 print "ok 242\n";
1253} else {
1254 print "not ok 242\n";
1255}
1256
1257if ("\xce" =~ /[\xc9-\xd1]/) {
db615365 1258 print "ok 243\n";
fd291da9
JH
1259} else {
1260 print "not ok 243\n";
1261}
1262
1263# In most places these tests would succeed since \x8e does not
1264# in most character sets match 'i' or 'j' nor would \xce match
1265# 'I' or 'J', but strictly speaking these tests are here for
1266# the good of EBCDIC, so let's test these only there.
1267if (ord('i') == 0x89 && ord('J') == 0xd1) { # EBCDIC
1268 if ("\x8e" !~ /[i-j]/) {
1269 print "ok 244\n";
1270 } else {
1271 print "not ok 244\n";
1272 }
1273 if ("\xce" !~ /[I-J]/) {
1274 print "ok 245\n";
1275 } else {
1276 print "not ok 245\n";
1277 }
1278} else {
1279 for (244..245) {
60425c38 1280 print "ok $_ # Skip: only in EBCDIC\n";
fd291da9 1281 }
db615365 1282}
4765795a
JH
1283
1284print "not " unless "\x{ab}" =~ /\x{ab}/;
1285print "ok 246\n";
1286
1287print "not " unless "\x{abcd}" =~ /\x{abcd}/;
1288print "ok 247\n";
1289
66b1de87 1290{
f9969324
JH
1291 # bug id 20001008.001
1292
fc8cd66c 1293 $test = 248;
f9969324
JH
1294 my @x = ("stra\337e 138","stra\337e 138");
1295 for (@x) {
1296 s/(\d+)\s*([\w\-]+)/$1 . uc $2/e;
1297 my($latin) = /^(.+)(?:\s+\d)/;
1298 print $latin eq "stra\337e" ? "ok $test\n" : # 248,249
1299 "#latin[$latin]\nnot ok $test\n";
1300 $test++;
1301 $latin =~ s/stra\337e/straße/; # \303\237 after the 2nd a
1302 use utf8; # needed for the raw UTF-8
1303 $latin =~ s!(s)tr(?:aß|s+e)!$1tr.!; # \303\237 after the a
4765795a
JH
1304 }
1305}
1306
1307{
1308 print "not " unless "ba\xd4c" =~ /([a\xd4]+)/ && $1 eq "a\xd4";
1309 print "ok 250\n";
1310
1311 print "not " unless "ba\xd4c" =~ /([a\xd4]+)/ && $1 eq "a\x{d4}";
1312 print "ok 251\n";
1313
1314 print "not " unless "ba\x{d4}c" =~ /([a\xd4]+)/ && $1 eq "a\x{d4}";
1315 print "ok 252\n";
1316
1317 print "not " unless "ba\x{d4}c" =~ /([a\xd4]+)/ && $1 eq "a\xd4";
1318 print "ok 253\n";
1319
1320 print "not " unless "ba\xd4c" =~ /([a\x{d4}]+)/ && $1 eq "a\xd4";
1321 print "ok 254\n";
1322
1323 print "not " unless "ba\xd4c" =~ /([a\x{d4}]+)/ && $1 eq "a\x{d4}";
1324 print "ok 255\n";
1325
1326 print "not " unless "ba\x{d4}c" =~ /([a\x{d4}]+)/ && $1 eq "a\x{d4}";
1327 print "ok 256\n";
1328
1329 print "not " unless "ba\x{d4}c" =~ /([a\x{d4}]+)/ && $1 eq "a\xd4";
1330 print "ok 257\n";
1331}
1332
1333{
1334 # the first half of 20001028.003
1335
1336 my $X = chr(1448);
1337 my ($Y) = $X =~ /(.*)/;
1338 print "not " unless $Y eq v1448 && length($Y) == 1;
1339 print "ok 258\n";
1340}
1341
1342{
1343 # 20001108.001
1344
1345 my $X = "Szab\x{f3},Bal\x{e1}zs";
1346 my $Y = $X;
1347 $Y =~ s/(B)/$1/ for 0..3;
1348 print "not " unless $Y eq $X && $X eq "Szab\x{f3},Bal\x{e1}zs";
1349 print "ok 259\n";
1350}
1351
1352{
1353 # the second half of 20001028.003
1354
3568d838 1355 my $X = '';
4765795a
JH
1356 $X =~ s/^/chr(1488)/e;
1357 print "not " unless length $X == 1 && ord($X) == 1488;
1358 print "ok 260\n";
1359}
1360
1361{
1362 # 20000517.001
1363
1364 my $x = "\x{100}A";
1365
1366 $x =~ s/A/B/;
1367
1368 print "not " unless $x eq "\x{100}B" && length($x) == 2;
1369 print "ok 261\n";
1370}
1371
1372{
1373 # bug id 20001230.002
1374
1375 print "not " unless "École" =~ /^\C\C(.)/ && $1 eq 'c';
1376 print "ok 262\n";
1377
1378 print "not " unless "École" =~ /^\C\C(c)/;
1379 print "ok 263\n";
1380}
1381
8dcad4ad 1382SKIP: {
fc8cd66c 1383 $test = 264; # till 575
4765795a 1384
bb99e5f1 1385 use charnames ":full";
4765795a
JH
1386
1387 # This is far from complete testing, there are dozens of character
1388 # classes in Unicode. The mixing of literals and \N{...} is
1389 # intentional so that in non-Latin-1 places we test the native
1390 # characters, not the Unicode code points.
1391
1392 my %s = (
1393 "a" => 'Ll',
1394 "\N{CYRILLIC SMALL LETTER A}" => 'Ll',
1395 "A" => 'Lu',
1396 "\N{GREEK CAPITAL LETTER ALPHA}" => 'Lu',
1397 "\N{HIRAGANA LETTER SMALL A}" => 'Lo',
1398 "\N{COMBINING GRAVE ACCENT}" => 'Mn',
1399 "0" => 'Nd',
1400 "\N{ARABIC-INDIC DIGIT ZERO}" => 'Nd',
1401 "_" => 'N',
1402 "!" => 'P',
1403 " " => 'Zs',
1404 "\0" => 'Cc',
1405 );
73d6d589 1406
3568d838
JH
1407 for my $char (map { s/^\S+ //; $_ }
1408 sort map { sprintf("%06x", ord($_))." $_" } keys %s) {
4765795a 1409 my $class = $s{$char};
3568d838
JH
1410 my $code = sprintf("%06x", ord($char));
1411 printf "#\n# 0x$code\n#\n";
4765795a
JH
1412 print "# IsAlpha\n";
1413 if ($class =~ /^[LM]/) {
1414 print "not " unless $char =~ /\p{IsAlpha}/;
1415 print "ok $test\n"; $test++;
1416 print "not " if $char =~ /\P{IsAlpha}/;
1417 print "ok $test\n"; $test++;
1418 } else {
1419 print "not " if $char =~ /\p{IsAlpha}/;
1420 print "ok $test\n"; $test++;
1421 print "not " unless $char =~ /\P{IsAlpha}/;
1422 print "ok $test\n"; $test++;
1423 }
1424 print "# IsAlnum\n";
1425 if ($class =~ /^[LMN]/ && $char ne "_") {
1426 print "not " unless $char =~ /\p{IsAlnum}/;
1427 print "ok $test\n"; $test++;
1428 print "not " if $char =~ /\P{IsAlnum}/;
1429 print "ok $test\n"; $test++;
1430 } else {
1431 print "not " if $char =~ /\p{IsAlnum}/;
1432 print "ok $test\n"; $test++;
1433 print "not " unless $char =~ /\P{IsAlnum}/;
1434 print "ok $test\n"; $test++;
1435 }
1436 print "# IsASCII\n";
24817587
JH
1437 if (ord("A") == 193) {
1438 print "ok $test # Skip: in EBCDIC\n"; $test++;
1439 print "ok $test # Skip: in EBCDIC\n"; $test++;
4765795a 1440 } else {
24817587
JH
1441 if ($code le '00007f') {
1442 print "not " unless $char =~ /\p{IsASCII}/;
1443 print "ok $test\n"; $test++;
1444 print "not " if $char =~ /\P{IsASCII}/;
1445 print "ok $test\n"; $test++;
1446 } else {
1447 print "not " if $char =~ /\p{IsASCII}/;
1448 print "ok $test\n"; $test++;
1449 print "not " unless $char =~ /\P{IsASCII}/;
1450 print "ok $test\n"; $test++;
1451 }
4765795a
JH
1452 }
1453 print "# IsCntrl\n";
1454 if ($class =~ /^C/) {
1455 print "not " unless $char =~ /\p{IsCntrl}/;
1456 print "ok $test\n"; $test++;
1457 print "not " if $char =~ /\P{IsCntrl}/;
1458 print "ok $test\n"; $test++;
1459 } else {
1460 print "not " if $char =~ /\p{IsCntrl}/;
1461 print "ok $test\n"; $test++;
1462 print "not " unless $char =~ /\P{IsCntrl}/;
1463 print "ok $test\n"; $test++;
1464 }
1465 print "# IsBlank\n";
1466 if ($class =~ /^Z[lp]/ || $char eq " ") {
1467 print "not " unless $char =~ /\p{IsBlank}/;
1468 print "ok $test\n"; $test++;
1469 print "not " if $char =~ /\P{IsBlank}/;
1470 print "ok $test\n"; $test++;
1471 } else {
1472 print "not " if $char =~ /\p{IsBlank}/;
1473 print "ok $test\n"; $test++;
1474 print "not " unless $char =~ /\P{IsBlank}/;
1475 print "ok $test\n"; $test++;
1476 }
1477 print "# IsDigit\n";
1478 if ($class =~ /^Nd$/) {
1479 print "not " unless $char =~ /\p{IsDigit}/;
1480 print "ok $test\n"; $test++;
1481 print "not " if $char =~ /\P{IsDigit}/;
1482 print "ok $test\n"; $test++;
1483 } else {
1484 print "not " if $char =~ /\p{IsDigit}/;
1485 print "ok $test\n"; $test++;
1486 print "not " unless $char =~ /\P{IsDigit}/;
1487 print "ok $test\n"; $test++;
1488 }
1489 print "# IsGraph\n";
1490 if ($class =~ /^([LMNPS])|Co/) {
1491 print "not " unless $char =~ /\p{IsGraph}/;
1492 print "ok $test\n"; $test++;
1493 print "not " if $char =~ /\P{IsGraph}/;
1494 print "ok $test\n"; $test++;
1495 } else {
1496 print "not " if $char =~ /\p{IsGraph}/;
1497 print "ok $test\n"; $test++;
1498 print "not " unless $char =~ /\P{IsGraph}/;
1499 print "ok $test\n"; $test++;
1500 }
1501 print "# IsLower\n";
1502 if ($class =~ /^Ll$/) {
1503 print "not " unless $char =~ /\p{IsLower}/;
1504 print "ok $test\n"; $test++;
1505 print "not " if $char =~ /\P{IsLower}/;
1506 print "ok $test\n"; $test++;
1507 } else {
1508 print "not " if $char =~ /\p{IsLower}/;
1509 print "ok $test\n"; $test++;
1510 print "not " unless $char =~ /\P{IsLower}/;
1511 print "ok $test\n"; $test++;
1512 }
1513 print "# IsPrint\n";
1514 if ($class =~ /^([LMNPS])|Co|Zs/) {
1515 print "not " unless $char =~ /\p{IsPrint}/;
1516 print "ok $test\n"; $test++;
1517 print "not " if $char =~ /\P{IsPrint}/;
1518 print "ok $test\n"; $test++;
1519 } else {
1520 print "not " if $char =~ /\p{IsPrint}/;
1521 print "ok $test\n"; $test++;
1522 print "not " unless $char =~ /\P{IsPrint}/;
1523 print "ok $test\n"; $test++;
1524 }
1525 print "# IsPunct\n";
1526 if ($class =~ /^P/ || $char eq "_") {
1527 print "not " unless $char =~ /\p{IsPunct}/;
1528 print "ok $test\n"; $test++;
1529 print "not " if $char =~ /\P{IsPunct}/;
1530 print "ok $test\n"; $test++;
1531 } else {
1532 print "not " if $char =~ /\p{IsPunct}/;
1533 print "ok $test\n"; $test++;
1534 print "not " unless $char =~ /\P{IsPunct}/;
1535 print "ok $test\n"; $test++;
1536 }
1537 print "# IsSpace\n";
1538 if ($class =~ /^Z/ || ($code =~ /^(0009|000A|000B|000C|000D)$/)) {
1539 print "not " unless $char =~ /\p{IsSpace}/;
1540 print "ok $test\n"; $test++;
1541 print "not " if $char =~ /\P{IsSpace}/;
1542 print "ok $test\n"; $test++;
1543 } else {
1544 print "not " if $char =~ /\p{IsSpace}/;
1545 print "ok $test\n"; $test++;
1546 print "not " unless $char =~ /\P{IsSpace}/;
1547 print "ok $test\n"; $test++;
1548 }
1549 print "# IsUpper\n";
1550 if ($class =~ /^L[ut]/) {
1551 print "not " unless $char =~ /\p{IsUpper}/;
1552 print "ok $test\n"; $test++;
1553 print "not " if $char =~ /\P{IsUpper}/;
1554 print "ok $test\n"; $test++;
1555 } else {
1556 print "not " if $char =~ /\p{IsUpper}/;
1557 print "ok $test\n"; $test++;
1558 print "not " unless $char =~ /\P{IsUpper}/;
1559 print "ok $test\n"; $test++;
1560 }
1561 print "# IsWord\n";
1562 if ($class =~ /^[LMN]/ || $char eq "_") {
1563 print "not " unless $char =~ /\p{IsWord}/;
1564 print "ok $test\n"; $test++;
1565 print "not " if $char =~ /\P{IsWord}/;
1566 print "ok $test\n"; $test++;
1567 } else {
1568 print "not " if $char =~ /\p{IsWord}/;
1569 print "ok $test\n"; $test++;
1570 print "not " unless $char =~ /\P{IsWord}/;
1571 print "ok $test\n"; $test++;
1572 }
1573 }
1574}
1575
1576{
1577 $_ = "abc\x{100}\x{200}\x{300}\x{380}\x{400}defg";
1578
1579 if (/(.\x{300})./) {
1580 print "ok 576\n";
1581
1582 print "not " unless $` eq "abc\x{100}" && length($`) == 4;
73d6d589 1583 print "ok 577\n";
4765795a
JH
1584
1585 print "not " unless $& eq "\x{200}\x{300}\x{380}" && length($&) == 3;
73d6d589 1586 print "ok 578\n";
4765795a
JH
1587
1588 print "not " unless $' eq "\x{400}defg" && length($') == 5;
73d6d589 1589 print "ok 579\n";
4765795a
JH
1590
1591 print "not " unless $1 eq "\x{200}\x{300}" && length($1) == 2;
73d6d589 1592 print "ok 580\n";
a8a2fe91
JH
1593 } else {
1594 for (576..580) { print "not ok $_\n" }
4765795a
JH
1595 }
1596}
8269fa76
JH
1597
1598{
1599 # bug id 20010306.008
1600
1601 $a = "a\x{1234}";
1602 # The original bug report had 'no utf8' here but that was irrelevant.
