J. Duke | 319a3b9 | 2007-12-01 00:00:00 +0000 | [diff] [blame^] | 1 | /* |
| 2 | * Copyright 1999-2007 Sun Microsystems, Inc. All Rights Reserved. |
| 3 | * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. |
| 4 | * |
| 5 | * This code is free software; you can redistribute it and/or modify it |
| 6 | * under the terms of the GNU General Public License version 2 only, as |
| 7 | * published by the Free Software Foundation. Sun designates this |
| 8 | * particular file as subject to the "Classpath" exception as provided |
| 9 | * by Sun in the LICENSE file that accompanied this code. |
| 10 | * |
| 11 | * This code is distributed in the hope that it will be useful, but WITHOUT |
| 12 | * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
| 13 | * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
| 14 | * version 2 for more details (a copy is included in the LICENSE file that |
| 15 | * accompanied this code). |
| 16 | * |
| 17 | * You should have received a copy of the GNU General Public License version |
| 18 | * 2 along with this work; if not, write to the Free Software Foundation, |
| 19 | * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. |
| 20 | * |
| 21 | * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, |
| 22 | * CA 95054 USA or visit www.sun.com if you need additional information or |
| 23 | * have any questions. |
| 24 | */ |
| 25 | |
| 26 | package java.util.regex; |
| 27 | |
| 28 | import java.security.AccessController; |
| 29 | import java.security.PrivilegedAction; |
| 30 | import java.text.CharacterIterator; |
| 31 | import java.text.Normalizer; |
| 32 | import java.util.ArrayList; |
| 33 | import java.util.HashMap; |
| 34 | import java.util.Arrays; |
| 35 | |
| 36 | |
| 37 | /** |
| 38 | * A compiled representation of a regular expression. |
| 39 | * |
| 40 | * <p> A regular expression, specified as a string, must first be compiled into |
| 41 | * an instance of this class. The resulting pattern can then be used to create |
| 42 | * a {@link Matcher} object that can match arbitrary {@link |
| 43 | * java.lang.CharSequence </code>character sequences<code>} against the regular |
| 44 | * expression. All of the state involved in performing a match resides in the |
| 45 | * matcher, so many matchers can share the same pattern. |
| 46 | * |
| 47 | * <p> A typical invocation sequence is thus |
| 48 | * |
| 49 | * <blockquote><pre> |
| 50 | * Pattern p = Pattern.{@link #compile compile}("a*b"); |
| 51 | * Matcher m = p.{@link #matcher matcher}("aaaaab"); |
| 52 | * boolean b = m.{@link Matcher#matches matches}();</pre></blockquote> |
| 53 | * |
| 54 | * <p> A {@link #matches matches} method is defined by this class as a |
| 55 | * convenience for when a regular expression is used just once. This method |
| 56 | * compiles an expression and matches an input sequence against it in a single |
| 57 | * invocation. The statement |
| 58 | * |
| 59 | * <blockquote><pre> |
| 60 | * boolean b = Pattern.matches("a*b", "aaaaab");</pre></blockquote> |
| 61 | * |
| 62 | * is equivalent to the three statements above, though for repeated matches it |
| 63 | * is less efficient since it does not allow the compiled pattern to be reused. |
| 64 | * |
| 65 | * <p> Instances of this class are immutable and are safe for use by multiple |
| 66 | * concurrent threads. Instances of the {@link Matcher} class are not safe for |
| 67 | * such use. |
| 68 | * |
| 69 | * |
| 70 | * <a name="sum"> |
| 71 | * <h4> Summary of regular-expression constructs </h4> |
| 72 | * |
| 73 | * <table border="0" cellpadding="1" cellspacing="0" |
| 74 | * summary="Regular expression constructs, and what they match"> |
| 75 | * |
| 76 | * <tr align="left"> |
| 77 | * <th bgcolor="#CCCCFF" align="left" id="construct">Construct</th> |
| 78 | * <th bgcolor="#CCCCFF" align="left" id="matches">Matches</th> |
| 79 | * </tr> |
| 80 | * |
| 81 | * <tr><th> </th></tr> |
| 82 | * <tr align="left"><th colspan="2" id="characters">Characters</th></tr> |
| 83 | * |
| 84 | * <tr><td valign="top" headers="construct characters"><i>x</i></td> |
| 85 | * <td headers="matches">The character <i>x</i></td></tr> |
| 86 | * <tr><td valign="top" headers="construct characters"><tt>\\</tt></td> |
| 87 | * <td headers="matches">The backslash character</td></tr> |
| 88 | * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>n</i></td> |
| 89 | * <td headers="matches">The character with octal value <tt>0</tt><i>n</i> |
| 90 | * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> |
| 91 | * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>nn</i></td> |
| 92 | * <td headers="matches">The character with octal value <tt>0</tt><i>nn</i> |
| 93 | * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> |
| 94 | * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>mnn</i></td> |
| 95 | * <td headers="matches">The character with octal value <tt>0</tt><i>mnn</i> |
| 96 | * (0 <tt><=</tt> <i>m</i> <tt><=</tt> 3, |
| 97 | * 0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> |
| 98 | * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>hh</i></td> |
| 99 | * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hh</i></td></tr> |
| 100 | * <tr><td valign="top" headers="construct characters"><tt>\u</tt><i>hhhh</i></td> |
| 101 | * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hhhh</i></td></tr> |
| 102 | * <tr><td valign="top" headers="matches"><tt>\t</tt></td> |
| 103 | * <td headers="matches">The tab character (<tt>'\u0009'</tt>)</td></tr> |
| 104 | * <tr><td valign="top" headers="construct characters"><tt>\n</tt></td> |
| 105 | * <td headers="matches">The newline (line feed) character (<tt>'\u000A'</tt>)</td></tr> |
| 106 | * <tr><td valign="top" headers="construct characters"><tt>\r</tt></td> |
| 107 | * <td headers="matches">The carriage-return character (<tt>'\u000D'</tt>)</td></tr> |
| 108 | * <tr><td valign="top" headers="construct characters"><tt>\f</tt></td> |
| 109 | * <td headers="matches">The form-feed character (<tt>'\u000C'</tt>)</td></tr> |
| 110 | * <tr><td valign="top" headers="construct characters"><tt>\a</tt></td> |
| 111 | * <td headers="matches">The alert (bell) character (<tt>'\u0007'</tt>)</td></tr> |
| 112 | * <tr><td valign="top" headers="construct characters"><tt>\e</tt></td> |
| 113 | * <td headers="matches">The escape character (<tt>'\u001B'</tt>)</td></tr> |
| 114 | * <tr><td valign="top" headers="construct characters"><tt>\c</tt><i>x</i></td> |
| 115 | * <td headers="matches">The control character corresponding to <i>x</i></td></tr> |
| 116 | * |
| 117 | * <tr><th> </th></tr> |
| 118 | * <tr align="left"><th colspan="2" id="classes">Character classes</th></tr> |
| 119 | * |
| 120 | * <tr><td valign="top" headers="construct classes"><tt>[abc]</tt></td> |
| 121 | * <td headers="matches"><tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (simple class)</td></tr> |
| 122 | * <tr><td valign="top" headers="construct classes"><tt>[^abc]</tt></td> |
| 123 | * <td headers="matches">Any character except <tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (negation)</td></tr> |
| 124 | * <tr><td valign="top" headers="construct classes"><tt>[a-zA-Z]</tt></td> |
| 125 | * <td headers="matches"><tt>a</tt> through <tt>z</tt> |
| 126 | * or <tt>A</tt> through <tt>Z</tt>, inclusive (range)</td></tr> |
| 127 | * <tr><td valign="top" headers="construct classes"><tt>[a-d[m-p]]</tt></td> |
| 128 | * <td headers="matches"><tt>a</tt> through <tt>d</tt>, |
| 129 | * or <tt>m</tt> through <tt>p</tt>: <tt>[a-dm-p]</tt> (union)</td></tr> |
| 130 | * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[def]]</tt></td> |
| 131 | * <td headers="matches"><tt>d</tt>, <tt>e</tt>, or <tt>f</tt> (intersection)</tr> |
| 132 | * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^bc]]</tt></td> |
| 133 | * <td headers="matches"><tt>a</tt> through <tt>z</tt>, |
| 134 | * except for <tt>b</tt> and <tt>c</tt>: <tt>[ad-z]</tt> (subtraction)</td></tr> |
| 135 | * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^m-p]]</tt></td> |
| 136 | * <td headers="matches"><tt>a</tt> through <tt>z</tt>, |
| 137 | * and not <tt>m</tt> through <tt>p</tt>: <tt>[a-lq-z]</tt>(subtraction)</td></tr> |
| 138 | * <tr><th> </th></tr> |
| 139 | * |
| 140 | * <tr align="left"><th colspan="2" id="predef">Predefined character classes</th></tr> |
| 141 | * |
| 142 | * <tr><td valign="top" headers="construct predef"><tt>.</tt></td> |
| 143 | * <td headers="matches">Any character (may or may not match <a href="#lt">line terminators</a>)</td></tr> |
| 144 | * <tr><td valign="top" headers="construct predef"><tt>\d</tt></td> |
| 145 | * <td headers="matches">A digit: <tt>[0-9]</tt></td></tr> |
| 146 | * <tr><td valign="top" headers="construct predef"><tt>\D</tt></td> |
| 147 | * <td headers="matches">A non-digit: <tt>[^0-9]</tt></td></tr> |
| 148 | * <tr><td valign="top" headers="construct predef"><tt>\s</tt></td> |
| 149 | * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> |
| 150 | * <tr><td valign="top" headers="construct predef"><tt>\S</tt></td> |
| 151 | * <td headers="matches">A non-whitespace character: <tt>[^\s]</tt></td></tr> |
| 152 | * <tr><td valign="top" headers="construct predef"><tt>\w</tt></td> |
| 153 | * <td headers="matches">A word character: <tt>[a-zA-Z_0-9]</tt></td></tr> |
| 154 | * <tr><td valign="top" headers="construct predef"><tt>\W</tt></td> |
| 155 | * <td headers="matches">A non-word character: <tt>[^\w]</tt></td></tr> |
| 156 | * |
| 157 | * <tr><th> </th></tr> |
| 158 | * <tr align="left"><th colspan="2" id="posix">POSIX character classes</b> (US-ASCII only)<b></th></tr> |
| 159 | * |
| 160 | * <tr><td valign="top" headers="construct posix"><tt>\p{Lower}</tt></td> |
| 161 | * <td headers="matches">A lower-case alphabetic character: <tt>[a-z]</tt></td></tr> |
| 162 | * <tr><td valign="top" headers="construct posix"><tt>\p{Upper}</tt></td> |
