This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Worrying about insecure directories now is a bit too late.
[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$| = 1;
3568d838 8
f79b3095 9print "1..639\n";
8d37f932 10
e4d48cc9
GS
11BEGIN {
12 chdir 't' if -d 't';
20822f61 13 @INC = '../lib';
e4d48cc9 14}
ffbc6a93 15
8d37f932 16eval 'use Config'; # Defaults assumed if this fails
8d063cd8
LW
17
18$x = "abc\ndef\n";
19
20if ($x =~ /^abc/) {print "ok 1\n";} else {print "not ok 1\n";}
21if ($x !~ /^def/) {print "ok 2\n";} else {print "not ok 2\n";}
22
23$* = 1;
24if ($x =~ /^def/) {print "ok 3\n";} else {print "not ok 3\n";}
25$* = 0;
26
27$_ = '123';
28if (/^([0-9][0-9]*)/) {print "ok 4\n";} else {print "not ok 4\n";}
29
30if ($x =~ /^xxx/) {print "not ok 5\n";} else {print "ok 5\n";}
31if ($x !~ /^abc/) {print "not ok 6\n";} else {print "ok 6\n";}
32
33if ($x =~ /def/) {print "ok 7\n";} else {print "not ok 7\n";}
34if ($x !~ /def/) {print "not ok 8\n";} else {print "ok 8\n";}
35
36if ($x !~ /.def/) {print "ok 9\n";} else {print "not ok 9\n";}
37if ($x =~ /.def/) {print "not ok 10\n";} else {print "ok 10\n";}
38
39if ($x =~ /\ndef/) {print "ok 11\n";} else {print "not ok 11\n";}
40if ($x !~ /\ndef/) {print "not ok 12\n";} else {print "ok 12\n";}
41
42$_ = 'aaabbbccc';
43if (/(a*b*)(c*)/ && $1 eq 'aaabbb' && $2 eq 'ccc') {
44 print "ok 13\n";
45} else {
46 print "not ok 13\n";
47}
48if (/(a+b+c+)/ && $1 eq 'aaabbbccc') {
49 print "ok 14\n";
50} else {
51 print "not ok 14\n";
52}
53
54if (/a+b?c+/) {print "not ok 15\n";} else {print "ok 15\n";}
55
56$_ = 'aaabccc';
57if (/a+b?c+/) {print "ok 16\n";} else {print "not ok 16\n";}
58if (/a*b+c*/) {print "ok 17\n";} else {print "not ok 17\n";}
59
60$_ = 'aaaccc';
61if (/a*b?c*/) {print "ok 18\n";} else {print "not ok 18\n";}
62if (/a*b+c*/) {print "not ok 19\n";} else {print "ok 19\n";}
63
64$_ = 'abcdef';
65if (/bcd|xyz/) {print "ok 20\n";} else {print "not ok 20\n";}
66if (/xyz|bcd/) {print "ok 21\n";} else {print "not ok 21\n";}
67
68if (m|bc/*d|) {print "ok 22\n";} else {print "not ok 22\n";}
378cc40b
LW
69
70if (/^$_$/) {print "ok 23\n";} else {print "not ok 23\n";}
71
72$* = 1; # test 3 only tested the optimized version--this one is for real
73if ("ab\ncd\n" =~ /^cd/) {print "ok 24\n";} else {print "not ok 24\n";}
74$* = 0;
75
cb55de95
JH
76$XXX{123} = 123;
77$XXX{234} = 234;
78$XXX{345} = 345;
79
80@XXX = ('ok 25','not ok 25', 'ok 26','not ok 26','not ok 27');
81while ($_ = shift(@XXX)) {
82 ?(.*)? && (print $1,"\n");
83 /not/ && reset;
84 /not ok 26/ && reset 'X';
85}
86
87while (($key,$val) = each(%XXX)) {
88 print "not ok 27\n";
89 exit;
90}
91
92print "ok 27\n";
378cc40b 93
378cc40b
LW
94'cde' =~ /[^ab]*/;
95'xyz' =~ //;
96if ($& eq 'xyz') {print "ok 28\n";} else {print "not ok 28\n";}
97
98$foo = '[^ab]*';
99'cde' =~ /$foo/;
100'xyz' =~ //;
101if ($& eq 'xyz') {print "ok 29\n";} else {print "not ok 29\n";}
102
103$foo = '[^ab]*';
104'cde' =~ /$foo/;
105'xyz' =~ /$null/;
106if ($& eq 'xyz') {print "ok 30\n";} else {print "not ok 30\n";}
a687059c
LW
107
108$_ = 'abcdefghi';
109/def/; # optimized up to cmd
110if ("$`:$&:$'" eq 'abc:def:ghi') {print "ok 31\n";} else {print "not ok 31\n";}
111
112/cde/ + 0; # optimized only to spat
113if ("$`:$&:$'" eq 'ab:cde:fghi') {print "ok 32\n";} else {print "not ok 32\n";}
114
115/[d][e][f]/; # not optimized
116if ("$`:$&:$'" eq 'abc:def:ghi') {print "ok 33\n";} else {print "not ok 33\n";}
117
118$_ = 'now is the {time for all} good men to come to.';
119/ {([^}]*)}/;
120if ($1 eq 'time for all') {print "ok 34\n";} else {print "not ok 34 $1\n";}
121
122$_ = 'xxx {3,4} yyy zzz';
123print /( {3,4})/ ? "ok 35\n" : "not ok 35\n";
124print $1 eq ' ' ? "ok 36\n" : "not ok 36\n";
125print /( {4,})/ ? "not ok 37\n" : "ok 37\n";
126print /( {2,3}.)/ ? "ok 38\n" : "not ok 38\n";
127print $1 eq ' y' ? "ok 39\n" : "not ok 39\n";
128print /(y{2,3}.)/ ? "ok 40\n" : "not ok 40\n";
129print $1 eq 'yyy ' ? "ok 41\n" : "not ok 41\n";
130print /x {3,4}/ ? "not ok 42\n" : "ok 42\n";
131print /^xxx {3,4}/ ? "not ok 43\n" : "ok 43\n";
352d5a3a
LW
132
133$_ = "now is the time for all good men to come to.";
134@words = /(\w+)/g;
135print join(':',@words) eq "now:is:the:time:for:all:good:men:to:come:to"
136 ? "ok 44\n"
137 : "not ok 44\n";
138
139@words = ();
140while (/\w+/g) {
141 push(@words, $&);
142}
143print join(':',@words) eq "now:is:the:time:for:all:good:men:to:come:to"
144 ? "ok 45\n"
145 : "not ok 45\n";
146
147@words = ();
71be2cbc 148pos = 0;
352d5a3a
LW
149while (/to/g) {
150 push(@words, $&);
151}
152print join(':',@words) eq "to:to"
153 ? "ok 46\n"
71be2cbc 154 : "not ok 46 `@words'\n";
352d5a3a 155
71be2cbc 156pos $_ = 0;
352d5a3a
LW
157@words = /to/g;
158print join(':',@words) eq "to:to"
159 ? "ok 47\n"
71be2cbc 160 : "not ok 47 `@words'\n";
352d5a3a
LW
161
162$_ = "abcdefghi";
163
164$pat1 = 'def';
165$pat2 = '^def';
166$pat3 = '.def.';
167$pat4 = 'abc';
168$pat5 = '^abc';
169$pat6 = 'abc$';
170$pat7 = 'ghi';
171$pat8 = '\w*ghi';
172$pat9 = 'ghi$';
173
174$t1=$t2=$t3=$t4=$t5=$t6=$t7=$t8=$t9=0;
175
176for $iter (1..5) {
177 $t1++ if /$pat1/o;
178 $t2++ if /$pat2/o;
179 $t3++ if /$pat3/o;
180 $t4++ if /$pat4/o;
181 $t5++ if /$pat5/o;
182 $t6++ if /$pat6/o;
183 $t7++ if /$pat7/o;
184 $t8++ if /$pat8/o;
185 $t9++ if /$pat9/o;
186}
187
188$x = "$t1$t2$t3$t4$t5$t6$t7$t8$t9";
189print $x eq '505550555' ? "ok 48\n" : "not ok 48 $x\n";
1462b684
LW
190
191$xyz = 'xyz';
192print "abc" =~ /^abc$|$xyz/ ? "ok 49\n" : "not ok 49\n";
193
194# perl 4.009 says "unmatched ()"
195eval '"abc" =~ /a(bc$)|$xyz/; $result = "$&:$1"';
196print $@ eq "" ? "ok 50\n" : "not ok 50\n";
197print $result eq "abc:bc" ? "ok 51\n" : "not ok 51\n";
a0d0e21e
LW
198
199
200$_="abcfooabcbar";
201$x=/abc/g;
202print $` eq "" ? "ok 52\n" : "not ok 52\n" if $x;
203$x=/abc/g;
204print $` eq "abcfoo" ? "ok 53\n" : "not ok 53\n" if $x;
205$x=/abc/g;
206print $x == 0 ? "ok 54\n" : "not ok 54\n";
71be2cbc 207pos = 0;
a0d0e21e
LW
208$x=/ABC/gi;
209print $` eq "" ? "ok 55\n" : "not ok 55\n" if $x;
210$x=/ABC/gi;
211print $` eq "abcfoo" ? "ok 56\n" : "not ok 56\n" if $x;
212$x=/ABC/gi;
213print $x == 0 ? "ok 57\n" : "not ok 57\n";
71be2cbc 214pos = 0;
a0d0e21e
LW
215$x=/abc/g;
216print $' eq "fooabcbar" ? "ok 58\n" : "not ok 58\n" if $x;
217$x=/abc/g;
218print $' eq "bar" ? "ok 59\n" : "not ok 59\n" if $x;
219$_ .= '';
220@x=/abc/g;
221print scalar @x == 2 ? "ok 60\n" : "not ok 60\n";
71be2cbc
PP
222
223$_ = "abdc";
224pos $_ = 2;
c90c0ff4 225/\Gc/gc;
71be2cbc
PP
226print "not " if (pos $_) != 2;
227print "ok 61\n";
c90c0ff4
PP
228/\Gc/g;
229print "not " if defined pos $_;
230print "ok 62\n";
c277df42
IZ
231
232$out = 1;
233'abc' =~ m'a(?{ $out = 2 })b';
234print "not " if $out != 2;
235print "ok 63\n";
236
237$out = 1;
238'abc' =~ m'a(?{ $out = 3 })c';
239print "not " if $out != 1;
240print "ok 64\n";
241
242$_ = 'foobar1 bar2 foobar3 barfoobar5 foobar6';
243@out = /(?<!foo)bar./g;
244print "not " if "@out" ne 'bar2 barf';
245print "ok 65\n";
246
8d37f932
DD
247# Tests which depend on REG_INFTY
248$reg_infty = defined $Config{reg_infty} ? $Config{reg_infty} : 32767;
249$reg_infty_m = $reg_infty - 1; $reg_infty_p = $reg_infty + 1;
250
251# As well as failing if the pattern matches do unexpected things, the
252# next three tests will fail if you should have picked up a lower-than-
253# default value for $reg_infty from Config.pm, but have not.