1603 $a =~ m/\w/; # used to core dump
1604
1605 print "ok 581\n";
1606}
b8ef571c
JH
1607
1608{
339e86bc
JH
1609 $test = 582;
1610
b8ef571c
JH
1611 # bugid 20010410.006
1612 for my $rx (
1613 '/(.*?)\{(.*?)\}/csg',
1614 '/(.*?)\{(.*?)\}/cg',
1615 '/(.*?)\{(.*?)\}/sg',
1616 '/(.*?)\{(.*?)\}/g',
1617 '/(.+?)\{(.+?)\}/csg',
1618 )
1619 {
1620 my($input, $i);
1621
1622 $i = 0;
1623 $input = "a{b}c{d}";
1624 eval <<EOT;
1625 while (eval \$input =~ $rx) {
1626 print "# \\\$1 = '\$1' \\\$2 = '\$2'\n";
1627 ++\$i;
1628 }
1629EOT
1630 print "not " unless $i == 2;
1631 print "ok " . $test++ . "\n";
1632 }
1633}
209a9bc1
JH
1634
1635{
1636 # from Robin Houston
1637
b851fbc1 1638 my $x = "\x{10FFFD}";
209a9bc1 1639 $x =~ s/(.)/$1/g;
b851fbc1 1640 print "not " unless ord($x) == 0x10FFFD && length($x) == 1;
209a9bc1
JH
1641 print "ok 587\n";
1642}
3568d838
JH
1643
1644{
1645 my $x = "\x7f";
1646
1647 print "not " if $x =~ /[\x80-\xff]/;
1648 print "ok 588\n";
1649
1650 print "not " if $x =~ /[\x80-\x{100}]/;
1651 print "ok 589\n";
1652
1653 print "not " if $x =~ /[\x{100}]/;
1654 print "ok 590\n";
1655
1656 print "not " if $x =~ /\p{InLatin1Supplement}/;
1657 print "ok 591\n";
1658
1659 print "not " unless $x =~ /\P{InLatin1Supplement}/;
1660 print "ok 592\n";
1661
1662 print "not " if $x =~ /\p{InLatinExtendedA}/;
1663 print "ok 593\n";
1664
1665 print "not " unless $x =~ /\P{InLatinExtendedA}/;
1666 print "ok 594\n";
1667}
1668
1669{
1670 my $x = "\x80";
1671
1672 print "not " unless $x =~ /[\x80-\xff]/;
1673 print "ok 595\n";
1674
1675 print "not " unless $x =~ /[\x80-\x{100}]/;
1676 print "ok 596\n";
1677
1678 print "not " if $x =~ /[\x{100}]/;
1679 print "ok 597\n";
1680
1681 print "not " unless $x =~ /\p{InLatin1Supplement}/;
1682 print "ok 598\n";
1683
1684 print "not " if $x =~ /\P{InLatin1Supplement}/;
1685 print "ok 599\n";
1686
1687 print "not " if $x =~ /\p{InLatinExtendedA}/;
1688 print "ok 600\n";
1689
1690 print "not " unless $x =~ /\P{InLatinExtendedA}/;
1691 print "ok 601\n";
1692}
1693
1694{
1695 my $x = "\xff";
1696
1697 print "not " unless $x =~ /[\x80-\xff]/;
1698 print "ok 602\n";
1699
1700 print "not " unless $x =~ /[\x80-\x{100}]/;
1701 print "ok 603\n";
1702
1703 print "not " if $x =~ /[\x{100}]/;
1704 print "ok 604\n";
1705
cc8040a1
DT
1706 # the next two tests must be ignored on EBCDIC
1707 print "not " unless $x =~ /\p{InLatin1Supplement}/ or ord("A") == 193;
3568d838
JH
1708 print "ok 605\n";
1709
cc8040a1 1710 print "not " if $x =~ /\P{InLatin1Supplement}/ and ord("A") != 193;
3568d838
JH
1711 print "ok 606\n";
1712
1713 print "not " if $x =~ /\p{InLatinExtendedA}/;
1714 print "ok 607\n";
1715
1716 print "not " unless $x =~ /\P{InLatinExtendedA}/;
1717 print "ok 608\n";
1718}
1719
1720{
1721 my $x = "\x{100}";
1722
1723 print "not " if $x =~ /[\x80-\xff]/;
1724 print "ok 609\n";
1725
1726 print "not " unless $x =~ /[\x80-\x{100}]/;
1727 print "ok 610\n";
1728
1729 print "not " unless $x =~ /[\x{100}]/;
1730 print "ok 611\n";
1731
1732 print "not " if $x =~ /\p{InLatin1Supplement}/;
1733 print "ok 612\n";
1734
1735 print "not " unless $x =~ /\P{InLatin1Supplement}/;
1736 print "ok 613\n";
1737
1738 print "not " unless $x =~ /\p{InLatinExtendedA}/;
1739 print "ok 614\n";
1740
1741 print "not " if $x =~ /\P{InLatinExtendedA}/;
1742 print "ok 615\n";
1743}
1744
9d1d55b5
JP
1745{
1746 # from japhy
1747 my $w;
1748 use warnings;
1749 local $SIG{__WARN__} = sub { $w .= shift };
1750
1751 $w = "";
1752 eval 'qr/(?c)/';
1753 print "not " if $w !~ /^Useless \(\?c\)/;
1754 print "ok 616\n";
1755
1756 $w = "";
1757 eval 'qr/(?-c)/';
1758 print "not " if $w !~ /^Useless \(\?-c\)/;
1759 print "ok 617\n";
1760
1761 $w = "";
1762 eval 'qr/(?g)/';
1763 print "not " if $w !~ /^Useless \(\?g\)/;
1764 print "ok 618\n";
1765
1766 $w = "";
1767 eval 'qr/(?-g)/';
1768 print "not " if $w !~ /^Useless \(\?-g\)/;
1769 print "ok 619\n";
1770
1771 $w = "";
1772 eval 'qr/(?o)/';
1773 print "not " if $w !~ /^Useless \(\?o\)/;
1774 print "ok 620\n";
1775
1776 $w = "";
1777 eval 'qr/(?-o)/';
1778 print "not " if $w !~ /^Useless \(\?-o\)/;
1779 print "ok 621\n";
1780
1781 # now test multi-error regexes
1782
1783 $w = "";
1784 eval 'qr/(?g-o)/';
1785 print "not " if $w !~ /^Useless \(\?g\).*\nUseless \(\?-o\)/;
1786 print "ok 622\n";
1787
1788 $w = "";
1789 eval 'qr/(?g-c)/';
1790 print "not " if $w !~ /^Useless \(\?g\).*\nUseless \(\?-c\)/;
1791 print "ok 623\n";
1792
1793 $w = "";
1794 eval 'qr/(?o-cg)/'; # (?c) means (?g) error won't be thrown
1795 print "not " if $w !~ /^Useless \(\?o\).*\nUseless \(\?-c\)/;
1796 print "ok 624\n";
1797
1798 $w = "";
1799 eval 'qr/(?ogc)/';
1800 print "not " if $w !~ /^Useless \(\?o\).*\nUseless \(\?g\).*\nUseless \(\?c\)/;
1801 print "ok 625\n";
1802}
a72deede
JH
1803
1804# More Unicode "class" tests
1805
1806{
1807 use charnames ':full';
1808
1809 print "not " unless "\N{LATIN CAPITAL LETTER A}" =~ /\p{InBasicLatin}/;
1810 print "ok 626\n";
1811
1812 print "not " unless "\N{LATIN CAPITAL LETTER A WITH GRAVE}" =~ /\p{InLatin1Supplement}/;
1813 print "ok 627\n";
1814
1815 print "not " unless "\N{LATIN CAPITAL LETTER A WITH MACRON}" =~ /\p{InLatinExtendedA}/;
1816 print "ok 628\n";
1817
1818 print "not " unless "\N{LATIN SMALL LETTER B WITH STROKE}" =~ /\p{InLatinExtendedB}/;
1819 print "ok 629\n";
1820
1821 print "not " unless "\N{KATAKANA LETTER SMALL A}" =~ /\p{InKatakana}/;
1822 print "ok 630\n";
1823}
1824
6002328a
JH
1825$_ = "foo";
1826
1827eval <<"EOT"; die if $@;
1828 /f
1829 o\r
1830 o
1831 \$
1832 /x && print "ok 631\n";
1833EOT
1834
1835eval <<"EOT"; die if $@;
1836 /f
1837 o
1838 o
1839 \$\r
1840 /x && print "ok 632\n";
1841EOT
1842
569b5e07
AB
1843#test /o feature
1844sub test_o { $_[0] =~/$_[1]/o; return $1}
1845if(test_o('abc','(.)..') eq 'a') {
395ddfe6 1846 print "ok 633\n";
569b5e07 1847} else {
395ddfe6 1848 print "not ok 633\n";
569b5e07
AB
1849}
1850if(test_o('abc','..(.)') eq 'a') {
395ddfe6 1851 print "ok 634\n";
569b5e07 1852} else {
395ddfe6 1853 print "not ok 634\n";
569b5e07
AB
1854}
1855
f79b3095
JH
1856# 635..639: ID 20010619.003 (only the space character is
1857# supposed to be [:print:], not the whole isprint()).
1858
1859print "not " if "\n" =~ /[[:print:]]/;
1860print "ok 635\n";
1861
1862print "not " if "\t" =~ /[[:print:]]/;
1863print "ok 636\n";
1864
e857312d 1865# Amazingly vertical tabulator is the same in ASCII and EBCDIC.
f79b3095
JH
1866print "not " if "\014" =~ /[[:print:]]/;
1867print "ok 637\n";
1868
1869print "not " if "\r" =~ /[[:print:]]/;
1870print "ok 638\n";
1871
1872print "not " unless " " =~ /[[:print:]]/;
1873print "ok 639\n";
1874
a01268b5
JH
1875##
1876## Test basic $^N usage outside of a regex
1877##
1878$x = "abcdef";
1879$T="ok 640\n";if ($x =~ /cde/ and not defined $^N) {print $T} else {print "not $T"};
1880$T="ok 641\n";if ($x =~ /(cde)/ and $^N eq "cde") {print $T} else {print "not $T"};
1881$T="ok 642\n";if ($x =~ /(c)(d)(e)/ and $^N eq "e") {print $T} else {print "not $T"};
1882$T="ok 643\n";if ($x =~ /(c(d)e)/ and $^N eq "cde") {print $T} else {print "not $T"};
1883$T="ok 644\n";if ($x =~ /(foo)|(c(d)e)/ and $^N eq "cde") {print $T} else {print "not $T"};
1884$T="ok 645\n";if ($x =~ /(c(d)e)|(foo)/ and $^N eq "cde") {print $T} else {print "not $T"};
1885$T="ok 646\n";if ($x =~ /(c(d)e)|(abc)/ and $^N eq "abc") {print $T} else {print "not $T"};
1886$T="ok 647\n";if ($x =~ /(c(d)e)|(abc)x/ and $^N eq "cde") {print $T} else {print "not $T"};
1887$T="ok 648\n";if ($x =~ /(c(d)e)(abc)?/ and $^N eq "cde") {print $T} else {print "not $T"};
1888$T="ok 649\n";if ($x =~ /(?:c(d)e)/ and $^N eq "d" ) {print $T} else {print "not $T"};
1889$T="ok 650\n";if ($x =~ /(?:c(d)e)(?:f)/ and $^N eq "d" ) {print $T} else {print "not $T"};
1890$T="ok 651\n";if ($x =~ /(?:([abc])|([def]))*/ and $^N eq "f" ){print $T} else {print "not $T"};
1891$T="ok 652\n";if ($x =~ /(?:([ace])|([bdf]))*/ and $^N eq "f" ){print $T} else {print "not $T"};
1892$T="ok 653\n";if ($x =~ /(([ace])|([bd]))*/ and $^N eq "e" ){print $T} else {print "not $T"};
1893{
1894 $T="ok 654\n";if($x =~ /(([ace])|([bdf]))*/ and $^N eq "f" ){print $T} else {print "not $T"};
1895}
1896## test to see if $^N is automatically localized -- it should now
1897## have the value set in test 653
1898$T="ok 655\n";if ($^N eq "e" ){print $T} else {print "not $T"};
1899
1900##
1901## Now test inside (?{...})
1902##
1903$T="ok 656\n";if ($x =~ /a([abc])(?{$y=$^N})c/ and $y eq "b" ){print $T} else {print "not $T"};
1904$T="ok 657\n";if ($x =~ /a([abc]+)(?{$y=$^N})d/ and $y eq "bc"){print $T} else {print "not $T"};
1905$T="ok 658\n";if ($x =~ /a([abcdefg]+)(?{$y=$^N})d/ and $y eq "bc"){print $T} else {print "not $T"};
1906$T="ok 659\n";if ($x =~ /(a([abcdefg]+)(?{$y=$^N})d)(?{$z=$^N})e/ and $y eq "bc" and $z eq "abcd")
1907 {print $T} else {print "not $T"};
1908$T="ok 660\n";if ($x =~ /(a([abcdefg]+)(?{$y=$^N})de)(?{$z=$^N})/ and $y eq "bc" and $z eq "abcde")
1909 {print $T} else {print "not $T"};
2796c109
JH
1910
1911# Test the Unicode script classes
1912
cf25bb62 1913print "not " unless chr(0x100) =~ /\p{IsLatin}/; # outside Latin-1
2796c109
JH
1914print "ok 661\n";
1915
cf25bb62 1916print "not " unless chr(0x212b) =~ /\p{IsLatin}/; # Angstrom sign, very outside
2796c109
JH
1917print "ok 662\n";
1918
cf25bb62 1919print "not " unless chr(0x5d0) =~ /\p{IsHebrew}/; # inside InHebrew
2796c109
JH
1920print "ok 663\n";
1921
cf25bb62 1922print "not " unless chr(0xfb4f) =~ /\p{IsHebrew}/; # outside InHebrew
2796c109
JH
1923print "ok 664\n";
1924
7be0dac3
NC
1925# # singleton (not in a range, this test must be ignored on EBCDIC)
1926# print "not " unless chr(0xb5) =~ /\p{IsGreek}/ or ord("A") == 193;
1927# print "ok 665\n";
1928print "ok 665 # 0xb5 moved from Greek to Common with Unicode 4.0.1\n";
5f9563ea 1929
cf25bb62 1930print "not " unless chr(0x37a) =~ /\p{IsGreek}/; # singleton
5f9563ea
JH
1931print "ok 666\n";
1932
cf25bb62 1933print "not " unless chr(0x386) =~ /\p{IsGreek}/; # singleton
5f9563ea
JH
1934print "ok 667\n";
1935
cf25bb62 1936print "not " unless chr(0x387) =~ /\P{IsGreek}/; # not there
5f9563ea
JH
1937print "ok 668\n";
1938
cf25bb62 1939print "not " unless chr(0x388) =~ /\p{IsGreek}/; # range
5f9563ea
JH
1940print "ok 669\n";
1941
cf25bb62 1942print "not " unless chr(0x38a) =~ /\p{IsGreek}/; # range
5f9563ea
JH
1943print "ok 670\n";
1944
cf25bb62 1945print "not " unless chr(0x38b) =~ /\P{IsGreek}/; # not there
5f9563ea
JH
1946print "ok 671\n";
1947
cf25bb62 1948print "not " unless chr(0x38c) =~ /\p{IsGreek}/; # singleton
5f9563ea
JH
1949print "ok 672\n";
1950
550cec39 1951if (ord("A") == 65) {
7be5a6cf
JF
1952##
1953## Test [:cntrl:]...
1954##
1955## Should probably put in tests for all the POSIX stuff, but not sure how to
1956## guarantee a specific locale......
1957##
550cec39
JH
1958 $AllBytes = join('', map { chr($_) } 0..255);
1959 ($x = $AllBytes) =~ s/[[:cntrl:]]//g;
1960 if ($x ne join('', map { chr($_) } 0x20..0x7E, 0x80..0xFF)) {
1961 print "not ";
1962 }
1963 print "ok 673\n";
1964
1965 ($x = $AllBytes) =~ s/[^[:cntrl:]]//g;
1966 if ($x ne join('', map { chr($_) } 0..0x1F, 0x7F)) { print "not " }
1967 print "ok 674\n";
1968} else {
1969 print "ok $_ # Skip: EBCDIC\n" for 673..674;
1970}
f33976b4
DB
1971
1972# With /s modifier UTF8 chars were interpreted as bytes
1973{
1974 my $a = "Hello \x{263A} World";
1975
1976 my @a = ($a =~ /./gs);
1977
1978 print "not " unless $#a == 12;
1979 print "ok 675\n";
1980}
cce850e4
JH
1981
1982@a = ("foo\nbar" =~ /./g);
1983print "ok 676\n" if @a == 6 && "@a" eq "f o o b a r";
1984
1985@a = ("foo\nbar" =~ /./gs);
1986print "ok 677\n" if @a == 7 && "@a" eq "f o o \n b a r";
1987
1988@a = ("foo\nbar" =~ /\C/g);
1989print "ok 678\n" if @a == 7 && "@a" eq "f o o \n b a r";
1990
1991@a = ("foo\nbar" =~ /\C/gs);
1992print "ok 679\n" if @a == 7 && "@a" eq "f o o \n b a r";
1993
1994@a = ("foo\n\x{100}bar" =~ /./g);
1995print "ok 680\n" if @a == 7 && "@a" eq "f o o \x{100} b a r";
1996
1997@a = ("foo\n\x{100}bar" =~ /./gs);
1998print "ok 681\n" if @a == 8 && "@a" eq "f o o \n \x{100} b a r";
1999
2000($a, $b) = map { chr } ord('A') == 65 ? (0xc4, 0x80) : (0x8c, 0x41);
2001
2002@a = ("foo\n\x{100}bar" =~ /\C/g);
2003print "ok 682\n" if @a == 9 && "@a" eq "f o o \n $a $b b a r";
2004
2005@a = ("foo\n\x{100}bar" =~ /\C/gs);
2006print "ok 683\n" if @a == 9 && "@a" eq "f o o \n $a $b b a r";
2007
0af80b60
HS
2008{
2009 # [ID 20010814.004] pos() doesn't work when using =~m// in list context
2010 $_ = "ababacadaea";
2011 $a = join ":", /b./gc;
2012 $b = join ":", /a./gc;
2013 $c = pos;
2014 print "$a $b $c" eq 'ba:ba ad:ae 10' ? "ok 684\n" : "not ok 684\t# $a $b $c\n";
2015}
d9f424b2
JH
2016
2017{
75685a94
JH
2018 # [ID 20010407.006] matching utf8 return values from functions does not work
2019
d9f424b2
JH
2020 package ID_20010407_006;
2021
2022 sub x {
2023 "a\x{1234}";
2024 }
2025
2026 my $x = x;
2027 my $y;
2028
2029 $x =~ /(..)/; $y = $1;
2030 print "not " unless length($y) == 2 && $y eq $x;
75685a94 2031 print "ok 685\n";
d9f424b2
JH
2032
2033 x =~ /(..)/; $y = $1;
2034 print "not " unless length($y) == 2 && $y eq $x;
2035 print "ok 686\n";
2036}
a4c04bdc 2037
e2d8ce26 2038
fc8cd66c 2039$test = 687;
a4c04bdc 2040
987aaf07 2041# Force scalar context on the pattern match
fc8cd66c 2042sub ok ($;$) {
a4c04bdc 2043 my($ok, $name) = @_;
03949bff 2044 my $todo = $TODO ? " # TODO $TODO" : '';
a4c04bdc 2045
f0852a51 2046 printf "%sok %d - %s\n", ($ok ? "" : "not "), $test,
03949bff 2047 ($name||$Message)."$todo\tLine ".((caller)[2]);
a4c04bdc
NC
2048
2049 printf "# Failed test at line %d\n", (caller)[2] unless $ok;
2050
2051 $test++;
2052 return $ok;
2053}
2054
987aaf07
RGS
2055sub skip {
2056 my $why = shift;
eb5a2f3c 2057 $why =~ s/\n.*//s;
987aaf07
RGS
2058 my $n = @_ ? shift : 1;
2059 for (1..$n) {
2060 print "ok $test # skip: $why\n";
2061 $test++;
2062 }
2063 local $^W = 0;
2064 last SKIP;
2065}
2066
a4c04bdc
NC
2067{
2068 # Check that \x## works. 5.6.1 and 5.005_03 fail some of these.