| 163 | * <td headers="matches">An upper-case alphabetic character:<tt>[A-Z]</tt></td></tr> |
| 164 | * <tr><td valign="top" headers="construct posix"><tt>\p{ASCII}</tt></td> |
| 165 | * <td headers="matches">All ASCII:<tt>[\x00-\x7F]</tt></td></tr> |
| 166 | * <tr><td valign="top" headers="construct posix"><tt>\p{Alpha}</tt></td> |
| 167 | * <td headers="matches">An alphabetic character:<tt>[\p{Lower}\p{Upper}]</tt></td></tr> |
| 168 | * <tr><td valign="top" headers="construct posix"><tt>\p{Digit}</tt></td> |
| 169 | * <td headers="matches">A decimal digit: <tt>[0-9]</tt></td></tr> |
| 170 | * <tr><td valign="top" headers="construct posix"><tt>\p{Alnum}</tt></td> |
| 171 | * <td headers="matches">An alphanumeric character:<tt>[\p{Alpha}\p{Digit}]</tt></td></tr> |
| 172 | * <tr><td valign="top" headers="construct posix"><tt>\p{Punct}</tt></td> |
| 173 | * <td headers="matches">Punctuation: One of <tt>!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~</tt></td></tr> |
| 174 | * <!-- <tt>[\!"#\$%&'\(\)\*\+,\-\./:;\<=\>\?@\[\\\]\^_`\{\|\}~]</tt> |
| 175 | * <tt>[\X21-\X2F\X31-\X40\X5B-\X60\X7B-\X7E]</tt> --> |
| 176 | * <tr><td valign="top" headers="construct posix"><tt>\p{Graph}</tt></td> |
| 177 | * <td headers="matches">A visible character: <tt>[\p{Alnum}\p{Punct}]</tt></td></tr> |
| 178 | * <tr><td valign="top" headers="construct posix"><tt>\p{Print}</tt></td> |
| 179 | * <td headers="matches">A printable character: <tt>[\p{Graph}\x20]</tt></td></tr> |
| 180 | * <tr><td valign="top" headers="construct posix"><tt>\p{Blank}</tt></td> |
| 181 | * <td headers="matches">A space or a tab: <tt>[ \t]</tt></td></tr> |
| 182 | * <tr><td valign="top" headers="construct posix"><tt>\p{Cntrl}</tt></td> |
| 183 | * <td headers="matches">A control character: <tt>[\x00-\x1F\x7F]</tt></td></tr> |
| 184 | * <tr><td valign="top" headers="construct posix"><tt>\p{XDigit}</tt></td> |
| 185 | * <td headers="matches">A hexadecimal digit: <tt>[0-9a-fA-F]</tt></td></tr> |
| 186 | * <tr><td valign="top" headers="construct posix"><tt>\p{Space}</tt></td> |
| 187 | * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> |
| 188 | * |
| 189 | * <tr><th> </th></tr> |
| 190 | * <tr align="left"><th colspan="2">java.lang.Character classes (simple <a href="#jcc">java character type</a>)</th></tr> |
| 191 | * |
| 192 | * <tr><td valign="top"><tt>\p{javaLowerCase}</tt></td> |
| 193 | * <td>Equivalent to java.lang.Character.isLowerCase()</td></tr> |
| 194 | * <tr><td valign="top"><tt>\p{javaUpperCase}</tt></td> |
| 195 | * <td>Equivalent to java.lang.Character.isUpperCase()</td></tr> |
| 196 | * <tr><td valign="top"><tt>\p{javaWhitespace}</tt></td> |
| 197 | * <td>Equivalent to java.lang.Character.isWhitespace()</td></tr> |
| 198 | * <tr><td valign="top"><tt>\p{javaMirrored}</tt></td> |
| 199 | * <td>Equivalent to java.lang.Character.isMirrored()</td></tr> |
| 200 | * |
| 201 | * <tr><th> </th></tr> |
| 202 | * <tr align="left"><th colspan="2" id="unicode">Classes for Unicode blocks and categories</th></tr> |
| 203 | * |
| 204 | * <tr><td valign="top" headers="construct unicode"><tt>\p{InGreek}</tt></td> |
| 205 | * <td headers="matches">A character in the Greek block (simple <a href="#ubc">block</a>)</td></tr> |
| 206 | * <tr><td valign="top" headers="construct unicode"><tt>\p{Lu}</tt></td> |
| 207 | * <td headers="matches">An uppercase letter (simple <a href="#ubc">category</a>)</td></tr> |
| 208 | * <tr><td valign="top" headers="construct unicode"><tt>\p{Sc}</tt></td> |
| 209 | * <td headers="matches">A currency symbol</td></tr> |
| 210 | * <tr><td valign="top" headers="construct unicode"><tt>\P{InGreek}</tt></td> |
| 211 | * <td headers="matches">Any character except one in the Greek block (negation)</td></tr> |
| 212 | * <tr><td valign="top" headers="construct unicode"><tt>[\p{L}&&[^\p{Lu}]] </tt></td> |
| 213 | * <td headers="matches">Any letter except an uppercase letter (subtraction)</td></tr> |
| 214 | * |
| 215 | * <tr><th> </th></tr> |
| 216 | * <tr align="left"><th colspan="2" id="bounds">Boundary matchers</th></tr> |
| 217 | * |
| 218 | * <tr><td valign="top" headers="construct bounds"><tt>^</tt></td> |
| 219 | * <td headers="matches">The beginning of a line</td></tr> |
| 220 | * <tr><td valign="top" headers="construct bounds"><tt>$</tt></td> |
| 221 | * <td headers="matches">The end of a line</td></tr> |
| 222 | * <tr><td valign="top" headers="construct bounds"><tt>\b</tt></td> |
| 223 | * <td headers="matches">A word boundary</td></tr> |
| 224 | * <tr><td valign="top" headers="construct bounds"><tt>\B</tt></td> |
| 225 | * <td headers="matches">A non-word boundary</td></tr> |
| 226 | * <tr><td valign="top" headers="construct bounds"><tt>\A</tt></td> |
| 227 | * <td headers="matches">The beginning of the input</td></tr> |
| 228 | * <tr><td valign="top" headers="construct bounds"><tt>\G</tt></td> |
| 229 | * <td headers="matches">The end of the previous match</td></tr> |
| 230 | * <tr><td valign="top" headers="construct bounds"><tt>\Z</tt></td> |
| 231 | * <td headers="matches">The end of the input but for the final |
| 232 | * <a href="#lt">terminator</a>, if any</td></tr> |
| 233 | * <tr><td valign="top" headers="construct bounds"><tt>\z</tt></td> |
| 234 | * <td headers="matches">The end of the input</td></tr> |
| 235 | * |
| 236 | * <tr><th> </th></tr> |
| 237 | * <tr align="left"><th colspan="2" id="greedy">Greedy quantifiers</th></tr> |
| 238 | * |
| 239 | * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>?</tt></td> |
| 240 | * <td headers="matches"><i>X</i>, once or not at all</td></tr> |
| 241 | * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>*</tt></td> |
| 242 | * <td headers="matches"><i>X</i>, zero or more times</td></tr> |
| 243 | * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>+</tt></td> |
| 244 | * <td headers="matches"><i>X</i>, one or more times</td></tr> |
| 245 | * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>}</tt></td> |
| 246 | * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> |
| 247 | * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,}</tt></td> |
| 248 | * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> |
| 249 | * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}</tt></td> |
| 250 | * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> |
| 251 | * |
| 252 | * <tr><th> </th></tr> |
| 253 | * <tr align="left"><th colspan="2" id="reluc">Reluctant quantifiers</th></tr> |
| 254 | * |
| 255 | * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>??</tt></td> |
| 256 | * <td headers="matches"><i>X</i>, once or not at all</td></tr> |
| 257 | * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>*?</tt></td> |
| 258 | * <td headers="matches"><i>X</i>, zero or more times</td></tr> |
| 259 | * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>+?</tt></td> |
| 260 | * <td headers="matches"><i>X</i>, one or more times</td></tr> |
| 261 | * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>}?</tt></td> |
| 262 | * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> |
| 263 | * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,}?</tt></td> |
| 264 | * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> |
| 265 | * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}?</tt></td> |
| 266 | * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> |
| 267 | * |
| 268 | * <tr><th> </th></tr> |
| 269 | * <tr align="left"><th colspan="2" id="poss">Possessive quantifiers</th></tr> |
| 270 | * |
| 271 | * <tr><td valign="top" headers="construct poss"><i>X</i><tt>?+</tt></td> |
| 272 | * <td headers="matches"><i>X</i>, once or not at all</td></tr> |
| 273 | * <tr><td valign="top" headers="construct poss"><i>X</i><tt>*+</tt></td> |
| 274 | * <td headers="matches"><i>X</i>, zero or more times</td></tr> |
| 275 | * <tr><td valign="top" headers="construct poss"><i>X</i><tt>++</tt></td> |
| 276 | * <td headers="matches"><i>X</i>, one or more times</td></tr> |
| 277 | * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>}+</tt></td> |
| 278 | * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> |
| 279 | * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,}+</tt></td> |
| 280 | * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> |
| 281 | * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}+</tt></td> |
| 282 | * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> |
| 283 | * |
| 284 | * <tr><th> </th></tr> |
| 285 | * <tr align="left"><th colspan="2" id="logical">Logical operators</th></tr> |
| 286 | * |
| 287 | * <tr><td valign="top" headers="construct logical"><i>XY</i></td> |
| 288 | * <td headers="matches"><i>X</i> followed by <i>Y</i></td></tr> |
| 289 | * <tr><td valign="top" headers="construct logical"><i>X</i><tt>|</tt><i>Y</i></td> |
| 290 | * <td headers="matches">Either <i>X</i> or <i>Y</i></td></tr> |
| 291 | * <tr><td valign="top" headers="construct logical"><tt>(</tt><i>X</i><tt>)</tt></td> |
| 292 | * <td headers="matches">X, as a <a href="#cg">capturing group</a></td></tr> |
| 293 | * |
| 294 | * <tr><th> </th></tr> |
| 295 | * <tr align="left"><th colspan="2" id="backref">Back references</th></tr> |
| 296 | * |
| 297 | * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>n</i></td> |
| 298 | * <td valign="bottom" headers="matches">Whatever the <i>n</i><sup>th</sup> |
| 299 | * <a href="#cg">capturing group</a> matched</td></tr> |
| 300 | * |
| 301 | * <tr><th> </th></tr> |
| 302 | * <tr align="left"><th colspan="2" id="quot">Quotation</th></tr> |
| 303 | * |
| 304 | * <tr><td valign="top" headers="construct quot"><tt>\</tt></td> |
| 305 | * <td headers="matches">Nothing, but quotes the following character</td></tr> |
| 306 | * <tr><td valign="top" headers="construct quot"><tt>\Q</tt></td> |
| 307 | * <td headers="matches">Nothing, but quotes all characters until <tt>\E</tt></td></tr> |
| 308 | * <tr><td valign="top" headers="construct quot"><tt>\E</tt></td> |