254
255undef $@;
256print "not " if eval q(('aaa' =~ /(a{1,$reg_infty_m})/)[0] ne 'aaa') || $@;
257print "ok 66\n";
258
259undef $@;
260print "not " if eval q(('a' x $reg_infty_m) !~ /a{$reg_infty_m}/) || $@;
261print "ok 67\n";
262
263undef $@;
264print "not " if eval q(('a' x ($reg_infty_m - 1)) =~ /a{$reg_infty_m}/) || $@;
265print "ok 68\n";
266
267undef $@;
268eval "'aaa' =~ /a{1,$reg_infty}/";
9baa0206 269print "not " if $@ !~ m%^\QQuantifier in {,} bigger than%;
8d37f932
DD
270print "ok 69\n";
271
272eval "'aaa' =~ /a{1,$reg_infty_p}/";
273print "not "
9baa0206 274 if $@ !~ m%^\QQuantifier in {,} bigger than%;
8d37f932
DD
275print "ok 70\n";
276undef $@;
277
278# Poke a couple more parse failures
279
280$context = 'x' x 256;
281eval qq("${context}y" =~ /(?<=$context)y/);
9baa0206 282print "not " if $@ !~ m%^\QLookbehind longer than 255 not%;
8d37f932
DD
283print "ok 71\n";
284
b8c5462f 285# removed test
8d37f932
DD
286print "ok 72\n";
287
c277df42 288# Long Monsters
8d37f932 289$test = 73;
c277df42
IZ
290for $l (125, 140, 250, 270, 300000, 30) { # Ordered to free memory
291 $a = 'a' x $l;
292 print "# length=$l\nnot " unless "ba$a=" =~ /a$a=/;
293 print "ok $test\n";
294 $test++;
73d6d589 295
c277df42
IZ
296 print "not " if "b$a=" =~ /a$a=/;
297 print "ok $test\n";
298 $test++;
299}
300
301# 20000 nodes, each taking 3 words per string, and 1 per branch
302$long_constant_len = join '|', 12120 .. 32645;
303$long_var_len = join '|', 8120 .. 28645;
304%ans = ( 'ax13876y25677lbc' => 1,
305 'ax13876y25677mcb' => 0, # not b.
306 'ax13876y35677nbc' => 0, # Num too big
307 'ax13876y25677y21378obc' => 1,
308 'ax13876y25677y21378zbc' => 0, # Not followed by [k-o]
309 'ax13876y25677y21378y21378kbc' => 1,
310 'ax13876y25677y21378y21378kcb' => 0, # Not b.
311 'ax13876y25677y21378y21378y21378kbc' => 0, # 5 runs
312 );
313
314for ( keys %ans ) {
73d6d589 315 print "# const-len `$_' not => $ans{$_}\nnot "
c277df42
IZ
316 if $ans{$_} xor /a(?=([yx]($long_constant_len)){2,4}[k-o]).*b./o;
317 print "ok $test\n";
318 $test++;
73d6d589 319 print "# var-len `$_' not => $ans{$_}\nnot "
c277df42
IZ
320 if $ans{$_} xor /a(?=([yx]($long_var_len)){2,4}[k-o]).*b./o;
321 print "ok $test\n";
322 $test++;
323}
324
325$_ = " a (bla()) and x(y b((l)u((e))) and b(l(e)e)e";
326$expect = "(bla()) ((l)u((e))) (l(e)e)";
327
73d6d589 328sub matchit {
cc6b7395 329 m/
c277df42 330 (
73d6d589 331 \(
c277df42
IZ
332 (?{ $c = 1 }) # Initialize
333 (?:
334 (?(?{ $c == 0 }) # PREVIOUS iteration was OK, stop the loop
335 (?!
336 ) # Fail: will unwind one iteration back
73d6d589 337 )
c277df42
IZ
338 (?:
339 [^()]+ # Match a big chunk
340 (?=
341 [()]
342 ) # Do not try to match subchunks
343 |
73d6d589 344 \(
c277df42
IZ
345 (?{ ++$c })
346 |
73d6d589 347 \)
c277df42
IZ
348 (?{ --$c })
349 )
350 )+ # This may not match with different subblocks
351 )
352 (?(?{ $c != 0 })
353 (?!
