rt/emul/compact/src/main/java/java/util/regex/ASCII.java
author Jaroslav Tulach <jtulach@netbeans.org>
Mon, 07 Oct 2013 16:13:27 +0200
branchjdk7-b147
changeset 1348 bca65655b36b
permissions -rw-r--r--
Adding RegEx implementation
jtulach@1348
     1
/*
jtulach@1348
     2
 * Copyright (c) 1999, 2000, Oracle and/or its affiliates. All rights reserved.
jtulach@1348
     3
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
jtulach@1348
     4
 *
jtulach@1348
     5
 * This code is free software; you can redistribute it and/or modify it
jtulach@1348
     6
 * under the terms of the GNU General Public License version 2 only, as
jtulach@1348
     7
 * published by the Free Software Foundation.  Oracle designates this
jtulach@1348
     8
 * particular file as subject to the "Classpath" exception as provided
jtulach@1348
     9
 * by Oracle in the LICENSE file that accompanied this code.
jtulach@1348
    10
 *
jtulach@1348
    11
 * This code is distributed in the hope that it will be useful, but WITHOUT
jtulach@1348
    12
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
jtulach@1348
    13
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
jtulach@1348
    14
 * version 2 for more details (a copy is included in the LICENSE file that
jtulach@1348
    15
 * accompanied this code).
jtulach@1348
    16
 *
jtulach@1348
    17
 * You should have received a copy of the GNU General Public License version
jtulach@1348
    18
 * 2 along with this work; if not, write to the Free Software Foundation,
jtulach@1348
    19
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
jtulach@1348
    20
 *
jtulach@1348
    21
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
jtulach@1348
    22
 * or visit www.oracle.com if you need additional information or have any
jtulach@1348
    23
 * questions.
jtulach@1348
    24
 */
jtulach@1348
    25
jtulach@1348
    26
package java.util.regex;
jtulach@1348
    27
jtulach@1348
    28
jtulach@1348
    29
/**
jtulach@1348
    30
 * Utility class that implements the standard C ctype functionality.
jtulach@1348
    31
 *
jtulach@1348
    32
 * @author Hong Zhang
jtulach@1348
    33
 */
jtulach@1348
    34
jtulach@1348
    35
final class ASCII {
jtulach@1348
    36
jtulach@1348
    37
    static final int UPPER   = 0x00000100;
jtulach@1348
    38
jtulach@1348
    39
    static final int LOWER   = 0x00000200;
jtulach@1348
    40
jtulach@1348
    41
    static final int DIGIT   = 0x00000400;
jtulach@1348
    42
jtulach@1348
    43
    static final int SPACE   = 0x00000800;
jtulach@1348
    44
jtulach@1348
    45
    static final int PUNCT   = 0x00001000;
jtulach@1348
    46
jtulach@1348
    47
    static final int CNTRL   = 0x00002000;
jtulach@1348
    48
jtulach@1348
    49
    static final int BLANK   = 0x00004000;
jtulach@1348
    50
jtulach@1348
    51
    static final int HEX     = 0x00008000;
jtulach@1348
    52
jtulach@1348
    53
    static final int UNDER   = 0x00010000;
jtulach@1348
    54
jtulach@1348
    55
    static final int ASCII   = 0x0000FF00;
jtulach@1348
    56
jtulach@1348
    57
    static final int ALPHA   = (UPPER|LOWER);
jtulach@1348
    58
jtulach@1348
    59
    static final int ALNUM   = (UPPER|LOWER|DIGIT);
jtulach@1348
    60
jtulach@1348
    61
    static final int GRAPH   = (PUNCT|UPPER|LOWER|DIGIT);
jtulach@1348
    62
jtulach@1348
    63
    static final int WORD    = (UPPER|LOWER|UNDER|DIGIT);
jtulach@1348
    64
jtulach@1348
    65
    static final int XDIGIT  = (HEX);
jtulach@1348
    66
jtulach@1348
    67
    private static final int[] ctype = new int[] {
jtulach@1348
    68
        CNTRL,                  /* 00 (NUL) */
jtulach@1348
    69
        CNTRL,                  /* 01 (SOH) */
jtulach@1348
    70
        CNTRL,                  /* 02 (STX) */
jtulach@1348
    71
        CNTRL,                  /* 03 (ETX) */
