blob: bc345eda8531fdeda219dc2ebf01c5d407faea6f [file] [log] [blame]
sherman0b4d42d2009-02-23 21:06:15 -08001/*
alanb0d058232012-11-02 15:50:11 +00002 * Copyright (c) 1999, 2012, Oracle and/or its affiliates. All rights reserved.
sherman0b4d42d2009-02-23 21:06:15 -08003 * 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
ohair2283b9d2010-05-25 15:58:33 -07007 * published by the Free Software Foundation. Oracle designates this
sherman0b4d42d2009-02-23 21:06:15 -08008 * particular file as subject to the "Classpath" exception as provided
ohair2283b9d2010-05-25 15:58:33 -07009 * by Oracle in the LICENSE file that accompanied this code.
sherman0b4d42d2009-02-23 21:06:15 -080010 *
11 * This code is distributed in the hope that it will be useful, but WITHOUT
12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14 * version 2 for more details (a copy is included in the LICENSE file that
15 * accompanied this code).
16 *
17 * You should have received a copy of the GNU General Public License version
18 * 2 along with this work; if not, write to the Free Software Foundation,
19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
20 *
ohair2283b9d2010-05-25 15:58:33 -070021 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
22 * or visit www.oracle.com if you need additional information or have any
23 * questions.
sherman0b4d42d2009-02-23 21:06:15 -080024 */
25
26/**
27 * @test
28 * @summary tests RegExp framework
29 * @author Mike McCloskey
30 * @bug 4481568 4482696 4495089 4504687 4527731 4599621 4631553 4619345
31 * 4630911 4672616 4711773 4727935 4750573 4792284 4803197 4757029 4808962
32 * 4872664 4803179 4892980 4900747 4945394 4938995 4979006 4994840 4997476
33 * 5013885 5003322 4988891 5098443 5110268 6173522 4829857 5027748 6376940
34 * 6358731 6178785 6284152 6231989 6497148 6486934 6233084 6504326 6635133
shermanb16229d2011-12-19 14:14:14 -080035 * 6350801 6676425 6878475 6919132 6931676 6948903 6990617 7014645 7039066
shermanf6f35a12013-04-26 13:59:10 -070036 * 7067045 7014640 7189363 8007395
sherman0b4d42d2009-02-23 21:06:15 -080037 */
38
39import java.util.regex.*;
40import java.util.Random;
41import java.io.*;
42import java.util.*;
43import java.nio.CharBuffer;
44
45/**
46 * This is a test class created to check the operation of
47 * the Pattern and Matcher classes.
48 */
49public class RegExTest {
50
51 private static Random generator = new Random();
52 private static boolean failure = false;
53 private static int failCount = 0;
shermanb16229d2011-12-19 14:14:14 -080054 private static String firstFailure = null;
sherman0b4d42d2009-02-23 21:06:15 -080055
56 /**
57 * Main to interpret arguments and run several tests.
58 *
59 */
60 public static void main(String[] args) throws Exception {
61 // Most of the tests are in a file
62 processFile("TestCases.txt");
63 //processFile("PerlCases.txt");
64 processFile("BMPTestCases.txt");
65 processFile("SupplementaryTestCases.txt");
66
67 // These test many randomly generated char patterns
68 bm();
69 slice();
70
71 // These are hard to put into the file
72 escapes();
73 blankInput();
74
75 // Substitition tests on randomly generated sequences
76 globalSubstitute();
77 stringbufferSubstitute();
78 substitutionBasher();
79
80 // Canonical Equivalence
81 ceTest();
82
83 // Anchors
84 anchorTest();
85
86 // boolean match calls
87 matchesTest();
88 lookingAtTest();
89
90 // Pattern API
91 patternMatchesTest();
92
93 // Misc
94 lookbehindTest();
95 nullArgumentTest();
96 backRefTest();
97 groupCaptureTest();
98 caretTest();
99 charClassTest();
100 emptyPatternTest();
101 findIntTest();
102 group0Test();
103 longPatternTest();
104 octalTest();
105 ampersandTest();
106 negationTest();
107 splitTest();
108 appendTest();
109 caseFoldingTest();
110 commentsTest();
111 unixLinesTest();
112 replaceFirstTest();
113 gTest();
114 zTest();
115 serializeTest();
116 reluctantRepetitionTest();
117 multilineDollarTest();
118 dollarAtEndTest();
119 caretBetweenTerminatorsTest();
120 // This RFE rejected in Tiger numOccurrencesTest();
121 javaCharClassTest();
122 nonCaptureRepetitionTest();
123 notCapturedGroupCurlyMatchTest();
124 escapedSegmentTest();
125 literalPatternTest();
126 literalReplacementTest();
127 regionTest();
128 toStringTest();
129 negatedCharClassTest();
130 findFromTest();
131 boundsTest();
132 unicodeWordBoundsTest();
133 caretAtEndTest();
134 wordSearchTest();
135 hitEndTest();
136 toMatchResultTest();
137 surrogatesInClassTest();
shermanb16229d2011-12-19 14:14:14 -0800138 removeQEQuotingTest();
sherman0b4d42d2009-02-23 21:06:15 -0800139 namedGroupCaptureTest();
sherman6782c962010-02-05 00:10:42 -0800140 nonBmpClassComplementTest();
shermancc01ef52010-05-18 15:36:47 -0700141 unicodePropertiesTest();
shermanf03c78b2011-02-03 13:49:25 -0800142 unicodeHexNotationTest();
sherman85bbd8b2011-04-28 20:48:36 -0700143 unicodeClassesTest();
shermanecb65472012-05-08 10:57:13 -0700144 horizontalAndVerticalWSTest();
145 linebreakTest();
sherman36e2c8f2012-08-09 10:15:26 -0700146 branchTest();
shermanf6f35a12013-04-26 13:59:10 -0700147 groupCurlyNotFoundSuppTest();
shermanb16229d2011-12-19 14:14:14 -0800148 if (failure) {
149 throw new
150 RuntimeException("RegExTest failed, 1st failure: " +
151 firstFailure);
152 } else {
sherman0b4d42d2009-02-23 21:06:15 -0800153 System.err.println("OKAY: All tests passed.");
shermanb16229d2011-12-19 14:14:14 -0800154 }
sherman0b4d42d2009-02-23 21:06:15 -0800155 }
156
157 // Utility functions
158
159 private static String getRandomAlphaString(int length) {
160 StringBuffer buf = new StringBuffer(length);
161 for (int i=0; i<length; i++) {
162 char randChar = (char)(97 + generator.nextInt(26));
163 buf.append(randChar);
164 }
165 return buf.toString();
166 }
167
168 private static void check(Matcher m, String expected) {
169 m.find();
170 if (!m.group().equals(expected))
171 failCount++;
172 }
173
174 private static void check(Matcher m, String result, boolean expected) {
175 m.find();
shermanf03c78b2011-02-03 13:49:25 -0800176 if (m.group().equals(result) != expected)
177 failCount++;
sherman0b4d42d2009-02-23 21:06:15 -0800178 }
179
180 private static void check(Pattern p, String s, boolean expected) {
shermanf03c78b2011-02-03 13:49:25 -0800181 if (p.matcher(s).find() != expected)
182 failCount++;
183 }
184
185 private static void check(String p, String s, boolean expected) {
186 Matcher matcher = Pattern.compile(p).matcher(s);
187 if (matcher.find() != expected)
188 failCount++;
sherman0b4d42d2009-02-23 21:06:15 -0800189 }
190
191 private static void check(String p, char c, boolean expected) {
192 String propertyPattern = expected ? "\\p" + p : "\\P" + p;
193 Pattern pattern = Pattern.compile(propertyPattern);
194 char[] ca = new char[1]; ca[0] = c;
195 Matcher matcher = pattern.matcher(new String(ca));
196 if (!matcher.find())
197 failCount++;
198 }
199
200 private static void check(String p, int codePoint, boolean expected) {
201 String propertyPattern = expected ? "\\p" + p : "\\P" + p;
202 Pattern pattern = Pattern.compile(propertyPattern);
203 char[] ca = Character.toChars(codePoint);
204 Matcher matcher = pattern.matcher(new String(ca));
205 if (!matcher.find())
206 failCount++;
207 }
208
209 private static void check(String p, int flag, String input, String s,
210 boolean expected)
211 {
212 Pattern pattern = Pattern.compile(p, flag);
213 Matcher matcher = pattern.matcher(input);
214 if (expected)
215 check(matcher, s, expected);
216 else
217 check(pattern, input, false);
218 }
219
220 private static void report(String testName) {
221 int spacesToAdd = 30 - testName.length();
222 StringBuffer paddedNameBuffer = new StringBuffer(testName);
223 for (int i=0; i<spacesToAdd; i++)
224 paddedNameBuffer.append(" ");
225 String paddedName = paddedNameBuffer.toString();
226 System.err.println(paddedName + ": " +
227 (failCount==0 ? "Passed":"Failed("+failCount+")"));
shermanb16229d2011-12-19 14:14:14 -0800228 if (failCount > 0) {
sherman0b4d42d2009-02-23 21:06:15 -0800229 failure = true;
shermanb16229d2011-12-19 14:14:14 -0800230
231 if (firstFailure == null) {
232 firstFailure = testName;
233 }
234 }
235
sherman0b4d42d2009-02-23 21:06:15 -0800236 failCount = 0;
237 }
238
239 /**
240 * Converts ASCII alphabet characters [A-Za-z] in the given 's' to
241 * supplementary characters. This method does NOT fully take care
242 * of the regex syntax.
243 */
244 private static String toSupplementaries(String s) {
245 int length = s.length();
246 StringBuffer sb = new StringBuffer(length * 2);
247
248 for (int i = 0; i < length; ) {
249 char c = s.charAt(i++);
250 if (c == '\\') {
251 sb.append(c);
252 if (i < length) {
253 c = s.charAt(i++);
254 sb.append(c);
255 if (c == 'u') {
256 // assume no syntax error
257 sb.append(s.charAt(i++));
258 sb.append(s.charAt(i++));
259 sb.append(s.charAt(i++));
260 sb.append(s.charAt(i++));
261 }
262 }
263 } else if ((c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z')) {
264 sb.append('\ud800').append((char)('\udc00'+c));
265 } else {
266 sb.append(c);
267 }
268 }
269 return sb.toString();
270 }
271
272 // Regular expression tests
273
274 // This is for bug 6178785
275 // Test if an expected NPE gets thrown when passing in a null argument
276 private static boolean check(Runnable test) {
277 try {
278 test.run();
279 failCount++;
280 return false;
281 } catch (NullPointerException npe) {
282 return true;
283 }
284 }
285
286 private static void nullArgumentTest() {
287 check(new Runnable() { public void run() { Pattern.compile(null); }});
288 check(new Runnable() { public void run() { Pattern.matches(null, null); }});
289 check(new Runnable() { public void run() { Pattern.matches("xyz", null);}});
290 check(new Runnable() { public void run() { Pattern.quote(null);}});
291 check(new Runnable() { public void run() { Pattern.compile("xyz").split(null);}});
292 check(new Runnable() { public void run() { Pattern.compile("xyz").matcher(null);}});
293
294 final Matcher m = Pattern.compile("xyz").matcher("xyz");
295 m.matches();
296 check(new Runnable() { public void run() { m.appendTail(null);}});
297 check(new Runnable() { public void run() { m.replaceAll(null);}});
298 check(new Runnable() { public void run() { m.replaceFirst(null);}});
299 check(new Runnable() { public void run() { m.appendReplacement(null, null);}});
300 check(new Runnable() { public void run() { m.reset(null);}});
301 check(new Runnable() { public void run() { Matcher.quoteReplacement(null);}});
302 //check(new Runnable() { public void run() { m.usePattern(null);}});
303
304 report("Null Argument");
305 }
306
307 // This is for bug6635133
308 // Test if surrogate pair in Unicode escapes can be handled correctly.
309 private static void surrogatesInClassTest() throws Exception {
310 Pattern pattern = Pattern.compile("[\\ud834\\udd21-\\ud834\\udd24]");
311 Matcher matcher = pattern.matcher("\ud834\udd22");
312 if (!matcher.find())
313 failCount++;
shermanb16229d2011-12-19 14:14:14 -0800314
315 report("Surrogate pair in Unicode escape");
316 }
317
318 // This is for bug6990617
319 // Test if Pattern.RemoveQEQuoting works correctly if the octal unicode
320 // char encoding is only 2 or 3 digits instead of 4 and the first quoted
321 // char is an octal digit.
322 private static void removeQEQuotingTest() throws Exception {
323 Pattern pattern =
324 Pattern.compile("\\011\\Q1sometext\\E\\011\\Q2sometext\\E");
325 Matcher matcher = pattern.matcher("\t1sometext\t2sometext");
326 if (!matcher.find())
327 failCount++;
328
329 report("Remove Q/E Quoting");
sherman0b4d42d2009-02-23 21:06:15 -0800330 }
331
332 // This is for bug 4988891
333 // Test toMatchResult to see that it is a copy of the Matcher
334 // that is not affected by subsequent operations on the original
335 private static void toMatchResultTest() throws Exception {
336 Pattern pattern = Pattern.compile("squid");
337 Matcher matcher = pattern.matcher(
338 "agiantsquidofdestinyasmallsquidoffate");
339 matcher.find();
340 int matcherStart1 = matcher.start();
341 MatchResult mr = matcher.toMatchResult();
342 if (mr == matcher)
343 failCount++;
344 int resultStart1 = mr.start();
345 if (matcherStart1 != resultStart1)
346 failCount++;
347 matcher.find();
348 int matcherStart2 = matcher.start();
349 int resultStart2 = mr.start();
350 if (matcherStart2 == resultStart2)
351 failCount++;
352 if (resultStart1 != resultStart2)
353 failCount++;
354 MatchResult mr2 = matcher.toMatchResult();
355 if (mr == mr2)
356 failCount++;
357 if (mr2.start() != matcherStart2)
358 failCount++;
359 report("toMatchResult is a copy");
360 }
361
362 // This is for bug 5013885
363 // Must test a slice to see if it reports hitEnd correctly
364 private static void hitEndTest() throws Exception {
365 // Basic test of Slice node
366 Pattern p = Pattern.compile("^squidattack");
367 Matcher m = p.matcher("squack");
368 m.find();
369 if (m.hitEnd())
370 failCount++;
371 m.reset("squid");
372 m.find();
373 if (!m.hitEnd())
374 failCount++;
375
376 // Test Slice, SliceA and SliceU nodes
377 for (int i=0; i<3; i++) {
378 int flags = 0;
379 if (i==1) flags = Pattern.CASE_INSENSITIVE;
380 if (i==2) flags = Pattern.UNICODE_CASE;
381 p = Pattern.compile("^abc", flags);
382 m = p.matcher("ad");
383 m.find();
384 if (m.hitEnd())
385 failCount++;
386 m.reset("ab");
387 m.find();
388 if (!m.hitEnd())
389 failCount++;
390 }
391
392 // Test Boyer-Moore node
393 p = Pattern.compile("catattack");
394 m = p.matcher("attack");
395 m.find();
396 if (!m.hitEnd())
397 failCount++;
398
399 p = Pattern.compile("catattack");
400 m = p.matcher("attackattackattackcatatta");
401 m.find();
402 if (!m.hitEnd())
403 failCount++;
sherman0b4d42d2009-02-23 21:06:15 -0800404 report("hitEnd from a Slice");
405 }
406
407 // This is for bug 4997476
408 // It is weird code submitted by customer demonstrating a regression
409 private static void wordSearchTest() throws Exception {
410 String testString = new String("word1 word2 word3");
411 Pattern p = Pattern.compile("\\b");
412 Matcher m = p.matcher(testString);
413 int position = 0;
414 int start = 0;
415 while (m.find(position)) {
416 start = m.start();
417 if (start == testString.length())
418 break;
419 if (m.find(start+1)) {
420 position = m.start();
421 } else {
422 position = testString.length();
423 }
424 if (testString.substring(start, position).equals(" "))
425 continue;
426 if (!testString.substring(start, position-1).startsWith("word"))
427 failCount++;
428 }
429 report("Customer word search");
430 }
431
432 // This is for bug 4994840
433 private static void caretAtEndTest() throws Exception {
434 // Problem only occurs with multiline patterns
435 // containing a beginning-of-line caret "^" followed
436 // by an expression that also matches the empty string.