354 ) # Fail
355 ) # Otherwise the chunk 1 may succeed with $c>0
cc6b7395 356 /xg;
c277df42
IZ
357}
358
0f5d15d6 359@ans = ();
c277df42
IZ
360push @ans, $res while $res = matchit;
361
362print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne "1 1 1";
363print "ok $test\n";
364$test++;
365
366@ans = matchit;
367
368print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne $expect;
369print "ok $test\n";
370$test++;
371
96776eda
GS
372print "not " unless "abc" =~ /^(??{"a"})b/;
373print "ok $test\n";
374$test++;
375
0f5d15d6 376my $matched;
14455d6c 377$matched = qr/\((?:(?>[^()]+)|(??{$matched}))*\)/;
0f5d15d6
IZ
378
379@ans = @ans1 = ();
380push(@ans, $res), push(@ans1, $&) while $res = m/$matched/g;
381
382print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne "1 1 1";
383print "ok $test\n";
384$test++;
385
386print "# ans1='@ans1'\n# expect='$expect'\nnot " if "@ans1" ne $expect;
387print "ok $test\n";
388$test++;
389
390@ans = m/$matched/g;
391
392print "# ans='@ans'\n# expect='$expect'\nnot " if "@ans" ne $expect;
393print "ok $test\n";
394$test++;
395
c277df42
IZ
396@ans = ('a/b' =~ m%(.*/)?(.*)%); # Stack may be bad
397print "not " if "@ans" ne 'a/ b';
398print "ok $test\n";
399$test++;
400
cc6b7395 401$code = '{$blah = 45}';
c277df42 402$blah = 12;
2cd61cdb
IZ
403eval { /(?$code)/ };
404print "not " unless $@ and $@ =~ /not allowed at runtime/ and $blah == 12;
e4d48cc9
GS
405print "ok $test\n";
406$test++;
407
2cd61cdb
IZ
408for $code ('{$blah = 45}','=xx') {
409 $blah = 12;
410 $res = eval { "xx" =~ /(?$code)/o };
411 if ($code eq '=xx') {
412 print "#'$@','$res','$blah'\nnot " unless not $@ and $res;
413 } else {
73d6d589 414 print "#'$@','$res','$blah'\nnot " unless $@ and $@ =~ /not allowed at runtime/ and $blah == 12;
2cd61cdb
IZ
415 }
416 print "ok $test\n";
417 $test++;
418}
419
e4d48cc9
GS
420$code = '{$blah = 45}';
421$blah = 12;
422eval "/(?$code)/";
cc6b7395
IZ
423print "not " if $blah != 45;
424print "ok $test\n";
425$test++;
426
427$blah = 12;
428/(?{$blah = 45})/;
c277df42
IZ
429print "not " if $blah != 45;
430print "ok $test\n";
431$test++;
432
74d6a13a
MB
433$x = 'banana';
434$x =~ /.a/g;
435print "not " unless pos($x) == 2;
436print "ok $test\n";
437$test++;
438
439$x =~ /.z/gc;
440print "not " unless pos($x) == 2;
441print "ok $test\n";
442$test++;
443
444sub f {
445 my $p = $_[0];
446 return $p;
447}
448
449$x =~ /.a/g;
450print "not " unless f(pos($x)) == 4;
451print "ok $test\n";
452$test++;
4599a1de 453
ce862d02
IZ
454$x = $^R = 67;
455'foot' =~ /foo(?{$x = 12; 75})[t]/;
456print "not " unless $^R eq '75';
457print "ok $test\n";
458$test++;
459
460$x = $^R = 67;
461'foot' =~ /foo(?{$x = 12; 75})[xy]/;
462print "not " unless $^R eq '67' and $x eq '12';
463print "ok $test\n";
464$test++;
465
466$x = $^R = 67;
467'foot' =~ /foo(?{ $^R + 12 })((?{ $x = 12; $^R + 17 })[xy])?/;
468print "not " unless $^R eq '79' and $x eq '12';
469print "ok $test\n";
470$test++;
471
8782bef2
GB
472print "not " unless qr/\b\v$/i eq '(?i-xsm:\bv$)';
473print "ok $test\n";
474$test++;
475
476print "not " unless qr/\b\v$/s eq '(?s-xim:\bv$)';
477print "ok $test\n";
478$test++;
479
480print "not " unless qr/\b\v$/m eq '(?m-xis:\bv$)';
481print "ok $test\n";
482$test++;
483
484print "not " unless qr/\b\v$/x eq '(?x-ism:\bv$)';
485print "ok $test\n";
486$test++;
487
488print "not " unless qr/\b\v$/xism eq '(?msix:\bv$)';
489print "ok $test\n";
490$test++;
491
492print "not " unless qr/\b\v$/ eq '(?-xism:\bv$)';
97197631
IZ
493print "ok $test\n";
494$test++;
495
7e5428c5
IZ
496$_ = 'xabcx';
497foreach $ans ('', 'c') {
498 /(?<=(?=a)..)((?=c)|.)/g;
02db2b7b 499 print "# \$1 ='$1'\n# \$ans='$ans'\nnot " unless $1 eq $ans;
7e5428c5
IZ
500 print "ok $test\n";
501 $test++;
502}
503
504$_ = 'a';
505foreach $ans ('', 'a', '') {
506 /^|a|$/g;
02db2b7b 507 print "# \$& ='$&'\n# \$ans='$ans'\nnot " unless $& eq $ans;
7e5428c5
IZ
508 print "ok $test\n";
509 $test++;
510}
511
09f25ae4 512sub prefixify {
73d6d589
NIS
513 my($v,$a,$b,$res) = @_;
514 $v =~ s/\Q$a\E/$b/;
515 print "not " unless $res eq $v;
09f25ae4
IZ
516 print "ok $test\n";
517 $test++;
518}
519prefixify('/a/b/lib/arch', "/a/b/lib", 'X/lib', 'X/lib/arch');
520prefixify('/a/b/man/arch', "/a/b/man", 'X/man', 'X/man/arch');
521
522$_ = 'var="foo"';
523/(\")/;
524print "not " unless $1 and /$1/;
525print "ok $test\n";
526$test++;
527
73d6d589 528$a=qr/(?{++$b})/;
2cd61cdb 529$b = 7;
73d6d589
NIS
530/$a$a/;
531print "not " unless $b eq '9';
2cd61cdb
IZ
532print "ok $test\n";
533$test++;
534
73d6d589
NIS
535$c="$a";
536/$a$a/;
537print "not " unless $b eq '11';
2cd61cdb
IZ
538print "ok $test\n";
539$test++;
540
541{
73d6d589
NIS
542 use re "eval";
543 /$a$c$a/;
544 print "not " unless $b eq '14';
2cd61cdb
IZ
545 print "ok $test\n";
546 $test++;
547
160cb429
JH
548 local $lex_a = 2;
549 my $lex_a = 43;
550 my $lex_b = 17;
551 my $lex_c = 27;
552 my $lex_res = ($lex_b =~ qr/$lex_b(?{ $lex_c = $lex_a++ })/);
553 print "not " unless $lex_res eq '1';
554 print "ok $test\n";
555 $test++;
556 print "not " unless $lex_a eq '44';
557 print "ok $test\n";
558 $test++;
559 print "not " unless $lex_c eq '43';
560 print "ok $test\n";
561 $test++;
562
563
73d6d589 564 no re "eval";
2cd61cdb 565 $match = eval { /$a$c$a/ };
73d6d589 566 print "not "
2cd61cdb
IZ
567 unless $b eq '14' and $@ =~ /Eval-group not allowed/ and not $match;
568 print "ok $test\n";
569 $test++;
570}
cbce877f
IZ
571
572{
160cb429
JH
573 local $lex_a = 2;
574 my $lex_a = 43;
575 my $lex_b = 17;
576 my $lex_c = 27;
577 my $lex_res = ($lex_b =~ qr/17(?{ $lex_c = $lex_a++ })/);
578 print "not " unless $lex_res eq '1';
579 print "ok $test\n";
580 $test++;
581 print "not " unless $lex_a eq '44';
582 print "ok $test\n";
583 $test++;
584 print "not " unless $lex_c eq '43';
585 print "ok $test\n";
586 $test++;
587}
588
589{
cbce877f
IZ
590 package aa;
591 $c = 2;
592 $::c = 3;
593 '' =~ /(?{ $c = 4 })/;
594 print "not " unless $c == 4;
595}
596print "ok $test\n";
597$test++;
598print "not " unless $c == 3;
599print "ok $test\n";
73d6d589
NIS
600$test++;
601
4599a1de
JH
602sub must_warn_pat {
603 my $warn_pat = shift;
604 return sub { print "not " unless $_[0] =~ /$warn_pat/ }
605}
606
607sub must_warn {
608 my ($warn_pat, $code) = @_;
9f1b1f2d
GS
609 local %SIG;
610 eval 'BEGIN { use warnings; $SIG{__WARN__} = $warn_pat };' . $code;
4599a1de
JH
611 print "ok $test\n";
612 $test++;
613}
614
615
616sub make_must_warn {
617 my $warn_pat = shift;
618 return sub { must_warn(must_warn_pat($warn_pat)) }
619}
620
621my $for_future = make_must_warn('reserved for future extensions');
622
623&$for_future('q(a:[b]:) =~ /[x[:foo:]]/');
9baa0206
HS
624
625#&$for_future('q(a=[b]=) =~ /[x[=foo=]]/');
626print "ok $test\n"; $test++; # now a fatal croak
627
628#&$for_future('q(a.[b].) =~ /[x[.foo.]]/');
629print "ok $test\n"; $test++; # now a fatal croak
f7e33566
GS
630
631# test if failure of patterns returns empty list
632$_ = 'aaa';
633@_ = /bbb/;
634print "not " if @_;
635print "ok $test\n";
636$test++;
637
638@_ = /bbb/g;
639print "not " if @_;
640print "ok $test\n";
641$test++;
642
643@_ = /(bbb)/;
644print "not " if @_;
645print "ok $test\n";
646$test++;