2069 $x = "\x4e" . "E";
2070 ok ($x =~ /^\x4EE$/, "Check only 2 bytes of hex are matched.");
2071
2072 $x = "\x4e" . "i";
2073 ok ($x =~ /^\x4Ei$/, "Check that invalid hex digit stops it (2)");
2074
2075 $x = "\x4" . "j";
2076 ok ($x =~ /^\x4j$/, "Check that invalid hex digit stops it (1)");
2077
2078 $x = "\x0" . "k";
2079 ok ($x =~ /^\xk$/, "Check that invalid hex digit stops it (0)");
2080
2081 $x = "\x0" . "x";
2082 ok ($x =~ /^\xx$/, "\\xx isn't to be treated as \\0");
2083
2084 $x = "\x0" . "xa";
2085 ok ($x =~ /^\xxa$/, "\\xxa isn't to be treated as \\xa");
2086
2087 $x = "\x9" . "_b";
2088 ok ($x =~ /^\x9_b$/, "\\x9_b isn't to be treated as \\x9b");
2089
2090 print "# and now again in [] ranges\n";
2091
2092 $x = "\x4e" . "E";
2093 ok ($x =~ /^[\x4EE]{2}$/, "Check only 2 bytes of hex are matched.");
2094
2095 $x = "\x4e" . "i";
2096 ok ($x =~ /^[\x4Ei]{2}$/, "Check that invalid hex digit stops it (2)");
2097
2098 $x = "\x4" . "j";
2099 ok ($x =~ /^[\x4j]{2}$/, "Check that invalid hex digit stops it (1)");
2100
2101 $x = "\x0" . "k";
2102 ok ($x =~ /^[\xk]{2}$/, "Check that invalid hex digit stops it (0)");
2103
2104 $x = "\x0" . "x";
2105 ok ($x =~ /^[\xx]{2}$/, "\\xx isn't to be treated as \\0");
2106
2107 $x = "\x0" . "xa";
2108 ok ($x =~ /^[\xxa]{3}$/, "\\xxa isn't to be treated as \\xa");
2109
2110 $x = "\x9" . "_b";
2111 ok ($x =~ /^[\x9_b]{3}$/, "\\x9_b isn't to be treated as \\x9b");
2112
2113}
2114
2115{
2116 # Check that \x{##} works. 5.6.1 fails quite a few of these.
2117
2118 $x = "\x9b";
2119 ok ($x =~ /^\x{9_b}$/, "\\x{9_b} is to be treated as \\x9b");
2120
2121 $x = "\x9b" . "y";
2122 ok ($x =~ /^\x{9_b}y$/, "\\x{9_b} is to be treated as \\x9b (again)");
2123
2124 $x = "\x9b" . "y";
2125 ok ($x =~ /^\x{9b_}y$/, "\\x{9b_} is to be treated as \\x9b");
2126
2127 $x = "\x9b" . "y";
2128 ok ($x =~ /^\x{9_bq}y$/, "\\x{9_bc} is to be treated as \\x9b");
2129
2130 $x = "\x0" . "y";
2131 ok ($x =~ /^\x{x9b}y$/, "\\x{x9b} is to be treated as \\x0");
2132
2133 $x = "\x0" . "y";
2134 ok ($x =~ /^\x{0x9b}y$/, "\\x{0x9b} is to be treated as \\x0");
2135
2136 $x = "\x9b" . "y";
2137 ok ($x =~ /^\x{09b}y$/, "\\x{09b} is to be treated as \\x9b");
2138
2139 print "# and now again in [] ranges\n";
2140
2141 $x = "\x9b";
2142 ok ($x =~ /^[\x{9_b}]$/, "\\x{9_b} is to be treated as \\x9b");
2143
2144 $x = "\x9b" . "y";
2145 ok ($x =~ /^[\x{9_b}y]{2}$/, "\\x{9_b} is to be treated as \\x9b (again)");
2146
2147 $x = "\x9b" . "y";
2148 ok ($x =~ /^[\x{9b_}y]{2}$/, "\\x{9b_} is to be treated as \\x9b");
2149
2150 $x = "\x9b" . "y";
2151 ok ($x =~ /^[\x{9_bq}y]{2}$/, "\\x{9_bc} is to be treated as \\x9b");
2152
2153 $x = "\x0" . "y";
2154 ok ($x =~ /^[\x{x9b}y]{2}$/, "\\x{x9b} is to be treated as \\x0");
2155
2156 $x = "\x0" . "y";
2157 ok ($x =~ /^[\x{0x9b}y]{2}$/, "\\x{0x9b} is to be treated as \\x0");
2158
2159 $x = "\x9b" . "y";
2160 ok ($x =~ /^[\x{09b}y]{2}$/, "\\x{09b} is to be treated as \\x9b");
2161}
e2d8ce26
JP
2162
2163{
d9efae67
JH
2164 # high bit bug -- japhy
2165 my $x = "ab\200d";
2166 $x =~ /.*?\200/ or print "not ";
2167 print "ok 715\n";
e2d8ce26
JP
2168}
2169
4193bef7
JH
2170print "# some Unicode properties\n";
2171
d9efae67 2172{
4193bef7 2173 # Dashes, underbars, case.
d9efae67
JH
2174 print "not " unless "\x80" =~ /\p{in-latin1_SUPPLEMENT}/;
2175 print "ok 716\n";
ab13f0c7 2176
4193bef7 2177 # Complement, leading and trailing whitespace.
ab13f0c7
JH
2178 print "not " unless "\x80" =~ /\P{ ^ In Latin 1 Supplement }/;
2179 print "ok 717\n";
4193bef7 2180
f173cd49
JH
2181 # No ^In, dashes, case, dash, any intervening (word-break) whitespace.
2182 # (well, newlines don't work...)
2183 print "not " unless "\x80" =~ /\p{latin-1 supplement}/;
4193bef7
JH
2184 print "ok 718\n";
2185}
2186
2187{
2188 print "not " unless "a" =~ /\pL/;
2189 print "ok 719\n";
4193bef7 2190
4193bef7
JH
2191 print "not " unless "a" =~ /\p{IsLl}/;
2192 print "ok 720\n";
4193bef7 2193
c87b7cc2 2194 print "not " if "a" =~ /\p{IsLu}/;
4193bef7 2195 print "ok 721\n";
4193bef7 2196
61247495 2197 print "not " unless "a" =~ /\p{Ll}/;
4193bef7 2198 print "ok 722\n";
c87b7cc2 2199
61247495 2200 print "not " if "a" =~ /\p{Lu}/;
c87b7cc2
JH
2201 print "ok 723\n";
2202
61247495 2203 print "not " unless "A" =~ /\pL/;
c87b7cc2
JH
2204 print "ok 724\n";
2205
61247495 2206 print "not " unless "A" =~ /\p{IsLu}/;
c87b7cc2
JH
2207 print "ok 725\n";
2208
61247495 2209 print "not " if "A" =~ /\p{IsLl}/;
c87b7cc2
JH
2210 print "ok 726\n";
2211
61247495 2212 print "not " unless "A" =~ /\p{Lu}/;
c87b7cc2
JH
2213 print "ok 727\n";
2214
61247495 2215 print "not " if "A" =~ /\p{Ll}/;
c87b7cc2
JH
2216 print "ok 728\n";
2217
61247495 2218 print "not " if "a" =~ /\PL/;
c87b7cc2
JH
2219 print "ok 729\n";
2220
61247495 2221 print "not " if "a" =~ /\P{IsLl}/;
c87b7cc2 2222 print "ok 730\n";
61247495
JH
2223
2224 print "not " unless "a" =~ /\P{IsLu}/;
2225 print "ok 731\n";
2226
2227 print "not " if "a" =~ /\P{Ll}/;
2228 print "ok 732\n";
2229
2230 print "not " unless "a" =~ /\P{Lu}/;
2231 print "ok 733\n";
2232
2233 print "not " if "A" =~ /\PL/;
2234 print "ok 734\n";
2235
2236 print "not " if "A" =~ /\P{IsLu}/;
2237 print "ok 735\n";
2238
2239 print "not " unless "A" =~ /\P{IsLl}/;
2240 print "ok 736\n";
2241
2242 print "not " if "A" =~ /\P{Lu}/;
2243 print "ok 737\n";
2244
2245 print "not " unless "A" =~ /\P{Ll}/;
2246 print "ok 738\n";
2247
4193bef7 2248}
9b4e380a
JH
2249
2250{
2251 print "not " if "a" =~ /\p{Common}/;
2252 print "ok 739\n";
2253
2254 print "not " unless "1" =~ /\p{Common}/;
2255 print "ok 740\n";
2256}
2257
2258{
2259 print "not " if "a" =~ /\p{Inherited}/;
2260 print "ok 741\n";
2261
2262 print "not " unless "\x{300}" =~ /\p{Inherited}/;
2263 print "ok 742\n";
2264}
2265
2266{
12ac2576
JP
2267 # L& and LC are the same
2268 print "not " unless "a" =~ /\p{LC}/ and "a" =~ /\p{L&}/;
9b4e380a
JH
2269 print "ok 743\n";
2270
12ac2576 2271 print "not " if "1" =~ /\p{LC}/ or "1" =~ /\p{L&}/;
9b4e380a
JH
2272 print "ok 744\n";
2273}
d73e5302
JH
2274
2275{
904bc114 2276 print "not " unless "a" =~ /\p{Lowercase Letter}/;
d73e5302
JH
2277 print "ok 745\n";
2278
904bc114 2279 print "not " if "A" =~ /\p{lowercaseletter}/;
d73e5302
JH
2280 print "ok 746\n";
2281}
2282
2283{
cf25bb62 2284 print "not " unless "\x{AC00}" =~ /\p{HangulSyllables}/;
d73e5302
JH
2285 print "ok 747\n";
2286}
71d929cb
JH
2287
2288{
701a277b
JH
2289 # Script=, Block=, Category=
2290
71d929cb
JH
2291 print "not " unless "\x{0100}" =~ /\p{Script=Latin}/;
2292 print "ok 748\n";
2293
2294 print "not " unless "\x{0100}" =~ /\p{Block=LatinExtendedA}/;
2295 print "ok 749\n";
2296
2297 print "not " unless "\x{0100}" =~ /\p{Category=UppercaseLetter}/;
2298 print "ok 750\n";
2299}
2300
ef54fa25 2301{
701a277b
JH
2302 print "# the basic character classes and Unicode \n";
2303
ef54fa25
JH
2304 # 0100;LATIN CAPITAL LETTER A WITH MACRON;Lu;0;L;0041 0304;;;;N;LATIN CAPITAL LETTER A MACRON;;;0101;
2305 print "not " unless "\x{0100}" =~ /\w/;
2306 print "ok 751\n";
2307
2308 # 0660;ARABIC-INDIC DIGIT ZERO;Nd;0;AN;;0;0;0;N;;;;;
2309 print "not " unless "\x{0660}" =~ /\d/;
2310 print "ok 752\n";
2311
2312 # 1680;OGHAM SPACE MARK;Zs;0;WS;;;;;N;;;;;
2313 print "not " unless "\x{1680}" =~ /\s/;
2314 print "ok 753\n";
2315}
701a277b
JH
2316
2317{
2318 print "# folding matches and Unicode\n";
2319
2320 print "not " unless "a\x{100}" =~ /A/i;
2321 print "ok 754\n";
2322
27e28630 2323 print "not " unless "A\x{100}" =~ /a/i;
701a277b
JH
2324 print "ok 755\n";
2325
2326 print "not " unless "a\x{100}" =~ /a/i;
2327 print "ok 756\n";
2328
2329 print "not " unless "A\x{100}" =~ /A/i;
2330 print "ok 757\n";
bc517b45
JH
2331
2332 print "not " unless "\x{101}a" =~ /\x{100}/i;
2333 print "ok 758\n";
2334
2335 print "not " unless "\x{100}a" =~ /\x{100}/i;
2336 print "ok 759\n";
2337
2338 print "not " unless "\x{101}a" =~ /\x{101}/i;
2339 print "ok 760\n";
2340
2341 print "not " unless "\x{100}a" =~ /\x{101}/i;
2342 print "ok 761\n";
2343
2344 print "not " unless "a\x{100}" =~ /A\x{100}/i;
2345 print "ok 762\n";
2346
27e28630 2347 print "not " unless "A\x{100}" =~ /a\x{100}/i;
bc517b45
JH
2348 print "ok 763\n";
2349
2350 print "not " unless "a\x{100}" =~ /a\x{100}/i;
2351 print "ok 764\n";
2352
2353 print "not " unless "A\x{100}" =~ /A\x{100}/i;
2354 print "ok 765\n";
2355
2356 print "not " unless "a\x{100}" =~ /[A]/i;
2357 print "ok 766\n";
2358
27e28630 2359 print "not " unless "A\x{100}" =~ /[a]/i;
bc517b45
JH
2360 print "ok 767\n";
2361
2362 print "not " unless "a\x{100}" =~ /[a]/i;
2363 print "ok 768\n";
2364
2365 print "not " unless "A\x{100}" =~ /[A]/i;
2366 print "ok 769\n";
2367
2368 print "not " unless "\x{101}a" =~ /[\x{100}]/i;
2369 print "ok 770\n";
2370
2371 print "not " unless "\x{100}a" =~ /[\x{100}]/i;
2372 print "ok 771\n";
2373
2374 print "not " unless "\x{101}a" =~ /[\x{101}]/i;
2375 print "ok 772\n";
2376
2377 print "not " unless "\x{100}a" =~ /[\x{101}]/i;
2378 print "ok 773\n";
2379
701a277b 2380}
a5961de5
JH
2381
2382{
2383 use charnames ':full';
2384
2385 print "# LATIN LETTER A WITH GRAVE\n";
2386 my $lower = "\N{LATIN SMALL LETTER A WITH GRAVE}";
2387 my $UPPER = "\N{LATIN CAPITAL LETTER A WITH GRAVE}";
2388
bc517b45
JH
2389 print $lower =~ m/$UPPER/i ? "ok 774\n" : "not ok 774\n";
2390 print $UPPER =~ m/$lower/i ? "ok 775\n" : "not ok 775\n";
2391 print $lower =~ m/[$UPPER]/i ? "ok 776\n" : "not ok 776\n";
2392 print $UPPER =~ m/[$lower]/i ? "ok 777\n" : "not ok 777\n";
a5961de5
JH
2393
2394 print "# GREEK LETTER ALPHA WITH VRACHY\n";
2395
2396 $lower = "\N{GREEK CAPITAL LETTER ALPHA WITH VRACHY}";
2397 $UPPER = "\N{GREEK SMALL LETTER ALPHA WITH VRACHY}";
2398
bc517b45
JH
2399 print $lower =~ m/$UPPER/i ? "ok 778\n" : "not ok 778\n";
2400 print $UPPER =~ m/$lower/i ? "ok 779\n" : "not ok 779\n";
2401 print $lower =~ m/[$UPPER]/i ? "ok 780\n" : "not ok 780\n";
2402 print $UPPER =~ m/[$lower]/i ? "ok 781\n" : "not ok 781\n";
a5961de5
JH
2403
2404 print "# LATIN LETTER Y WITH DIAERESIS\n";
2405
2406 $lower = "\N{LATIN CAPITAL LETTER Y WITH DIAERESIS}";
2407 $UPPER = "\N{LATIN SMALL LETTER Y WITH DIAERESIS}";
bc517b45
JH
2408 print $lower =~ m/$UPPER/i ? "ok 782\n" : "not ok 782\n";
2409 print $UPPER =~ m/$lower/i ? "ok 783\n" : "not ok 783\n";
2410 print $lower =~ m/[$UPPER]/i ? "ok 784\n" : "not ok 784\n";
2411 print $UPPER =~ m/[$lower]/i ? "ok 785\n" : "not ok 785\n";
a5961de5 2412}
55da9344
JH
2413
2414{
2415 use warnings;
2416 use charnames ':full';
2417
2418 print "# GREEK CAPITAL LETTER SIGMA vs COMBINING GREEK PERISPOMENI\n";
2419
2420 my $SIGMA = "\N{GREEK CAPITAL LETTER SIGMA}";
d07ddd77 2421 my $char = "\N{COMBINING GREEK PERISPOMENI}";
55da9344 2422
bc517b45
JH
2423 # Before #13843 this was failing by matching falsely.
