test/jdk/java/util/regex/RegExTest.java
author igerasim
Wed, 29 May 2019 13:44:27 -0700
changeset 55096 234673929e0a
parent 55013 8dae495a59e7
child 57624 67e58672c503
permissions -rw-r--r--
8224789: Parsing repetition count in regex does not detect numeric overflow Reviewed-by: rriggs, bchristi
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     1
/*
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
     2
 * Copyright (c) 1999, 2019, Oracle and/or its affiliates. All rights reserved.
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     3
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     4
 *
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     5
 * This code is free software; you can redistribute it and/or modify it
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     6
 * under the terms of the GNU General Public License version 2 only, as
21596
0e3a39f29dbc 8027696: Incorrect copyright header in the tests
serb
parents: 19604
diff changeset
     7
 * published by the Free Software Foundation.
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     8
 *
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
     9
 * This code is distributed in the hope that it will be useful, but WITHOUT
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    10
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    11
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    12
 * version 2 for more details (a copy is included in the LICENSE file that
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    13
 * accompanied this code).
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    14
 *
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    15
 * You should have received a copy of the GNU General Public License version
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    16
 * 2 along with this work; if not, write to the Free Software Foundation,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    17
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    18
 *
5506
202f599c92aa 6943119: Rebrand source copyright notices
ohair
parents: 5295
diff changeset
    19
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
202f599c92aa 6943119: Rebrand source copyright notices
ohair
parents: 5295
diff changeset
    20
 * or visit www.oracle.com if you need additional information or have any
202f599c92aa 6943119: Rebrand source copyright notices
ohair
parents: 5295
diff changeset
    21
 * questions.
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    22
 */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    23
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    24
/**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    25
 * @test
30370
984bf40cdc67 8079419: Update to RegEx test to use random number library
darcy
parents: 30046
diff changeset
    26
 * @summary tests RegExp framework (use -Dseed=X to set PRNG seed)
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    27
 * @author Mike McCloskey
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    28
 * @bug 4481568 4482696 4495089 4504687 4527731 4599621 4631553 4619345
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    29
 * 4630911 4672616 4711773 4727935 4750573 4792284 4803197 4757029 4808962
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    30
 * 4872664 4803179 4892980 4900747 4945394 4938995 4979006 4994840 4997476
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    31
 * 5013885 5003322 4988891 5098443 5110268 6173522 4829857 5027748 6376940
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    32
 * 6358731 6178785 6284152 6231989 6497148 6486934 6233084 6504326 6635133
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
    33
 * 6350801 6676425 6878475 6919132 6931676 6948903 6990617 7014645 7039066
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
    34
 * 7067045 7014640 7189363 8007395 8013252 8013254 8012646 8023647 6559590
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
    35
 * 8027645 8035076 8039124 8035975 8074678 6854417 8143854 8147531 7071819
45888
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
    36
 * 8151481 4867170 7080302 6728861 6995635 6736245 4916384 6328855 6192895
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
    37
 * 6345469 6988218 6693451 7006761 8140212 8143282 8158482 8176029 8184706
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    38
 * 8194667 8197462 8184692 8221431 8224789
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
    39
 *
45288
58be10a068c2 8180805: move RandomFactory to the top level testlibrary
iignatyev
parents: 44122
diff changeset
    40
 * @library /test/lib
55013
8dae495a59e7 8221431: Support for Unicode 12.1
naoto
parents: 49763
diff changeset
    41
 * @library /lib/testlibrary/java/lang
45466
faf1c55d2046 8181759: add explicit @build actions for jdk.test.lib classes in all :tier1 tests
iignatyev
parents: 45288
diff changeset
    42
 * @build jdk.test.lib.RandomFactory
30370
984bf40cdc67 8079419: Update to RegEx test to use random number library
darcy
parents: 30046
diff changeset
    43
 * @run main RegExTest
31645
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
    44
 * @key randomness
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    45
 */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    46
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    47
import java.io.BufferedReader;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    48
import java.io.ByteArrayInputStream;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    49
import java.io.ByteArrayOutputStream;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    50
import java.io.File;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    51
import java.io.FileInputStream;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    52
import java.io.InputStreamReader;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    53
import java.io.ObjectInputStream;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    54
import java.io.ObjectOutputStream;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    55
import java.math.BigInteger;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    56
import java.nio.CharBuffer;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    57
import java.nio.file.Files;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    58
import java.util.ArrayList;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    59
import java.util.Arrays;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    60
import java.util.List;
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    61
import java.util.Random;
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
    62
import java.util.Scanner;
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    63
import java.util.function.Function;
17447
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
    64
import java.util.function.Predicate;
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    65
import java.util.regex.Matcher;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    66
import java.util.regex.MatchResult;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    67
import java.util.regex.Pattern;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
    68
import java.util.regex.PatternSyntaxException;
45288
58be10a068c2 8180805: move RandomFactory to the top level testlibrary
iignatyev
parents: 44122
diff changeset
    69
import jdk.test.lib.RandomFactory;
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    70
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    71
/**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    72
 * This is a test class created to check the operation of
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    73
 * the Pattern and Matcher classes.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    74
 */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    75
public class RegExTest {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    76
30370
984bf40cdc67 8079419: Update to RegEx test to use random number library
darcy
parents: 30046
diff changeset
    77
    private static Random generator = RandomFactory.getRandom();
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    78
    private static boolean failure = false;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    79
    private static int failCount = 0;
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
    80
    private static String firstFailure = null;
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    81
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    82
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    83
     * Main to interpret arguments and run several tests.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    84
     *
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    85
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    86
    public static void main(String[] args) throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    87
        // Most of the tests are in a file
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    88
        processFile("TestCases.txt");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    89
        //processFile("PerlCases.txt");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    90
        processFile("BMPTestCases.txt");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    91
        processFile("SupplementaryTestCases.txt");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    92
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    93
        // These test many randomly generated char patterns
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    94
        bm();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    95
        slice();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    96
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    97
        // These are hard to put into the file
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    98
        escapes();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
    99
        blankInput();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   100
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   101
        // Substitition tests on randomly generated sequences
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   102
        globalSubstitute();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   103
        stringbufferSubstitute();
23734
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
   104
        stringbuilderSubstitute();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
   105
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   106
        substitutionBasher();
23734
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
   107
        substitutionBasher2();
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   108
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   109
        // Canonical Equivalence
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   110
        ceTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   111
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   112
        // Anchors
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   113
        anchorTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   114
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   115
        // boolean match calls
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   116
        matchesTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   117
        lookingAtTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   118
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   119
        // Pattern API
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   120
        patternMatchesTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   121
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   122
        // Misc
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   123
        lookbehindTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   124
        nullArgumentTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   125
        backRefTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   126
        groupCaptureTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   127
        caretTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   128
        charClassTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   129
        emptyPatternTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   130
        findIntTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   131
        group0Test();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   132
        longPatternTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   133
        octalTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   134
        ampersandTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   135
        negationTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   136
        splitTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   137
        appendTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   138
        caseFoldingTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   139
        commentsTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   140
        unixLinesTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   141
        replaceFirstTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   142
        gTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   143
        zTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   144
        serializeTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   145
        reluctantRepetitionTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   146
        multilineDollarTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   147
        dollarAtEndTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   148
        caretBetweenTerminatorsTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   149
        // This RFE rejected in Tiger numOccurrencesTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   150
        javaCharClassTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   151
        nonCaptureRepetitionTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   152
        notCapturedGroupCurlyMatchTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   153
        escapedSegmentTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   154
        literalPatternTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   155
        literalReplacementTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   156
        regionTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   157
        toStringTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   158
        negatedCharClassTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   159
        findFromTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   160
        boundsTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   161
        unicodeWordBoundsTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   162
        caretAtEndTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   163
        wordSearchTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   164
        hitEndTest();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   165
        toMatchResultTest();
29503
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   166
        toMatchResultTest2();
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   167
        surrogatesInClassTest();
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   168
        removeQEQuotingTest();
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   169
        namedGroupCaptureTest();
4820
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
   170
        nonBmpClassComplementTest();
5610
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
   171
        unicodePropertiesTest();
8173
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   172
        unicodeHexNotationTest();
9536
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
   173
        unicodeClassesTest();
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
   174
        unicodeCharacterNameTest();
12675
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   175
        horizontalAndVerticalWSTest();
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   176
        linebreakTest();
13554
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
   177
        branchTest();
17183
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
   178
        groupCurlyNotFoundSuppTest();
19604
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
   179
        groupCurlyBackoffTest();
17447
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
   180
        patternAsPredicate();
49763
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
   181
        patternAsMatchPredicate();
25523
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
   182
        invalidFlags();
38777
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
   183
        embeddedFlags();
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
   184
        grapheme();
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
   185
        expoBacktracking();
48853
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
   186
        invalidGroupName();
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
   187
        illegalRepetitionRange();
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
   188
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   189
        if (failure) {
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   190
            throw new
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   191
                RuntimeException("RegExTest failed, 1st failure: " +
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   192
                                 firstFailure);
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   193
        } else {
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   194
            System.err.println("OKAY: All tests passed.");
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   195
        }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   196
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   197
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   198
    // Utility functions
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   199
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   200
    private static String getRandomAlphaString(int length) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   201
        StringBuffer buf = new StringBuffer(length);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   202
        for (int i=0; i<length; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   203
            char randChar = (char)(97 + generator.nextInt(26));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   204
            buf.append(randChar);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   205
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   206
        return buf.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   207
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   208
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   209
    private static void check(Matcher m, String expected) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   210
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   211
        if (!m.group().equals(expected))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   212
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   213
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   214
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   215
    private static void check(Matcher m, String result, boolean expected) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   216
        m.find();
8173
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   217
        if (m.group().equals(result) != expected)
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   218
            failCount++;
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   219
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   220
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   221
    private static void check(Pattern p, String s, boolean expected) {
8173
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   222
        if (p.matcher(s).find() != expected)
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   223
            failCount++;
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   224
    }
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   225
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   226
    private static void check(String p, String s, boolean expected) {
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   227
        Matcher matcher = Pattern.compile(p).matcher(s);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   228
        if (matcher.find() != expected)
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
   229
            failCount++;
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   230
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   231
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   232
    private static void check(String p, char c, boolean expected) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   233
        String propertyPattern = expected ? "\\p" + p : "\\P" + p;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   234
        Pattern pattern = Pattern.compile(propertyPattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   235
        char[] ca = new char[1]; ca[0] = c;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   236
        Matcher matcher = pattern.matcher(new String(ca));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   237
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   238
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   239
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   240
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   241
    private static void check(String p, int codePoint, boolean expected) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   242
        String propertyPattern = expected ? "\\p" + p : "\\P" + p;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   243
        Pattern pattern = Pattern.compile(propertyPattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   244
        char[] ca = Character.toChars(codePoint);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   245
        Matcher matcher = pattern.matcher(new String(ca));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   246
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   247
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   248
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   249
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   250
    private static void check(String p, int flag, String input, String s,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   251
                              boolean expected)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   252
    {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   253
        Pattern pattern = Pattern.compile(p, flag);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   254
        Matcher matcher = pattern.matcher(input);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   255
        if (expected)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   256
            check(matcher, s, expected);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   257
        else
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   258
            check(pattern, input, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   259
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   260
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   261
    private static void report(String testName) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   262
        int spacesToAdd = 30 - testName.length();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   263
        StringBuffer paddedNameBuffer = new StringBuffer(testName);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   264
        for (int i=0; i<spacesToAdd; i++)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   265
            paddedNameBuffer.append(" ");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   266
        String paddedName = paddedNameBuffer.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   267
        System.err.println(paddedName + ": " +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   268
                           (failCount==0 ? "Passed":"Failed("+failCount+")"));
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   269
        if (failCount > 0) {
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   270
            failure = true;
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   271
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   272
            if (firstFailure == null) {
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   273
                firstFailure = testName;
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   274
            }
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   275
        }
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   276
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   277
        failCount = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   278
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   279
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   280
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   281
     * Converts ASCII alphabet characters [A-Za-z] in the given 's' to
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   282
     * supplementary characters. This method does NOT fully take care
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   283
     * of the regex syntax.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   284
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   285
    private static String toSupplementaries(String s) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   286
        int length = s.length();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   287
        StringBuffer sb = new StringBuffer(length * 2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   288
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   289
        for (int i = 0; i < length; ) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   290
            char c = s.charAt(i++);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   291
            if (c == '\\') {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   292
                sb.append(c);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   293
                if (i < length) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   294
                    c = s.charAt(i++);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   295
                    sb.append(c);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   296
                    if (c == 'u') {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   297
                        // assume no syntax error
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   298
                        sb.append(s.charAt(i++));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   299
                        sb.append(s.charAt(i++));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   300
                        sb.append(s.charAt(i++));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   301
                        sb.append(s.charAt(i++));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   302
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   303
                }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   304
            } else if ((c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z')) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   305
                sb.append('\ud800').append((char)('\udc00'+c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   306
            } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   307
                sb.append(c);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   308
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   309
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   310
        return sb.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   311
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   312
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   313
    // Regular expression tests
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   314
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   315
    // This is for bug 6178785
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   316
    // Test if an expected NPE gets thrown when passing in a null argument
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   317
    private static boolean check(Runnable test) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   318
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   319
            test.run();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   320
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   321
            return false;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   322
        } catch (NullPointerException npe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   323
            return true;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   324
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   325
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   326
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   327
    private static void nullArgumentTest() {
29243
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   328
        check(() -> Pattern.compile(null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   329
        check(() -> Pattern.matches(null, null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   330
        check(() -> Pattern.matches("xyz", null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   331
        check(() -> Pattern.quote(null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   332
        check(() -> Pattern.compile("xyz").split(null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   333
        check(() -> Pattern.compile("xyz").matcher(null));
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   334
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   335
        final Matcher m = Pattern.compile("xyz").matcher("xyz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   336
        m.matches();
29243
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   337
        check(() -> m.appendTail((StringBuffer) null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   338
        check(() -> m.appendTail((StringBuilder)null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   339
        check(() -> m.replaceAll((String) null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   340
        check(() -> m.replaceAll((Function<MatchResult, String>)null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   341
        check(() -> m.replaceFirst((String)null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   342
        check(() -> m.replaceFirst((Function<MatchResult, String>) null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   343
        check(() -> m.appendReplacement((StringBuffer)null, null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   344
        check(() -> m.appendReplacement((StringBuilder)null, null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   345
        check(() -> m.reset(null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   346
        check(() -> Matcher.quoteReplacement(null));
80ea8d3d39d0 8071479: Stream and lambdafication improvements to j.u.regex.Matcher
psandoz
parents: 25523
diff changeset
   347
        //check(() -> m.usePattern(null));
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   348
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   349
        report("Null Argument");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   350
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   351
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   352
    // This is for bug6635133
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   353
    // Test if surrogate pair in Unicode escapes can be handled correctly.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   354
    private static void surrogatesInClassTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   355
        Pattern pattern = Pattern.compile("[\\ud834\\udd21-\\ud834\\udd24]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   356
        Matcher matcher = pattern.matcher("\ud834\udd22");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   357
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   358
            failCount++;
11287
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   359
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   360
        report("Surrogate pair in Unicode escape");
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   361
    }
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   362
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   363
    // This is for bug6990617
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   364
    // Test if Pattern.RemoveQEQuoting works correctly if the octal unicode
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   365
    // char encoding is only 2 or 3 digits instead of 4 and the first quoted
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   366
    // char is an octal digit.
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   367
    private static void removeQEQuotingTest() throws Exception {
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   368
        Pattern pattern =
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   369
            Pattern.compile("\\011\\Q1sometext\\E\\011\\Q2sometext\\E");
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   370
        Matcher matcher = pattern.matcher("\t1sometext\t2sometext");
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   371
        if (!matcher.find())
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   372
            failCount++;
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   373
3db172a5433c 6990617: Regular expression doesn't match if unicode character next to a digit.
sherman
parents: 9536
diff changeset
   374
        report("Remove Q/E Quoting");
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   375
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   376
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   377
    // This is for bug 4988891
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   378
    // Test toMatchResult to see that it is a copy of the Matcher
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   379
    // that is not affected by subsequent operations on the original
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   380
    private static void toMatchResultTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   381
        Pattern pattern = Pattern.compile("squid");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   382
        Matcher matcher = pattern.matcher(
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   383
            "agiantsquidofdestinyasmallsquidoffate");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   384
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   385
        int matcherStart1 = matcher.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   386
        MatchResult mr = matcher.toMatchResult();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   387
        if (mr == matcher)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   388
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   389
        int resultStart1 = mr.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   390
        if (matcherStart1 != resultStart1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   391
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   392
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   393
        int matcherStart2 = matcher.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   394
        int resultStart2 = mr.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   395
        if (matcherStart2 == resultStart2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   396
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   397
        if (resultStart1 != resultStart2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   398
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   399
        MatchResult mr2 = matcher.toMatchResult();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   400
        if (mr == mr2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   401
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   402
        if (mr2.start() != matcherStart2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   403
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   404
        report("toMatchResult is a copy");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   405
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   406
29503
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   407
    private static void checkExpectedISE(Runnable test) {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   408
        try {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   409
            test.run();
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   410
            failCount++;
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   411
        } catch (IllegalStateException x) {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   412
        } catch (IndexOutOfBoundsException xx) {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   413
            failCount++;
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   414
        }
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   415
    }
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   416
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   417
    private static void checkExpectedIOOE(Runnable test) {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   418
        try {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   419
            test.run();
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   420
            failCount++;
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   421
        } catch (IndexOutOfBoundsException x) {}
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   422
    }
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   423
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   424
    // This is for bug 8074678
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   425
    // Test the result of toMatchResult throws ISE if no match is availble
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   426
    private static void toMatchResultTest2() throws Exception {
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   427
        Matcher matcher = Pattern.compile("nomatch").matcher("hello world");
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   428
        matcher.find();
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   429
        MatchResult mr = matcher.toMatchResult();
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   430
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   431
        checkExpectedISE(() -> mr.start());
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   432
        checkExpectedISE(() -> mr.start(2));
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   433
        checkExpectedISE(() -> mr.end());
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   434
        checkExpectedISE(() -> mr.end(2));
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   435
        checkExpectedISE(() -> mr.group());
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   436
        checkExpectedISE(() -> mr.group(2));
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   437
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   438
        matcher = Pattern.compile("(match)").matcher("there is a match");
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   439
        matcher.find();
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   440
        MatchResult mr2 = matcher.toMatchResult();
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   441
        checkExpectedIOOE(() -> mr2.start(2));
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   442
        checkExpectedIOOE(() -> mr2.end(2));
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   443
        checkExpectedIOOE(() -> mr2.group(2));
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   444
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   445
        report("toMatchResult2 appropriate exceptions");
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   446
    }
95318f222149 8074678: JCK test java_util/regex/MatchResult/index.html starts failing after JDK-8071479
sherman
parents: 29243
diff changeset
   447
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   448
    // This is for bug 5013885
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   449
    // Must test a slice to see if it reports hitEnd correctly
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   450
    private static void hitEndTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   451
        // Basic test of Slice node
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   452
        Pattern p = Pattern.compile("^squidattack");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   453
        Matcher m = p.matcher("squack");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   454
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   455
        if (m.hitEnd())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   456
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   457
        m.reset("squid");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   458
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   459
        if (!m.hitEnd())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   460
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   461
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   462
        // Test Slice, SliceA and SliceU nodes
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   463
        for (int i=0; i<3; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   464
            int flags = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   465
            if (i==1) flags = Pattern.CASE_INSENSITIVE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   466
            if (i==2) flags = Pattern.UNICODE_CASE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   467
            p = Pattern.compile("^abc", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   468
            m = p.matcher("ad");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   469
            m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   470
            if (m.hitEnd())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   471
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   472
            m.reset("ab");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   473
            m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   474
            if (!m.hitEnd())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   475
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   476
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   477
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   478
        // Test Boyer-Moore node
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   479
        p = Pattern.compile("catattack");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   480
        m = p.matcher("attack");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   481
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   482
        if (!m.hitEnd())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   483
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   484
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   485
        p = Pattern.compile("catattack");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   486
        m = p.matcher("attackattackattackcatatta");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   487
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   488
        if (!m.hitEnd())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   489
            failCount++;
45888
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   490
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   491
        // 8184706: Matching u+0d at EOL against \R should hit-end
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   492
        p = Pattern.compile("...\\R");
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   493
        m = p.matcher("cat" + (char)0x0a);
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   494
        m.find();
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   495
        if (m.hitEnd())
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   496
            failCount++;
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   497
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   498
        m = p.matcher("cat" + (char)0x0d);
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   499
        m.find();
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   500
        if (!m.hitEnd())
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   501
            failCount++;
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   502
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   503
        m = p.matcher("cat" + (char)0x0d + (char)0x0a);
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   504
        m.find();
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   505
        if (m.hitEnd())
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   506
            failCount++;
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   507
ae448cf92e00 8184706: Matcher doesn't indicate hitEnd after matching \u0D with \R at EOL
igerasim
parents: 45466
diff changeset
   508
        report("hitEnd");
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   509
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   510
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   511
    // This is for bug 4997476
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   512
    // It is weird code submitted by customer demonstrating a regression
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   513
    private static void wordSearchTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   514
        String testString = new String("word1 word2 word3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   515
        Pattern p = Pattern.compile("\\b");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   516
        Matcher m = p.matcher(testString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   517
        int position = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   518
        int start = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   519
        while (m.find(position)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   520
            start = m.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   521
            if (start == testString.length())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   522
                break;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   523
            if (m.find(start+1)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   524
                position = m.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   525
            } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   526
                position = testString.length();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   527
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   528
            if (testString.substring(start, position).equals(" "))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   529
                continue;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   530
            if (!testString.substring(start, position-1).startsWith("word"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   531
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   532
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   533
        report("Customer word search");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   534
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   535
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   536
    // This is for bug 4994840
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   537
    private static void caretAtEndTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   538
        // Problem only occurs with multiline patterns
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   539
        // containing a beginning-of-line caret "^" followed
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   540
        // by an expression that also matches the empty string.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   541
        Pattern pattern = Pattern.compile("^x?", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   542
        Matcher matcher = pattern.matcher("\r");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   543
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   544
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   545
        report("Caret at end");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   546
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   547
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   548
    // This test is for 4979006
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   549
    // Check to see if word boundary construct properly handles unicode
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   550
    // non spacing marks
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   551
    private static void unicodeWordBoundsTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   552
        String spaces = "  ";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   553
        String wordChar = "a";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   554
        String nsm = "\u030a";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   555
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   556
        assert (Character.getType('\u030a') == Character.NON_SPACING_MARK);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   557
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   558
        Pattern pattern = Pattern.compile("\\b");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   559
        Matcher matcher = pattern.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   560
        // S=other B=word character N=non spacing mark .=word boundary
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   561
        // SS.BB.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   562
        String input = spaces + wordChar + wordChar + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   563
        twoFindIndexes(input, matcher, 2, 4);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   564
        // SS.BBN.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   565
        input = spaces + wordChar +wordChar + nsm + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   566
        twoFindIndexes(input, matcher, 2, 5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   567
        // SS.BN.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   568
        input = spaces + wordChar + nsm + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   569
        twoFindIndexes(input, matcher, 2, 4);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   570
        // SS.BNN.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   571
        input = spaces + wordChar + nsm + nsm + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   572
        twoFindIndexes(input, matcher, 2, 5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   573
        // SSN.BB.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   574
        input = spaces + nsm + wordChar + wordChar + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   575
        twoFindIndexes(input, matcher, 3, 5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   576
        // SS.BNB.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   577
        input = spaces + wordChar + nsm + wordChar + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   578
        twoFindIndexes(input, matcher, 2, 5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   579
        // SSNNSS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   580
        input = spaces + nsm + nsm + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   581
        matcher.reset(input);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   582
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   583
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   584
        // SSN.BBN.SS
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   585
        input = spaces + nsm + wordChar + wordChar + nsm + spaces;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   586
        twoFindIndexes(input, matcher, 3, 6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   587
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   588
        report("Unicode word boundary");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   589
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   590
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   591
    private static void twoFindIndexes(String input, Matcher matcher, int a,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   592
                                       int b) throws Exception
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   593
    {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   594
        matcher.reset(input);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   595
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   596
        if (matcher.start() != a)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   597
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   598
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   599
        if (matcher.start() != b)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   600
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   601
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   602
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   603
    // This test is for 6284152
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   604
    static void check(String regex, String input, String[] expected) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   605
        List<String> result = new ArrayList<String>();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   606
        Pattern p = Pattern.compile(regex);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   607
        Matcher m = p.matcher(input);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   608
        while (m.find()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   609
            result.add(m.group());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   610
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   611
        if (!Arrays.asList(expected).equals(result))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   612
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   613
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   614
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   615
    private static void lookbehindTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   616
        //Positive
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   617
        check("(?<=%.{0,5})foo\\d",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   618
              "%foo1\n%bar foo2\n%bar  foo3\n%blahblah foo4\nfoo5",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   619
              new String[]{"foo1", "foo2", "foo3"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   620