437 Pattern pattern = Pattern.compile("^x?", Pattern.MULTILINE);
438 Matcher matcher = pattern.matcher("\r");
439 matcher.find();
440 matcher.find();
441 report("Caret at end");
442 }
443
444 // This test is for 4979006
445 // Check to see if word boundary construct properly handles unicode
446 // non spacing marks
447 private static void unicodeWordBoundsTest() throws Exception {
448 String spaces = " ";
449 String wordChar = "a";
450 String nsm = "\u030a";
451
452 assert (Character.getType('\u030a') == Character.NON_SPACING_MARK);
453
454 Pattern pattern = Pattern.compile("\\b");
455 Matcher matcher = pattern.matcher("");
456 // S=other B=word character N=non spacing mark .=word boundary
457 // SS.BB.SS
458 String input = spaces + wordChar + wordChar + spaces;
459 twoFindIndexes(input, matcher, 2, 4);
460 // SS.BBN.SS
461 input = spaces + wordChar +wordChar + nsm + spaces;
462 twoFindIndexes(input, matcher, 2, 5);
463 // SS.BN.SS
464 input = spaces + wordChar + nsm + spaces;
465 twoFindIndexes(input, matcher, 2, 4);
466 // SS.BNN.SS
467 input = spaces + wordChar + nsm + nsm + spaces;
468 twoFindIndexes(input, matcher, 2, 5);
469 // SSN.BB.SS
470 input = spaces + nsm + wordChar + wordChar + spaces;
471 twoFindIndexes(input, matcher, 3, 5);
472 // SS.BNB.SS
473 input = spaces + wordChar + nsm + wordChar + spaces;
474 twoFindIndexes(input, matcher, 2, 5);
475 // SSNNSS
476 input = spaces + nsm + nsm + spaces;
477 matcher.reset(input);
478 if (matcher.find())
479 failCount++;
480 // SSN.BBN.SS
481 input = spaces + nsm + wordChar + wordChar + nsm + spaces;
482 twoFindIndexes(input, matcher, 3, 6);
483
484 report("Unicode word boundary");
485 }
486
487 private static void twoFindIndexes(String input, Matcher matcher, int a,
488 int b) throws Exception
489 {
490 matcher.reset(input);
491 matcher.find();
492 if (matcher.start() != a)
493 failCount++;
494 matcher.find();
495 if (matcher.start() != b)
496 failCount++;
497 }
498
499 // This test is for 6284152
500 static void check(String regex, String input, String[] expected) {
501 List<String> result = new ArrayList<String>();
502 Pattern p = Pattern.compile(regex);
503 Matcher m = p.matcher(input);
504 while (m.find()) {
505 result.add(m.group());
506 }
507 if (!Arrays.asList(expected).equals(result))
508 failCount++;
509 }
510
511 private static void lookbehindTest() throws Exception {
512 //Positive
513 check("(?<=%.{0,5})foo\\d",
514 "%foo1\n%bar foo2\n%bar foo3\n%blahblah foo4\nfoo5",
515 new String[]{"foo1", "foo2", "foo3"});
516
517 //boundary at end of the lookbehind sub-regex should work consistently
518 //with the boundary just after the lookbehind sub-regex
519 check("(?<=.*\\b)foo", "abcd foo", new String[]{"foo"});
520 check("(?<=.*)\\bfoo", "abcd foo", new String[]{"foo"});
521 check("(?<!abc )\\bfoo", "abc foo", new String[0]);
522 check("(?<!abc \\b)foo", "abc foo", new String[0]);
523
524 //Negative
525 check("(?<!%.{0,5})foo\\d",
526 "%foo1\n%bar foo2\n%bar foo3\n%blahblah foo4\nfoo5",
527 new String[] {"foo4", "foo5"});
528
529 //Positive greedy
530 check("(?<=%b{1,4})foo", "%bbbbfoo", new String[] {"foo"});
531
532 //Positive reluctant
533 check("(?<=%b{1,4}?)foo", "%bbbbfoo", new String[] {"foo"});
534
535 //supplementary
536 check("(?<=%b{1,4})fo\ud800\udc00o", "%bbbbfo\ud800\udc00o",
537 new String[] {"fo\ud800\udc00o"});
538 check("(?<=%b{1,4}?)fo\ud800\udc00o", "%bbbbfo\ud800\udc00o",
539 new String[] {"fo\ud800\udc00o"});
540 check("(?<!%b{1,4})fo\ud800\udc00o", "%afo\ud800\udc00o",
541 new String[] {"fo\ud800\udc00o"});
542 check("(?<!%b{1,4}?)fo\ud800\udc00o", "%afo\ud800\udc00o",
543 new String[] {"fo\ud800\udc00o"});
544 report("Lookbehind");
545 }
546
547 // This test is for 4938995
548 // Check to see if weak region boundaries are transparent to
549 // lookahead and lookbehind constructs
550 private static void boundsTest() throws Exception {
551 String fullMessage = "catdogcat";
552 Pattern pattern = Pattern.compile("(?<=cat)dog(?=cat)");
553 Matcher matcher = pattern.matcher("catdogca");
554 matcher.useTransparentBounds(true);
555 if (matcher.find())
556 failCount++;
557 matcher.reset("atdogcat");
558 if (matcher.find())
559 failCount++;
560 matcher.reset(fullMessage);
561 if (!matcher.find())
562 failCount++;
563 matcher.reset(fullMessage);
564 matcher.region(0,9);
565 if (!matcher.find())
566 failCount++;
567 matcher.reset(fullMessage);
568 matcher.region(0,6);
569 if (!matcher.find())
570 failCount++;
571 matcher.reset(fullMessage);
572 matcher.region(3,6);
573 if (!matcher.find())
574 failCount++;
575 matcher.useTransparentBounds(false);
576 if (matcher.find())
577 failCount++;
578
579 // Negative lookahead/lookbehind
580 pattern = Pattern.compile("(?<!cat)dog(?!cat)");
581 matcher = pattern.matcher("dogcat");
582 matcher.useTransparentBounds(true);
583 matcher.region(0,3);
584 if (matcher.find())
585 failCount++;
586 matcher.reset("catdog");
587 matcher.region(3,6);
588 if (matcher.find())
589 failCount++;
590 matcher.useTransparentBounds(false);
591 matcher.reset("dogcat");
592 matcher.region(0,3);
593 if (!matcher.find())
594 failCount++;
595 matcher.reset("catdog");
596 matcher.region(3,6);
597 if (!matcher.find())
598 failCount++;
599
600 report("Region bounds transparency");
601 }
602
603 // This test is for 4945394
604 private static void findFromTest() throws Exception {
605 String message = "This is 40 $0 message.";
606 Pattern pat = Pattern.compile("\\$0");
607 Matcher match = pat.matcher(message);
608 if (!match.find())
609 failCount++;
610 if (match.find())
611 failCount++;
612 if (match.find())
613 failCount++;
614 report("Check for alternating find");
615 }
616
617 // This test is for 4872664 and 4892980
618 private static void negatedCharClassTest() throws Exception {
619 Pattern pattern = Pattern.compile("[^>]");
620 Matcher matcher = pattern.matcher("\u203A");
621 if (!matcher.matches())
622 failCount++;
623 pattern = Pattern.compile("[^fr]");
624 matcher = pattern.matcher("a");
625 if (!matcher.find())
626 failCount++;
627 matcher.reset("\u203A");
628 if (!matcher.find())
629 failCount++;
630 String s = "for";
631 String result[] = s.split("[^fr]");
632 if (!result[0].equals("f"))
633 failCount++;
634 if (!result[1].equals("r"))
635 failCount++;
636 s = "f\u203Ar";
637 result = s.split("[^fr]");
638 if (!result[0].equals("f"))
639 failCount++;
640 if (!result[1].equals("r"))
641 failCount++;
642
643 // Test adding to bits, subtracting a node, then adding to bits again
644 pattern = Pattern.compile("[^f\u203Ar]");
645 matcher = pattern.matcher("a");
646 if (!matcher.find())
647 failCount++;
648 matcher.reset("f");
649 if (matcher.find())
650 failCount++;
651 matcher.reset("\u203A");
652 if (matcher.find())
653 failCount++;
654 matcher.reset("r");
655 if (matcher.find())
656 failCount++;
657 matcher.reset("\u203B");
658 if (!matcher.find())
659 failCount++;
660
661 // Test subtracting a node, adding to bits, subtracting again
662 pattern = Pattern.compile("[^\u203Ar\u203B]");
663 matcher = pattern.matcher("a");
664 if (!matcher.find())
665 failCount++;
666 matcher.reset("\u203A");
667 if (matcher.find())
668 failCount++;
669 matcher.reset("r");
670 if (matcher.find())
671 failCount++;
672 matcher.reset("\u203B");
673 if (matcher.find())
674 failCount++;
675 matcher.reset("\u203C");
676 if (!matcher.find())
677 failCount++;
678
679 report("Negated Character Class");
680 }
681
682 // This test is for 4628291
683 private static void toStringTest() throws Exception {
684 Pattern pattern = Pattern.compile("b+");
685 if (pattern.toString() != "b+")
686 failCount++;
687 Matcher matcher = pattern.matcher("aaabbbccc");
688 String matcherString = matcher.toString(); // unspecified
689 matcher.find();
690 matcherString = matcher.toString(); // unspecified
691 matcher.region(0,3);
692 matcherString = matcher.toString(); // unspecified
693 matcher.reset();
694 matcherString = matcher.toString(); // unspecified
695 report("toString");
696 }
697
698 // This test is for 4808962
699 private static void literalPatternTest() throws Exception {
700 int flags = Pattern.LITERAL;
701
702 Pattern pattern = Pattern.compile("abc\\t$^", flags);
703 check(pattern, "abc\\t$^", true);
704
705 pattern = Pattern.compile(Pattern.quote("abc\\t$^"));
706 check(pattern, "abc\\t$^", true);
707
708 pattern = Pattern.compile("\\Qa^$bcabc\\E", flags);
709 check(pattern, "\\Qa^$bcabc\\E", true);
710 check(pattern, "a^$bcabc", false);
711
712 pattern = Pattern.compile("\\\\Q\\\\E");
713 check(pattern, "\\Q\\E", true);
714
715 pattern = Pattern.compile("\\Qabc\\Eefg\\\\Q\\\\Ehij");
716 check(pattern, "abcefg\\Q\\Ehij", true);
717
718 pattern = Pattern.compile("\\\\\\Q\\\\E");
719 check(pattern, "\\\\\\\\", true);
720
721 pattern = Pattern.compile(Pattern.quote("\\Qa^$bcabc\\E"));
722 check(pattern, "\\Qa^$bcabc\\E", true);
723 check(pattern, "a^$bcabc", false);
724
725 pattern = Pattern.compile(Pattern.quote("\\Qabc\\Edef"));
726 check(pattern, "\\Qabc\\Edef", true);
727 check(pattern, "abcdef", false);
728
729 pattern = Pattern.compile(Pattern.quote("abc\\Edef"));
730 check(pattern, "abc\\Edef", true);
731 check(pattern, "abcdef", false);
732
733 pattern = Pattern.compile(Pattern.quote("\\E"));
734 check(pattern, "\\E", true);
735
736 pattern = Pattern.compile("((((abc.+?:)", flags);
737 check(pattern, "((((abc.+?:)", true);
738
739 flags |= Pattern.MULTILINE;
740
741 pattern = Pattern.compile("^cat$", flags);
742 check(pattern, "abc^cat$def", true);
743 check(pattern, "cat", false);
744
745 flags |= Pattern.CASE_INSENSITIVE;
746
747 pattern = Pattern.compile("abcdef", flags);
748 check(pattern, "ABCDEF", true);
749 check(pattern, "AbCdEf", true);
750
751 flags |= Pattern.DOTALL;
752
753 pattern = Pattern.compile("a...b", flags);
754 check(pattern, "A...b", true);
755 check(pattern, "Axxxb", false);
756
757 flags |= Pattern.CANON_EQ;
758
759 Pattern p = Pattern.compile("testa\u030a", flags);
760 check(pattern, "testa\u030a", false);
761 check(pattern, "test\u00e5", false);
762
763 // Supplementary character test
764 flags = Pattern.LITERAL;
765
766 pattern = Pattern.compile(toSupplementaries("abc\\t$^"), flags);
767 check(pattern, toSupplementaries("abc\\t$^"), true);
768
769 pattern = Pattern.compile(Pattern.quote(toSupplementaries("abc\\t$^")));
770 check(pattern, toSupplementaries("abc\\t$^"), true);
771
772 pattern = Pattern.compile(toSupplementaries("\\Qa^$bcabc\\E"), flags);
773 check(pattern, toSupplementaries("\\Qa^$bcabc\\E"), true);
774 check(pattern, toSupplementaries("a^$bcabc"), false);
775
776 pattern = Pattern.compile(Pattern.quote(toSupplementaries("\\Qa^$bcabc\\E")));
777 check(pattern, toSupplementaries("\\Qa^$bcabc\\E"), true);
778 check(pattern, toSupplementaries("a^$bcabc"), false);
779
780 pattern = Pattern.compile(Pattern.quote(toSupplementaries("\\Qabc\\Edef")));
781 check(pattern, toSupplementaries("\\Qabc\\Edef"), true);
782 check(pattern, toSupplementaries("abcdef"), false);
783
784 pattern = Pattern.compile(Pattern.quote(toSupplementaries("abc\\Edef")));
785 check(pattern, toSupplementaries("abc\\Edef"), true);
786 check(pattern, toSupplementaries("abcdef"), false);
787
788 pattern = Pattern.compile(toSupplementaries("((((abc.+?:)"), flags);
789 check(pattern, toSupplementaries("((((abc.+?:)"), true);
790
791 flags |= Pattern.MULTILINE;
792
793 pattern = Pattern.compile(toSupplementaries("^cat$"), flags);
794 check(pattern, toSupplementaries("abc^cat$def"), true);
795 check(pattern, toSupplementaries("cat"), false);
796
797 flags |= Pattern.DOTALL;
798
799 // note: this is case-sensitive.
800 pattern = Pattern.compile(toSupplementaries("a...b"), flags);
801 check(pattern, toSupplementaries("a...b"), true);
802 check(pattern, toSupplementaries("axxxb"), false);
803
804 flags |= Pattern.CANON_EQ;
805
806 String t = toSupplementaries("test");
807 p = Pattern.compile(t + "a\u030a", flags);
808 check(pattern, t + "a\u030a", false);
809 check(pattern, t + "\u00e5", false);
810
811 report("Literal pattern");
812 }
813
814 // This test is for 4803179
815 // This test is also for 4808962, replacement parts
816 private static void literalReplacementTest() throws Exception {
817 int flags = Pattern.LITERAL;
818
819 Pattern pattern = Pattern.compile("abc", flags);
820 Matcher matcher = pattern.matcher("zzzabczzz");
821 String replaceTest = "$0";
822 String result = matcher.replaceAll(replaceTest);
823 if (!result.equals("zzzabczzz"))
824 failCount++;
825
826 matcher.reset();
827 String literalReplacement = matcher.quoteReplacement(replaceTest);
828 result = matcher.replaceAll(literalReplacement);
829 if (!result.equals("zzz$0zzz"))
830 failCount++;
831
832 matcher.reset();
833 replaceTest = "\\t$\\$";
834 literalReplacement = matcher.quoteReplacement(replaceTest);
835 result = matcher.replaceAll(literalReplacement);
836 if (!result.equals("zzz\\t$\\$zzz"))
837 failCount++;
838
839 // Supplementary character test
840 pattern = Pattern.compile(toSupplementaries("abc"), flags);
841 matcher = pattern.matcher(toSupplementaries("zzzabczzz"));
842 replaceTest = "$0";
843 result = matcher.replaceAll(replaceTest);
844 if (!result.equals(toSupplementaries("zzzabczzz")))
845 failCount++;
846
847 matcher.reset();
848 literalReplacement = matcher.quoteReplacement(replaceTest);
849 result = matcher.replaceAll(literalReplacement);
850 if (!result.equals(toSupplementaries("zzz$0zzz")))
851 failCount++;
852
853 matcher.reset();
854 replaceTest = "\\t$\\$";
855 literalReplacement = matcher.quoteReplacement(replaceTest);
856 result = matcher.replaceAll(literalReplacement);
857 if (!result.equals(toSupplementaries("zzz\\t$\\$zzz")))
858 failCount++;
859
sherman5c8f3492012-04-12 15:01:41 -0700860 // IAE should be thrown if backslash or '$' is the last character
861 // in replacement string
862 try {
863 "\uac00".replaceAll("\uac00", "$");
shermanecb65472012-05-08 10:57:13 -0700864 failCount++;
865 } catch (IllegalArgumentException iie) {
866 } catch (Exception e) {
867 failCount++;
868 }
869 try {
sherman5c8f3492012-04-12 15:01:41 -0700870 "\uac00".replaceAll("\uac00", "\\");
871 failCount++;
872 } catch (IllegalArgumentException iie) {
873 } catch (Exception e) {
874 failCount++;
875 }
sherman0b4d42d2009-02-23 21:06:15 -0800876 report("Literal replacement");
877 }
878
879 // This test is for 4757029
880 private static void regionTest() throws Exception {
881 Pattern pattern = Pattern.compile("abc");
882 Matcher matcher = pattern.matcher("abcdefabc");
883
884 matcher.region(0,9);
885 if (!matcher.find())
886 failCount++;
887 if (!matcher.find())
888 failCount++;
889 matcher.region(0,3);
890 if (!matcher.find())
891 failCount++;
892 matcher.region(3,6);
893 if (matcher.find())
894 failCount++;
895 matcher.region(0,2);
896 if (matcher.find())
897 failCount++;
898
899 expectRegionFail(matcher, 1, -1);
900 expectRegionFail(matcher, -1, -1);
901 expectRegionFail(matcher, -1, 1);
902 expectRegionFail(matcher, 5, 3);
903 expectRegionFail(matcher, 5, 12);
904 expectRegionFail(matcher, 12, 12);
905
906 pattern = Pattern.compile("^abc$");
907 matcher = pattern.matcher("zzzabczzz");
908 matcher.region(0,9);
909 if (matcher.find())
910 failCount++;
911 matcher.region(3,6);
912 if (!matcher.find())
913 failCount++;
914 matcher.region(3,6);
915 matcher.useAnchoringBounds(false);
916 if (matcher.find())
917 failCount++;
918
919 // Supplementary character test
920 pattern = Pattern.compile(toSupplementaries("abc"));
921 matcher = pattern.matcher(toSupplementaries("abcdefabc"));
922 matcher.region(0,9*2);
923 if (!matcher.find())
924 failCount++;
925 if (!matcher.find())
926 failCount++;
927 matcher.region(0,3*2);
928 if (!matcher.find())
929 failCount++;
930 matcher.region(1,3*2);
931 if (matcher.find())
932 failCount++;
933 matcher.region(3*2,6*2);
934 if (matcher.find())
935 failCount++;
936 matcher.region(0,2*2);
937 if (matcher.find())
938 failCount++;
939 matcher.region(0,2*2+1);
940 if (matcher.find())
941 failCount++;
942
943 expectRegionFail(matcher, 1*2, -1);
944 expectRegionFail(matcher, -1, -1);
945 expectRegionFail(matcher, -1, 1*2);
946 expectRegionFail(matcher, 5*2, 3*2);
947 expectRegionFail(matcher, 5*2, 12*2);