| 309 | * <td headers="matches">Nothing, but ends quoting started by <tt>\Q</tt></td></tr> |
| 310 | * <!-- Metachars: !$()*+.<>?[\]^{|} --> |
| 311 | * |
| 312 | * <tr><th> </th></tr> |
| 313 | * <tr align="left"><th colspan="2" id="special">Special constructs (non-capturing)</th></tr> |
| 314 | * |
| 315 | * <tr><td valign="top" headers="construct special"><tt>(?:</tt><i>X</i><tt>)</tt></td> |
| 316 | * <td headers="matches"><i>X</i>, as a non-capturing group</td></tr> |
| 317 | * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux) </tt></td> |
| 318 | * <td headers="matches">Nothing, but turns match flags <a href="#CASE_INSENSITIVE">i</a> |
| 319 | * <a href="#UNIX_LINES">d</a> <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> |
| 320 | * <a href="#UNICODE_CASE">u</a> <a href="#COMMENTS">x</a> on - off</td></tr> |
| 321 | * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux:</tt><i>X</i><tt>)</tt> </td> |
| 322 | * <td headers="matches"><i>X</i>, as a <a href="#cg">non-capturing group</a> with the |
| 323 | * given flags <a href="#CASE_INSENSITIVE">i</a> <a href="#UNIX_LINES">d</a> |
| 324 | * <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> <a href="#UNICODE_CASE">u</a > |
| 325 | * <a href="#COMMENTS">x</a> on - off</td></tr> |
| 326 | * <tr><td valign="top" headers="construct special"><tt>(?=</tt><i>X</i><tt>)</tt></td> |
| 327 | * <td headers="matches"><i>X</i>, via zero-width positive lookahead</td></tr> |
| 328 | * <tr><td valign="top" headers="construct special"><tt>(?!</tt><i>X</i><tt>)</tt></td> |
| 329 | * <td headers="matches"><i>X</i>, via zero-width negative lookahead</td></tr> |
| 330 | * <tr><td valign="top" headers="construct special"><tt>(?<=</tt><i>X</i><tt>)</tt></td> |
| 331 | * <td headers="matches"><i>X</i>, via zero-width positive lookbehind</td></tr> |
| 332 | * <tr><td valign="top" headers="construct special"><tt>(?<!</tt><i>X</i><tt>)</tt></td> |
| 333 | * <td headers="matches"><i>X</i>, via zero-width negative lookbehind</td></tr> |
| 334 | * <tr><td valign="top" headers="construct special"><tt>(?></tt><i>X</i><tt>)</tt></td> |
| 335 | * <td headers="matches"><i>X</i>, as an independent, non-capturing group</td></tr> |
| 336 | * |
| 337 | * </table> |
| 338 | * |
| 339 | * <hr> |
| 340 | * |
| 341 | * |
| 342 | * <a name="bs"> |
| 343 | * <h4> Backslashes, escapes, and quoting </h4> |
| 344 | * |
| 345 | * <p> The backslash character (<tt>'\'</tt>) serves to introduce escaped |
| 346 | * constructs, as defined in the table above, as well as to quote characters |
| 347 | * that otherwise would be interpreted as unescaped constructs. Thus the |
| 348 | * expression <tt>\\</tt> matches a single backslash and <tt>\{</tt> matches a |
| 349 | * left brace. |
| 350 | * |
| 351 | * <p> It is an error to use a backslash prior to any alphabetic character that |
| 352 | * does not denote an escaped construct; these are reserved for future |
| 353 | * extensions to the regular-expression language. A backslash may be used |
| 354 | * prior to a non-alphabetic character regardless of whether that character is |
| 355 | * part of an unescaped construct. |
| 356 | * |
| 357 | * <p> Backslashes within string literals in Java source code are interpreted |
| 358 | * as required by the <a |
| 359 | * href="http://java.sun.com/docs/books/jls">Java Language |
| 360 | * Specification</a> as either <a |
| 361 | * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#100850">Unicode |
| 362 | * escapes</a> or other <a |
| 363 | * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#101089">character |
| 364 | * escapes</a>. It is therefore necessary to double backslashes in string |
| 365 | * literals that represent regular expressions to protect them from |
| 366 | * interpretation by the Java bytecode compiler. The string literal |
| 367 | * <tt>"\b"</tt>, for example, matches a single backspace character when |
| 368 | * interpreted as a regular expression, while <tt>"\\b"</tt> matches a |
| 369 | * word boundary. The string literal <tt>"\(hello\)"</tt> is illegal |
| 370 | * and leads to a compile-time error; in order to match the string |
| 371 | * <tt>(hello)</tt> the string literal <tt>"\\(hello\\)"</tt> |
| 372 | * must be used. |
| 373 | * |
| 374 | * <a name="cc"> |
| 375 | * <h4> Character Classes </h4> |
| 376 | * |
| 377 | * <p> Character classes may appear within other character classes, and |
| 378 | * may be composed by the union operator (implicit) and the intersection |
| 379 | * operator (<tt>&&</tt>). |
| 380 | * The union operator denotes a class that contains every character that is |
| 381 | * in at least one of its operand classes. The intersection operator |
| 382 | * denotes a class that contains every character that is in both of its |
| 383 | * operand classes. |
| 384 | * |
| 385 | * <p> The precedence of character-class operators is as follows, from |
| 386 | * highest to lowest: |
| 387 | * |
| 388 | * <blockquote><table border="0" cellpadding="1" cellspacing="0" |
| 389 | * summary="Precedence of character class operators."> |
| 390 | * <tr><th>1 </th> |
| 391 | * <td>Literal escape </td> |
| 392 | * <td><tt>\x</tt></td></tr> |
| 393 | * <tr><th>2 </th> |
| 394 | * <td>Grouping</td> |
| 395 | * <td><tt>[...]</tt></td></tr> |
| 396 | * <tr><th>3 </th> |
| 397 | * <td>Range</td> |
| 398 | * <td><tt>a-z</tt></td></tr> |
| 399 | * <tr><th>4 </th> |
| 400 | * <td>Union</td> |
| 401 | * <td><tt>[a-e][i-u]</tt></td></tr> |
| 402 | * <tr><th>5 </th> |
| 403 | * <td>Intersection</td> |
| 404 | * <td><tt>[a-z&&[aeiou]]</tt></td></tr> |
| 405 | * </table></blockquote> |
| 406 | * |
| 407 | * <p> Note that a different set of metacharacters are in effect inside |
| 408 | * a character class than outside a character class. For instance, the |
| 409 | * regular expression <tt>.</tt> loses its special meaning inside a |
| 410 | * character class, while the expression <tt>-</tt> becomes a range |
| 411 | * forming metacharacter. |
| 412 | * |
| 413 | * <a name="lt"> |
| 414 | * <h4> Line terminators </h4> |
| 415 | * |
| 416 | * <p> A <i>line terminator</i> is a one- or two-character sequence that marks |
| 417 | * the end of a line of the input character sequence. The following are |
| 418 | * recognized as line terminators: |
| 419 | * |
| 420 | * <ul> |
| 421 | * |
| 422 | * <li> A newline (line feed) character (<tt>'\n'</tt>), |
| 423 | * |
| 424 | * <li> A carriage-return character followed immediately by a newline |
| 425 | * character (<tt>"\r\n"</tt>), |
| 426 | * |
| 427 | * <li> A standalone carriage-return character (<tt>'\r'</tt>), |
| 428 | * |
| 429 | * <li> A next-line character (<tt>'\u0085'</tt>), |
| 430 | * |
| 431 | * <li> A line-separator character (<tt>'\u2028'</tt>), or |
| 432 | * |
| 433 | * <li> A paragraph-separator character (<tt>'\u2029</tt>). |
| 434 | * |
| 435 | * </ul> |
| 436 | * <p>If {@link #UNIX_LINES} mode is activated, then the only line terminators |
| 437 | * recognized are newline characters. |
| 438 | * |
| 439 | * <p> The regular expression <tt>.</tt> matches any character except a line |
| 440 | * terminator unless the {@link #DOTALL} flag is specified. |
| 441 | * |
| 442 | * <p> By default, the regular expressions <tt>^</tt> and <tt>$</tt> ignore |
| 443 | * line terminators and only match at the beginning and the end, respectively, |
| 444 | * of the entire input sequence. If {@link #MULTILINE} mode is activated then |
| 445 | * <tt>^</tt> matches at the beginning of input and after any line terminator |
| 446 | * except at the end of input. When in {@link #MULTILINE} mode <tt>$</tt> |
| 447 | * matches just before a line terminator or the end of the input sequence. |
| 448 | * |
| 449 | * <a name="cg"> |
| 450 | * <h4> Groups and capturing </h4> |
| 451 | * |
| 452 | * <p> Capturing groups are numbered by counting their opening parentheses from |
| 453 | * left to right. In the expression <tt>((A)(B(C)))</tt>, for example, there |
| 454 | * are four such groups: </p> |
| 455 | * |
| 456 | * <blockquote><table cellpadding=1 cellspacing=0 summary="Capturing group numberings"> |
| 457 | * <tr><th>1 </th> |
| 458 | * <td><tt>((A)(B(C)))</tt></td></tr> |
| 459 | * <tr><th>2 </th> |
| 460 | * <td><tt>(A)</tt></td></tr> |
| 461 | * <tr><th>3 </th> |
| 462 | * <td><tt>(B(C))</tt></td></tr> |
| 463 | * <tr><th>4 </th> |
| 464 | * <td><tt>(C)</tt></td></tr> |
| 465 | * </table></blockquote> |
| 466 | * |
| 467 | * <p> Group zero always stands for the entire expression. |
| 468 | * |
| 469 | * <p> Capturing groups are so named because, during a match, each subsequence |
| 470 | * of the input sequence that matches such a group is saved. The captured |
| 471 | * subsequence may be used later in the expression, via a back reference, and |
| 472 | * may also be retrieved from the matcher once the match operation is complete. |
| 473 | * |
| 474 | * <p> The captured input associated with a group is always the subsequence |
| 475 | * that the group most recently matched. If a group is evaluated a second time |
| 476 | * because of quantification then its previously-captured value, if any, will |
| 477 | * be retained if the second evaluation fails. Matching the string |
| 478 | * <tt>"aba"</tt> against the expression <tt>(a(b)?)+</tt>, for example, leaves |
| 479 | * group two set to <tt>"b"</tt>. All captured input is discarded at the |
| 480 | * beginning of each match. |
| 481 | * |
| 482 | * <p> Groups beginning with <tt>(?</tt> are pure, <i>non-capturing</i> groups |
| 483 | * that do not capture text and do not count towards the group total. |
| 484 | * |
| 485 | * |
| 486 | * <h4> Unicode support </h4> |
| 487 | * |
| 488 | * <p> This class is in conformance with Level 1 of <a |
| 489 | * href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical |
| 490 | * Standard #18: Unicode Regular Expression Guidelines</i></a>, plus RL2.1 |
| 491 | * Canonical Equivalents. |
