1 //
2 // Copyright (c) 1999, 2009, Oracle and/or its affiliates. All rights reserved.
3 // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 //
5 // This code is free software; you can redistribute it and/or modify it
6 // under the terms of the GNU General Public License version 2 only, as
7 // published by the Free Software Foundation.
8 //
9 // This code is distributed in the hope that it will be useful, but WITHOUT
10 // ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 // FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 // version 2 for more details (a copy is included in the LICENSE file that
13 // accompanied this code).
14 //
15 // You should have received a copy of the GNU General Public License version
16 // 2 along with this work; if not, write to the Free Software Foundation,
17 // Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 //
19 // Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 // or visit www.oracle.com if you need additional information or have any
21 // questions.
22 //
23 //
24 // This file contains test cases for regular expressions.
25 // A test case consists of three lines:
26 // The first line is a pattern used in the test
27 // The second line is the input to search for the pattern in
28 // The third line is a concatentation of the match, the number of groups,
29 // and the contents of the first four subexpressions.
30 // Empty lines and lines beginning with comment slashes are ignored.
31 //
32 // Test unsetting of backed off groups
33 ^(a)?a
34 a
35 true a 1
36
37 ^(aa(bb)?)+$
38 aabbaa
39 true aabbaa 2 aa bb
40
41 ((a|b)?b)+
42 b
43 true b 2 b
44
45 (aaa)?aaa
46 aaa
47 true aaa 1
48
49 ^(a(b)?)+$
50 aba
51 true aba 2 a b
52
53 ^(a(b(c)?)?)?abc
54 abc
55 true abc 3
56
57 ^(a(b(c))).*
58 abc
59 true abc 3 abc bc c
60
61 // use of x modifier
62 abc(?x)blah
63 abcblah
64 true abcblah 0
65
66 abc(?x) blah
67 abcblah
68 true abcblah 0
69
70 abc(?x) blah blech
71 abcblahblech
72 true abcblahblech 0
73
74 abc(?x) blah # ignore comment
75 abcblah
76 true abcblah 0
77
78 // Simple alternation
79 a|b
80 a
81 true a 0
82
83 a|b
84 z
85 false 0
86
87 a|b
88 b
89 true b 0
90
91 a|b|cd
92 cd
93 true cd 0
94
95 a|ad
96 ad
97 true a 0
98
99 z(a|ac)b
100 zacb
101 true zacb 1 ac
102
103 // Simple char class
104 [abc]+
105 ababab
106 true ababab 0
107
108 [abc]+
109 defg
110 false 0
111
112 [abc]+[def]+[ghi]+
113 zzzaaddggzzz
114 true aaddgg 0
115
116 // Range char class
117 [a-g]+
118 zzzggg
119 true ggg 0
120
121 [a-g]+
122 mmm
123 false 0
124
125 [a-]+
126 za-9z
127 true a- 0
128
129 [a-\\u4444]+
130 za-9z
131 true za 0
132
133 // Negated char class
134 [^abc]+
135 ababab
136 false 0
137
138 [^abc]+
139 aaabbbcccdefg
140 true defg 0
141
142 // Negation with nested char class and intersection
143 [^[c]]
144 c
145 false 0
146
147 [^[a-z]]
148 e
149 false 0
150
151 [^[a-z][A-Z]]
152 E