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   621
        //boundary at end of the lookbehind sub-regex should work consistently
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   622
        //with the boundary just after the lookbehind sub-regex
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   623
        check("(?<=.*\\b)foo", "abcd foo", new String[]{"foo"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   624
        check("(?<=.*)\\bfoo", "abcd foo", new String[]{"foo"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   625
        check("(?<!abc )\\bfoo", "abc foo", new String[0]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   626
        check("(?<!abc \\b)foo", "abc foo", new String[0]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   627
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   628
        //Negative
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   629
        check("(?<!%.{0,5})foo\\d",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   630
              "%foo1\n%bar foo2\n%bar  foo3\n%blahblah foo4\nfoo5",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   631
              new String[] {"foo4", "foo5"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   632
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   633
        //Positive greedy
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   634
        check("(?<=%b{1,4})foo", "%bbbbfoo", new String[] {"foo"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   635
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   636
        //Positive reluctant
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   637
        check("(?<=%b{1,4}?)foo", "%bbbbfoo", new String[] {"foo"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   638
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   639
        //supplementary
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   640
        check("(?<=%b{1,4})fo\ud800\udc00o", "%bbbbfo\ud800\udc00o",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   641
              new String[] {"fo\ud800\udc00o"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   642
        check("(?<=%b{1,4}?)fo\ud800\udc00o", "%bbbbfo\ud800\udc00o",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   643
              new String[] {"fo\ud800\udc00o"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   644
        check("(?<!%b{1,4})fo\ud800\udc00o", "%afo\ud800\udc00o",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   645
              new String[] {"fo\ud800\udc00o"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   646
        check("(?<!%b{1,4}?)fo\ud800\udc00o", "%afo\ud800\udc00o",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   647
              new String[] {"fo\ud800\udc00o"});
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   648
        report("Lookbehind");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   649
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   650
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   651
    // This test is for 4938995
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   652
    // Check to see if weak region boundaries are transparent to
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   653
    // lookahead and lookbehind constructs
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   654
    private static void boundsTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   655
        String fullMessage = "catdogcat";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   656
        Pattern pattern = Pattern.compile("(?<=cat)dog(?=cat)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   657
        Matcher matcher = pattern.matcher("catdogca");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   658
        matcher.useTransparentBounds(true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   659
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   660
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   661
        matcher.reset("atdogcat");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   662
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   663
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   664
        matcher.reset(fullMessage);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   665
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   666
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   667
        matcher.reset(fullMessage);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   668
        matcher.region(0,9);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   669
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   670
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   671
        matcher.reset(fullMessage);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   672
        matcher.region(0,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   673
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   674
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   675
        matcher.reset(fullMessage);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   676
        matcher.region(3,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   677
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   678
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   679
        matcher.useTransparentBounds(false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   680
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   681
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   682
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   683
        // Negative lookahead/lookbehind
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   684
        pattern = Pattern.compile("(?<!cat)dog(?!cat)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   685
        matcher = pattern.matcher("dogcat");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   686
        matcher.useTransparentBounds(true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   687
        matcher.region(0,3);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   688
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   689
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   690
        matcher.reset("catdog");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   691
        matcher.region(3,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   692
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   693
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   694
        matcher.useTransparentBounds(false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   695
        matcher.reset("dogcat");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   696
        matcher.region(0,3);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   697
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   698
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   699
        matcher.reset("catdog");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   700
        matcher.region(3,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   701
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   702
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   703
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   704
        report("Region bounds transparency");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   705
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   706
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   707
    // This test is for 4945394
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   708
    private static void findFromTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   709
        String message = "This is 40 $0 message.";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   710
        Pattern pat = Pattern.compile("\\$0");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   711
        Matcher match = pat.matcher(message);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   712
        if (!match.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   713
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   714
        if (match.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   715
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   716
        if (match.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   717
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   718
        report("Check for alternating find");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   719
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   720
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   721
    // This test is for 4872664 and 4892980
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   722
    private static void negatedCharClassTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   723
        Pattern pattern = Pattern.compile("[^>]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   724
        Matcher matcher = pattern.matcher("\u203A");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   725
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   726
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   727
        pattern = Pattern.compile("[^fr]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   728
        matcher = pattern.matcher("a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   729
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   730
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   731
        matcher.reset("\u203A");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   732
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   733
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   734
        String s = "for";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   735
        String result[] = s.split("[^fr]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   736
        if (!result[0].equals("f"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   737
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   738
        if (!result[1].equals("r"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   739
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   740
        s = "f\u203Ar";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   741
        result = s.split("[^fr]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   742
        if (!result[0].equals("f"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   743
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   744
        if (!result[1].equals("r"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   745
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   746
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   747
        // Test adding to bits, subtracting a node, then adding to bits again
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   748
        pattern = Pattern.compile("[^f\u203Ar]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   749
        matcher = pattern.matcher("a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   750
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   751
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   752
        matcher.reset("f");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   753
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   754
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   755
        matcher.reset("\u203A");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   756
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   757
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   758
        matcher.reset("r");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   759
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   760
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   761
        matcher.reset("\u203B");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   762
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   763
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   764
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   765
        // Test subtracting a node, adding to bits, subtracting again
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   766
        pattern = Pattern.compile("[^\u203Ar\u203B]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   767
        matcher = pattern.matcher("a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   768
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   769
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   770
        matcher.reset("\u203A");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   771
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   772
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   773
        matcher.reset("r");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   774
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   775
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   776
        matcher.reset("\u203B");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   777
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   778
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   779
        matcher.reset("\u203C");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   780
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   781
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   782
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   783
        report("Negated Character Class");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   784
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   785
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   786
    // This test is for 4628291
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   787
    private static void toStringTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   788
        Pattern pattern = Pattern.compile("b+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   789
        if (pattern.toString() != "b+")
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   790
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   791
        Matcher matcher = pattern.matcher("aaabbbccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   792
        String matcherString = matcher.toString(); // unspecified
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   793
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   794
        matcherString = matcher.toString(); // unspecified
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   795
        matcher.region(0,3);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   796
        matcherString = matcher.toString(); // unspecified
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   797
        matcher.reset();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   798
        matcherString = matcher.toString(); // unspecified
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   799
        report("toString");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   800
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   801
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   802
    // This test is for 4808962
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   803
    private static void literalPatternTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   804
        int flags = Pattern.LITERAL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   805
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   806
        Pattern pattern = Pattern.compile("abc\\t$^", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   807
        check(pattern, "abc\\t$^", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   808
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   809
        pattern = Pattern.compile(Pattern.quote("abc\\t$^"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   810
        check(pattern, "abc\\t$^", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   811
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   812
        pattern = Pattern.compile("\\Qa^$bcabc\\E", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   813
        check(pattern, "\\Qa^$bcabc\\E", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   814
        check(pattern, "a^$bcabc", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   815
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   816
        pattern = Pattern.compile("\\\\Q\\\\E");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   817
        check(pattern, "\\Q\\E", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   818
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   819
        pattern = Pattern.compile("\\Qabc\\Eefg\\\\Q\\\\Ehij");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   820
        check(pattern, "abcefg\\Q\\Ehij", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   821
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   822
        pattern = Pattern.compile("\\\\\\Q\\\\E");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   823
        check(pattern, "\\\\\\\\", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   824
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   825
        pattern = Pattern.compile(Pattern.quote("\\Qa^$bcabc\\E"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   826
        check(pattern, "\\Qa^$bcabc\\E", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   827
        check(pattern, "a^$bcabc", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   828
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   829
        pattern = Pattern.compile(Pattern.quote("\\Qabc\\Edef"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   830
        check(pattern, "\\Qabc\\Edef", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   831
        check(pattern, "abcdef", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   832
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   833
        pattern = Pattern.compile(Pattern.quote("abc\\Edef"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   834
        check(pattern, "abc\\Edef", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   835
        check(pattern, "abcdef", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   836
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   837
        pattern = Pattern.compile(Pattern.quote("\\E"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   838
        check(pattern, "\\E", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   839
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   840
        pattern = Pattern.compile("((((abc.+?:)", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   841
        check(pattern, "((((abc.+?:)", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   842
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   843
        flags |= Pattern.MULTILINE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   844
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   845
        pattern = Pattern.compile("^cat$", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   846
        check(pattern, "abc^cat$def", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   847
        check(pattern, "cat", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   848
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   849
        flags |= Pattern.CASE_INSENSITIVE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   850
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   851
        pattern = Pattern.compile("abcdef", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   852
        check(pattern, "ABCDEF", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   853
        check(pattern, "AbCdEf", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   854
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   855
        flags |= Pattern.DOTALL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   856
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   857
        pattern = Pattern.compile("a...b", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   858
        check(pattern, "A...b", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   859
        check(pattern, "Axxxb", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   860
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   861
        flags |= Pattern.CANON_EQ;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   862
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   863
        Pattern p = Pattern.compile("testa\u030a", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   864
        check(pattern, "testa\u030a", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   865
        check(pattern, "test\u00e5", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   866
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   867
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   868
        flags = Pattern.LITERAL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   869
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   870
        pattern = Pattern.compile(toSupplementaries("abc\\t$^"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   871
        check(pattern, toSupplementaries("abc\\t$^"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   872
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   873
        pattern = Pattern.compile(Pattern.quote(toSupplementaries("abc\\t$^")));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   874
        check(pattern, toSupplementaries("abc\\t$^"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   875
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   876
        pattern = Pattern.compile(toSupplementaries("\\Qa^$bcabc\\E"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   877
        check(pattern, toSupplementaries("\\Qa^$bcabc\\E"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   878
        check(pattern, toSupplementaries("a^$bcabc"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   879
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   880
        pattern = Pattern.compile(Pattern.quote(toSupplementaries("\\Qa^$bcabc\\E")));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   881
        check(pattern, toSupplementaries("\\Qa^$bcabc\\E"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   882
        check(pattern, toSupplementaries("a^$bcabc"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   883
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   884
        pattern = Pattern.compile(Pattern.quote(toSupplementaries("\\Qabc\\Edef")));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   885
        check(pattern, toSupplementaries("\\Qabc\\Edef"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   886
        check(pattern, toSupplementaries("abcdef"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   887
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   888
        pattern = Pattern.compile(Pattern.quote(toSupplementaries("abc\\Edef")));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   889
        check(pattern, toSupplementaries("abc\\Edef"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   890
        check(pattern, toSupplementaries("abcdef"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   891
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   892
        pattern = Pattern.compile(toSupplementaries("((((abc.+?:)"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   893
        check(pattern, toSupplementaries("((((abc.+?:)"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   894
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   895
        flags |= Pattern.MULTILINE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   896
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   897
        pattern = Pattern.compile(toSupplementaries("^cat$"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   898
        check(pattern, toSupplementaries("abc^cat$def"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   899
        check(pattern, toSupplementaries("cat"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   900
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   901
        flags |= Pattern.DOTALL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   902
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   903
        // note: this is case-sensitive.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   904
        pattern = Pattern.compile(toSupplementaries("a...b"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   905
        check(pattern, toSupplementaries("a...b"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   906
        check(pattern, toSupplementaries("axxxb"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   907
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   908
        flags |= Pattern.CANON_EQ;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   909
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   910
        String t = toSupplementaries("test");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   911
        p = Pattern.compile(t + "a\u030a", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   912
        check(pattern, t + "a\u030a", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   913
        check(pattern, t + "\u00e5", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   914
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   915
        report("Literal pattern");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   916
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   917
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   918
    // This test is for 4803179
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   919
    // This test is also for 4808962, replacement parts
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   920
    private static void literalReplacementTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   921
        int flags = Pattern.LITERAL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   922
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   923
        Pattern pattern = Pattern.compile("abc", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   924
        Matcher matcher = pattern.matcher("zzzabczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   925
        String replaceTest = "$0";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   926
        String result = matcher.replaceAll(replaceTest);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   927
        if (!result.equals("zzzabczzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   928
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   929
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   930
        matcher.reset();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   931
        String literalReplacement = matcher.quoteReplacement(replaceTest);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   932
        result = matcher.replaceAll(literalReplacement);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   933
        if (!result.equals("zzz$0zzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   934
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   935
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   936
        matcher.reset();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   937
        replaceTest = "\\t$\\$";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   938
        literalReplacement = matcher.quoteReplacement(replaceTest);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   939
        result = matcher.replaceAll(literalReplacement);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   940
        if (!result.equals("zzz\\t$\\$zzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   941
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   942
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   943
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   944
        pattern = Pattern.compile(toSupplementaries("abc"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   945
        matcher = pattern.matcher(toSupplementaries("zzzabczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   946
        replaceTest = "$0";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   947
        result = matcher.replaceAll(replaceTest);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   948
        if (!result.equals(toSupplementaries("zzzabczzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   949
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   950
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   951
        matcher.reset();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   952
        literalReplacement = matcher.quoteReplacement(replaceTest);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   953
        result = matcher.replaceAll(literalReplacement);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   954
        if (!result.equals(toSupplementaries("zzz$0zzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   955
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   956
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   957
        matcher.reset();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   958
        replaceTest = "\\t$\\$";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   959
        literalReplacement = matcher.quoteReplacement(replaceTest);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   960
        result = matcher.replaceAll(literalReplacement);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   961
        if (!result.equals(toSupplementaries("zzz\\t$\\$zzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   962
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   963
12432
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   964
        // IAE should be thrown if backslash or '$' is the last character
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   965
        // in replacement string
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   966
        try {
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   967
            "\uac00".replaceAll("\uac00", "$");
12675
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   968
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   969
        } catch (IllegalArgumentException iie) {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   970
        } catch (Exception e) {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   971
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   972
        }
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
   973
        try {
12432
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   974
            "\uac00".replaceAll("\uac00", "\\");
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   975
            failCount++;
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   976
        } catch (IllegalArgumentException iie) {
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   977
        } catch (Exception e) {
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   978
            failCount++;
748fcaa0e0c1 7067045: replaceAll("\u20ac", "$"); causses java.lang.StringIndexOutOfBoundsExceptio
sherman
parents: 11287
diff changeset
   979
        }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   980
        report("Literal replacement");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   981
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   982
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   983
    // This test is for 4757029
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   984
    private static void regionTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   985
        Pattern pattern = Pattern.compile("abc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   986
        Matcher matcher = pattern.matcher("abcdefabc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   987
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   988
        matcher.region(0,9);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   989
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   990
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   991
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   992
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   993
        matcher.region(0,3);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   994
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   995
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   996
        matcher.region(3,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   997
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   998
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
   999
        matcher.region(0,2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1000
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1001
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1002
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1003
        expectRegionFail(matcher, 1, -1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1004
        expectRegionFail(matcher, -1, -1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1005
        expectRegionFail(matcher, -1, 1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1006
        expectRegionFail(matcher, 5, 3);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1007
        expectRegionFail(matcher, 5, 12);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1008
        expectRegionFail(matcher, 12, 12);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1009
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1010
        pattern = Pattern.compile("^abc$");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1011
        matcher = pattern.matcher("zzzabczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1012
        matcher.region(0,9);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1013
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1014
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1015
        matcher.region(3,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1016
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1017
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1018
        matcher.region(3,6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1019
        matcher.useAnchoringBounds(false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1020
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1021
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1022
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1023
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1024
        pattern = Pattern.compile(toSupplementaries("abc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1025
        matcher = pattern.matcher(toSupplementaries("abcdefabc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1026
        matcher.region(0,9*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1027
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1028
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1029
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1030
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1031
        matcher.region(0,3*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1032
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1033
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1034
        matcher.region(1,3*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1035
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1036
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1037
        matcher.region(3*2,6*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1038
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1039
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1040
        matcher.region(0,2*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1041
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1042
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1043
        matcher.region(0,2*2+1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1044
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1045
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1046
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1047
        expectRegionFail(matcher, 1*2, -1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1048
        expectRegionFail(matcher, -1, -1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1049
        expectRegionFail(matcher, -1, 1*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1050
        expectRegionFail(matcher, 5*2, 3*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1051
        expectRegionFail(matcher, 5*2, 12*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1052
        expectRegionFail(matcher, 12*2, 12*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1053
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1054
        pattern = Pattern.compile(toSupplementaries("^abc$"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1055
        matcher = pattern.matcher(toSupplementaries("zzzabczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1056
        matcher.region(0,9*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1057
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1058
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1059
        matcher.region(3*2,6*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1060
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1061
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1062
        matcher.region(3*2+1,6*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1063
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1064
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1065
        matcher.region(3*2,6*2-1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1066
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1067
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1068
        matcher.region(3*2,6*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1069
        matcher.useAnchoringBounds(false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1070
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1071
           failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1072
        report("Regions");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1073
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1074
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1075
    private static void expectRegionFail(Matcher matcher, int index1,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1076
                                         int index2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1077
    {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1078
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1079
            matcher.region(index1, index2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1080
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1081
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1082
            // Correct result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1083
        } catch (IllegalStateException ise) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1084
            // Correct result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1085
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1086
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1087
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1088
    // This test is for 4803197
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1089
    private static void escapedSegmentTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1090
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1091
        Pattern pattern = Pattern.compile("\\Qdir1\\dir2\\E");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1092
        check(pattern, "dir1\\dir2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1093
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1094
        pattern = Pattern.compile("\\Qdir1\\dir2\\\\E");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1095
        check(pattern, "dir1\\dir2\\", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1096
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1097
        pattern = Pattern.compile("(\\Qdir1\\dir2\\\\E)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1098
        check(pattern, "dir1\\dir2\\", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1099
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1100
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1101
        pattern = Pattern.compile(toSupplementaries("\\Qdir1\\dir2\\E"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1102
        check(pattern, toSupplementaries("dir1\\dir2"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1103
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1104
        pattern = Pattern.compile(toSupplementaries("\\Qdir1\\dir2")+"\\\\E");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1105
        check(pattern, toSupplementaries("dir1\\dir2\\"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1106
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1107
        pattern = Pattern.compile(toSupplementaries("(\\Qdir1\\dir2")+"\\\\E)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1108
        check(pattern, toSupplementaries("dir1\\dir2\\"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1109
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1110
        report("Escaped segment");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1111
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1112
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1113
    // This test is for 4792284
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1114
    private static void nonCaptureRepetitionTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1115
        String input = "abcdefgh;";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1116
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1117
        String[] patterns = new String[] {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1118
            "(?:\\w{4})+;",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1119
            "(?:\\w{8})*;",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1120
            "(?:\\w{2}){2,4};",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1121
            "(?:\\w{4}){2,};",   // only matches the
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1122
            ".*?(?:\\w{5})+;",   //     specified minimum
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1123
            ".*?(?:\\w{9})*;",   //     number of reps - OK
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1124
            "(?:\\w{4})+?;",     // lazy repetition - OK
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1125
            "(?:\\w{4})++;",     // possessive repetition - OK
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1126
            "(?:\\w{2,}?)+;",    // non-deterministic - OK
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1127
            "(\\w{4})+;",        // capturing group - OK
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1128
        };
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1129
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1130
        for (int i = 0; i < patterns.length; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1131
            // Check find()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1132
            check(patterns[i], 0, input, input, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1133
            // Check matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1134
            Pattern p = Pattern.compile(patterns[i]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1135
            Matcher m = p.matcher(input);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1136
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1137
            if (m.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1138
                if (!m.group(0).equals(input))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1139
                    failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1140
            } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1141
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1142
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1143
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1144
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1145
        report("Non capturing repetition");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1146
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1147
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1148
    // This test is for 6358731
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1149
    private static void notCapturedGroupCurlyMatchTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1150
        Pattern pattern = Pattern.compile("(abc)+|(abcd)+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1151
        Matcher matcher = pattern.matcher("abcd");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1152
        if (!matcher.matches() ||
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1153
             matcher.group(1) != null ||
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1154
             !matcher.group(2).equals("abcd")) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1155
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1156
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1157
        report("Not captured GroupCurly");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1158
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1159
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1160
    // This test is for 4706545
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1161
    private static void javaCharClassTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1162
        for (int i=0; i<1000; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1163
            char c = (char)generator.nextInt();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1164
            check("{javaLowerCase}", c, Character.isLowerCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1165
            check("{javaUpperCase}", c, Character.isUpperCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1166
            check("{javaUpperCase}+", c, Character.isUpperCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1167
            check("{javaTitleCase}", c, Character.isTitleCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1168
            check("{javaDigit}", c, Character.isDigit(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1169
            check("{javaDefined}", c, Character.isDefined(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1170
            check("{javaLetter}", c, Character.isLetter(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1171
            check("{javaLetterOrDigit}", c, Character.isLetterOrDigit(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1172
            check("{javaJavaIdentifierStart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1173
                  Character.isJavaIdentifierStart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1174
            check("{javaJavaIdentifierPart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1175
                  Character.isJavaIdentifierPart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1176
            check("{javaUnicodeIdentifierStart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1177
                  Character.isUnicodeIdentifierStart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1178
            check("{javaUnicodeIdentifierPart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1179
                  Character.isUnicodeIdentifierPart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1180
            check("{javaIdentifierIgnorable}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1181
                  Character.isIdentifierIgnorable(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1182
            check("{javaSpaceChar}", c, Character.isSpaceChar(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1183
            check("{javaWhitespace}", c, Character.isWhitespace(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1184
            check("{javaISOControl}", c, Character.isISOControl(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1185
            check("{javaMirrored}", c, Character.isMirrored(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1186
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1187
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1188
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1189
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1190
        for (int i=0; i<1000; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1191
            int c = generator.nextInt(Character.MAX_CODE_POINT
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1192
                                      - Character.MIN_SUPPLEMENTARY_CODE_POINT)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1193
                        + Character.MIN_SUPPLEMENTARY_CODE_POINT;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1194
            check("{javaLowerCase}", c, Character.isLowerCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1195
            check("{javaUpperCase}", c, Character.isUpperCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1196
            check("{javaUpperCase}+", c, Character.isUpperCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1197
            check("{javaTitleCase}", c, Character.isTitleCase(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1198
            check("{javaDigit}", c, Character.isDigit(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1199
            check("{javaDefined}", c, Character.isDefined(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1200
            check("{javaLetter}", c, Character.isLetter(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1201
            check("{javaLetterOrDigit}", c, Character.isLetterOrDigit(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1202
            check("{javaJavaIdentifierStart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1203
                  Character.isJavaIdentifierStart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1204
            check("{javaJavaIdentifierPart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1205
                  Character.isJavaIdentifierPart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1206
            check("{javaUnicodeIdentifierStart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1207
                  Character.isUnicodeIdentifierStart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1208
            check("{javaUnicodeIdentifierPart}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1209
                  Character.isUnicodeIdentifierPart(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1210
            check("{javaIdentifierIgnorable}", c,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1211
                  Character.isIdentifierIgnorable(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1212
            check("{javaSpaceChar}", c, Character.isSpaceChar(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1213
            check("{javaWhitespace}", c, Character.isWhitespace(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1214
            check("{javaISOControl}", c, Character.isISOControl(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1215
            check("{javaMirrored}", c, Character.isMirrored(c));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1216
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1217
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1218
        report("Java character classes");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1219
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1220
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1221
    // This test is for 4523620
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1222
    /*
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1223
    private static void numOccurrencesTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1224
        Pattern pattern = Pattern.compile("aaa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1225
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1226
        if (pattern.numOccurrences("aaaaaa", false) != 2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1227
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1228
        if (pattern.numOccurrences("aaaaaa", true) != 4)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1229
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1230
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1231
        pattern = Pattern.compile("^");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1232
        if (pattern.numOccurrences("aaaaaa", false) != 1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1233
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1234
        if (pattern.numOccurrences("aaaaaa", true) != 1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1235
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1236
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1237
        report("Number of Occurrences");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1238
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1239
    */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1240
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1241
    // This test is for 4776374
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1242
    private static void caretBetweenTerminatorsTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1243
        int flags1 = Pattern.DOTALL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1244
        int flags2 = Pattern.DOTALL | Pattern.UNIX_LINES;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1245
        int flags3 = Pattern.DOTALL | Pattern.UNIX_LINES | Pattern.MULTILINE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1246
        int flags4 = Pattern.DOTALL | Pattern.MULTILINE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1247
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1248
        check("^....", flags1, "test\ntest", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1249
        check(".....^", flags1, "test\ntest", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1250
        check(".....^", flags1, "test\n", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1251
        check("....^", flags1, "test\r\n", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1252
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1253
        check("^....", flags2, "test\ntest", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1254
        check("....^", flags2, "test\ntest", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1255
        check(".....^", flags2, "test\n", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1256
        check("....^", flags2, "test\r\n", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1257
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1258
        check("^....", flags3, "test\ntest", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1259
        check(".....^", flags3, "test\ntest", "test\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1260
        check(".....^", flags3, "test\u0085test", "test\u0085", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1261
        check(".....^", flags3, "test\n", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1262
        check(".....^", flags3, "test\r\n", "test", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1263
        check("......^", flags3, "test\r\ntest", "test\r\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1264
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1265
        check("^....", flags4, "test\ntest", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1266
        check(".....^", flags3, "test\ntest", "test\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1267
        check(".....^", flags4, "test\u0085test", "test\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1268