| 492 | * |
| 493 | * <p> Unicode escape sequences such as <tt>\u2014</tt> in Java source code |
| 494 | * are processed as described in <a |
| 495 | * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#100850">\u00A73.3</a> |
| 496 | * of the Java Language Specification. Such escape sequences are also |
| 497 | * implemented directly by the regular-expression parser so that Unicode |
| 498 | * escapes can be used in expressions that are read from files or from the |
| 499 | * keyboard. Thus the strings <tt>"\u2014"</tt> and <tt>"\\u2014"</tt>, |
| 500 | * while not equal, compile into the same pattern, which matches the character |
| 501 | * with hexadecimal value <tt>0x2014</tt>. |
| 502 | * |
| 503 | * <a name="ubc"> <p>Unicode blocks and categories are written with the |
| 504 | * <tt>\p</tt> and <tt>\P</tt> constructs as in |
| 505 | * Perl. <tt>\p{</tt><i>prop</i><tt>}</tt> matches if the input has the |
| 506 | * property <i>prop</i>, while <tt>\P{</tt><i>prop</i><tt>}</tt> does not match if |
| 507 | * the input has that property. Blocks are specified with the prefix |
| 508 | * <tt>In</tt>, as in <tt>InMongolian</tt>. Categories may be specified with |
| 509 | * the optional prefix <tt>Is</tt>: Both <tt>\p{L}</tt> and <tt>\p{IsL}</tt> |
| 510 | * denote the category of Unicode letters. Blocks and categories can be used |
| 511 | * both inside and outside of a character class. |
| 512 | * |
| 513 | * <p> The supported categories are those of |
| 514 | * <a href="http://www.unicode.org/unicode/standard/standard.html"> |
| 515 | * <i>The Unicode Standard</i></a> in the version specified by the |
| 516 | * {@link java.lang.Character Character} class. The category names are those |
| 517 | * defined in the Standard, both normative and informative. |
| 518 | * The block names supported by <code>Pattern</code> are the valid block names |
| 519 | * accepted and defined by |
| 520 | * {@link java.lang.Character.UnicodeBlock#forName(String) UnicodeBlock.forName}. |
| 521 | * |
| 522 | * <a name="jcc"> <p>Categories that behave like the java.lang.Character |
| 523 | * boolean is<i>methodname</i> methods (except for the deprecated ones) are |
| 524 | * available through the same <tt>\p{</tt><i>prop</i><tt>}</tt> syntax where |
| 525 | * the specified property has the name <tt>java<i>methodname</i></tt>. |
| 526 | * |
| 527 | * <h4> Comparison to Perl 5 </h4> |
| 528 | * |
| 529 | * <p>The <code>Pattern</code> engine performs traditional NFA-based matching |
| 530 | * with ordered alternation as occurs in Perl 5. |
| 531 | * |
| 532 | * <p> Perl constructs not supported by this class: </p> |
| 533 | * |
| 534 | * <ul> |
| 535 | * |
| 536 | * <li><p> The conditional constructs <tt>(?{</tt><i>X</i><tt>})</tt> and |
| 537 | * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>|</tt><i>Y</i><tt>)</tt>, |
| 538 | * </p></li> |
| 539 | * |
| 540 | * <li><p> The embedded code constructs <tt>(?{</tt><i>code</i><tt>})</tt> |
| 541 | * and <tt>(??{</tt><i>code</i><tt>})</tt>,</p></li> |
| 542 | * |
| 543 | * <li><p> The embedded comment syntax <tt>(?#comment)</tt>, and </p></li> |
| 544 | * |
| 545 | * <li><p> The preprocessing operations <tt>\l</tt> <tt>\u</tt>, |
| 546 | * <tt>\L</tt>, and <tt>\U</tt>. </p></li> |
| 547 | * |
| 548 | * </ul> |
| 549 | * |
| 550 | * <p> Constructs supported by this class but not by Perl: </p> |
| 551 | * |
| 552 | * <ul> |
| 553 | * |
| 554 | * <li><p> Possessive quantifiers, which greedily match as much as they can |
| 555 | * and do not back off, even when doing so would allow the overall match to |
| 556 | * succeed. </p></li> |
| 557 | * |
| 558 | * <li><p> Character-class union and intersection as described |
| 559 | * <a href="#cc">above</a>.</p></li> |
| 560 | * |
| 561 | * </ul> |
| 562 | * |
| 563 | * <p> Notable differences from Perl: </p> |
| 564 | * |
| 565 | * <ul> |
| 566 | * |
| 567 | * <li><p> In Perl, <tt>\1</tt> through <tt>\9</tt> are always interpreted |
| 568 | * as back references; a backslash-escaped number greater than <tt>9</tt> is |
| 569 | * treated as a back reference if at least that many subexpressions exist, |
| 570 | * otherwise it is interpreted, if possible, as an octal escape. In this |
| 571 | * class octal escapes must always begin with a zero. In this class, |
| 572 | * <tt>\1</tt> through <tt>\9</tt> are always interpreted as back |
| 573 | * references, and a larger number is accepted as a back reference if at |
| 574 | * least that many subexpressions exist at that point in the regular |
| 575 | * expression, otherwise the parser will drop digits until the number is |
| 576 | * smaller or equal to the existing number of groups or it is one digit. |
| 577 | * </p></li> |
| 578 | * |
| 579 | * <li><p> Perl uses the <tt>g</tt> flag to request a match that resumes |
| 580 | * where the last match left off. This functionality is provided implicitly |
| 581 | * by the {@link Matcher} class: Repeated invocations of the {@link |
| 582 | * Matcher#find find} method will resume where the last match left off, |
| 583 | * unless the matcher is reset. </p></li> |
| 584 | * |
| 585 | * <li><p> In Perl, embedded flags at the top level of an expression affect |
| 586 | * the whole expression. In this class, embedded flags always take effect |
| 587 | * at the point at which they appear, whether they are at the top level or |
| 588 | * within a group; in the latter case, flags are restored at the end of the |
| 589 | * group just as in Perl. </p></li> |
| 590 | * |
| 591 | * <li><p> Perl is forgiving about malformed matching constructs, as in the |
| 592 | * expression <tt>*a</tt>, as well as dangling brackets, as in the |
| 593 | * expression <tt>abc]</tt>, and treats them as literals. This |
| 594 | * class also accepts dangling brackets but is strict about dangling |
| 595 | * metacharacters like +, ? and *, and will throw a |
| 596 | * {@link PatternSyntaxException} if it encounters them. </p></li> |
| 597 | * |
| 598 | * </ul> |
| 599 | * |
| 600 | * |
| 601 | * <p> For a more precise description of the behavior of regular expression |
| 602 | * constructs, please see <a href="http://www.oreilly.com/catalog/regex3/"> |
| 603 | * <i>Mastering Regular Expressions, 3nd Edition</i>, Jeffrey E. F. Friedl, |
| 604 | * O'Reilly and Associates, 2006.</a> |
| 605 | * </p> |
| 606 | * |
| 607 | * @see java.lang.String#split(String, int) |
| 608 | * @see java.lang.String#split(String) |
| 609 | * |
| 610 | * @author Mike McCloskey |
| 611 | * @author Mark Reinhold |
| 612 | * @author JSR-51 Expert Group |
| 613 | * @since 1.4 |
| 614 | * @spec JSR-51 |
| 615 | */ |
| 616 | |
| 617 | public final class Pattern |
| 618 | implements java.io.Serializable |
| 619 | { |
| 620 | |
| 621 | /** |
| 622 | * Regular expression modifier values. Instead of being passed as |
| 623 | * arguments, they can also be passed as inline modifiers. |
| 624 | * For example, the following statements have the same effect. |
| 625 | * <pre> |
| 626 | * RegExp r1 = RegExp.compile("abc", Pattern.I|Pattern.M); |
| 627 | * RegExp r2 = RegExp.compile("(?im)abc", 0); |
| 628 | * </pre> |
| 629 | * |
| 630 | * The flags are duplicated so that the familiar Perl match flag |
| 631 | * names are available. |
| 632 | */ |
| 633 | |
| 634 | /** |
| 635 | * Enables Unix lines mode. |
| 636 | * |
| 637 | * <p> In this mode, only the <tt>'\n'</tt> line terminator is recognized |
| 638 | * in the behavior of <tt>.</tt>, <tt>^</tt>, and <tt>$</tt>. |
| 639 | * |
| 640 | * <p> Unix lines mode can also be enabled via the embedded flag |
| 641 | * expression <tt>(?d)</tt>. |
| 642 | */ |
| 643 | public static final int UNIX_LINES = 0x01; |
| 644 | |
| 645 | /** |
| 646 | * Enables case-insensitive matching. |
| 647 | * |
| 648 | * <p> By default, case-insensitive matching assumes that only characters |
| 649 | * in the US-ASCII charset are being matched. Unicode-aware |
| 650 | * case-insensitive matching can be enabled by specifying the {@link |
| 651 | * #UNICODE_CASE} flag in conjunction with this flag. |
| 652 | * |
| 653 | * <p> Case-insensitive matching can also be enabled via the embedded flag |
| 654 | * expression <tt>(?i)</tt>. |
| 655 | * |
| 656 | * <p> Specifying this flag may impose a slight performance penalty. </p> |
| 657 | */ |
| 658 | public static final int CASE_INSENSITIVE = 0x02; |
| 659 | |
| 660 | /** |
| 661 | * Permits whitespace and comments in pattern. |
| 662 | * |
| 663 | * <p> In this mode, whitespace is ignored, and embedded comments starting |
| 664 | * with <tt>#</tt> are ignored until the end of a line. |
| 665 | * |
| 666 | * <p> Comments mode can also be enabled via the embedded flag |
| 667 | * expression <tt>(?x)</tt>. |
| 668 | */ |
| 669 | public static final int COMMENTS = 0x04; |
| 670 | |
| 671 | /** |
| 672 | * Enables multiline mode. |
| 673 | * |
| 674 | * <p> In multiline mode the expressions <tt>^</tt> and <tt>$</tt> match |
| 675 | * just after or just before, respectively, a line terminator or the end of |
| 676 | * the input sequence. By default these expressions only match at the |
| 677 | * beginning and the end of the entire input sequence. |
| 678 | * |
| 679 | * <p> Multiline mode can also be enabled via the embedded flag |
| 680 | * expression <tt>(?m)</tt>. </p> |
| 681 | */ |
| 682 | public static final int MULTILINE = 0x08; |
| 683 | |
| 684 | /** |
| 685 | * Enables literal parsing of the pattern. |
| 686 | * |
| 687 | * <p> When this flag is specified then the input string that specifies |
| 688 | * the pattern is treated as a sequence of literal characters. |
| 689 | * Metacharacters or escape sequences in the input sequence will be |
| 690 | * given no special meaning. |
| 691 | * |
| 692 | * <p>The flags CASE_INSENSITIVE and UNICODE_CASE retain their impact on |
| 693 | * matching when used in conjunction with this flag. The other flags |
| 694 | * become superfluous. |