153 false 0
154
155 [^a-d[0-9][m-p]]
156 e
157 true e 0
158
159 [^a-d[0-9][m-p]]
160 8
161 false 0
162
163 [^[a-c]&&[d-f]]
164 z
165 true z 0
166
167 [^a-c&&d-f]
168 a
169 true a 0
170
171 [^a-m&&m-z]
172 m
173 false 0
174
175 [^a-m&&m-z&&a-c]
176 m
177 true m 0
178
179 [^a-cd-f&&[d-f]]
180 c
181 true c 0
182
183 [^[a-c][d-f]&&abc]
184 a
185 false 0
186
187 [^[a-c][d-f]&&abc]
188 d
189 true d 0
190
191 [^[a-c][d-f]&&abc[def]]
192 a
193 false 0
194
195 [^[a-c][d-f]&&abc[def]]
196 e
197 false 0
198
199 [^[a-c]&&[b-d]&&[c-e]]
200 a
201 true a 0
202
203 [^[a-c]&&[b-d]&&[c-e]]
204 c
205 false 0
206
207 // Making sure a ^ not in first position matches literal ^
208 [abc^b]
209 b
210 true b 0
211
212 [abc^b]
213 ^
214 true ^ 0
215
216 // Class union and intersection
217 [abc[def]]
218 b
219 true b 0
220
221 [abc[def]]
222 e
223 true e 0
224
225 [a-d[0-9][m-p]]
226 a
227 true a 0
228
229 [a-d[0-9][m-p]]
230 o
231 true o 0
232
233 [a-d[0-9][m-p]]
234 4
235 true 4 0
236
237 [a-d[0-9][m-p]]
238 e
239 false 0
240
241 [a-d[0-9][m-p]]
242 u
243 false 0
244
245 [[a-d][0-9][m-p]]
246 b
247 true b 0
248
249 [[a-d][0-9][m-p]]
250 z
251 false 0
252
253 [a-c[d-f[g-i]]]
254 a
255 true a 0
256
257 [a-c[d-f[g-i]]]
258 e
259 true e 0
260
261 [a-c[d-f[g-i]]]
262 h
263 true h 0
264
265 [a-c[d-f[g-i]]]
266 m
267 false 0
268
269 [a-c[d-f[g-i]]m]
270 m
271 true m 0
272
273 [abc[def]ghi]
274 a
275 true a 0
276
277 [abc[def]ghi]
278 d
279 true d 0
280
281 [abc[def]ghi]
282 h
283 true h 0
284
285 [abc[def]ghi]
286 w
287 false 0
288
289 [a-c&&[d-f]]
290 a
291 false 0
292
293 [a-c&&[d-f]]
294 e
295 false 0
296
297 [a-c&&[d-f]]
298 z
299 false 0
300
301 [[a-c]&&[d-f]]
302 a
303 false 0
304
305 [[a-c]&&[d-f]]
306 e
307 false 0
308
309 [[a-c]&&[d-f]]
310 z
311 false 0
312
313 [a-c&&d-f]
314 a
315 false 0
316
317 [a-m&&m-z]
318 m
319 true m 0
320
321 [a-m&&m-z&&a-c]
322 m
323 false 0
324
325 [a-m&&m-z&&a-z]
326 m
327 true m 0
328
329 [[a-m]&&[m-z]]
330 a
331 false 0
332
333 [[a-m]&&[m-z]]
334 m
335 true m 0
336
337 [[a-m]&&[m-z]]
338 z
339 false 0
340
341 [[a-m]&&[^a-c]]
342 a
343 false 0
344
345 [[a-m]&&[^a-c]]
346 d
347 true d 0
348
349 [a-m&&[^a-c]]
350 a
351 false 0
352
353 [a-m&&[^a-c]]
354 d
355 true d 0
356
357 [a-cd-f&&[d-f]]
358 a
359 false 0
360
361 [a-cd-f&&[d-f]]
362 e
363 true e 0
364
365 [[a-c]&&d-fa-c]
366 a
367 true a 0
368
369 [[a-c]&&[d-f][a-c]]
370 a
371 true a 0
372
373 [[a-c][d-f]&&abc]
374 a
375 true a 0
376
377 [[a-c][d-f]&&abc[def]]
378 e
379 true e 0
380
381 [[a-c]&&[b-d]&&[c-e]]
382 a
383 false 0
384
385 [[a-c]&&[b-d]&&[c-e]]
386 c
387 true c 0
388
389 [[a-c]&&[b-d][c-e]&&[u-z]]
390 c
391 false 0
392
393 [abc[^bcd]]
394 a
395 true a 0
396
397 [abc[^bcd]]
398 d
399 false 0
400
401 [a-c&&a-d&&a-eghi]
402 b
403 true b 0
404
405 [a-c&&a-d&&a-eghi]
406 g
407 false 0
408
409 [[a[b]]&&[b[a]]]
410 a
411 true a 0
412
413 [[a]&&[b][c][a]&&[^d]]