        check(".....^", flags4, "test\n", "test\n", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1269
        check(".....^", flags4, "test\r\n", "test\r", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1270
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1271
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1272
        String t = toSupplementaries("test");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1273
        check("^....", flags1, t+"\n"+t, t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1274
        check(".....^", flags1, t+"\n"+t, t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1275
        check(".....^", flags1, t+"\n", t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1276
        check("....^", flags1, t+"\r\n", t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1277
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1278
        check("^....", flags2, t+"\n"+t, t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1279
        check("....^", flags2, t+"\n"+t, t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1280
        check(".....^", flags2, t+"\n", t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1281
        check("....^", flags2, t+"\r\n", t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1282
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1283
        check("^....", flags3, t+"\n"+t, t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1284
        check(".....^", flags3, t+"\n"+t, t+"\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1285
        check(".....^", flags3, t+"\u0085"+t, t+"\u0085", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1286
        check(".....^", flags3, t+"\n", t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1287
        check(".....^", flags3, t+"\r\n", t, false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1288
        check("......^", flags3, t+"\r\n"+t, t+"\r\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1289
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1290
        check("^....", flags4, t+"\n"+t, t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1291
        check(".....^", flags3, t+"\n"+t, t+"\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1292
        check(".....^", flags4, t+"\u0085"+t, t+"\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1293
        check(".....^", flags4, t+"\n", t+"\n", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1294
        check(".....^", flags4, t+"\r\n", t+"\r", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1295
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1296
        report("Caret between terminators");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1297
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1298
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1299
    // This test is for 4727935
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1300
    private static void dollarAtEndTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1301
        int flags1 = Pattern.DOTALL;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1302
        int flags2 = Pattern.DOTALL | Pattern.UNIX_LINES;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1303
        int flags3 = Pattern.DOTALL | Pattern.MULTILINE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1304
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1305
        check("....$", flags1, "test\n", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1306
        check("....$", flags1, "test\r\n", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1307
        check(".....$", flags1, "test\n", "test\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1308
        check(".....$", flags1, "test\u0085", "test\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1309
        check("....$", flags1, "test\u0085", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1310
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1311
        check("....$", flags2, "test\n", "test", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1312
        check(".....$", flags2, "test\n", "test\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1313
        check(".....$", flags2, "test\u0085", "test\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1314
        check("....$", flags2, "test\u0085", "est\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1315
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1316
        check("....$.blah", flags3, "test\nblah", "test\nblah", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1317
        check(".....$.blah", flags3, "test\n\nblah", "test\n\nblah", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1318
        check("....$blah", flags3, "test\nblah", "!!!!", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1319
        check(".....$blah", flags3, "test\nblah", "!!!!", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1320
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1321
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1322
        String t = toSupplementaries("test");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1323
        String b = toSupplementaries("blah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1324
        check("....$", flags1, t+"\n", t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1325
        check("....$", flags1, t+"\r\n", t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1326
        check(".....$", flags1, t+"\n", t+"\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1327
        check(".....$", flags1, t+"\u0085", t+"\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1328
        check("....$", flags1, t+"\u0085", t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1329
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1330
        check("....$", flags2, t+"\n", t, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1331
        check(".....$", flags2, t+"\n", t+"\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1332
        check(".....$", flags2, t+"\u0085", t+"\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1333
        check("....$", flags2, t+"\u0085", toSupplementaries("est\u0085"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1334
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1335
        check("....$."+b, flags3, t+"\n"+b, t+"\n"+b, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1336
        check(".....$."+b, flags3, t+"\n\n"+b, t+"\n\n"+b, true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1337
        check("....$"+b, flags3, t+"\n"+b, "!!!!", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1338
        check(".....$"+b, flags3, t+"\n"+b, "!!!!", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1339
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1340
        report("Dollar at End");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1341
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1342
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1343
    // This test is for 4711773
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1344
    private static void multilineDollarTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1345
        Pattern findCR = Pattern.compile("$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1346
        Matcher matcher = findCR.matcher("first bit\nsecond bit");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1347
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1348
        if (matcher.start(0) != 9)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1349
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1350
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1351
        if (matcher.start(0) != 20)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1352
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1353
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1354
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1355
        matcher = findCR.matcher(toSupplementaries("first  bit\n second  bit")); // double BMP chars
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1356
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1357
        if (matcher.start(0) != 9*2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1358
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1359
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1360
        if (matcher.start(0) != 20*2)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1361
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1362
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1363
        report("Multiline Dollar");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1364
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1365
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1366
    private static void reluctantRepetitionTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1367
        Pattern p = Pattern.compile("1(\\s\\S+?){1,3}?[\\s,]2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1368
        check(p, "1 word word word 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1369
        check(p, "1 wor wo w 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1370
        check(p, "1 word word 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1371
        check(p, "1 word 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1372
        check(p, "1 wo w w 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1373
        check(p, "1 wo w 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1374
        check(p, "1 wor w 2", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1375
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1376
        p = Pattern.compile("([a-z])+?c");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1377
        Matcher m = p.matcher("ababcdefdec");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1378
        check(m, "ababc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1379
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1380
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1381
        p = Pattern.compile(toSupplementaries("([a-z])+?c"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1382
        m = p.matcher(toSupplementaries("ababcdefdec"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1383
        check(m, toSupplementaries("ababc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1384
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1385
        report("Reluctant Repetition");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1386
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1387
48491
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1388
    private static Pattern serializedPattern(Pattern p) throws Exception {
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1389
        ByteArrayOutputStream baos = new ByteArrayOutputStream();
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1390
        ObjectOutputStream oos = new ObjectOutputStream(baos);
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1391
        oos.writeObject(p);
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1392
        oos.close();
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1393
        try (ObjectInputStream ois = new ObjectInputStream(
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1394
                new ByteArrayInputStream(baos.toByteArray()))) {
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1395
            return (Pattern)ois.readObject();
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1396
        }
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1397
    }
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1398
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1399
    private static void serializeTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1400
        String patternStr = "(b)";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1401
        String matchStr = "b";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1402
        Pattern pattern = Pattern.compile(patternStr);
48491
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1403
        Pattern serializedPattern = serializedPattern(pattern);
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1404
        Matcher matcher = serializedPattern.matcher(matchStr);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1405
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1406
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1407
        if (matcher.groupCount() != 1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1408
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1409
48491
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1410
        pattern = Pattern.compile("a(?-i)b", Pattern.CASE_INSENSITIVE);
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1411
        serializedPattern = serializedPattern(pattern);
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1412
        if (!serializedPattern.matcher("Ab").matches())
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1413
            failCount++;
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1414
        if (serializedPattern.matcher("AB").matches())
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1415
            failCount++;
7f11a1699ef6 8194667: Regex: Serialization doesn't work with match flags
sherman
parents: 47216
diff changeset
  1416
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1417
        report("Serialization");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1418
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1419
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1420
    private static void gTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1421
        Pattern pattern = Pattern.compile("\\G\\w");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1422
        Matcher matcher = pattern.matcher("abc#x#x");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1423
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1424
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1425
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1426
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1427
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1428
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1429
        pattern = Pattern.compile("\\GA*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1430
        matcher = pattern.matcher("1A2AA3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1431
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1432
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1433
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1434
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1435
        pattern = Pattern.compile("\\GA*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1436
        matcher = pattern.matcher("1A2AA3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1437
        if (!matcher.find(1))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1438
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1439
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1440
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1441
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1442
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1443
        report("\\G");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1444
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1445
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1446
    private static void zTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1447
        Pattern pattern = Pattern.compile("foo\\Z");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1448
        // Positives
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1449
        check(pattern, "foo\u0085", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1450
        check(pattern, "foo\u2028", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1451
        check(pattern, "foo\u2029", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1452
        check(pattern, "foo\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1453
        check(pattern, "foo\r", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1454
        check(pattern, "foo\r\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1455
        // Negatives
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1456
        check(pattern, "fooo", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1457
        check(pattern, "foo\n\r", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1458
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1459
        pattern = Pattern.compile("foo\\Z", Pattern.UNIX_LINES);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1460
        // Positives
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1461
        check(pattern, "foo", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1462
        check(pattern, "foo\n", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1463
        // Negatives
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1464
        check(pattern, "foo\r", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1465
        check(pattern, "foo\u0085", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1466
        check(pattern, "foo\u2028", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1467
        check(pattern, "foo\u2029", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1468
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1469
        report("\\Z");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1470
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1471
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1472
    private static void replaceFirstTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1473
        Pattern pattern = Pattern.compile("(ab)(c*)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1474
        Matcher matcher = pattern.matcher("abccczzzabcczzzabccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1475
        if (!matcher.replaceFirst("test").equals("testzzzabcczzzabccc"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1476
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1477
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1478
        matcher.reset("zzzabccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1479
        if (!matcher.replaceFirst("test").equals("zzztestzzzabcczzzabccczzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1480
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1481
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1482
        matcher.reset("zzzabccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1483
        String result = matcher.replaceFirst("$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1484
        if (!result.equals("zzzabzzzabcczzzabccczzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1485
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1486
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1487
        matcher.reset("zzzabccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1488
        result = matcher.replaceFirst("$2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1489
        if (!result.equals("zzzccczzzabcczzzabccczzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1490
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1491
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1492
        pattern = Pattern.compile("a*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1493
        matcher = pattern.matcher("aaaaaaaaaa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1494
        if (!matcher.replaceFirst("test").equals("test"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1495
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1496
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1497
        pattern = Pattern.compile("a+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1498
        matcher = pattern.matcher("zzzaaaaaaaaaa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1499
        if (!matcher.replaceFirst("test").equals("zzztest"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1500
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1501
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1502
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1503
        pattern = Pattern.compile(toSupplementaries("(ab)(c*)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1504
        matcher = pattern.matcher(toSupplementaries("abccczzzabcczzzabccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1505
        if (!matcher.replaceFirst(toSupplementaries("test"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1506
                .equals(toSupplementaries("testzzzabcczzzabccc")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1507
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1508
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1509
        matcher.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1510
        if (!matcher.replaceFirst(toSupplementaries("test")).
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1511
            equals(toSupplementaries("zzztestzzzabcczzzabccczzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1512
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1513
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1514
        matcher.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1515
        result = matcher.replaceFirst("$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1516
        if (!result.equals(toSupplementaries("zzzabzzzabcczzzabccczzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1517
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1518
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1519
        matcher.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1520
        result = matcher.replaceFirst("$2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1521
        if (!result.equals(toSupplementaries("zzzccczzzabcczzzabccczzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1522
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1523
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1524
        pattern = Pattern.compile(toSupplementaries("a*"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1525
        matcher = pattern.matcher(toSupplementaries("aaaaaaaaaa"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1526
        if (!matcher.replaceFirst(toSupplementaries("test")).equals(toSupplementaries("test")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1527
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1528
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1529
        pattern = Pattern.compile(toSupplementaries("a+"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1530
        matcher = pattern.matcher(toSupplementaries("zzzaaaaaaaaaa"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1531
        if (!matcher.replaceFirst(toSupplementaries("test")).equals(toSupplementaries("zzztest")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1532
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1533
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1534
        report("Replace First");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1535
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1536
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1537
    private static void unixLinesTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1538
        Pattern pattern = Pattern.compile(".*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1539
        Matcher matcher = pattern.matcher("aa\u2028blah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1540
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1541
        if (!matcher.group(0).equals("aa"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1542
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1543
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1544
        pattern = Pattern.compile(".*", Pattern.UNIX_LINES);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1545
        matcher = pattern.matcher("aa\u2028blah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1546
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1547
        if (!matcher.group(0).equals("aa\u2028blah"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1548
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1549
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1550
        pattern = Pattern.compile("[az]$",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1551
                                  Pattern.MULTILINE | Pattern.UNIX_LINES);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1552
        matcher = pattern.matcher("aa\u2028zz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1553
        check(matcher, "a\u2028", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1554
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1555
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1556
        pattern = Pattern.compile(".*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1557
        matcher = pattern.matcher(toSupplementaries("aa\u2028blah"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1558
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1559
        if (!matcher.group(0).equals(toSupplementaries("aa")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1560
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1561
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1562
        pattern = Pattern.compile(".*", Pattern.UNIX_LINES);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1563
        matcher = pattern.matcher(toSupplementaries("aa\u2028blah"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1564
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1565
        if (!matcher.group(0).equals(toSupplementaries("aa\u2028blah")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1566
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1567
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1568
        pattern = Pattern.compile(toSupplementaries("[az]$"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1569
                                  Pattern.MULTILINE | Pattern.UNIX_LINES);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1570
        matcher = pattern.matcher(toSupplementaries("aa\u2028zz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1571
        check(matcher, toSupplementaries("a\u2028"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1572
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1573
        report("Unix Lines");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1574
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1575
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1576
    private static void commentsTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1577
        int flags = Pattern.COMMENTS;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1578
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1579
        Pattern pattern = Pattern.compile("aa \\# aa", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1580
        Matcher matcher = pattern.matcher("aa#aa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1581
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1582
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1583
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1584
        pattern = Pattern.compile("aa  # blah", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1585
        matcher = pattern.matcher("aa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1586
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1587
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1588
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1589
        pattern = Pattern.compile("aa blah", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1590
        matcher = pattern.matcher("aablah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1591
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1592
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1593
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1594
        pattern = Pattern.compile("aa  # blah blech  ", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1595
        matcher = pattern.matcher("aa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1596
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1597
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1598
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1599
        pattern = Pattern.compile("aa  # blah\n  ", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1600
        matcher = pattern.matcher("aa");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1601
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1602
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1603
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1604
        pattern = Pattern.compile("aa  # blah\nbc # blech", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1605
        matcher = pattern.matcher("aabc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1606
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1607
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1608
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1609
        pattern = Pattern.compile("aa  # blah\nbc# blech", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1610
        matcher = pattern.matcher("aabc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1611
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1612
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1613
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1614
        pattern = Pattern.compile("aa  # blah\nbc\\# blech", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1615
        matcher = pattern.matcher("aabc#blech");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1616
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1617
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1618
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1619
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1620
        pattern = Pattern.compile(toSupplementaries("aa \\# aa"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1621
        matcher = pattern.matcher(toSupplementaries("aa#aa"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1622
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1623
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1624
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1625
        pattern = Pattern.compile(toSupplementaries("aa  # blah"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1626
        matcher = pattern.matcher(toSupplementaries("aa"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1627
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1628
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1629
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1630
        pattern = Pattern.compile(toSupplementaries("aa blah"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1631
        matcher = pattern.matcher(toSupplementaries("aablah"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1632
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1633
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1634
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1635
        pattern = Pattern.compile(toSupplementaries("aa  # blah blech  "), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1636
        matcher = pattern.matcher(toSupplementaries("aa"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1637
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1638
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1639
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1640
        pattern = Pattern.compile(toSupplementaries("aa  # blah\n  "), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1641
        matcher = pattern.matcher(toSupplementaries("aa"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1642
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1643
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1644
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1645
        pattern = Pattern.compile(toSupplementaries("aa  # blah\nbc # blech"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1646
        matcher = pattern.matcher(toSupplementaries("aabc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1647
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1648
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1649
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1650
        pattern = Pattern.compile(toSupplementaries("aa  # blah\nbc# blech"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1651
        matcher = pattern.matcher(toSupplementaries("aabc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1652
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1653
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1654
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1655
        pattern = Pattern.compile(toSupplementaries("aa  # blah\nbc\\# blech"), flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1656
        matcher = pattern.matcher(toSupplementaries("aabc#blech"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1657
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1658
             failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1659
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1660
        report("Comments");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1661
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1662
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1663
    private static void caseFoldingTest() { // bug 4504687
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1664
        int flags = Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1665
        Pattern pattern = Pattern.compile("aa", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1666
        Matcher matcher = pattern.matcher("ab");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1667
        if (matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1668
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1669
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1670
        pattern = Pattern.compile("aA", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1671
        matcher = pattern.matcher("ab");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1672
        if (matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1673
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1674
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1675
        pattern = Pattern.compile("aa", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1676
        matcher = pattern.matcher("aB");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1677
        if (matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1678
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1679
        matcher = pattern.matcher("Ab");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1680
        if (matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1681
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1682
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1683
        // ASCII               "a"
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1684
        // Latin-1 Supplement  "a" + grave
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1685
        // Cyrillic            "a"
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1686
        String[] patterns = new String[] {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1687
            //single
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1688
            "a", "\u00e0", "\u0430",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1689
            //slice
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1690
            "ab", "\u00e0\u00e1", "\u0430\u0431",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1691
            //class single
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1692
            "[a]", "[\u00e0]", "[\u0430]",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1693
            //class range
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1694
            "[a-b]", "[\u00e0-\u00e5]", "[\u0430-\u0431]",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1695
            //back reference
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1696
            "(a)\\1", "(\u00e0)\\1", "(\u0430)\\1"
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1697
        };
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1698
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1699
        String[] texts = new String[] {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1700
            "A", "\u00c0", "\u0410",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1701
            "AB", "\u00c0\u00c1", "\u0410\u0411",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1702
            "A", "\u00c0", "\u0410",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1703
            "B", "\u00c2", "\u0411",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1704
            "aA", "\u00e0\u00c0", "\u0430\u0410"
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1705
        };
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1706
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1707
        boolean[] expected = new boolean[] {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1708
            true, false, false,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1709
            true, false, false,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1710
            true, false, false,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1711
            true, false, false,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1712
            true, false, false
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1713
        };
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1714
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1715
        flags = Pattern.CASE_INSENSITIVE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1716
        for (int i = 0; i < patterns.length; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1717
            pattern = Pattern.compile(patterns[i], flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1718
            matcher = pattern.matcher(texts[i]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1719
            if (matcher.matches() != expected[i]) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1720
                System.out.println("<1> Failed at " + i);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1721
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1722
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1723
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1724
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1725
        flags = Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1726
        for (int i = 0; i < patterns.length; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1727
            pattern = Pattern.compile(patterns[i], flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1728
            matcher = pattern.matcher(texts[i]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1729
            if (!matcher.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1730
                System.out.println("<2> Failed at " + i);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1731
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1732
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1733
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1734
        // flag unicode_case alone should do nothing
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1735
        flags = Pattern.UNICODE_CASE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1736
        for (int i = 0; i < patterns.length; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1737
            pattern = Pattern.compile(patterns[i], flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1738
            matcher = pattern.matcher(texts[i]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1739
            if (matcher.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1740
                System.out.println("<3> Failed at " + i);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1741
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1742
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1743
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1744
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1745
        // Special cases: i, I, u+0131 and u+0130
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1746
        flags = Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1747
        pattern = Pattern.compile("[h-j]+", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1748
        if (!pattern.matcher("\u0131\u0130").matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1749
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1750
        report("Case Folding");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1751
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1752
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1753
    private static void appendTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1754
        Pattern pattern = Pattern.compile("(ab)(cd)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1755
        Matcher matcher = pattern.matcher("abcd");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1756
        String result = matcher.replaceAll("$2$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1757
        if (!result.equals("cdab"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1758
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1759
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1760
        String  s1 = "Swap all: first = 123, second = 456";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1761
        String  s2 = "Swap one: first = 123, second = 456";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1762
        String  r  = "$3$2$1";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1763
        pattern = Pattern.compile("([a-z]+)( *= *)([0-9]+)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1764
        matcher = pattern.matcher(s1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1765
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1766
        result = matcher.replaceAll(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1767
        if (!result.equals("Swap all: 123 = first, 456 = second"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1768
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1769
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1770
        matcher = pattern.matcher(s2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1771
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1772
        if (matcher.find()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1773
            StringBuffer sb = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1774
            matcher.appendReplacement(sb, r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1775
            matcher.appendTail(sb);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1776
            result = sb.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1777
            if (!result.equals("Swap one: 123 = first, second = 456"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1778
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1779
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1780
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1781
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1782
        pattern = Pattern.compile(toSupplementaries("(ab)(cd)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1783
        matcher = pattern.matcher(toSupplementaries("abcd"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1784
        result = matcher.replaceAll("$2$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1785
        if (!result.equals(toSupplementaries("cdab")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1786
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1787
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1788
        s1 = toSupplementaries("Swap all: first = 123, second = 456");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1789
        s2 = toSupplementaries("Swap one: first = 123, second = 456");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1790
        r  = toSupplementaries("$3$2$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1791
        pattern = Pattern.compile(toSupplementaries("([a-z]+)( *= *)([0-9]+)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1792
        matcher = pattern.matcher(s1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1793
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1794
        result = matcher.replaceAll(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1795
        if (!result.equals(toSupplementaries("Swap all: 123 = first, 456 = second")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1796
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1797
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1798
        matcher = pattern.matcher(s2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1799
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1800
        if (matcher.find()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1801
            StringBuffer sb = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1802
            matcher.appendReplacement(sb, r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1803
            matcher.appendTail(sb);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1804
            result = sb.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1805
            if (!result.equals(toSupplementaries("Swap one: 123 = first, second = 456")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1806
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1807
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1808
        report("Append");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1809
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1810
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1811
    private static void splitTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1812
        Pattern pattern = Pattern.compile(":");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1813
        String[] result = pattern.split("foo:and:boo", 2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1814
        if (!result[0].equals("foo"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1815
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1816
        if (!result[1].equals("and:boo"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1817
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1818
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1819
        Pattern patternX = Pattern.compile(toSupplementaries("X"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1820
        result = patternX.split(toSupplementaries("fooXandXboo"), 2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1821
        if (!result[0].equals(toSupplementaries("foo")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1822
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1823
        if (!result[1].equals(toSupplementaries("andXboo")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1824
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1825
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1826
        CharBuffer cb = CharBuffer.allocate(100);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1827
        cb.put("foo:and:boo");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1828
        cb.flip();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1829
        result = pattern.split(cb);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1830
        if (!result[0].equals("foo"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1831
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1832
        if (!result[1].equals("and"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1833
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1834
        if (!result[2].equals("boo"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1835
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1836
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1837
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1838
        CharBuffer cbs = CharBuffer.allocate(100);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1839
        cbs.put(toSupplementaries("fooXandXboo"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1840
        cbs.flip();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1841
        result = patternX.split(cbs);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1842
        if (!result[0].equals(toSupplementaries("foo")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1843
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1844
        if (!result[1].equals(toSupplementaries("and")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1845
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1846
        if (!result[2].equals(toSupplementaries("boo")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1847
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1848
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1849
        String source = "0123456789";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1850
        for (int limit=-2; limit<3; limit++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1851
            for (int x=0; x<10; x++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1852
                result = source.split(Integer.toString(x), limit);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1853
                int expectedLength = limit < 1 ? 2 : limit;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1854
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1855
                if ((limit == 0) && (x == 9)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1856
                    // expected dropping of ""
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1857
                    if (result.length != 1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1858
                        failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1859
                    if (!result[0].equals("012345678")) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1860
                        failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1861
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1862
                } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1863
                    if (result.length != expectedLength) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1864
                        failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1865
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1866
                    if (!result[0].equals(source.substring(0,x))) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1867
                        if (limit != 1) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1868
                            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1869
                        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1870
                            if (!result[0].equals(source.substring(0,10))) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1871
                                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1872
                            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1873
                        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1874
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1875
                    if (expectedLength > 1) { // Check segment 2
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1876
                        if (!result[1].equals(source.substring(x+1,10)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1877
                            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1878
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1879
                }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1880
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1881
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1882
        // Check the case for no match found
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1883
        for (int limit=-2; limit<3; limit++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1884
            result = source.split("e", limit);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1885
            if (result.length != 1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1886
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1887
            if (!result[0].equals(source))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1888
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1889
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1890
        // Check the case for limit == 0, source = "";
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1891
        // split() now returns 0-length for empty source "" see #6559590
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1892
        source = "";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1893
        result = source.split("e", 0);
21670
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1894
        if (result.length != 1)
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1895
            failCount++;
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1896
        if (!result[0].equals(source))
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1897
            failCount++;
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1898
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1899
        // Check both split() and splitAsStraem(), especially for zero-lenth
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1900
        // input and zero-lenth match cases
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1901
        String[][] input = new String[][] {
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1902
            { " ",           "Abc Efg Hij" },   // normal non-zero-match
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1903
            { " ",           " Abc Efg Hij" },  // leading empty str for non-zero-match
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1904
            { " ",           "Abc  Efg Hij" },  // non-zero-match in the middle
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1905
            { "(?=\\p{Lu})", "AbcEfgHij" },     // no leading empty str for zero-match
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1906
            { "(?=\\p{Lu})", "AbcEfg" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1907
            { "(?=\\p{Lu})", "Abc" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1908
            { " ",           "" },              // zero-length input
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1909
            { ".*",          "" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1910
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1911
            // some tests from PatternStreamTest.java
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1912
            { "4",       "awgqwefg1fefw4vssv1vvv1" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1913
            { "\u00a3a", "afbfq\u00a3abgwgb\u00a3awngnwggw\u00a3a\u00a3ahjrnhneerh" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1914
            { "1",       "awgqwefg1fefw4vssv1vvv1" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1915
            { "1",       "a\u4ebafg1fefw\u4eba4\u9f9cvssv\u9f9c1v\u672c\u672cvv" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1916
            { "\u56da",  "1\u56da23\u56da456\u56da7890" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1917
            { "\u56da",  "1\u56da23\u9f9c\u672c\u672c\u56da456\u56da\u9f9c\u672c7890" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1918
            { "\u56da",  "" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1919
            { "[ \t,:.]","This is,testing: with\tdifferent separators." }, //multiple septs
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1920