| 695 | * |
| 696 | * <p> There is no embedded flag character for enabling literal parsing. |
| 697 | * @since 1.5 |
| 698 | */ |
| 699 | public static final int LITERAL = 0x10; |
| 700 | |
| 701 | /** |
| 702 | * Enables dotall mode. |
| 703 | * |
| 704 | * <p> In dotall mode, the expression <tt>.</tt> matches any character, |
| 705 | * including a line terminator. By default this expression does not match |
| 706 | * line terminators. |
| 707 | * |
| 708 | * <p> Dotall mode can also be enabled via the embedded flag |
| 709 | * expression <tt>(?s)</tt>. (The <tt>s</tt> is a mnemonic for |
| 710 | * "single-line" mode, which is what this is called in Perl.) </p> |
| 711 | */ |
| 712 | public static final int DOTALL = 0x20; |
| 713 | |
| 714 | /** |
| 715 | * Enables Unicode-aware case folding. |
| 716 | * |
| 717 | * <p> When this flag is specified then case-insensitive matching, when |
| 718 | * enabled by the {@link #CASE_INSENSITIVE} flag, is done in a manner |
| 719 | * consistent with the Unicode Standard. By default, case-insensitive |
| 720 | * matching assumes that only characters in the US-ASCII charset are being |
| 721 | * matched. |
| 722 | * |
| 723 | * <p> Unicode-aware case folding can also be enabled via the embedded flag |
| 724 | * expression <tt>(?u)</tt>. |
| 725 | * |
| 726 | * <p> Specifying this flag may impose a performance penalty. </p> |
| 727 | */ |
| 728 | public static final int UNICODE_CASE = 0x40; |
| 729 | |
| 730 | /** |
| 731 | * Enables canonical equivalence. |
| 732 | * |
| 733 | * <p> When this flag is specified then two characters will be considered |
| 734 | * to match if, and only if, their full canonical decompositions match. |
| 735 | * The expression <tt>"a\u030A"</tt>, for example, will match the |
| 736 | * string <tt>"\u00E5"</tt> when this flag is specified. By default, |
| 737 | * matching does not take canonical equivalence into account. |
| 738 | * |
| 739 | * <p> There is no embedded flag character for enabling canonical |
| 740 | * equivalence. |
| 741 | * |
| 742 | * <p> Specifying this flag may impose a performance penalty. </p> |
| 743 | */ |
| 744 | public static final int CANON_EQ = 0x80; |
| 745 | |
| 746 | /* Pattern has only two serialized components: The pattern string |
| 747 | * and the flags, which are all that is needed to recompile the pattern |
| 748 | * when it is deserialized. |
| 749 | */ |
| 750 | |
| 751 | /** use serialVersionUID from Merlin b59 for interoperability */ |
| 752 | private static final long serialVersionUID = 5073258162644648461L; |
| 753 | |
| 754 | /** |
| 755 | * The original regular-expression pattern string. |
| 756 | * |
| 757 | * @serial |
| 758 | */ |
| 759 | private String pattern; |
| 760 | |
| 761 | /** |
| 762 | * The original pattern flags. |
| 763 | * |
| 764 | * @serial |
| 765 | */ |
| 766 | private int flags; |
| 767 | |
| 768 | /** |
| 769 | * Boolean indicating this Pattern is compiled; this is necessary in order |
| 770 | * to lazily compile deserialized Patterns. |
| 771 | */ |
| 772 | private transient volatile boolean compiled = false; |
| 773 | |
| 774 | /** |
| 775 | * The normalized pattern string. |
| 776 | */ |
| 777 | private transient String normalizedPattern; |
| 778 | |
| 779 | /** |
| 780 | * The starting point of state machine for the find operation. This allows |
| 781 | * a match to start anywhere in the input. |
| 782 | */ |
| 783 | transient Node root; |
| 784 | |
| 785 | /** |
| 786 | * The root of object tree for a match operation. The pattern is matched |
| 787 | * at the beginning. This may include a find that uses BnM or a First |
| 788 | * node. |
| 789 | */ |
| 790 | transient Node matchRoot; |
| 791 | |
| 792 | /** |
| 793 | * Temporary storage used by parsing pattern slice. |
| 794 | */ |
| 795 | transient int[] buffer; |
| 796 | |
| 797 | /** |
| 798 | * Temporary storage used while parsing group references. |
| 799 | */ |
| 800 | transient GroupHead[] groupNodes; |
| 801 | |
| 802 | /** |
| 803 | * Temporary null terminated code point array used by pattern compiling. |
| 804 | */ |
| 805 | private transient int[] temp; |
| 806 | |
| 807 | /** |
| 808 | * The number of capturing groups in this Pattern. Used by matchers to |
| 809 | * allocate storage needed to perform a match. |
| 810 | */ |
| 811 | transient int capturingGroupCount; |
| 812 | |
| 813 | /** |
| 814 | * The local variable count used by parsing tree. Used by matchers to |
| 815 | * allocate storage needed to perform a match. |
| 816 | */ |
| 817 | transient int localCount; |
| 818 | |
| 819 | /** |
| 820 | * Index into the pattern string that keeps track of how much has been |
| 821 | * parsed. |
| 822 | */ |
| 823 | private transient int cursor; |
| 824 | |
| 825 | /** |
| 826 | * Holds the length of the pattern string. |
| 827 | */ |
| 828 | private transient int patternLength; |
| 829 | |
| 830 | /** |
| 831 | * Compiles the given regular expression into a pattern. </p> |
| 832 | * |
| 833 | * @param regex |
| 834 | * The expression to be compiled |
| 835 | * |
| 836 | * @throws PatternSyntaxException |
| 837 | * If the expression's syntax is invalid |
| 838 | */ |
| 839 | public static Pattern compile(String regex) { |
| 840 | return new Pattern(regex, 0); |
| 841 | } |
| 842 | |
| 843 | /** |
| 844 | * Compiles the given regular expression into a pattern with the given |
| 845 | * flags. </p> |
| 846 | * |
| 847 | * @param regex |
| 848 | * The expression to be compiled |
| 849 | * |
| 850 | * @param flags |
| 851 | * Match flags, a bit mask that may include |
| 852 | * {@link #CASE_INSENSITIVE}, {@link #MULTILINE}, {@link #DOTALL}, |
| 853 | * {@link #UNICODE_CASE}, {@link #CANON_EQ}, {@link #UNIX_LINES}, |
| 854 | * {@link #LITERAL} and {@link #COMMENTS} |
| 855 | * |
| 856 | * @throws IllegalArgumentException |
| 857 | * If bit values other than those corresponding to the defined |
| 858 | * match flags are set in <tt>flags</tt> |
| 859 | * |
| 860 | * @throws PatternSyntaxException |
| 861 | * If the expression's syntax is invalid |
| 862 | */ |
| 863 | public static Pattern compile(String regex, int flags) { |
| 864 | return new Pattern(regex, flags); |
| 865 | } |
| 866 | |
| 867 | /** |
| 868 | * Returns the regular expression from which this pattern was compiled. |
| 869 | * </p> |
| 870 | * |
| 871 | * @return The source of this pattern |
| 872 | */ |
| 873 | public String pattern() { |
| 874 | return pattern; |
| 875 | } |
| 876 | |
| 877 | /** |
| 878 | * <p>Returns the string representation of this pattern. This |
| 879 | * is the regular expression from which this pattern was |
| 880 | * compiled.</p> |
| 881 | * |
| 882 | * @return The string representation of this pattern |
| 883 | * @since 1.5 |
| 884 | */ |
| 885 | public String toString() { |
| 886 | return pattern; |
| 887 | } |
| 888 | |
| 889 | /** |
| 890 | * Creates a matcher that will match the given input against this pattern. |
| 891 | * </p> |
| 892 | * |
| 893 | * @param input |
| 894 | * The character sequence to be matched |
| 895 | * |
| 896 | * @return A new matcher for this pattern |
| 897 | */ |
| 898 | public Matcher matcher(CharSequence input) { |
| 899 | if (!compiled) { |
| 900 | synchronized(this) { |
| 901 | if (!compiled) |
| 902 | compile(); |
| 903 | } |
| 904 | } |
| 905 | Matcher m = new Matcher(this, input); |
| 906 | return m; |
| 907 | } |
| 908 | |
| 909 | /** |
| 910 | * Returns this pattern's match flags. </p> |
| 911 | * |
| 912 | * @return The match flags specified when this pattern was compiled |
| 913 | */ |
| 914 | public int flags() { |
| 915 | return flags; |
| 916 | } |
| 917 | |
| 918 | /** |
| 919 | * Compiles the given regular expression and attempts to match the given |
| 920 | * input against it. |
| 921 | * |
| 922 | * <p> An invocation of this convenience method of the form |
| 923 | * |
| 924 | * <blockquote><pre> |
| 925 | * Pattern.matches(regex, input);</pre></blockquote> |
| 926 | * |
| 927 | * behaves in exactly the same way as the expression |
| 928 | * |
| 929 | * <blockquote><pre> |
| 930 | * Pattern.compile(regex).matcher(input).matches()</pre></blockquote> |
| 931 | * |
| 932 | * <p> If a pattern is to be used multiple times, compiling it once and reusing |
| 933 | * it will be more efficient than invoking this method each time. </p> |
| 934 | * |
| 935 | * @param regex |
| 936 | * The expression to be compiled |
| 937 | * |
| 938 | * @param input |
| 939 | * The character sequence to be matched |
| 940 | * |
| 941 | * @throws PatternSyntaxException |
| 942 | * If the expression's syntax is invalid |
| 943 | */ |
| 944 | public static boolean matches(String regex, CharSequence input) { |
| 945 | Pattern p = Pattern.compile(regex); |
| 946 | Matcher m = p.matcher(input); |
| 947 | return m.matches(); |
| 948 | } |
| 949 | |
| 950 | /** |
| 951 | * Splits the given input sequence around matches of this pattern. |
| 952 | * |
| 953 | * <p> The array returned by this method contains each substring of the |
| 954 | * input sequence that is terminated by another subsequence that matches |
| 955 | * this pattern or is terminated by the end of the input sequence. The |
| 956 | * substrings in the array are in the order in which they occur in the |
| 957 | * input. If this pattern does not match any subsequence of the input then |
| 958 | * the resulting array has just one element, namely the input sequence in |
| 959 | * string form. |
| 960 | * |
| 961 | * <p> The <tt>limit</tt> parameter controls the number of times the |
| 962 | * pattern is applied and therefore affects the length of the resulting |
| 963 | * array. If the limit <i>n</i> is greater than zero then the pattern |
| 964 | * will be applied at most <i>n</i> - 1 times, the array's |
| 965 | * length will be no greater than <i>n</i>, and the array's last entry |
| 966 | * will contain all input beyond the last matched delimiter. If <i>n</i> |