647
648@_ = /(bbb)/g;
649print "not " if @_;
650print "ok $test\n";
651$test++;
652
6cef1e77
IZ
653/a(?=.$)/;
654print "not " if $#+ != 0 or $#- != 0;
655print "ok $test\n";
656$test++;
657
658print "not " if $+[0] != 2 or $-[0] != 1;
659print "ok $test\n";
660$test++;
661
73d6d589 662print "not "
6cef1e77
IZ
663 if defined $+[1] or defined $-[1] or defined $+[2] or defined $-[2];
664print "ok $test\n";
665$test++;
666
667/a(a)(a)/;
668print "not " if $#+ != 2 or $#- != 2;
669print "ok $test\n";
670$test++;
671
672print "not " if $+[0] != 3 or $-[0] != 0;
673print "ok $test\n";
674$test++;
675
676print "not " if $+[1] != 2 or $-[1] != 1;
677print "ok $test\n";
678$test++;
679
680print "not " if $+[2] != 3 or $-[2] != 2;
681print "ok $test\n";
682$test++;
683
73d6d589 684print "not "
6cef1e77
IZ
685 if defined $+[3] or defined $-[3] or defined $+[4] or defined $-[4];
686print "ok $test\n";
687$test++;
688
689/.(a)(b)?(a)/;
690print "not " if $#+ != 3 or $#- != 3;
691print "ok $test\n";
692$test++;
693
694print "not " if $+[0] != 3 or $-[0] != 0;
695print "ok $test\n";
696$test++;
697
698print "not " if $+[1] != 2 or $-[1] != 1;
699print "ok $test\n";
700$test++;
701
702print "not " if $+[3] != 3 or $-[3] != 2;
703print "ok $test\n";
704$test++;
705
73d6d589 706print "not "
6cef1e77
IZ
707 if defined $+[2] or defined $-[2] or defined $+[4] or defined $-[4];
708print "ok $test\n";
709$test++;
710
711/.(a)/;
712print "not " if $#+ != 1 or $#- != 1;
713print "ok $test\n";
714$test++;
715
716print "not " if $+[0] != 2 or $-[0] != 0;
717print "ok $test\n";
718$test++;
719
720print "not " if $+[1] != 2 or $-[1] != 1;
721print "ok $test\n";
722$test++;
723
73d6d589 724print "not "
6cef1e77
IZ
725 if defined $+[2] or defined $-[2] or defined $+[3] or defined $-[3];
726print "ok $test\n";
727$test++;
728
03a27ae7 729eval { $+[0] = 13; };
73d6d589 730print "not "
03a27ae7
MG
731 if $@ !~ /^Modification of a read-only value attempted/;
732print "ok $test\n";
733$test++;
734
735eval { $-[0] = 13; };
73d6d589 736print "not "
03a27ae7
MG
737 if $@ !~ /^Modification of a read-only value attempted/;
738print "ok $test\n";
739$test++;
740
741eval { @+ = (7, 6, 5); };
73d6d589 742print "not "
03a27ae7
MG
743 if $@ !~ /^Modification of a read-only value attempted/;
744print "ok $test\n";
745$test++;
746
747eval { @- = qw(foo bar); };
73d6d589 748print "not "
03a27ae7
MG
749 if $@ !~ /^Modification of a read-only value attempted/;
750print "ok $test\n";
751$test++;
752
8f580fb8
IZ
753/.(a)(ba*)?/;
754print "#$#-..$#+\nnot " if $#+ != 2 or $#- != 1;
755print "ok $test\n";
756$test++;
757
ad94a511
IZ
758$_ = 'aaa';
759pos = 1;
760@a = /\Ga/g;
761print "not " unless "@a" eq "a a";
762print "ok $test\n";
763$test++;
764
22e551b9
IZ
765$str = 'abcde';
766pos $str = 2;
767
768print "not " if $str =~ /^\G/;
769print "ok $test\n";
770$test++;
771
772print "not " if $str =~ /^.\G/;
773print "ok $test\n";
774$test++;
775
776print "not " unless $str =~ /^..\G/;
777print "ok $test\n";
778$test++;
779
780print "not " if $str =~ /^...\G/;
781print "ok $test\n";
782$test++;
783
784print "not " unless $str =~ /.\G./ and $& eq 'bc';
785print "ok $test\n";
786$test++;
787
788print "not " unless $str =~ /\G../ and $& eq 'cd';
789print "ok $test\n";
790$test++;
791
9661b544
IZ
792undef $foo; undef $bar;
793print "#'$str','$foo','$bar'\nnot "
73d6d589 794 unless $str =~ /b(?{$foo = $_; $bar = pos})c/
9661b544
IZ
795 and $foo eq 'abcde' and $bar eq 2;
796print "ok $test\n";
797$test++;
798
799undef $foo; undef $bar;
800pos $str = undef;
801print "#'$str','$foo','$bar'\nnot "
73d6d589 802 unless $str =~ /b(?{$foo = $_; $bar = pos})c/g
9661b544
IZ
803 and $foo eq 'abcde' and $bar eq 2 and pos $str eq 3;
804print "ok $test\n";
805$test++;
806
807$_ = $str;
808
809undef $foo; undef $bar;
810print "#'$str','$foo','$bar'\nnot "
73d6d589 811 unless /b(?{$foo = $_; $bar = pos})c/
9661b544
IZ
812 and $foo eq 'abcde' and $bar eq 2;
813print "ok $test\n";
814$test++;
815
816undef $foo; undef $bar;
817print "#'$str','$foo','$bar'\nnot "
73d6d589 818 unless /b(?{$foo = $_; $bar = pos})c/g
9661b544
IZ
819 and $foo eq 'abcde' and $bar eq 2 and pos eq 3;
820print "ok $test\n";
821$test++;
822
823undef $foo; undef $bar;
824pos = undef;
8251 while /b(?{$foo = $_; $bar = pos})c/g;
826print "#'$str','$foo','$bar'\nnot "
827 unless $foo eq 'abcde' and $bar eq 2 and not defined pos;
828print "ok $test\n";
829$test++;
830
831undef $foo; undef $bar;
832$_ = 'abcde|abcde';
833print "#'$str','$foo','$bar','$_'\nnot "
73d6d589 834 unless s/b(?{$foo = $_; $bar = pos})c/x/g and $foo eq 'abcde|abcde'
9661b544
IZ
835 and $bar eq 8 and $_ eq 'axde|axde';
836print "ok $test\n";
837$test++;
838
5c5e4c24
IZ
839@res = ();
840# List context:
841$_ = 'abcde|abcde';
842@dummy = /([ace]).(?{push @res, $1,$2})([ce])(?{push @res, $1,$2})/g;
843@res = map {defined $_ ? "'$_'" : 'undef'} @res;
844$res = "@res";
845print "#'@res' '$_'\nnot "
846 unless "@res" eq "'a' undef 'a' 'c' 'e' undef 'a' undef 'a' 'c'";
847print "ok $test\n";
848$test++;
849
850@res = ();
851@dummy = /([ace]).(?{push @res, $`,$&,$'})([ce])(?{push @res, $`,$&,$'})/g;
852@res = map {defined $_ ? "'$_'" : 'undef'} @res;
853$res = "@res";
854print "#'@res' '$_'\nnot "
855 unless "@res" eq
856 "'' 'ab' 'cde|abcde' " .
857 "'' 'abc' 'de|abcde' " .
858 "'abcd' 'e|' 'abcde' " .
859 "'abcde|' 'ab' 'cde' " .
860 "'abcde|' 'abc' 'de'" ;
861print "ok $test\n";
862$test++;
863
b7a35066
IZ
864#Some more \G anchor checks
865$foo='aabbccddeeffgg';
866
867pos($foo)=1;
868
869$foo=~/.\G(..)/g;
870print "not " unless($1 eq 'ab');
871print "ok $test\n";
872$test++;
873
874pos($foo) += 1;
875$foo=~/.\G(..)/g;
876print "not " unless($1 eq 'cc');
877print "ok $test\n";
878$test++;
879
880pos($foo) += 1;
881$foo=~/.\G(..)/g;
882print "not " unless($1 eq 'de');
883print "ok $test\n";
884$test++;
885
0ef3e39e
HS
886print "not " unless $foo =~ /\Gef/g;
887print "ok $test\n";
888$test++;
889
b7a35066
IZ
890undef pos $foo;
891
892$foo=~/\G(..)/g;
893print "not " unless($1 eq 'aa');
894print "ok $test\n";
895$test++;
896
897$foo=~/\G(..)/g;
898print "not " unless($1 eq 'bb');
899print "ok $test\n";
900$test++;
901
902pos($foo)=5;
903$foo=~/\G(..)/g;
904print "not " unless($1 eq 'cd');
905print "ok $test\n";
906$test++;
907
73d6d589 908$_='123x123';
e60df1fa
IZ
909@res = /(\d*|x)/g;
910print "not " unless('123||x|123|' eq join '|', @res);
911print "ok $test\n";
912$test++;
913
9d080a66
GS
914# see if matching against temporaries (created via pp_helem()) is safe
915{ foo => "ok $test\n".$^X }->{foo} =~ /^(.*)\n/g;
916print "$1\n";
917$test++;
918
cf93c79d
IZ
919# See if $i work inside (?{}) in the presense of saved substrings and
920# changing $_
921@a = qw(foo bar);
922@b = ();
923s/(\w)(?{push @b, $1})/,$1,/g for @a;
924
925print "# \@b='@b', expect 'f o o b a r'\nnot " unless("@b" eq "f o o b a r");
926print "ok $test\n";
927$test++;
928
929print "not " unless("@a" eq ",f,,o,,o, ,b,,a,,r,");
930print "ok $test\n";
931$test++;
932
2c914db6 933$brackets = qr{
14455d6c 934 { (?> [^{}]+ | (??{ $brackets }) )* }
2c914db6
IZ
935 }x;
936
937"{{}" =~ $brackets;
938print "ok $test\n"; # Did we survive?