            { "o",       "boo:and:foo" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1921
            { "o",       "booooo:and:fooooo" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1922
            { "o",       "fooooo:" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1923
        };
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1924
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1925
        String[][] expected = new String[][] {
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1926
            { "Abc", "Efg", "Hij" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1927
            { "", "Abc", "Efg", "Hij" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1928
            { "Abc", "", "Efg", "Hij" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1929
            { "Abc", "Efg", "Hij" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1930
            { "Abc", "Efg" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1931
            { "Abc" },
21670
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1932
            { "" },
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1933
            { "" },
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1934
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1935
            { "awgqwefg1fefw", "vssv1vvv1" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1936
            { "afbfq", "bgwgb", "wngnwggw", "", "hjrnhneerh" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1937
            { "awgqwefg", "fefw4vssv", "vvv" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1938
            { "a\u4ebafg", "fefw\u4eba4\u9f9cvssv\u9f9c", "v\u672c\u672cvv" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1939
            { "1", "23", "456", "7890" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1940
            { "1", "23\u9f9c\u672c\u672c", "456", "\u9f9c\u672c7890" },
21670
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1941
            { "" },
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1942
            { "This", "is", "testing", "", "with", "different", "separators" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1943
            { "b", "", ":and:f" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1944
            { "b", "", "", "", "", ":and:f" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1945
            { "f", "", "", "", "", ":" },
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1946
        };
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1947
        for (int i = 0; i < input.length; i++) {
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1948
            pattern = Pattern.compile(input[i][0]);
21670
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1949
            if (!Arrays.equals(pattern.split(input[i][1]), expected[i])) {
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1950
                failCount++;
21670
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1951
            }
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1952
            if (input[i][1].length() > 0 &&  // splitAsStream() return empty resulting
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1953
                                             // array for zero-length input for now
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1954
                !Arrays.equals(pattern.splitAsStream(input[i][1]).toArray(),
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1955
                               expected[i])) {
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1956
                failCount++;
21670
ca3553133ede 8028321: Fix for String.split() empty input sequence/JDK-6559590 triggers regression
sherman
parents: 21668
diff changeset
  1957
            }
21668
b62ce4a9635f 8027645: Pattern.split() with positive lookahead
sherman
parents: 19604
diff changeset
  1958
        }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1959
        report("Split");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1960
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1961
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1962
    private static void negationTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1963
        Pattern pattern = Pattern.compile("[\\[@^]+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1964
        Matcher matcher = pattern.matcher("@@@@[[[[^^^^");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1965
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1966
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1967
        if (!matcher.group(0).equals("@@@@[[[[^^^^"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1968
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1969
        pattern = Pattern.compile("[@\\[^]+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1970
        matcher = pattern.matcher("@@@@[[[[^^^^");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1971
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1972
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1973
        if (!matcher.group(0).equals("@@@@[[[[^^^^"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1974
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1975
        pattern = Pattern.compile("[@\\[^@]+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1976
        matcher = pattern.matcher("@@@@[[[[^^^^");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1977
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1978
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1979
        if (!matcher.group(0).equals("@@@@[[[[^^^^"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1980
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1981
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1982
        pattern = Pattern.compile("\\)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1983
        matcher = pattern.matcher("xxx)xxx");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1984
        if (!matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1985
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1986
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1987
        report("Negation");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1988
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1989
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1990
    private static void ampersandTest() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1991
        Pattern pattern = Pattern.compile("[&@]+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1992
        check(pattern, "@@@@&&&&", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1993
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1994
        pattern = Pattern.compile("[@&]+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1995
        check(pattern, "@@@@&&&&", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1996
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1997
        pattern = Pattern.compile("[@\\&]+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1998
        check(pattern, "@@@@&&&&", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  1999
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2000
        report("Ampersand");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2001
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2002
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2003
    private static void octalTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2004
        Pattern pattern = Pattern.compile("\\u0007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2005
        Matcher matcher = pattern.matcher("\u0007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2006
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2007
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2008
        pattern = Pattern.compile("\\07");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2009
        matcher = pattern.matcher("\u0007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2010
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2011
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2012
        pattern = Pattern.compile("\\007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2013
        matcher = pattern.matcher("\u0007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2014
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2015
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2016
        pattern = Pattern.compile("\\0007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2017
        matcher = pattern.matcher("\u0007");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2018
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2019
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2020
        pattern = Pattern.compile("\\040");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2021
        matcher = pattern.matcher("\u0020");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2022
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2023
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2024
        pattern = Pattern.compile("\\0403");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2025
        matcher = pattern.matcher("\u00203");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2026
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2027
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2028
        pattern = Pattern.compile("\\0103");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2029
        matcher = pattern.matcher("\u0043");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2030
        if (!matcher.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2031
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2032
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2033
        report("Octal");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2034
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2035
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2036
    private static void longPatternTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2037
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2038
            Pattern pattern = Pattern.compile(
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2039
                "a 32-character-long pattern xxxx");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2040
            pattern = Pattern.compile("a 33-character-long pattern xxxxx");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2041
            pattern = Pattern.compile("a thirty four character long regex");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2042
            StringBuffer patternToBe = new StringBuffer(101);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2043
            for (int i=0; i<100; i++)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2044
                patternToBe.append((char)(97 + i%26));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2045
            pattern = Pattern.compile(patternToBe.toString());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2046
        } catch (PatternSyntaxException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2047
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2048
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2049
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2050
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2051
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2052
            Pattern pattern = Pattern.compile(
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2053
                toSupplementaries("a 32-character-long pattern xxxx"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2054
            pattern = Pattern.compile(toSupplementaries("a 33-character-long pattern xxxxx"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2055
            pattern = Pattern.compile(toSupplementaries("a thirty four character long regex"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2056
            StringBuffer patternToBe = new StringBuffer(101*2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2057
            for (int i=0; i<100; i++)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2058
                patternToBe.append(Character.toChars(Character.MIN_SUPPLEMENTARY_CODE_POINT
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2059
                                                     + 97 + i%26));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2060
            pattern = Pattern.compile(patternToBe.toString());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2061
        } catch (PatternSyntaxException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2062
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2063
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2064
        report("LongPattern");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2065
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2066
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2067
    private static void group0Test() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2068
        Pattern pattern = Pattern.compile("(tes)ting");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2069
        Matcher matcher = pattern.matcher("testing");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2070
        check(matcher, "testing");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2071
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2072
        matcher.reset("testing");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2073
        if (matcher.lookingAt()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2074
            if (!matcher.group(0).equals("testing"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2075
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2076
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2077
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2078
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2079
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2080
        matcher.reset("testing");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2081
        if (matcher.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2082
            if (!matcher.group(0).equals("testing"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2083
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2084
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2085
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2086
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2087
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2088
        pattern = Pattern.compile("(tes)ting");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2089
        matcher = pattern.matcher("testing");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2090
        if (matcher.lookingAt()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2091
            if (!matcher.group(0).equals("testing"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2092
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2093
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2094
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2095
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2096
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2097
        pattern = Pattern.compile("^(tes)ting");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2098
        matcher = pattern.matcher("testing");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2099
        if (matcher.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2100
            if (!matcher.group(0).equals("testing"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2101
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2102
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2103
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2104
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2105
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2106
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2107
        pattern = Pattern.compile(toSupplementaries("(tes)ting"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2108
        matcher = pattern.matcher(toSupplementaries("testing"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2109
        check(matcher, toSupplementaries("testing"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2110
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2111
        matcher.reset(toSupplementaries("testing"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2112
        if (matcher.lookingAt()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2113
            if (!matcher.group(0).equals(toSupplementaries("testing")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2114
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2115
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2116
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2117
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2118
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2119
        matcher.reset(toSupplementaries("testing"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2120
        if (matcher.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2121
            if (!matcher.group(0).equals(toSupplementaries("testing")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2122
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2123
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2124
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2125
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2126
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2127
        pattern = Pattern.compile(toSupplementaries("(tes)ting"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2128
        matcher = pattern.matcher(toSupplementaries("testing"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2129
        if (matcher.lookingAt()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2130
            if (!matcher.group(0).equals(toSupplementaries("testing")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2131
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2132
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2133
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2134
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2135
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2136
        pattern = Pattern.compile(toSupplementaries("^(tes)ting"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2137
        matcher = pattern.matcher(toSupplementaries("testing"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2138
        if (matcher.matches()) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2139
            if (!matcher.group(0).equals(toSupplementaries("testing")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2140
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2141
        } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2142
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2143
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2144
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2145
        report("Group0");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2146
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2147
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2148
    private static void findIntTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2149
        Pattern p = Pattern.compile("blah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2150
        Matcher m = p.matcher("zzzzblahzzzzzblah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2151
        boolean result = m.find(2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2152
        if (!result)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2153
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2154
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2155
        p = Pattern.compile("$");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2156
        m = p.matcher("1234567890");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2157
        result = m.find(10);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2158
        if (!result)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2159
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2160
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2161
            result = m.find(11);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2162
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2163
        } catch (IndexOutOfBoundsException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2164
            // correct result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2165
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2166
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2167
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2168
        p = Pattern.compile(toSupplementaries("blah"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2169
        m = p.matcher(toSupplementaries("zzzzblahzzzzzblah"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2170
        result = m.find(2);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2171
        if (!result)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2172
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2173
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2174
        report("FindInt");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2175
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2176
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2177
    private static void emptyPatternTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2178
        Pattern p = Pattern.compile("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2179
        Matcher m = p.matcher("foo");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2180
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2181
        // Should find empty pattern at beginning of input
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2182
        boolean result = m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2183
        if (result != true)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2184
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2185
        if (m.start() != 0)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2186
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2187
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2188
        // Should not match entire input if input is not empty
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2189
        m.reset();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2190
        result = m.matches();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2191
        if (result == true)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2192
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2193
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2194
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2195
            m.start(0);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2196
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2197
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2198
            // Correct result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2199
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2200
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2201
        // Should match entire input if input is empty
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2202
        m.reset("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2203
        result = m.matches();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2204
        if (result != true)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2205
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2206
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2207
        result = Pattern.matches("", "");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2208
        if (result != true)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2209
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2210
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2211
        result = Pattern.matches("", "foo");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2212
        if (result == true)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2213
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2214
        report("EmptyPattern");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2215
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2216
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2217
    private static void charClassTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2218
        Pattern pattern = Pattern.compile("blah[ab]]blech");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2219
        check(pattern, "blahb]blech", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2220
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2221
        pattern = Pattern.compile("[abc[def]]");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2222
        check(pattern, "b", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2223
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2224
        // Supplementary character tests
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2225
        pattern = Pattern.compile(toSupplementaries("blah[ab]]blech"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2226
        check(pattern, toSupplementaries("blahb]blech"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2227
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2228
        pattern = Pattern.compile(toSupplementaries("[abc[def]]"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2229
        check(pattern, toSupplementaries("b"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2230
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2231
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2232
            // u00ff when UNICODE_CASE
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2233
            pattern = Pattern.compile("[ab\u00ffcd]",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2234
                                      Pattern.CASE_INSENSITIVE|
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2235
                                      Pattern.UNICODE_CASE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2236
            check(pattern, "ab\u00ffcd", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2237
            check(pattern, "Ab\u0178Cd", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2238
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2239
            // u00b5 when UNICODE_CASE
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2240
            pattern = Pattern.compile("[ab\u00b5cd]",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2241
                                      Pattern.CASE_INSENSITIVE|
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2242
                                      Pattern.UNICODE_CASE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2243
            check(pattern, "ab\u00b5cd", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2244
            check(pattern, "Ab\u039cCd", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2245
        } catch (Exception e) { failCount++; }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2246
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2247
        /* Special cases
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2248
           (1)LatinSmallLetterLongS u+017f
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2249
           (2)LatinSmallLetterDotlessI u+0131
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2250
           (3)LatineCapitalLetterIWithDotAbove u+0130
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2251
           (4)KelvinSign u+212a
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2252
           (5)AngstromSign u+212b
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2253
        */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2254
        int flags = Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2255
        pattern = Pattern.compile("[sik\u00c5]+", flags);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2256
        if (!pattern.matcher("\u017f\u0130\u0131\u212a\u212b").matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2257
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2258
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2259
        report("CharClass");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2260
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2261
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2262
    private static void caretTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2263
        Pattern pattern = Pattern.compile("\\w*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2264
        Matcher matcher = pattern.matcher("a#bc#def##g");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2265
        check(matcher, "a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2266
        check(matcher, "");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2267
        check(matcher, "bc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2268
        check(matcher, "");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2269
        check(matcher, "def");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2270
        check(matcher, "");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2271
        check(matcher, "");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2272
        check(matcher, "g");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2273
        check(matcher, "");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2274
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2275
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2276
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2277
        pattern = Pattern.compile("^\\w*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2278
        matcher = pattern.matcher("a#bc#def##g");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2279
        check(matcher, "a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2280
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2281
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2282
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2283
        pattern = Pattern.compile("\\w");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2284
        matcher = pattern.matcher("abc##x");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2285
        check(matcher, "a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2286
        check(matcher, "b");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2287
        check(matcher, "c");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2288
        check(matcher, "x");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2289
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2290
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2291
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2292
        pattern = Pattern.compile("^\\w");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2293
        matcher = pattern.matcher("abc##x");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2294
        check(matcher, "a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2295
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2296
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2297
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2298
        pattern = Pattern.compile("\\A\\p{Alpha}{3}");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2299
        matcher = pattern.matcher("abcdef-ghi\njklmno");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2300
        check(matcher, "abc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2301
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2302
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2303
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2304
        pattern = Pattern.compile("^\\p{Alpha}{3}", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2305
        matcher = pattern.matcher("abcdef-ghi\njklmno");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2306
        check(matcher, "abc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2307
        check(matcher, "jkl");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2308
        if (matcher.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2309
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2310
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2311
        pattern = Pattern.compile("^", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2312
        matcher = pattern.matcher("this is some text");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2313
        String result = matcher.replaceAll("X");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2314
        if (!result.equals("Xthis is some text"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2315
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2316
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2317
        pattern = Pattern.compile("^");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2318
        matcher = pattern.matcher("this is some text");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2319
        result = matcher.replaceAll("X");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2320
        if (!result.equals("Xthis is some text"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2321
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2322
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2323
        pattern = Pattern.compile("^", Pattern.MULTILINE | Pattern.UNIX_LINES);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2324
        matcher = pattern.matcher("this is some text\n");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2325
        result = matcher.replaceAll("X");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2326
        if (!result.equals("Xthis is some text\n"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2327
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2328
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2329
        report("Caret");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2330
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2331
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2332
    private static void groupCaptureTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2333
        // Independent group
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2334
        Pattern pattern = Pattern.compile("x+(?>y+)z+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2335
        Matcher matcher = pattern.matcher("xxxyyyzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2336
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2337
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2338
            String blah = matcher.group(1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2339
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2340
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2341
            // Good result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2342
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2343
        // Pure group
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2344
        pattern = Pattern.compile("x+(?:y+)z+");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2345
        matcher = pattern.matcher("xxxyyyzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2346
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2347
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2348
            String blah = matcher.group(1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2349
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2350
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2351
            // Good result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2352
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2353
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2354
        // Supplementary character tests
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2355
        // Independent group
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2356
        pattern = Pattern.compile(toSupplementaries("x+(?>y+)z+"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2357
        matcher = pattern.matcher(toSupplementaries("xxxyyyzzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2358
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2359
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2360
            String blah = matcher.group(1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2361
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2362
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2363
            // Good result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2364
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2365
        // Pure group
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2366
        pattern = Pattern.compile(toSupplementaries("x+(?:y+)z+"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2367
        matcher = pattern.matcher(toSupplementaries("xxxyyyzzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2368
        matcher.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2369
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2370
            String blah = matcher.group(1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2371
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2372
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2373
            // Good result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2374
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2375
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2376
        report("GroupCapture");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2377
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2378
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2379
    private static void backRefTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2380
        Pattern pattern = Pattern.compile("(a*)bc\\1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2381
        check(pattern, "zzzaabcazzz", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2382
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2383
        pattern = Pattern.compile("(a*)bc\\1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2384
        check(pattern, "zzzaabcaazzz", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2385
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2386
        pattern = Pattern.compile("(abc)(def)\\1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2387
        check(pattern, "abcdefabc", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2388
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2389
        pattern = Pattern.compile("(abc)(def)\\3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2390
        check(pattern, "abcdefabc", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2391
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2392
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2393
            for (int i = 1; i < 10; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2394
                // Make sure backref 1-9 are always accepted
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2395
                pattern = Pattern.compile("abcdef\\" + i);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2396
                // and fail to match if the target group does not exit
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2397
                check(pattern, "abcdef", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2398
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2399
        } catch(PatternSyntaxException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2400
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2401
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2402
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2403
        pattern = Pattern.compile("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)\\11");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2404
        check(pattern, "abcdefghija", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2405
        check(pattern, "abcdefghija1", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2406
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2407
        pattern = Pattern.compile("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)(k)\\11");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2408
        check(pattern, "abcdefghijkk", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2409
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2410
        pattern = Pattern.compile("(a)bcdefghij\\11");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2411
        check(pattern, "abcdefghija1", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2412
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2413
        // Supplementary character tests
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2414
        pattern = Pattern.compile(toSupplementaries("(a*)bc\\1"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2415
        check(pattern, toSupplementaries("zzzaabcazzz"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2416
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2417
        pattern = Pattern.compile(toSupplementaries("(a*)bc\\1"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2418
        check(pattern, toSupplementaries("zzzaabcaazzz"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2419
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2420
        pattern = Pattern.compile(toSupplementaries("(abc)(def)\\1"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2421
        check(pattern, toSupplementaries("abcdefabc"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2422
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2423
        pattern = Pattern.compile(toSupplementaries("(abc)(def)\\3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2424
        check(pattern, toSupplementaries("abcdefabc"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2425
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2426
        pattern = Pattern.compile(toSupplementaries("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)\\11"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2427
        check(pattern, toSupplementaries("abcdefghija"), false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2428
        check(pattern, toSupplementaries("abcdefghija1"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2429
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2430
        pattern = Pattern.compile(toSupplementaries("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)(k)\\11"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2431
        check(pattern, toSupplementaries("abcdefghijkk"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2432
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2433
        report("BackRef");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2434
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2435
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2436
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2437
     * Unicode Technical Report #18, section 2.6 End of Line
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2438
     * There is no empty line to be matched in the sequence \u000D\u000A
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2439
     * but there is an empty line in the sequence \u000A\u000D.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2440
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2441
    private static void anchorTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2442
        Pattern p = Pattern.compile("^.*$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2443
        Matcher m = p.matcher("blah1\r\nblah2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2444
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2445
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2446
        if (!m.group().equals("blah2"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2447
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2448
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2449
        m.reset("blah1\n\rblah2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2450
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2451
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2452
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2453
        if (!m.group().equals("blah2"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2454
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2455
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2456
        // Test behavior of $ with \r\n at end of input
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2457
        p = Pattern.compile(".+$");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2458
        m = p.matcher("blah1\r\n");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2459
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2460
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2461
       if (!m.group().equals("blah1"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2462
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2463
        if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2464
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2465
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2466
        // Test behavior of $ with \r\n at end of input in multiline
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2467
        p = Pattern.compile(".+$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2468
        m = p.matcher("blah1\r\n");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2469
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2470
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2471
        if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2472
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2473
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2474
        // Test for $ recognition of \u0085 for bug 4527731
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2475
        p = Pattern.compile(".+$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2476
        m = p.matcher("blah1\u0085");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2477
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2478
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2479
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2480
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2481
        p = Pattern.compile("^.*$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2482
        m = p.matcher(toSupplementaries("blah1\r\nblah2"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2483
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2484
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2485
        if (!m.group().equals(toSupplementaries("blah2")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2486
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2487
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2488
        m.reset(toSupplementaries("blah1\n\rblah2"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2489
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2490
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2491
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2492
        if (!m.group().equals(toSupplementaries("blah2")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2493
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2494
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2495
        // Test behavior of $ with \r\n at end of input
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2496
        p = Pattern.compile(".+$");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2497
        m = p.matcher(toSupplementaries("blah1\r\n"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2498
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2499
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2500
        if (!m.group().equals(toSupplementaries("blah1")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2501
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2502
        if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2503
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2504
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2505
        // Test behavior of $ with \r\n at end of input in multiline
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2506
        p = Pattern.compile(".+$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2507
        m = p.matcher(toSupplementaries("blah1\r\n"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2508
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2509
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2510
        if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2511
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2512
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2513
        // Test for $ recognition of \u0085 for bug 4527731
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2514
        p = Pattern.compile(".+$", Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2515
        m = p.matcher(toSupplementaries("blah1\u0085"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2516
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2517
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2518
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2519
        report("Anchors");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2520
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2521
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2522
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2523
     * A basic sanity test of Matcher.lookingAt().
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2524
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2525
    private static void lookingAtTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2526
        Pattern p = Pattern.compile("(ab)(c*)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2527
        Matcher m = p.matcher("abccczzzabcczzzabccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2528
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2529
        if (!m.lookingAt())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2530
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2531
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2532
        if (!m.group().equals(m.group(0)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2533
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2534
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2535
        m = p.matcher("zzzabccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2536
        if (m.lookingAt())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2537
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2538
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2539
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2540
        p = Pattern.compile(toSupplementaries("(ab)(c*)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2541
        m = p.matcher(toSupplementaries("abccczzzabcczzzabccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2542
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2543
        if (!m.lookingAt())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2544
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2545
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2546
        if (!m.group().equals(m.group(0)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2547
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2548
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2549
        m = p.matcher(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2550
        if (m.lookingAt())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2551
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2552
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2553
        report("Looking At");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2554
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2555
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2556
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2557
     * A basic sanity test of Matcher.matches().
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2558
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2559
    private static void matchesTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2560
        // matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2561
        Pattern p = Pattern.compile("ulb(c*)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2562
        Matcher m = p.matcher("ulbcccccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2563
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2564
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2565
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2566
        // find() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2567
        m.reset("zzzulbcccccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2568
        if (m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2569
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2570
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2571
        // lookingAt() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2572
        m.reset("ulbccccccdef");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2573
        if (m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2574
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2575
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2576
        // matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2577
        p = Pattern.compile("a|ad");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2578
        m = p.matcher("ad");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2579
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2580
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2581
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2582
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2583
        // matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2584
        p = Pattern.compile(toSupplementaries("ulb(c*)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2585
        m = p.matcher(toSupplementaries("ulbcccccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2586
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2587
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2588
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2589
        // find() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2590
        m.reset(toSupplementaries("zzzulbcccccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2591
        if (m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2592
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2593
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2594
        // lookingAt() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2595
        m.reset(toSupplementaries("ulbccccccdef"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2596
        if (m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2597
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2598
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2599
        // matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2600
        p = Pattern.compile(toSupplementaries("a|ad"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2601
        m = p.matcher(toSupplementaries("ad"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2602
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2603
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2604
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2605
        report("Matches");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2606
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2607
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2608
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2609
     * A basic sanity test of Pattern.matches().