948 expectRegionFail(matcher, 12*2, 12*2);
949
950 pattern = Pattern.compile(toSupplementaries("^abc$"));
951 matcher = pattern.matcher(toSupplementaries("zzzabczzz"));
952 matcher.region(0,9*2);
953 if (matcher.find())
954 failCount++;
955 matcher.region(3*2,6*2);
956 if (!matcher.find())
957 failCount++;
958 matcher.region(3*2+1,6*2);
959 if (matcher.find())
960 failCount++;
961 matcher.region(3*2,6*2-1);
962 if (matcher.find())
963 failCount++;
964 matcher.region(3*2,6*2);
965 matcher.useAnchoringBounds(false);
966 if (matcher.find())
967 failCount++;
968 report("Regions");
969 }
970
971 private static void expectRegionFail(Matcher matcher, int index1,
972 int index2)
973 {
974 try {
975 matcher.region(index1, index2);
976 failCount++;
977 } catch (IndexOutOfBoundsException ioobe) {
978 // Correct result
979 } catch (IllegalStateException ise) {
980 // Correct result
981 }
982 }
983
984 // This test is for 4803197
985 private static void escapedSegmentTest() throws Exception {
986
987 Pattern pattern = Pattern.compile("\\Qdir1\\dir2\\E");
988 check(pattern, "dir1\\dir2", true);
989
990 pattern = Pattern.compile("\\Qdir1\\dir2\\\\E");
991 check(pattern, "dir1\\dir2\\", true);
992
993 pattern = Pattern.compile("(\\Qdir1\\dir2\\\\E)");
994 check(pattern, "dir1\\dir2\\", true);
995
996 // Supplementary character test
997 pattern = Pattern.compile(toSupplementaries("\\Qdir1\\dir2\\E"));
998 check(pattern, toSupplementaries("dir1\\dir2"), true);
999
1000 pattern = Pattern.compile(toSupplementaries("\\Qdir1\\dir2")+"\\\\E");
1001 check(pattern, toSupplementaries("dir1\\dir2\\"), true);
1002
1003 pattern = Pattern.compile(toSupplementaries("(\\Qdir1\\dir2")+"\\\\E)");
1004 check(pattern, toSupplementaries("dir1\\dir2\\"), true);
1005
1006 report("Escaped segment");
1007 }
1008
1009 // This test is for 4792284
1010 private static void nonCaptureRepetitionTest() throws Exception {
1011 String input = "abcdefgh;";
1012
1013 String[] patterns = new String[] {
1014 "(?:\\w{4})+;",
1015 "(?:\\w{8})*;",
1016 "(?:\\w{2}){2,4};",
1017 "(?:\\w{4}){2,};", // only matches the
1018 ".*?(?:\\w{5})+;", // specified minimum
1019 ".*?(?:\\w{9})*;", // number of reps - OK
1020 "(?:\\w{4})+?;", // lazy repetition - OK
1021 "(?:\\w{4})++;", // possessive repetition - OK
1022 "(?:\\w{2,}?)+;", // non-deterministic - OK
1023 "(\\w{4})+;", // capturing group - OK
1024 };
1025
1026 for (int i = 0; i < patterns.length; i++) {
1027 // Check find()
1028 check(patterns[i], 0, input, input, true);
1029 // Check matches()
1030 Pattern p = Pattern.compile(patterns[i]);
1031 Matcher m = p.matcher(input);
1032
1033 if (m.matches()) {
1034 if (!m.group(0).equals(input))
1035 failCount++;
1036 } else {
1037 failCount++;
1038 }
1039 }
1040
1041 report("Non capturing repetition");
1042 }
1043
1044 // This test is for 6358731
1045 private static void notCapturedGroupCurlyMatchTest() throws Exception {
1046 Pattern pattern = Pattern.compile("(abc)+|(abcd)+");
1047 Matcher matcher = pattern.matcher("abcd");
1048 if (!matcher.matches() ||
1049 matcher.group(1) != null ||
1050 !matcher.group(2).equals("abcd")) {
1051 failCount++;
1052 }
1053 report("Not captured GroupCurly");
1054 }
1055
1056 // This test is for 4706545
1057 private static void javaCharClassTest() throws Exception {
1058 for (int i=0; i<1000; i++) {
1059 char c = (char)generator.nextInt();
1060 check("{javaLowerCase}", c, Character.isLowerCase(c));
1061 check("{javaUpperCase}", c, Character.isUpperCase(c));
1062 check("{javaUpperCase}+", c, Character.isUpperCase(c));
1063 check("{javaTitleCase}", c, Character.isTitleCase(c));
1064 check("{javaDigit}", c, Character.isDigit(c));
1065 check("{javaDefined}", c, Character.isDefined(c));
1066 check("{javaLetter}", c, Character.isLetter(c));
1067 check("{javaLetterOrDigit}", c, Character.isLetterOrDigit(c));
1068 check("{javaJavaIdentifierStart}", c,
1069 Character.isJavaIdentifierStart(c));
1070 check("{javaJavaIdentifierPart}", c,
1071 Character.isJavaIdentifierPart(c));
1072 check("{javaUnicodeIdentifierStart}", c,
1073 Character.isUnicodeIdentifierStart(c));
1074 check("{javaUnicodeIdentifierPart}", c,
1075 Character.isUnicodeIdentifierPart(c));
1076 check("{javaIdentifierIgnorable}", c,
1077 Character.isIdentifierIgnorable(c));
1078 check("{javaSpaceChar}", c, Character.isSpaceChar(c));
1079 check("{javaWhitespace}", c, Character.isWhitespace(c));
1080 check("{javaISOControl}", c, Character.isISOControl(c));
1081 check("{javaMirrored}", c, Character.isMirrored(c));
1082
1083 }
1084
1085 // Supplementary character test
1086 for (int i=0; i<1000; i++) {
1087 int c = generator.nextInt(Character.MAX_CODE_POINT
1088 - Character.MIN_SUPPLEMENTARY_CODE_POINT)
1089 + Character.MIN_SUPPLEMENTARY_CODE_POINT;
1090 check("{javaLowerCase}", c, Character.isLowerCase(c));
1091 check("{javaUpperCase}", c, Character.isUpperCase(c));
1092 check("{javaUpperCase}+", c, Character.isUpperCase(c));
1093 check("{javaTitleCase}", c, Character.isTitleCase(c));
1094 check("{javaDigit}", c, Character.isDigit(c));
1095 check("{javaDefined}", c, Character.isDefined(c));
1096 check("{javaLetter}", c, Character.isLetter(c));
1097 check("{javaLetterOrDigit}", c, Character.isLetterOrDigit(c));
1098 check("{javaJavaIdentifierStart}", c,
1099 Character.isJavaIdentifierStart(c));
1100 check("{javaJavaIdentifierPart}", c,
1101 Character.isJavaIdentifierPart(c));
1102 check("{javaUnicodeIdentifierStart}", c,
1103 Character.isUnicodeIdentifierStart(c));
1104 check("{javaUnicodeIdentifierPart}", c,
1105 Character.isUnicodeIdentifierPart(c));
1106 check("{javaIdentifierIgnorable}", c,
1107 Character.isIdentifierIgnorable(c));
1108 check("{javaSpaceChar}", c, Character.isSpaceChar(c));
1109 check("{javaWhitespace}", c, Character.isWhitespace(c));
1110 check("{javaISOControl}", c, Character.isISOControl(c));
1111 check("{javaMirrored}", c, Character.isMirrored(c));
1112 }
1113
1114 report("Java character classes");
1115 }
1116
1117 // This test is for 4523620
1118 /*
1119 private static void numOccurrencesTest() throws Exception {
1120 Pattern pattern = Pattern.compile("aaa");
1121
1122 if (pattern.numOccurrences("aaaaaa", false) != 2)
1123 failCount++;
1124 if (pattern.numOccurrences("aaaaaa", true) != 4)
1125 failCount++;
1126
1127 pattern = Pattern.compile("^");
1128 if (pattern.numOccurrences("aaaaaa", false) != 1)
1129 failCount++;
1130 if (pattern.numOccurrences("aaaaaa", true) != 1)
1131 failCount++;
1132
1133 report("Number of Occurrences");
1134 }
1135 */
1136
1137 // This test is for 4776374
1138 private static void caretBetweenTerminatorsTest() throws Exception {
1139 int flags1 = Pattern.DOTALL;
1140 int flags2 = Pattern.DOTALL | Pattern.UNIX_LINES;
1141 int flags3 = Pattern.DOTALL | Pattern.UNIX_LINES | Pattern.MULTILINE;
1142 int flags4 = Pattern.DOTALL | Pattern.MULTILINE;
1143
1144 check("^....", flags1, "test\ntest", "test", true);
1145 check(".....^", flags1, "test\ntest", "test", false);
1146 check(".....^", flags1, "test\n", "test", false);
1147 check("....^", flags1, "test\r\n", "test", false);
1148
1149 check("^....", flags2, "test\ntest", "test", true);
1150 check("....^", flags2, "test\ntest", "test", false);
1151 check(".....^", flags2, "test\n", "test", false);
1152 check("....^", flags2, "test\r\n", "test", false);
1153
1154 check("^....", flags3, "test\ntest", "test", true);
1155 check(".....^", flags3, "test\ntest", "test\n", true);
1156 check(".....^", flags3, "test\u0085test", "test\u0085", false);
1157 check(".....^", flags3, "test\n", "test", false);
1158 check(".....^", flags3, "test\r\n", "test", false);
1159 check("......^", flags3, "test\r\ntest", "test\r\n", true);
1160
1161 check("^....", flags4, "test\ntest", "test", true);
1162 check(".....^", flags3, "test\ntest", "test\n", true);
1163 check(".....^", flags4, "test\u0085test", "test\u0085", true);
1164 check(".....^", flags4, "test\n", "test\n", false);
1165 check(".....^", flags4, "test\r\n", "test\r", false);
1166
1167 // Supplementary character test
1168 String t = toSupplementaries("test");
1169 check("^....", flags1, t+"\n"+t, t, true);
1170 check(".....^", flags1, t+"\n"+t, t, false);
1171 check(".....^", flags1, t+"\n", t, false);
1172 check("....^", flags1, t+"\r\n", t, false);
1173
1174 check("^....", flags2, t+"\n"+t, t, true);
1175 check("....^", flags2, t+"\n"+t, t, false);
1176 check(".....^", flags2, t+"\n", t, false);
1177 check("....^", flags2, t+"\r\n", t, false);
1178
1179 check("^....", flags3, t+"\n"+t, t, true);
1180 check(".....^", flags3, t+"\n"+t, t+"\n", true);
1181 check(".....^", flags3, t+"\u0085"+t, t+"\u0085", false);
1182 check(".....^", flags3, t+"\n", t, false);
1183 check(".....^", flags3, t+"\r\n", t, false);
1184 check("......^", flags3, t+"\r\n"+t, t+"\r\n", true);
1185
1186 check("^....", flags4, t+"\n"+t, t, true);
1187 check(".....^", flags3, t+"\n"+t, t+"\n", true);
1188 check(".....^", flags4, t+"\u0085"+t, t+"\u0085", true);
1189 check(".....^", flags4, t+"\n", t+"\n", false);
1190 check(".....^", flags4, t+"\r\n", t+"\r", false);
1191
1192 report("Caret between terminators");
1193 }
1194
1195 // This test is for 4727935
1196 private static void dollarAtEndTest() throws Exception {
1197 int flags1 = Pattern.DOTALL;
1198 int flags2 = Pattern.DOTALL | Pattern.UNIX_LINES;
1199 int flags3 = Pattern.DOTALL | Pattern.MULTILINE;
1200
1201 check("....$", flags1, "test\n", "test", true);
1202 check("....$", flags1, "test\r\n", "test", true);
1203 check(".....$", flags1, "test\n", "test\n", true);
1204 check(".....$", flags1, "test\u0085", "test\u0085", true);
1205 check("....$", flags1, "test\u0085", "test", true);
1206
1207 check("....$", flags2, "test\n", "test", true);
1208 check(".....$", flags2, "test\n", "test\n", true);
1209 check(".....$", flags2, "test\u0085", "test\u0085", true);
1210 check("....$", flags2, "test\u0085", "est\u0085", true);
1211
1212 check("....$.blah", flags3, "test\nblah", "test\nblah", true);
1213 check(".....$.blah", flags3, "test\n\nblah", "test\n\nblah", true);
1214 check("....$blah", flags3, "test\nblah", "!!!!", false);
1215 check(".....$blah", flags3, "test\nblah", "!!!!", false);
1216
1217 // Supplementary character test
1218 String t = toSupplementaries("test");
1219 String b = toSupplementaries("blah");
1220 check("....$", flags1, t+"\n", t, true);
1221 check("....$", flags1, t+"\r\n", t, true);
1222 check(".....$", flags1, t+"\n", t+"\n", true);
1223 check(".....$", flags1, t+"\u0085", t+"\u0085", true);
1224 check("....$", flags1, t+"\u0085", t, true);
1225
1226 check("....$", flags2, t+"\n", t, true);
1227 check(".....$", flags2, t+"\n", t+"\n", true);
1228 check(".....$", flags2, t+"\u0085", t+"\u0085", true);
1229 check("....$", flags2, t+"\u0085", toSupplementaries("est\u0085"), true);
1230
1231 check("....$."+b, flags3, t+"\n"+b, t+"\n"+b, true);
1232 check(".....$."+b, flags3, t+"\n\n"+b, t+"\n\n"+b, true);
1233 check("....$"+b, flags3, t+"\n"+b, "!!!!", false);
1234 check(".....$"+b, flags3, t+"\n"+b, "!!!!", false);
1235
1236 report("Dollar at End");
1237 }
1238
1239 // This test is for 4711773
1240 private static void multilineDollarTest() throws Exception {
1241 Pattern findCR = Pattern.compile("$", Pattern.MULTILINE);
1242 Matcher matcher = findCR.matcher("first bit\nsecond bit");
1243 matcher.find();
1244 if (matcher.start(0) != 9)
1245 failCount++;
1246 matcher.find();
1247 if (matcher.start(0) != 20)
1248 failCount++;
1249
1250 // Supplementary character test
1251 matcher = findCR.matcher(toSupplementaries("first bit\n second bit")); // double BMP chars
1252 matcher.find();
1253 if (matcher.start(0) != 9*2)
1254 failCount++;
1255 matcher.find();
1256 if (matcher.start(0) != 20*2)
1257 failCount++;
1258
1259 report("Multiline Dollar");
1260 }
1261
1262 private static void reluctantRepetitionTest() throws Exception {
1263 Pattern p = Pattern.compile("1(\\s\\S+?){1,3}?[\\s,]2");
1264 check(p, "1 word word word 2", true);
1265 check(p, "1 wor wo w 2", true);
1266 check(p, "1 word word 2", true);
1267 check(p, "1 word 2", true);
1268 check(p, "1 wo w w 2", true);
1269 check(p, "1 wo w 2", true);
1270 check(p, "1 wor w 2", true);
1271
1272 p = Pattern.compile("([a-z])+?c");
1273 Matcher m = p.matcher("ababcdefdec");
1274 check(m, "ababc");
1275
1276 // Supplementary character test
1277 p = Pattern.compile(toSupplementaries("([a-z])+?c"));
1278 m = p.matcher(toSupplementaries("ababcdefdec"));
1279 check(m, toSupplementaries("ababc"));
1280
1281 report("Reluctant Repetition");
1282 }
1283
1284 private static void serializeTest() throws Exception {
1285 String patternStr = "(b)";
1286 String matchStr = "b";
1287 Pattern pattern = Pattern.compile(patternStr);
1288 ByteArrayOutputStream baos = new ByteArrayOutputStream();
1289 ObjectOutputStream oos = new ObjectOutputStream(baos);
1290 oos.writeObject(pattern);
1291 oos.close();
1292 ObjectInputStream ois = new ObjectInputStream(
1293 new ByteArrayInputStream(baos.toByteArray()));
1294 Pattern serializedPattern = (Pattern)ois.readObject();
1295 ois.close();
1296 Matcher matcher = serializedPattern.matcher(matchStr);
1297 if (!matcher.matches())
1298 failCount++;
1299 if (matcher.groupCount() != 1)
1300 failCount++;
1301
1302 report("Serialization");
1303 }
1304
1305 private static void gTest() {
1306 Pattern pattern = Pattern.compile("\\G\\w");
1307 Matcher matcher = pattern.matcher("abc#x#x");
1308 matcher.find();
1309 matcher.find();
1310 matcher.find();
1311 if (matcher.find())
1312 failCount++;
1313
1314 pattern = Pattern.compile("\\GA*");
1315 matcher = pattern.matcher("1A2AA3");
1316 matcher.find();
1317 if (matcher.find())
1318 failCount++;
1319
1320 pattern = Pattern.compile("\\GA*");
1321 matcher = pattern.matcher("1A2AA3");
1322 if (!matcher.find(1))
1323 failCount++;
1324 matcher.find();
1325 if (matcher.find())
1326 failCount++;
1327
1328 report("\\G");
1329 }
1330
1331 private static void zTest() {
1332 Pattern pattern = Pattern.compile("foo\\Z");
1333 // Positives
1334 check(pattern, "foo\u0085", true);
1335 check(pattern, "foo\u2028", true);
1336 check(pattern, "foo\u2029", true);
1337 check(pattern, "foo\n", true);
1338 check(pattern, "foo\r", true);
1339 check(pattern, "foo\r\n", true);
1340 // Negatives
1341 check(pattern, "fooo", false);
1342 check(pattern, "foo\n\r", false);
1343
1344 pattern = Pattern.compile("foo\\Z", Pattern.UNIX_LINES);
1345 // Positives
1346 check(pattern, "foo", true);
1347 check(pattern, "foo\n", true);
1348 // Negatives
1349 check(pattern, "foo\r", false);
1350 check(pattern, "foo\u0085", false);
1351 check(pattern, "foo\u2028", false);
1352 check(pattern, "foo\u2029", false);
1353
1354 report("\\Z");
1355 }
1356
1357 private static void replaceFirstTest() {
1358 Pattern pattern = Pattern.compile("(ab)(c*)");
1359 Matcher matcher = pattern.matcher("abccczzzabcczzzabccc");
1360 if (!matcher.replaceFirst("test").equals("testzzzabcczzzabccc"))
1361 failCount++;
1362
1363 matcher.reset("zzzabccczzzabcczzzabccczzz");
1364 if (!matcher.replaceFirst("test").equals("zzztestzzzabcczzzabccczzz"))
1365 failCount++;
1366
1367 matcher.reset("zzzabccczzzabcczzzabccczzz");
1368 String result = matcher.replaceFirst("$1");
1369 if (!result.equals("zzzabzzzabcczzzabccczzz"))
1370 failCount++;
1371
1372 matcher.reset("zzzabccczzzabcczzzabccczzz");
1373 result = matcher.replaceFirst("$2");
1374 if (!result.equals("zzzccczzzabcczzzabccczzz"))
1375 failCount++;
1376
1377 pattern = Pattern.compile("a*");
1378 matcher = pattern.matcher("aaaaaaaaaa");
1379 if (!matcher.replaceFirst("test").equals("test"))
1380 failCount++;
1381
1382 pattern = Pattern.compile("a+");
1383 matcher = pattern.matcher("zzzaaaaaaaaaa");
1384 if (!matcher.replaceFirst("test").equals("zzztest"))
1385 failCount++;
1386
1387 // Supplementary character test
1388 pattern = Pattern.compile(toSupplementaries("(ab)(c*)"));
1389 matcher = pattern.matcher(toSupplementaries("abccczzzabcczzzabccc"));
1390 if (!matcher.replaceFirst(toSupplementaries("test"))
1391 .equals(toSupplementaries("testzzzabcczzzabccc")))
1392 failCount++;
1393
1394 matcher.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
1395 if (!matcher.replaceFirst(toSupplementaries("test")).