939$test++;
940
941"something { long { and } hairy" =~ $brackets;
942print "ok $test\n"; # Did we survive?
943$test++;
944
14455d6c 945"something { long { and } hairy" =~ m/((??{ $brackets }))/;
2c914db6
IZ
946print "not " unless $1 eq "{ and }";
947print "ok $test\n";
948$test++;
949
30944b6d
IZ
950$_ = "a-a\nxbb";
951pos=1;
952m/^-.*bb/mg and print "not ";
953print "ok $test\n";
954$test++;
30382c73
IZ
955
956$text = "aaXbXcc";
957pos($text)=0;
958$text =~ /\GXb*X/g and print 'not ';
959print "ok $test\n";
960$test++;
3cf5c195
IZ
961
962$text = "xA\n" x 500;
963$text =~ /^\s*A/m and print 'not ';
964print "ok $test\n";
965$test++;
d506a20d
IZ
966
967$text = "abc dbf";
968@res = ($text =~ /.*?(b).*?\b/g);
969"@res" eq 'b b' or print 'not ';
970print "ok $test\n";
971$test++;
972
9442cb0e 973@a = map chr,0..255;
aeaf5620
GS
974
975@b = grep(/\S/,@a);
976@c = grep(/[^\s]/,@a);
977print "not " if "@b" ne "@c";
9442cb0e
GS
978print "ok $test\n";
979$test++;
980
aeaf5620
GS
981@b = grep(/\S/,@a);
982@c = grep(/[\S]/,@a);
983print "not " if "@b" ne "@c";
9442cb0e
GS
984print "ok $test\n";
985$test++;
986
aeaf5620
GS
987@b = grep(/\s/,@a);
988@c = grep(/[^\S]/,@a);
989print "not " if "@b" ne "@c";
9442cb0e
GS
990print "ok $test\n";
991$test++;
992
aeaf5620
GS
993@b = grep(/\s/,@a);
994@c = grep(/[\s]/,@a);
995print "not " if "@b" ne "@c";
9442cb0e
GS
996print "ok $test\n";
997$test++;
998
aeaf5620
GS
999@b = grep(/\D/,@a);
1000@c = grep(/[^\d]/,@a);
1001print "not " if "@b" ne "@c";
9442cb0e
GS
1002print "ok $test\n";
1003$test++;
1004
aeaf5620
GS
1005@b = grep(/\D/,@a);
1006@c = grep(/[\D]/,@a);
1007print "not " if "@b" ne "@c";
9442cb0e
GS
1008print "ok $test\n";
1009$test++;
1010
aeaf5620
GS
1011@b = grep(/\d/,@a);
1012@c = grep(/[^\D]/,@a);
1013print "not " if "@b" ne "@c";
9442cb0e
GS
1014print "ok $test\n";
1015$test++;
1016
aeaf5620
GS
1017@b = grep(/\d/,@a);
1018@c = grep(/[\d]/,@a);
1019print "not " if "@b" ne "@c";
9442cb0e
GS
1020print "ok $test\n";
1021$test++;
1022
aeaf5620
GS
1023@b = grep(/\W/,@a);
1024@c = grep(/[^\w]/,@a);
1025print "not " if "@b" ne "@c";
9442cb0e
GS
1026print "ok $test\n";
1027$test++;
1028
aeaf5620
GS
1029@b = grep(/\W/,@a);
1030@c = grep(/[\W]/,@a);
1031print "not " if "@b" ne "@c";
9442cb0e
GS
1032print "ok $test\n";
1033$test++;
1034
aeaf5620
GS
1035@b = grep(/\w/,@a);
1036@c = grep(/[^\W]/,@a);
1037print "not " if "@b" ne "@c";
9442cb0e
GS
1038print "ok $test\n";
1039$test++;
1040
aeaf5620
GS
1041@b = grep(/\w/,@a);
1042@c = grep(/[\w]/,@a);
1043print "not " if "@b" ne "@c";
9442cb0e
GS
1044print "ok $test\n";
1045$test++;
1aeab75a
GS
1046
1047# see if backtracking optimization works correctly
1048"\n\n" =~ /\n $ \n/x or print "not ";
1049print "ok $test\n";
1050$test++;
1051
1052"\n\n" =~ /\n* $ \n/x or print "not ";
1053print "ok $test\n";
1054$test++;
1055
1056"\n\n" =~ /\n+ $ \n/x or print "not ";
1057print "ok $test\n";
1058$test++;
05b4157f
GS
1059
1060[] =~ /^ARRAY/ or print "# [] \nnot ";
1061print "ok $test\n";
1062$test++;
1063
1064eval << 'EOE';
1065{
1066 package S;
1067 use overload '""' => sub { 'Object S' };
1068 sub new { bless [] }
1069}
1070$a = 'S'->new;
1071EOE
1072
1073$a and $a =~ /^Object\sS/ or print "# '$a' \nnot ";
1074print "ok $test\n";
1075$test++;
815d35b9
G
1076
1077# test result of match used as match (!)
1078'a1b' =~ ('xyz' =~ /y/) and $` eq 'a' or print "not ";
1079print "ok $test\n";
1080$test++;
1081
1082'a1b' =~ ('xyz' =~ /t/) and $` eq 'a' or print "not ";
1083print "ok $test\n";
1084$test++;
5e39e1e5
HS
1085
1086$w = 0;
1087{
1088 local $SIG{__WARN__} = sub { $w = 1 };
1089 local $^W = 1;
1090 $w = 1 if ("1\n" x 102) =~ /^\s*\n/m;
1091}
1092print $w ? "not " : "", "ok $test\n";
1093$test++;
aaa51d5e
JF
1094
1095my %space = ( spc => " ",
1096 tab => "\t",
1097 cr => "\r",
1098 lf => "\n",
1099 ff => "\f",
75369ccb
JH
1100# There's no \v but the vertical tabulator seems miraculously
1101# be 11 both in ASCII and EBCDIC.
aaa51d5e
JF
1102 vt => chr(11),
1103 false => "space" );
1104
1105my @space0 = sort grep { $space{$_} =~ /\s/ } keys %space;
1106my @space1 = sort grep { $space{$_} =~ /[[:space:]]/ } keys %space;
1107my @space2 = sort grep { $space{$_} =~ /[[:blank:]]/ } keys %space;
1108
1109print "not " unless "@space0" eq "cr ff lf spc tab";
3bec3564 1110print "ok $test # @space0\n";
aaa51d5e
JF
1111$test++;
1112
1113print "not " unless "@space1" eq "cr ff lf spc tab vt";
3bec3564 1114print "ok $test # @space1\n";
aaa51d5e
JF
1115$test++;
1116
1117print "not " unless "@space2" eq "spc tab";
3bec3564 1118print "ok $test # @space2\n";
aaa51d5e 1119$test++;
73d6d589 1120
a1933d95
HS
1121# bugid 20001021.005 - this caused a SEGV
1122print "not " unless undef =~ /^([^\/]*)(.*)$/;
1123print "ok $test\n";
1124$test++;
b91bb191
JH
1125
1126# bugid 20000731.001
1127
1128print "not " unless "A \x{263a} B z C" =~ /A . B (??{ "z" }) C/;
1129print "ok $test\n";
1130$test++;
1131
3baa4c62
JH
1132$_ = "a\x{100}b";
1133if (/(.)(\C)(\C)(.)/) {
1134 print "ok 232\n";
1135 if ($1 eq "a") {
1136 print "ok 233\n";
1137 } else {
1138 print "not ok 233\n";
1139 }
1140 if ($2 eq "\xC4") {
1141 print "ok 234\n";
1142 } else {
1143 print "not ok 234\n";
1144 }
1145 if ($3 eq "\x80") {
1146 print "ok 235\n";
1147 } else {
1148 print "not ok 235\n";
1149 }
1150 if ($4 eq "b") {
1151 print "ok 236\n";
1152 } else {
1153 print "not ok 236\n";
1154 }
1155} else {
1156 for (232..236) {
1157 print "not ok $_\n";
1158 }
1159}
1160$_ = "\x{100}";
1161if (/(\C)/g) {
1162 print "ok 237\n";
73d6d589 1163 # currently \C are still tagged as UTF-8
3baa4c62
JH
1164 if ($1 eq "\xC4") {
1165 print "ok 238\n";
1166 } else {
1167 print "not ok 238\n";
1168 }
1169} else {
1170 for (237..238) {
1171 print "not ok $_\n";
1172 }
1173}
1174if (/(\C)/g) {
1175 print "ok 239\n";
73d6d589 1176 # currently \C are still tagged as UTF-8
3baa4c62
JH
1177 if ($1 eq "\x80") {
1178 print "ok 240\n";
1179 } else {
1180 print "not ok 240\n";
1181 }
1182} else {
1183 for (239..240) {
1184 print "not ok $_\n";
1185 }
1186}
b485d051 1187
db615365
JP
1188{
1189 # japhy -- added 03/03/2001
1190 () = (my $str = "abc") =~ /(...)/;
1191 $str = "def";
1192 print "not " if $1 ne "abc";
fd291da9
JH
1193 print "ok 241\n";
1194}
1195
1196# The 242 and 243 go with the 244 and 245.