2424 print "_:$char:_" =~ m/_:$SIGMA:_/i ? "not ok 786\n" : "ok 786\n";
55da9344 2425}
b7c83a7e
JH
2426
2427{
2428 print "# \\X\n";
2429
2430 use charnames ':full';
2431
eb08e2da
JH
2432 print "a!" =~ /^(\X)!/ && $1 eq "a" ?
2433 "ok 787\n" : "not ok 787 # $1\n";
2434 print "\xDF!" =~ /^(\X)!/ && $1 eq "\xDF" ?
2435 "ok 788\n" : "not ok 788 # $1\n";
2436 print "\x{100}!" =~ /^(\X)!/ && $1 eq "\x{100}" ?
2437 "ok 789\n" : "not ok 789 # $1\n";
2438 print "\x{100}\x{300}!" =~ /^(\X)!/ && $1 eq "\x{100}\x{300}" ?
2439 "ok 790\n" : "not ok 790 # $1\n";
2440 print "\N{LATIN CAPITAL LETTER E}!" =~ /^(\X)!/ &&
2441 $1 eq "\N{LATIN CAPITAL LETTER E}" ?
2442 "ok 791\n" : "not ok 791 # $1\n";
2443 print "\N{LATIN CAPITAL LETTER E}\N{COMBINING GRAVE ACCENT}!" =~
2444 /^(\X)!/ &&
2445 $1 eq "\N{LATIN CAPITAL LETTER E}\N{COMBINING GRAVE ACCENT}" ?
2446 "ok 792\n" : "not ok 792 # $1\n";
b7c83a7e 2447}
112bedeb
JH
2448
2449{
2450 print "#\\C and \\X\n";
2451
2452 print "!abc!" =~ /a\Cc/ ? "ok 793\n" : "not ok 793\n";
2453 print "!abc!" =~ /a\Xc/ ? "ok 794\n" : "not ok 794\n";
2454}
09091399
JH
2455
2456{
2457 print "# FINAL SIGMA\n";
2458
2459 my $SIGMA = "\x{03A3}"; # CAPITAL
2460 my $Sigma = "\x{03C2}"; # SMALL FINAL
2461 my $sigma = "\x{03C3}"; # SMALL
2462
2463 print $SIGMA =~ /$SIGMA/i ? "ok 795\n" : "not ok 795\n";
2464 print $SIGMA =~ /$Sigma/i ? "ok 796\n" : "not ok 796\n";
2465 print $SIGMA =~ /$sigma/i ? "ok 797\n" : "not ok 797\n";
2466
2467 print $Sigma =~ /$SIGMA/i ? "ok 798\n" : "not ok 798\n";
2468 print $Sigma =~ /$Sigma/i ? "ok 799\n" : "not ok 799\n";
2469 print $Sigma =~ /$sigma/i ? "ok 800\n" : "not ok 800\n";
2470
2471 print $sigma =~ /$SIGMA/i ? "ok 801\n" : "not ok 801\n";
2472 print $sigma =~ /$Sigma/i ? "ok 802\n" : "not ok 802\n";
2473 print $sigma =~ /$sigma/i ? "ok 803\n" : "not ok 803\n";
2474
2475 print $SIGMA =~ /[$SIGMA]/i ? "ok 804\n" : "not ok 804\n";
2476 print $SIGMA =~ /[$Sigma]/i ? "ok 805\n" : "not ok 805\n";
2477 print $SIGMA =~ /[$sigma]/i ? "ok 806\n" : "not ok 806\n";
2478
2479 print $Sigma =~ /[$SIGMA]/i ? "ok 807\n" : "not ok 807\n";
2480 print $Sigma =~ /[$Sigma]/i ? "ok 808\n" : "not ok 808\n";
2481 print $Sigma =~ /[$sigma]/i ? "ok 809\n" : "not ok 809\n";
2482
2483 print $sigma =~ /[$SIGMA]/i ? "ok 810\n" : "not ok 810\n";
2484 print $sigma =~ /[$Sigma]/i ? "ok 811\n" : "not ok 811\n";
2485 print $sigma =~ /[$sigma]/i ? "ok 812\n" : "not ok 812\n";
2486}
e036fef9
JF
2487
2488{
2489 print "# parlez-vous?\n";
2490
2491 use charnames ':full';
2492
2493 print "fran\N{LATIN SMALL LETTER C}ais" =~
2494 /fran.ais/ &&
2495 $& eq "francais" ?
2496 "ok 813\n" : "not ok 813\n";
2497
2498 print "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" =~
2499 /fran.ais/ &&
2500 $& eq "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" ?
2501 "ok 814\n" : "not ok 814\n";
2502
2503 print "fran\N{LATIN SMALL LETTER C}ais" =~
2504 /fran\Cais/ &&
2505 $& eq "francais" ?
2506 "ok 815\n" : "not ok 815\n";
2507
2508 print "franc\N{COMBINING CEDILLA}ais" =~
2509 /franc\C\Cais/ ? # COMBINING CEDILLA is two bytes when encoded
2510 "ok 816\n" : "not ok 816\n";
2511
2512 print "fran\N{LATIN SMALL LETTER C}ais" =~
2513 /fran\Xais/ &&
2514 $& eq "francais" ?
2515 "ok 817\n" : "not ok 817\n";
2516
2517 print "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" =~
2518 /fran\Xais/ &&
2519 $& eq "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" ?
2520 "ok 818\n" : "not ok 818\n";
2521
2522 print "franc\N{COMBINING CEDILLA}ais" =~
2523 /fran\Xais/ &&
2524 $& eq "franc\N{COMBINING CEDILLA}ais" ?
2525 "ok 819\n" : "not ok 819\n";
2526
2527 print "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" =~
2528 /fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais/ &&
2529 $& eq "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" ?
2530 "ok 820\n" : "not ok 820\n";
2531
2532 print "franc\N{COMBINING CEDILLA}ais" =~
2533 /franc\N{COMBINING CEDILLA}ais/ &&
2534 $& eq "franc\N{COMBINING CEDILLA}ais" ?
2535 "ok 821\n" : "not ok 821\n";
2536
2537 print "fran\N{LATIN SMALL LETTER C}ais" =~
2538 /fran(?:c\N{COMBINING CEDILLA}?|\N{LATIN SMALL LETTER C WITH CEDILLA})ais/ &&
2539 $& eq "francais" ?
2540 "ok 822\n" : "not ok 822\n";
2541
2542 print "fran\N{LATIN SMALL LETTER C}ais" =~
2543 /fran(?:c\N{COMBINING CEDILLA}?|\N{LATIN SMALL LETTER C WITH CEDILLA})ais/ &&
2544 $& eq "francais" ?
2545 "ok 823\n" : "not ok 823\n";
2546
2547 print "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" =~
2548 /fran(?:c\N{COMBINING CEDILLA}?|\N{LATIN SMALL LETTER C WITH CEDILLA})ais/ &&
2549 $& eq "fran\N{LATIN SMALL LETTER C WITH CEDILLA}ais" ?
2550 "ok 824\n" : "not ok 824\n";
2551
2552 print "franc\N{COMBINING CEDILLA}ais" =~
2553 /fran(?:c\N{COMBINING CEDILLA}?|\N{LATIN SMALL LETTER C WITH CEDILLA})ais/ &&
2554 $& eq "franc\N{COMBINING CEDILLA}ais" ?
2555 "ok 825\n" : "not ok 825\n";
2556}
ffce6cc2
JH
2557
2558{
2559 print "# Does lingering (and useless) UTF8 flag mess up /i matching?\n";
2560
2561 {
2562 my $regex = "ABcde";
2563 my $string = "abcDE\x{100}";
2564 chop($string);
2565 if ($string =~ m/$regex/i) {
2566 print "ok 826\n";
2567 } else {
2568 print "not ok 826\n";
2569 }
2570 }
2571
2572 {
2573 my $regex = "ABcde\x{100}";
2574 my $string = "abcDE";
2575 chop($regex);
2576 if ($string =~ m/$regex/i) {
2577 print "ok 827\n";
2578 } else {
2579 print "not ok 827\n";
2580 }
2581 }
2582
2583 {
2584 my $regex = "ABcde\x{100}";
2585 my $string = "abcDE\x{100}";
2586 chop($regex);
2587 chop($string);
2588 if ($string =~ m/$regex/i) {
2589 print "ok 828\n";
2590 } else {
2591 print "not ok 828\n";
2592 }
2593 }
2594}
d07ddd77
JH
2595
2596{
2597 print "# more SIGMAs\n";
2598
2599 my $SIGMA = "\x{03A3}"; # CAPITAL
2600 my $Sigma = "\x{03C2}"; # SMALL FINAL
2601 my $sigma = "\x{03C3}"; # SMALL
2602
2603 my $S3 = "$SIGMA$Sigma$sigma";
2604
2605 print ":$S3:" =~ /:(($SIGMA)+):/i && $1 eq $S3 && $2 eq $sigma ?
2606 "ok 829\n" : "not ok 829\n";
2607 print ":$S3:" =~ /:(($Sigma)+):/i && $1 eq $S3 && $2 eq $sigma ?
2608 "ok 830\n" : "not ok 830\n";
2609 print ":$S3:" =~ /:(($sigma)+):/i && $1 eq $S3 && $2 eq $sigma ?
2610 "ok 831\n" : "not ok 831\n";
2611
2612 print ":$S3:" =~ /:(([$SIGMA])+):/i && $1 eq $S3 && $2 eq $sigma ?
2613 "ok 832\n" : "not ok 832\n";
2614 print ":$S3:" =~ /:(([$Sigma])+):/i && $1 eq $S3 && $2 eq $sigma ?
2615 "ok 833\n" : "not ok 833\n";
2616 print ":$S3:" =~ /:(([$sigma])+):/i && $1 eq $S3 && $2 eq $sigma ?
2617 "ok 834\n" : "not ok 834\n";
2618}
925f9e00
JH
2619
2620{
2621 print "# LATIN SMALL LETTER SHARP S\n";
2622
2623 use charnames ':full';
2624
fc8cd66c 2625 $test= 835;
925f9e00 2626
fc8cd66c
YO
2627 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /\N{LATIN SMALL LETTER SHARP S}/);
2628 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /\N{LATIN SMALL LETTER SHARP S}/i);
925f9e00 2629
fc8cd66c
YO
2630 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /[\N{LATIN SMALL LETTER SHARP S}]/);
2631 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i);
925f9e00 2632
fc8cd66c
YO
2633 ok("ss" =~ /\N{LATIN SMALL LETTER SHARP S}/i);
2634 ok("SS" =~ /\N{LATIN SMALL LETTER SHARP S}/i);
2635 ok("ss" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i);
2636 ok("SS" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i);
925f9e00 2637
fc8cd66c
YO
2638 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /ss/i);
2639 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /SS/i);
925f9e00 2640}
d8f6a732
JH
2641
2642{
2643 print "# more whitespace: U+0085, U+2028, U+2029\n";
2644
2645 # U+0085 needs to be forced to be Unicode, the \x{100} does that.
250d67eb
JH
2646 if ($ordA == 193) {
2647 print "<\x{100}\x{0085}>" =~ /<\x{100}e>/ ? "ok 845\n" : "not ok 845\n";
2648 } else {
2649 print "<\x{100}\x{0085}>" =~ /<\x{100}\s>/ ? "ok 845\n" : "not ok 845\n";
2650 }
5486206c
JH
2651 print "<\x{2028}>" =~ /<\s>/ ? "ok 846\n" : "not ok 846\n";
2652 print "<\x{2029}>" =~ /<\s>/ ? "ok 847\n" : "not ok 847\n";
d8f6a732
JH
2653}
2654
def8e4ea 2655{
229196e5 2656 print "# . with /s should work on characters, as opposed to bytes\n";
def8e4ea
JH
2657
2658 my $s = "\x{e4}\x{100}";
2659
2660 # This is not expected to match: the point is that
2661 # neither should we get "Malformed UTF-8" warnings.
2662 print $s =~ /\G(.+?)\n/gcs ?
5486206c 2663 "not ok 848\n" : "ok 848\n";
def8e4ea
JH
2664
2665 my @c;
2666
2667 while ($s =~ /\G(.)/gs) {
2668 push @c, $1;
2669 }
2670
5486206c 2671 print join("", @c) eq $s ? "ok 849\n" : "not ok 849\n";
a0804c9e
A
2672
2673 my $t1 = "Q003\n\n\x{e4}\x{f6}\n\nQ004\n\n\x{e7}"; # test only chars < 256
2674 my $r1 = "";
2675 while ($t1 =~ / \G ( .+? ) \n\s+ ( .+? ) ( $ | \n\s+ ) /xgcs) {
2676 $r1 .= $1 . $2;
2677 }
2678
2679 my $t2 = $t1 . "\x{100}"; # repeat with a larger char
2680 my $r2 = "";
2681 while ($t2 =~ / \G ( .+? ) \n\s+ ( .+? ) ( $ | \n\s+ ) /xgcs) {
2682 $r2 .= $1 . $2;
2683 }
2684 $r2 =~ s/\x{100}//;
5486206c 2685 print $r1 eq $r2 ? "ok 850\n" : "not ok 850\n";
def8e4ea 2686}
e54858b0
JH
2687
2688{
2689 print "# Unicode lookbehind\n";
2690
5486206c
JH
2691 print "A\x{100}B" =~ /(?<=A.)B/ ? "ok 851\n" : "not ok 851\n";
2692 print "A\x{200}\x{300}B" =~ /(?<=A..)B/ ? "ok 852\n" : "not ok 852\n";
2693 print "\x{400}AB" =~ /(?<=\x{400}.)B/ ? "ok 853\n" : "not ok 853\n";
2694 print "\x{500\x{600}}B" =~ /(?<=\x{500}.)B/ ? "ok 854\n" : "not ok 854\n";
e54858b0 2695}
c46248c1
JH
2696
2697{
85fd1718
JH
2698 print "# UTF-8 hash keys and /\$/\n";
2699 # http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-01/msg01327.html
2700
2701 my $u = "a\x{100}";
2702 my $v = substr($u,0,1);
2703 my $w = substr($u,1,1);
2704 my %u = ( $u => $u, $v => $v, $w => $w );
8d21bda2 2705 my $i = 855;
85fd1718
JH
2706 for (keys %u) {
2707 my $m1 = /^\w*$/ ? 1 : 0;
2708 my $m2 = $u{$_}=~/^\w*$/ ? 1 : 0;
2709 print $m1 == $m2 ? "ok $i\n" : "not ok $i # $m1 $m2\n";
2710 $i++;
2711 }
2712}
8d21bda2
JH
2713
2714{
2715 print "# [ID 20020124.005]\n";
2716 # Fixed by #14795.