jtulach@1348
    72
        CNTRL,                  /* 04 (EOT) */
jtulach@1348
    73
        CNTRL,                  /* 05 (ENQ) */
jtulach@1348
    74
        CNTRL,                  /* 06 (ACK) */
jtulach@1348
    75
        CNTRL,                  /* 07 (BEL) */
jtulach@1348
    76
        CNTRL,                  /* 08 (BS)  */
jtulach@1348
    77
        SPACE+CNTRL+BLANK,      /* 09 (HT)  */
jtulach@1348
    78
        SPACE+CNTRL,            /* 0A (LF)  */
jtulach@1348
    79
        SPACE+CNTRL,            /* 0B (VT)  */
jtulach@1348
    80
        SPACE+CNTRL,            /* 0C (FF)  */
jtulach@1348
    81
        SPACE+CNTRL,            /* 0D (CR)  */
jtulach@1348
    82
        CNTRL,                  /* 0E (SI)  */
jtulach@1348
    83
        CNTRL,                  /* 0F (SO)  */
jtulach@1348
    84
        CNTRL,                  /* 10 (DLE) */
jtulach@1348
    85
        CNTRL,                  /* 11 (DC1) */
jtulach@1348
    86
        CNTRL,                  /* 12 (DC2) */
jtulach@1348
    87
        CNTRL,                  /* 13 (DC3) */
jtulach@1348
    88
        CNTRL,                  /* 14 (DC4) */
jtulach@1348
    89
        CNTRL,                  /* 15 (NAK) */
jtulach@1348
    90
        CNTRL,                  /* 16 (SYN) */
jtulach@1348
    91
        CNTRL,                  /* 17 (ETB) */
jtulach@1348
    92
        CNTRL,                  /* 18 (CAN) */
jtulach@1348
    93
        CNTRL,                  /* 19 (EM)  */
jtulach@1348
    94
        CNTRL,                  /* 1A (SUB) */
jtulach@1348
    95
        CNTRL,                  /* 1B (ESC) */
jtulach@1348
    96
        CNTRL,                  /* 1C (FS)  */
jtulach@1348
    97
        CNTRL,                  /* 1D (GS)  */
jtulach@1348
    98
        CNTRL,                  /* 1E (RS)  */
jtulach@1348
    99
        CNTRL,                  /* 1F (US)  */
jtulach@1348
   100
        SPACE+BLANK,            /* 20 SPACE */
jtulach@1348
   101
        PUNCT,                  /* 21 !     */
jtulach@1348
   102
        PUNCT,                  /* 22 "     */
jtulach@1348
   103
        PUNCT,                  /* 23 #     */
jtulach@1348
   104
        PUNCT,                  /* 24 $     */
jtulach@1348
   105
        PUNCT,                  /* 25 %     */
jtulach@1348
   106
        PUNCT,                  /* 26 &     */
jtulach@1348
   107
        PUNCT,                  /* 27 '     */
jtulach@1348
   108
        PUNCT,                  /* 28 (     */
jtulach@1348
   109
        PUNCT,                  /* 29 )     */
jtulach@1348
   110
        PUNCT,                  /* 2A *     */
jtulach@1348
   111
        PUNCT,                  /* 2B +     */
jtulach@1348
   112
        PUNCT,                  /* 2C ,     */
jtulach@1348
   113
        PUNCT,                  /* 2D -     */
jtulach@1348
   114
        PUNCT,                  /* 2E .     */
jtulach@1348
   115
        PUNCT,                  /* 2F /     */
jtulach@1348
   116
        DIGIT+HEX+0,            /* 30 0     */
jtulach@1348
   117
        DIGIT+HEX+1,            /* 31 1     */
jtulach@1348
   118
        DIGIT+HEX+2,            /* 32 2     */
jtulach@1348
   119
        DIGIT+HEX+3,            /* 33 3     */
jtulach@1348
   120
        DIGIT+HEX+4,            /* 34 4     */
jtulach@1348
   121
        DIGIT+HEX+5,            /* 35 5     */
jtulach@1348
   122
        DIGIT+HEX+6,            /* 36 6     */
jtulach@1348
   123
        DIGIT+HEX+7,            /* 37 7     */
jtulach@1348
   124
        DIGIT+HEX+8,            /* 38 8     */
jtulach@1348
   125
        DIGIT+HEX+9,            /* 39 9     */
jtulach@1348
   126
        PUNCT,                  /* 3A :     */
jtulach@1348
   127
        PUNCT,                  /* 3B ;     */
jtulach@1348
   128
        PUNCT,                  /* 3C <     */