414 a
415 true a 0
416
417 [[a]&&[b][c][a]&&[^d]]
418 d
419 false 0
420
421 [[[a-d]&&[c-f]]]
422 a
423 false 0
424
425 [[[a-d]&&[c-f]]]
426 c
427 true c 0
428
429 [[[a-d]&&[c-f]]&&[c]]
430 c
431 true c 0
432
433 [[[a-d]&&[c-f]]&&[c]&&c]
434 c
435 true c 0
436
437 [[[a-d]&&[c-f]]&&[c]&&c&&c]
438 c
439 true c 0
440
441 [[[a-d]&&[c-f]]&&[c]&&c&&[cde]]
442 c
443 true c 0
444
445 [z[abc&&bcd]]
446 c
447 true c 0
448
449 [z[abc&&bcd]&&[u-z]]
450 z
451 true z 0
452
453 [x[abc&&bcd[z]]&&[u-z]]
454 z
455 false 0
456
457 [x[[wz]abc&&bcd[z]]&&[u-z]]
458 z
459 true z 0
460
461 [[abc]&&[def]abc]
462 a
463 true a 0
464
465 [[abc]&&[def]xyz[abc]]
466 a
467 true a 0
468
469 \pL
470 a
471 true a 0
472
473 \pL
474 7
475 false 0
476
477 \p{L}
478 a
479 true a 0
480
481 \p{LC}
482 a
483 true a 0
484
485 \p{LC}
486 A
487 true A 0
488
489 \p{IsL}
490 a
491 true a 0
492
493 \p{IsLC}
494 a
495 true a 0
496
497 \p{IsLC}
498 A
499 true A 0
500
501 \p{IsLC}
502 9
503 false 0
504
505 \P{IsLC}
506 9
507 true 9 0
508
509 // Guillemet left is initial quote punctuation
510 \p{Pi}
511 \u00ab
512 true \u00ab 0
513
514 \P{Pi}
515 \u00ac
516 true \u00ac 0
517
518 // Guillemet right is final quote punctuation
519 \p{IsPf}
520 \u00bb
521 true \u00bb 0
522
523 \p{P}
524 \u00bb
525 true \u00bb 0
526
527 \p{P}+
528 \u00bb
529 true \u00bb 0
530
531 \P{IsPf}
532 \u00bc
533 true \u00bc 0
534
535 \P{IsP}
536 \u00bc
537 true \u00bc 0
538
539 \p{L1}
540 \u00bc
541 true \u00bc 0
542
543 \p{L1}+
544 \u00bc
545 true \u00bc 0
546
547 \p{L1}
548 \u02bc
549 false 0
550
551 \p{ASCII}
552 a
553 true a 0
554
555 \p{IsASCII}
556 a
557 true a 0
558
559 \p{IsASCII}
560 \u0370
561 false 0
562
563 \pLbc
564 abc
565 true abc 0
566
567 a[r\p{InGreek}]c
568 a\u0370c
569 true a\u0370c 0
570
571 a\p{InGreek}
572 a\u0370
573 true a\u0370 0
574
575 a\P{InGreek}
576 a\u0370
577 false 0
578
579 a\P{InGreek}
580 ab
581 true ab 0
582
583 a{^InGreek}
584 -
585 error
586
587 a\p{^InGreek}
588 -
589 error
590
591 a\P{^InGreek}
592 -
593 error
594
595 a\p{InGreek}
596 a\u0370
597 true a\u0370 0
598
599 a[\p{InGreek}]c
600 a\u0370c
601 true a\u0370c 0
602
603 a[\P{InGreek}]c
604 a\u0370c
605 false 0
606
607 a[\P{InGreek}]c
608 abc
609 true abc 0
610
611 a[{^InGreek}]c
612 anc
613 true anc 0
614
615 a[{^InGreek}]c
616 azc
617 false 0
618
619 a[\p{^InGreek}]c
620 -
621 error
622
623 a[\P{^InGreek}]c
624 -
625 error
626
627 a[\p{InGreek}]
628 a\u0370
629 true a\u0370 0
630
631 a[r\p{InGreek}]c
632 arc
633 true arc 0
634
635 a[\p{InGreek}r]c
636 arc
637 true arc 0
638
639 a[r\p{InGreek}]c
640 arc
641 true arc 0
642
643 a[^\p{InGreek}]c
644 a\u0370c
645 false 0
646
647 a[^\P{InGreek}]c
648 a\u0370c
649 true a\u0370c 0
650
651 a[\p{InGreek}&&[^\u0370]]c
652 a\u0370c
653 false 0
654
655 // Test the dot metacharacter
656 a.c.+
657 a#c%&
658 true a#c%& 0
659
660 ab.
661 ab\n
662 false 0
663
664 (?s)ab.