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2610
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2611
    private static void patternMatchesTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2612
        // matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2613
        if (!Pattern.matches(toSupplementaries("ulb(c*)"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2614
                             toSupplementaries("ulbcccccc")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2615
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2616
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2617
        // find() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2618
        if (Pattern.matches(toSupplementaries("ulb(c*)"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2619
                            toSupplementaries("zzzulbcccccc")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2620
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2621
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2622
        // lookingAt() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2623
        if (Pattern.matches(toSupplementaries("ulb(c*)"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2624
                            toSupplementaries("ulbccccccdef")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2625
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2626
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2627
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2628
        // matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2629
        if (!Pattern.matches(toSupplementaries("ulb(c*)"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2630
                             toSupplementaries("ulbcccccc")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2631
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2632
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2633
        // find() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2634
        if (Pattern.matches(toSupplementaries("ulb(c*)"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2635
                            toSupplementaries("zzzulbcccccc")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2636
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2637
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2638
        // lookingAt() but not matches()
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2639
        if (Pattern.matches(toSupplementaries("ulb(c*)"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2640
                            toSupplementaries("ulbccccccdef")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2641
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2642
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2643
        report("Pattern Matches");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2644
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2645
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2646
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2647
     * Canonical equivalence testing. Tests the ability of the engine
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2648
     * to match sequences that are not explicitly specified in the
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2649
     * pattern when they are considered equivalent by the Unicode Standard.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2650
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2651
    private static void ceTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2652
        // Decomposed char outside char classes
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2653
        Pattern p = Pattern.compile("testa\u030a", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2654
        Matcher m = p.matcher("test\u00e5");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2655
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2656
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2657
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2658
        m.reset("testa\u030a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2659
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2660
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2661
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2662
        // Composed char outside char classes
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2663
        p = Pattern.compile("test\u00e5", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2664
        m = p.matcher("test\u00e5");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2665
        if (!m.matches())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2666
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2667
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2668
        m.reset("testa\u030a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2669
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2670
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2671
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2672
        // Decomposed char inside a char class
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2673
        p = Pattern.compile("test[abca\u030a]", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2674
        m = p.matcher("test\u00e5");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2675
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2676
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2677
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2678
        m.reset("testa\u030a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2679
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2680
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2681
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2682
        // Composed char inside a char class
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2683
        p = Pattern.compile("test[abc\u00e5def\u00e0]", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2684
        m = p.matcher("test\u00e5");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2685
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2686
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2687
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2688
        m.reset("testa\u0300");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2689
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2690
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2691
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2692
        m.reset("testa\u030a");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2693
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2694
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2695
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2696
        // Marks that cannot legally change order and be equivalent
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2697
        p = Pattern.compile("testa\u0308\u0300", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2698
        check(p, "testa\u0308\u0300", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2699
        check(p, "testa\u0300\u0308", false);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2700
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2701
        // Marks that can legally change order and be equivalent
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2702
        p = Pattern.compile("testa\u0308\u0323", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2703
        check(p, "testa\u0308\u0323", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2704
        check(p, "testa\u0323\u0308", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2705
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2706
        // Test all equivalences of the sequence a\u0308\u0323\u0300
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2707
        p = Pattern.compile("testa\u0308\u0323\u0300", Pattern.CANON_EQ);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2708
        check(p, "testa\u0308\u0323\u0300", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2709
        check(p, "testa\u0323\u0308\u0300", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2710
        check(p, "testa\u0308\u0300\u0323", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2711
        check(p, "test\u00e4\u0323\u0300", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2712
        check(p, "test\u00e4\u0300\u0323", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2713
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2714
        Object[][] data = new Object[][] {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2715
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2716
        // JDK-4867170
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2717
        { "[\u1f80-\u1f82]", "ab\u1f80cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2718
        { "[\u1f80-\u1f82]", "ab\u1f81cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2719
        { "[\u1f80-\u1f82]", "ab\u1f82cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2720
        { "[\u1f80-\u1f82]", "ab\u03b1\u0314\u0345cd", "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2721
        { "[\u1f80-\u1f82]", "ab\u03b1\u0345\u0314cd", "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2722
        { "[\u1f80-\u1f82]", "ab\u1f01\u0345cd",       "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2723
        { "[\u1f80-\u1f82]", "ab\u1f00\u0345cd",       "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2724
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2725
        { "\\p{IsGreek}",    "ab\u1f80cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2726
        { "\\p{IsGreek}",    "ab\u1f81cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2727
        { "\\p{IsGreek}",    "ab\u1f82cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2728
        { "\\p{IsGreek}",    "ab\u03b1\u0314\u0345cd", "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2729
        { "\\p{IsGreek}",    "ab\u1f01\u0345cd",       "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2730
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2731
        // backtracking, force to match "\u1f80", instead of \u1f82"
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2732
        { "ab\\p{IsGreek}\u0300cd", "ab\u03b1\u0313\u0345\u0300cd", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2733
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2734
        { "[\\p{IsGreek}]",  "\u03b1\u0314\u0345",     "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2735
        { "\\p{IsGreek}",    "\u03b1\u0314\u0345",     "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2736
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2737
        { "[^\u1f80-\u1f82]","\u1f81",                 "m", false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2738
        { "[^\u1f80-\u1f82]","\u03b1\u0314\u0345",     "m", false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2739
        { "[^\u1f01\u0345]", "\u1f81",                 "f", false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2740
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2741
        { "[^\u1f81]+",      "\u1f80\u1f82",           "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2742
        { "[\u1f80]",        "ab\u1f80cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2743
        { "\u1f80",          "ab\u1f80cd",             "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2744
        { "\u1f00\u0345\u0300",  "\u1f82", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2745
        { "\u1f80",          "-\u1f00\u0345\u0300-",   "f", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2746
        { "\u1f82",          "\u1f00\u0345\u0300",     "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2747
        { "\u1f82",          "\u1f80\u0300",           "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2748
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2749
        // JDK-7080302       # compile failed
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2750
        { "a(\u0041\u0301\u0328)", "a\u0041\u0301\u0328", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2751
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2752
        // JDK-6728861, same cause as above one
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2753
        { "\u00e9\u00e9n", "e\u0301e\u0301n", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2754
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2755
        // JDK-6995635
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2756
        { "(\u00e9)", "e\u0301", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2757
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2758
        // JDK-6736245
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2759
        // intereting special case, nfc(u2add+u0338) -> u2add+u0338) NOT u2adc
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2760
        { "\u2ADC", "\u2ADC", "m", true},          // NFC
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2761
        { "\u2ADC", "\u2ADD\u0338", "m", true},    // NFD
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2762
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2763
        //  4916384.
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2764
        // Decomposed hangul (jamos) works inside clazz
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2765
        { "[\u1100\u1161]", "\u1100\u1161", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2766
        { "[\u1100\u1161]", "\uac00", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2767
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2768
        { "[\uac00]", "\u1100\u1161", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2769
        { "[\uac00]", "\uac00", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2770
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2771
        // Decomposed hangul (jamos)
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2772
        { "\u1100\u1161", "\u1100\u1161", "m", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2773
        { "\u1100\u1161", "\uac00", "m", true},
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2774
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2775
        // Composed hangul
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2776
        { "\uac00",  "\u1100\u1161", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2777
        { "\uac00",  "\uac00", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2778
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2779
        /* Need a NFDSlice to nfd the source to solve this issue
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2780
           u+1d1c0 -> nfd: <u+1d1ba><u+1d165><u+1d16f>  -> nfc: <u+1d1ba><u+1d165><u+1d16f>
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2781
           u+1d1bc -> nfd: <u+1d1ba><u+1d165>           -> nfc: <u+1d1ba><u+1d165>
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2782
           <u+1d1bc><u+1d16f> -> nfd: <u+1d1ba><u+1d165><u+1d16f> -> nfc: <u+1d1ba><u+1d165><u+1d16f>
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2783
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2784
        // Decomposed supplementary outside char classes
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2785
        // { "test\ud834\uddbc\ud834\udd6f", "test\ud834\uddc0", "m", true },
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2786
        // Composed supplementary outside char classes
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2787
        // { "test\ud834\uddc0", "test\ud834\uddbc\ud834\udd6f", "m", true },
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2788
        */
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2789
        { "test\ud834\uddbc\ud834\udd6f", "test\ud834\uddbc\ud834\udd6f", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2790
        { "test\ud834\uddc0",             "test\ud834\uddbc\ud834\udd6f", "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2791
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2792
        { "test\ud834\uddc0",             "test\ud834\uddc0",             "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2793
        { "test\ud834\uddbc\ud834\udd6f", "test\ud834\uddc0",             "m", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2794
        };
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2795
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2796
        int failCount = 0;
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2797
        for (Object[] d : data) {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2798
            String pn = (String)d[0];
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2799
            String tt = (String)d[1];
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2800
            boolean isFind = "f".equals(((String)d[2]));
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2801
            boolean expected = (boolean)d[3];
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2802
            boolean ret = isFind ? Pattern.compile(pn, Pattern.CANON_EQ).matcher(tt).find()
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2803
                                 : Pattern.compile(pn, Pattern.CANON_EQ).matcher(tt).matches();
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2804
            if (ret != expected) {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2805
                failCount++;
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2806
                continue;
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2807
            }
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  2808
        }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2809
        report("Canonical Equivalence");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2810
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2811
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2812
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2813
     * A basic sanity test of Matcher.replaceAll().
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2814
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2815
    private static void globalSubstitute() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2816
        // Global substitution with a literal
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2817
        Pattern p = Pattern.compile("(ab)(c*)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2818
        Matcher m = p.matcher("abccczzzabcczzzabccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2819
        if (!m.replaceAll("test").equals("testzzztestzzztest"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2820
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2821
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2822
        m.reset("zzzabccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2823
        if (!m.replaceAll("test").equals("zzztestzzztestzzztestzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2824
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2825
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2826
        // Global substitution with groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2827
        m.reset("zzzabccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2828
        String result = m.replaceAll("$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2829
        if (!result.equals("zzzabzzzabzzzabzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2830
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2831
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2832
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2833
        // Global substitution with a literal
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2834
        p = Pattern.compile(toSupplementaries("(ab)(c*)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2835
        m = p.matcher(toSupplementaries("abccczzzabcczzzabccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2836
        if (!m.replaceAll(toSupplementaries("test")).
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2837
            equals(toSupplementaries("testzzztestzzztest")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2838
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2839
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2840
        m.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2841
        if (!m.replaceAll(toSupplementaries("test")).
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2842
            equals(toSupplementaries("zzztestzzztestzzztestzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2843
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2844
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2845
        // Global substitution with groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2846
        m.reset(toSupplementaries("zzzabccczzzabcczzzabccczzz"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2847
        result = m.replaceAll("$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2848
        if (!result.equals(toSupplementaries("zzzabzzzabzzzabzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2849
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2850
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2851
        report("Global Substitution");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2852
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2853
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2854
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2855
     * Tests the usage of Matcher.appendReplacement() with literal
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2856
     * and group substitutions.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2857
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2858
    private static void stringbufferSubstitute() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2859
        // SB substitution with literal
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2860
        String blah = "zzzblahzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2861
        Pattern p = Pattern.compile("blah");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2862
        Matcher m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2863
        StringBuffer result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2864
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2865
            m.appendReplacement(result, "blech");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2866
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2867
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2868
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2869
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2870
        m.appendReplacement(result, "blech");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2871
        if (!result.toString().equals("zzzblech"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2872
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2873
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2874
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2875
        if (!result.toString().equals("zzzblechzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2876
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2877
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2878
        // SB substitution with groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2879
        blah = "zzzabcdzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2880
        p = Pattern.compile("(ab)(cd)*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2881
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2882
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2883
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2884
            m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2885
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2886
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2887
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2888
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2889
        m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2890
        if (!result.toString().equals("zzzab"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2891
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2892
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2893
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2894
        if (!result.toString().equals("zzzabzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2895
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2896
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2897
        // SB substitution with 3 groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2898
        blah = "zzzabcdcdefzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2899
        p = Pattern.compile("(ab)(cd)*(ef)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2900
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2901
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2902
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2903
            m.appendReplacement(result, "$1w$2w$3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2904
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2905
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2906
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2907
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2908
        m.appendReplacement(result, "$1w$2w$3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2909
        if (!result.toString().equals("zzzabwcdwef"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2910
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2911
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2912
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2913
        if (!result.toString().equals("zzzabwcdwefzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2914
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2915
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2916
        // SB substitution with groups and three matches
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2917
        // skipping middle match
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2918
        blah = "zzzabcdzzzabcddzzzabcdzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2919
        p = Pattern.compile("(ab)(cd*)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2920
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2921
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2922
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2923
            m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2924
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2925
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2926
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2927
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2928
        m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2929
        if (!result.toString().equals("zzzab"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2930
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2931
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2932
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2933
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2934
        m.appendReplacement(result, "$2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2935
        if (!result.toString().equals("zzzabzzzabcddzzzcd"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2936
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2937
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2938
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2939
        if (!result.toString().equals("zzzabzzzabcddzzzcdzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2940
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2941
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2942
        // Check to make sure escaped $ is ignored
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2943
        blah = "zzzabcdcdefzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2944
        p = Pattern.compile("(ab)(cd)*(ef)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2945
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2946
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2947
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2948
        m.appendReplacement(result, "$1w\\$2w$3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2949
        if (!result.toString().equals("zzzabw$2wef"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2950
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2951
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2952
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2953
        if (!result.toString().equals("zzzabw$2wefzzz"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2954
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2955
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2956
        // Check to make sure a reference to nonexistent group causes error
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2957
        blah = "zzzabcdcdefzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2958
        p = Pattern.compile("(ab)(cd)*(ef)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2959
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2960
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2961
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2962
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2963
            m.appendReplacement(result, "$1w$5w$3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2964
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2965
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2966
            // Correct result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2967
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2968
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2969
        // Check double digit group references
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2970
        blah = "zzz123456789101112zzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2971
        p = Pattern.compile("(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2972
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2973
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2974
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2975
        m.appendReplacement(result, "$1w$11w$3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2976
        if (!result.toString().equals("zzz1w11w3"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2977
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2978
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2979
        // Check to make sure it backs off $15 to $1 if only three groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2980
        blah = "zzzabcdcdefzzz";
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2981
        p = Pattern.compile("(ab)(cd)*(ef)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2982
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2983
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2984
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2985
        m.appendReplacement(result, "$1w$15w$3");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2986
        if (!result.toString().equals("zzzabwab5wef"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2987
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2988
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2989
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2990
        // Supplementary character test
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2991
        // SB substitution with literal
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2992
        blah = toSupplementaries("zzzblahzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2993
        p = Pattern.compile(toSupplementaries("blah"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2994
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2995
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2996
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2997
            m.appendReplacement(result, toSupplementaries("blech"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2998
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  2999
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3000
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3001
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3002
        m.appendReplacement(result, toSupplementaries("blech"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3003
        if (!result.toString().equals(toSupplementaries("zzzblech")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3004
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3005
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3006
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3007
        if (!result.toString().equals(toSupplementaries("zzzblechzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3008
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3009
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3010
        // SB substitution with groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3011
        blah = toSupplementaries("zzzabcdzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3012
        p = Pattern.compile(toSupplementaries("(ab)(cd)*"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3013
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3014
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3015
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3016
            m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3017
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3018
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3019
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3020
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3021
        m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3022
        if (!result.toString().equals(toSupplementaries("zzzab")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3023
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3024
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3025
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3026
        if (!result.toString().equals(toSupplementaries("zzzabzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3027
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3028
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3029
        // SB substitution with 3 groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3030
        blah = toSupplementaries("zzzabcdcdefzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3031
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3032
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3033
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3034
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3035
            m.appendReplacement(result, toSupplementaries("$1w$2w$3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3036
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3037
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3038
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3039
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3040
        m.appendReplacement(result, toSupplementaries("$1w$2w$3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3041
        if (!result.toString().equals(toSupplementaries("zzzabwcdwef")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3042
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3043
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3044
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3045
        if (!result.toString().equals(toSupplementaries("zzzabwcdwefzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3046
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3047
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3048
        // SB substitution with groups and three matches
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3049
        // skipping middle match
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3050
        blah = toSupplementaries("zzzabcdzzzabcddzzzabcdzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3051
        p = Pattern.compile(toSupplementaries("(ab)(cd*)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3052
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3053
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3054
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3055
            m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3056
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3057
        } catch (IllegalStateException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3058
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3059
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3060
        m.appendReplacement(result, "$1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3061
        if (!result.toString().equals(toSupplementaries("zzzab")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3062
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3063
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3064
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3065
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3066
        m.appendReplacement(result, "$2");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3067
        if (!result.toString().equals(toSupplementaries("zzzabzzzabcddzzzcd")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3068
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3069
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3070
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3071
        if (!result.toString().equals(toSupplementaries("zzzabzzzabcddzzzcdzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3072
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3073
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3074
        // Check to make sure escaped $ is ignored
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3075
        blah = toSupplementaries("zzzabcdcdefzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3076
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3077
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3078
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3079
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3080
        m.appendReplacement(result, toSupplementaries("$1w\\$2w$3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3081
        if (!result.toString().equals(toSupplementaries("zzzabw$2wef")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3082
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3083
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3084
        m.appendTail(result);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3085
        if (!result.toString().equals(toSupplementaries("zzzabw$2wefzzz")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3086
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3087
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3088
        // Check to make sure a reference to nonexistent group causes error
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3089
        blah = toSupplementaries("zzzabcdcdefzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3090
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3091
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3092
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3093
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3094
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3095
            m.appendReplacement(result, toSupplementaries("$1w$5w$3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3096
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3097
        } catch (IndexOutOfBoundsException ioobe) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3098
            // Correct result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3099
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3100
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3101
        // Check double digit group references
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3102
        blah = toSupplementaries("zzz123456789101112zzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3103
        p = Pattern.compile("(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3104
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3105
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3106
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3107
        m.appendReplacement(result, toSupplementaries("$1w$11w$3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3108
        if (!result.toString().equals(toSupplementaries("zzz1w11w3")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3109
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3110
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3111
        // Check to make sure it backs off $15 to $1 if only three groups
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3112
        blah = toSupplementaries("zzzabcdcdefzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3113
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3114
        m = p.matcher(blah);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3115
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3116
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3117
        m.appendReplacement(result, toSupplementaries("$1w$15w$3"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3118
        if (!result.toString().equals(toSupplementaries("zzzabwab5wef")))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3119
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3120
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3121
        // Check nothing has been appended into the output buffer if
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3122
        // the replacement string triggers IllegalArgumentException.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3123
        p = Pattern.compile("(abc)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3124
        m = p.matcher("abcd");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3125
        result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3126
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3127
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3128
            m.appendReplacement(result, ("xyz$g"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3129
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3130
        } catch (IllegalArgumentException iae) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3131
            if (result.length() != 0)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3132
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3133
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3134
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3135
        report("SB Substitution");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3136
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3137
23734
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3138
    /**
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3139
     * Tests the usage of Matcher.appendReplacement() with literal
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3140
     * and group substitutions.
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3141
     */
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3142
    private static void stringbuilderSubstitute() throws Exception {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3143
        // SB substitution with literal
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3144
        String blah = "zzzblahzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3145
        Pattern p = Pattern.compile("blah");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3146
        Matcher m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3147
        StringBuilder result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3148
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3149
            m.appendReplacement(result, "blech");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3150
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3151
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3152
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3153
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3154
        m.appendReplacement(result, "blech");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3155
        if (!result.toString().equals("zzzblech"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3156
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3157
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3158
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3159
        if (!result.toString().equals("zzzblechzzz"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3160
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3161
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3162
        // SB substitution with groups
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3163
        blah = "zzzabcdzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3164
        p = Pattern.compile("(ab)(cd)*");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3165
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3166
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3167
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3168
            m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3169
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3170
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3171
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3172
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3173
        m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3174
        if (!result.toString().equals("zzzab"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3175
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3176
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3177
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3178
        if (!result.toString().equals("zzzabzzz"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3179
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3180
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3181
        // SB substitution with 3 groups
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3182
        blah = "zzzabcdcdefzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3183
        p = Pattern.compile("(ab)(cd)*(ef)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3184
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3185
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3186
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3187
            m.appendReplacement(result, "$1w$2w$3");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3188
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3189
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3190
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3191
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3192
        m.appendReplacement(result, "$1w$2w$3");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3193
        if (!result.toString().equals("zzzabwcdwef"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3194
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3195
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3196
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3197
        if (!result.toString().equals("zzzabwcdwefzzz"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3198
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3199
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3200
        // SB substitution with groups and three matches
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3201
        // skipping middle match
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3202
        blah = "zzzabcdzzzabcddzzzabcdzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3203
        p = Pattern.compile("(ab)(cd*)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3204
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3205
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3206
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3207
            m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3208
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3209
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3210
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3211
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3212
        m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3213
        if (!result.toString().equals("zzzab"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3214
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3215
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3216
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3217
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3218
        m.appendReplacement(result, "$2");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3219
        if (!result.toString().equals("zzzabzzzabcddzzzcd"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3220
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3221
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3222
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3223
        if (!result.toString().equals("zzzabzzzabcddzzzcdzzz"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3224
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3225
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3226
        // Check to make sure escaped $ is ignored
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3227
        blah = "zzzabcdcdefzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3228
        p = Pattern.compile("(ab)(cd)*(ef)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3229
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3230
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3231
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3232
        m.appendReplacement(result, "$1w\\$2w$3");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3233
        if (!result.toString().equals("zzzabw$2wef"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3234
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3235
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3236
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3237
        if (!result.toString().equals("zzzabw$2wefzzz"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3238
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3239
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3240
        // Check to make sure a reference to nonexistent group causes error
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3241
        blah = "zzzabcdcdefzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3242
        p = Pattern.compile("(ab)(cd)*(ef)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3243
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3244
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3245
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3246
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3247
            m.appendReplacement(result, "$1w$5w$3");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3248
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3249
        } catch (IndexOutOfBoundsException ioobe) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3250
            // Correct result
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3251
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3252
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3253
        // Check double digit group references
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3254
        blah = "zzz123456789101112zzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3255
        p = Pattern.compile("(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3256
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3257
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3258
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3259
        m.appendReplacement(result, "$1w$11w$3");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3260
        if (!result.toString().equals("zzz1w11w3"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3261
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3262
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3263
        // Check to make sure it backs off $15 to $1 if only three groups
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3264
        blah = "zzzabcdcdefzzz";
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3265
        p = Pattern.compile("(ab)(cd)*(ef)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3266
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3267
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3268
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3269
        m.appendReplacement(result, "$1w$15w$3");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3270
        if (!result.toString().equals("zzzabwab5wef"))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3271
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3272
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3273
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3274
        // Supplementary character test
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3275
        // SB substitution with literal
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3276
        blah = toSupplementaries("zzzblahzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3277
        p = Pattern.compile(toSupplementaries("blah"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3278
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3279
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3280
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3281
            m.appendReplacement(result, toSupplementaries("blech"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3282
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3283
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3284
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3285
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3286
        m.appendReplacement(result, toSupplementaries("blech"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3287
        if (!result.toString().equals(toSupplementaries("zzzblech")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3288
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3289
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3290
        if (!result.toString().equals(toSupplementaries("zzzblechzzz")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3291
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3292
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3293
        // SB substitution with groups
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3294
        blah = toSupplementaries("zzzabcdzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3295
        p = Pattern.compile(toSupplementaries("(ab)(cd)*"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3296
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3297
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3298
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3299
            m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3300
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3301
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3302
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3303
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3304
        m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3305
        if (!result.toString().equals(toSupplementaries("zzzab")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3306
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3307
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3308
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3309
        if (!result.toString().equals(toSupplementaries("zzzabzzz")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3310
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3311
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3312
        // SB substitution with 3 groups
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3313
        blah = toSupplementaries("zzzabcdcdefzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3314
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3315
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3316
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3317
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3318
            m.appendReplacement(result, toSupplementaries("$1w$2w$3"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3319
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3320
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3321
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3322
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3323
        m.appendReplacement(result, toSupplementaries("$1w$2w$3"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3324
        if (!result.toString().equals(toSupplementaries("zzzabwcdwef")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3325
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3326
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3327
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3328
        if (!result.toString().equals(toSupplementaries("zzzabwcdwefzzz")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3329
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3330
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3331
        // SB substitution with groups and three matches
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3332
        // skipping middle match
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3333
        blah = toSupplementaries("zzzabcdzzzabcddzzzabcdzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3334
        p = Pattern.compile(toSupplementaries("(ab)(cd*)"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3335
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3336
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3337
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3338
            m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3339
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3340
        } catch (IllegalStateException e) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3341
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3342
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3343
        m.appendReplacement(result, "$1");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3344
        if (!result.toString().equals(toSupplementaries("zzzab")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3345
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3346
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3347
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3348
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3349
        m.appendReplacement(result, "$2");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3350
        if (!result.toString().equals(toSupplementaries("zzzabzzzabcddzzzcd")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3351
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3352
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3353
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3354
        if (!result.toString().equals(toSupplementaries("zzzabzzzabcddzzzcdzzz")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3355
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3356
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3357
        // Check to make sure escaped $ is ignored
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3358
        blah = toSupplementaries("zzzabcdcdefzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3359
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3360
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3361
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3362
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3363
        m.appendReplacement(result, toSupplementaries("$1w\\$2w$3"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3364
        if (!result.toString().equals(toSupplementaries("zzzabw$2wef")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3365
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3366
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3367
        m.appendTail(result);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3368
        if (!result.toString().equals(toSupplementaries("zzzabw$2wefzzz")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3369
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3370
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3371
        // Check to make sure a reference to nonexistent group causes error
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3372
        blah = toSupplementaries("zzzabcdcdefzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3373
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3374
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3375
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3376
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3377
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3378
            m.appendReplacement(result, toSupplementaries("$1w$5w$3"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3379
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3380
        } catch (IndexOutOfBoundsException ioobe) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3381
            // Correct result
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3382
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3383
        // Check double digit group references
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3384
        blah = toSupplementaries("zzz123456789101112zzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3385
        p = Pattern.compile("(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3386
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3387
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3388
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3389
        m.appendReplacement(result, toSupplementaries("$1w$11w$3"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3390
        if (!result.toString().equals(toSupplementaries("zzz1w11w3")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3391
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3392
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3393
        // Check to make sure it backs off $15 to $1 if only three groups
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3394
        blah = toSupplementaries("zzzabcdcdefzzz");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3395
        p = Pattern.compile(toSupplementaries("(ab)(cd)*(ef)"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3396
        m = p.matcher(blah);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3397
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3398
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3399
        m.appendReplacement(result, toSupplementaries("$1w$15w$3"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3400
        if (!result.toString().equals(toSupplementaries("zzzabwab5wef")))
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3401
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3402
        // Check nothing has been appended into the output buffer if
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3403
        // the replacement string triggers IllegalArgumentException.