jtulach@1348
   129
        PUNCT,                  /* 3D =     */
jtulach@1348
   130
        PUNCT,                  /* 3E >     */
jtulach@1348
   131
        PUNCT,                  /* 3F ?     */
jtulach@1348
   132
        PUNCT,                  /* 40 @     */
jtulach@1348
   133
        UPPER+HEX+10,           /* 41 A     */
jtulach@1348
   134
        UPPER+HEX+11,           /* 42 B     */
jtulach@1348
   135
        UPPER+HEX+12,           /* 43 C     */
jtulach@1348
   136
        UPPER+HEX+13,           /* 44 D     */
jtulach@1348
   137
        UPPER+HEX+14,           /* 45 E     */
jtulach@1348
   138
        UPPER+HEX+15,           /* 46 F     */
jtulach@1348
   139
        UPPER+16,               /* 47 G     */
jtulach@1348
   140
        UPPER+17,               /* 48 H     */
jtulach@1348
   141
        UPPER+18,               /* 49 I     */
jtulach@1348
   142
        UPPER+19,               /* 4A J     */
jtulach@1348
   143
        UPPER+20,               /* 4B K     */
jtulach@1348
   144
        UPPER+21,               /* 4C L     */
jtulach@1348
   145
        UPPER+22,               /* 4D M     */
jtulach@1348
   146
        UPPER+23,               /* 4E N     */
jtulach@1348
   147
        UPPER+24,               /* 4F O     */
jtulach@1348
   148
        UPPER+25,               /* 50 P     */
jtulach@1348
   149
        UPPER+26,               /* 51 Q     */
jtulach@1348
   150
        UPPER+27,               /* 52 R     */
jtulach@1348
   151
        UPPER+28,               /* 53 S     */
jtulach@1348
   152
        UPPER+29,               /* 54 T     */
jtulach@1348
   153
        UPPER+30,               /* 55 U     */
jtulach@1348
   154
        UPPER+31,               /* 56 V     */
jtulach@1348
   155
        UPPER+32,               /* 57 W     */
jtulach@1348
   156
        UPPER+33,               /* 58 X     */
jtulach@1348
   157
        UPPER+34,               /* 59 Y     */
jtulach@1348
   158
        UPPER+35,               /* 5A Z     */
jtulach@1348
   159
        PUNCT,                  /* 5B [     */
jtulach@1348
   160
        PUNCT,                  /* 5C \     */
jtulach@1348
   161
        PUNCT,                  /* 5D ]     */
jtulach@1348
   162
        PUNCT,                  /* 5E ^     */
jtulach@1348
   163
        PUNCT|UNDER,            /* 5F _     */
jtulach@1348
   164
        PUNCT,                  /* 60 `     */
jtulach@1348
   165
        LOWER+HEX+10,           /* 61 a     */
jtulach@1348
   166
        LOWER+HEX+11,           /* 62 b     */
jtulach@1348
   167
        LOWER+HEX+12,           /* 63 c     */
jtulach@1348
   168
        LOWER+HEX+13,           /* 64 d     */
jtulach@1348
   169
        LOWER+HEX+14,           /* 65 e     */
jtulach@1348
   170
        LOWER+HEX+15,           /* 66 f     */
jtulach@1348
   171
        LOWER+16,               /* 67 g     */
jtulach@1348
   172
        LOWER+17,               /* 68 h     */
jtulach@1348
   173
        LOWER+18,               /* 69 i     */
jtulach@1348
   174
        LOWER+19,               /* 6A j     */
jtulach@1348
   175
        LOWER+20,               /* 6B k     */
jtulach@1348
   176
        LOWER+21,               /* 6C l     */
jtulach@1348
   177
        LOWER+22,               /* 6D m     */
jtulach@1348
   178
        LOWER+23,               /* 6E n     */
jtulach@1348
   179
        LOWER+24,               /* 6F o     */
jtulach@1348
   180
        LOWER+25,               /* 70 p     */
jtulach@1348
   181
        LOWER+26,               /* 71 q     */
jtulach@1348
   182
        LOWER+27,               /* 72 r     */
jtulach@1348
   183
        LOWER+28,               /* 73 s     */
jtulach@1348
   184
        LOWER+29,               /* 74 t     */
jtulach@1348
   185
        LOWER+30,               /* 75 u     */