1197# The trick is that in EBCDIC the explicit numeric range should match
1198# (as also in non-EBCDIC) but the explicit alphabetic range should not match.
1199
1200if ("\x8e" =~ /[\x89-\x91]/) {
1201 print "ok 242\n";
1202} else {
1203 print "not ok 242\n";
1204}
1205
1206if ("\xce" =~ /[\xc9-\xd1]/) {
db615365 1207 print "ok 243\n";
fd291da9
JH
1208} else {
1209 print "not ok 243\n";
1210}
1211
1212# In most places these tests would succeed since \x8e does not
1213# in most character sets match 'i' or 'j' nor would \xce match
1214# 'I' or 'J', but strictly speaking these tests are here for
1215# the good of EBCDIC, so let's test these only there.
1216if (ord('i') == 0x89 && ord('J') == 0xd1) { # EBCDIC
1217 if ("\x8e" !~ /[i-j]/) {
1218 print "ok 244\n";
1219 } else {
1220 print "not ok 244\n";
1221 }
1222 if ("\xce" !~ /[I-J]/) {
1223 print "ok 245\n";
1224 } else {
1225 print "not ok 245\n";
1226 }
1227} else {
1228 for (244..245) {
60425c38 1229 print "ok $_ # Skip: only in EBCDIC\n";
fd291da9 1230 }
db615365 1231}
4765795a
JH
1232
1233print "not " unless "\x{ab}" =~ /\x{ab}/;
1234print "ok 246\n";
1235
1236print "not " unless "\x{abcd}" =~ /\x{abcd}/;
1237print "ok 247\n";
1238
1239{
1240 # bug id 20001008.001
1241
4765795a
JH
1242 my $test = 248;
1243 my @x = ("stra\337e 138","stra\337e 138");
1244 for (@x) {
1245 s/(\d+)\s*([\w\-]+)/$1 . uc $2/e;
1246 my($latin) = /^(.+)(?:\s+\d)/;
1247 print $latin eq "stra\337e" ? "ok $test\n" : # 248,249
1248 "#latin[$latin]\nnot ok $test\n";
1249 $test++;
1250 $latin =~ s/stra\337e/straße/; # \303\237 after the 2nd a
1251 use utf8;
1252 $latin =~ s!(s)tr(?:aß|s+e)!$1tr.!; # \303\237 after the a
1253 }
1254}
1255
1256{
1257 print "not " unless "ba\xd4c" =~ /([a\xd4]+)/ && $1 eq "a\xd4";
1258 print "ok 250\n";
1259
1260 print "not " unless "ba\xd4c" =~ /([a\xd4]+)/ && $1 eq "a\x{d4}";
1261 print "ok 251\n";
1262
1263 print "not " unless "ba\x{d4}c" =~ /([a\xd4]+)/ && $1 eq "a\x{d4}";
1264 print "ok 252\n";
1265
1266 print "not " unless "ba\x{d4}c" =~ /([a\xd4]+)/ && $1 eq "a\xd4";
1267 print "ok 253\n";
1268
1269 print "not " unless "ba\xd4c" =~ /([a\x{d4}]+)/ && $1 eq "a\xd4";
1270 print "ok 254\n";
1271
1272 print "not " unless "ba\xd4c" =~ /([a\x{d4}]+)/ && $1 eq "a\x{d4}";
1273 print "ok 255\n";
1274
1275 print "not " unless "ba\x{d4}c" =~ /([a\x{d4}]+)/ && $1 eq "a\x{d4}";
1276 print "ok 256\n";
1277
1278 print "not " unless "ba\x{d4}c" =~ /([a\x{d4}]+)/ && $1 eq "a\xd4";
1279 print "ok 257\n";
1280}
1281
1282{
1283 # the first half of 20001028.003
1284
1285 my $X = chr(1448);
1286 my ($Y) = $X =~ /(.*)/;
1287 print "not " unless $Y eq v1448 && length($Y) == 1;
1288 print "ok 258\n";
1289}
1290
1291{
1292 # 20001108.001
1293
1294 my $X = "Szab\x{f3},Bal\x{e1}zs";
1295 my $Y = $X;
1296 $Y =~ s/(B)/$1/ for 0..3;
1297 print "not " unless $Y eq $X && $X eq "Szab\x{f3},Bal\x{e1}zs";
1298 print "ok 259\n";
1299}
1300
1301{
1302 # the second half of 20001028.003
1303
3568d838 1304 my $X = '';
4765795a
JH
1305 $X =~ s/^/chr(1488)/e;
1306 print "not " unless length $X == 1 && ord($X) == 1488;
1307 print "ok 260\n";
1308}
1309
1310{
1311 # 20000517.001
1312
1313 my $x = "\x{100}A";
1314
1315 $x =~ s/A/B/;
1316
1317 print "not " unless $x eq "\x{100}B" && length($x) == 2;
1318 print "ok 261\n";
1319}
1320
1321{
1322 # bug id 20001230.002
1323
1324 print "not " unless "École" =~ /^\C\C(.)/ && $1 eq 'c';
1325 print "ok 262\n";
1326
1327 print "not " unless "École" =~ /^\C\C(c)/;
1328 print "ok 263\n";
1329}
1330
1331{
1332 my $test = 264; # till 575
1333
1334 use charnames ':full';
1335
1336 # This is far from complete testing, there are dozens of character
1337 # classes in Unicode. The mixing of literals and \N{...} is
1338 # intentional so that in non-Latin-1 places we test the native
1339 # characters, not the Unicode code points.