665 ab\n
666 true ab\n 0
667
668 a[\p{L}&&[\P{InGreek}]]c
669 a\u6000c
670 true a\u6000c 0
671
672 a[\p{L}&&[\P{InGreek}]]c
673 arc
674 true arc 0
675
676 a[\p{L}&&[\P{InGreek}]]c
677 a\u0370c
678 false 0
679
680 a\p{InGreek}c
681 a\u0370c
682 true a\u0370c 0
683
684 a\p{Sc}
685 a$
686 true a$ 0
687
688 // Test the word char escape sequence
689 ab\wc
690 abcc
691 true abcc 0
692
693 \W\w\W
694 #r#
695 true #r# 0
696
697 \W\w\W
698 rrrr#ggg
699 false 0
700
701 abc[\w]
702 abcd
703 true abcd 0
704
705 abc[\sdef]*
706 abc def
707 true abc def 0
708
709 abc[\sy-z]*
710 abc y z
711 true abc y z 0
712
713 abc[a-d\sm-p]*
714 abcaa mn p
715 true abcaa mn p 0
716
717 // Test the whitespace escape sequence
718 ab\sc
719 ab c
720 true ab c 0
721
722 \s\s\s
723 blah err
724 false 0
725
726 \S\S\s
727 blah err
728 true ah 0
729
730 // Test the digit escape sequence
731 ab\dc
732 ab9c
733 true ab9c 0
734
735 \d\d\d
736 blah45
737 false 0
738
739 // Test the caret metacharacter
740 ^abc
741 abcdef
742 true abc 0
743
744 ^abc
745 bcdabc
746 false 0
747
748 // Greedy ? metacharacter
749 a?b
750 aaaab
751 true ab 0
752
753 a?b
754 b
755 true b 0
756
757 a?b
758 aaaccc
759 false 0
760
761 .?b
762 aaaab
763 true ab 0
764
765 // Reluctant ? metacharacter
766 a??b
767 aaaab
768 true ab 0
769
770 a??b
771 b
772 true b 0
773
774 a??b
775 aaaccc
776 false 0
777
778 .??b
779 aaaab
780 true ab 0
781
782 // Possessive ? metacharacter
783 a?+b
784 aaaab
785 true ab 0
786
787 a?+b
788 b
789 true b 0
790
791 a?+b
792 aaaccc
793 false 0
794
795 .?+b
796 aaaab
797 true ab 0
798
799 // Greedy + metacharacter
800 a+b
801 aaaab
802 true aaaab 0
803
804 a+b
805 b
806 false 0
807
808 a+b
809 aaaccc
810 false 0
811
812 .+b
813 aaaab
814 true aaaab 0
815
816 // Reluctant + metacharacter
817 a+?b
818 aaaab
819 true aaaab 0
820
821 a+?b
822 b
823 false 0
824
825 a+?b
826 aaaccc
827 false 0
828
829 .+?b
830 aaaab
831 true aaaab 0
832
833 // Possessive + metacharacter
834 a++b
835 aaaab
836 true aaaab 0
837
838 a++b
839 b
840 false 0
841
842 a++b
843 aaaccc
844 false 0
845
846 .++b
847 aaaab
848 false 0
849
850 // Greedy Repetition
851 a{2,3}
852 a
853 false 0
854
855 a{2,3}
856 aa
857 true aa 0
858
859 a{2,3}
860 aaa
861 true aaa 0
862
863 a{2,3}
864 aaaa
865 true aaa 0
866
867 a{3,}
868 zzzaaaazzz
869 true aaaa 0
870
871 a{3,}
872 zzzaazzz
873 false 0
874
875 // Reluctant Repetition
876 a{2,3}?
877 a
878 false 0
879
880 a{2,3}?
881 aa
882 true aa 0
883
884 a{2,3}?
885 aaa
886 true aa 0
887
888 a{2,3}?