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3404
        p = Pattern.compile("(abc)");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3405
        m = p.matcher("abcd");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3406
        result = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3407
        m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3408
        try {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3409
            m.appendReplacement(result, ("xyz$g"));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3410
            failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3411
        } catch (IllegalArgumentException iae) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3412
            if (result.length() != 0)
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3413
                failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3414
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3415
        report("SB Substitution 2");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3416
    }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3417
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3418
    /*
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3419
     * 5 groups of characters are created to make a substitution string.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3420
     * A base string will be created including random lead chars, the
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3421
     * substitution string, and random trailing chars.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3422
     * A pattern containing the 5 groups is searched for and replaced with:
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3423
     * random group + random string + random group.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3424
     * The results are checked for correctness.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3425
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3426
    private static void substitutionBasher() {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3427
        for (int runs = 0; runs<1000; runs++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3428
            // Create a base string to work in
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3429
            int leadingChars = generator.nextInt(10);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3430
            StringBuffer baseBuffer = new StringBuffer(100);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3431
            String leadingString = getRandomAlphaString(leadingChars);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3432
            baseBuffer.append(leadingString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3433
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3434
            // Create 5 groups of random number of random chars
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3435
            // Create the string to substitute
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3436
            // Create the pattern string to search for
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3437
            StringBuffer bufferToSub = new StringBuffer(25);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3438
            StringBuffer bufferToPat = new StringBuffer(50);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3439
            String[] groups = new String[5];
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3440
            for(int i=0; i<5; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3441
                int aGroupSize = generator.nextInt(5)+1;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3442
                groups[i] = getRandomAlphaString(aGroupSize);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3443
                bufferToSub.append(groups[i]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3444
                bufferToPat.append('(');
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3445
                bufferToPat.append(groups[i]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3446
                bufferToPat.append(')');
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3447
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3448
            String stringToSub = bufferToSub.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3449
            String pattern = bufferToPat.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3450
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3451
            // Place sub string into working string at random index
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3452
            baseBuffer.append(stringToSub);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3453
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3454
            // Append random chars to end
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3455
            int trailingChars = generator.nextInt(10);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3456
            String trailingString = getRandomAlphaString(trailingChars);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3457
            baseBuffer.append(trailingString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3458
            String baseString = baseBuffer.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3459
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3460
            // Create test pattern and matcher
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3461
            Pattern p = Pattern.compile(pattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3462
            Matcher m = p.matcher(baseString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3463
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3464
            // Reject candidate if pattern happens to start early
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3465
            m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3466
            if (m.start() < leadingChars)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3467
                continue;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3468
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3469
            // Reject candidate if more than one match
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3470
            if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3471
                continue;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3472
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3473
            // Construct a replacement string with :
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3474
            // random group + random string + random group
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3475
            StringBuffer bufferToRep = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3476
            int groupIndex1 = generator.nextInt(5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3477
            bufferToRep.append("$" + (groupIndex1 + 1));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3478
            String randomMidString = getRandomAlphaString(5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3479
            bufferToRep.append(randomMidString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3480
            int groupIndex2 = generator.nextInt(5);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3481
            bufferToRep.append("$" + (groupIndex2 + 1));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3482
            String replacement = bufferToRep.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3483
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3484
            // Do the replacement
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3485
            String result = m.replaceAll(replacement);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3486
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3487
            // Construct expected result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3488
            StringBuffer bufferToRes = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3489
            bufferToRes.append(leadingString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3490
            bufferToRes.append(groups[groupIndex1]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3491
            bufferToRes.append(randomMidString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3492
            bufferToRes.append(groups[groupIndex2]);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3493
            bufferToRes.append(trailingString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3494
            String expectedResult = bufferToRes.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3495
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3496
            // Check results
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3497
            if (!result.equals(expectedResult))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3498
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3499
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3500
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3501
        report("Substitution Basher");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3502
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3503
23734
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3504
    /*
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3505
     * 5 groups of characters are created to make a substitution string.
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3506
     * A base string will be created including random lead chars, the
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3507
     * substitution string, and random trailing chars.
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3508
     * A pattern containing the 5 groups is searched for and replaced with:
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3509
     * random group + random string + random group.
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3510
     * The results are checked for correctness.
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3511
     */
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3512
    private static void substitutionBasher2() {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3513
        for (int runs = 0; runs<1000; runs++) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3514
            // Create a base string to work in
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3515
            int leadingChars = generator.nextInt(10);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3516
            StringBuilder baseBuffer = new StringBuilder(100);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3517
            String leadingString = getRandomAlphaString(leadingChars);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3518
            baseBuffer.append(leadingString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3519
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3520
            // Create 5 groups of random number of random chars
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3521
            // Create the string to substitute
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3522
            // Create the pattern string to search for
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3523
            StringBuilder bufferToSub = new StringBuilder(25);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3524
            StringBuilder bufferToPat = new StringBuilder(50);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3525
            String[] groups = new String[5];
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3526
            for(int i=0; i<5; i++) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3527
                int aGroupSize = generator.nextInt(5)+1;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3528
                groups[i] = getRandomAlphaString(aGroupSize);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3529
                bufferToSub.append(groups[i]);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3530
                bufferToPat.append('(');
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3531
                bufferToPat.append(groups[i]);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3532
                bufferToPat.append(')');
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3533
            }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3534
            String stringToSub = bufferToSub.toString();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3535
            String pattern = bufferToPat.toString();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3536
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3537
            // Place sub string into working string at random index
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3538
            baseBuffer.append(stringToSub);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3539
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3540
            // Append random chars to end
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3541
            int trailingChars = generator.nextInt(10);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3542
            String trailingString = getRandomAlphaString(trailingChars);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3543
            baseBuffer.append(trailingString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3544
            String baseString = baseBuffer.toString();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3545
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3546
            // Create test pattern and matcher
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3547
            Pattern p = Pattern.compile(pattern);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3548
            Matcher m = p.matcher(baseString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3549
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3550
            // Reject candidate if pattern happens to start early
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3551
            m.find();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3552
            if (m.start() < leadingChars)
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3553
                continue;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3554
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3555
            // Reject candidate if more than one match
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3556
            if (m.find())
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3557
                continue;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3558
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3559
            // Construct a replacement string with :
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3560
            // random group + random string + random group
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3561
            StringBuilder bufferToRep = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3562
            int groupIndex1 = generator.nextInt(5);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3563
            bufferToRep.append("$" + (groupIndex1 + 1));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3564
            String randomMidString = getRandomAlphaString(5);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3565
            bufferToRep.append(randomMidString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3566
            int groupIndex2 = generator.nextInt(5);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3567
            bufferToRep.append("$" + (groupIndex2 + 1));
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3568
            String replacement = bufferToRep.toString();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3569
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3570
            // Do the replacement
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3571
            String result = m.replaceAll(replacement);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3572
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3573
            // Construct expected result
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3574
            StringBuilder bufferToRes = new StringBuilder();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3575
            bufferToRes.append(leadingString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3576
            bufferToRes.append(groups[groupIndex1]);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3577
            bufferToRes.append(randomMidString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3578
            bufferToRes.append(groups[groupIndex2]);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3579
            bufferToRes.append(trailingString);
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3580
            String expectedResult = bufferToRes.toString();
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3581
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3582
            // Check results
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3583
            if (!result.equals(expectedResult)) {
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3584
                failCount++;
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3585
            }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3586
        }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3587
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3588
        report("Substitution Basher 2");
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3589
    }
439905b27f94 8039124: j.u.regex.Matcher.appendReplace/Tail() should support StringBuilder variant
sherman
parents: 22997
diff changeset
  3590
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3591
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3592
     * Checks the handling of some escape sequences that the Pattern
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3593
     * class should process instead of the java compiler. These are
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3594
     * not in the file because the escapes should be be processed
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3595
     * by the Pattern class when the regex is compiled.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3596
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3597
    private static void escapes() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3598
        Pattern p = Pattern.compile("\\043");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3599
        Matcher m = p.matcher("#");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3600
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3601
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3602
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3603
        p = Pattern.compile("\\x23");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3604
        m = p.matcher("#");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3605
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3606
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3607
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3608
        p = Pattern.compile("\\u0023");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3609
        m = p.matcher("#");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3610
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3611
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3612
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3613
        report("Escape sequences");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3614
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3615
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3616
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3617
     * Checks the handling of blank input situations. These
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3618
     * tests are incompatible with my test file format.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3619
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3620
    private static void blankInput() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3621
        Pattern p = Pattern.compile("abc", Pattern.CASE_INSENSITIVE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3622
        Matcher m = p.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3623
        if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3624
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3625
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3626
        p = Pattern.compile("a*", Pattern.CASE_INSENSITIVE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3627
        m = p.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3628
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3629
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3630
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3631
        p = Pattern.compile("abc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3632
        m = p.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3633
        if (m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3634
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3635
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3636
        p = Pattern.compile("a*");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3637
        m = p.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3638
        if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3639
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3640
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3641
        report("Blank input");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3642
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3643
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3644
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3645
     * Tests the Boyer-Moore pattern matching of a character sequence
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3646
     * on randomly generated patterns.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3647
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3648
    private static void bm() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3649
        doBnM('a');
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3650
        report("Boyer Moore (ASCII)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3651
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3652
        doBnM(Character.MIN_SUPPLEMENTARY_CODE_POINT - 10);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3653
        report("Boyer Moore (Supplementary)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3654
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3655
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3656
    private static void doBnM(int baseCharacter) throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3657
        int achar=0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3658
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3659
        for (int i=0; i<100; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3660
            // Create a short pattern to search for
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3661
            int patternLength = generator.nextInt(7) + 4;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3662
            StringBuffer patternBuffer = new StringBuffer(patternLength);
31645
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3663
            String pattern;
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3664
            retry: for (;;) {
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3665
                for (int x=0; x<patternLength; x++) {
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3666
                    int ch = baseCharacter + generator.nextInt(26);
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3667
                    if (Character.isSupplementaryCodePoint(ch)) {
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3668
                        patternBuffer.append(Character.toChars(ch));
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3669
                    } else {
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3670
                        patternBuffer.append((char)ch);
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3671
                    }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3672
                }
31645
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3673
                pattern = patternBuffer.toString();
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3674
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3675
                // Avoid patterns that start and end with the same substring
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3676
                // See JDK-6854417
34436
33c20335507c 8143854: java/util/regex/RegExTest.java.RegExTest failed
sherman
parents: 31645
diff changeset
  3677
                for (int x=1; x < pattern.length(); x++) {
31645
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3678
                    if (pattern.startsWith(pattern.substring(x)))
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3679
                        continue retry;
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3680
                }
26db68d6758c 6854417: TESTBUG: java/util/regex/RegExTest.java fails intermittently
igerasim
parents: 30436
diff changeset
  3681
                break;
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3682
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3683
            Pattern p = Pattern.compile(pattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3684
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3685
            // Create a buffer with random ASCII chars that does
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3686
            // not match the sample
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3687
            String toSearch = null;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3688
            StringBuffer s = null;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3689
            Matcher m = p.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3690
            do {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3691
                s = new StringBuffer(100);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3692
                for (int x=0; x<100; x++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3693
                    int ch = baseCharacter + generator.nextInt(26);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3694
                    if (Character.isSupplementaryCodePoint(ch)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3695
                        s.append(Character.toChars(ch));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3696
                    } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3697
                        s.append((char)ch);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3698
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3699
                }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3700
                toSearch = s.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3701
                m.reset(toSearch);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3702
            } while (m.find());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3703
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3704
            // Insert the pattern at a random spot
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3705
            int insertIndex = generator.nextInt(99);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3706
            if (Character.isLowSurrogate(s.charAt(insertIndex)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3707
                insertIndex++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3708
            s = s.insert(insertIndex, pattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3709
            toSearch = s.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3710
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3711
            // Make sure that the pattern is found
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3712
            m.reset(toSearch);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3713
            if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3714
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3715
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3716
            // Make sure that the match text is the pattern
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3717
            if (!m.group().equals(pattern))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3718
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3719
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3720
            // Make sure match occured at insertion point
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3721
            if (m.start() != insertIndex)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3722
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3723
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3724
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3725
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3726
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3727
     * Tests the matching of slices on randomly generated patterns.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3728
     * The Boyer-Moore optimization is not done on these patterns
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3729
     * because it uses unicode case folding.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3730
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3731
    private static void slice() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3732
        doSlice(Character.MAX_VALUE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3733
        report("Slice");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3734
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3735
        doSlice(Character.MAX_CODE_POINT);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3736
        report("Slice (Supplementary)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3737
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3738
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3739
    private static void doSlice(int maxCharacter) throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3740
        Random generator = new Random();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3741
        int achar=0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3742
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3743
        for (int i=0; i<100; i++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3744
            // Create a short pattern to search for
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3745
            int patternLength = generator.nextInt(7) + 4;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3746
            StringBuffer patternBuffer = new StringBuffer(patternLength);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3747
            for (int x=0; x<patternLength; x++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3748
                int randomChar = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3749
                while (!Character.isLetterOrDigit(randomChar))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3750
                    randomChar = generator.nextInt(maxCharacter);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3751
                if (Character.isSupplementaryCodePoint(randomChar)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3752
                    patternBuffer.append(Character.toChars(randomChar));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3753
                } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3754
                    patternBuffer.append((char) randomChar);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3755
                }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3756
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3757
            String pattern =  patternBuffer.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3758
            Pattern p = Pattern.compile(pattern, Pattern.UNICODE_CASE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3759
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3760
            // Create a buffer with random chars that does not match the sample
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3761
            String toSearch = null;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3762
            StringBuffer s = null;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3763
            Matcher m = p.matcher("");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3764
            do {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3765
                s = new StringBuffer(100);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3766
                for (int x=0; x<100; x++) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3767
                    int randomChar = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3768
                    while (!Character.isLetterOrDigit(randomChar))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3769
                        randomChar = generator.nextInt(maxCharacter);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3770
                    if (Character.isSupplementaryCodePoint(randomChar)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3771
                        s.append(Character.toChars(randomChar));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3772
                    } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3773
                        s.append((char) randomChar);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3774
                    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3775
                }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3776
                toSearch = s.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3777
                m.reset(toSearch);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3778
            } while (m.find());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3779
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3780
            // Insert the pattern at a random spot
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3781
            int insertIndex = generator.nextInt(99);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3782
            if (Character.isLowSurrogate(s.charAt(insertIndex)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3783
                insertIndex++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3784
            s = s.insert(insertIndex, pattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3785
            toSearch = s.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3786
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3787
            // Make sure that the pattern is found
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3788
            m.reset(toSearch);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3789
            if (!m.find())
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3790
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3791
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3792
            // Make sure that the match text is the pattern
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3793
            if (!m.group().equals(pattern))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3794
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3795
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3796
            // Make sure match occured at insertion point
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3797
            if (m.start() != insertIndex)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3798
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3799
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3800
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3801
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3802
    private static void explainFailure(String pattern, String data,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3803
                                       String expected, String actual) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3804
        System.err.println("----------------------------------------");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3805
        System.err.println("Pattern = "+pattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3806
        System.err.println("Data = "+data);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3807
        System.err.println("Expected = " + expected);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3808
        System.err.println("Actual   = " + actual);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3809
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3810
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3811
    private static void explainFailure(String pattern, String data,
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3812
                                       Throwable t) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3813
        System.err.println("----------------------------------------");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3814
        System.err.println("Pattern = "+pattern);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3815
        System.err.println("Data = "+data);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3816
        t.printStackTrace(System.err);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3817
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3818
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3819
    // Testing examples from a file
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3820
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3821
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3822
     * Goes through the file "TestCases.txt" and creates many patterns
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3823
     * described in the file, matching the patterns against input lines in
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3824
     * the file, and comparing the results against the correct results
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3825
     * also found in the file. The file format is described in comments
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3826
     * at the head of the file.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3827
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3828
    private static void processFile(String fileName) throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3829
        File testCases = new File(System.getProperty("test.src", "."),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3830
                                  fileName);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3831
        FileInputStream in = new FileInputStream(testCases);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3832
        BufferedReader r = new BufferedReader(new InputStreamReader(in));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3833
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3834
        // Process next test case.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3835
        String aLine;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3836
        while((aLine = r.readLine()) != null) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3837
            // Read a line for pattern
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3838
            String patternString = grabLine(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3839
            Pattern p = null;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3840
            try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3841
                p = compileTestPattern(patternString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3842
            } catch (PatternSyntaxException e) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3843
                String dataString = grabLine(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3844
                String expectedResult = grabLine(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3845
                if (expectedResult.startsWith("error"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3846
                    continue;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3847
                explainFailure(patternString, dataString, e);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3848
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3849
                continue;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3850
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3851
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3852
            // Read a line for input string
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3853
            String dataString = grabLine(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3854
            Matcher m = p.matcher(dataString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3855
            StringBuffer result = new StringBuffer();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3856
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3857
            // Check for IllegalStateExceptions before a match
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3858
            failCount += preMatchInvariants(m);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3859
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3860
            boolean found = m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3861
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3862
            if (found)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3863
                failCount += postTrueMatchInvariants(m);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3864
            else
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3865
                failCount += postFalseMatchInvariants(m);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3866
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3867
            if (found) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3868
                result.append("true ");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3869
                result.append(m.group(0) + " ");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3870
            } else {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3871
                result.append("false ");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3872
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3873
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3874
            result.append(m.groupCount());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3875
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3876
            if (found) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3877
                for (int i=1; i<m.groupCount()+1; i++)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3878
                    if (m.group(i) != null)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3879
                        result.append(" " +m.group(i));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3880
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3881
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3882
            // Read a line for the expected result
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3883
            String expectedResult = grabLine(r);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3884
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3885
            if (!result.toString().equals(expectedResult)) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3886
                explainFailure(patternString, dataString, expectedResult, result.toString());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3887
                failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3888
            }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3889
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3890
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3891
        report(fileName);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3892
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3893
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3894
    private static int preMatchInvariants(Matcher m) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3895
        int failCount = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3896
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3897
            m.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3898
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3899
        } catch (IllegalStateException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3900
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3901
            m.end();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3902
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3903
        } catch (IllegalStateException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3904
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3905
            m.group();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3906
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3907
        } catch (IllegalStateException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3908
        return failCount;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3909
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3910
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3911
    private static int postFalseMatchInvariants(Matcher m) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3912
        int failCount = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3913
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3914
            m.group();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3915
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3916
        } catch (IllegalStateException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3917
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3918
            m.start();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3919
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3920
        } catch (IllegalStateException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3921
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3922
            m.end();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3923
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3924
        } catch (IllegalStateException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3925
        return failCount;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3926
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3927
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3928
    private static int postTrueMatchInvariants(Matcher m) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3929
        int failCount = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3930
        //assert(m.start() = m.start(0);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3931
        if (m.start() != m.start(0))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3932
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3933
        //assert(m.end() = m.end(0);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3934
        if (m.start() != m.start(0))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3935
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3936
        //assert(m.group() = m.group(0);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3937
        if (!m.group().equals(m.group(0)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3938
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3939
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3940
            m.group(50);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3941
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3942
        } catch (IndexOutOfBoundsException ise) {}
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3943
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3944
        return failCount;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3945
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3946
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3947
    private static Pattern compileTestPattern(String patternString) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3948
        if (!patternString.startsWith("'")) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3949
            return Pattern.compile(patternString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3950
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3951
        int break1 = patternString.lastIndexOf("'");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3952
        String flagString = patternString.substring(
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3953
                                          break1+1, patternString.length());
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3954
        patternString = patternString.substring(1, break1);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3955
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3956
        if (flagString.equals("i"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3957
            return Pattern.compile(patternString, Pattern.CASE_INSENSITIVE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3958
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3959
        if (flagString.equals("m"))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3960
            return Pattern.compile(patternString, Pattern.MULTILINE);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3961
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3962
        return Pattern.compile(patternString);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3963
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3964
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3965
    /**
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3966
     * Reads a line from the input file. Keeps reading lines until a non
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3967
     * empty non comment line is read. If the line contains a \n then
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3968
     * these two characters are replaced by a newline char. If a \\uxxxx
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3969
     * sequence is read then the sequence is replaced by the unicode char.