1396 equals(toSupplementaries("zzztestzzzabcczzzabccczzz")))
1397 failCount++;
1398
1399 matcher.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
1400 result = matcher.replaceFirst("$1");
1401 if (!result.equals(toSupplementaries("zzzabzzzabcczzzabccczzz")))
1402 failCount++;
1403
1404 matcher.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
1405 result = matcher.replaceFirst("$2");
1406 if (!result.equals(toSupplementaries("zzzccczzzabcczzzabccczzz")))
1407 failCount++;
1408
1409 pattern = Pattern.compile(toSupplementaries("a*"));
1410 matcher = pattern.matcher(toSupplementaries("aaaaaaaaaa"));
1411 if (!matcher.replaceFirst(toSupplementaries("test")).equals(toSupplementaries("test")))
1412 failCount++;
1413
1414 pattern = Pattern.compile(toSupplementaries("a+"));
1415 matcher = pattern.matcher(toSupplementaries("zzzaaaaaaaaaa"));
1416 if (!matcher.replaceFirst(toSupplementaries("test")).equals(toSupplementaries("zzztest")))
1417 failCount++;
1418
1419 report("Replace First");
1420 }
1421
1422 private static void unixLinesTest() {
1423 Pattern pattern = Pattern.compile(".*");
1424 Matcher matcher = pattern.matcher("aa\u2028blah");
1425 matcher.find();
1426 if (!matcher.group(0).equals("aa"))
1427 failCount++;
1428
1429 pattern = Pattern.compile(".*", Pattern.UNIX_LINES);
1430 matcher = pattern.matcher("aa\u2028blah");
1431 matcher.find();
1432 if (!matcher.group(0).equals("aa\u2028blah"))
1433 failCount++;
1434
1435 pattern = Pattern.compile("[az]$",
1436 Pattern.MULTILINE | Pattern.UNIX_LINES);
1437 matcher = pattern.matcher("aa\u2028zz");
1438 check(matcher, "a\u2028", false);
1439
1440 // Supplementary character test
1441 pattern = Pattern.compile(".*");
1442 matcher = pattern.matcher(toSupplementaries("aa\u2028blah"));
1443 matcher.find();
1444 if (!matcher.group(0).equals(toSupplementaries("aa")))
1445 failCount++;
1446
1447 pattern = Pattern.compile(".*", Pattern.UNIX_LINES);
1448 matcher = pattern.matcher(toSupplementaries("aa\u2028blah"));
1449 matcher.find();
1450 if (!matcher.group(0).equals(toSupplementaries("aa\u2028blah")))
1451 failCount++;
1452
1453 pattern = Pattern.compile(toSupplementaries("[az]$"),
1454 Pattern.MULTILINE | Pattern.UNIX_LINES);
1455 matcher = pattern.matcher(toSupplementaries("aa\u2028zz"));
1456 check(matcher, toSupplementaries("a\u2028"), false);
1457
1458 report("Unix Lines");
1459 }
1460
1461 private static void commentsTest() {
1462 int flags = Pattern.COMMENTS;
1463
1464 Pattern pattern = Pattern.compile("aa \\# aa", flags);
1465 Matcher matcher = pattern.matcher("aa#aa");
1466 if (!matcher.matches())
1467 failCount++;
1468
1469 pattern = Pattern.compile("aa # blah", flags);
1470 matcher = pattern.matcher("aa");
1471 if (!matcher.matches())
1472 failCount++;
1473
1474 pattern = Pattern.compile("aa blah", flags);
1475 matcher = pattern.matcher("aablah");
1476 if (!matcher.matches())
1477 failCount++;
1478
1479 pattern = Pattern.compile("aa # blah blech ", flags);
1480 matcher = pattern.matcher("aa");
1481 if (!matcher.matches())
1482 failCount++;
1483
1484 pattern = Pattern.compile("aa # blah\n ", flags);
1485 matcher = pattern.matcher("aa");
1486 if (!matcher.matches())
1487 failCount++;
1488
1489 pattern = Pattern.compile("aa # blah\nbc # blech", flags);
1490 matcher = pattern.matcher("aabc");
1491 if (!matcher.matches())
1492 failCount++;
1493
1494 pattern = Pattern.compile("aa # blah\nbc# blech", flags);
1495 matcher = pattern.matcher("aabc");
1496 if (!matcher.matches())
1497 failCount++;
1498
1499 pattern = Pattern.compile("aa # blah\nbc\\# blech", flags);
1500 matcher = pattern.matcher("aabc#blech");
1501 if (!matcher.matches())
1502 failCount++;
1503
1504 // Supplementary character test
1505 pattern = Pattern.compile(toSupplementaries("aa \\# aa"), flags);
1506 matcher = pattern.matcher(toSupplementaries("aa#aa"));
1507 if (!matcher.matches())
1508 failCount++;
1509
1510 pattern = Pattern.compile(toSupplementaries("aa # blah"), flags);
1511 matcher = pattern.matcher(toSupplementaries("aa"));
1512 if (!matcher.matches())
1513 failCount++;
1514
1515 pattern = Pattern.compile(toSupplementaries("aa blah"), flags);
1516 matcher = pattern.matcher(toSupplementaries("aablah"));
1517 if (!matcher.matches())
1518 failCount++;
1519
1520 pattern = Pattern.compile(toSupplementaries("aa # blah blech "), flags);
1521 matcher = pattern.matcher(toSupplementaries("aa"));
1522 if (!matcher.matches())
1523 failCount++;
1524
1525 pattern = Pattern.compile(toSupplementaries("aa # blah\n "), flags);
1526 matcher = pattern.matcher(toSupplementaries("aa"));
1527 if (!matcher.matches())
1528 failCount++;
1529
1530 pattern = Pattern.compile(toSupplementaries("aa # blah\nbc # blech"), flags);
1531 matcher = pattern.matcher(toSupplementaries("aabc"));
1532 if (!matcher.matches())
1533 failCount++;
1534
1535 pattern = Pattern.compile(toSupplementaries("aa # blah\nbc# blech"), flags);
1536 matcher = pattern.matcher(toSupplementaries("aabc"));
1537 if (!matcher.matches())
1538 failCount++;
1539
1540 pattern = Pattern.compile(toSupplementaries("aa # blah\nbc\\# blech"), flags);
1541 matcher = pattern.matcher(toSupplementaries("aabc#blech"));
1542 if (!matcher.matches())
1543 failCount++;
1544
1545 report("Comments");
1546 }
1547
1548 private static void caseFoldingTest() { // bug 4504687
1549 int flags = Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE;
1550 Pattern pattern = Pattern.compile("aa", flags);
1551 Matcher matcher = pattern.matcher("ab");
1552 if (matcher.matches())
1553 failCount++;
1554
1555 pattern = Pattern.compile("aA", flags);
1556 matcher = pattern.matcher("ab");
1557 if (matcher.matches())
1558 failCount++;
1559
1560 pattern = Pattern.compile("aa", flags);
1561 matcher = pattern.matcher("aB");
1562 if (matcher.matches())
1563 failCount++;
1564 matcher = pattern.matcher("Ab");
1565 if (matcher.matches())
1566 failCount++;
1567
1568 // ASCII "a"
1569 // Latin-1 Supplement "a" + grave
1570 // Cyrillic "a"
1571 String[] patterns = new String[] {
1572 //single
1573 "a", "\u00e0", "\u0430",
1574 //slice
1575 "ab", "\u00e0\u00e1", "\u0430\u0431",
1576 //class single
1577 "[a]", "[\u00e0]", "[\u0430]",
1578 //class range
1579 "[a-b]", "[\u00e0-\u00e5]", "[\u0430-\u0431]",
1580 //back reference
1581 "(a)\\1", "(\u00e0)\\1", "(\u0430)\\1"
1582 };
1583
1584 String[] texts = new String[] {
1585 "A", "\u00c0", "\u0410",
1586 "AB", "\u00c0\u00c1", "\u0410\u0411",
1587 "A", "\u00c0", "\u0410",
1588 "B", "\u00c2", "\u0411",
1589 "aA", "\u00e0\u00c0", "\u0430\u0410"
1590 };
1591
1592 boolean[] expected = new boolean[] {
1593 true, false, false,
1594 true, false, false,
1595 true, false, false,
1596 true, false, false,
1597 true, false, false
1598 };
1599
1600 flags = Pattern.CASE_INSENSITIVE;
1601 for (int i = 0; i < patterns.length; i++) {
1602 pattern = Pattern.compile(patterns[i], flags);
1603 matcher = pattern.matcher(texts[i]);
1604 if (matcher.matches() != expected[i]) {
1605 System.out.println("<1> Failed at " + i);
1606 failCount++;
1607 }
1608 }
1609
1610 flags = Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE;
1611 for (int i = 0; i < patterns.length; i++) {
1612 pattern = Pattern.compile(patterns[i], flags);
1613 matcher = pattern.matcher(texts[i]);
1614 if (!matcher.matches()) {
1615 System.out.println("<2> Failed at " + i);
1616 failCount++;
1617 }
1618 }
1619 // flag unicode_case alone should do nothing
1620 flags = Pattern.UNICODE_CASE;
1621 for (int i = 0; i < patterns.length; i++) {
1622 pattern = Pattern.compile(patterns[i], flags);
1623 matcher = pattern.matcher(texts[i]);
1624 if (matcher.matches()) {
1625 System.out.println("<3> Failed at " + i);
1626 failCount++;
1627 }
1628 }
1629
1630 // Special cases: i, I, u+0131 and u+0130
1631 flags = Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE;
1632 pattern = Pattern.compile("[h-j]+", flags);
1633 if (!pattern.matcher("\u0131\u0130").matches())
1634 failCount++;
1635 report("Case Folding");
1636 }
1637
1638 private static void appendTest() {
1639 Pattern pattern = Pattern.compile("(ab)(cd)");
1640 Matcher matcher = pattern.matcher("abcd");
1641 String result = matcher.replaceAll("$2$1");
1642 if (!result.equals("cdab"))
1643 failCount++;
1644
1645 String s1 = "Swap all: first = 123, second = 456";
1646 String s2 = "Swap one: first = 123, second = 456";
1647 String r = "$3$2$1";
1648 pattern = Pattern.compile("([a-z]+)( *= *)([0-9]+)");
1649 matcher = pattern.matcher(s1);
1650
1651 result = matcher.replaceAll(r);
1652 if (!result.equals("Swap all: 123 = first, 456 = second"))
1653 failCount++;
1654
1655 matcher = pattern.matcher(s2);
1656
1657 if (matcher.find()) {
1658 StringBuffer sb = new StringBuffer();
1659 matcher.appendReplacement(sb, r);
1660 matcher.appendTail(sb);
1661 result = sb.toString();
1662 if (!result.equals("Swap one: 123 = first, second = 456"))
1663 failCount++;
1664 }
1665
1666 // Supplementary character test
1667 pattern = Pattern.compile(toSupplementaries("(ab)(cd)"));
1668 matcher = pattern.matcher(toSupplementaries("abcd"));
1669 result = matcher.replaceAll("$2$1");
1670 if (!result.equals(toSupplementaries("cdab")))
1671 failCount++;
1672
1673 s1 = toSupplementaries("Swap all: first = 123, second = 456");
1674 s2 = toSupplementaries("Swap one: first = 123, second = 456");
1675 r = toSupplementaries("$3$2$1");
1676 pattern = Pattern.compile(toSupplementaries("([a-z]+)( *= *)([0-9]+)"));
1677 matcher = pattern.matcher(s1);
1678
1679 result = matcher.replaceAll(r);
1680 if (!result.equals(toSupplementaries("Swap all: 123 = first, 456 = second")))
1681 failCount++;
1682
1683 matcher = pattern.matcher(s2);
1684
1685 if (matcher.find()) {
1686 StringBuffer sb = new StringBuffer();
1687 matcher.appendReplacement(sb, r);
1688 matcher.appendTail(sb);
1689 result = sb.toString();
1690 if (!result.equals(toSupplementaries("Swap one: 123 = first, second = 456")))
1691 failCount++;
1692 }
1693 report("Append");
1694 }
1695
1696 private static void splitTest() {
1697 Pattern pattern = Pattern.compile(":");
1698 String[] result = pattern.split("foo:and:boo", 2);
1699 if (!result[0].equals("foo"))
1700 failCount++;
1701 if (!result[1].equals("and:boo"))
1702 failCount++;
1703 // Supplementary character test
1704 Pattern patternX = Pattern.compile(toSupplementaries("X"));
1705 result = patternX.split(toSupplementaries("fooXandXboo"), 2);
1706 if (!result[0].equals(toSupplementaries("foo")))
1707 failCount++;
1708 if (!result[1].equals(toSupplementaries("andXboo")))
1709 failCount++;
1710
1711 CharBuffer cb = CharBuffer.allocate(100);
1712 cb.put("foo:and:boo");
1713 cb.flip();
1714 result = pattern.split(cb);
1715 if (!result[0].equals("foo"))
1716 failCount++;
1717 if (!result[1].equals("and"))
1718 failCount++;
1719 if (!result[2].equals("boo"))
1720 failCount++;
1721
1722 // Supplementary character test
1723 CharBuffer cbs = CharBuffer.allocate(100);
1724 cbs.put(toSupplementaries("fooXandXboo"));
1725 cbs.flip();
1726 result = patternX.split(cbs);
1727 if (!result[0].equals(toSupplementaries("foo")))
1728 failCount++;
1729 if (!result[1].equals(toSupplementaries("and")))
1730 failCount++;
1731 if (!result[2].equals(toSupplementaries("boo")))
1732 failCount++;
1733
1734 String source = "0123456789";
1735 for (int limit=-2; limit<3; limit++) {
1736 for (int x=0; x<10; x++) {
1737 result = source.split(Integer.toString(x), limit);
1738 int expectedLength = limit < 1 ? 2 : limit;
1739
1740 if ((limit == 0) && (x == 9)) {
1741 // expected dropping of ""
1742 if (result.length != 1)
1743 failCount++;
1744 if (!result[0].equals("012345678")) {
1745 failCount++;
1746 }
1747 } else {
1748 if (result.length != expectedLength) {
1749 failCount++;
1750 }
1751 if (!result[0].equals(source.substring(0,x))) {
1752 if (limit != 1) {
1753 failCount++;
1754 } else {
1755 if (!result[0].equals(source.substring(0,10))) {
1756 failCount++;
1757 }
1758 }
1759 }
1760 if (expectedLength > 1) { // Check segment 2
1761 if (!result[1].equals(source.substring(x+1,10)))
1762 failCount++;
1763 }
1764 }
1765 }
1766 }
1767 // Check the case for no match found
1768 for (int limit=-2; limit<3; limit++) {
1769 result = source.split("e", limit);
1770 if (result.length != 1)
1771 failCount++;
1772 if (!result[0].equals(source))
1773 failCount++;
1774 }
1775 // Check the case for limit == 0, source = "";
1776 source = "";
1777 result = source.split("e", 0);
1778 if (result.length != 1)
1779 failCount++;
1780 if (!result[0].equals(source))
1781 failCount++;
1782
1783 report("Split");
1784 }
1785
1786 private static void negationTest() {
1787 Pattern pattern = Pattern.compile("[\\[@^]+");
1788 Matcher matcher = pattern.matcher("@@@@[[[[^^^^");
1789 if (!matcher.find())
1790 failCount++;
1791 if (!matcher.group(0).equals("@@@@[[[[^^^^"))
1792 failCount++;
1793 pattern = Pattern.compile("[@\\[^]+");
1794 matcher = pattern.matcher("@@@@[[[[^^^^");
1795 if (!matcher.find())
1796 failCount++;
1797 if (!matcher.group(0).equals("@@@@[[[[^^^^"))
1798 failCount++;
1799 pattern = Pattern.compile("[@\\[^@]+");
1800 matcher = pattern.matcher("@@@@[[[[^^^^");
1801 if (!matcher.find())
1802 failCount++;
1803 if (!matcher.group(0).equals("@@@@[[[[^^^^"))
1804 failCount++;
1805
1806 pattern = Pattern.compile("\\)");
1807 matcher = pattern.matcher("xxx)xxx");
1808 if (!matcher.find())
1809 failCount++;
1810
1811 report("Negation");
1812 }
1813
1814 private static void ampersandTest() {
1815 Pattern pattern = Pattern.compile("[&@]+");
1816 check(pattern, "@@@@&&&&", true);
1817
1818 pattern = Pattern.compile("[@&]+");
1819 check(pattern, "@@@@&&&&", true);
1820
1821 pattern = Pattern.compile("[@\\&]+");
1822 check(pattern, "@@@@&&&&", true);
1823
1824 report("Ampersand");
1825 }
1826
1827 private static void octalTest() throws Exception {
1828 Pattern pattern = Pattern.compile("\\u0007");
1829 Matcher matcher = pattern.matcher("\u0007");
1830 if (!matcher.matches())
1831 failCount++;
1832 pattern = Pattern.compile("\\07");
1833 matcher = pattern.matcher("\u0007");
1834 if (!matcher.matches())
1835 failCount++;
1836 pattern = Pattern.compile("\\007");
1837 matcher = pattern.matcher("\u0007");
1838 if (!matcher.matches())
1839 failCount++;
1840 pattern = Pattern.compile("\\0007");
1841 matcher = pattern.matcher("\u0007");
1842 if (!matcher.matches())
1843 failCount++;
1844 pattern = Pattern.compile("\\040");
1845 matcher = pattern.matcher("\u0020");
1846 if (!matcher.matches())
1847 failCount++;
1848 pattern = Pattern.compile("\\0403");
1849 matcher = pattern.matcher("\u00203");
1850 if (!matcher.matches())
1851 failCount++;
1852 pattern = Pattern.compile("\\0103");
1853 matcher = pattern.matcher("\u0043");
1854 if (!matcher.matches())
1855 failCount++;
1856
1857 report("Octal");
1858 }
1859
1860 private static void longPatternTest() throws Exception {
1861 try {
1862 Pattern pattern = Pattern.compile(
1863 "a 32-character-long pattern xxxx");
1864 pattern = Pattern.compile("a 33-character-long pattern xxxxx");
1865 pattern = Pattern.compile("a thirty four character long regex");
1866 StringBuffer patternToBe = new StringBuffer(101);
1867 for (int i=0; i<100; i++)
1868 patternToBe.append((char)(97 + i%26));
1869 pattern = Pattern.compile(patternToBe.toString());
1870 } catch (PatternSyntaxException e) {
1871 failCount++;
1872 }
1873
1874 // Supplementary character test
1875 try {
1876 Pattern pattern = Pattern.compile(
1877 toSupplementaries("a 32-character-long pattern xxxx"));
1878 pattern = Pattern.compile(toSupplementaries("a 33-character-long pattern xxxxx"));
1879 pattern = Pattern.compile(toSupplementaries("a thirty four character long regex"));
1880 StringBuffer patternToBe = new StringBuffer(101*2);
1881 for (int i=0; i<100; i++)
1882 patternToBe.append(Character.toChars(Character.MIN_SUPPLEMENTARY_CODE_POINT
1883 + 97 + i%26));
1884 pattern = Pattern.compile(patternToBe.toString());
1885 } catch (PatternSyntaxException e) {
1886 failCount++;
1887 }
1888 report("LongPattern");
1889 }
1890
1891 private static void group0Test() throws Exception {
1892 Pattern pattern = Pattern.compile("(tes)ting");
1893 Matcher matcher = pattern.matcher("testing");
1894 check(matcher, "testing");
1895
1896 matcher.reset("testing");
1897 if (matcher.lookingAt()) {
1898 if (!matcher.group(0).equals("testing"))
1899 failCount++;
1900 } else {
1901 failCount++;
1902 }
1903
1904 matcher.reset("testing");
1905 if (matcher.matches()) {
1906 if (!matcher.group(0).equals("testing"))
1907 failCount++;
1908 } else {
1909 failCount++;
1910 }
1911
1912 pattern = Pattern.compile("(tes)ting");
1913 matcher = pattern.matcher("testing");
1914 if (matcher.lookingAt()) {
1915 if (!matcher.group(0).equals("testing"))
1916 failCount++;
1917 } else {
1918 failCount++;
1919 }
1920
1921 pattern = Pattern.compile("^(tes)ting");
1922 matcher = pattern.matcher("testing");
1923 if (matcher.matches()) {
1924 if (!matcher.group(0).equals("testing"))
1925 failCount++;
1926 } else {
1927 failCount++;
1928 }
1929
1930 // Supplementary character test
1931 pattern = Pattern.compile(toSupplementaries("(tes)ting"));
1932 matcher = pattern.matcher(toSupplementaries("testing"));
1933 check(matcher, toSupplementaries("testing"));
1934
1935 matcher.reset(toSupplementaries("testing"));
1936 if (matcher.lookingAt()) {
1937 if (!matcher.group(0).equals(toSupplementaries("testing")))
1938 failCount++;
1939 } else {
1940 failCount++;
1941 }
1942
1943 matcher.reset(toSupplementaries("testing"));
1944 if (matcher.matches()) {
1945 if (!matcher.group(0).equals(toSupplementaries("testing")))
1946 failCount++;
1947 } else {
1948 failCount++;
1949 }
1950
1951 pattern = Pattern.compile(toSupplementaries("(tes)ting"));
1952 matcher = pattern.matcher(toSupplementaries("testing"));
1953 if (matcher.lookingAt()) {
1954 if (!matcher.group(0).equals(toSupplementaries("testing")))
1955 failCount++;
1956 } else {
1957 failCount++;
1958 }
1959
1960 pattern = Pattern.compile(toSupplementaries("^(tes)ting"));
1961 matcher = pattern.matcher(toSupplementaries("testing"));
1962 if (matcher.matches()) {
1963 if (!matcher.group(0).equals(toSupplementaries("testing")))
1964 failCount++;
1965 } else {
1966 failCount++;
1967 }
1968
1969 report("Group0");
1970 }
1971
1972 private static void findIntTest() throws Exception {
1973 Pattern p = Pattern.compile("blah");
1974 Matcher m = p.matcher("zzzzblahzzzzzblah");
1975 boolean result = m.find(2);
1976 if (!result)
1977 failCount++;
1978
1979 p = Pattern.compile("$");
1980 m = p.matcher("1234567890");
1981 result = m.find(10);
1982 if (!result)
1983 failCount++;
1984 try {
1985 result = m.find(11);
1986 failCount++;