889 aaaa
890 true aa 0
891
892 // Zero width Positive lookahead
893 abc(?=d)
894 zzzabcd
895 true abc 0
896
897 abc(?=d)
898 zzzabced
899 false 0
900
901 // Zero width Negative lookahead
902 abc(?!d)
903 zzabcd
904 false 0
905
906 abc(?!d)
907 zzabced
908 true abc 0
909
910 // Zero width Positive lookbehind
911 \w(?<=a)
912 ###abc###
913 true a 0
914
915 \w(?<=a)
916 ###ert###
917 false 0
918
919 // Zero width Negative lookbehind
920 (?<!a)\w
921 ###abc###
922 true a 0
923
924 (?<!a)c
925 bc
926 true c 0
927
928 (?<!a)c
929 ac
930 false 0
931
932 // Nondeterministic group
933 (a+b)+
934 ababab
935 true ababab 1 ab
936
937 (a|b)+
938 ccccd
939 false 1
940
941 // Deterministic group
942 (ab)+
943 ababab
944 true ababab 1 ab
945
946 (ab)+
947 accccd
948 false 1
949
950 (ab)*
951 ababab
952 true ababab 1 ab
953
954 (ab)(cd*)
955 zzzabczzz
956 true abc 2 ab c
957
958 abc(d)*abc
959 abcdddddabc
960 true abcdddddabc 1 d
961
962 // Escaped metacharacter
963 \*
964 *
965 true * 0
966
967 \\
968 \
969 true \ 0
970
971 \\
972 \\\\
973 true \ 0
974
975 // Back references
976 (a*)bc\1
977 zzzaabcaazzz
978 true aabcaa 1 aa
979
980 (a*)bc\1
981 zzzaabcazzz
982 true abca 1 a
983
984 (gt*)(dde)*(yu)\1\3(vv)
985 zzzgttddeddeyugttyuvvzzz
986 true gttddeddeyugttyuvv 4 gtt dde yu vv
987
988 // Greedy * metacharacter
989 a*b
990 aaaab
991 true aaaab 0
992
993 a*b
994 b
995 true b 0
996
997 a*b
998 aaaccc
999 false 0
1000
1001 .*b
1002 aaaab
1003 true aaaab 0
1004
1005 // Reluctant * metacharacter
1006 a*?b
1007 aaaab
1008 true aaaab 0
1009
1010 a*?b
1011 b
1012 true b 0
1013
1014 a*?b
1015 aaaccc
1016 false 0
1017
1018 .*?b
1019 aaaab
1020 true aaaab 0
1021
1022 // Possessive * metacharacter
1023 a*+b
1024 aaaab
1025 true aaaab 0
1026
1027 a*+b
1028 b
1029 true b 0
1030
1031 a*+b
1032 aaaccc
1033 false 0
1034
1035 .*+b
1036 aaaab
1037 false 0
1038
1039 // Case insensitivity
1040 (?i)foobar
1041 fOobAr
1042 true fOobAr 0
1043
1044 f(?i)oobar
1045 fOobAr
1046 true fOobAr 0
1047
1048 foo(?i)bar
1049 fOobAr
1050 false 0
1051
1052 (?i)foo[bar]+
1053 foObAr
1054 true foObAr 0
1055
1056 (?i)foo[a-r]+
1057 foObAr
1058 true foObAr 0
1059
1060 // Disable metacharacters- test both length <=3 and >3
1061 // So that the BM optimization is part of test
1062 \Q***\Eabc
1063 ***abc
1064 true ***abc 0
1065
1066 bl\Q***\Eabc
1067 bl***abc
1068 true bl***abc 0
1069
1070 \Q***abc
1071 ***abc
1072 true ***abc 0
1073
1074 blah\Q***\Eabc
1075 blah***abc
1076 true blah***abc 0
1077
1078 \Q***abc
1079 ***abc
1080 true ***abc 0
1081
1082 \Q*ab
1083 *ab
1084 true *ab 0
1085
1086 blah\Q***abc
1087 blah***abc
1088 true blah***abc 0
1089
1090 bla\Q***abc
1091 bla***abc
1092 true bla***abc 0
1093
1094 // Escapes in char classes
1095 [ab\Qdef\E]
1096 d
1097 true d 0
1098
1099 [ab\Q[\E]
1100 [
1101 true [ 0
1102
1103 [\Q]\E]
1104 ]
1105 true ] 0
1106
1107 [\Q\\E]
1108 \
1109 true \ 0
1110
1111 [\Q(\E]
1112 (
1113 true ( 0
1114
1115 [\n-#]
1116 !
1117 true ! 0
1118
1119 [\n-#]
1120 -
1121 false 0
1122
1123 [\w-#]
1124 !
1125 false 0
1126
1127 [\w-#]
1128 a
1129 true a 0
1130
1131 [\w-#]
1132 -
1133 true - 0
1134
1135 [\w-#]
1136 #
1137 true # 0
1138
1139 [\043]+
1140 blahblah#blech
1141 true # 0
1142
1143 [\042-\044]+
1144 blahblah#blech
1145 true # 0
1146
1147 [\u1234-\u1236]
1148 blahblah\u1235blech
1149 true \u1235 0
1150
1151 [^\043]*
1152 blahblah#blech
1153 true blahblah 0
1154
1155 (|f)?+
1156 foo
1157 true 1