2717 my $i = 858;
2718 for my $char ("a", "\x{df}", "\x{100}"){
2719 $x = "$char b $char";
2720 $x =~ s{($char)}{
2721 "c" =~ /c/;
2722 "x";
2723 }ge;
2724 print substr($x,0,1) eq substr($x,-1,1) ?
2725 "ok $i\n" : "not ok $i # debug: $x\n";
2726 $i++;
2727 }
2728}
6e602e29
A
2729
2730{
2731 print "# SEGV in s/// and UTF-8\n";
2732 $s = "s#\x{100}" x 4;
2733 $s =~ s/[^\w]/ /g;
2734 print $s eq "s \x{100}" x 4 ? "ok 861\n" : "not ok 861\n";
2735}
f272994b
A
2736
2737{
2738 print "# UTF-8 bug (maybe alreayd known?)\n";
2739 my $u;
2740
2741 $u = "foo";
2742 $u =~ s/./\x{100}/g;
2743 print $u eq "\x{100}\x{100}\x{100}" ? "ok 862\n" : "not ok 862\n";
2744
2745 $u = "foobar";
2746 $u =~ s/[ao]/\x{100}/g;
2747 print $u eq "f\x{100}\x{100}b\x{100}r" ? "ok 863\n" : "not ok 863\n";
2748
2749 $u =~ s/\x{100}/e/g;
2750 print $u eq "feeber" ? "ok 864\n" : "not ok 864\n";
2751}
2752
33b8afdf
JH
2753{
2754 print "# UTF-8 bug with s///\n";
2755 # check utf8/non-utf8 mixtures
2756 # try to force all float/anchored check combinations
2757 my $c = "\x{100}";
fc8cd66c 2758 $test = 865;
33b8afdf
JH
2759 my $subst;
2760 for my $re (
2761 "xx.*$c", "x.*$c$c", "$c.*xx", "$c$c.*x", "xx.*(?=$c)", "(?=$c).*xx",
2762 ) {
2763 print "xxx" =~ /$re/ ? "not ok $test\n" : "ok $test\n";
2764 ++$test;
2765 print +($subst = "xxx") =~ s/$re// ? "not ok $test\n" : "ok $test\n";
2766 ++$test;
2767 }
2768 for my $re ("xx.*$c*", "$c*.*xx") {
2769 print "xxx" =~ /$re/ ? "ok $test\n" : "not ok $test\n";
2770 ++$test;
2771 ($subst = "xxx") =~ s/$re//;
2772 print $subst eq '' ? "ok $test\n" : "not ok $test\t# $subst\n";
2773 ++$test;
2774 }
2775 for my $re ("xxy*", "y*xx") {
2776 print "xx$c" =~ /$re/ ? "ok $test\n" : "not ok $test\n";
2777 ++$test;
2778 ($subst = "xx$c") =~ s/$re//;
2779 print $subst eq $c ? "ok $test\n" : "not ok $test\n";
2780 ++$test;
2781 print "xy$c" =~ /$re/ ? "not ok $test\n" : "ok $test\n";
2782 ++$test;
2783 print +($subst = "xy$c") =~ /$re/ ? "not ok $test\n" : "ok $test\n";
2784 ++$test;
2785 }
2786 for my $re ("xy$c*z", "x$c*yz") {
2787 print "xyz" =~ /$re/ ? "ok $test\n" : "not ok $test\n";
2788 ++$test;
2789 ($subst = "xyz") =~ s/$re//;
2790 print $subst eq '' ? "ok $test\n" : "not ok $test\n";
2791 ++$test;
2792 }
2793}
574c8022 2794
ff385a1b
JF
2795{
2796 print "# qr/.../x\n";
fc8cd66c 2797 $test = 893;
ff385a1b
JF
2798
2799 my $R = qr/ A B C # D E/x;
2800
2801 print eval {"ABCDE" =~ $R} ? "ok $test\n" : "not ok $test\n";
2802 $test++;
2803
2804 print eval {"ABCDE" =~ m/$R/} ? "ok $test\n" : "not ok $test\n";
2805 $test++;
2806
2807 print eval {"ABCDE" =~ m/($R)/} ? "ok $test\n" : "not ok $test\n";
2808 $test++;
2809}
bc45ce41
JH
2810
2811{
2812 print "# illegal Unicode properties\n";
fc8cd66c 2813 $test = 896;
bc45ce41
JH
2814
2815 print eval { "a" =~ /\pq / } ? "not ok $test\n" : "ok $test\n";
2816 $test++;
2817
2818 print eval { "a" =~ /\p{qrst} / } ? "not ok $test\n" : "ok $test\n";
2819 $test++;
2820}
d65afb4b
HS
2821
2822{
491fd90a 2823 print "# [ID 20020412.005] wrong pmop flags checked when empty pattern\n";
d65afb4b 2824 # requires reuse of last successful pattern
fc8cd66c 2825 $test = 898;
d65afb4b
HS
2826 $test =~ /\d/;
2827 for (0 .. 1) {
2828 my $match = ?? + 0;
2829 if ($match != $_) {
2830 print "ok $test\n";
2831 } else {
2832 printf "not ok %s\t# 'match once' %s on %s iteration\n", $test,
2833 $match ? 'succeeded' : 'failed', $_ ? 'second' : 'first';
2834 }
2835 ++$test;
2836 }
2837 $test =~ /(\d)/;
2838 my $result = join '', $test =~ //g;
2839 if ($result eq $test) {
2840 print "ok $test\n";
2841 } else {
2842 printf "not ok %s\t# expected '%s', got '%s'\n", $test, $test, $result;
2843 }
2844 ++$test;
2845}
491fd90a
JH
2846
2847print "# user-defined character properties\n";
2848
2849sub InKana1 {
2850 return <<'END';
28513040 309F
285230A0 30FF
2853END
2854}
2855
2856sub InKana2 {
2857 return <<'END';
2858+utf8::InHiragana
2859+utf8::InKatakana
2860END
2861}
2862
2863sub InKana3 {
2864 return <<'END';
2865+utf8::InHiragana
2866+utf8::InKatakana
2867-utf8::IsCn
2868END
2869}
2870
2871sub InNotKana {
2872 return <<'END';
2873!utf8::InHiragana
2874-utf8::InKatakana
2875+utf8::IsCn
2876END
2877}
2878
2879$test = 901;
2880
2881print "\x{3040}" =~ /\p{InKana1}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2882print "\x{303F}" =~ /\P{InKana1}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2883
2884print "\x{3040}" =~ /\p{InKana2}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2885print "\x{303F}" =~ /\P{InKana2}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2886
2887print "\x{3041}" =~ /\p{InKana3}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2888print "\x{3040}" =~ /\P{InKana3}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2889
2890print "\x{3040}" =~ /\p{InNotKana}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2891print "\x{3041}" =~ /\P{InNotKana}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2892
11ef8fdd
JH
2893sub InConsonant { # Not EBCDIC-aware.
2894 return <<EOF;
28950061 007f
2896-0061
2897-0065
2898-0069
2899-006f
2900-0075
2901EOF
2902}
2903
2904print "d" =~ /\p{InConsonant}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2905print "e" =~ /\P{InConsonant}/ ? "ok $test\n" : "not ok $test\n"; $test++;
2906
faf11cac
HS
2907{
2908 print "# [ID 20020630.002] utf8 regex only matches 32k\n";
2909 $test = 911;
2910 for ([ 'byte', "\x{ff}" ], [ 'utf8', "\x{1ff}" ]) {
2911 my($type, $char) = @$_;
2912 for my $len (32000, 32768, 33000) {
2913 my $s = $char . "f" x $len;
2914 my $r = $s =~ /$char([f]*)/gc;
2915 print $r ? "ok $test\n" : "not ok $test\t# <$type x $len> fail\n";
2916 ++$test;
2917 print +(!$r or pos($s) == $len + 1) ? "ok $test\n"
2918 : "not ok $test\t# <$type x $len> pos @{[ pos($s) ]}\n";
2919 ++$test;
2920 }
2921 }
2922}
2923
2924$test = 923;
f3b1e556
YST
2925
2926$a = bless qr/foo/, 'Foo';
2927print(('goodfood' =~ $a ? '' : 'not '),
2928 "ok $test\t# reblessed qr// matches\n");
2929++$test;
2930
2931print(($a eq '(?-xism:foo)' ? '' : 'not '),
2932 "ok $test\t# reblessed qr// stringizes\n");
2933++$test;
446eaa42
YST
2934
2935$x = "\x{3fe}";
cb50f42d
YST
2936$z=$y = "\317\276"; # $y is byte representation of $x
2937
446eaa42
YST
2938$a = qr/$x/;
2939print(($x =~ $a ? '' : 'not '), "ok $test - utf8 interpolation in qr//\n");
2940++$test;
2941
2942print(("a$a" =~ $x ? '' : 'not '),
cb50f42d
YST
2943 "ok $test - stringifed qr// preserves utf8\n");
2944++$test;
2945
2946print(("a$x" =~ /^a$a\z/ ? '' : 'not '),
2947 "ok $test - interpolated qr// preserves utf8\n");
2948++$test;
2949
2950print(("a$x" =~ /^a(??{$a})\z/ ? '' : 'not '),
2951 "ok $test - postponed interpolation of qr// preserves utf8\n");
2952++$test;
2953
ab01544f
RGS
2954print((length(qr/##/x) == 12 ? '' : 'not '),
2955 "ok $test - ## in qr// doesn't corrupt memory [perl #17776]\n");
2956++$test;
2957
cb50f42d
YST
2958{ use re 'eval';
2959
2960print(("$x$x" =~ /^$x(??{$x})\z/ ? '' : 'not '),
2961 "ok $test - postponed utf8 string in utf8 re matches utf8\n");
2962++$test;
2963
2964print(("$y$x" =~ /^$y(??{$x})\z/ ? '' : 'not '),
2965 "ok $test - postponed utf8 string in non-utf8 re matches utf8\n");
446eaa42
YST
2966++$test;
2967
cb50f42d
YST
2968print(("$y$x" !~ /^$y(??{$y})\z/ ? '' : 'not '),
2969 "ok $test - postponed non-utf8 string in non-utf8 re doesn't match utf8\n");
446eaa42
YST
2970++$test;
2971
cb50f42d
YST
2972print(("$x$x" !~ /^$x(??{$y})\z/ ? '' : 'not '),
2973 "ok $test - postponed non-utf8 string in utf8 re doesn't match utf8\n");
446eaa42
YST
2974++$test;
2975
cb50f42d
YST
2976print(("$y$y" =~ /^$y(??{$y})\z/ ? '' : 'not '),
2977 "ok $test - postponed non-utf8 string in non-utf8 re matches non-utf8\n");
2978++$test;
2979
2980print(("$x$y" =~ /^$x(??{$y})\z/ ? '' : 'not '),
2981 "ok $test - postponed non-utf8 string in utf8 re matches non-utf8\n");
2982++$test;
2983$y = $z; # reset $y after upgrade
2984
2985print(("$x$y" !~ /^$x(??{$x})\z/ ? '' : 'not '),
2986 "ok $test - postponed utf8 string in utf8 re doesn't match non-utf8\n");
2987++$test;
2988$y = $z; # reset $y after upgrade
2989
2990print(("$y$y" !~ /^$y(??{$x})\z/ ? '' : 'not '),
2991 "ok $test - postponed utf8 string in non-utf8 re doesn't match non-utf8\n");
2992++$test;
2993
2994} # no re 'eval'
2995
2d92f8a0
JH
2996print "# more user-defined character properties\n";
2997
2998sub IsSyriac1 {
2999 return <<'END';
30000712 072C
30010730 074A
3002END
3003}
3004
711a919c
ST
3005ok("\x{0712}" =~ /\p{IsSyriac1}/, '\x{0712}, \p{IsSyriac1}');
3006ok("\x{072F}" =~ /\P{IsSyriac1}/, '\x{072F}, \P{IsSyriac1}');
2d92f8a0
JH
3007
3008sub Syriac1 {
3009 return <<'END';
30100712 072C
30110730 074A
3012END
3013}
3014
711a919c
ST
3015ok("\x{0712}" =~ /\p{Syriac1}/, '\x{0712}, \p{Syriac1}');
3016ok("\x{072F}" =~ /\P{Syriac1}/, '\x{072F}, \p{Syriac1}');
2d92f8a0 3017
4a818d86
ST
3018print "# user-defined character properties may lack \\n at the end\n";
3019sub InGreekSmall { return "03B1\t03C9" }
3020sub InGreekCapital { return "0391\t03A9\n-03A2" }
3021
3022ok("\x{03C0}" =~ /\p{InGreekSmall}/, "Small pi");
3023ok("\x{03C2}" =~ /\p{InGreekSmall}/, "Final sigma");
3024ok("\x{03A0}" =~ /\p{InGreekCapital}/, "Capital PI");
3025ok("\x{03A2}" =~ /\P{InGreekCapital}/, "Reserved");
3026
711a919c
ST
3027sub AsciiHexAndDash {
3028 return <<'END';
3029+utf8::ASCII_Hex_Digit
3030+utf8::Dash
3031END
3032}
3033
3034ok("-" =~ /\p{Dash}/, "'-' is Dash");
3035ok("A" =~ /\p{ASCII_Hex_Digit}/, "'A' is ASCII_Hex_Digit");
3036ok("-" =~ /\p{AsciiHexAndDash}/, "'-' is AsciiHexAndDash");
3037ok("A" =~ /\p{AsciiHexAndDash}/, "'A' is AsciiHexAndDash");
3038
0e933229 3039{
3a2263fe
RGS
3040 print "# Change #18179\n";
3041 # previously failed with "panic: end_shift
0e933229
IH
3042 my $s = "\x{100}" x 5;
3043 my $ok = $s =~ /(\x{100}{4})/;
3044 my($ord, $len) = (ord $1, length $1);
3045 print +($ok && $ord == 0x100 && $len == 4)
fc8cd66c 3046 ? "ok $test\n" : "not ok $test\t# [#18179] $ok/$ord/$len\n";
0e933229
IH
3047 ++$test;
3048}
3049
f14c76ed
RGS
3050{
3051 print "# [perl #15763]\n";
3052
3053 $a = "x\x{100}";
3054 chop $a; # but leaves the UTF-8 flag
3055 $a .= "y"; # 1 byte before "y"
3056
3057 ok($a =~ /^\C/, 'match one \C on 1-byte UTF-8');
3058 ok($a =~ /^\C{1}/, 'match \C{1}');
3059
3060 ok($a =~ /^\Cy/, 'match \Cy');
3061 ok($a =~ /^\C{1}y/, 'match \C{1}y');
3062
3063 $a = "\x{100}y"; # 2 bytes before "y"
3064
3065 ok($a =~ /^\C/, 'match one \C on 2-byte UTF-8');
3066 ok($a =~ /^\C{1}/, 'match \C{1}');
3067 ok($a =~ /^\C\C/, 'match two \C');
3068 ok($a =~ /^\C{2}/, 'match \C{2}');
3069
3070 ok($a =~ /^\C\C\C/, 'match three \C on 2-byte UTF-8 and a byte');
3071 ok($a =~ /^\C{3}/, 'match \C{3}');
3072
3073 ok($a =~ /^\C\Cy/, 'match two \C');
3074 ok($a =~ /^\C{2}y/, 'match \C{2}');
3075
836995da
RGS
3076 ok($a !~ /^\C\C\Cy/, q{don't match three \Cy});
3077 ok($a !~ /^\C{2}\Cy/, q{don't match \C{3}y});
f14c76ed
RGS
3078
3079 $a = "\x{1000}y"; # 3 bytes before "y"
3080
3081 ok($a =~ /^\C/, 'match one \C on three-byte UTF-8');
3082 ok($a =~ /^\C{1}/, 'match \C{1}');
3083 ok($a =~ /^\C\C/, 'match two \C');
3084 ok($a =~ /^\C{2}/, 'match \C{2}');
3085 ok($a =~ /^\C\C\C/, 'match three \C');
3086 ok($a =~ /^\C{3}/, 'match \C{3}');
3087
3088 ok($a =~ /^\C\C\C\C/, 'match four \C on three-byte UTF-8 and a byte');
3089 ok($a =~ /^\C{4}/, 'match \C{4}');