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3970
     */
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3971
    private static String grabLine(BufferedReader r) throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3972
        int index = 0;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3973
        String line = r.readLine();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3974
        while (line.startsWith("//") || line.length() < 1)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3975
            line = r.readLine();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3976
        while ((index = line.indexOf("\\n")) != -1) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3977
            StringBuffer temp = new StringBuffer(line);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3978
            temp.replace(index, index+2, "\n");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3979
            line = temp.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3980
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3981
        while ((index = line.indexOf("\\u")) != -1) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3982
            StringBuffer temp = new StringBuffer(line);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3983
            String value = temp.substring(index+2, index+6);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3984
            char aChar = (char)Integer.parseInt(value, 16);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3985
            String unicodeChar = "" + aChar;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3986
            temp.replace(index, index+6, unicodeChar);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3987
            line = temp.toString();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3988
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3989
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3990
        return line;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3991
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3992
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3993
    private static void check(Pattern p, String s, String g, String expected) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3994
        Matcher m = p.matcher(s);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3995
        m.find();
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  3996
        if (!m.group(g).equals(expected) ||
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  3997
            s.charAt(m.start(g)) != expected.charAt(0) ||
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  3998
            s.charAt(m.end(g) - 1) != expected.charAt(expected.length() - 1))
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  3999
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4000
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4001
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4002
    private static void checkReplaceFirst(String p, String s, String r, String expected)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4003
    {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4004
        if (!expected.equals(Pattern.compile(p)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4005
                                    .matcher(s)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4006
                                    .replaceFirst(r)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4007
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4008
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4009
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4010
    private static void checkReplaceAll(String p, String s, String r, String expected)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4011
    {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4012
        if (!expected.equals(Pattern.compile(p)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4013
                                    .matcher(s)
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4014
                                    .replaceAll(r)))
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4015
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4016
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4017
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4018
    private static void checkExpectedFail(String p) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4019
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4020
            Pattern.compile(p);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4021
        } catch (PatternSyntaxException pse) {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4022
            //pse.printStackTrace();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4023
            return;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4024
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4025
        failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4026
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4027
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4028
    private static void checkExpectedIAE(Matcher m, String g) {
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4029
        m.find();
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4030
        try {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4031
            m.group(g);
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4032
        } catch (IllegalArgumentException x) {
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4033
            //iae.printStackTrace();
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4034
            try {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4035
                m.start(g);
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4036
            } catch (IllegalArgumentException xx) {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4037
                try {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4038
                    m.start(g);
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4039
                } catch (IllegalArgumentException xxx) {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4040
                    return;
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4041
                }
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4042
            }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4043
        }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4044
        failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4045
    }
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4046
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4047
    private static void checkExpectedNPE(Matcher m) {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4048
        m.find();
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4049
        try {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4050
            m.group(null);
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4051
        } catch (NullPointerException x) {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4052
            try {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4053
                m.start(null);
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4054
            } catch (NullPointerException xx) {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4055
                try {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4056
                    m.end(null);
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4057
                } catch (NullPointerException xxx) {
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4058
                    return;
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4059
                }
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4060
            }
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4061
        }
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4062
        failCount++;
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4063
    }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4064
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4065
    private static void namedGroupCaptureTest() throws Exception {
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4066
        check(Pattern.compile("x+(?<gname>y+)z+"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4067
              "xxxyyyzzz",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4068
              "gname",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4069
              "yyy");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4070
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4071
        check(Pattern.compile("x+(?<gname8>y+)z+"),
2290
3a3bde061968 6817475: named-capturing group name started with digit causes PSE exception
sherman
parents: 2070
diff changeset
  4072
              "xxxyyyzzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4073
              "gname8",
2290
3a3bde061968 6817475: named-capturing group name started with digit causes PSE exception
sherman
parents: 2070
diff changeset
  4074
              "yyy");
3a3bde061968 6817475: named-capturing group name started with digit causes PSE exception
sherman
parents: 2070
diff changeset
  4075
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4076
        //backref
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4077
        Pattern pattern = Pattern.compile("(a*)bc\\1");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4078
        check(pattern, "zzzaabcazzz", true);  // found "abca"
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4079
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4080
        check(Pattern.compile("(?<gname>a*)bc\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4081
              "zzzaabcaazzz", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4082
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4083
        check(Pattern.compile("(?<gname>abc)(def)\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4084
              "abcdefabc", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4085
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4086
        check(Pattern.compile("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)(?<gname>k)\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4087
              "abcdefghijkk", true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4088
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4089
        // Supplementary character tests
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4090
        check(Pattern.compile("(?<gname>" + toSupplementaries("a*)bc") + "\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4091
              toSupplementaries("zzzaabcazzz"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4092
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4093
        check(Pattern.compile("(?<gname>" + toSupplementaries("a*)bc") + "\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4094
              toSupplementaries("zzzaabcaazzz"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4095
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4096
        check(Pattern.compile("(?<gname>" + toSupplementaries("abc)(def)") + "\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4097
              toSupplementaries("abcdefabc"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4098
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4099
        check(Pattern.compile(toSupplementaries("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)") +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4100
                              "(?<gname>" +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4101
                              toSupplementaries("k)") + "\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4102
              toSupplementaries("abcdefghijkk"), true);
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4103
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4104
        check(Pattern.compile("x+(?<gname>y+)z+\\k<gname>"),
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4105
              "xxxyyyzzzyyy",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4106
              "gname",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4107
              "yyy");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4108
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4109
        //replaceFirst/All
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4110
        checkReplaceFirst("(?<gn>ab)(c*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4111
                          "abccczzzabcczzzabccc",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4112
                          "${gn}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4113
                          "abzzzabcczzzabccc");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4114
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4115
        checkReplaceAll("(?<gn>ab)(c*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4116
                        "abccczzzabcczzzabccc",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4117
                        "${gn}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4118
                        "abzzzabzzzab");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4119
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4120
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4121
        checkReplaceFirst("(?<gn>ab)(c*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4122
                          "zzzabccczzzabcczzzabccczzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4123
                          "${gn}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4124
                          "zzzabzzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4125
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4126
        checkReplaceAll("(?<gn>ab)(c*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4127
                        "zzzabccczzzabcczzzabccczzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4128
                        "${gn}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4129
                        "zzzabzzzabzzzabzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4130
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4131
        checkReplaceFirst("(?<gn1>ab)(?<gn2>c*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4132
                          "zzzabccczzzabcczzzabccczzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4133
                          "${gn2}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4134
                          "zzzccczzzabcczzzabccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4135
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4136
        checkReplaceAll("(?<gn1>ab)(?<gn2>c*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4137
                        "zzzabccczzzabcczzzabccczzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4138
                        "${gn2}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4139
                        "zzzccczzzcczzzccczzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4140
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4141
        //toSupplementaries("(ab)(c*)"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4142
        checkReplaceFirst("(?<gn1>" + toSupplementaries("ab") +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4143
                           ")(?<gn2>" + toSupplementaries("c") + "*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4144
                          toSupplementaries("abccczzzabcczzzabccc"),
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4145
                          "${gn1}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4146
                          toSupplementaries("abzzzabcczzzabccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4147
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4148
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4149
        checkReplaceAll("(?<gn1>" + toSupplementaries("ab") +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4150
                        ")(?<gn2>" + toSupplementaries("c") + "*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4151
                        toSupplementaries("abccczzzabcczzzabccc"),
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4152
                        "${gn1}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4153
                        toSupplementaries("abzzzabzzzab"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4154
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4155
        checkReplaceFirst("(?<gn1>" + toSupplementaries("ab") +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4156
                           ")(?<gn2>" + toSupplementaries("c") + "*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4157
                          toSupplementaries("abccczzzabcczzzabccc"),
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4158
                          "${gn2}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4159
                          toSupplementaries("ccczzzabcczzzabccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4160
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4161
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4162
        checkReplaceAll("(?<gn1>" + toSupplementaries("ab") +
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4163
                        ")(?<gn2>" + toSupplementaries("c") + "*)",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4164
                        toSupplementaries("abccczzzabcczzzabccc"),
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4165
                        "${gn2}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4166
                        toSupplementaries("ccczzzcczzzccc"));
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4167
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4168
        checkReplaceFirst("(?<dog>Dog)AndCat",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4169
                          "zzzDogAndCatzzzDogAndCatzzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4170
                          "${dog}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4171
                          "zzzDogzzzDogAndCatzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4172
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4173
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4174
        checkReplaceAll("(?<dog>Dog)AndCat",
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4175
                          "zzzDogAndCatzzzDogAndCatzzz",
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4176
                          "${dog}",
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4177
                          "zzzDogzzzDogzzz");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4178
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4179
        // backref in Matcher & String
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4180
        if (!"abcdefghij".replaceFirst("cd(?<gn>ef)gh", "${gn}").equals("abefij") ||
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4181
            !"abbbcbdbefgh".replaceAll("(?<gn>[a-e])b", "${gn}").equals("abcdefgh"))
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4182
            failCount++;
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4183
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4184
        // negative
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4185
        checkExpectedFail("(?<groupnamehasnoascii.in>abc)(def)");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4186
        checkExpectedFail("(?<groupnamehasnoascii_in>abc)(def)");
4161
679d00486dc6 6878475: Better syntax for the named capture group in regex
sherman
parents: 2290
diff changeset
  4187
        checkExpectedFail("(?<6groupnamestartswithdigit>abc)(def)");
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4188
        checkExpectedFail("(?<gname>abc)(def)\\k<gnameX>");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4189
        checkExpectedFail("(?<gname>abc)(?<gname>def)\\k<gnameX>");
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4190
        checkExpectedIAE(Pattern.compile("(?<gname>abc)(def)").matcher("abcdef"),
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4191
                         "gnameX");
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4192
        checkExpectedNPE(Pattern.compile("(?<gname>abc)(def)").matcher("abcdef"));
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4193
        report("NamedGroupCapture");
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4194
    }
4820
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4195
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4196
    // This is for bug 6919132
4820
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4197
    private static void nonBmpClassComplementTest() throws Exception {
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4198
        Pattern p = Pattern.compile("\\P{Lu}");
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4199
        Matcher m = p.matcher(new String(new int[] {0x1d400}, 0, 1));
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4200
4820
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4201
        if (m.find() && m.start() == 1)
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4202
            failCount++;
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4203
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4204
        // from a unicode category
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4205
        p = Pattern.compile("\\P{Lu}");
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4206
        m = p.matcher(new String(new int[] {0x1d400}, 0, 1));
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4207
        if (m.find())
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4208
            failCount++;
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4209
        if (!m.hitEnd())
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4210
            failCount++;
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4211
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4212
        // block
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4213
        p = Pattern.compile("\\P{InMathematicalAlphanumericSymbols}");
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4214
        m = p.matcher(new String(new int[] {0x1d400}, 0, 1));
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4215
        if (m.find() && m.start() == 1)
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4216
            failCount++;
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4217
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4218
        p = Pattern.compile("\\P{sc=GRANTHA}");
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4219
        m = p.matcher(new String(new int[] {0x11350}, 0, 1));
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4220
        if (m.find() && m.start() == 1)
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4221
            failCount++;
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4222
4820
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4223
        report("NonBmpClassComplement");
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4224
    }
e32f5d2ba062 6919132: Regex \P{Lu} selects half of a surrogate pari
sherman
parents: 4161
diff changeset
  4225
5610
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4226
    private static void unicodePropertiesTest() throws Exception {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4227
        // different forms
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4228
        if (!Pattern.compile("\\p{IsLu}").matcher("A").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4229
            !Pattern.compile("\\p{Lu}").matcher("A").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4230
            !Pattern.compile("\\p{gc=Lu}").matcher("A").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4231
            !Pattern.compile("\\p{general_category=Lu}").matcher("A").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4232
            !Pattern.compile("\\p{IsLatin}").matcher("B").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4233
            !Pattern.compile("\\p{sc=Latin}").matcher("B").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4234
            !Pattern.compile("\\p{script=Latin}").matcher("B").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4235
            !Pattern.compile("\\p{InBasicLatin}").matcher("c").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4236
            !Pattern.compile("\\p{blk=BasicLatin}").matcher("c").matches() ||
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4237
            !Pattern.compile("\\p{block=BasicLatin}").matcher("c").matches())
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4238
            failCount++;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4239
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4240
        Matcher common  = Pattern.compile("\\p{script=Common}").matcher("");
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4241
        Matcher unknown = Pattern.compile("\\p{IsUnknown}").matcher("");
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4242
        Matcher lastSM  = common;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4243
        Character.UnicodeScript lastScript = Character.UnicodeScript.of(0);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4244
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4245
        Matcher latin  = Pattern.compile("\\p{block=basic_latin}").matcher("");
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4246
        Matcher greek  = Pattern.compile("\\p{InGreek}").matcher("");
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4247
        Matcher lastBM = latin;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4248
        Character.UnicodeBlock lastBlock = Character.UnicodeBlock.of(0);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4249
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4250
        for (int cp = 1; cp < Character.MAX_CODE_POINT; cp++) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4251
            if (cp >= 0x30000 && (cp & 0x70) == 0){
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4252
                continue;  // only pick couple code points, they are the same
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4253
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4254
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4255
            // Unicode Script
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4256
            Character.UnicodeScript script = Character.UnicodeScript.of(cp);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4257
            Matcher m;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4258
            String str = new String(Character.toChars(cp));
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4259
            if (script == lastScript) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4260
                 m = lastSM;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4261
                 m.reset(str);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4262
            } else {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4263
                 m  = Pattern.compile("\\p{Is" + script.name() + "}").matcher(str);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4264
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4265
            if (!m.matches()) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4266
                failCount++;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4267
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4268
            Matcher other = (script == Character.UnicodeScript.COMMON)? unknown : common;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4269
            other.reset(str);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4270
            if (other.matches()) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4271
                failCount++;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4272
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4273
            lastSM = m;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4274
            lastScript = script;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4275
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4276
            // Unicode Block
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4277
            Character.UnicodeBlock block = Character.UnicodeBlock.of(cp);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4278
            if (block == null) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4279
                //System.out.printf("Not a Block: cp=%x%n", cp);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4280
                continue;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4281
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4282
            if (block == lastBlock) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4283
                 m = lastBM;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4284
                 m.reset(str);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4285
            } else {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4286
                 m  = Pattern.compile("\\p{block=" + block.toString() + "}").matcher(str);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4287
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4288
            if (!m.matches()) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4289
                failCount++;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4290
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4291
            other = (block == Character.UnicodeBlock.BASIC_LATIN)? greek : latin;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4292
            other.reset(str);
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4293
            if (other.matches()) {
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4294
                failCount++;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4295
            }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4296
            lastBM = m;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4297
            lastBlock = block;
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4298
        }
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4299
        report("unicodeProperties");
fd2427610c7f 6945564: Unicode script support in Character class
sherman
parents: 5295
diff changeset
  4300
    }
8173
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4301
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4302
    private static void unicodeHexNotationTest() throws Exception {
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4303
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4304
        // negative
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4305
        checkExpectedFail("\\x{-23}");
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4306
        checkExpectedFail("\\x{110000}");
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4307
        checkExpectedFail("\\x{}");
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4308
        checkExpectedFail("\\x{AB[ef]");
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4309
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4310
        // codepoint
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4311
        check("^\\x{1033c}$",              "\uD800\uDF3C", true);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4312
        check("^\\xF0\\x90\\x8C\\xBC$",    "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4313
        check("^\\x{D800}\\x{DF3c}+$",     "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4314
        check("^\\xF0\\x90\\x8C\\xBC$",    "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4315
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4316
        // in class
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4317
        check("^[\\x{D800}\\x{DF3c}]+$",   "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4318
        check("^[\\xF0\\x90\\x8C\\xBC]+$", "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4319
        check("^[\\x{D800}\\x{DF3C}]+$",   "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4320
        check("^[\\x{DF3C}\\x{D800}]+$",   "\uD800\uDF3C", false);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4321
        check("^[\\x{D800}\\x{DF3C}]+$",   "\uDF3C\uD800", true);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4322
        check("^[\\x{DF3C}\\x{D800}]+$",   "\uDF3C\uD800", true);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4323
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4324
        for (int cp = 0; cp <= 0x10FFFF; cp++) {
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4325
             String s = "A" + new String(Character.toChars(cp)) + "B";
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4326
             String hexUTF16 = (cp <= 0xFFFF)? String.format("\\u%04x", cp)
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4327
                                             : String.format("\\u%04x\\u%04x",
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4328
                                               (int) Character.toChars(cp)[0],
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4329
                                               (int) Character.toChars(cp)[1]);
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4330
             String hexCodePoint = "\\x{" + Integer.toHexString(cp) + "}";
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4331
             if (!Pattern.matches("A" + hexUTF16 + "B", s))
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4332
                 failCount++;
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4333
             if (!Pattern.matches("A[" + hexUTF16 + "]B", s))
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4334
                 failCount++;
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4335
             if (!Pattern.matches("A" + hexCodePoint + "B", s))
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4336
                 failCount++;
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4337
             if (!Pattern.matches("A[" + hexCodePoint + "]B", s))
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4338
                 failCount++;
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4339
         }
a3a39b98e05a 7014645: Support perl style Unicode hex notation \x{...}
sherman
parents: 7668
diff changeset
  4340
         report("unicodeHexNotation");
9536
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4341
    }
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4342
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4343
    private static void unicodeClassesTest() throws Exception {
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4344
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4345
        Matcher lower  = Pattern.compile("\\p{Lower}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4346
        Matcher upper  = Pattern.compile("\\p{Upper}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4347
        Matcher ASCII  = Pattern.compile("\\p{ASCII}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4348
        Matcher alpha  = Pattern.compile("\\p{Alpha}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4349
        Matcher digit  = Pattern.compile("\\p{Digit}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4350
        Matcher alnum  = Pattern.compile("\\p{Alnum}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4351
        Matcher punct  = Pattern.compile("\\p{Punct}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4352
        Matcher graph  = Pattern.compile("\\p{Graph}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4353
        Matcher print  = Pattern.compile("\\p{Print}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4354
        Matcher blank  = Pattern.compile("\\p{Blank}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4355
        Matcher cntrl  = Pattern.compile("\\p{Cntrl}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4356
        Matcher xdigit = Pattern.compile("\\p{XDigit}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4357
        Matcher space  = Pattern.compile("\\p{Space}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4358
        Matcher bound  = Pattern.compile("\\b").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4359
        Matcher word   = Pattern.compile("\\w++").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4360
        // UNICODE_CHARACTER_CLASS
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4361
        Matcher lowerU  = Pattern.compile("\\p{Lower}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4362
        Matcher upperU  = Pattern.compile("\\p{Upper}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4363
        Matcher ASCIIU  = Pattern.compile("\\p{ASCII}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4364
        Matcher alphaU  = Pattern.compile("\\p{Alpha}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4365
        Matcher digitU  = Pattern.compile("\\p{Digit}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4366
        Matcher alnumU  = Pattern.compile("\\p{Alnum}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4367
        Matcher punctU  = Pattern.compile("\\p{Punct}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4368
        Matcher graphU  = Pattern.compile("\\p{Graph}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4369
        Matcher printU  = Pattern.compile("\\p{Print}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4370
        Matcher blankU  = Pattern.compile("\\p{Blank}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4371
        Matcher cntrlU  = Pattern.compile("\\p{Cntrl}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4372
        Matcher xdigitU = Pattern.compile("\\p{XDigit}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4373
        Matcher spaceU  = Pattern.compile("\\p{Space}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4374
        Matcher boundU  = Pattern.compile("\\b", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4375
        Matcher wordU   = Pattern.compile("\\w", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4376
        // embedded flag (?U)
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4377
        Matcher lowerEU  = Pattern.compile("(?U)\\p{Lower}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4378
        Matcher graphEU  = Pattern.compile("(?U)\\p{Graph}", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4379
        Matcher wordEU   = Pattern.compile("(?U)\\w", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4380
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4381
        Matcher bwb    = Pattern.compile("\\b\\w\\b").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4382
        Matcher bwbU   = Pattern.compile("\\b\\w++\\b", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4383
        Matcher bwbEU  = Pattern.compile("(?U)\\b\\w++\\b", Pattern.UNICODE_CHARACTER_CLASS).matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4384
        // properties
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4385
        Matcher lowerP  = Pattern.compile("\\p{IsLowerCase}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4386
        Matcher upperP  = Pattern.compile("\\p{IsUpperCase}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4387
        Matcher titleP  = Pattern.compile("\\p{IsTitleCase}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4388
        Matcher letterP = Pattern.compile("\\p{IsLetter}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4389
        Matcher alphaP  = Pattern.compile("\\p{IsAlphabetic}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4390
        Matcher ideogP  = Pattern.compile("\\p{IsIdeographic}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4391
        Matcher cntrlP  = Pattern.compile("\\p{IsControl}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4392
        Matcher spaceP  = Pattern.compile("\\p{IsWhiteSpace}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4393
        Matcher definedP = Pattern.compile("\\p{IsAssigned}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4394
        Matcher nonCCPP = Pattern.compile("\\p{IsNoncharacterCodePoint}").matcher("");
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4395
        Matcher joinCrtl = Pattern.compile("\\p{IsJoinControl}").matcher("");
9536
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4396
        // javaMethod
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4397
        Matcher lowerJ  = Pattern.compile("\\p{javaLowerCase}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4398
        Matcher upperJ  = Pattern.compile("\\p{javaUpperCase}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4399
        Matcher alphaJ  = Pattern.compile("\\p{javaAlphabetic}").matcher("");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4400
        Matcher ideogJ  = Pattern.compile("\\p{javaIdeographic}").matcher("");
38450
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4401
        // GC/C
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4402
        Matcher gcC  = Pattern.compile("\\p{C}").matcher("");
9536
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4403
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4404
        for (int cp = 1; cp < 0x30000; cp++) {
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4405
            String str = new String(Character.toChars(cp));
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4406
            int type = Character.getType(cp);
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4407
            if (// lower
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4408
                POSIX_ASCII.isLower(cp)   != lower.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4409
                Character.isLowerCase(cp) != lowerU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4410
                Character.isLowerCase(cp) != lowerP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4411
                Character.isLowerCase(cp) != lowerEU.reset(str).matches()||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4412
                Character.isLowerCase(cp) != lowerJ.reset(str).matches()||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4413
                // upper
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4414
                POSIX_ASCII.isUpper(cp)   != upper.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4415
                POSIX_Unicode.isUpper(cp) != upperU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4416
                Character.isUpperCase(cp) != upperP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4417
                Character.isUpperCase(cp) != upperJ.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4418
                // alpha
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4419
                POSIX_ASCII.isAlpha(cp)   != alpha.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4420
                POSIX_Unicode.isAlpha(cp) != alphaU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4421
                Character.isAlphabetic(cp)!= alphaP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4422
                Character.isAlphabetic(cp)!= alphaJ.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4423
                // digit
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4424
                POSIX_ASCII.isDigit(cp)   != digit.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4425
                Character.isDigit(cp)     != digitU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4426
                // alnum
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4427
                POSIX_ASCII.isAlnum(cp)   != alnum.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4428
                POSIX_Unicode.isAlnum(cp) != alnumU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4429
                // punct
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4430
                POSIX_ASCII.isPunct(cp)   != punct.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4431
                POSIX_Unicode.isPunct(cp) != punctU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4432
                // graph
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4433
                POSIX_ASCII.isGraph(cp)   != graph.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4434
                POSIX_Unicode.isGraph(cp) != graphU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4435
                POSIX_Unicode.isGraph(cp) != graphEU.reset(str).matches()||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4436
                // blank
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4437
                POSIX_ASCII.isType(cp, POSIX_ASCII.BLANK)
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4438
                                          != blank.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4439
                POSIX_Unicode.isBlank(cp) != blankU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4440
                // print
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4441
                POSIX_ASCII.isPrint(cp)   != print.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4442
                POSIX_Unicode.isPrint(cp) != printU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4443
                // cntrl
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4444
                POSIX_ASCII.isCntrl(cp)   != cntrl.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4445
                POSIX_Unicode.isCntrl(cp) != cntrlU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4446
                (Character.CONTROL == type) != cntrlP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4447
                // hexdigit
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4448
                POSIX_ASCII.isHexDigit(cp)   != xdigit.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4449
                POSIX_Unicode.isHexDigit(cp) != xdigitU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4450
                // space
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4451
                POSIX_ASCII.isSpace(cp)   != space.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4452
                POSIX_Unicode.isSpace(cp) != spaceU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4453
                POSIX_Unicode.isSpace(cp) != spaceP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4454
                // word
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4455
                POSIX_ASCII.isWord(cp)   != word.reset(str).matches()  ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4456
                POSIX_Unicode.isWord(cp) != wordU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4457
                POSIX_Unicode.isWord(cp) != wordEU.reset(str).matches()||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4458
                // bwordb
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4459
                POSIX_ASCII.isWord(cp) != bwb.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4460
                POSIX_Unicode.isWord(cp) != bwbU.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4461
                // properties
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4462
                Character.isTitleCase(cp) != titleP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4463
                Character.isLetter(cp)    != letterP.reset(str).matches()||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4464
                Character.isIdeographic(cp) != ideogP.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4465
                Character.isIdeographic(cp) != ideogJ.reset(str).matches() ||
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4466
                (Character.UNASSIGNED == type) == definedP.reset(str).matches() ||
17434
4a04d7127e80 8013252: Regex Matcher .start and .end should be accessible by group name
sherman
parents: 17183
diff changeset
  4467
                POSIX_Unicode.isNoncharacterCodePoint(cp) != nonCCPP.reset(str).matches() ||
38450
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4468
                POSIX_Unicode.isJoinControl(cp) != joinCrtl.reset(str).matches() ||
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4469
                // gc_C
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4470
                (Character.CONTROL == type || Character.FORMAT == type ||
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4471
                 Character.PRIVATE_USE == type || Character.SURROGATE == type ||
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4472
                 Character.UNASSIGNED == type)
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4473
                != gcC.reset(str).matches()) {
9536
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4474
                failCount++;
38450
516990ff3a4c 8143282: \p{Cn} unassigned code points should be included in \p{C}
sherman
parents: 37882
diff changeset
  4475
            }
9536
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4476
        }
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4477
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4478
        // bounds/word align
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4479
        twoFindIndexes(" \u0180sherman\u0400 ", bound, 1, 10);
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4480
        if (!bwbU.reset("\u0180sherman\u0400").matches())
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4481
            failCount++;
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4482
        twoFindIndexes(" \u0180sh\u0345erman\u0400 ", bound, 1, 11);
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4483
        if (!bwbU.reset("\u0180sh\u0345erman\u0400").matches())
