src/java.base/share/classes/java/util/regex/Grapheme.java
author erikj
Tue, 12 Sep 2017 19:03:39 +0200
changeset 47216 71c04702a3d5
parent 35783 jdk/src/java.base/share/classes/java/util/regex/Grapheme.java@2690535d72cc
child 55013 8dae495a59e7
permissions -rw-r--r--
8187443: Forest Consolidation: Move files to unified layout Reviewed-by: darcy, ihse
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
35783
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     1
/*
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     2
 * Copyright (c) 2016, Oracle and/or its affiliates. All rights reserved.
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     3
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     4
 *
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     5
 * This code is free software; you can redistribute it and/or modify it
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     6
 * under the terms of the GNU General Public License version 2 only, as
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     7
 * published by the Free Software Foundation.  Oracle designates this
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     8
 * particular file as subject to the "Classpath" exception as provided
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
     9
 * by Oracle in the LICENSE file that accompanied this code.
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    10
 *
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    11
 * This code is distributed in the hope that it will be useful, but WITHOUT
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    12
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    13
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    14
 * version 2 for more details (a copy is included in the LICENSE file that
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    15
 * accompanied this code).
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    16
 *
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    17
 * You should have received a copy of the GNU General Public License version
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    18
 * 2 along with this work; if not, write to the Free Software Foundation,
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    19
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    20
 *
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    21
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    22
 * or visit www.oracle.com if you need additional information or have any
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    23
 * questions.
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    24
 */
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    25
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    26
package java.util.regex;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    27
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    28
final class Grapheme {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    29
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    30
    /**
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    31
     * Determines if there is an extended  grapheme cluster boundary between two
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    32
     * continuing characters {@code cp1} and {@code cp2}.
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    33
     * <p>
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    34
     * See Unicode Standard Annex #29 Unicode Text Segmentation for the specification
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    35
     * for the extended grapheme cluster boundary rules
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    36
     */
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    37
    static boolean isBoundary(int cp1, int cp2) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    38
        return rules[getType(cp1)][getType(cp2)];
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    39
    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    40
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    41
    // types
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    42
    private static final int OTHER = 0;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    43
    private static final int CR = 1;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    44
    private static final int LF = 2;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    45
    private static final int CONTROL = 3;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    46
    private static final int EXTEND = 4;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    47
    private static final int RI = 5;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    48
    private static final int PREPEND = 6;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    49
    private static final int SPACINGMARK = 7;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    50
    private static final int L = 8;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    51
    private static final int V = 9;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    52
    private static final int T = 10;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    53
    private static final int LV = 11;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    54
    private static final int LVT = 12;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    55
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    56
    private static final int FIRST_TYPE = 0;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    57
    private static final int LAST_TYPE = 12;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    58
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    59
    private static boolean[][] rules;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    60
    static {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    61
        rules = new boolean[LAST_TYPE + 1][LAST_TYPE + 1];
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    62
        // default, any + any
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    63
        for (int i = FIRST_TYPE; i <= LAST_TYPE; i++)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    64
            for (int j = FIRST_TYPE; j <= LAST_TYPE; j++)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    65
                rules[i][j] = true;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    66
        // GB 6 L x (L | V | LV | VT)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    67
        rules[L][L] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    68
        rules[L][V] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    69
        rules[L][LV] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    70
        rules[L][LVT] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    71
        // GB 7 (LV | V) x (V | T)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    72
        rules[LV][V] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    73
        rules[LV][T] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    74
        rules[V][V] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    75
        rules[V][T] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    76
        // GB 8 (LVT | T) x T
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    77
        rules[LVT][T] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    78
        rules[T][T] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    79
        // GB 8a RI x RI
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    80
        rules[RI][RI] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    81
        // GB 9 x Extend
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    82
        // GB 9a x Spacing Mark
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    83
        // GB 9b Prepend x
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    84
        for (int i = FIRST_TYPE; i <= LAST_TYPE; i++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    85
            rules[i][EXTEND] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    86
            rules[i][SPACINGMARK] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    87
            rules[PREPEND][i] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    88
        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    89
        // GB 4  (Control | CR | LF) +
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    90
        // GB 5  + (Control | CR | LF)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    91
        for (int i = FIRST_TYPE; i <= LAST_TYPE; i++)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    92
            for (int j = CR; j <= CONTROL; j++) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    93
                rules[i][j] = true;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    94
                rules[j][i] = true;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    95
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    96
        // GB 3 CR x LF
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    97
        rules[CR][LF] = false;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    98
        // GB 10 Any + Any  -> default
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
    99
    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   100
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   101
    // Hangul syllables
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   102
    private static final int SYLLABLE_BASE = 0xAC00;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   103
    private static final int LCOUNT = 19;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   104