jtulach@1348
   186
        LOWER+31,               /* 76 v     */
jtulach@1348
   187
        LOWER+32,               /* 77 w     */
jtulach@1348
   188
        LOWER+33,               /* 78 x     */
jtulach@1348
   189
        LOWER+34,               /* 79 y     */
jtulach@1348
   190
        LOWER+35,               /* 7A z     */
jtulach@1348
   191
        PUNCT,                  /* 7B {     */
jtulach@1348
   192
        PUNCT,                  /* 7C |     */
jtulach@1348
   193
        PUNCT,                  /* 7D }     */
jtulach@1348
   194
        PUNCT,                  /* 7E ~     */
jtulach@1348
   195
        CNTRL,                  /* 7F (DEL) */
jtulach@1348
   196
    };
jtulach@1348
   197
jtulach@1348
   198
    static int getType(int ch) {
jtulach@1348
   199
        return ((ch & 0xFFFFFF80) == 0 ? ctype[ch] : 0);
jtulach@1348
   200
    }
jtulach@1348
   201
jtulach@1348
   202
    static boolean isType(int ch, int type) {
jtulach@1348
   203
        return (getType(ch) & type) != 0;
jtulach@1348
   204
    }
jtulach@1348
   205
jtulach@1348
   206
    static boolean isAscii(int ch) {
jtulach@1348
   207
        return ((ch & 0xFFFFFF80) == 0);
jtulach@1348
   208
    }
jtulach@1348
   209
jtulach@1348
   210
    static boolean isAlpha(int ch) {
jtulach@1348
   211
        return isType(ch, ALPHA);
jtulach@1348
   212
    }
jtulach@1348
   213
jtulach@1348
   214
    static boolean isDigit(int ch) {
jtulach@1348
   215
        return ((ch-'0')|('9'-ch)) >= 0;
jtulach@1348
   216
    }
jtulach@1348
   217
jtulach@1348
   218
    static boolean isAlnum(int ch) {
jtulach@1348
   219
        return isType(ch, ALNUM);
jtulach@1348
   220
    }
jtulach@1348
   221
jtulach@1348
   222
    static boolean isGraph(int ch) {
jtulach@1348
   223
        return isType(ch, GRAPH);
jtulach@1348
   224
    }
jtulach@1348
   225
jtulach@1348
   226
    static boolean isPrint(int ch) {
jtulach@1348
   227
        return ((ch-0x20)|(0x7E-ch)) >= 0;
jtulach@1348
   228
    }
jtulach@1348
   229
jtulach@1348
   230
    static boolean isPunct(int ch) {
jtulach@1348
   231
        return isType(ch, PUNCT);
jtulach@1348
   232
    }
jtulach@1348
   233
jtulach@1348
   234
    static boolean isSpace(int ch) {
jtulach@1348
   235
        return isType(ch, SPACE);
jtulach@1348
   236
    }
jtulach@1348
   237
jtulach@1348
   238
    static boolean isHexDigit(int ch) {
jtulach@1348
   239
        return isType(ch, HEX);
jtulach@1348
   240
    }
jtulach@1348
   241
jtulach@1348
   242
    static boolean isOctDigit(int ch) {
jtulach@1348
   243
        return ((ch-'0')|('7'-ch)) >= 0;
jtulach@1348
   244
    }
jtulach@1348
   245
jtulach@1348
   246
    static boolean isCntrl(int ch) {
jtulach@1348
   247
        return isType(ch, CNTRL);
jtulach@1348
   248
    }
jtulach@1348
   249
jtulach@1348
   250
    static boolean isLower(int ch) {
jtulach@1348
   251
        return ((ch-'a')|('z'-ch)) >= 0;
jtulach@1348
   252
    }
jtulach@1348
   253
jtulach@1348
   254
    static boolean isUpper(int ch) {
jtulach@1348
   255
        return ((ch-'A')|('Z'-ch)) >= 0;
jtulach@1348
   256
    }
jtulach@1348
   257
jtulach@1348
   258
    static boolean isWord(int ch) {
jtulach@1348
   259
        return isType(ch, WORD);
jtulach@1348
   260
    }
jtulach@1348
   261
jtulach@1348
   262
    static int toDigit(int ch) {
jtulach@1348
   263
        return (ctype[ch & 0x7F] & 0x3F);
jtulach@1348
   264
    }
jtulach@1348
   265
jtulach@1348
   266
    static int toLower(int ch) {
jtulach@1348
   267
        return isUpper(ch) ? (ch + 0x20) : ch;
jtulach@1348
   268
    }
jtulach@1348
   269
jtulach@1348
   270
    static int toUpper(int ch) {
jtulach@1348
   271
        return isLower(ch) ? (ch - 0x20) : ch;
jtulach@1348
   272
    }
jtulach@1348
   273
jtulach@1348
   274
}