| 967 | * is non-positive then the pattern will be applied as many times as |
| 968 | * possible and the array can have any length. If <i>n</i> is zero then |
| 969 | * the pattern will be applied as many times as possible, the array can |
| 970 | * have any length, and trailing empty strings will be discarded. |
| 971 | * |
| 972 | * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following |
| 973 | * results with these parameters: |
| 974 | * |
| 975 | * <blockquote><table cellpadding=1 cellspacing=0 |
| 976 | * summary="Split examples showing regex, limit, and result"> |
| 977 | * <tr><th><P align="left"><i>Regex </i></th> |
| 978 | * <th><P align="left"><i>Limit </i></th> |
| 979 | * <th><P align="left"><i>Result </i></th></tr> |
| 980 | * <tr><td align=center>:</td> |
| 981 | * <td align=center>2</td> |
| 982 | * <td><tt>{ "boo", "and:foo" }</tt></td></tr> |
| 983 | * <tr><td align=center>:</td> |
| 984 | * <td align=center>5</td> |
| 985 | * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> |
| 986 | * <tr><td align=center>:</td> |
| 987 | * <td align=center>-2</td> |
| 988 | * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> |
| 989 | * <tr><td align=center>o</td> |
| 990 | * <td align=center>5</td> |
| 991 | * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> |
| 992 | * <tr><td align=center>o</td> |
| 993 | * <td align=center>-2</td> |
| 994 | * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> |
| 995 | * <tr><td align=center>o</td> |
| 996 | * <td align=center>0</td> |
| 997 | * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> |
| 998 | * </table></blockquote> |
| 999 | * |
| 1000 | * |
| 1001 | * @param input |
| 1002 | * The character sequence to be split |
| 1003 | * |
| 1004 | * @param limit |
| 1005 | * The result threshold, as described above |
| 1006 | * |
| 1007 | * @return The array of strings computed by splitting the input |
| 1008 | * around matches of this pattern |
| 1009 | */ |
| 1010 | public String[] split(CharSequence input, int limit) { |
| 1011 | int index = 0; |
| 1012 | boolean matchLimited = limit > 0; |
| 1013 | ArrayList<String> matchList = new ArrayList<String>(); |
| 1014 | Matcher m = matcher(input); |
| 1015 | |
| 1016 | // Add segments before each match found |
| 1017 | while(m.find()) { |
| 1018 | if (!matchLimited || matchList.size() < limit - 1) { |
| 1019 | String match = input.subSequence(index, m.start()).toString(); |
| 1020 | matchList.add(match); |
| 1021 | index = m.end(); |
| 1022 | } else if (matchList.size() == limit - 1) { // last one |
| 1023 | String match = input.subSequence(index, |
| 1024 | input.length()).toString(); |
| 1025 | matchList.add(match); |
| 1026 | index = m.end(); |
| 1027 | } |
| 1028 | } |
| 1029 | |
| 1030 | // If no match was found, return this |
| 1031 | if (index == 0) |
| 1032 | return new String[] {input.toString()}; |
| 1033 | |
| 1034 | // Add remaining segment |
| 1035 | if (!matchLimited || matchList.size() < limit) |
| 1036 | matchList.add(input.subSequence(index, input.length()).toString()); |
| 1037 | |
| 1038 | // Construct result |
| 1039 | int resultSize = matchList.size(); |
| 1040 | if (limit == 0) |
| 1041 | while (resultSize > 0 && matchList.get(resultSize-1).equals("")) |
| 1042 | resultSize--; |
| 1043 | String[] result = new String[resultSize]; |
| 1044 | return matchList.subList(0, resultSize).toArray(result); |
| 1045 | } |
| 1046 | |
| 1047 | /** |
| 1048 | * Splits the given input sequence around matches of this pattern. |
| 1049 | * |
| 1050 | * <p> This method works as if by invoking the two-argument {@link |
| 1051 | * #split(java.lang.CharSequence, int) split} method with the given input |
| 1052 | * sequence and a limit argument of zero. Trailing empty strings are |
| 1053 | * therefore not included in the resulting array. </p> |
| 1054 | * |
| 1055 | * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following |
| 1056 | * results with these expressions: |
| 1057 | * |
| 1058 | * <blockquote><table cellpadding=1 cellspacing=0 |
| 1059 | * summary="Split examples showing regex and result"> |
| 1060 | * <tr><th><P align="left"><i>Regex </i></th> |
| 1061 | * <th><P align="left"><i>Result</i></th></tr> |
| 1062 | * <tr><td align=center>:</td> |
| 1063 | * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> |
| 1064 | * <tr><td align=center>o</td> |
| 1065 | * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> |
| 1066 | * </table></blockquote> |
| 1067 | * |
| 1068 | * |
| 1069 | * @param input |
| 1070 | * The character sequence to be split |
| 1071 | * |
| 1072 | * @return The array of strings computed by splitting the input |
| 1073 | * around matches of this pattern |
| 1074 | */ |
| 1075 | public String[] split(CharSequence input) { |
| 1076 | return split(input, 0); |
| 1077 | } |
| 1078 | |
| 1079 | /** |
| 1080 | * Returns a literal pattern <code>String</code> for the specified |
| 1081 | * <code>String</code>. |
| 1082 | * |
| 1083 | * <p>This method produces a <code>String</code> that can be used to |
| 1084 | * create a <code>Pattern</code> that would match the string |
| 1085 | * <code>s</code> as if it were a literal pattern.</p> Metacharacters |
| 1086 | * or escape sequences in the input sequence will be given no special |
| 1087 | * meaning. |
| 1088 | * |
| 1089 | * @param s The string to be literalized |
| 1090 | * @return A literal string replacement |
| 1091 | * @since 1.5 |
| 1092 | */ |
| 1093 | public static String quote(String s) { |
| 1094 | int slashEIndex = s.indexOf("\\E"); |
| 1095 | if (slashEIndex == -1) |
| 1096 | return "\\Q" + s + "\\E"; |
| 1097 | |
| 1098 | StringBuilder sb = new StringBuilder(s.length() * 2); |
| 1099 | sb.append("\\Q"); |
| 1100 | slashEIndex = 0; |
| 1101 | int current = 0; |
| 1102 | while ((slashEIndex = s.indexOf("\\E", current)) != -1) { |
| 1103 | sb.append(s.substring(current, slashEIndex)); |
| 1104 | current = slashEIndex + 2; |
| 1105 | sb.append("\\E\\\\E\\Q"); |
| 1106 | } |
| 1107 | sb.append(s.substring(current, s.length())); |
| 1108 | sb.append("\\E"); |
| 1109 | return sb.toString(); |
| 1110 | } |
| 1111 | |
| 1112 | /** |
| 1113 | * Recompile the Pattern instance from a stream. The original pattern |
| 1114 | * string is read in and the object tree is recompiled from it. |
| 1115 | */ |
| 1116 | private void readObject(java.io.ObjectInputStream s) |
| 1117 | throws java.io.IOException, ClassNotFoundException { |
| 1118 | |
| 1119 | // Read in all fields |
| 1120 | s.defaultReadObject(); |
| 1121 | |
| 1122 | // Initialize counts |
| 1123 | capturingGroupCount = 1; |
| 1124 | localCount = 0; |
| 1125 | |
| 1126 | // if length > 0, the Pattern is lazily compiled |
| 1127 | compiled = false; |
| 1128 | if (pattern.length() == 0) { |
| 1129 | root = new Start(lastAccept); |
| 1130 | matchRoot = lastAccept; |
| 1131 | compiled = true; |
| 1132 | } |
| 1133 | } |
| 1134 | |
| 1135 | /** |
| 1136 | * This private constructor is used to create all Patterns. The pattern |
| 1137 | * string and match flags are all that is needed to completely describe |
| 1138 | * a Pattern. An empty pattern string results in an object tree with |
| 1139 | * only a Start node and a LastNode node. |
| 1140 | */ |
| 1141 | private Pattern(String p, int f) { |
| 1142 | pattern = p; |
| 1143 | flags = f; |
| 1144 | |
| 1145 | // Reset group index count |
| 1146 | capturingGroupCount = 1; |
| 1147 | localCount = 0; |
| 1148 | |
| 1149 | if (pattern.length() > 0) { |
| 1150 | compile(); |
| 1151 | } else { |
| 1152 | root = new Start(lastAccept); |
| 1153 | matchRoot = lastAccept; |
| 1154 | } |
| 1155 | } |
| 1156 | |
| 1157 | /** |
| 1158 | * The pattern is converted to normalizedD form and then a pure group |
| 1159 | * is constructed to match canonical equivalences of the characters. |
| 1160 | */ |
| 1161 | private void normalize() { |
| 1162 | boolean inCharClass = false; |
| 1163 | int lastCodePoint = -1; |
| 1164 | |
| 1165 | // Convert pattern into normalizedD form |
| 1166 | normalizedPattern = Normalizer.normalize(pattern, Normalizer.Form.NFD); |
| 1167 | patternLength = normalizedPattern.length(); |
| 1168 | |
| 1169 | // Modify pattern to match canonical equivalences |
| 1170 | StringBuilder newPattern = new StringBuilder(patternLength); |
| 1171 | for(int i=0; i<patternLength; ) { |
| 1172 | int c = normalizedPattern.codePointAt(i); |
| 1173 | StringBuilder sequenceBuffer; |
| 1174 | if ((Character.getType(c) == Character.NON_SPACING_MARK) |
| 1175 | && (lastCodePoint != -1)) { |
| 1176 | sequenceBuffer = new StringBuilder(); |
| 1177 | sequenceBuffer.appendCodePoint(lastCodePoint); |
| 1178 | sequenceBuffer.appendCodePoint(c); |
| 1179 | while(Character.getType(c) == Character.NON_SPACING_MARK) { |
| 1180 | i += Character.charCount(c); |
| 1181 | if (i >= patternLength) |
| 1182 | break; |
| 1183 | c = normalizedPattern.codePointAt(i); |
| 1184 | sequenceBuffer.appendCodePoint(c); |
| 1185 | } |
| 1186 | String ea = produceEquivalentAlternation( |
| 1187 | sequenceBuffer.toString()); |
| 1188 | newPattern.setLength(newPattern.length()-Character.charCount(lastCodePoint)); |
| 1189 | newPattern.append("(?:").append(ea).append(")"); |
| 1190 | } else if (c == '[' && lastCodePoint != '\\') { |
| 1191 | i = normalizeCharClass(newPattern, i); |
| 1192 | } else { |
| 1193 | newPattern.appendCodePoint(c); |
| 1194 | } |
| 1195 | lastCodePoint = c; |
| 1196 | i += Character.charCount(c); |
| 1197 | } |
| 1198 | normalizedPattern = newPattern.toString(); |
| 1199 | } |
| 1200 | |
| 1201 | /** |
| 1202 | * Complete the character class being parsed and add a set |
| 1203 | * of alternations to it that will match the canonical equivalences |
| 1204 | * of the characters within the class. |
| 1205 | */ |
| 1206 | private int normalizeCharClass(StringBuilder newPattern, int i) { |
| 1207 | StringBuilder charClass = new StringBuilder(); |
| 1208 | StringBuilder eq = null; |