1340
1341 my %s = (
1342 "a" => 'Ll',
1343 "\N{CYRILLIC SMALL LETTER A}" => 'Ll',
1344 "A" => 'Lu',
1345 "\N{GREEK CAPITAL LETTER ALPHA}" => 'Lu',
1346 "\N{HIRAGANA LETTER SMALL A}" => 'Lo',
1347 "\N{COMBINING GRAVE ACCENT}" => 'Mn',
1348 "0" => 'Nd',
1349 "\N{ARABIC-INDIC DIGIT ZERO}" => 'Nd',
1350 "_" => 'N',
1351 "!" => 'P',
1352 " " => 'Zs',
1353 "\0" => 'Cc',
1354 );
73d6d589 1355
3568d838
JH
1356 for my $char (map { s/^\S+ //; $_ }
1357 sort map { sprintf("%06x", ord($_))." $_" } keys %s) {
4765795a 1358 my $class = $s{$char};
3568d838
JH
1359 my $code = sprintf("%06x", ord($char));
1360 printf "#\n# 0x$code\n#\n";
4765795a
JH
1361 print "# IsAlpha\n";
1362 if ($class =~ /^[LM]/) {
1363 print "not " unless $char =~ /\p{IsAlpha}/;
1364 print "ok $test\n"; $test++;
1365 print "not " if $char =~ /\P{IsAlpha}/;
1366 print "ok $test\n"; $test++;
1367 } else {
1368 print "not " if $char =~ /\p{IsAlpha}/;
1369 print "ok $test\n"; $test++;
1370 print "not " unless $char =~ /\P{IsAlpha}/;
1371 print "ok $test\n"; $test++;
1372 }
1373 print "# IsAlnum\n";
1374 if ($class =~ /^[LMN]/ && $char ne "_") {
1375 print "not " unless $char =~ /\p{IsAlnum}/;
1376 print "ok $test\n"; $test++;
1377 print "not " if $char =~ /\P{IsAlnum}/;
1378 print "ok $test\n"; $test++;
1379 } else {
1380 print "not " if $char =~ /\p{IsAlnum}/;
1381 print "ok $test\n"; $test++;
1382 print "not " unless $char =~ /\P{IsAlnum}/;
1383 print "ok $test\n"; $test++;
1384 }
1385 print "# IsASCII\n";
3568d838 1386 if ($code le '00007f') {
4765795a
JH
1387 print "not " unless $char =~ /\p{IsASCII}/;
1388 print "ok $test\n"; $test++;
1389 print "not " if $char =~ /\P{IsASCII}/;
1390 print "ok $test\n"; $test++;
1391 } else {
1392 print "not " if $char =~ /\p{IsASCII}/;
1393 print "ok $test\n"; $test++;
1394 print "not " unless $char =~ /\P{IsASCII}/;
1395 print "ok $test\n"; $test++;
1396 }
1397 print "# IsCntrl\n";
1398 if ($class =~ /^C/) {
1399 print "not " unless $char =~ /\p{IsCntrl}/;
1400 print "ok $test\n"; $test++;
1401 print "not " if $char =~ /\P{IsCntrl}/;
1402 print "ok $test\n"; $test++;
1403 } else {
1404 print "not " if $char =~ /\p{IsCntrl}/;
1405 print "ok $test\n"; $test++;
1406 print "not " unless $char =~ /\P{IsCntrl}/;
1407 print "ok $test\n"; $test++;
1408 }
1409 print "# IsBlank\n";
1410 if ($class =~ /^Z[lp]/ || $char eq " ") {
1411 print "not " unless $char =~ /\p{IsBlank}/;
1412 print "ok $test\n"; $test++;
1413 print "not " if $char =~ /\P{IsBlank}/;
1414 print "ok $test\n"; $test++;
1415 } else {
1416 print "not " if $char =~ /\p{IsBlank}/;
1417 print "ok $test\n"; $test++;
1418 print "not " unless $char =~ /\P{IsBlank}/;
1419 print "ok $test\n"; $test++;
1420 }
1421 print "# IsDigit\n";
1422 if ($class =~ /^Nd$/) {
1423 print "not " unless $char =~ /\p{IsDigit}/;
1424 print "ok $test\n"; $test++;
1425 print "not " if $char =~ /\P{IsDigit}/;
1426 print "ok $test\n"; $test++;
1427 } else {
1428 print "not " if $char =~ /\p{IsDigit}/;
1429 print "ok $test\n"; $test++;
1430 print "not " unless $char =~ /\P{IsDigit}/;
1431 print "ok $test\n"; $test++;
1432 }
1433 print "# IsGraph\n";
1434 if ($class =~ /^([LMNPS])|Co/) {
1435 print "not " unless $char =~ /\p{IsGraph}/;
1436 print "ok $test\n"; $test++;
1437 print "not " if $char =~ /\P{IsGraph}/;
1438 print "ok $test\n"; $test++;
1439 } else {
1440 print "not " if $char =~ /\p{IsGraph}/;
1441 print "ok $test\n"; $test++;
1442 print "not " unless $char =~ /\P{IsGraph}/;
1443 print "ok $test\n"; $test++;
1444 }
1445 print "# IsLower\n";
1446 if ($class =~ /^Ll$/) {
1447 print "not " unless $char =~ /\p{IsLower}/;
1448 print "ok $test\n"; $test++;
1449 print "not " if $char =~ /\P{IsLower}/;
1450 print "ok $test\n"; $test++;
1451 } else {
1452 print "not " if $char =~ /\p{IsLower}/;
1453 print "ok $test\n"; $test++;
1454 print "not " unless $char =~ /\P{IsLower}/;
1455 print "ok $test\n"; $test++;
1456 }
1457 print "# IsPrint\n";
1458 if ($class =~ /^([LMNPS])|Co|Zs/) {
1459 print "not " unless $char =~ /\p{IsPrint}/;
1460 print "ok $test\n"; $test++;
1461 print "not " if $char =~ /\P{IsPrint}/;
1462 print "ok $test\n"; $test++;
1463 } else {
1464 print "not " if $char =~ /\p{IsPrint}/;
1465 print "ok $test\n"; $test++;
1466 print "not " unless $char =~ /\P{IsPrint}/;
1467 print "ok $test\n"; $test++;
1468 }
1469 print "# IsPunct\n";
1470 if ($class =~ /^P/ || $char eq "_") {
1471 print "not " unless $char =~ /\p{IsPunct}/;
1472 print "ok $test\n"; $test++;
1473 print "not " if $char =~ /\P{IsPunct}/;
1474 print "ok $test\n"; $test++;
1475 } else {
1476 print "not " if $char =~ /\p{IsPunct}/;
1477 print "ok $test\n"; $test++;
1478 print "not " unless $char =~ /\P{IsPunct}/;
1479 print "ok $test\n"; $test++;
1480 }
1481 print "# IsSpace\n";
1482 if ($class =~ /^Z/ || ($code =~ /^(0009|000A|000B|000C|000D)$/)) {
1483 print "not " unless $char =~ /\p{IsSpace}/;
1484 print "ok $test\n"; $test++;
1485 print "not " if $char =~ /\P{IsSpace}/;
1486 print "ok $test\n"; $test++;
1487 } else {
1488 print "not " if $char =~ /\p{IsSpace}/;
1489 print "ok $test\n"; $test++;
1490 print "not " unless $char =~ /\P{IsSpace}/;
1491 print "ok $test\n"; $test++;
1492 }
1493 print "# IsUpper\n";
1494 if ($class =~ /^L[ut]/) {
1495 print "not " unless $char =~ /\p{IsUpper}/;
1496 print "ok $test\n"; $test++;
1497 print "not " if $char =~ /\P{IsUpper}/;
1498 print "ok $test\n"; $test++;
1499 } else {
1500 print "not " if $char =~ /\p{IsUpper}/;
1501 print "ok $test\n"; $test++;
1502 print "not " unless $char =~ /\P{IsUpper}/;
1503 print "ok $test\n"; $test++;
1504 }
1505 print "# IsWord\n";
1506 if ($class =~ /^[LMN]/ || $char eq "_") {
1507 print "not " unless $char =~ /\p{IsWord}/;
1508 print "ok $test\n"; $test++;
1509 print "not " if $char =~ /\P{IsWord}/;
1510 print "ok $test\n"; $test++;
1511 } else {
1512 print "not " if $char =~ /\p{IsWord}/;
1513 print "ok $test\n"; $test++;
1514 print "not " unless $char =~ /\P{IsWord}/;
1515 print "ok $test\n"; $test++;
1516 }
1517 }
1518}
1519
1520{
1521 $_ = "abc\x{100}\x{200}\x{300}\x{380}\x{400}defg";
1522
1523 if (/(.\x{300})./) {
1524 print "ok 576\n";
1525
1526 print "not " unless $` eq "abc\x{100}" && length($`) == 4;
73d6d589 1527 print "ok 577\n";
4765795a
JH
1528
1529 print "not " unless $& eq "\x{200}\x{300}\x{380}" && length($&) == 3;
73d6d589 1530 print "ok 578\n";
4765795a
JH
1531
1532 print "not " unless $' eq "\x{400}defg" && length($') == 5;
73d6d589 1533 print "ok 579\n";
4765795a
JH
1534
1535 print "not " unless $1 eq "\x{200}\x{300}" && length($1) == 2;
73d6d589 1536 print "ok 580\n";
a8a2fe91
JH
1537 } else {
1538 for (576..580) { print "not ok $_\n" }
4765795a
JH
1539 }
1540}
8269fa76
JH
1541
1542{
1543 # bug id 20010306.008
1544
1545 $a = "a\x{1234}";
1546 # The original bug report had 'no utf8' here but that was irrelevant.