1987 } catch (IndexOutOfBoundsException e) {
1988 // correct result
1989 }
1990
1991 // Supplementary character test
1992 p = Pattern.compile(toSupplementaries("blah"));
1993 m = p.matcher(toSupplementaries("zzzzblahzzzzzblah"));
1994 result = m.find(2);
1995 if (!result)
1996 failCount++;
1997
1998 report("FindInt");
1999 }
2000
2001 private static void emptyPatternTest() throws Exception {
2002 Pattern p = Pattern.compile("");
2003 Matcher m = p.matcher("foo");
2004
2005 // Should find empty pattern at beginning of input
2006 boolean result = m.find();
2007 if (result != true)
2008 failCount++;
2009 if (m.start() != 0)
2010 failCount++;
2011
2012 // Should not match entire input if input is not empty
2013 m.reset();
2014 result = m.matches();
2015 if (result == true)
2016 failCount++;
2017
2018 try {
2019 m.start(0);
2020 failCount++;
2021 } catch (IllegalStateException e) {
2022 // Correct result
2023 }
2024
2025 // Should match entire input if input is empty
2026 m.reset("");
2027 result = m.matches();
2028 if (result != true)
2029 failCount++;
2030
2031 result = Pattern.matches("", "");
2032 if (result != true)
2033 failCount++;
2034
2035 result = Pattern.matches("", "foo");
2036 if (result == true)
2037 failCount++;
2038 report("EmptyPattern");
2039 }
2040
2041 private static void charClassTest() throws Exception {
2042 Pattern pattern = Pattern.compile("blah[ab]]blech");
2043 check(pattern, "blahb]blech", true);
2044
2045 pattern = Pattern.compile("[abc[def]]");
2046 check(pattern, "b", true);
2047
2048 // Supplementary character tests
2049 pattern = Pattern.compile(toSupplementaries("blah[ab]]blech"));
2050 check(pattern, toSupplementaries("blahb]blech"), true);
2051
2052 pattern = Pattern.compile(toSupplementaries("[abc[def]]"));
2053 check(pattern, toSupplementaries("b"), true);
2054
2055 try {
2056 // u00ff when UNICODE_CASE
2057 pattern = Pattern.compile("[ab\u00ffcd]",
2058 Pattern.CASE_INSENSITIVE|
2059 Pattern.UNICODE_CASE);
2060 check(pattern, "ab\u00ffcd", true);
2061 check(pattern, "Ab\u0178Cd", true);
2062
2063 // u00b5 when UNICODE_CASE
2064 pattern = Pattern.compile("[ab\u00b5cd]",
2065 Pattern.CASE_INSENSITIVE|
2066 Pattern.UNICODE_CASE);
2067 check(pattern, "ab\u00b5cd", true);
2068 check(pattern, "Ab\u039cCd", true);
2069 } catch (Exception e) { failCount++; }
2070
2071 /* Special cases
2072 (1)LatinSmallLetterLongS u+017f
2073 (2)LatinSmallLetterDotlessI u+0131
2074 (3)LatineCapitalLetterIWithDotAbove u+0130
2075 (4)KelvinSign u+212a
2076 (5)AngstromSign u+212b
2077 */
2078 int flags = Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE;
2079 pattern = Pattern.compile("[sik\u00c5]+", flags);
2080 if (!pattern.matcher("\u017f\u0130\u0131\u212a\u212b").matches())
2081 failCount++;
2082
2083 report("CharClass");
2084 }
2085
2086 private static void caretTest() throws Exception {
2087 Pattern pattern = Pattern.compile("\\w*");
2088 Matcher matcher = pattern.matcher("a#bc#def##g");
2089 check(matcher, "a");
2090 check(matcher, "");
2091 check(matcher, "bc");
2092 check(matcher, "");
2093 check(matcher, "def");
2094 check(matcher, "");
2095 check(matcher, "");
2096 check(matcher, "g");
2097 check(matcher, "");
2098 if (matcher.find())
2099 failCount++;
2100
2101 pattern = Pattern.compile("^\\w*");
2102 matcher = pattern.matcher("a#bc#def##g");
2103 check(matcher, "a");
2104 if (matcher.find())
2105 failCount++;
2106
2107 pattern = Pattern.compile("\\w");
2108 matcher = pattern.matcher("abc##x");
2109 check(matcher, "a");
2110 check(matcher, "b");
2111 check(matcher, "c");
2112 check(matcher, "x");
2113 if (matcher.find())
2114 failCount++;
2115
2116 pattern = Pattern.compile("^\\w");
2117 matcher = pattern.matcher("abc##x");
2118 check(matcher, "a");
2119 if (matcher.find())
2120 failCount++;
2121
2122 pattern = Pattern.compile("\\A\\p{Alpha}{3}");
2123 matcher = pattern.matcher("abcdef-ghi\njklmno");
2124 check(matcher, "abc");
2125 if (matcher.find())
2126 failCount++;
2127
2128 pattern = Pattern.compile("^\\p{Alpha}{3}", Pattern.MULTILINE);
2129 matcher = pattern.matcher("abcdef-ghi\njklmno");
2130 check(matcher, "abc");
2131 check(matcher, "jkl");
2132 if (matcher.find())
2133 failCount++;
2134
2135 pattern = Pattern.compile("^", Pattern.MULTILINE);
2136 matcher = pattern.matcher("this is some text");
2137 String result = matcher.replaceAll("X");
2138 if (!result.equals("Xthis is some text"))
2139 failCount++;
2140
2141 pattern = Pattern.compile("^");
2142 matcher = pattern.matcher("this is some text");
2143 result = matcher.replaceAll("X");
2144 if (!result.equals("Xthis is some text"))
2145 failCount++;
2146
2147 pattern = Pattern.compile("^", Pattern.MULTILINE | Pattern.UNIX_LINES);
2148 matcher = pattern.matcher("this is some text\n");
2149 result = matcher.replaceAll("X");
2150 if (!result.equals("Xthis is some text\n"))
2151 failCount++;
2152
2153 report("Caret");
2154 }
2155
2156 private static void groupCaptureTest() throws Exception {
2157 // Independent group
2158 Pattern pattern = Pattern.compile("x+(?>y+)z+");
2159 Matcher matcher = pattern.matcher("xxxyyyzzz");
2160 matcher.find();
2161 try {
2162 String blah = matcher.group(1);
2163 failCount++;
2164 } catch (IndexOutOfBoundsException ioobe) {
2165 // Good result
2166 }
2167 // Pure group
2168 pattern = Pattern.compile("x+(?:y+)z+");
2169 matcher = pattern.matcher("xxxyyyzzz");
2170 matcher.find();
2171 try {
2172 String blah = matcher.group(1);
2173 failCount++;
2174 } catch (IndexOutOfBoundsException ioobe) {
2175 // Good result
2176 }
2177
2178 // Supplementary character tests
2179 // Independent group
2180 pattern = Pattern.compile(toSupplementaries("x+(?>y+)z+"));
2181 matcher = pattern.matcher(toSupplementaries("xxxyyyzzz"));
2182 matcher.find();
2183 try {
2184 String blah = matcher.group(1);
2185 failCount++;
2186 } catch (IndexOutOfBoundsException ioobe) {
2187 // Good result
2188 }
2189 // Pure group
2190 pattern = Pattern.compile(toSupplementaries("x+(?:y+)z+"));
2191 matcher = pattern.matcher(toSupplementaries("xxxyyyzzz"));
2192 matcher.find();
2193 try {
2194 String blah = matcher.group(1);
2195 failCount++;
2196 } catch (IndexOutOfBoundsException ioobe) {
2197 // Good result
2198 }
2199
2200 report("GroupCapture");
2201 }
2202
2203 private static void backRefTest() throws Exception {
2204 Pattern pattern = Pattern.compile("(a*)bc\\1");
2205 check(pattern, "zzzaabcazzz", true);
2206
2207 pattern = Pattern.compile("(a*)bc\\1");
2208 check(pattern, "zzzaabcaazzz", true);
2209
2210 pattern = Pattern.compile("(abc)(def)\\1");
2211 check(pattern, "abcdefabc", true);
2212
2213 pattern = Pattern.compile("(abc)(def)\\3");
2214 check(pattern, "abcdefabc", false);
2215
2216 try {
2217 for (int i = 1; i < 10; i++) {
2218 // Make sure backref 1-9 are always accepted
2219 pattern = Pattern.compile("abcdef\\" + i);
2220 // and fail to match if the target group does not exit
2221 check(pattern, "abcdef", false);
2222 }
2223 } catch(PatternSyntaxException e) {
2224 failCount++;
2225 }
2226
2227 pattern = Pattern.compile("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)\\11");
2228 check(pattern, "abcdefghija", false);
2229 check(pattern, "abcdefghija1", true);
2230
2231 pattern = Pattern.compile("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)(k)\\11");
2232 check(pattern, "abcdefghijkk", true);
2233
2234 pattern = Pattern.compile("(a)bcdefghij\\11");
2235 check(pattern, "abcdefghija1", true);
2236
2237 // Supplementary character tests
2238 pattern = Pattern.compile(toSupplementaries("(a*)bc\\1"));
2239 check(pattern, toSupplementaries("zzzaabcazzz"), true);
2240
2241 pattern = Pattern.compile(toSupplementaries("(a*)bc\\1"));
2242 check(pattern, toSupplementaries("zzzaabcaazzz"), true);
2243
2244 pattern = Pattern.compile(toSupplementaries("(abc)(def)\\1"));
2245 check(pattern, toSupplementaries("abcdefabc"), true);
2246
2247 pattern = Pattern.compile(toSupplementaries("(abc)(def)\\3"));
2248 check(pattern, toSupplementaries("abcdefabc"), false);
2249
2250 pattern = Pattern.compile(toSupplementaries("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)\\11"));
2251 check(pattern, toSupplementaries("abcdefghija"), false);
2252 check(pattern, toSupplementaries("abcdefghija1"), true);
2253
2254 pattern = Pattern.compile(toSupplementaries("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)(k)\\11"));
2255 check(pattern, toSupplementaries("abcdefghijkk"), true);
2256
2257 report("BackRef");
2258 }
2259
2260 /**
2261 * Unicode Technical Report #18, section 2.6 End of Line
2262 * There is no empty line to be matched in the sequence \u000D\u000A
2263 * but there is an empty line in the sequence \u000A\u000D.
2264 */
2265 private static void anchorTest() throws Exception {
2266 Pattern p = Pattern.compile("^.*$", Pattern.MULTILINE);
2267 Matcher m = p.matcher("blah1\r\nblah2");
2268 m.find();
2269 m.find();
2270 if (!m.group().equals("blah2"))
2271 failCount++;
2272
2273 m.reset("blah1\n\rblah2");
2274 m.find();
2275 m.find();
2276 m.find();
2277 if (!m.group().equals("blah2"))
2278 failCount++;
2279
2280 // Test behavior of $ with \r\n at end of input
2281 p = Pattern.compile(".+$");
2282 m = p.matcher("blah1\r\n");
2283 if (!m.find())
2284 failCount++;
2285 if (!m.group().equals("blah1"))
2286 failCount++;
2287 if (m.find())
2288 failCount++;
2289
2290 // Test behavior of $ with \r\n at end of input in multiline
2291 p = Pattern.compile(".+$", Pattern.MULTILINE);
2292 m = p.matcher("blah1\r\n");
2293 if (!m.find())
2294 failCount++;
2295 if (m.find())
2296 failCount++;
2297
2298 // Test for $ recognition of \u0085 for bug 4527731
2299 p = Pattern.compile(".+$", Pattern.MULTILINE);
2300 m = p.matcher("blah1\u0085");
2301 if (!m.find())
2302 failCount++;
2303
2304 // Supplementary character test
2305 p = Pattern.compile("^.*$", Pattern.MULTILINE);
2306 m = p.matcher(toSupplementaries("blah1\r\nblah2"));
2307 m.find();
2308 m.find();
2309 if (!m.group().equals(toSupplementaries("blah2")))
2310 failCount++;
2311
2312 m.reset(toSupplementaries("blah1\n\rblah2"));
2313 m.find();
2314 m.find();
2315 m.find();
2316 if (!m.group().equals(toSupplementaries("blah2")))
2317 failCount++;
2318
2319 // Test behavior of $ with \r\n at end of input
2320 p = Pattern.compile(".+$");
2321 m = p.matcher(toSupplementaries("blah1\r\n"));
2322 if (!m.find())
2323 failCount++;
2324 if (!m.group().equals(toSupplementaries("blah1")))
2325 failCount++;
2326 if (m.find())
2327 failCount++;
2328
2329 // Test behavior of $ with \r\n at end of input in multiline
2330 p = Pattern.compile(".+$", Pattern.MULTILINE);
2331 m = p.matcher(toSupplementaries("blah1\r\n"));
2332 if (!m.find())
2333 failCount++;
2334 if (m.find())
2335 failCount++;
2336
2337 // Test for $ recognition of \u0085 for bug 4527731
2338 p = Pattern.compile(".+$", Pattern.MULTILINE);
2339 m = p.matcher(toSupplementaries("blah1\u0085"));
2340 if (!m.find())
2341 failCount++;
2342
2343 report("Anchors");
2344 }
2345
2346 /**
2347 * A basic sanity test of Matcher.lookingAt().
2348 */
2349 private static void lookingAtTest() throws Exception {
2350 Pattern p = Pattern.compile("(ab)(c*)");
2351 Matcher m = p.matcher("abccczzzabcczzzabccc");
2352
2353 if (!m.lookingAt())
2354 failCount++;
2355
2356 if (!m.group().equals(m.group(0)))
2357 failCount++;
2358
2359 m = p.matcher("zzzabccczzzabcczzzabccczzz");
2360 if (m.lookingAt())
2361 failCount++;
2362
2363 // Supplementary character test
2364 p = Pattern.compile(toSupplementaries("(ab)(c*)"));
2365 m = p.matcher(toSupplementaries("abccczzzabcczzzabccc"));
2366
2367 if (!m.lookingAt())
2368 failCount++;
2369
2370 if (!m.group().equals(m.group(0)))
2371 failCount++;
2372
2373 m = p.matcher(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
2374 if (m.lookingAt())
2375 failCount++;
2376
2377 report("Looking At");
2378 }
2379
2380 /**
2381 * A basic sanity test of Matcher.matches().
2382 */
2383 private static void matchesTest() throws Exception {
2384 // matches()
2385 Pattern p = Pattern.compile("ulb(c*)");
2386 Matcher m = p.matcher("ulbcccccc");
2387 if (!m.matches())
2388 failCount++;
2389
2390 // find() but not matches()
2391 m.reset("zzzulbcccccc");
2392 if (m.matches())
2393 failCount++;
2394
2395 // lookingAt() but not matches()
2396 m.reset("ulbccccccdef");
2397 if (m.matches())
2398 failCount++;
2399
2400 // matches()
2401 p = Pattern.compile("a|ad");
2402 m = p.matcher("ad");
2403 if (!m.matches())
2404 failCount++;
2405
2406 // Supplementary character test
2407 // matches()
2408 p = Pattern.compile(toSupplementaries("ulb(c*)"));
2409 m = p.matcher(toSupplementaries("ulbcccccc"));
2410 if (!m.matches())
2411 failCount++;
2412
2413 // find() but not matches()
2414 m.reset(toSupplementaries("zzzulbcccccc"));
2415 if (m.matches())
2416 failCount++;
2417
2418 // lookingAt() but not matches()
2419 m.reset(toSupplementaries("ulbccccccdef"));
2420 if (m.matches())
2421 failCount++;
2422
2423 // matches()
2424 p = Pattern.compile(toSupplementaries("a|ad"));
2425 m = p.matcher(toSupplementaries("ad"));
2426 if (!m.matches())
2427 failCount++;
2428
2429 report("Matches");
2430 }
2431
2432 /**
2433 * A basic sanity test of Pattern.matches().
2434 */
2435 private static void patternMatchesTest() throws Exception {
2436 // matches()
2437 if (!Pattern.matches(toSupplementaries("ulb(c*)"),
2438 toSupplementaries("ulbcccccc")))
2439 failCount++;
2440
2441 // find() but not matches()
2442 if (Pattern.matches(toSupplementaries("ulb(c*)"),
2443 toSupplementaries("zzzulbcccccc")))
2444 failCount++;
2445
2446 // lookingAt() but not matches()
2447 if (Pattern.matches(toSupplementaries("ulb(c*)"),
2448 toSupplementaries("ulbccccccdef")))
2449 failCount++;
2450
2451 // Supplementary character test
2452 // matches()
2453 if (!Pattern.matches(toSupplementaries("ulb(c*)"),
2454 toSupplementaries("ulbcccccc")))
2455 failCount++;
2456
2457 // find() but not matches()
2458 if (Pattern.matches(toSupplementaries("ulb(c*)"),
2459 toSupplementaries("zzzulbcccccc")))
2460 failCount++;
2461
2462 // lookingAt() but not matches()
2463 if (Pattern.matches(toSupplementaries("ulb(c*)"),
2464 toSupplementaries("ulbccccccdef")))
2465 failCount++;
2466
2467 report("Pattern Matches");
2468 }
2469
2470 /**
2471 * Canonical equivalence testing. Tests the ability of the engine
2472 * to match sequences that are not explicitly specified in the
2473 * pattern when they are considered equivalent by the Unicode Standard.
2474 */
2475 private static void ceTest() throws Exception {
2476 // Decomposed char outside char classes
2477 Pattern p = Pattern.compile("testa\u030a", Pattern.CANON_EQ);
2478 Matcher m = p.matcher("test\u00e5");
2479 if (!m.matches())
2480 failCount++;
2481
2482 m.reset("testa\u030a");
2483 if (!m.matches())
2484 failCount++;
2485
2486 // Composed char outside char classes
2487 p = Pattern.compile("test\u00e5", Pattern.CANON_EQ);
2488 m = p.matcher("test\u00e5");
2489 if (!m.matches())
2490 failCount++;
2491
2492 m.reset("testa\u030a");
2493 if (!m.find())
2494 failCount++;
2495
2496 // Decomposed char inside a char class
2497 p = Pattern.compile("test[abca\u030a]", Pattern.CANON_EQ);
2498 m = p.matcher("test\u00e5");
2499 if (!m.find())
2500 failCount++;
2501
2502 m.reset("testa\u030a");
2503 if (!m.find())
2504 failCount++;
2505
2506 // Composed char inside a char class
2507 p = Pattern.compile("test[abc\u00e5def\u00e0]", Pattern.CANON_EQ);
2508 m = p.matcher("test\u00e5");
2509 if (!m.find())
2510 failCount++;
2511
2512 m.reset("testa\u0300");
2513 if (!m.find())
2514 failCount++;
2515
2516 m.reset("testa\u030a");
2517 if (!m.find())
2518 failCount++;
2519
2520 // Marks that cannot legally change order and be equivalent
2521 p = Pattern.compile("testa\u0308\u0300", Pattern.CANON_EQ);
2522 check(p, "testa\u0308\u0300", true);
2523 check(p, "testa\u0300\u0308", false);
2524
2525 // Marks that can legally change order and be equivalent
2526 p = Pattern.compile("testa\u0308\u0323", Pattern.CANON_EQ);
2527 check(p, "testa\u0308\u0323", true);
2528 check(p, "testa\u0323\u0308", true);
2529
2530 // Test all equivalences of the sequence a\u0308\u0323\u0300
2531 p = Pattern.compile("testa\u0308\u0323\u0300", Pattern.CANON_EQ);
2532 check(p, "testa\u0308\u0323\u0300", true);
2533 check(p, "testa\u0323\u0308\u0300", true);
2534 check(p, "testa\u0308\u0300\u0323", true);
2535 check(p, "test\u00e4\u0323\u0300", true);
2536 check(p, "test\u00e4\u0300\u0323", true);
2537
2538 /*
2539 * The following canonical equivalence tests don't work. Bug id: 4916384.
2540 *
2541 // Decomposed hangul (jamos)
2542 p = Pattern.compile("\u1100\u1161", Pattern.CANON_EQ);
2543 m = p.matcher("\u1100\u1161");
2544 if (!m.matches())
2545 failCount++;
2546
2547 m.reset("\uac00");
2548 if (!m.matches())
2549 failCount++;
2550
2551 // Composed hangul
2552 p = Pattern.compile("\uac00", Pattern.CANON_EQ);
2553 m = p.matcher("\u1100\u1161");
2554 if (!m.matches())
2555 failCount++;
2556
2557 m.reset("\uac00");
2558 if (!m.matches())
2559 failCount++;
2560
2561 // Decomposed supplementary outside char classes
2562 p = Pattern.compile("test\ud834\uddbc\ud834\udd6f", Pattern.CANON_EQ);
2563 m = p.matcher("test\ud834\uddc0");
2564 if (!m.matches())
2565 failCount++;
2566
2567 m.reset("test\ud834\uddbc\ud834\udd6f");
2568 if (!m.matches())
2569 failCount++;
2570
2571 // Composed supplementary outside char classes
2572 p = Pattern.compile("test\ud834\uddc0", Pattern.CANON_EQ);
2573 m.reset("test\ud834\uddbc\ud834\udd6f");
2574 if (!m.matches())
2575 failCount++;
2576
2577 m = p.matcher("test\ud834\uddc0");
2578 if (!m.matches())
2579 failCount++;
2580
2581 */
2582
2583 report("Canonical Equivalence");
2584 }
2585
2586 /**
2587 * A basic sanity test of Matcher.replaceAll().
2588 */
2589 private static void globalSubstitute() throws Exception {
2590 // Global substitution with a literal
2591 Pattern p = Pattern.compile("(ab)(c*)");
2592 Matcher m = p.matcher("abccczzzabcczzzabccc");
2593 if (!m.replaceAll("test").equals("testzzztestzzztest"))
2594 failCount++;
2595
2596 m.reset("zzzabccczzzabcczzzabccczzz");
2597 if (!m.replaceAll("test").equals("zzztestzzztestzzztestzzz"))
2598 failCount++;
2599
2600 // Global substitution with groups
2601 m.reset("zzzabccczzzabcczzzabccczzz");
2602 String result = m.replaceAll("$1");
2603 if (!result.equals("zzzabzzzabzzzabzzz"))
2604 failCount++;
2605
2606 // Supplementary character test
2607 // Global substitution with a literal
2608 p = Pattern.compile(toSupplementaries("(ab)(c*)"));
2609 m = p.matcher(toSupplementaries("abccczzzabcczzzabccc"));
2610 if (!m.replaceAll(toSupplementaries("test")).