| 1209 | int lastCodePoint = -1; |
| 1210 | String result; |
| 1211 | |
| 1212 | i++; |
| 1213 | charClass.append("["); |
| 1214 | while(true) { |
| 1215 | int c = normalizedPattern.codePointAt(i); |
| 1216 | StringBuilder sequenceBuffer; |
| 1217 | |
| 1218 | if (c == ']' && lastCodePoint != '\\') { |
| 1219 | charClass.append((char)c); |
| 1220 | break; |
| 1221 | } else if (Character.getType(c) == Character.NON_SPACING_MARK) { |
| 1222 | sequenceBuffer = new StringBuilder(); |
| 1223 | sequenceBuffer.appendCodePoint(lastCodePoint); |
| 1224 | while(Character.getType(c) == Character.NON_SPACING_MARK) { |
| 1225 | sequenceBuffer.appendCodePoint(c); |
| 1226 | i += Character.charCount(c); |
| 1227 | if (i >= normalizedPattern.length()) |
| 1228 | break; |
| 1229 | c = normalizedPattern.codePointAt(i); |
| 1230 | } |
| 1231 | String ea = produceEquivalentAlternation( |
| 1232 | sequenceBuffer.toString()); |
| 1233 | |
| 1234 | charClass.setLength(charClass.length()-Character.charCount(lastCodePoint)); |
| 1235 | if (eq == null) |
| 1236 | eq = new StringBuilder(); |
| 1237 | eq.append('|'); |
| 1238 | eq.append(ea); |
| 1239 | } else { |
| 1240 | charClass.appendCodePoint(c); |
| 1241 | i++; |
| 1242 | } |
| 1243 | if (i == normalizedPattern.length()) |
| 1244 | throw error("Unclosed character class"); |
| 1245 | lastCodePoint = c; |
| 1246 | } |
| 1247 | |
| 1248 | if (eq != null) { |
| 1249 | result = "(?:"+charClass.toString()+eq.toString()+")"; |
| 1250 | } else { |
| 1251 | result = charClass.toString(); |
| 1252 | } |
| 1253 | |
| 1254 | newPattern.append(result); |
| 1255 | return i; |
| 1256 | } |
| 1257 | |
| 1258 | /** |
| 1259 | * Given a specific sequence composed of a regular character and |
| 1260 | * combining marks that follow it, produce the alternation that will |
| 1261 | * match all canonical equivalences of that sequence. |
| 1262 | */ |
| 1263 | private String produceEquivalentAlternation(String source) { |
| 1264 | int len = countChars(source, 0, 1); |
| 1265 | if (source.length() == len) |
| 1266 | // source has one character. |
| 1267 | return source; |
| 1268 | |
| 1269 | String base = source.substring(0,len); |
| 1270 | String combiningMarks = source.substring(len); |
| 1271 | |
| 1272 | String[] perms = producePermutations(combiningMarks); |
| 1273 | StringBuilder result = new StringBuilder(source); |
| 1274 | |
| 1275 | // Add combined permutations |
| 1276 | for(int x=0; x<perms.length; x++) { |
| 1277 | String next = base + perms[x]; |
| 1278 | if (x>0) |
| 1279 | result.append("|"+next); |
| 1280 | next = composeOneStep(next); |
| 1281 | if (next != null) |
| 1282 | result.append("|"+produceEquivalentAlternation(next)); |
| 1283 | } |
| 1284 | return result.toString(); |
| 1285 | } |
| 1286 | |
| 1287 | /** |
| 1288 | * Returns an array of strings that have all the possible |
| 1289 | * permutations of the characters in the input string. |
| 1290 | * This is used to get a list of all possible orderings |
| 1291 | * of a set of combining marks. Note that some of the permutations |
| 1292 | * are invalid because of combining class collisions, and these |
| 1293 | * possibilities must be removed because they are not canonically |
| 1294 | * equivalent. |
| 1295 | */ |
| 1296 | private String[] producePermutations(String input) { |
| 1297 | if (input.length() == countChars(input, 0, 1)) |
| 1298 | return new String[] {input}; |
| 1299 | |
| 1300 | if (input.length() == countChars(input, 0, 2)) { |
| 1301 | int c0 = Character.codePointAt(input, 0); |
| 1302 | int c1 = Character.codePointAt(input, Character.charCount(c0)); |
| 1303 | if (getClass(c1) == getClass(c0)) { |
| 1304 | return new String[] {input}; |
| 1305 | } |
| 1306 | String[] result = new String[2]; |
| 1307 | result[0] = input; |
| 1308 | StringBuilder sb = new StringBuilder(2); |
| 1309 | sb.appendCodePoint(c1); |
| 1310 | sb.appendCodePoint(c0); |
| 1311 | result[1] = sb.toString(); |
| 1312 | return result; |
| 1313 | } |
| 1314 | |
| 1315 | int length = 1; |
| 1316 | int nCodePoints = countCodePoints(input); |
| 1317 | for(int x=1; x<nCodePoints; x++) |
| 1318 | length = length * (x+1); |
| 1319 | |
| 1320 | String[] temp = new String[length]; |
| 1321 | |
| 1322 | int combClass[] = new int[nCodePoints]; |
| 1323 | for(int x=0, i=0; x<nCodePoints; x++) { |
| 1324 | int c = Character.codePointAt(input, i); |
| 1325 | combClass[x] = getClass(c); |
| 1326 | i += Character.charCount(c); |
| 1327 | } |
| 1328 | |
| 1329 | // For each char, take it out and add the permutations |
| 1330 | // of the remaining chars |
| 1331 | int index = 0; |
| 1332 | int len; |
| 1333 | // offset maintains the index in code units. |
| 1334 | loop: for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) { |
| 1335 | len = countChars(input, offset, 1); |
| 1336 | boolean skip = false; |
| 1337 | for(int y=x-1; y>=0; y--) { |
| 1338 | if (combClass[y] == combClass[x]) { |
| 1339 | continue loop; |
| 1340 | } |
| 1341 | } |
| 1342 | StringBuilder sb = new StringBuilder(input); |
| 1343 | String otherChars = sb.delete(offset, offset+len).toString(); |
| 1344 | String[] subResult = producePermutations(otherChars); |
| 1345 | |
| 1346 | String prefix = input.substring(offset, offset+len); |
| 1347 | for(int y=0; y<subResult.length; y++) |
| 1348 | temp[index++] = prefix + subResult[y]; |
| 1349 | } |
| 1350 | String[] result = new String[index]; |
| 1351 | for (int x=0; x<index; x++) |
| 1352 | result[x] = temp[x]; |
| 1353 | return result; |
| 1354 | } |
| 1355 | |
| 1356 | private int getClass(int c) { |
| 1357 | return sun.text.Normalizer.getCombiningClass(c); |
| 1358 | } |
| 1359 | |
| 1360 | /** |
| 1361 | * Attempts to compose input by combining the first character |
| 1362 | * with the first combining mark following it. Returns a String |
| 1363 | * that is the composition of the leading character with its first |
| 1364 | * combining mark followed by the remaining combining marks. Returns |
| 1365 | * null if the first two characters cannot be further composed. |
| 1366 | */ |
| 1367 | private String composeOneStep(String input) { |
| 1368 | int len = countChars(input, 0, 2); |
| 1369 | String firstTwoCharacters = input.substring(0, len); |
| 1370 | String result = Normalizer.normalize(firstTwoCharacters, Normalizer.Form.NFC); |
| 1371 | |
| 1372 | if (result.equals(firstTwoCharacters)) |
| 1373 | return null; |
| 1374 | else { |
| 1375 | String remainder = input.substring(len); |
| 1376 | return result + remainder; |
| 1377 | } |
| 1378 | } |
| 1379 | |
| 1380 | /** |
| 1381 | * Preprocess any \Q...\E sequences in `temp', meta-quoting them. |
| 1382 | * See the description of `quotemeta' in perlfunc(1). |
| 1383 | */ |
| 1384 | private void RemoveQEQuoting() { |
| 1385 | final int pLen = patternLength; |
| 1386 | int i = 0; |
| 1387 | while (i < pLen-1) { |
| 1388 | if (temp[i] != '\\') |
| 1389 | i += 1; |
| 1390 | else if (temp[i + 1] != 'Q') |
| 1391 | i += 2; |
| 1392 | else |
| 1393 | break; |
| 1394 | } |
| 1395 | if (i >= pLen - 1) // No \Q sequence found |
| 1396 | return; |
| 1397 | int j = i; |
| 1398 | i += 2; |
| 1399 | int[] newtemp = new int[j + 2*(pLen-i) + 2]; |
| 1400 | System.arraycopy(temp, 0, newtemp, 0, j); |
| 1401 | |
| 1402 | boolean inQuote = true; |
| 1403 | while (i < pLen) { |
| 1404 | int c = temp[i++]; |
| 1405 | if (! ASCII.isAscii(c) || ASCII.isAlnum(c)) { |
| 1406 | newtemp[j++] = c; |
| 1407 | } else if (c != '\\') { |
| 1408 | if (inQuote) newtemp[j++] = '\\'; |
| 1409 | newtemp[j++] = c; |
| 1410 | } else if (inQuote) { |
| 1411 | if (temp[i] == 'E') { |
| 1412 | i++; |
| 1413 | inQuote = false; |
| 1414 | } else { |
| 1415 | newtemp[j++] = '\\'; |
| 1416 | newtemp[j++] = '\\'; |
| 1417 | } |
| 1418 | } else { |
| 1419 | if (temp[i] == 'Q') { |
| 1420 | i++; |
| 1421 | inQuote = true; |
| 1422 | } else { |
| 1423 | newtemp[j++] = c; |
| 1424 | if (i != pLen) |
| 1425 | newtemp[j++] = temp[i++]; |
| 1426 | } |
| 1427 | } |
| 1428 | } |
| 1429 | |
| 1430 | patternLength = j; |
| 1431 | temp = Arrays.copyOf(newtemp, j + 2); // double zero termination |
| 1432 | } |
| 1433 | |
| 1434 | /** |
| 1435 | * Copies regular expression to an int array and invokes the parsing |
| 1436 | * of the expression which will create the object tree. |
| 1437 | */ |
| 1438 | private void compile() { |
| 1439 | // Handle canonical equivalences |
| 1440 | if (has(CANON_EQ) && !has(LITERAL)) { |
| 1441 | normalize(); |
| 1442 | } else { |
| 1443 | normalizedPattern = pattern; |
| 1444 | } |
| 1445 | patternLength = normalizedPattern.length(); |
| 1446 | |
| 1447 | // Copy pattern to int array for convenience |
| 1448 | // Use double zero to terminate pattern |
| 1449 | temp = new int[patternLength + 2]; |
| 1450 | |
| 1451 | boolean hasSupplementary = false; |
| 1452 | int c, count = 0; |
| 1453 | // Convert all chars into code points |
| 1454 | for (int x = 0; x < patternLength; x += Character.charCount(c)) { |
| 1455 | c = normalizedPattern.codePointAt(x); |
| 1456 | if (isSupplementary(c)) { |
| 1457 | hasSupplementary = true; |
| 1458 | } |
| 1459 | temp[count++] = c; |
| 1460 | } |
| 1461 | |
| 1462 | patternLength = count; // patternLength now in code points |
| 1463 | |
| 1464 | if (! has(LITERAL)) |
| 1465 | RemoveQEQuoting(); |
| 1466 | |
| 1467 | // Allocate all temporary objects here. |
| 1468 | buffer = new int[32]; |
| 1469 | groupNodes = new GroupHead[10]; |
| 1470 | |
| 1471 | if (has(LITERAL)) { |
| 1472 | // Literal pattern handling |
| 1473 | matchRoot = newSlice(temp, patternLength, hasSupplementary); |
| 1474 | matchRoot.next = lastAccept; |
| 1475 | } else { |
| 1476 | // Start recursive descent parsing |
| 1477 | matchRoot = expr(lastAccept); |
| 1478 | // Check extra pattern characters |
| 1479 | if (patternLength != cursor) { |
| 1480 | if (peek() == ')') { |
| 1481 | throw error("Unmatched closing ')'"); |