3090
3091 ok($a =~ /^\C\C\Cy/, 'match three \Cy');
3092 ok($a =~ /^\C{3}y/, 'match \C{3}y');
3093
836995da
RGS
3094 ok($a !~ /^\C\C\C\C\y/, q{don't match four \Cy});
3095 ok($a !~ /^\C{4}y/, q{don't match \C{4}y});
f14c76ed
RGS
3096}
3097
88044e37
NC
3098{
3099 local $\;
3100 $_ = 'aaaaaaaaaa';
3101 utf8::upgrade($_); chop $_; $\="\n";
3102 ok(/[^\s]+/, "m/[^\s]/ utf8");
3103 ok(/[^\d]+/, "m/[^\d]/ utf8");
3104 ok(($a = $_, $_ =~ s/[^\s]+/./g), "s/[^\s]/ utf8");
3105 ok(($a = $_, $a =~ s/[^\d]+/./g), "s/[^\s]/ utf8");
3106}
f14c76ed 3107
ce97f9d6
JH
3108ok("\x{100}" =~ /\x{100}/, "[perl #15397]");
3109ok("\x{100}" =~ /(\x{100})/, "[perl #15397]");
3110ok("\x{100}" =~ /(\x{100}){1}/, "[perl #15397]");
3111ok("\x{100}\x{100}" =~ /(\x{100}){2}/, "[perl #15397]");
3112ok("\x{100}\x{100}" =~ /(\x{100})(\x{100})/, "[perl #15397]");
3113
fbeb8e69
JH
3114$x = "CD";
3115$x =~ /(AB)*?CD/;
3116ok(!defined $1, "[perl #7471]");
ce97f9d6 3117
fbeb8e69
JH
3118$x = "CD";
3119$x =~ /(AB)*CD/;
3120ok(!defined $1, "[perl #7471]");
3121
ec391688
JH
3122$pattern = "^(b+?|a){1,2}c";
3123ok("bac" =~ /$pattern/ && $1 eq 'a', "[perl #3547]");
3124ok("bbac" =~ /$pattern/ && $1 eq 'a', "[perl #3547]");
3125ok("bbbac" =~ /$pattern/ && $1 eq 'a', "[perl #3547]");
3126ok("bbbbac" =~ /$pattern/ && $1 eq 'a', "[perl #3547]");
3127
a30b2f1f
RGS
3128{
3129 # [perl #18232]
3130 "\x{100}" =~ /(.)/;
3131 ok( $1 eq "\x{100}", '$1 is utf-8 [perl #18232]' );
3132 { 'a' =~ /./; }
3133 ok( $1 eq "\x{100}", '$1 is still utf-8' );
3134 ok( $1 ne "\xC4\x80", '$1 is not non-utf-8' );
3135}
3136
f119b0fb
JH
3137{
3138 use utf8;
3139 my $attr = 'Name-1' ;
3140
3141 my $NormalChar = qr/[\p{IsDigit}\p{IsLower}\p{IsUpper}]/;
3142 my $NormalWord = qr/${NormalChar}+?/;
3143 my $PredNameHyphen = qr/^${NormalWord}(\-${NormalWord})*?$/;
3144
3145 $attr =~ /^$/;
3146 ok( $attr =~ $PredNameHyphen, "[perl #19767] original test" );
3147}
3148
3149{
3150 use utf8;
3151 "a" =~ m/[b]/;
3152 ok ( "0" =~ /\p{N}+\z/, "[perl #19767] variant test" );
3153}
3154
faf82a0b
AE
3155{
3156
3157 $p = 1;
3158 foreach (1,2,3,4) {
3159 $p++ if /(??{ $p })/
3160 }
96704b71 3161 iseq ($p, 5, "[perl #20683] (??{ }) returns stale values");
faf82a0b
AE
3162 { package P; $a=1; sub TIESCALAR { bless[] } sub FETCH { $a++ } }
3163 tie $p, P;
3164 foreach (1,2,3,4) {
3165 /(??{ $p })/
3166 }
96704b71 3167 iseq ( $p, 5, "(??{ }) returns stale values");
faf82a0b
AE
3168}
3169
351208f1
JH
3170{
3171 # Subject: Odd regexp behavior
3172 # From: Markus Kuhn <Markus.Kuhn@cl.cam.ac.uk>
3173 # Date: Wed, 26 Feb 2003 16:53:12 +0000
3174 # Message-Id: <E18o4nw-0008Ly-00@wisbech.cl.cam.ac.uk>
3175 # To: perl-unicode@perl.org
3176
3177 $x = "\x{2019}\nk"; $x =~ s/(\S)\n(\S)/$1 $2/sg;
3178 ok($x eq "\x{2019} k", "Markus Kuhn 2003-02-26");
3179
3180 $x = "b\nk"; $x =~ s/(\S)\n(\S)/$1 $2/sg;
3181 ok($x eq "b k", "Markus Kuhn 2003-02-26");
3182
3183 ok("\x{2019}" =~ /\S/, "Markus Kuhn 2003-02-26");
3184}
3185
91e09a61
AE
3186{
3187 my $i;
3188 ok('-1-3-5-' eq join('', split /((??{$i++}))/, '-1-3-5-'),
080c2dec
AE
3189 "[perl #21411] (??{ .. }) corrupts split's stack");
3190 split /(?{'WOW'})/, 'abc';
3191 ok('a|b|c' eq join ('|', @_),
3192 "[perl #21411] (?{ .. }) version of the above");
3193}
3194
3195{
36daacbf
DM
3196 # XXX DAPM 13-Apr-06. Recursive split is still broken. It's only luck it
3197 # hasn't been crashing. Disable this test until it is fixed properly.
3198 # XXX also check what it returns rather than just doing ok(1,...)
3199 # split /(?{ split "" })/, "abc";
080c2dec 3200 ok(1,'cache_re & "(?{": it dumps core in 5.6.1 & 5.8.0');
91e09a61
AE
3201}
3202
c517dc2b
JH
3203{
3204 ok("\x{100}\n" =~ /\x{100}\n$/, "UTF8 length cache and fbm_compile");
3205}
ec391688 3206
16e1b944
JH
3207{
3208 package Str;
3209 use overload q/""/ => sub { ${$_[0]}; };
3210 sub new { my ($c, $v) = @_; bless \$v, $c; }
3211
3212 package main;
3213 $_ = Str->new("a\x{100}/\x{100}b");
3214 ok(join(":", /\b(.)\x{100}/g) eq "a:/", "re_intuit_start and PL_bostr");
3215}
3216
7ef91622
IH
3217{
3218 $_ = "code: 'x' { '...' }\n"; study;
3219 my @x; push @x, $& while m/'[^\']*'/gx;
3220 ok(join(":", @x) eq "'x':'...'",
3221 "[perl #17757] Parse::RecDescent triggers infinite loop");
3222}
3223
14ebb1a2
JH
3224{
3225 my $re = qq/^([^X]*)X/;
3226 utf8::upgrade($re);
3227 ok("\x{100}X" =~ /$re/, "S_cl_and ANYOF_UNICODE & ANYOF_INVERTED");
3228}
3229
f02c194e
RGS
3230# bug #22354
3231sub func ($) {
3232 ok( "a\nb" !~ /^b/, $_[0] );
3233 ok( "a\nb" =~ /^b/m, "$_[0] - with /m" );
3234}
3235func "standalone";
3236$_ = "x"; s/x/func "in subst"/e;
3237$_ = "x"; s/x/func "in multiline subst"/em;
3238#$_ = "x"; /x(?{func "in regexp"})/;
3239#$_ = "x"; /x(?{func "in multiline regexp"})/m;
3240
c74340f9 3241# bug RT#19049
5aca4364
AG
3242$_="abcdef\n";
3243@x = m/./g;
c74340f9 3244ok("abcde" eq "$`", 'RT#19049 - global match not setting $`');
5aca4364 3245
6b43b216
DM
3246ok("123\x{100}" =~ /^.*1.*23\x{100}$/, 'uft8 + multiple floating substr');
3247
89d0f8e1 3248# LATIN SMALL/CAPITAL LETTER A WITH MACRON
c1e0e3d2
NIS
3249ok(" \x{101}" =~ qr/\x{100}/i,
3250 "<20030808193656.5109.1@llama.ni-s.u-net.com>");
3251
89d0f8e1 3252# LATIN SMALL/CAPITAL LETTER A WITH RING BELOW
5dab1207
NIS
3253ok(" \x{1E01}" =~ qr/\x{1E00}/i,
3254 "<20030808193656.5109.1@llama.ni-s.u-net.com>");
3255
89d0f8e1
JH
3256# DESERET SMALL/CAPITAL LETTER LONG I
3257ok(" \x{10428}" =~ qr/\x{10400}/i,
3258 "<20030808193656.5109.1@llama.ni-s.u-net.com>");
3259
3260# LATIN SMALL/CAPITAL LETTER A WITH RING BELOW + 'X'
3261ok(" \x{1E01}x" =~ qr/\x{1E00}X/i,
3262 "<20030808193656.5109.1@llama.ni-s.u-net.com>");
3263
a7913593
JH
3264{
3265 # [perl #23769] Unicode regex broken on simple example
3266 # regrepeat() didn't handle UTF-8 EXACT case right.
3267
3268 my $s = "\x{a0}\x{a0}\x{a0}\x{100}"; chop $s;
3269
3270 ok($s =~ /\x{a0}/, "[perl #23769]");
3271 ok($s =~ /\x{a0}+/, "[perl #23769]");
3272 ok($s =~ /\x{a0}\x{a0}/, "[perl #23769]");
090f7165
JH
3273
3274 ok("aaa\x{100}" =~ /(a+)/, "[perl #23769] easy invariant");
3275 ok($1 eq "aaa", "[perl #23769]");
3276
3277 ok("\xa0\xa0\xa0\x{100}" =~ /(\xa0+)/, "[perl #23769] regrepeat invariant");
3278 ok($1 eq "\xa0\xa0\xa0", "[perl #23769]");
3279
3280 ok("ababab\x{100} " =~ /((?:ab)+)/, "[perl #23769] hard invariant");
3281 ok($1 eq "ababab", "[perl #23769]");
3282
3283 ok("\xa0\xa1\xa0\xa1\xa0\xa1\x{100}" =~ /((?:\xa0\xa1)+)/, "[perl #23769] hard variant");
3284 ok($1 eq "\xa0\xa1\xa0\xa1\xa0\xa1", "[perl #23769]");
3285
3286 ok("aaa\x{100} " =~ /(a+?)/, "[perl #23769] easy invariant");
3287 ok($1 eq "a", "[perl #23769]");
3288
3289 ok("\xa0\xa0\xa0\x{100} " =~ /(\xa0+?)/, "[perl #23769] regrepeat variant");
3290 ok($1 eq "\xa0", "[perl #23769]");
3291
3292 ok("ababab\x{100} " =~ /((?:ab)+?)/, "[perl #23769] hard invariant");
3293 ok($1 eq "ab", "[perl #23769]");
3294
3295 ok("\xa0\xa1\xa0\xa1\xa0\xa1\x{100}" =~ /((?:\xa0\xa1)+?)/, "[perl #23769] hard variant");
3296 ok($1 eq "\xa0\xa1", "[perl #23769]");
3297
3298 ok("\xc4\xc4\xc4" !~ /(\x{100}+)/, "[perl #23769] don't match first byte of utf8 representation");
3299 ok("\xc4\xc4\xc4" !~ /(\x{100}+?)/, "[perl #23769] don't match first byte of utf8 representation");
a7913593
JH
3300}
3301
4be7a33f
HS
3302for (120 .. 130) {
3303 my $head = 'x' x $_;
3304 for my $tail ('\x{0061}', '\x{1234}') {
3305 ok(
3306 eval qq{ "$head$tail" =~ /$head$tail/ },
3307 '\x{...} misparsed in regexp near 127 char EXACT limit'
3308 );
3309 }
3310}
3311
289555fe
HS
3312# perl #25269: panic: pp_match start/end pointers
3313ok("a-bc" eq eval {
3314 my($x, $y) = "bca" =~ /^(?=.*(a)).*(bc)/;
3315 "$x-$y";
3316}, 'captures can move backwards in string');
3317
b6d5fef8
RGS
3318# perl #27940: \cA not recognized in character classes
3319ok("a\cAb" =~ /\cA/, '\cA in pattern');
3320ok("a\cAb" =~ /[\cA]/, '\cA in character class');
3321ok("a\cAb" =~ /[\cA-\cB]/, '\cA in character class range');
3322ok("abc" =~ /[^\cA-\cB]/, '\cA in negated character class range');
3323ok("a\cBb" =~ /[\cA-\cC]/, '\cB in character class range');
3324ok("a\cCbc" =~ /[^\cA-\cB]/, '\cC in negated character class range');
3325ok("a\cAb" =~ /(??{"\cA"})/, '\cA in ??{} pattern');
25f684f7 3326ok("ab" !~ /a\cIb/x, '\cI in pattern');
5dab1207 3327
0e788c72
HS
3328# perl #28532: optional zero-width match at end of string is ignored
3329ok(("abc" =~ /^abc(\z)?/) && defined($1),
3330 'optional zero-width match at end of string');
3331ok(("abc" =~ /^abc(\z)??/) && !defined($1),
3332 'optional zero-width match at end of string');
3333
a3621e74
YO
3334
3335
3336{ # TRIE related
3337 my @got=();
3338 "words"=~/(word|word|word)(?{push @got,$1})s$/;
3339 ok(@got==1,"TRIE optimation is working") or warn "# @got";
3340 @got=();
3341 "words"=~/(word|word|word)(?{push @got,$1})s$/i;
3342 ok(@got==1,"TRIEF optimisation is working") or warn "# @got";
3343
3344 my @nums=map {int rand 1000} 1..100;
3345 my $re="(".(join "|",@nums).")";
3346 $re=qr/\b$re\b/;
3347
3348 foreach (@nums) {
3349 ok($_=~/$re/,"Trie nums");
3350 }
3351 $_=join " ", @nums;
3352 @got=();
3353 push @got,$1 while /$re/g;
3354
3355 my %count;
3356 $count{$_}++ for @got;
3357 my $ok=1;
3358 for (@nums) {
3359 $ok=0 if --$count{$_}<0;
3360 }
3361 ok($ok,"Trie min count matches");
3362}
3363
3364
3365# TRIE related
3366# LATIN SMALL/CAPITAL LETTER A WITH MACRON
3367ok(("foba \x{101}foo" =~ qr/(foo|\x{100}foo|bar)/i) && $1 eq "\x{101}foo",
3368 "TRIEF + LATIN SMALL/CAPITAL LETTER A WITH MACRON");
3369
3370# LATIN SMALL/CAPITAL LETTER A WITH RING BELOW
3371ok(("foba \x{1E01}foo" =~ qr/(foo|\x{1E00}foo|bar)/i) && $1 eq "\x{1E01}foo",
3372 "TRIEF + LATIN SMALL/CAPITAL LETTER A WITH RING BELOW");
3373
3374# DESERET SMALL/CAPITAL LETTER LONG I
3375ok(("foba \x{10428}foo" =~ qr/(foo|\x{10400}foo|bar)/i) && $1 eq "\x{10428}foo",
3376 "TRIEF + DESERET SMALL/CAPITAL LETTER LONG I");
3377
3378# LATIN SMALL/CAPITAL LETTER A WITH RING BELOW + 'X'
3379ok(("foba \x{1E01}xfoo" =~ qr/(foo|\x{1E00}Xfoo|bar)/i) && $1 eq "\x{1E01}xfoo",
3380 "TRIEF + LATIN SMALL/CAPITAL LETTER A WITH RING BELOW + 'X'");
3381
3382{# TRIE related
3383
3384use charnames ':full';
3385
3386$s="\N{LATIN SMALL LETTER SHARP S}";
3387ok(("foba ba$s" =~ qr/(foo|Ba$s|bar)/i)
3388 && $1 eq "ba$s",
3389 "TRIEF + LATIN SMALL LETTER SHARP S =~ ss");
3390ok(("foba ba$s" =~ qr/(Ba$s|foo|bar)/i)
3391 && $1 eq "ba$s",
3392 "TRIEF + LATIN SMALL LETTER SHARP S =~ ss");
3393ok(("foba ba$s" =~ qr/(foo|bar|Ba$s)/i)
3394 && $1 eq "ba$s",
3395 "TRIEF + LATIN SMALL LETTER SHARP S =~ ss");
3396
3397ok(("foba ba$s" =~ qr/(foo|Bass|bar)/i)
3398 && $1 eq "ba$s",
3399 "TRIEF + LATIN SMALL LETTER SHARP S =~ ss");
3400
3401ok(("foba ba$s" =~ qr/(foo|BaSS|bar)/i)
3402 && $1 eq "ba$s",
3403 "TRIEF + LATIN SMALL LETTER SHARP S =~ SS");
786e8c11
YO
3404
3405ok(("foba ba${s}pxySS$s$s" =~ qr/(b(?:a${s}t|a${s}f|a${s}p)[xy]+$s*)/i)
3406 && $1 eq "ba${s}pxySS$s$s",
3407 "COMMON PREFIX TRIEF + LATIN SMALL LETTER SHARP S");
3408
3409
a3621e74
YO
3410}
3411
3412
e1d1eefb 3413print "# set PERL_SKIP_PSYCHO_TEST to skip this test\n";
fc8cd66c 3414if (!$ENV{PERL_SKIP_PSYCHO_TEST}){
a3621e74
YO
3415 my @normal=qw(these are some normal words);
3416 my $psycho=join "|",@normal,map chr $_,255..20000;
3417 ok(('these'=~/($psycho)/) && $1 eq 'these','Pyscho');
fc8cd66c
YO
3418} else {