1547 $a =~ m/\w/; # used to core dump
1548
1549 print "ok 581\n";
1550}
b8ef571c
JH
1551
1552{
339e86bc
JH
1553 $test = 582;
1554
b8ef571c
JH
1555 # bugid 20010410.006
1556 for my $rx (
1557 '/(.*?)\{(.*?)\}/csg',
1558 '/(.*?)\{(.*?)\}/cg',
1559 '/(.*?)\{(.*?)\}/sg',
1560 '/(.*?)\{(.*?)\}/g',
1561 '/(.+?)\{(.+?)\}/csg',
1562 )
1563 {
1564 my($input, $i);
1565
1566 $i = 0;
1567 $input = "a{b}c{d}";
1568 eval <<EOT;
1569 while (eval \$input =~ $rx) {
1570 print "# \\\$1 = '\$1' \\\$2 = '\$2'\n";
1571 ++\$i;
1572 }
1573EOT
1574 print "not " unless $i == 2;
1575 print "ok " . $test++ . "\n";
1576 }
1577}
209a9bc1
JH
1578
1579{
1580 # from Robin Houston
1581
1582 my $x = "\x{12345678}";
1583 $x =~ s/(.)/$1/g;
1584 print "not " unless ord($x) == 0x12345678 && length($x) == 1;
1585 print "ok 587\n";
1586}
3568d838
JH
1587
1588{
1589 my $x = "\x7f";
1590
1591 print "not " if $x =~ /[\x80-\xff]/;
1592 print "ok 588\n";
1593
1594 print "not " if $x =~ /[\x80-\x{100}]/;
1595 print "ok 589\n";
1596
1597 print "not " if $x =~ /[\x{100}]/;
1598 print "ok 590\n";
1599
1600 print "not " if $x =~ /\p{InLatin1Supplement}/;
1601 print "ok 591\n";
1602
1603 print "not " unless $x =~ /\P{InLatin1Supplement}/;
1604 print "ok 592\n";
1605
1606 print "not " if $x =~ /\p{InLatinExtendedA}/;
1607 print "ok 593\n";
1608
1609 print "not " unless $x =~ /\P{InLatinExtendedA}/;
1610 print "ok 594\n";
1611}
1612
1613{
1614 my $x = "\x80";
1615
1616 print "not " unless $x =~ /[\x80-\xff]/;
1617 print "ok 595\n";
1618
1619 print "not " unless $x =~ /[\x80-\x{100}]/;
1620 print "ok 596\n";
1621
1622 print "not " if $x =~ /[\x{100}]/;
1623 print "ok 597\n";
1624
1625 print "not " unless $x =~ /\p{InLatin1Supplement}/;
1626 print "ok 598\n";
1627
1628 print "not " if $x =~ /\P{InLatin1Supplement}/;
1629 print "ok 599\n";
1630
1631 print "not " if $x =~ /\p{InLatinExtendedA}/;
1632 print "ok 600\n";
1633
1634 print "not " unless $x =~ /\P{InLatinExtendedA}/;
1635 print "ok 601\n";
1636}
1637
1638{
1639 my $x = "\xff";
1640
1641 print "not " unless $x =~ /[\x80-\xff]/;
1642 print "ok 602\n";
1643
1644 print "not " unless $x =~ /[\x80-\x{100}]/;
1645 print "ok 603\n";
1646
1647 print "not " if $x =~ /[\x{100}]/;
1648 print "ok 604\n";
1649
1650 print "not " unless $x =~ /\p{InLatin1Supplement}/;
1651 print "ok 605\n";
1652
1653 print "not " if $x =~ /\P{InLatin1Supplement}/;
1654 print "ok 606\n";
1655
1656 print "not " if $x =~ /\p{InLatinExtendedA}/;
1657 print "ok 607\n";
1658
1659 print "not " unless $x =~ /\P{InLatinExtendedA}/;
1660 print "ok 608\n";
1661}
1662
1663{
1664 my $x = "\x{100}";
1665
1666 print "not " if $x =~ /[\x80-\xff]/;
1667 print "ok 609\n";
1668
1669 print "not " unless $x =~ /[\x80-\x{100}]/;
1670 print "ok 610\n";
1671
1672 print "not " unless $x =~ /[\x{100}]/;
1673 print "ok 611\n";
1674
1675 print "not " if $x =~ /\p{InLatin1Supplement}/;
1676 print "ok 612\n";
1677
1678 print "not " unless $x =~ /\P{InLatin1Supplement}/;
1679 print "ok 613\n";
1680
1681 print "not " unless $x =~ /\p{InLatinExtendedA}/;
1682 print "ok 614\n";
1683
1684 print "not " if $x =~ /\P{InLatinExtendedA}/;
1685 print "ok 615\n";
1686}
1687
9d1d55b5
JP
1688{
1689 # from japhy
1690 my $w;
1691 use warnings;
1692 local $SIG{__WARN__} = sub { $w .= shift };
1693
1694 $w = "";
1695 eval 'qr/(?c)/';
1696 print "not " if $w !~ /^Useless \(\?c\)/;
1697 print "ok 616\n";
1698
1699 $w = "";
1700 eval 'qr/(?-c)/';
1701 print "not " if $w !~ /^Useless \(\?-c\)/;
1702 print "ok 617\n";
1703
1704 $w = "";
1705 eval 'qr/(?g)/';
1706 print "not " if $w !~ /^Useless \(\?g\)/;
1707 print "ok 618\n";
1708
1709 $w = "";
1710 eval 'qr/(?-g)/';
1711 print "not " if $w !~ /^Useless \(\?-g\)/;
1712 print "ok 619\n";
1713
1714 $w = "";
1715 eval 'qr/(?o)/';
1716 print "not " if $w !~ /^Useless \(\?o\)/;
1717 print "ok 620\n";
1718
1719 $w = "";
1720 eval 'qr/(?-o)/';
1721 print "not " if $w !~ /^Useless \(\?-o\)/;
1722 print "ok 621\n";
1723
1724 # now test multi-error regexes
1725
1726 $w = "";
1727 eval 'qr/(?g-o)/';
1728 print "not " if $w !~ /^Useless \(\?g\).*\nUseless \(\?-o\)/;
1729 print "ok 622\n";
1730
1731 $w = "";
1732 eval 'qr/(?g-c)/';
1733 print "not " if $w !~ /^Useless \(\?g\).*\nUseless \(\?-c\)/;
1734 print "ok 623\n";
1735
1736 $w = "";
1737 eval 'qr/(?o-cg)/'; # (?c) means (?g) error won't be thrown
1738 print "not " if $w !~ /^Useless \(\?o\).*\nUseless \(\?-c\)/;
1739 print "ok 624\n";
1740
1741 $w = "";
1742 eval 'qr/(?ogc)/';
1743 print "not " if $w !~ /^Useless \(\?o\).*\nUseless \(\?g\).*\nUseless \(\?c\)/;
1744 print "ok 625\n";
1745}
a72deede
JH
1746
1747# More Unicode "class" tests
1748
1749{
1750 use charnames ':full';
1751
1752 print "not " unless "\N{LATIN CAPITAL LETTER A}" =~ /\p{InBasicLatin}/;
1753 print "ok 626\n";
1754
1755 print "not " unless "\N{LATIN CAPITAL LETTER A WITH GRAVE}" =~ /\p{InLatin1Supplement}/;
1756 print "ok 627\n";
1757
1758 print "not " unless "\N{LATIN CAPITAL LETTER A WITH MACRON}" =~ /\p{InLatinExtendedA}/;
1759 print "ok 628\n";
1760
1761 print "not " unless "\N{LATIN SMALL LETTER B WITH STROKE}" =~ /\p{InLatinExtendedB}/;
1762 print "ok 629\n";
1763
1764 print "not " unless "\N{KATAKANA LETTER SMALL A}" =~ /\p{InKatakana}/;
1765 print "ok 630\n";
1766}
1767
6002328a
JH
1768$_ = "foo";
1769
1770eval <<"EOT"; die if $@;
1771 /f
1772 o\r
1773 o
1774 \$
1775 /x && print "ok 631\n";
1776EOT
1777
1778eval <<"EOT"; die if $@;
1779 /f
1780 o
1781 o
1782 \$\r
1783 /x && print "ok 632\n";
1784EOT
1785
569b5e07
AB
1786#test /o feature
1787sub test_o { $_[0] =~/$_[1]/o; return $1}
1788if(test_o('abc','(.)..') eq 'a') {
395ddfe6 1789 print "ok 633\n";
569b5e07 1790} else {
395ddfe6 1791 print "not ok 633\n";
569b5e07
AB
1792}
1793if(test_o('abc','..(.)') eq 'a') {
395ddfe6 1794 print "ok 634\n";
569b5e07 1795} else {
395ddfe6 1796 print "not ok 634\n";
569b5e07
AB
1797}
1798
f79b3095
JH
1799# 635..639: ID 20010619.003 (only the space character is
1800# supposed to be [:print:], not the whole isprint()).
1801
1802print "not " if "\n" =~ /[[:print:]]/;
1803print "ok 635\n";
1804
1805print "not " if "\t" =~ /[[:print:]]/;
1806print "ok 636\n";
1807
e857312d 1808# Amazingly vertical tabulator is the same in ASCII and EBCDIC.
f79b3095
JH
1809print "not " if "\014" =~ /[[:print:]]/;
1810print "ok 637\n";
1811
1812print "not " if "\r" =~ /[[:print:]]/;
1813print "ok 638\n";
1814
1815print "not " unless " " =~ /[[:print:]]/;
1816print "ok 639\n";
1817