2611 equals(toSupplementaries("testzzztestzzztest")))
2612 failCount++;
2613
2614 m.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
2615 if (!m.replaceAll(toSupplementaries("test")).
2616 equals(toSupplementaries("zzztestzzztestzzztestzzz")))
2617 failCount++;
2618
2619 // Global substitution with groups
2620 m.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
2621 result = m.replaceAll("$1");
2622 if (!result.equals(toSupplementaries("zzzabzzzabzzzabzzz")))
2623 failCount++;
2624
2625 report("Global Substitution");
2626 }
2627
2628 /**
2629 * Tests the usage of Matcher.appendReplacement() with literal
2630 * and group substitutions.
2631 */
2632 private static void stringbufferSubstitute() throws Exception {
2633 // SB substitution with literal
2634 String blah = "zzzblahzzz";
2635 Pattern p = Pattern.compile("blah");
2636 Matcher m = p.matcher(blah);
2637 StringBuffer result = new StringBuffer();
2638 try {
2639 m.appendReplacement(result, "blech");
2640 failCount++;
2641 } catch (IllegalStateException e) {
2642 }
2643 m.find();
2644 m.appendReplacement(result, "blech");
2645 if (!result.toString().equals("zzzblech"))
2646 failCount++;
2647
2648 m.appendTail(result);
2649 if (!result.toString().equals("zzzblechzzz"))
2650 failCount++;
2651
2652 // SB substitution with groups
2653 blah = "zzzabcdzzz";
2654 p = Pattern.compile("(ab)(cd)*");
2655 m = p.matcher(blah);
2656 result = new StringBuffer();
2657 try {
2658 m.appendReplacement(result, "$1");
2659 failCount++;
2660 } catch (IllegalStateException e) {
2661 }
2662 m.find();
2663 m.appendReplacement(result, "$1");
2664 if (!result.toString().equals("zzzab"))
2665 failCount++;
2666
2667 m.appendTail(result);
2668 if (!result.toString().equals("zzzabzzz"))
2669 failCount++;
2670
2671 // SB substitution with 3 groups
2672 blah = "zzzabcdcdefzzz";
2673 p = Pattern.compile("(ab)(cd)*(ef)");
2674 m = p.matcher(blah);
2675 result = new StringBuffer();
2676 try {
2677 m.appendReplacement(result, "$1w$2w$3");
2678 failCount++;
2679 } catch (IllegalStateException e) {
2680 }
2681 m.find();
2682 m.appendReplacement(result, "$1w$2w$3");
2683 if (!result.toString().equals("zzzabwcdwef"))
2684 failCount++;
2685
2686 m.appendTail(result);
2687 if (!result.toString().equals("zzzabwcdwefzzz"))
2688 failCount++;
2689
2690 // SB substitution with groups and three matches
2691 // skipping middle match
2692 blah = "zzzabcdzzzabcddzzzabcdzzz";
2693 p = Pattern.compile("(ab)(cd*)");
2694 m = p.matcher(blah);
2695 result = new StringBuffer();
2696 try {
2697 m.appendReplacement(result, "$1");
2698 failCount++;
2699 } catch (IllegalStateException e) {
2700 }
2701 m.find();
2702 m.appendReplacement(result, "$1");
2703 if (!result.toString().equals("zzzab"))
2704 failCount++;
2705
2706 m.find();
2707 m.find();
2708 m.appendReplacement(result, "$2");
2709 if (!result.toString().equals("zzzabzzzabcddzzzcd"))
2710 failCount++;
2711
2712 m.appendTail(result);
2713 if (!result.toString().equals("zzzabzzzabcddzzzcdzzz"))
2714 failCount++;
2715
2716 // Check to make sure escaped $ is ignored
2717 blah = "zzzabcdcdefzzz";
2718 p = Pattern.compile("(ab)(cd)*(ef)");
2719 m = p.matcher(blah);
2720 result = new StringBuffer();
2721 m.find();
2722 m.appendReplacement(result, "$1w\\$2w$3");
2723 if (!result.toString().equals("zzzabw$2wef"))
2724 failCount++;
2725
2726 m.appendTail(result);
2727 if (!result.toString().equals("zzzabw$2wefzzz"))
2728 failCount++;
2729
2730 // Check to make sure a reference to nonexistent group causes error
2731 blah = "zzzabcdcdefzzz";
2732 p = Pattern.compile("(ab)(cd)*(ef)");
2733 m = p.matcher(blah);
2734 result = new StringBuffer();
2735 m.find();
2736 try {
2737 m.appendReplacement(result, "$1w$5w$3");
2738 failCount++;
2739 } catch (IndexOutOfBoundsException ioobe) {
2740 // Correct result
2741 }
2742
2743 // Check double digit group references
2744 blah = "zzz123456789101112zzz";
2745 p = Pattern.compile("(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)");
2746 m = p.matcher(blah);
2747 result = new StringBuffer();
2748 m.find();
2749 m.appendReplacement(result, "$1w$11w$3");
2750 if (!result.toString().equals("zzz1w11w3"))
2751 failCount++;
2752
2753 // Check to make sure it backs off $15 to $1 if only three groups
2754 blah = "zzzabcdcdefzzz";
2755 p = Pattern.compile("(ab)(cd)*(ef)");
2756 m = p.matcher(blah);
2757 result = new StringBuffer();
2758 m.find();
2759 m.appendReplacement(result, "$1w$15w$3");
2760 if (!result.toString().equals("zzzabwab5wef"))
2761 failCount++;
2762
2763
2764 // Supplementary character test
2765 // SB substitution with literal
2766 blah = toSupplementaries("zzzblahzzz");
2767 p = Pattern.compile(toSupplementaries("blah"));
2768 m = p.matcher(blah);
2769 result = new StringBuffer();
2770 try {
2771 m.appendReplacement(result, toSupplementaries("blech"));
2772 failCount++;
2773 } catch (IllegalStateException e) {
2774 }
2775 m.find();
2776 m.appendReplacement(result, toSupplementaries("blech"));
2777 if (!result.toString().equals(toSupplementaries("zzzblech")))
2778 failCount++;
2779
2780 m.appendTail(result);
2781 if (!result.toString().equals(toSupplementaries("zzzblechzzz")))
2782 failCount++;
2783
2784 // SB substitution with groups
2785 blah = toSupplementaries("zzzabcdzzz");
2786 p = Pattern.compile(toSupplementaries("(ab)(cd)*"));
2787 m = p.matcher(blah);
2788 result = new StringBuffer();
2789 try {
2790 m.appendReplacement(result, "$1");
2791 failCount++;
2792 } catch (IllegalStateException e) {
2793 }
2794 m.find();
2795 m.appendReplacement(result, "$1");
2796 if (!result.toString().equals(toSupplementaries("zzzab")))
2797 failCount++;
2798
2799 m.appendTail(result);
2800 if (!result.toString().equals(toSupplementaries("zzzabzzz")))
2801 failCount++;
2802
2803 // SB substitution with 3 groups
2804 blah = toSupplementaries("zzzabcdcdefzzz");
2805 p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
2806 m = p.matcher(blah);
2807 result = new StringBuffer();
2808 try {
2809 m.appendReplacement(result, toSupplementaries("$1w$2w$3"));
2810 failCount++;
2811 } catch (IllegalStateException e) {
2812 }
2813 m.find();
2814 m.appendReplacement(result, toSupplementaries("$1w$2w$3"));
2815 if (!result.toString().equals(toSupplementaries("zzzabwcdwef")))
2816 failCount++;
2817
2818 m.appendTail(result);
2819 if (!result.toString().equals(toSupplementaries("zzzabwcdwefzzz")))
2820 failCount++;
2821
2822 // SB substitution with groups and three matches
2823 // skipping middle match
2824 blah = toSupplementaries("zzzabcdzzzabcddzzzabcdzzz");
2825 p = Pattern.compile(toSupplementaries("(ab)(cd*)"));
2826 m = p.matcher(blah);
2827 result = new StringBuffer();
2828 try {
2829 m.appendReplacement(result, "$1");
2830 failCount++;
2831 } catch (IllegalStateException e) {
2832 }
2833 m.find();
2834 m.appendReplacement(result, "$1");
2835 if (!result.toString().equals(toSupplementaries("zzzab")))
2836 failCount++;
2837
2838 m.find();
2839 m.find();
2840 m.appendReplacement(result, "$2");
2841 if (!result.toString().equals(toSupplementaries("zzzabzzzabcddzzzcd")))
2842 failCount++;
2843
2844 m.appendTail(result);
2845 if (!result.toString().equals(toSupplementaries("zzzabzzzabcddzzzcdzzz")))
2846 failCount++;
2847
2848 // Check to make sure escaped $ is ignored
2849 blah = toSupplementaries("zzzabcdcdefzzz");
2850 p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
2851 m = p.matcher(blah);
2852 result = new StringBuffer();
2853 m.find();
2854 m.appendReplacement(result, toSupplementaries("$1w\\$2w$3"));
2855 if (!result.toString().equals(toSupplementaries("zzzabw$2wef")))
2856 failCount++;
2857
2858 m.appendTail(result);
2859 if (!result.toString().equals(toSupplementaries("zzzabw$2wefzzz")))
2860 failCount++;
2861
2862 // Check to make sure a reference to nonexistent group causes error
2863 blah = toSupplementaries("zzzabcdcdefzzz");
2864 p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
2865 m = p.matcher(blah);
2866 result = new StringBuffer();
2867 m.find();
2868 try {
2869 m.appendReplacement(result, toSupplementaries("$1w$5w$3"));
2870 failCount++;
2871 } catch (IndexOutOfBoundsException ioobe) {
2872 // Correct result
2873 }
2874
2875 // Check double digit group references
2876 blah = toSupplementaries("zzz123456789101112zzz");
2877 p = Pattern.compile("(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)");
2878 m = p.matcher(blah);
2879 result = new StringBuffer();
2880 m.find();
2881 m.appendReplacement(result, toSupplementaries("$1w$11w$3"));
2882 if (!result.toString().equals(toSupplementaries("zzz1w11w3")))
2883 failCount++;
2884
2885 // Check to make sure it backs off $15 to $1 if only three groups
2886 blah = toSupplementaries("zzzabcdcdefzzz");
2887 p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
2888 m = p.matcher(blah);
2889 result = new StringBuffer();
2890 m.find();
2891 m.appendReplacement(result, toSupplementaries("$1w$15w$3"));
2892 if (!result.toString().equals(toSupplementaries("zzzabwab5wef")))
2893 failCount++;
2894
2895 // Check nothing has been appended into the output buffer if
2896 // the replacement string triggers IllegalArgumentException.
2897 p = Pattern.compile("(abc)");
2898 m = p.matcher("abcd");
2899 result = new StringBuffer();
2900 m.find();
2901 try {
2902 m.appendReplacement(result, ("xyz$g"));
2903 failCount++;
2904 } catch (IllegalArgumentException iae) {
2905 if (result.length() != 0)
2906 failCount++;
2907 }
2908
2909 report("SB Substitution");
2910 }
2911
2912 /*
2913 * 5 groups of characters are created to make a substitution string.
2914 * A base string will be created including random lead chars, the
2915 * substitution string, and random trailing chars.
2916 * A pattern containing the 5 groups is searched for and replaced with:
2917 * random group + random string + random group.
2918 * The results are checked for correctness.
2919 */
2920 private static void substitutionBasher() {
2921 for (int runs = 0; runs<1000; runs++) {
2922 // Create a base string to work in
2923 int leadingChars = generator.nextInt(10);
2924 StringBuffer baseBuffer = new StringBuffer(100);
2925 String leadingString = getRandomAlphaString(leadingChars);
2926 baseBuffer.append(leadingString);
2927
2928 // Create 5 groups of random number of random chars
2929 // Create the string to substitute
2930 // Create the pattern string to search for
2931 StringBuffer bufferToSub = new StringBuffer(25);
2932 StringBuffer bufferToPat = new StringBuffer(50);
2933 String[] groups = new String[5];
2934 for(int i=0; i<5; i++) {
2935 int aGroupSize = generator.nextInt(5)+1;
2936 groups[i] = getRandomAlphaString(aGroupSize);
2937 bufferToSub.append(groups[i]);
2938 bufferToPat.append('(');
2939 bufferToPat.append(groups[i]);
2940 bufferToPat.append(')');
2941 }
2942 String stringToSub = bufferToSub.toString();
2943 String pattern = bufferToPat.toString();
2944
2945 // Place sub string into working string at random index
2946 baseBuffer.append(stringToSub);
2947
2948 // Append random chars to end
2949 int trailingChars = generator.nextInt(10);
2950 String trailingString = getRandomAlphaString(trailingChars);
2951 baseBuffer.append(trailingString);
2952 String baseString = baseBuffer.toString();
2953
2954 // Create test pattern and matcher
2955 Pattern p = Pattern.compile(pattern);
2956 Matcher m = p.matcher(baseString);
2957
2958 // Reject candidate if pattern happens to start early
2959 m.find();
2960 if (m.start() < leadingChars)
2961 continue;
2962
2963 // Reject candidate if more than one match
2964 if (m.find())
2965 continue;
2966
2967 // Construct a replacement string with :
2968 // random group + random string + random group
2969 StringBuffer bufferToRep = new StringBuffer();
2970 int groupIndex1 = generator.nextInt(5);
2971 bufferToRep.append("$" + (groupIndex1 + 1));
2972 String randomMidString = getRandomAlphaString(5);
2973 bufferToRep.append(randomMidString);
2974 int groupIndex2 = generator.nextInt(5);
2975 bufferToRep.append("$" + (groupIndex2 + 1));
2976 String replacement = bufferToRep.toString();
2977
2978 // Do the replacement
2979 String result = m.replaceAll(replacement);
2980
2981 // Construct expected result
2982 StringBuffer bufferToRes = new StringBuffer();
2983 bufferToRes.append(leadingString);
2984 bufferToRes.append(groups[groupIndex1]);
2985 bufferToRes.append(randomMidString);
2986 bufferToRes.append(groups[groupIndex2]);
2987 bufferToRes.append(trailingString);
2988 String expectedResult = bufferToRes.toString();
2989
2990 // Check results
2991 if (!result.equals(expectedResult))
2992 failCount++;
2993 }
2994
2995 report("Substitution Basher");
2996 }
2997
2998 /**
2999 * Checks the handling of some escape sequences that the Pattern
3000 * class should process instead of the java compiler. These are
3001 * not in the file because the escapes should be be processed
3002 * by the Pattern class when the regex is compiled.
3003 */
3004 private static void escapes() throws Exception {
3005 Pattern p = Pattern.compile("\\043");
3006 Matcher m = p.matcher("#");
3007 if (!m.find())
3008 failCount++;
3009
3010 p = Pattern.compile("\\x23");
3011 m = p.matcher("#");
3012 if (!m.find())
3013 failCount++;
3014
3015 p = Pattern.compile("\\u0023");
3016 m = p.matcher("#");
3017 if (!m.find())
3018 failCount++;
3019
3020 report("Escape sequences");
3021 }
3022
3023 /**
3024 * Checks the handling of blank input situations. These
3025 * tests are incompatible with my test file format.
3026 */
3027 private static void blankInput() throws Exception {
3028 Pattern p = Pattern.compile("abc", Pattern.CASE_INSENSITIVE);
3029 Matcher m = p.matcher("");
3030 if (m.find())
3031 failCount++;
3032
3033 p = Pattern.compile("a*", Pattern.CASE_INSENSITIVE);
3034 m = p.matcher("");
3035 if (!m.find())
3036 failCount++;
3037
3038 p = Pattern.compile("abc");
3039 m = p.matcher("");
3040 if (m.find())
3041 failCount++;
3042
3043 p = Pattern.compile("a*");
3044 m = p.matcher("");
3045 if (!m.find())
3046 failCount++;
3047
3048 report("Blank input");
3049 }
3050
3051 /**
3052 * Tests the Boyer-Moore pattern matching of a character sequence
3053 * on randomly generated patterns.
3054 */
3055 private static void bm() throws Exception {
3056 doBnM('a');
3057 report("Boyer Moore (ASCII)");
3058
3059 doBnM(Character.MIN_SUPPLEMENTARY_CODE_POINT - 10);
3060 report("Boyer Moore (Supplementary)");
3061 }
3062
3063 private static void doBnM(int baseCharacter) throws Exception {
3064 int achar=0;
3065
3066 for (int i=0; i<100; i++) {
3067 // Create a short pattern to search for
3068 int patternLength = generator.nextInt(7) + 4;
3069 StringBuffer patternBuffer = new StringBuffer(patternLength);
3070 for (int x=0; x<patternLength; x++) {
3071 int ch = baseCharacter + generator.nextInt(26);
3072 if (Character.isSupplementaryCodePoint(ch)) {
3073 patternBuffer.append(Character.toChars(ch));
3074 } else {
3075 patternBuffer.append((char)ch);
3076 }
3077 }
3078 String pattern = patternBuffer.toString();
3079 Pattern p = Pattern.compile(pattern);
3080
3081 // Create a buffer with random ASCII chars that does
3082 // not match the sample
3083 String toSearch = null;
3084 StringBuffer s = null;
3085 Matcher m = p.matcher("");
3086 do {
3087 s = new StringBuffer(100);
3088 for (int x=0; x<100; x++) {
3089 int ch = baseCharacter + generator.nextInt(26);
3090 if (Character.isSupplementaryCodePoint(ch)) {
3091 s.append(Character.toChars(ch));
3092 } else {
3093 s.append((char)ch);
3094 }
3095 }
3096 toSearch = s.toString();
3097 m.reset(toSearch);
3098 } while (m.find());
3099
3100 // Insert the pattern at a random spot
3101 int insertIndex = generator.nextInt(99);
3102 if (Character.isLowSurrogate(s.charAt(insertIndex)))
3103 insertIndex++;
3104 s = s.insert(insertIndex, pattern);
3105 toSearch = s.toString();
3106
3107 // Make sure that the pattern is found
3108 m.reset(toSearch);
3109 if (!m.find())
3110 failCount++;
3111
3112 // Make sure that the match text is the pattern
3113 if (!m.group().equals(pattern))
3114 failCount++;
3115
3116 // Make sure match occured at insertion point
3117 if (m.start() != insertIndex)
3118 failCount++;
3119 }
3120 }
3121
3122 /**
3123 * Tests the matching of slices on randomly generated patterns.
3124 * The Boyer-Moore optimization is not done on these patterns
3125 * because it uses unicode case folding.