3419 ok(1,'Skipped Psycho');
a3621e74 3420}
809e8e66
DM
3421
3422# [perl #36207] mixed utf8 / latin-1 and case folding
3423
3424{
108bb1ad
YO
3425 my $utf8 = "\xe9\x{100}"; chop $utf8;
3426 my $latin1 = "\xe9";
3427
3428 ok($utf8 =~ /\xe9/i, "utf8/latin");
3429 ok($utf8 =~ /$latin1/i, "utf8/latin runtime");
3430 ok($utf8 =~ /(abc|\xe9)/i, "utf8/latin trie");
3431 ok($utf8 =~ /(abc|$latin1)/i, "utf8/latin trie runtime");
3432
a0a388a1 3433 ok("\xe9" =~ /$utf8/i, "# latin/utf8");
108bb1ad 3434 ok("\xe9" =~ /(abc|$utf8)/i, "# latin/utf8 trie");
a0a388a1 3435 ok($latin1 =~ /$utf8/i, "# latin/utf8 runtime");
108bb1ad 3436 ok($latin1 =~ /(abc|$utf8)/i, "# latin/utf8 trie runtime");
809e8e66
DM
3437}
3438
740266bf
NC
3439# [perl #37038] Global regular matches generate invalid pointers
3440
3441{
3442 my $s = "abcd";
3443 $s =~ /(..)(..)/g;
3444 $s = $1;
3445 $s = $2;
3446 ok($s eq 'cd',
4f5010f2 3447 "# assigning to original string should not corrupt match vars");
740266bf
NC
3448}
3449
0523e772
NC
3450{
3451 package wooosh;
3452 sub gloople {
3453 "!";
3454 }
3455 package main;
3456
3457 my $aeek = bless {}, 'wooosh';
3458 eval {$aeek->gloople() =~ /(.)/g;};
547ba77e 3459 ok($@ eq "", "//g match against return value of sub") or print "# $@\n";
0523e772 3460}
740266bf 3461
86f12da2
NC
3462{
3463 sub gloople {
3464 "!";
3465 }
3466 eval {gloople() =~ /(.)/g;};
3467 ok($@ eq "", "# 26410 didn't affect sub calls for some reason")
3468 or print "# $@\n";
3469}
3470
3471{
03949bff 3472 local $TODO = "See changes 26925-26928, which reverted change 26410";
86f12da2
NC
3473 package lv;
3474 $var = "abc";
3475 sub variable : lvalue { $var }
3476
3477 package main;
3478 my $o = bless [], "lv";
3479 my $f = "";
3480 eval { for (1..2) { $f .= $1 if $o->variable =~ /(.)/g } };
03949bff 3481 ok($f eq "ab", "pos retained between calls") or print "# $@\n";
86f12da2
NC
3482}
3483
3484{
03949bff 3485 local $TODO = "See changes 26925-26928, which reverted change 26410";
86f12da2
NC
3486 $var = "abc";
3487 sub variable : lvalue { $var }
3488
3489 my $f = "";
3490 eval { for (1..2) { $f .= $1 if variable() =~ /(.)/g } };
03949bff 3491 ok($f eq "ab", "pos retained between calls") or print "# $@\n";
86f12da2
NC
3492}
3493
9f7f3913
ST
3494# [perl #37836] Simple Regex causes SEGV when run on specific data
3495if ($ordA == 193) {
3496 print "ok $test # Skip: in EBCDIC\n"; $test++;
3497} else {
3498 no warnings 'utf8';
3499 $_ = pack('U0C2', 0xa2, 0xf8); # ill-formed UTF-8
3500 my $ret = 0;
3501 eval { $ret = s/[\0]+//g };
3502 ok($ret == 0, "ill-formed UTF-8 doesn't match NUL in class");
3503}
3504
3505{ # [perl #38293] chr(65535) should be allowed in regexes
3506 no warnings 'utf8'; # to allow non-characters
3507 my($c, $r, $s);
3508
3509 $c = chr 0xffff;
3510 $c =~ s/$c//g;
3511 ok($c eq "", "U+FFFF, parsed as atom");
3512
3513 $c = chr 0xffff;
3514 $r = "\\$c";
3515 $c =~ s/$r//g;
3516 ok($c eq "", "U+FFFF backslashed, parsed as atom");
3517
3518 $c = chr 0xffff;
3519 $c =~ s/[$c]//g;
3520 ok($c eq "", "U+FFFF, parsed in class");
3521
3522 $c = chr 0xffff;
3523 $r = "[\\$c]";
3524 $c =~ s/$r//g;
3525 ok($c eq "", "U+FFFF backslashed, parsed in class");
3526
3527 $s = "A\x{ffff}B";
3528 $s =~ s/\x{ffff}//i;
3529 ok($s eq "AB", "U+FFFF, EXACTF");
3530
3531 $s = "\x{ffff}A";
3532 $s =~ s/\bA//;
3533 ok($s eq "\x{ffff}", "U+FFFF, BOUND");
3534
3535 $s = "\x{ffff}!";
3536 $s =~ s/\B!//;
3537 ok($s eq "\x{ffff}", "U+FFFF, NBOUND");
3538} # non-characters end
3539
a30987ac
YO
3540{
3541 # https://rt.perl.org/rt3/Ticket/Display.html?id=39583
3542
3543 # The printing characters
3544 my @chars = ("A".."Z");
3545 my $delim = ",";
3546 my $size = 32771 - 4;
fc8cd66c 3547 my $str = '';
a30987ac
YO
3548
3549 # create some random junk. Inefficient, but it works.
3550 for ($i = 0 ; $i < $size ; $i++) {
fc8cd66c 3551 $str .= $chars[int(rand(@chars))];
a30987ac
YO
3552 }
3553
fc8cd66c 3554 $str .= ($delim x 4);
a30987ac
YO
3555 my $res;
3556 my $matched;
fc8cd66c 3557 if ($str =~ s/^(.*?)${delim}{4}//s) {
a30987ac
YO
3558 $res = $1;
3559 $matched=1;
3560 }
3561 ok($matched,'pattern matches');
fc8cd66c 3562 ok(length($str)==0,"Empty string");
a30987ac
YO
3563 ok(defined($res) && length($res)==$size,"\$1 is correct size");
3564}
9f7f3913 3565
1749ea0d
ST
3566{ # related to [perl #27940]
3567 ok("\0-A" =~ /\c@-A/, '@- should not be interpolated in a pattern');
3568 ok("\0\0A" =~ /\c@+A/, '@+ should not be interpolated in a pattern');
3569 ok("X\@-A" =~ /X@-A/, '@- should not be interpolated in a pattern');
3570 ok("X\@\@A" =~ /X@+A/, '@+ should not be interpolated in a pattern');
3571
3572 ok("X\0A" =~ /X\c@?A/, '\c@?');
3573 ok("X\0A" =~ /X\c@*A/, '\c@*');
3574 ok("X\0A" =~ /X\c@(A)/, '\c@(');
3575 ok("X\0A" =~ /X(\c@)A/, '\c@)');
3576 ok("X\0A" =~ /X\c@|ZA/, '\c@|');
3577
3578 ok("X\@A" =~ /X@?A/, '@?');
3579 ok("X\@A" =~ /X@*A/, '@*');
3580 ok("X\@A" =~ /X@(A)/, '@(');
3581 ok("X\@A" =~ /X(@)A/, '@)');
3582 ok("X\@A" =~ /X@|ZA/, '@|');
3583
3584 local $" = ','; # non-whitespace and non-RE-specific
3585 ok('abc' =~ /(.)(.)(.)/, 'the last successful match is bogus');
3586 ok("A@+B" =~ /A@{+}B/, 'interpolation of @+ in /@{+}/');
3587 ok("A@-B" =~ /A@{-}B/, 'interpolation of @- in /@{-}/');
3588 ok("A@+B" =~ /A@{+}B/x, 'interpolation of @+ in /@{+}/x');
3589 ok("A@-B" =~ /A@{-}B/x, 'interpolation of @- in /@{-}/x');
3590}
3591
fc8cd66c
YO
3592{
3593 use lib 'lib';
3594 use Cname;
3595
3596 ok('fooB'=~/\N{foo}[\N{B}\N{b}]/,"Passthrough charname");
3597 $test=1233; my $handle=make_must_warn('Ignoring excess chars from');
3598 $handle->('q(xxWxx) =~ /[\N{WARN}]/');
3599 {
3600 my $code;
3601 my $w="";
3602 local $SIG{__WARN__} = sub { $w.=shift };
3603 eval($code=<<'EOFTEST') or die "$@\n$code\n";
3604 {
3605 use warnings;
3606
3607 #1234
3608 ok("\0" !~ /[\N{EMPTY-STR}XY]/,
3609 "Zerolength charname in charclass doesnt match \0");
3610 1;
3611 }
3612EOFTEST
3613 ok($w=~/Ignoring zero length/,
3614 "Got expected zero length warning");
3615 warn $code;
3616
3617 }
3618 $handle= make_must_warn('Ignoring zero length');
3619 $handle->('qq(\\0) =~ /[\N{EMPTY-STR}XY]/');
3620 ok('AB'=~/(\N{EVIL})/ && $1 eq 'A',"Charname caching $1");
3621 ok('ABC'=~/(\N{EVIL})/,"Charname caching $1");
3622 ok('xy'=~/x\N{EMPTY-STR}y/, 'Empty string charname produces NOTHING node');
3623 ok(''=~/\N{EMPTY-STR}/, 'Empty string charname produces NOTHING node 2');
3624
3625}
3626{
3627 print "# MORE LATIN SMALL LETTER SHARP S\n";
3628
3629 use charnames ':full';
3630
3631 #see also test #835
3632 ok("ss" =~ /[\N{LATIN SMALL LETTER SHARP S}x]/i,
3633 "unoptimized named sequence in class 1");
3634 ok("SS" =~ /[\N{LATIN SMALL LETTER SHARP S}x]/i,
3635 "unoptimized named sequence in class 2");
3636 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /[\N{LATIN SMALL LETTER SHARP S}x]/,
3637 "unoptimized named sequence in class 3");
3638 ok("\N{LATIN SMALL LETTER SHARP S}" =~ /[\N{LATIN SMALL LETTER SHARP S}x]/i,
3639 "unoptimized named sequence in class 4");
3640
3641 ok('aabc' !~ /a\N{PLUS SIGN}b/,'/a\N{PLUS SIGN}b/ against aabc');
3642 ok('a+bc' =~ /a\N{PLUS SIGN}b/,'/a\N{PLUS SIGN}b/ against a+bc');
3643 ok('a+bc' =~ /a\N{PLUS SIGN}b/,'/a\N{PLUS SIGN}b/ against a+bc');
3644
3645 ok(' A B'=~/\N{SPACE}\N{U+0041}\N{SPACE}\N{U+0042}/,
3646 'Intermixed named and unicode escapes 1');
3647 ok("\N{SPACE}\N{U+0041}\N{SPACE}\N{U+0042}"=~
3648 /\N{SPACE}\N{U+0041}\N{SPACE}\N{U+0042}/,
3649 'Intermixed named and unicode escapes 2');
3650 ok("\N{SPACE}\N{U+0041}\N{SPACE}\N{U+0042} 3"=~
3651 /[\N{SPACE}\N{U+0041}][\N{SPACE}\N{U+0042}]/,
3652 'Intermixed named and unicode escapes');
3653}
faec1544
DM
3654$brackets = qr{
3655 { (?> [^{}]+ | (??{ $brackets }) )* }
3656 }x;
3657ok("{b{c}d" !~ m/^((??{ $brackets }))/, "bracket mismatch");
3658
6bda09f9
YO
3659SKIP:{
3660 our @stack=();
3661 my @expect=qw(
3662 stuff1
3663 stuff2
3664 <stuff1>and<stuff2>
3665 right
3666 <right>
3667 <<right>>
3668 <<<right>>>
3669 <<stuff1>and<stuff2>><<<<right>>>>
3670 );
3671
3672 local $_='<<<stuff1>and<stuff2>><<<<right>>>>>';
3673 ok(/^(<((?:(?>[^<>]+)|(?1))*)>(?{push @stack, $2 }))$/,
3674 "Recursion should match");
3675 ok(@stack==@expect)
3676 or skip("Won't test individual results as count isn't equal",
3677 0+@expect);
3678 foreach my $idx (@expect) {
3679 ok($expect[$idx] eq $stack[$idx],
3680 "Expecting '$expect' at stack pos #$idx");
3681 }
3682
3683}
81714fb9
YO
3684{
3685 my $s='123453456';
3686 $s=~s/(?<digits>\d+)\k<digits>/$+{digits}/;
3687 ok($s eq '123456','Named capture (angle brackets) s///');
3688 $s='123453456';
3689 $s=~s/(?'digits'\d+)\k'digits'/$+{digits}/;
3690 ok($s eq '123456','Named capture (single quotes) s///');
3691}
1f1031fe
YO
3692
3693{
1b0dc72f
ST
3694 my @ary = (
3695 pack('U', 0x00F1), # n-tilde
3696 '_'.pack('U', 0x00F1), # _ + n-tilde
3697 'c'.pack('U', 0x0327), # c + cedilla
3698 pack('U*', 0x00F1, 0x0327), # n-tilde + cedilla
3699 'a'.pack('U', 0x00B2), # a + superscript two
3700 pack('U', 0x0391), # ALPHA
3701 pack('U', 0x0391).'2', # ALPHA + 2
3702 pack('U', 0x0391).'_', # ALPHA + _
3703 );
3704 for my $uni (@ary) {
3705 my ($r1, $c1, $r2, $c2) = eval qq{
3706 use utf8;
3707 scalar("..foo foo.." =~ /(?'${uni}'foo) \\k'${uni}'/),
3708 \$+{${uni}},
3709 scalar("..bar bar.." =~ /(?<${uni}>bar) \\k<${uni}>/),
3710 \$+{${uni}};
3711 };
3712 ok($r1, "Named capture UTF (?'')");
3713 ok(defined $c1 && $c1 eq 'foo', "Named capture UTF \%+");
3714 ok($r2, "Named capture UTF (?<>)");
3715 ok(defined $c2 && $c2 eq 'bar', "Named capture UTF \%+");
1f1031fe
YO
3716 }
3717}
3718
e62cc96a
YO
3719sub iseq($$;$) {
3720 my ( $got, $expect, $name)=@_;
03949bff 3721 my $todo = $TODO ? " # TODO $TODO" : '';
e62cc96a
YO
3722
3723 $_=defined($_) ? "'$_'" : "undef"
3724 for $got, $expect;
3725
3726 my $ok= $got eq $expect;
3727
03949bff 3728 printf "%sok %d - %s$todo\n", ($ok ? "" : "not "), $test,
de734bd5 3729 ($name||$Message)."\tLine ".((caller)[2]);
e62cc96a 3730
320adbb0 3731 no warnings 'utf8';
e62cc96a
YO
3732 printf "# Failed test at line %d\n".
3733 "# expected: %s\n".
3734 "# result: %s\n",
3735 (caller)[2], $expect, $got
3736 unless $ok;
3737
3738 $test++;
3739 return $ok;
3740}
81714fb9
YO
3741{
3742 my $s='foo bar baz';
e62cc96a
YO
3743 my (@k,@v,@fetch,$res);
3744 my $count= 0;
3745 my @names=qw($+{A} $+{B} $+{C});
81714fb9
YO
3746 if ($s=~/(?<A>foo)\s+(?<B>bar)?\s+(?<C>baz)/) {
3747 while (my ($k,$v)=each(%+)) {
3748 $count++;
3749 }
3750 @k=sort keys(%+);
3751 @v=sort values(%+);
e62cc96a
YO
3752 $res=1;
3753 push @fetch,
3754 [ "$+{A}", "$1" ],
3755 [ "$+{B}", "$2" ],
3756 [ "$+{C}", "$3" ],
3757 ;
3758 }
3759 foreach (0..2) {
3760 if ($fetch[$_]) {
3761 iseq($fetch[$_][0],$fetch[$_][1],$names[$_]);
3762 } else {
3763 ok(0, $names[$_]);
3764 }
81714fb9 3765 }
e62cc96a
YO
3766 iseq($res,1,"$s~=/(?<A>foo)\s+(?<B>bar)?\s+(?<C>baz)/");
3767 iseq($count,3,"Got 3 keys in %+ via each");
3768 iseq(0+@k, 3, 'Got 3 keys in %+ via keys');
3769 iseq("@k","A B C", "Got expected keys");
3770 iseq("@v","bar baz foo", "Got expected values");
0a4db386
YO
3771 eval'
3772 print for $+{this_key_doesnt_exist};
3773 ';
3774 ok(!$@,'lvalue $+{...} should not throw an exception');
81714fb9 3775}
44a2ac75 3776{
66072c92
A
3777 #
3778 # Almost the same as the block above, except that the capture is nested.