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4484
            failCount++;
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4485
        twoFindIndexes(" \u0724\u0739\u0724 ", bound, 1, 4);
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4486
        if (!bwbU.reset("\u0724\u0739\u0724").matches())
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4487
            failCount++;
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4488
        if (!bwbEU.reset("\u0724\u0739\u0724").matches())
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4489
            failCount++;
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4490
        report("unicodePredefinedClasses");
648c9add2a74 7039066: j.u.rgex does not match TR18 RL1.4 Simple Word Boundaries and RL1.2 Properties
sherman
parents: 9035
diff changeset
  4491
    }
12675
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4492
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4493
    private static void unicodeCharacterNameTest() throws Exception {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4494
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4495
        for (int cp = 0; cp < Character.MAX_CODE_POINT; cp++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4496
            if (!Character.isValidCodePoint(cp) ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4497
                Character.getType(cp) == Character.UNASSIGNED)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4498
                continue;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4499
            String str = new String(Character.toChars(cp));
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4500
            // single
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4501
            String p = "\\N{" + Character.getName(cp) + "}";
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4502
            if (!Pattern.compile(p).matcher(str).matches()) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4503
                failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4504
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4505
            // class[c]
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4506
            p = "[\\N{" + Character.getName(cp) + "}]";
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4507
            if (!Pattern.compile(p).matcher(str).matches()) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4508
                failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4509
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4510
        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4511
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4512
        // range
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4513
        for (int i = 0; i < 10; i++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4514
            int start = generator.nextInt(20);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4515
            int end = start + generator.nextInt(200);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4516
            String p = "[\\N{" + Character.getName(start) + "}-\\N{" + Character.getName(end) + "}]";
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4517
            String str;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4518
            for (int cp = start; cp < end; cp++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4519
                str = new String(Character.toChars(cp));
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4520
                if (!Pattern.compile(p).matcher(str).matches()) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4521
                    failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4522
                }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4523
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4524
            str = new String(Character.toChars(end + 10));
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4525
            if (Pattern.compile(p).matcher(str).matches()) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4526
                failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4527
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4528
        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4529
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4530
        // slice
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4531
        for (int i = 0; i < 10; i++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4532
            int n = generator.nextInt(256);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4533
            int[] buf = new int[n];
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4534
            StringBuffer sb = new StringBuffer(1024);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4535
            for (int j = 0; j < n; j++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4536
                int cp = generator.nextInt(1000);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4537
                if (!Character.isValidCodePoint(cp) ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4538
                    Character.getType(cp) == Character.UNASSIGNED)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4539
                    cp = 0x4e00;    // just use 4e00
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4540
                sb.append("\\N{" + Character.getName(cp) + "}");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4541
                buf[j] = cp;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4542
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4543
            String p = sb.toString();
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4544
            String str = new String(buf, 0, buf.length);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4545
            if (!Pattern.compile(p).matcher(str).matches()) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4546
                failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4547
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4548
        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4549
        report("unicodeCharacterName");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4550
    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4551
12675
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4552
    private static void horizontalAndVerticalWSTest() throws Exception {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4553
        String hws = new String (new char[] {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4554
                                     0x09, 0x20, 0xa0, 0x1680, 0x180e,
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4555
                                     0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005,
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4556
                                     0x2006, 0x2007, 0x2008, 0x2009, 0x200a,
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4557
                                     0x202f, 0x205f, 0x3000 });
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4558
        String vws = new String (new char[] {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4559
                                     0x0a, 0x0b, 0x0c, 0x0d, 0x85, 0x2028, 0x2029 });
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4560
        if (!Pattern.compile("\\h+").matcher(hws).matches() ||
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4561
            !Pattern.compile("[\\h]+").matcher(hws).matches())
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4562
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4563
        if (Pattern.compile("\\H").matcher(hws).find() ||
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4564
            Pattern.compile("[\\H]").matcher(hws).find())
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4565
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4566
        if (!Pattern.compile("\\v+").matcher(vws).matches() ||
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4567
            !Pattern.compile("[\\v]+").matcher(vws).matches())
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4568
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4569
        if (Pattern.compile("\\V").matcher(vws).find() ||
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4570
            Pattern.compile("[\\V]").matcher(vws).find())
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4571
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4572
        String prefix = "abcd";
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4573
        String suffix = "efgh";
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4574
        String ng = "A";
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4575
        for (int i = 0; i < hws.length(); i++) {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4576
            String c = String.valueOf(hws.charAt(i));
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4577
            Matcher m = Pattern.compile("\\h").matcher(prefix + c + suffix);
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4578
            if (!m.find() || !c.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4579
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4580
            m = Pattern.compile("[\\h]").matcher(prefix + c + suffix);
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4581
            if (!m.find() || !c.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4582
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4583
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4584
            m = Pattern.compile("\\H").matcher(hws.substring(0, i) + ng + hws.substring(i));
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4585
            if (!m.find() || !ng.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4586
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4587
            m = Pattern.compile("[\\H]").matcher(hws.substring(0, i) + ng + hws.substring(i));
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4588
            if (!m.find() || !ng.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4589
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4590
        }
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4591
        for (int i = 0; i < vws.length(); i++) {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4592
            String c = String.valueOf(vws.charAt(i));
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4593
            Matcher m = Pattern.compile("\\v").matcher(prefix + c + suffix);
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4594
            if (!m.find() || !c.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4595
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4596
            m = Pattern.compile("[\\v]").matcher(prefix + c + suffix);
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4597
            if (!m.find() || !c.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4598
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4599
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4600
            m = Pattern.compile("\\V").matcher(vws.substring(0, i) + ng + vws.substring(i));
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4601
            if (!m.find() || !ng.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4602
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4603
            m = Pattern.compile("[\\V]").matcher(vws.substring(0, i) + ng + vws.substring(i));
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4604
            if (!m.find() || !ng.equals(m.group()))
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4605
                failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4606
        }
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4607
        // \v in range is interpreted as 0x0B. This is the undocumented behavior
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4608
        if (!Pattern.compile("[\\v-\\v]").matcher(String.valueOf((char)0x0B)).matches())
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4609
            failCount++;
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4610
        report("horizontalAndVerticalWSTest");
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4611
    }
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4612
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4613
    private static void linebreakTest() throws Exception {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4614
        String linebreaks = new String (new char[] {
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4615
            0x0A, 0x0B, 0x0C, 0x0D, 0x85, 0x2028, 0x2029 });
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4616
        String crnl = "\r\n";
44122
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4617
        if (!(Pattern.compile("\\R+").matcher(linebreaks).matches() &&
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4618
              Pattern.compile("\\R").matcher(crnl).matches() &&
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4619
              Pattern.compile("\\Rabc").matcher(crnl + "abc").matches() &&
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4620
              Pattern.compile("\\Rabc").matcher("\rabc").matches() &&
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4621
              Pattern.compile("\\R\\R").matcher(crnl).matches() &&  // backtracking
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4622
              Pattern.compile("\\R\\n").matcher(crnl).matches()) && // backtracking
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4623
              !Pattern.compile("((?<!\\R)\\s)*").matcher(crnl).matches()) { // #8176029
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4624
            failCount++;
65a14579a2ae 8176029: Linebreak matcher is not equivalent to the pattern as stated in javadoc
sherman
parents: 38777
diff changeset
  4625
        }
12675
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4626
        report("linebreakTest");
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4627
    }
3b7d71495649 7014640: To add a metachar \R for line ending and character classes for vertical/horizontal ws \v \V \h \H
sherman
parents: 12432
diff changeset
  4628
13554
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4629
    // #7189363
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4630
    private static void branchTest() throws Exception {
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4631
        if (!Pattern.compile("(a)?bc|d").matcher("d").find() ||     // greedy
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4632
            !Pattern.compile("(a)+bc|d").matcher("d").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4633
            !Pattern.compile("(a)*bc|d").matcher("d").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4634
            !Pattern.compile("(a)??bc|d").matcher("d").find() ||    // reluctant
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4635
            !Pattern.compile("(a)+?bc|d").matcher("d").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4636
            !Pattern.compile("(a)*?bc|d").matcher("d").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4637
            !Pattern.compile("(a)?+bc|d").matcher("d").find() ||    // possessive
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4638
            !Pattern.compile("(a)++bc|d").matcher("d").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4639
            !Pattern.compile("(a)*+bc|d").matcher("d").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4640
            !Pattern.compile("(a)?bc|d").matcher("d").matches() ||  // greedy
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4641
            !Pattern.compile("(a)+bc|d").matcher("d").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4642
            !Pattern.compile("(a)*bc|d").matcher("d").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4643
            !Pattern.compile("(a)??bc|d").matcher("d").matches() || // reluctant
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4644
            !Pattern.compile("(a)+?bc|d").matcher("d").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4645
            !Pattern.compile("(a)*?bc|d").matcher("d").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4646
            !Pattern.compile("(a)?+bc|d").matcher("d").matches() || // possessive
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4647
            !Pattern.compile("(a)++bc|d").matcher("d").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4648
            !Pattern.compile("(a)*+bc|d").matcher("d").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4649
            !Pattern.compile("(a)?bc|de").matcher("de").find() ||   // others
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4650
            !Pattern.compile("(a)??bc|de").matcher("de").find() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4651
            !Pattern.compile("(a)?bc|de").matcher("de").matches() ||
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4652
            !Pattern.compile("(a)??bc|de").matcher("de").matches())
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4653
            failCount++;
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4654
        report("branchTest");
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4655
    }
ad737436c54a 7189363: Regex Pattern compilation buggy for special sequences
sherman
parents: 12675
diff changeset
  4656
17183
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4657
    // This test is for 8007395
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4658
    private static void groupCurlyNotFoundSuppTest() throws Exception {
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4659
        String input = "test this as \ud83d\ude0d";
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4660
        for (String pStr : new String[] { "test(.)+(@[a-zA-Z.]+)",
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4661
                                          "test(.)*(@[a-zA-Z.]+)",
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4662
                                          "test([^B])+(@[a-zA-Z.]+)",
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4663
                                          "test([^B])*(@[a-zA-Z.]+)",
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4664
                                          "test(\\P{IsControl})+(@[a-zA-Z.]+)",
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4665
                                          "test(\\P{IsControl})*(@[a-zA-Z.]+)",
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4666
                                        }) {
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4667
            Matcher m = Pattern.compile(pStr, Pattern.CASE_INSENSITIVE)
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4668
                               .matcher(input);
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4669
            try {
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4670
                if (m.find()) {
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4671
                    failCount++;
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4672
                }
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4673
            } catch (Exception x) {
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4674
                failCount++;
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4675
            }
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4676
        }
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4677
        report("GroupCurly NotFoundSupp");
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4678
    }
1e21e0bc10e7 8007395: StringIndexOutofBoundsException in Match.find() when input String contains surrogate UTF-16 characters
sherman
parents: 14342
diff changeset
  4679
19604
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4680
    // This test is for 8023647
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4681
    private static void groupCurlyBackoffTest() throws Exception {
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4682
        if (!"abc1c".matches("(\\w)+1\\1") ||
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4683
            "abc11".matches("(\\w)+1\\1")) {
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4684
            failCount++;
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4685
        }
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4686
        report("GroupCurly backoff");
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4687
    }
f96e3aef2081 8023647: "abc1c".matches("(\\w)+1\\1")) returns false
sherman
parents: 17447
diff changeset
  4688
17447
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4689
    // This test is for 8012646
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4690
    private static void patternAsPredicate() throws Exception {
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4691
        Predicate<String> p = Pattern.compile("[a-z]+").asPredicate();
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4692
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4693
        if (p.test("")) {
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4694
            failCount++;
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4695
        }
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4696
        if (!p.test("word")) {
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4697
            failCount++;
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4698
        }
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4699
        if (p.test("1234")) {
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4700
            failCount++;
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4701
        }
49554
f088ec60bed5 8164781: Pattern.asPredicate specification is incomplete
vtheeyarath
parents: 48853
diff changeset
  4702
        if (!p.test("word1234")) {
f088ec60bed5 8164781: Pattern.asPredicate specification is incomplete
vtheeyarath
parents: 48853
diff changeset
  4703
            failCount++;
f088ec60bed5 8164781: Pattern.asPredicate specification is incomplete
vtheeyarath
parents: 48853
diff changeset
  4704
        }
17447
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4705
        report("Pattern.asPredicate");
3031deba3fbe 8012646: Pattern.splitAsStream
psandoz
parents: 17434
diff changeset
  4706
    }
25523
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4707
49763
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4708
    // This test is for 8184692
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4709
    private static void patternAsMatchPredicate() throws Exception {
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4710
        Predicate<String> p = Pattern.compile("[a-z]+").asMatchPredicate();
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4711
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4712
        if (p.test("")) {
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4713
            failCount++;
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4714
        }
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4715
        if (!p.test("word")) {
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4716
            failCount++;
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4717
        }
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4718
        if (p.test("1234word")) {
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4719
            failCount++;
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4720
        }
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4721
        if (p.test("1234")) {
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4722
            failCount++;
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4723
        }
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4724
        report("Pattern.asMatchPredicate");
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4725
    }
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4726
1ccf5fae9664 8184692: add Pattern.asMatchPredicate
vtheeyarath
parents: 49554
diff changeset
  4727
25523
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4728
    // This test is for 8035975
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4729
    private static void invalidFlags() throws Exception {
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4730
        for (int flag = 1; flag != 0; flag <<= 1) {
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4731
            switch (flag) {
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4732
            case Pattern.CASE_INSENSITIVE:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4733
            case Pattern.MULTILINE:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4734
            case Pattern.DOTALL:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4735
            case Pattern.UNICODE_CASE:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4736
            case Pattern.CANON_EQ:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4737
            case Pattern.UNIX_LINES:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4738
            case Pattern.LITERAL:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4739
            case Pattern.UNICODE_CHARACTER_CLASS:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4740
            case Pattern.COMMENTS:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4741
                // valid flag, continue
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4742
                break;
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4743
            default:
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4744
                try {
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4745
                    Pattern.compile(".", flag);
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4746
                    failCount++;
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4747
                } catch (IllegalArgumentException expected) {
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4748
                }
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4749
            }
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4750
        }
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4751
        report("Invalid compile flags");
c751d1010164 8035975: Pattern.compile(String, int) fails to throw IllegalArgumentException
igerasim
parents: 23734
diff changeset
  4752
    }
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4753
38777
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4754
    // This test is for 8158482
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4755
    private static void embeddedFlags() throws Exception {
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4756
        try {
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4757
            Pattern.compile("(?i).(?-i).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4758
            Pattern.compile("(?m).(?-m).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4759
            Pattern.compile("(?s).(?-s).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4760
            Pattern.compile("(?d).(?-d).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4761
            Pattern.compile("(?u).(?-u).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4762
            Pattern.compile("(?c).(?-c).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4763
            Pattern.compile("(?x).(?-x).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4764
            Pattern.compile("(?U).(?-U).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4765
            Pattern.compile("(?imsducxU).(?-imsducxU).");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4766
        } catch (PatternSyntaxException x) {
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4767
            failCount++;
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4768
        }
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4769
        report("Embedded flags");
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4770
    }
826eb7091523 8158482: regex UNICODE_CHARACTER_CLASS flag cannot be disabled with an embedded flag expression
sherman
parents: 38450
diff changeset
  4771
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4772
    private static void grapheme() throws Exception {
55013
8dae495a59e7 8221431: Support for Unicode 12.1
naoto
parents: 49763
diff changeset
  4773
        Files.lines(UCDFiles.GRAPHEME_BREAK_TEST)
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4774
            .filter( ln -> ln.length() != 0 && !ln.startsWith("#") )
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4775
            .forEach( ln -> {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4776
                    ln = ln.replaceAll("\\s+|\\([a-zA-Z]+\\)|\\[[a-zA-Z]]+\\]|#.*", "");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4777
                    // System.out.println(str);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4778
                    String[] strs = ln.split("\u00f7|\u00d7");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4779
                    StringBuilder src = new StringBuilder();
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4780
                    ArrayList<String> graphemes = new ArrayList<>();
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4781
                    StringBuilder buf = new StringBuilder();
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4782
                    int offBk = 0;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4783
                    for (String str : strs) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4784
                        if (str.length() == 0)  // first empty str
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4785
                            continue;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4786
                        int cp = Integer.parseInt(str, 16);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4787
                        src.appendCodePoint(cp);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4788
                        buf.appendCodePoint(cp);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4789
                        offBk += (str.length() + 1);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4790
                        if (ln.charAt(offBk) == '\u00f7') {    // DIV
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4791
                            graphemes.add(buf.toString());
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4792
                            buf = new StringBuilder();
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4793
                        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4794
                    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4795
                    Pattern p = Pattern.compile("\\X");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4796
                    Matcher m = p.matcher(src.toString());
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4797
                    Scanner s = new Scanner(src.toString()).useDelimiter("\\b{g}");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4798
                    for (String g : graphemes) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4799
                        // System.out.printf("     grapheme:=[%s]%n", g);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4800
                        // (1) test \\X directly
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4801
                        if (!m.find() || !m.group().equals(g)) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4802
                            System.out.println("Failed \\X [" + ln + "] : " + g);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4803
                            failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4804
                        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4805
                        // (2) test \\b{g} + \\X  via Scanner
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4806
                        boolean hasNext = s.hasNext(p);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4807
                        // if (!s.hasNext() || !s.next().equals(next)) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4808
                        if (!s.hasNext(p) || !s.next(p).equals(g)) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4809
                            System.out.println("Failed b{g} [" + ln + "] : " + g);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4810
                            failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4811
                        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4812
                    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4813
                });
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4814
        // some sanity checks
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4815
        if (!Pattern.compile("\\X{10}").matcher("abcdefghij").matches() ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4816
            !Pattern.compile("\\b{g}(?:\\X\\b{g}){5}\\b{g}").matcher("abcde").matches() ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4817
            !Pattern.compile("(?:\\X\\b{g}){2}").matcher("\ud800\udc00\ud801\udc02").matches())
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4818
            failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4819
        // make sure "\b{n}" still works
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4820
        if (!Pattern.compile("\\b{1}hello\\b{1} \\b{1}world\\b{1}").matcher("hello world").matches())
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4821
            failCount++;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4822
        report("Unicode extended grapheme cluster");
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents: 34436
diff changeset
  4823
    }
37882
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4824
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4825
    // hangup/timeout if go into exponential backtracking
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4826
    private static void expoBacktracking() throws Exception {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4827
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4828
        Object[][] patternMatchers = {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4829
            // 6328855
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4830
            { "(.*\n*)*",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4831
              "this little fine string lets\r\njava.lang.String.matches\r\ncrash\r\n(We don't know why but adding \r* to the regex makes it work again)",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4832
              false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4833
            // 6192895
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4834
            { " *([a-zA-Z0-9/\\-\\?:\\(\\)\\.,'\\+\\{\\}]+ *)+",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4835
              "Hello World this is a test this is a test this is a test A",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4836
              true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4837
            { " *([a-zA-Z0-9/\\-\\?:\\(\\)\\.,'\\+\\{\\}]+ *)+",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4838
              "Hello World this is a test this is a test this is a test \u4e00 ",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4839
              false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4840
            { " *([a-z0-9]+ *)+",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4841
              "hello world this is a test this is a test this is a test A",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4842
              false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4843
            // 4771934 [FIXED] #5013651?
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4844
            { "^(\\w+([\\.-]?\\w+)*@\\w+([\\.-]?\\w+)*(\\.\\w{2,4})+[,;]?)+$",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4845
              "abc@efg.abc,efg@abc.abc,abc@xyz.mno;abc@sdfsd.com",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4846
              true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4847
            // 4866249 [FIXED]
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4848
            { "<\\s*" + "(meta|META)" + "(\\s|[^>])+" + "(CHARSET|charset)=" + "(\\s|[^>])+>",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4849
              "<META http-equiv=\"Content-Type\" content=\"text/html; charset=ISO-8859-5\">",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4850
              true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4851
            { "^(\\w+([\\.-]?\\w+)*@\\w+([\\.-]?\\w+)*(\\.\\w{2,4})+[,;]?)+$",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4852
              "abc@efg.abc,efg@abc.abc,abc@xyz.mno;sdfsd.com",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4853
              false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4854
            // 6345469
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4855
            { "((<[^>]+>)?(((\\s)?)*(\\&nbsp;)?)*((\\s)?)*)+",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4856
              "&nbsp;&nbsp; < br/> &nbsp; < / p> <p> <html> <adfasfdasdf>&nbsp; </p>",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4857
              true }, // --> matched
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4858
            { "((<[^>]+>)?(((\\s)?)*(\\&nbsp;)?)*((\\s)?)*)+",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4859
              "&nbsp;&nbsp; < br/> &nbsp; < / p> <p> <html> <adfasfdasdf>&nbsp; p </p>",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4860
              false },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4861
            // 5026912
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4862
            { "^\\s*" + "(\\w|\\d|[\\xC0-\\xFF]|/)+" + "\\s+|$",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4863
              "156580451111112225588087755221111111566969655555555",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4864
              false},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4865
            // 6988218
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4866
            { "^([+-]?((0[xX](\\p{XDigit}+))|(((\\p{Digit}+)(\\.)?((\\p{Digit}+)?)([eE][+-]?(\\p{Digit}+))?)|(\\.((\\p{Digit}+))([eE][+-]?(\\p{Digit}+))?)))|[n|N]?'([^']*(?:'')*[^']*)*')",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4867
              "'%)) order by ANGEBOT.ID",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4868
              false},    // find
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4869
            // 6693451
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4870
            { "^(\\s*foo\\s*)*$",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4871
              "foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4872
              true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4873
            { "^(\\s*foo\\s*)*$",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4874
              "foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo fo",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4875
              false
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4876
            },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4877
            // 7006761
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4878
            { "(([0-9A-Z]+)([_]?+)*)*", "FOOOOO_BAAAR_FOOOOOOOOO_BA_", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4879
            { "(([0-9A-Z]+)([_]?+)*)*", "FOOOOO_BAAAR_FOOOOOOOOO_BA_ ", false},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4880
            // 8140212
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4881
            { "(?<before>.*)\\{(?<reflection>\\w+):(?<innerMethod>\\w+(\\.?\\w+(\\(((?<args>(('[^']*')|((/|\\w)+))(,(('[^']*')|((/|\\w)+)))*))?\\))?)*)\\}(?<after>.*)",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4882
              "{CeGlobal:getSodCutoff.getGui.getAmqp.getSimpleModeEnabled()",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4883
              false
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4884
            },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4885
            { "^(a+)+$", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4886
            { "^(a+)+$", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!", false},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4887
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4888
            { "(x+)*y",  "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxy", true },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4889
            { "(x+)*y",  "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxz", false},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4890
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4891
            { "(x+x+)+y", "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxy", true},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4892
            { "(x+x+)+y", "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxz", false},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4893
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4894
            { "(([0-9A-Z]+)([_]?+)*)*", "--------------------------------------", false},
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4895
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4896
            /* not fixed
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4897
            //8132141   --->    second level exponential backtracking
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4898
            { "(h|h|ih(((i|a|c|c|a|i|i|j|b|a|i|b|a|a|j))+h)ahbfhba|c|i)*",
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4899
              "hchcchicihcchciiicichhcichcihcchiihichiciiiihhcchicchhcihchcihiihciichhccciccichcichiihcchcihhicchcciicchcccihiiihhihihihichicihhcciccchihhhcchichchciihiicihciihcccciciccicciiiiiiiiicihhhiiiihchccchchhhhiiihchihcccchhhiiiiiiiicicichicihcciciihichhhhchihciiihhiccccccciciihhichiccchhicchicihihccichicciihcichccihhiciccccccccichhhhihihhcchchihihiihhihihihicichihiiiihhhhihhhchhichiicihhiiiiihchccccchichci" },
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4900
            */
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4901
        };
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4902
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4903
        for (Object[] pm : patternMatchers) {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4904
            String p = (String)pm[0];
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4905
            String s = (String)pm[1];
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4906
            boolean r = (Boolean)pm[2];
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4907
            if (r != Pattern.compile(p).matcher(s).matches()) {
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4908
                failCount++;
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4909
            }
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4910
        }
e7f3cf12e739 6328855: String: Matches hangs at short and easy Strings containing \r \n
sherman
parents: 35783
diff changeset
  4911
    }
48853
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4912
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4913
    private static void invalidGroupName() {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4914
        // Invalid start of a group name
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4915
        for (String groupName : List.of("", ".", "0", "\u0040", "\u005b",
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4916
                "\u0060", "\u007b", "\u0416")) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4917
            for (String pat : List.of("(?<" + groupName + ">)",
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4918
                    "\\k<" + groupName + ">")) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4919
                try {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4920
                    Pattern.compile(pat);
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4921
                    failCount++;
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4922
                } catch (PatternSyntaxException e) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4923
                    if (!e.getMessage().startsWith(
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4924
                            "capturing group name does not start with a"
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4925
                            + " Latin letter")) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4926
                        failCount++;
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4927
                    }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4928
                }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4929
            }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4930
        }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4931
        // Invalid char in a group name
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4932
        for (String groupName : List.of("a.", "b\u0040", "c\u005b",
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4933
                "d\u0060", "e\u007b", "f\u0416")) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4934
            for (String pat : List.of("(?<" + groupName + ">)",
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4935
                    "\\k<" + groupName + ">")) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4936
                try {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4937
                    Pattern.compile(pat);
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4938
                    failCount++;
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4939
                } catch (PatternSyntaxException e) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4940
                    if (!e.getMessage().startsWith(
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4941
                            "named capturing group is missing trailing '>'")) {
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4942
                        failCount++;
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4943
                    }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4944
                }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4945
            }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4946
        }
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4947
        report("Invalid capturing group names");
84b4ffbba8b0 8197462: Inconsistent exception messages for invalid capturing group names
igerasim
parents: 48491
diff changeset
  4948
    }
55096
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4949
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4950
    private static void illegalRepetitionRange() {
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4951
        // huge integers > (2^31 - 1)
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4952
        String n = BigInteger.valueOf(1L << 32)
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4953
            .toString();
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4954
        String m = BigInteger.valueOf(1L << 31)
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4955
            .add(new BigInteger(80, generator))
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4956
            .toString();
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4957
        for (String rep : List.of("", "x", ".", ",", "-1", "2,1",
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4958
                n, n + ",", "0," + n, n + "," + m, m, m + ",", "0," + m)) {
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4959
            String pat = ".{" + rep + "}";
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4960
            try {
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4961
                Pattern.compile(pat);
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4962
                failCount++;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4963
                System.out.println("Expected to fail. Pattern: " + pat);
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4964
            } catch (PatternSyntaxException e) {
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4965
                if (!e.getMessage().startsWith("Illegal repetition")) {
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4966
                    failCount++;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4967
                    System.out.println("Unexpected error message: " + e.getMessage());
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4968
                }
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4969
            } catch (Throwable t) {
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4970
                failCount++;
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4971
                System.out.println("Unexpected exception: " + t);
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4972
            }
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4973
        }
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4974
        report("illegalRepetitionRange");
234673929e0a 8224789: Parsing repetition count in regex does not detect numeric overflow
igerasim
parents: 55013
diff changeset
  4975
    }
2070
6e9972fbd965 6350801: Add support for named (instead of numbered) capture groups in regular expression
sherman
parents:
diff changeset
  4976
}