3126 */
3127 private static void slice() throws Exception {
3128 doSlice(Character.MAX_VALUE);
3129 report("Slice");
3130
3131 doSlice(Character.MAX_CODE_POINT);
3132 report("Slice (Supplementary)");
3133 }
3134
3135 private static void doSlice(int maxCharacter) throws Exception {
3136 Random generator = new Random();
3137 int achar=0;
3138
3139 for (int i=0; i<100; i++) {
3140 // Create a short pattern to search for
3141 int patternLength = generator.nextInt(7) + 4;
3142 StringBuffer patternBuffer = new StringBuffer(patternLength);
3143 for (int x=0; x<patternLength; x++) {
3144 int randomChar = 0;
3145 while (!Character.isLetterOrDigit(randomChar))
3146 randomChar = generator.nextInt(maxCharacter);
3147 if (Character.isSupplementaryCodePoint(randomChar)) {
3148 patternBuffer.append(Character.toChars(randomChar));
3149 } else {
3150 patternBuffer.append((char) randomChar);
3151 }
3152 }
3153 String pattern = patternBuffer.toString();
3154 Pattern p = Pattern.compile(pattern, Pattern.UNICODE_CASE);
3155
3156 // Create a buffer with random chars that does not match the sample
3157 String toSearch = null;
3158 StringBuffer s = null;
3159 Matcher m = p.matcher("");
3160 do {
3161 s = new StringBuffer(100);
3162 for (int x=0; x<100; x++) {
3163 int randomChar = 0;
3164 while (!Character.isLetterOrDigit(randomChar))
3165 randomChar = generator.nextInt(maxCharacter);
3166 if (Character.isSupplementaryCodePoint(randomChar)) {
3167 s.append(Character.toChars(randomChar));
3168 } else {
3169 s.append((char) randomChar);
3170 }
3171 }
3172 toSearch = s.toString();
3173 m.reset(toSearch);
3174 } while (m.find());
3175
3176 // Insert the pattern at a random spot
3177 int insertIndex = generator.nextInt(99);
3178 if (Character.isLowSurrogate(s.charAt(insertIndex)))
3179 insertIndex++;
3180 s = s.insert(insertIndex, pattern);
3181 toSearch = s.toString();
3182
3183 // Make sure that the pattern is found
3184 m.reset(toSearch);
3185 if (!m.find())
3186 failCount++;
3187
3188 // Make sure that the match text is the pattern
3189 if (!m.group().equals(pattern))
3190 failCount++;
3191
3192 // Make sure match occured at insertion point
3193 if (m.start() != insertIndex)
3194 failCount++;
3195 }
3196 }
3197
3198 private static void explainFailure(String pattern, String data,
3199 String expected, String actual) {
3200 System.err.println("----------------------------------------");
3201 System.err.println("Pattern = "+pattern);
3202 System.err.println("Data = "+data);
3203 System.err.println("Expected = " + expected);
3204 System.err.println("Actual = " + actual);
3205 }
3206
3207 private static void explainFailure(String pattern, String data,
3208 Throwable t) {
3209 System.err.println("----------------------------------------");
3210 System.err.println("Pattern = "+pattern);
3211 System.err.println("Data = "+data);
3212 t.printStackTrace(System.err);
3213 }
3214
3215 // Testing examples from a file
3216
3217 /**
3218 * Goes through the file "TestCases.txt" and creates many patterns
3219 * described in the file, matching the patterns against input lines in
3220 * the file, and comparing the results against the correct results
3221 * also found in the file. The file format is described in comments
3222 * at the head of the file.
3223 */
3224 private static void processFile(String fileName) throws Exception {
3225 File testCases = new File(System.getProperty("test.src", "."),
3226 fileName);
3227 FileInputStream in = new FileInputStream(testCases);
3228 BufferedReader r = new BufferedReader(new InputStreamReader(in));
3229
3230 // Process next test case.
3231 String aLine;
3232 while((aLine = r.readLine()) != null) {
3233 // Read a line for pattern
3234 String patternString = grabLine(r);
3235 Pattern p = null;
3236 try {
3237 p = compileTestPattern(patternString);
3238 } catch (PatternSyntaxException e) {
3239 String dataString = grabLine(r);
3240 String expectedResult = grabLine(r);
3241 if (expectedResult.startsWith("error"))
3242 continue;
3243 explainFailure(patternString, dataString, e);
3244 failCount++;
3245 continue;
3246 }
3247
3248 // Read a line for input string
3249 String dataString = grabLine(r);
3250 Matcher m = p.matcher(dataString);
3251 StringBuffer result = new StringBuffer();
3252
3253 // Check for IllegalStateExceptions before a match
3254 failCount += preMatchInvariants(m);
3255
3256 boolean found = m.find();
3257
3258 if (found)
3259 failCount += postTrueMatchInvariants(m);
3260 else
3261 failCount += postFalseMatchInvariants(m);
3262
3263 if (found) {
3264 result.append("true ");
3265 result.append(m.group(0) + " ");
3266 } else {
3267 result.append("false ");
3268 }
3269
3270 result.append(m.groupCount());
3271
3272 if (found) {
3273 for (int i=1; i<m.groupCount()+1; i++)
3274 if (m.group(i) != null)
3275 result.append(" " +m.group(i));
3276 }
3277
3278 // Read a line for the expected result
3279 String expectedResult = grabLine(r);
3280
3281 if (!result.toString().equals(expectedResult)) {
3282 explainFailure(patternString, dataString, expectedResult, result.toString());
3283 failCount++;
3284 }
3285 }
3286
3287 report(fileName);
3288 }
3289
3290 private static int preMatchInvariants(Matcher m) {
3291 int failCount = 0;
3292 try {
3293 m.start();
3294 failCount++;
3295 } catch (IllegalStateException ise) {}
3296 try {
3297 m.end();
3298 failCount++;
3299 } catch (IllegalStateException ise) {}
3300 try {
3301 m.group();
3302 failCount++;
3303 } catch (IllegalStateException ise) {}
3304 return failCount;
3305 }
3306
3307 private static int postFalseMatchInvariants(Matcher m) {
3308 int failCount = 0;
3309 try {
3310 m.group();
3311 failCount++;
3312 } catch (IllegalStateException ise) {}
3313 try {
3314 m.start();
3315 failCount++;
3316 } catch (IllegalStateException ise) {}
3317 try {
3318 m.end();
3319 failCount++;
3320 } catch (IllegalStateException ise) {}
3321 return failCount;
3322 }
3323
3324 private static int postTrueMatchInvariants(Matcher m) {
3325 int failCount = 0;
3326 //assert(m.start() = m.start(0);
3327 if (m.start() != m.start(0))
3328 failCount++;
3329 //assert(m.end() = m.end(0);
3330 if (m.start() != m.start(0))
3331 failCount++;
3332 //assert(m.group() = m.group(0);
3333 if (!m.group().equals(m.group(0)))
3334 failCount++;
3335 try {
3336 m.group(50);
3337 failCount++;
3338 } catch (IndexOutOfBoundsException ise) {}
3339
3340 return failCount;
3341 }
3342
3343 private static Pattern compileTestPattern(String patternString) {
3344 if (!patternString.startsWith("'")) {
3345 return Pattern.compile(patternString);
3346 }
3347
3348 int break1 = patternString.lastIndexOf("'");
3349 String flagString = patternString.substring(
3350 break1+1, patternString.length());
3351 patternString = patternString.substring(1, break1);
3352
3353 if (flagString.equals("i"))
3354 return Pattern.compile(patternString, Pattern.CASE_INSENSITIVE);
3355
3356 if (flagString.equals("m"))
3357 return Pattern.compile(patternString, Pattern.MULTILINE);
3358
3359 return Pattern.compile(patternString);
3360 }
3361
3362 /**
3363 * Reads a line from the input file. Keeps reading lines until a non
3364 * empty non comment line is read. If the line contains a \n then
3365 * these two characters are replaced by a newline char. If a \\uxxxx
3366 * sequence is read then the sequence is replaced by the unicode char.
3367 */
3368 private static String grabLine(BufferedReader r) throws Exception {
3369 int index = 0;
3370 String line = r.readLine();
3371 while (line.startsWith("//") || line.length() < 1)
3372 line = r.readLine();
3373 while ((index = line.indexOf("\\n")) != -1) {
3374 StringBuffer temp = new StringBuffer(line);
3375 temp.replace(index, index+2, "\n");
3376 line = temp.toString();
3377 }
3378 while ((index = line.indexOf("\\u")) != -1) {
3379 StringBuffer temp = new StringBuffer(line);
3380 String value = temp.substring(index+2, index+6);
3381 char aChar = (char)Integer.parseInt(value, 16);
3382 String unicodeChar = "" + aChar;
3383 temp.replace(index, index+6, unicodeChar);
3384 line = temp.toString();
3385 }
3386
3387 return line;
3388 }
3389
3390 private static void check(Pattern p, String s, String g, String expected) {
3391 Matcher m = p.matcher(s);
3392 m.find();
3393 if (!m.group(g).equals(expected))
3394 failCount++;
3395 }
3396
3397 private static void checkReplaceFirst(String p, String s, String r, String expected)
3398 {
3399 if (!expected.equals(Pattern.compile(p)
3400 .matcher(s)
3401 .replaceFirst(r)))
3402 failCount++;
3403 }
3404
3405 private static void checkReplaceAll(String p, String s, String r, String expected)
3406 {
3407 if (!expected.equals(Pattern.compile(p)
3408 .matcher(s)
3409 .replaceAll(r)))
3410 failCount++;
3411 }
3412
3413 private static void checkExpectedFail(String p) {
3414 try {
3415 Pattern.compile(p);
3416 } catch (PatternSyntaxException pse) {
3417 //pse.printStackTrace();
3418 return;
3419 }
3420 failCount++;
3421 }
3422
3423 private static void checkExpectedFail(Matcher m, String g) {
3424 m.find();
3425 try {
3426 m.group(g);
3427 } catch (IllegalArgumentException iae) {
3428 //iae.printStackTrace();
3429 return;
3430 } catch (NullPointerException npe) {
3431 return;
3432 }
3433 failCount++;
3434 }
3435
3436
3437 private static void namedGroupCaptureTest() throws Exception {
3438 check(Pattern.compile("x+(?<gname>y+)z+"),
3439 "xxxyyyzzz",
3440 "gname",
3441 "yyy");
3442
shermand9337e02009-10-21 11:40:40 -07003443 check(Pattern.compile("x+(?<gname8>y+)z+"),
shermanffaf4922009-03-20 16:22:59 -07003444 "xxxyyyzzz",
shermand9337e02009-10-21 11:40:40 -07003445 "gname8",
shermanffaf4922009-03-20 16:22:59 -07003446 "yyy");
3447
sherman0b4d42d2009-02-23 21:06:15 -08003448 //backref
3449 Pattern pattern = Pattern.compile("(a*)bc\\1");
3450 check(pattern, "zzzaabcazzz", true); // found "abca"
3451
3452 check(Pattern.compile("(?<gname>a*)bc\\k<gname>"),
3453 "zzzaabcaazzz", true);
3454
3455 check(Pattern.compile("(?<gname>abc)(def)\\k<gname>"),
3456 "abcdefabc", true);
3457
3458 check(Pattern.compile("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)(?<gname>k)\\k<gname>"),
3459 "abcdefghijkk", true);
3460
3461 // Supplementary character tests
3462 check(Pattern.compile("(?<gname>" + toSupplementaries("a*)bc") + "\\k<gname>"),
3463 toSupplementaries("zzzaabcazzz"), true);
3464
3465 check(Pattern.compile("(?<gname>" + toSupplementaries("a*)bc") + "\\k<gname>"),
3466 toSupplementaries("zzzaabcaazzz"), true);
3467
3468 check(Pattern.compile("(?<gname>" + toSupplementaries("abc)(def)") + "\\k<gname>"),
3469 toSupplementaries("abcdefabc"), true);
3470
3471 check(Pattern.compile(toSupplementaries("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)") +
3472 "(?<gname>" +
3473 toSupplementaries("k)") + "\\k<gname>"),
3474 toSupplementaries("abcdefghijkk"), true);
3475
3476 check(Pattern.compile("x+(?<gname>y+)z+\\k<gname>"),
3477 "xxxyyyzzzyyy",
3478 "gname",
3479 "yyy");
3480
3481 //replaceFirst/All
3482 checkReplaceFirst("(?<gn>ab)(c*)",
3483 "abccczzzabcczzzabccc",
shermand9337e02009-10-21 11:40:40 -07003484 "${gn}",
sherman0b4d42d2009-02-23 21:06:15 -08003485 "abzzzabcczzzabccc");
3486
3487 checkReplaceAll("(?<gn>ab)(c*)",
3488 "abccczzzabcczzzabccc",
shermand9337e02009-10-21 11:40:40 -07003489 "${gn}",
sherman0b4d42d2009-02-23 21:06:15 -08003490 "abzzzabzzzab");
3491
3492
3493 checkReplaceFirst("(?<gn>ab)(c*)",
3494 "zzzabccczzzabcczzzabccczzz",
shermand9337e02009-10-21 11:40:40 -07003495 "${gn}",
sherman0b4d42d2009-02-23 21:06:15 -08003496 "zzzabzzzabcczzzabccczzz");
3497
3498 checkReplaceAll("(?<gn>ab)(c*)",
3499 "zzzabccczzzabcczzzabccczzz",
shermand9337e02009-10-21 11:40:40 -07003500 "${gn}",
sherman0b4d42d2009-02-23 21:06:15 -08003501 "zzzabzzzabzzzabzzz");
3502
3503 checkReplaceFirst("(?<gn1>ab)(?<gn2>c*)",
3504 "zzzabccczzzabcczzzabccczzz",
shermand9337e02009-10-21 11:40:40 -07003505 "${gn2}",
sherman0b4d42d2009-02-23 21:06:15 -08003506 "zzzccczzzabcczzzabccczzz");
3507
3508 checkReplaceAll("(?<gn1>ab)(?<gn2>c*)",
3509 "zzzabccczzzabcczzzabccczzz",
shermand9337e02009-10-21 11:40:40 -07003510 "${gn2}",
sherman0b4d42d2009-02-23 21:06:15 -08003511 "zzzccczzzcczzzccczzz");
3512
3513 //toSupplementaries("(ab)(c*)"));
3514 checkReplaceFirst("(?<gn1>" + toSupplementaries("ab") +
3515 ")(?<gn2>" + toSupplementaries("c") + "*)",
3516 toSupplementaries("abccczzzabcczzzabccc"),
shermand9337e02009-10-21 11:40:40 -07003517 "${gn1}",
sherman0b4d42d2009-02-23 21:06:15 -08003518 toSupplementaries("abzzzabcczzzabccc"));
3519
3520
3521 checkReplaceAll("(?<gn1>" + toSupplementaries("ab") +
3522 ")(?<gn2>" + toSupplementaries("c") + "*)",
3523 toSupplementaries("abccczzzabcczzzabccc"),
shermand9337e02009-10-21 11:40:40 -07003524 "${gn1}",
sherman0b4d42d2009-02-23 21:06:15 -08003525 toSupplementaries("abzzzabzzzab"));
3526
3527 checkReplaceFirst("(?<gn1>" + toSupplementaries("ab") +
3528 ")(?<gn2>" + toSupplementaries("c") + "*)",
3529 toSupplementaries("abccczzzabcczzzabccc"),
shermand9337e02009-10-21 11:40:40 -07003530 "${gn2}",
sherman0b4d42d2009-02-23 21:06:15 -08003531 toSupplementaries("ccczzzabcczzzabccc"));
3532
3533
3534 checkReplaceAll("(?<gn1>" + toSupplementaries("ab") +
3535 ")(?<gn2>" + toSupplementaries("c") + "*)",
3536 toSupplementaries("abccczzzabcczzzabccc"),
shermand9337e02009-10-21 11:40:40 -07003537 "${gn2}",
sherman0b4d42d2009-02-23 21:06:15 -08003538 toSupplementaries("ccczzzcczzzccc"));
3539
3540 checkReplaceFirst("(?<dog>Dog)AndCat",
3541 "zzzDogAndCatzzzDogAndCatzzz",
shermand9337e02009-10-21 11:40:40 -07003542 "${dog}",
sherman0b4d42d2009-02-23 21:06:15 -08003543 "zzzDogzzzDogAndCatzzz");
3544
3545
3546 checkReplaceAll("(?<dog>Dog)AndCat",
3547 "zzzDogAndCatzzzDogAndCatzzz",
shermand9337e02009-10-21 11:40:40 -07003548 "${dog}",
sherman0b4d42d2009-02-23 21:06:15 -08003549 "zzzDogzzzDogzzz");
3550
3551 // backref in Matcher & String
shermand9337e02009-10-21 11:40:40 -07003552 if (!"abcdefghij".replaceFirst("cd(?<gn>ef)gh", "${gn}").equals("abefij") ||
3553 !"abbbcbdbefgh".replaceAll("(?<gn>[a-e])b", "${gn}").equals("abcdefgh"))
sherman0b4d42d2009-02-23 21:06:15 -08003554 failCount++;
3555
3556 // negative
3557 checkExpectedFail("(?<groupnamehasnoascii.in>abc)(def)");
3558 checkExpectedFail("(?<groupnamehasnoascii_in>abc)(def)");
shermand9337e02009-10-21 11:40:40 -07003559 checkExpectedFail("(?<6groupnamestartswithdigit>abc)(def)");
sherman0b4d42d2009-02-23 21:06:15 -08003560 checkExpectedFail("(?<gname>abc)(def)\\k<gnameX>");
3561 checkExpectedFail("(?<gname>abc)(?<gname>def)\\k<gnameX>");
3562 checkExpectedFail(Pattern.compile("(?<gname>abc)(def)").matcher("abcdef"),
3563 "gnameX");
3564 checkExpectedFail(Pattern.compile("(?<gname>abc)(def)").matcher("abcdef"),
3565 null);
3566 report("NamedGroupCapture");
3567 }
sherman6782c962010-02-05 00:10:42 -08003568
shermancc01ef52010-05-18 15:36:47 -07003569 // This is for bug 6969132
sherman6782c962010-02-05 00:10:42 -08003570 private static void nonBmpClassComplementTest() throws Exception {
3571 Pattern p = Pattern.compile("\\P{Lu}");
3572 Matcher m = p.matcher(new String(new int[] {0x1d400}, 0, 1));
3573 if (m.find() && m.start() == 1)
3574 failCount++;
3575
3576 // from a unicode category
3577 p = Pattern.compile("\\P{Lu}");
3578 m = p.matcher(new String(new int[] {0x1d400}, 0, 1));
3579 if (m.find())
3580 failCount++;
3581 if (!m.hitEnd())
3582 failCount++;
3583
3584 // block
3585 p = Pattern.compile("\\P{InMathematicalAlphanumericSymbols}");
3586 m = p.matcher(new String(new int[] {0x1d400}, 0, 1));
3587 if (m.find() && m.start() == 1)
3588 failCount++;
3589
3590 report("NonBmpClassComplement");
3591 }
3592
shermancc01ef52010-05-18 15:36:47 -07003593 private static void unicodePropertiesTest() throws Exception {
3594 // different forms
3595 if (!Pattern.compile("\\p{IsLu}").matcher("A").matches() ||
3596 !Pattern.compile("\\p{Lu}").matcher("A").matches() ||
3597 !Pattern.compile("\\p{gc=Lu}").matcher("A").matches() ||
3598 !Pattern.compile("\\p{general_category=Lu}").matcher("A").matches() ||
3599 !Pattern.compile("\\p{IsLatin}").matcher("B").matches() ||
3600 !Pattern.compile("\\p{sc=Latin}").matcher("B").matches() ||
3601 !Pattern.compile("\\p{script=Latin}").matcher("B").matches() ||
3602 !Pattern.compile("\\p{InBasicLatin}").matcher("c").matches() ||
3603 !Pattern.compile("\\p{blk=BasicLatin}").matcher("c").matches() ||
3604 !Pattern.compile("\\p{block=BasicLatin}").matcher("c").matches())
3605 failCount++;
3606
3607 Matcher common = Pattern.compile("\\p{script=Common}").matcher("");
3608 Matcher unknown = Pattern.compile("\\p{IsUnknown}").matcher("");
3609 Matcher lastSM = common;
3610 Character.UnicodeScript lastScript = Character.UnicodeScript.of(0);
3611
3612 Matcher latin = Pattern.compile("\\p{block=basic_latin}").matcher("");