3779 #
3780 my $s = 'foo bar baz';
3781 my (@k,@v,@fetch,$res);
3782 my $count = 0;
3783 my @names = qw($+{A} $+{B} $+{C} $+{D});
ac29ea61 3784 if ($s =~ /(?<D>(?<A>foo)\s+(?<B>bar)?\s+(?<C>baz))/) {
66072c92
A
3785 while (my ($k,$v) = each(%+)) {
3786 $count++;
3787 }
3788 @k = sort keys(%+);
3789 @v = sort values(%+);
3790 $res = 1;
3791 push @fetch,
3792 [ "$+{A}", "$2" ],
3793 [ "$+{B}", "$3" ],
3794 [ "$+{C}", "$4" ],
3795 [ "$+{D}", $1 ],
3796 ;
3797 }
3798 foreach (0..3) {
3799 if ($fetch[$_]) {
66072c92
A
3800 iseq($fetch[$_][0],$fetch[$_][1],$names[$_]);
3801 } else {
3802 ok(0, $names[$_]);
3803 }
3804 }
3805 iseq($res,1,"$s~=/(?<D>(?<A>foo)\s+(?<B>bar)?\s+(?<C>baz))/");
250257bb
YO
3806 iseq($count,4,"Got 4 keys in %+ via each -- bug 50496");
3807 iseq(0+@k, 4, 'Got 4 keys in %+ via keys -- bug 50496');
3808 iseq("@k","A B C D", "Got expected keys -- bug 50496");
3809 iseq("@v","bar baz foo foo bar baz", "Got expected values -- bug = 50496");
66072c92
A
3810 eval'
3811 print for $+{this_key_doesnt_exist};
3812 ';
3813 ok(!$@,'lvalue $+{...} should not throw an exception');
3814}
3815{
44a2ac75
YO
3816 my $s='foo bar baz';
3817 my @res;
3818 if ('1234'=~/(?<A>1)(?<B>2)(?<A>3)(?<B>4)/) {
3819 foreach my $name (sort keys(%-)) {
3820 my $ary = $-{$name};
3821 foreach my $idx (0..$#$ary) {
3822 push @res,"$name:$idx:$ary->[$idx]";
3823 }
3824 }
3825 }
3826 my @expect=qw(A:0:1 A:1:3 B:0:2 B:1:4);
3827 iseq("@res","@expect","Check %-");
3828 eval'
3829 print for $-{this_key_doesnt_exist};
3830 ';
3831 ok(!$@,'lvalue $-{...} should not throw an exception');
3832}
bfac009d
DM
3833# stress test CURLYX/WHILEM.
3834#
3835# This test includes varying levels of nesting, and according to
3836# profiling done against build 28905, exercises every code line in the
3837# CURLYX and WHILEM blocks, except those related to LONGJMP, the
3838# super-linear cache and warnings. It executes about 0.5M regexes
3839
24b23f37
YO
3840if ($ENV{PERL_SKIP_PSYCHO_TEST}){
3841 printf "ok %d Skip: No psycho tests\n", $test++;
3842} else {
e1d1eefb 3843 print "# set PERL_SKIP_PSYCHO_TEST to skip this test\n";
bfac009d
DM
3844 my $r = qr/^
3845 (?:
3846 ( (?:a|z+)+ )
3847 (?:
3848 ( (?:b|z+){3,}? )
3849 (
3850 (?:
c476f425
DM
3851 (?:
3852 (?:c|z+){1,1}?z
3853 )?
bfac009d
DM
3854 (?:c|z+){1,1}
3855 )*
3856 )
3857 (?:z*){2,}
3858 ( (?:z+|d)+ )
3859 (?:
3860 ( (?:e|z+)+ )
3861 )*
3862 ( (?:f|z+)+ )
3863 )*
3864 ( (?:z+|g)+ )
3865 (?:
3866 ( (?:h|z+)+ )
3867 )*
3868 ( (?:i|z+)+ )
3869 )+
3870 ( (?:j|z+)+ )
3871 (?:
3872 ( (?:k|z+)+ )
3873 )*
3874 ( (?:l|z+)+ )
3875 $/x;
3876
3877
3878 my $ok = 1;
3879 my $msg = "CURLYX stress test";
3880 OUTER:
3881 for my $a ("x","a","aa") {
3882 for my $b ("x","bbb","bbbb") {
3883 my $bs = $a.$b;
3884 for my $c ("x","c","cc") {
3885 my $cs = $bs.$c;
3886 for my $d ("x","d","dd") {
3887 my $ds = $cs.$d;
3888 for my $e ("x","e","ee") {
3889 my $es = $ds.$e;
3890 for my $f ("x","f","ff") {
3891 my $fs = $es.$f;
3892 for my $g ("x","g","gg") {
3893 my $gs = $fs.$g;
3894 for my $h ("x","h","hh") {
3895 my $hs = $gs.$h;
3896 for my $i ("x","i","ii") {
3897 my $is = $hs.$i;
3898 for my $j ("x","j","jj") {
3899 my $js = $is.$j;
3900 for my $k ("x","k","kk") {
3901 my $ks = $js.$k;
3902 for my $l ("x","l","ll") {
3903 my $ls = $ks.$l;
3904 if ($ls =~ $r) {
3905 if ($ls =~ /x/) {
3906 $msg .= ": unexpected match for [$ls]";
3907 $ok = 0;
3908 last OUTER;
3909 }
3910 my $cap = "$1$2$3$4$5$6$7$8$9$10$11$12";
3911 unless ($ls eq $cap) {
3912 $msg .= ": capture: [$ls], got [$cap]";
3913 $ok = 0;
3914 last OUTER;
3915 }
3916 }
3917 else {
3918 unless ($ls =~ /x/) {
3919 $msg = ": failed for [$ls]";
3920 $ok = 0;
3921 last OUTER;
3922 }
3923 }
3924 }
3925 }
3926 }
3927 }
3928 }
3929 }
3930 }
3931 }
3932 }
3933 }
3934 }
3935 }
3936 ok($ok, $msg);
3937}
3938
cc74c5bd
ST
3939# \, breaks {3,4}
3940ok("xaaay" !~ /xa{3\,4}y/, "\, in a pattern");
3941ok("xa{3,4}y" =~ /xa{3\,4}y/, "\, in a pattern");
3942
3943# \c\ followed by _
3944ok("x\c_y" !~ /x\c\_y/, "\_ in a pattern");
3945ok("x\c\_y" =~ /x\c\_y/, "\_ in a pattern");
3946
3947# \c\ followed by other characters
3948for my $c ("z", "\0", "!", chr(254), chr(256)) {
3949 my $targ = "a\034$c";
3950 my $reg = "a\\c\\$c";
3951 ok(eval("qq/$targ/ =~ /$reg/"), "\\c\\ in pattern");
3952}
bfac009d 3953
581d1b5b
RGS
3954{
3955 my $str='abc';
3956 my $count=0;
3957 my $mval=0;
3958 my $pval=0;
3959 while ($str=~/b/g) { $mval=$#-; $pval=$#+; $count++ }
3960 iseq($mval,0,"\@- should be empty [RT#36046]");
3961 iseq($pval,0,"\@+ should be empty [RT#36046]");
3962 iseq($count,1,"should have matched once only [RT#36046]");
3963}
3964
5d458dd8 3965{ # Test the (*PRUNE) pattern
24b23f37 3966 our $count = 0;
e2e6a0f1 3967 'aaab'=~/a+b?(?{$count++})(*FAIL)/;
5d458dd8 3968 iseq($count,9,"expect 9 for no (*PRUNE)");
24b23f37 3969 $count = 0;
5d458dd8
YO
3970 'aaab'=~/a+b?(*PRUNE)(?{$count++})(*FAIL)/;
3971 iseq($count,3,"expect 3 with (*PRUNE)");
24b23f37
YO
3972 local $_='aaab';
3973 $count=0;
5d458dd8
YO
3974 1 while /.(*PRUNE)(?{$count++})(*FAIL)/g;
3975 iseq($count,4,"/.(*PRUNE)/");
24b23f37 3976 $count = 0;
5d458dd8
YO
3977 'aaab'=~/a+b?(??{'(*PRUNE)'})(?{$count++})(*FAIL)/;
3978 iseq($count,3,"expect 3 with (*PRUNE)");
24b23f37
YO
3979 local $_='aaab';
3980 $count=0;
5d458dd8
YO
3981 1 while /.(??{'(*PRUNE)'})(?{$count++})(*FAIL)/g;
3982 iseq($count,4,"/.(*PRUNE)/");
24b23f37 3983}
5d458dd8 3984{ # Test the (*SKIP) pattern
24b23f37 3985 our $count = 0;
5d458dd8
YO
3986 'aaab'=~/a+b?(*SKIP)(?{$count++})(*FAIL)/;
3987 iseq($count,1,"expect 1 with (*SKIP)");
24b23f37
YO
3988 local $_='aaab';
3989 $count=0;
5d458dd8
YO
3990 1 while /.(*SKIP)(?{$count++})(*FAIL)/g;
3991 iseq($count,4,"/.(*SKIP)/");
24b23f37
YO
3992 $_='aaabaaab';
3993 $count=0;
3994 our @res=();
5d458dd8
YO
3995 1 while /(a+b?)(*SKIP)(?{$count++; push @res,$1})(*FAIL)/g;
3996 iseq($count,2,"Expect 2 with (*SKIP)" );
3997 iseq("@res","aaab aaab","adjacent (*SKIP) works as expected" );
24b23f37 3998}
5d458dd8 3999{ # Test the (*SKIP) pattern
24b23f37 4000 our $count = 0;
5d458dd8
YO
4001 'aaab'=~/a+b?(*MARK:foo)(*SKIP)(?{$count++})(*FAIL)/;
4002 iseq($count,1,"expect 1 with (*SKIP)");
24b23f37
YO
4003 local $_='aaab';
4004 $count=0;
5d458dd8
YO
4005 1 while /.(*MARK:foo)(*SKIP)(?{$count++})(*FAIL)/g;
4006 iseq($count,4,"/.(*SKIP)/");
24b23f37
YO
4007 $_='aaabaaab';
4008 $count=0;
4009 our @res=();
5d458dd8
YO
4010 1 while /(a+b?)(*MARK:foo)(*SKIP)(?{$count++; push @res,$1})(*FAIL)/g;
4011 iseq($count,2,"Expect 2 with (*SKIP)" );
4012 iseq("@res","aaab aaab","adjacent (*SKIP) works as expected" );
e2e6a0f1 4013}
5d458dd8 4014{ # Test the (*SKIP) pattern
e2e6a0f1 4015 our $count = 0;
5d458dd8
YO
4016 'aaab'=~/a*(*MARK:a)b?(*MARK:b)(*SKIP:a)(?{$count++})(*FAIL)/;
4017 iseq($count,3,"expect 3 with *MARK:a)b?(*MARK:b)(*SKIP:a)");
e2e6a0f1
YO
4018 local $_='aaabaaab';
4019 $count=0;
4020 our @res=();
5d458dd8
YO
4021 1 while /(a*(*MARK:a)b?)(*MARK:x)(*SKIP:a)(?{$count++; push @res,$1})(*FAIL)/g;
4022 iseq($count,5,"Expect 5 with (*MARK:a)b?)(*MARK:x)(*SKIP:a)" );
4023 iseq("@res","aaab b aaab b ","adjacent (*MARK:a)b?)(*MARK:x)(*SKIP:a) works as expected" );
e2e6a0f1
YO
4024}
4025{ # Test the (*COMMIT) pattern
4026 our $count = 0;
4027 'aaabaaab'=~/a+b?(*COMMIT)(?{$count++})(*FAIL)/;
4028 iseq($count,1,"expect 1 with (*COMMIT)");
4029 local $_='aaab';
4030 $count=0;
4031 1 while /.(*COMMIT)(?{$count++})(*FAIL)/g;
4032 iseq($count,1,"/.(*COMMIT)/");
4033 $_='aaabaaab';
4034 $count=0;
4035 our @res=();
4036 1 while /(a+b?)(*COMMIT)(?{$count++; push @res,$1})(*FAIL)/g;
4037 iseq($count,1,"Expect 1 with (*COMMIT)" );
4038 iseq("@res","aaab","adjacent (*COMMIT) works as expected" );
4039}
4040{
4041 # Test named commits and the $REGERROR var
4042 our $REGERROR;
4043 for my $name ('',':foo')
4044 {
5d458dd8
YO
4045 for my $pat ("(*PRUNE$name)",
4046 ($name? "(*MARK$name)" : "")
4047 . "(*SKIP$name)",
4048 "(*COMMIT$name)")
e2e6a0f1
YO
4049 {
4050 for my $suffix ('(*FAIL)','')
4051 {
4052 'aaaab'=~/a+b$pat$suffix/;
4053 iseq(
4054 $REGERROR,
4055 ($suffix ? ($name ? 'foo' : "1") : ""),
4056 "Test $pat and \$REGERROR $suffix"
4057 );
4058 }
4059 }
4060 }
4061}
4062{
4063 # Test named commits and the $REGERROR var
4064 package Fnorble;
4065 our $REGERROR;
4066 for my $name ('',':foo')
4067 {
5d458dd8
YO
4068 for my $pat ("(*PRUNE$name)",
4069 ($name? "(*MARK$name)" : "")
4070 . "(*SKIP$name)",
4071 "(*COMMIT$name)")
e2e6a0f1
YO
4072 {
4073 for my $suffix ('(*FAIL)','')
4074 {
4075 'aaaab'=~/a+b$pat$suffix/;
4076 ::iseq(
4077 $REGERROR,
4078 ($suffix ? ($name ? 'foo' : "1") : ""),
4079 "Test $pat and \$REGERROR $suffix"
4080 );
4081 }
4082 }
4083 }
4084}
4085{
4086 # Test named commits and the $REGERROR var
f0852a51 4087 local $Message = "\$REGERROR";
e2e6a0f1
YO
4088 our $REGERROR;
4089 for $word (qw(bar baz bop)) {
4090 $REGERROR="";
4091 "aaaaa$word"=~/a+(?:bar(*COMMIT:bar)|baz(*COMMIT:baz)|bop(*COMMIT:bop))(*FAIL)/;
4092 iseq($REGERROR,$word);
4093 }
4094}
4095{ #Regression test for perlbug 40684
f0852a51 4096 local $Message = "RT#40684 tests:";
e2e6a0f1
YO
4097 my $s = "abc\ndef";
4098 my $rex = qr'^abc$'m;
4099 ok($s =~ m/$rex/);
4100 ok($s =~ m/^abc$/m);
24b23f37 4101}
5d458dd8
YO
4102{
4103 #Mindnumbingly simple test of (*THEN)
4104 for ("ABC","BAX") {
4105 ok(/A (*THEN) X | B (*THEN) C/x,"Simple (*THEN) test");
4106 }
542fa716
YO
4107}
4108
4109{
f0852a51 4110 local $Message = "Relative Recursion";
542fa716
YO
4111 my $parens=qr/(\((?:[^()]++|(?-1))*+\))/;
4112 local $_='foo((2*3)+4-3) + bar(2*(3+4)-1*(2-3))';
4113 my ($all,$one,$two)=('','','');
4114 if (/foo $parens \s* \+ \s* bar $parens/x) {
4115 $all=$&;
4116 $one=$1;
4117 $two=$2;
4118 }
4119 iseq($one, '((2*3)+4-3)');
4120 iseq($two, '(2*(3+4)-1*(2-3))');
4121 iseq($all, 'foo((2*3)+4-3) + bar(2*(3+4)-1*(2-3))');
4122 iseq($all, $_);
4123}
b515a41d
YO
4124{
4125 my $spaces=" ";
4126 local $_=join 'bar',$spaces,$spaces;
4127 our $count=0;
4128 s/(?>\s+bar)(?{$count++})//g;
4129 iseq($_,$spaces,"SUSPEND final string");
4130 iseq($count,1,"Optimiser should have prevented more than one match");
4131}
f0852a51
YO
4132{
4133 local $Message = "RT#36909 test";
4134 $^R = 'Nothing