    private static final int VCOUNT = 21;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   105
    private static final int TCOUNT = 28;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   106
    private static final int NCOUNT = VCOUNT * TCOUNT; // 588
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   107
    private static final int SCOUNT = LCOUNT * NCOUNT; // 11172
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   108
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   109
    // #tr29: SpacingMark exceptions: The following (which have
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   110
    // General_Category = Spacing_Mark and would otherwise be included)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   111
    // are specifically excluded
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   112
    private static boolean isExcludedSpacingMark(int cp) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   113
       return  cp == 0x102B || cp == 0x102C || cp == 0x1038 ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   114
               cp >= 0x1062 && cp <= 0x1064 ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   115
               cp >= 0x1062 && cp <= 0x106D ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   116
               cp == 0x1083 ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   117
               cp >= 0x1087 && cp <= 0x108C ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   118
               cp == 0x108F ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   119
               cp >= 0x109A && cp <= 0x109C ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   120
               cp == 0x1A61 || cp == 0x1A63 || cp == 0x1A64 ||
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   121
               cp == 0xAA7B || cp == 0xAA7D;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   122
    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   123
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   124
    @SuppressWarnings("fallthrough")
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   125
    private static int getType(int cp) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   126
        int type = Character.getType(cp);
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   127
        switch(type) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   128
        case Character.CONTROL:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   129
            if (cp == 0x000D)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   130
                return CR;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   131
            if (cp == 0x000A)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   132
                return LF;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   133
            return CONTROL;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   134
         case Character.UNASSIGNED:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   135
            // NOTE: #tr29 lists "Unassigned and Default_Ignorable_Code_Point" as Control
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   136
            // but GraphemeBreakTest.txt lists u+0378/reserved-0378 as "Other"
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   137
            // so type it as "Other" to make the test happy
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   138
             if (cp == 0x0378)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   139
                 return OTHER;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   140
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   141
        case Character.LINE_SEPARATOR:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   142
        case Character.PARAGRAPH_SEPARATOR:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   143
        case Character.SURROGATE:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   144
            return CONTROL;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   145
        case Character.FORMAT:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   146
            if (cp == 0x200C || cp == 0x200D)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   147
                return EXTEND;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   148
            return CONTROL;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   149
        case Character.NON_SPACING_MARK:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   150
        case Character.ENCLOSING_MARK:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   151
             // NOTE:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   152
             // #tr29 "plus a few General_Category = Spacing_Mark needed for
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   153
             // canonical equivalence."
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   154
             // but for "extended grapheme clusters" support, there is no
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   155
             // need actually to diff "extend" and "spackmark" given GB9, GB9a
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   156
             return EXTEND;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   157
        case  Character.COMBINING_SPACING_MARK:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   158
            if (isExcludedSpacingMark(cp))
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   159
                return OTHER;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   160
            // NOTE:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   161
            // 0x11720 and 0x11721 are mentioned in #tr29 as
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   162
            // OTHER_LETTER but it appears their category has been updated to
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   163
            // COMBING_SPACING_MARK already (verified in ver.8)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   164
            return SPACINGMARK;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   165
        case Character.OTHER_SYMBOL:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   166
            if (cp >= 0x1F1E6 && cp <= 0x1F1FF)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   167
                return RI;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   168
            return OTHER;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   169
        case Character.MODIFIER_LETTER:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   170
            // WARNING:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   171
            // not mentioned in #tr29 but listed in GraphemeBreakProperty.txt
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   172
            if (cp == 0xFF9E || cp == 0xFF9F)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   173
                return EXTEND;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   174
            return OTHER;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   175
        case Character.OTHER_LETTER:
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   176
            if (cp == 0x0E33 || cp == 0x0EB3)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   177
                return SPACINGMARK;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   178
            // hangul jamo
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   179
            if (cp >= 0x1100 && cp <= 0x11FF) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   180
                if (cp <= 0x115F)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   181
                    return L;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   182
                if (cp <= 0x11A7)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   183
                    return V;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   184
                return T;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   185
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   186
            // hangul syllables
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   187
            int sindex = cp - SYLLABLE_BASE;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   188
            if (sindex >= 0 && sindex < SCOUNT) {
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   189
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   190
                if (sindex % TCOUNT == 0)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   191
                    return LV;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   192
                return LVT;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   193
            }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   194
            //  hangul jamo_extended A
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   195
            if (cp >= 0xA960 && cp <= 0xA97C)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   196
                return L;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   197
            //  hangul jamo_extended B
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   198
            if (cp >= 0xD7B0 && cp <= 0xD7C6)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   199
                return V;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   200
            if (cp >= 0xD7CB && cp <= 0xD7FB)
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   201
                return T;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   202
        }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   203
        return OTHER;
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   204
    }
2690535d72cc 7071819: To support Extended Grapheme Clusters in Regex
sherman
parents:
diff changeset
   205
}