3613 Matcher greek = Pattern.compile("\\p{InGreek}").matcher("");
3614 Matcher lastBM = latin;
3615 Character.UnicodeBlock lastBlock = Character.UnicodeBlock.of(0);
3616
3617 for (int cp = 1; cp < Character.MAX_CODE_POINT; cp++) {
3618 if (cp >= 0x30000 && (cp & 0x70) == 0){
3619 continue; // only pick couple code points, they are the same
3620 }
3621
3622 // Unicode Script
3623 Character.UnicodeScript script = Character.UnicodeScript.of(cp);
3624 Matcher m;
3625 String str = new String(Character.toChars(cp));
3626 if (script == lastScript) {
3627 m = lastSM;
3628 m.reset(str);
3629 } else {
3630 m = Pattern.compile("\\p{Is" + script.name() + "}").matcher(str);
3631 }
3632 if (!m.matches()) {
3633 failCount++;
3634 }
3635 Matcher other = (script == Character.UnicodeScript.COMMON)? unknown : common;
3636 other.reset(str);
3637 if (other.matches()) {
3638 failCount++;
3639 }
3640 lastSM = m;
3641 lastScript = script;
3642
3643 // Unicode Block
3644 Character.UnicodeBlock block = Character.UnicodeBlock.of(cp);
3645 if (block == null) {
3646 //System.out.printf("Not a Block: cp=%x%n", cp);
3647 continue;
3648 }
3649 if (block == lastBlock) {
3650 m = lastBM;
3651 m.reset(str);
3652 } else {
3653 m = Pattern.compile("\\p{block=" + block.toString() + "}").matcher(str);
3654 }
3655 if (!m.matches()) {
3656 failCount++;
3657 }
3658 other = (block == Character.UnicodeBlock.BASIC_LATIN)? greek : latin;
3659 other.reset(str);
3660 if (other.matches()) {
3661 failCount++;
3662 }
3663 lastBM = m;
3664 lastBlock = block;
3665 }
3666 report("unicodeProperties");
3667 }
shermanf03c78b2011-02-03 13:49:25 -08003668
3669 private static void unicodeHexNotationTest() throws Exception {
3670
3671 // negative
3672 checkExpectedFail("\\x{-23}");
3673 checkExpectedFail("\\x{110000}");
3674 checkExpectedFail("\\x{}");
3675 checkExpectedFail("\\x{AB[ef]");
3676
3677 // codepoint
3678 check("^\\x{1033c}$", "\uD800\uDF3C", true);
3679 check("^\\xF0\\x90\\x8C\\xBC$", "\uD800\uDF3C", false);
3680 check("^\\x{D800}\\x{DF3c}+$", "\uD800\uDF3C", false);
3681 check("^\\xF0\\x90\\x8C\\xBC$", "\uD800\uDF3C", false);
3682
3683 // in class
3684 check("^[\\x{D800}\\x{DF3c}]+$", "\uD800\uDF3C", false);
3685 check("^[\\xF0\\x90\\x8C\\xBC]+$", "\uD800\uDF3C", false);
3686 check("^[\\x{D800}\\x{DF3C}]+$", "\uD800\uDF3C", false);
3687 check("^[\\x{DF3C}\\x{D800}]+$", "\uD800\uDF3C", false);
3688 check("^[\\x{D800}\\x{DF3C}]+$", "\uDF3C\uD800", true);
3689 check("^[\\x{DF3C}\\x{D800}]+$", "\uDF3C\uD800", true);
3690
3691 for (int cp = 0; cp <= 0x10FFFF; cp++) {
3692 String s = "A" + new String(Character.toChars(cp)) + "B";
3693 String hexUTF16 = (cp <= 0xFFFF)? String.format("\\u%04x", cp)
3694 : String.format("\\u%04x\\u%04x",
3695 (int) Character.toChars(cp)[0],
3696 (int) Character.toChars(cp)[1]);
3697 String hexCodePoint = "\\x{" + Integer.toHexString(cp) + "}";
3698 if (!Pattern.matches("A" + hexUTF16 + "B", s))
3699 failCount++;
3700 if (!Pattern.matches("A[" + hexUTF16 + "]B", s))
3701 failCount++;
3702 if (!Pattern.matches("A" + hexCodePoint + "B", s))
3703 failCount++;
3704 if (!Pattern.matches("A[" + hexCodePoint + "]B", s))
3705 failCount++;
3706 }
3707 report("unicodeHexNotation");
sherman85bbd8b2011-04-28 20:48:36 -07003708 }
3709
3710 private static void unicodeClassesTest() throws Exception {
3711
3712 Matcher lower = Pattern.compile("\\p{Lower}").matcher("");
3713 Matcher upper = Pattern.compile("\\p{Upper}").matcher("");
3714 Matcher ASCII = Pattern.compile("\\p{ASCII}").matcher("");
3715 Matcher alpha = Pattern.compile("\\p{Alpha}").matcher("");
3716 Matcher digit = Pattern.compile("\\p{Digit}").matcher("");
3717 Matcher alnum = Pattern.compile("\\p{Alnum}").matcher("");
3718 Matcher punct = Pattern.compile("\\p{Punct}").matcher("");
3719 Matcher graph = Pattern.compile("\\p{Graph}").matcher("");
3720 Matcher print = Pattern.compile("\\p{Print}").matcher("");
3721 Matcher blank = Pattern.compile("\\p{Blank}").matcher("");
3722 Matcher cntrl = Pattern.compile("\\p{Cntrl}").matcher("");
3723 Matcher xdigit = Pattern.compile("\\p{XDigit}").matcher("");
3724 Matcher space = Pattern.compile("\\p{Space}").matcher("");
3725 Matcher bound = Pattern.compile("\\b").matcher("");
3726 Matcher word = Pattern.compile("\\w++").matcher("");
3727 // UNICODE_CHARACTER_CLASS
3728 Matcher lowerU = Pattern.compile("\\p{Lower}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3729 Matcher upperU = Pattern.compile("\\p{Upper}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3730 Matcher ASCIIU = Pattern.compile("\\p{ASCII}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3731 Matcher alphaU = Pattern.compile("\\p{Alpha}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3732 Matcher digitU = Pattern.compile("\\p{Digit}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3733 Matcher alnumU = Pattern.compile("\\p{Alnum}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3734 Matcher punctU = Pattern.compile("\\p{Punct}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3735 Matcher graphU = Pattern.compile("\\p{Graph}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3736 Matcher printU = Pattern.compile("\\p{Print}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3737 Matcher blankU = Pattern.compile("\\p{Blank}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3738 Matcher cntrlU = Pattern.compile("\\p{Cntrl}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3739 Matcher xdigitU = Pattern.compile("\\p{XDigit}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3740 Matcher spaceU = Pattern.compile("\\p{Space}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3741 Matcher boundU = Pattern.compile("\\b", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3742 Matcher wordU = Pattern.compile("\\w", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3743 // embedded flag (?U)
3744 Matcher lowerEU = Pattern.compile("(?U)\\p{Lower}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3745 Matcher graphEU = Pattern.compile("(?U)\\p{Graph}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3746 Matcher wordEU = Pattern.compile("(?U)\\w", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3747
3748 Matcher bwb = Pattern.compile("\\b\\w\\b").matcher("");
3749 Matcher bwbU = Pattern.compile("\\b\\w++\\b", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3750 Matcher bwbEU = Pattern.compile("(?U)\\b\\w++\\b", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
3751 // properties
3752 Matcher lowerP = Pattern.compile("\\p{IsLowerCase}").matcher("");
3753 Matcher upperP = Pattern.compile("\\p{IsUpperCase}").matcher("");
3754 Matcher titleP = Pattern.compile("\\p{IsTitleCase}").matcher("");
3755 Matcher letterP = Pattern.compile("\\p{IsLetter}").matcher("");
3756 Matcher alphaP = Pattern.compile("\\p{IsAlphabetic}").matcher("");
3757 Matcher ideogP = Pattern.compile("\\p{IsIdeographic}").matcher("");
3758 Matcher cntrlP = Pattern.compile("\\p{IsControl}").matcher("");
3759 Matcher spaceP = Pattern.compile("\\p{IsWhiteSpace}").matcher("");
3760 Matcher definedP = Pattern.compile("\\p{IsAssigned}").matcher("");
3761 Matcher nonCCPP = Pattern.compile("\\p{IsNoncharacterCodePoint}").matcher("");
3762
3763 // javaMethod
3764 Matcher lowerJ = Pattern.compile("\\p{javaLowerCase}").matcher("");
3765 Matcher upperJ = Pattern.compile("\\p{javaUpperCase}").matcher("");
3766 Matcher alphaJ = Pattern.compile("\\p{javaAlphabetic}").matcher("");
3767 Matcher ideogJ = Pattern.compile("\\p{javaIdeographic}").matcher("");
3768
3769 for (int cp = 1; cp < 0x30000; cp++) {
3770 String str = new String(Character.toChars(cp));
3771 int type = Character.getType(cp);
3772 if (// lower
3773 POSIX_ASCII.isLower(cp) != lower.reset(str).matches() ||
3774 Character.isLowerCase(cp) != lowerU.reset(str).matches() ||
3775 Character.isLowerCase(cp) != lowerP.reset(str).matches() ||
3776 Character.isLowerCase(cp) != lowerEU.reset(str).matches()||
3777 Character.isLowerCase(cp) != lowerJ.reset(str).matches()||
3778 // upper
3779 POSIX_ASCII.isUpper(cp) != upper.reset(str).matches() ||
3780 POSIX_Unicode.isUpper(cp) != upperU.reset(str).matches() ||
3781 Character.isUpperCase(cp) != upperP.reset(str).matches() ||
3782 Character.isUpperCase(cp) != upperJ.reset(str).matches() ||
3783 // alpha
3784 POSIX_ASCII.isAlpha(cp) != alpha.reset(str).matches() ||
3785 POSIX_Unicode.isAlpha(cp) != alphaU.reset(str).matches() ||
3786 Character.isAlphabetic(cp)!= alphaP.reset(str).matches() ||
3787 Character.isAlphabetic(cp)!= alphaJ.reset(str).matches() ||
3788 // digit
3789 POSIX_ASCII.isDigit(cp) != digit.reset(str).matches() ||
3790 Character.isDigit(cp) != digitU.reset(str).matches() ||
3791 // alnum
3792 POSIX_ASCII.isAlnum(cp) != alnum.reset(str).matches() ||
3793 POSIX_Unicode.isAlnum(cp) != alnumU.reset(str).matches() ||
3794 // punct
3795 POSIX_ASCII.isPunct(cp) != punct.reset(str).matches() ||
3796 POSIX_Unicode.isPunct(cp) != punctU.reset(str).matches() ||
3797 // graph
3798 POSIX_ASCII.isGraph(cp) != graph.reset(str).matches() ||
3799 POSIX_Unicode.isGraph(cp) != graphU.reset(str).matches() ||
3800 POSIX_Unicode.isGraph(cp) != graphEU.reset(str).matches()||
3801 // blank
3802 POSIX_ASCII.isType(cp, POSIX_ASCII.BLANK)
3803 != blank.reset(str).matches() ||
3804 POSIX_Unicode.isBlank(cp) != blankU.reset(str).matches() ||
3805 // print
3806 POSIX_ASCII.isPrint(cp) != print.reset(str).matches() ||
3807 POSIX_Unicode.isPrint(cp) != printU.reset(str).matches() ||
3808 // cntrl
3809 POSIX_ASCII.isCntrl(cp) != cntrl.reset(str).matches() ||
3810 POSIX_Unicode.isCntrl(cp) != cntrlU.reset(str).matches() ||
3811 (Character.CONTROL == type) != cntrlP.reset(str).matches() ||
3812 // hexdigit
3813 POSIX_ASCII.isHexDigit(cp) != xdigit.reset(str).matches() ||
3814 POSIX_Unicode.isHexDigit(cp) != xdigitU.reset(str).matches() ||
3815 // space
3816 POSIX_ASCII.isSpace(cp) != space.reset(str).matches() ||
3817 POSIX_Unicode.isSpace(cp) != spaceU.reset(str).matches() ||
3818 POSIX_Unicode.isSpace(cp) != spaceP.reset(str).matches() ||
3819 // word
3820 POSIX_ASCII.isWord(cp) != word.reset(str).matches() ||
3821 POSIX_Unicode.isWord(cp) != wordU.reset(str).matches() ||
3822 POSIX_Unicode.isWord(cp) != wordEU.reset(str).matches()||
3823 // bwordb
3824 POSIX_ASCII.isWord(cp) != bwb.reset(str).matches() ||
3825 POSIX_Unicode.isWord(cp) != bwbU.reset(str).matches() ||
3826 // properties
3827 Character.isTitleCase(cp) != titleP.reset(str).matches() ||
3828 Character.isLetter(cp) != letterP.reset(str).matches()||
3829 Character.isIdeographic(cp) != ideogP.reset(str).matches() ||
3830 Character.isIdeographic(cp) != ideogJ.reset(str).matches() ||
3831 (Character.UNASSIGNED == type) == definedP.reset(str).matches() ||
3832 POSIX_Unicode.isNoncharacterCodePoint(cp) != nonCCPP.reset(str).matches())
3833 failCount++;
3834 }
3835
3836 // bounds/word align
3837 twoFindIndexes(" \u0180sherman\u0400 ", bound, 1, 10);
3838 if (!bwbU.reset("\u0180sherman\u0400").matches())
3839 failCount++;
3840 twoFindIndexes(" \u0180sh\u0345erman\u0400 ", bound, 1, 11);
3841 if (!bwbU.reset("\u0180sh\u0345erman\u0400").matches())
3842 failCount++;
3843 twoFindIndexes(" \u0724\u0739\u0724 ", bound, 1, 4);
3844 if (!bwbU.reset("\u0724\u0739\u0724").matches())
3845 failCount++;
3846 if (!bwbEU.reset("\u0724\u0739\u0724").matches())
3847 failCount++;
3848 report("unicodePredefinedClasses");
3849 }
shermanecb65472012-05-08 10:57:13 -07003850
3851 private static void horizontalAndVerticalWSTest() throws Exception {
3852 String hws = new String (new char[] {
3853 0x09, 0x20, 0xa0, 0x1680, 0x180e,
3854 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005,
3855 0x2006, 0x2007, 0x2008, 0x2009, 0x200a,
3856 0x202f, 0x205f, 0x3000 });
3857 String vws = new String (new char[] {
3858 0x0a, 0x0b, 0x0c, 0x0d, 0x85, 0x2028, 0x2029 });
3859 if (!Pattern.compile("\\h+").matcher(hws).matches() ||
3860 !Pattern.compile("[\\h]+").matcher(hws).matches())
3861 failCount++;
3862 if (Pattern.compile("\\H").matcher(hws).find() ||
3863 Pattern.compile("[\\H]").matcher(hws).find())
3864 failCount++;
3865 if (!Pattern.compile("\\v+").matcher(vws).matches() ||
3866 !Pattern.compile("[\\v]+").matcher(vws).matches())
3867 failCount++;
3868 if (Pattern.compile("\\V").matcher(vws).find() ||
3869 Pattern.compile("[\\V]").matcher(vws).find())
3870 failCount++;
3871 String prefix = "abcd";
3872 String suffix = "efgh";
3873 String ng = "A";
3874 for (int i = 0; i < hws.length(); i++) {
3875 String c = String.valueOf(hws.charAt(i));
3876 Matcher m = Pattern.compile("\\h").matcher(prefix + c + suffix);
3877 if (!m.find() || !c.equals(m.group()))
3878 failCount++;
3879 m = Pattern.compile("[\\h]").matcher(prefix + c + suffix);
3880 if (!m.find() || !c.equals(m.group()))
3881 failCount++;
3882
3883 m = Pattern.compile("\\H").matcher(hws.substring(0, i) + ng + hws.substring(i));
3884 if (!m.find() || !ng.equals(m.group()))
3885 failCount++;
3886 m = Pattern.compile("[\\H]").matcher(hws.substring(0, i) + ng + hws.substring(i));
3887 if (!m.find() || !ng.equals(m.group()))
3888 failCount++;
3889 }
3890 for (int i = 0; i < vws.length(); i++) {
3891 String c = String.valueOf(vws.charAt(i));
3892 Matcher m = Pattern.compile("\\v").matcher(prefix + c + suffix);
3893 if (!m.find() || !c.equals(m.group()))
3894 failCount++;
3895 m = Pattern.compile("[\\v]").matcher(prefix + c + suffix);
3896 if (!m.find() || !c.equals(m.group()))
3897 failCount++;
3898
3899 m = Pattern.compile("\\V").matcher(vws.substring(0, i) + ng + vws.substring(i));
3900 if (!m.find() || !ng.equals(m.group()))
3901 failCount++;
3902 m = Pattern.compile("[\\V]").matcher(vws.substring(0, i) + ng + vws.substring(i));
3903 if (!m.find() || !ng.equals(m.group()))
3904 failCount++;
3905 }
3906 // \v in range is interpreted as 0x0B. This is the undocumented behavior
3907 if (!Pattern.compile("[\\v-\\v]").matcher(String.valueOf((char)0x0B)).matches())
3908 failCount++;
3909 report("horizontalAndVerticalWSTest");
3910 }
3911
3912 private static void linebreakTest() throws Exception {
3913 String linebreaks = new String (new char[] {
3914 0x0A, 0x0B, 0x0C, 0x0D, 0x85, 0x2028, 0x2029 });
3915 String crnl = "\r\n";
3916 if (!Pattern.compile("\\R+").matcher(linebreaks).matches() ||
3917 !Pattern.compile("\\R").matcher(crnl).matches() ||
3918 Pattern.compile("\\R\\R").matcher(crnl).matches())
3919 failCount++;
3920 report("linebreakTest");
3921 }
3922
sherman36e2c8f2012-08-09 10:15:26 -07003923 // #7189363
3924 private static void branchTest() throws Exception {
3925 if (!Pattern.compile("(a)?bc|d").matcher("d").find() || // greedy
3926 !Pattern.compile("(a)+bc|d").matcher("d").find() ||
3927 !Pattern.compile("(a)*bc|d").matcher("d").find() ||
3928 !Pattern.compile("(a)??bc|d").matcher("d").find() || // reluctant
3929 !Pattern.compile("(a)+?bc|d").matcher("d").find() ||
3930 !Pattern.compile("(a)*?bc|d").matcher("d").find() ||
3931 !Pattern.compile("(a)?+bc|d").matcher("d").find() || // possessive
3932 !Pattern.compile("(a)++bc|d").matcher("d").find() ||
3933 !Pattern.compile("(a)*+bc|d").matcher("d").find() ||
3934 !Pattern.compile("(a)?bc|d").matcher("d").matches() || // greedy
3935 !Pattern.compile("(a)+bc|d").matcher("d").matches() ||
3936 !Pattern.compile("(a)*bc|d").matcher("d").matches() ||
3937 !Pattern.compile("(a)??bc|d").matcher("d").matches() || // reluctant
3938 !Pattern.compile("(a)+?bc|d").matcher("d").matches() ||
3939 !Pattern.compile("(a)*?bc|d").matcher("d").matches() ||
3940 !Pattern.compile("(a)?+bc|d").matcher("d").matches() || // possessive
3941 !Pattern.compile("(a)++bc|d").matcher("d").matches() ||
3942 !Pattern.compile("(a)*+bc|d").matcher("d").matches() ||
3943 !Pattern.compile("(a)?bc|de").matcher("de").find() || // others
3944 !Pattern.compile("(a)??bc|de").matcher("de").find() ||
3945 !Pattern.compile("(a)?bc|de").matcher("de").matches() ||
3946 !Pattern.compile("(a)??bc|de").matcher("de").matches())
3947 failCount++;
3948 report("branchTest");
3949 }
3950
shermanf6f35a12013-04-26 13:59:10 -07003951 // This test is for 8007395
3952 private static void groupCurlyNotFoundSuppTest() throws Exception {
3953 String input = "test this as \ud83d\ude0d";
3954 for (String pStr : new String[] { "test(.)+(@[a-zA-Z.]+)",
3955 "test(.)*(@[a-zA-Z.]+)",
3956 "test([^B])+(@[a-zA-Z.]+)",
3957 "test([^B])*(@[a-zA-Z.]+)",
3958 "test(\\P{IsControl})+(@[a-zA-Z.]+)",
3959 "test(\\P{IsControl})*(@[a-zA-Z.]+)",
3960 }) {
3961 Matcher m = Pattern.compile(pStr, Pattern.CASE_INSENSITIVE)
3962 .matcher(input);
3963 try {
3964 if (m.find()) {
3965 failCount++;
3966 }
3967 } catch (Exception x) {
3968 failCount++;
3969 }
3970 }
3971 report("GroupCurly NotFoundSupp");
3972 }
3973
sherman0b4d42d2009-02-23 21:06:15 -08003974}