| 1482 | } else { |
| 1483 | throw error("Unexpected internal error"); |
| 1484 | } |
| 1485 | } |
| 1486 | } |
| 1487 | |
| 1488 | // Peephole optimization |
| 1489 | if (matchRoot instanceof Slice) { |
| 1490 | root = BnM.optimize(matchRoot); |
| 1491 | if (root == matchRoot) { |
| 1492 | root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); |
| 1493 | } |
| 1494 | } else if (matchRoot instanceof Begin || matchRoot instanceof First) { |
| 1495 | root = matchRoot; |
| 1496 | } else { |
| 1497 | root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); |
| 1498 | } |
| 1499 | |
| 1500 | // Release temporary storage |
| 1501 | temp = null; |
| 1502 | buffer = null; |
| 1503 | groupNodes = null; |
| 1504 | patternLength = 0; |
| 1505 | compiled = true; |
| 1506 | } |
| 1507 | |
| 1508 | /** |
| 1509 | * Used to print out a subtree of the Pattern to help with debugging. |
| 1510 | */ |
| 1511 | private static void printObjectTree(Node node) { |
| 1512 | while(node != null) { |
| 1513 | if (node instanceof Prolog) { |
| 1514 | System.out.println(node); |
| 1515 | printObjectTree(((Prolog)node).loop); |
| 1516 | System.out.println("**** end contents prolog loop"); |
| 1517 | } else if (node instanceof Loop) { |
| 1518 | System.out.println(node); |
| 1519 | printObjectTree(((Loop)node).body); |
| 1520 | System.out.println("**** end contents Loop body"); |
| 1521 | } else if (node instanceof Curly) { |
| 1522 | System.out.println(node); |
| 1523 | printObjectTree(((Curly)node).atom); |
| 1524 | System.out.println("**** end contents Curly body"); |
| 1525 | } else if (node instanceof GroupCurly) { |
| 1526 | System.out.println(node); |
| 1527 | printObjectTree(((GroupCurly)node).atom); |
| 1528 | System.out.println("**** end contents GroupCurly body"); |
| 1529 | } else if (node instanceof GroupTail) { |
| 1530 | System.out.println(node); |
| 1531 | System.out.println("Tail next is "+node.next); |
| 1532 | return; |
| 1533 | } else { |
| 1534 | System.out.println(node); |
| 1535 | } |
| 1536 | node = node.next; |
| 1537 | if (node != null) |
| 1538 | System.out.println("->next:"); |
| 1539 | if (node == Pattern.accept) { |
| 1540 | System.out.println("Accept Node"); |
| 1541 | node = null; |
| 1542 | } |
| 1543 | } |
| 1544 | } |
| 1545 | |
| 1546 | /** |
| 1547 | * Used to accumulate information about a subtree of the object graph |
| 1548 | * so that optimizations can be applied to the subtree. |
| 1549 | */ |
| 1550 | static final class TreeInfo { |
| 1551 | int minLength; |
| 1552 | int maxLength; |
| 1553 | boolean maxValid; |
| 1554 | boolean deterministic; |
| 1555 | |
| 1556 | TreeInfo() { |
| 1557 | reset(); |
| 1558 | } |
| 1559 | void reset() { |
| 1560 | minLength = 0; |
| 1561 | maxLength = 0; |
| 1562 | maxValid = true; |
| 1563 | deterministic = true; |
| 1564 | } |
| 1565 | } |
| 1566 | |
| 1567 | /* |
| 1568 | * The following private methods are mainly used to improve the |
| 1569 | * readability of the code. In order to let the Java compiler easily |
| 1570 | * inline them, we should not put many assertions or error checks in them. |
| 1571 | */ |
| 1572 | |
| 1573 | /** |
| 1574 | * Indicates whether a particular flag is set or not. |
| 1575 | */ |
| 1576 | private boolean has(int f) { |
| 1577 | return (flags & f) != 0; |
| 1578 | } |
| 1579 | |
| 1580 | /** |
| 1581 | * Match next character, signal error if failed. |
| 1582 | */ |
| 1583 | private void accept(int ch, String s) { |
| 1584 | int testChar = temp[cursor++]; |
| 1585 | if (has(COMMENTS)) |
| 1586 | testChar = parsePastWhitespace(testChar); |
| 1587 | if (ch != testChar) { |
| 1588 | throw error(s); |
| 1589 | } |
| 1590 | } |
| 1591 | |
| 1592 | /** |
| 1593 | * Mark the end of pattern with a specific character. |
| 1594 | */ |
| 1595 | private void mark(int c) { |
| 1596 | temp[patternLength] = c; |
| 1597 | } |
| 1598 | |
| 1599 | /** |
| 1600 | * Peek the next character, and do not advance the cursor. |
| 1601 | */ |
| 1602 | private int peek() { |
| 1603 | int ch = temp[cursor]; |
| 1604 | if (has(COMMENTS)) |
| 1605 | ch = peekPastWhitespace(ch); |
| 1606 | return ch; |
| 1607 | } |
| 1608 | |
| 1609 | /** |
| 1610 | * Read the next character, and advance the cursor by one. |
| 1611 | */ |
| 1612 | private int read() { |
| 1613 | int ch = temp[cursor++]; |
| 1614 | if (has(COMMENTS)) |
| 1615 | ch = parsePastWhitespace(ch); |
| 1616 | return ch; |
| 1617 | } |
| 1618 | |
| 1619 | /** |
| 1620 | * Read the next character, and advance the cursor by one, |
| 1621 | * ignoring the COMMENTS setting |
| 1622 | */ |
| 1623 | private int readEscaped() { |
| 1624 | int ch = temp[cursor++]; |
| 1625 | return ch; |
| 1626 | } |
| 1627 | |
| 1628 | /** |
| 1629 | * Advance the cursor by one, and peek the next character. |
| 1630 | */ |
| 1631 | private int next() { |
| 1632 | int ch = temp[++cursor]; |
| 1633 | if (has(COMMENTS)) |
| 1634 | ch = peekPastWhitespace(ch); |
| 1635 | return ch; |
| 1636 | } |
| 1637 | |
| 1638 | /** |
| 1639 | * Advance the cursor by one, and peek the next character, |
| 1640 | * ignoring the COMMENTS setting |
| 1641 | */ |
| 1642 | private int nextEscaped() { |
| 1643 | int ch = temp[++cursor]; |
| 1644 | return ch; |
| 1645 | } |
| 1646 | |
| 1647 | /** |
| 1648 | * If in xmode peek past whitespace and comments. |
| 1649 | */ |
| 1650 | private int peekPastWhitespace(int ch) { |
| 1651 | while (ASCII.isSpace(ch) || ch == '#') { |
| 1652 | while (ASCII.isSpace(ch)) |
| 1653 | ch = temp[++cursor]; |
| 1654 | if (ch == '#') { |
| 1655 | ch = peekPastLine(); |
| 1656 | } |
| 1657 | } |
| 1658 | return ch; |
| 1659 | } |
| 1660 | |
| 1661 | /** |
| 1662 | * If in xmode parse past whitespace and comments. |
| 1663 | */ |
| 1664 | private int parsePastWhitespace(int ch) { |
| 1665 | while (ASCII.isSpace(ch) || ch == '#') { |
| 1666 | while (ASCII.isSpace(ch)) |
| 1667 | ch = temp[cursor++]; |
| 1668 | if (ch == '#') |
| 1669 | ch = parsePastLine(); |
| 1670 | } |
| 1671 | return ch; |
| 1672 | } |
| 1673 | |
| 1674 | /** |
| 1675 | * xmode parse past comment to end of line. |
| 1676 | */ |
| 1677 | private int parsePastLine() { |
| 1678 | int ch = temp[cursor++]; |
| 1679 | while (ch != 0 && !isLineSeparator(ch)) |
| 1680 | ch = temp[cursor++]; |
| 1681 | return ch; |
| 1682 | } |
| 1683 | |
| 1684 | /** |
| 1685 | * xmode peek past comment to end of line. |
| 1686 | */ |
| 1687 | private int peekPastLine() { |
| 1688 | int ch = temp[++cursor]; |
| 1689 | while (ch != 0 && !isLineSeparator(ch)) |
| 1690 | ch = temp[++cursor]; |
| 1691 | return ch; |
| 1692 | } |
| 1693 | |
| 1694 | /** |
| 1695 | * Determines if character is a line separator in the current mode |
| 1696 | */ |
| 1697 | private boolean isLineSeparator(int ch) { |
| 1698 | if (has(UNIX_LINES)) { |
| 1699 | return ch == '\n'; |
| 1700 | } else { |
| 1701 | return (ch == '\n' || |
| 1702 | ch == '\r' || |
| 1703 | (ch|1) == '\u2029' || |
| 1704 | ch == '\u0085'); |
| 1705 | } |
| 1706 | } |
| 1707 | |
| 1708 | /** |
| 1709 | * Read the character after the next one, and advance the cursor by two. |
| 1710 | */ |
| 1711 | private int skip() { |
| 1712 | int i = cursor; |
| 1713 | int ch = temp[i+1]; |
| 1714 | cursor = i + 2; |
| 1715 | return ch; |
| 1716 | } |
| 1717 | |
| 1718 | /** |
| 1719 | * Unread one next character, and retreat cursor by one. |
| 1720 | */ |
| 1721 | private void unread() { |
| 1722 | cursor--; |
| 1723 | } |
| 1724 | |
| 1725 | /** |
| 1726 | * Internal method used for handling all syntax errors. The pattern is |
| 1727 | * displayed with a pointer to aid in locating the syntax error. |
| 1728 | */ |
| 1729 | private PatternSyntaxException error(String s) { |
| 1730 | return new PatternSyntaxException(s, normalizedPattern, cursor - 1); |
| 1731 | } |
| 1732 | |
| 1733 | /** |
| 1734 | * Determines if there is any supplementary character or unpaired |
| 1735 | * surrogate in the specified range. |
| 1736 | */ |
| 1737 | private boolean findSupplementary(int start, int end) { |
| 1738 | for (int i = start; i < end; i++) { |
| 1739 | if (isSupplementary(temp[i])) |
| 1740 | return true; |
| 1741 | } |
| 1742 | return false; |
| 1743 | } |
| 1744 | |
| 1745 | /** |
| 1746 | * Determines if the specified code point is a supplementary |
| 1747 | * character or unpaired surrogate. |
| 1748 | */ |
| 1749 | private static final boolean isSupplementary(int ch) { |
| 1750 | return ch >= Character.MIN_SUPPLEMENTARY_CODE_POINT || isSurrogate(ch); |
| 1751 | } |
| 1752 | |
| 1753 | /** |
| 1754 | * The following methods handle the main parsing. They are sorted |
| 1755 | * according to their precedence order, the lowest one first. |
| 1756 | */ |
| 1757 | |
| 1758 | /** |
| 1759 | * The expression is parsed with branch nodes added for alternations. |
| 1760 | * This may be called recursively to parse sub expressions that may |
| 1761 | * contain alternations. |
| 1762 | */ |
| 1763 | private Node expr(Node end) { |
| 1764 | Node prev = null; |
| 1765 | Node firstTail = null; |
| 1766 | Node branchConn = null; |
| 1767 | |
| 1768 | for (;;) { |
| 1769 | Node node = sequence(end); |
| 1770 | Node nodeTail = root; //double return |
| 1771 | if (prev == null) { |
| 1772 | prev = node; |
| 1773 | firstTail = nodeTail; |
| 1774 | } else { |
| 1775 | // Branch |
| 1776 | if (branchConn == null) { |
| 1777 | branchConn = new BranchConn(); |
| 1778 | branchConn.next = end; |
| 1779 | } |
| 1780 | if (node == end) { |
| 1781 | // if the node returned from sequence() is "end" |
| 1782 | // we have an empty expr, set a null atom into |
| 1783 | // the branch to indicate to go "next" directly. |
| 1784 | node = null; |
| 1785 | } else { |
| 1786 | // the "tail.next" of each atom goes to branchConn |
| 1787 | nodeTail.next = branchConn; |
| 1788 | } |
| 1789 | if (prev instanceof Branch) { |
| 1790 | ((Branch)prev).add(node); |
| 1791 | } else { |
| 1792 | if (prev == end) { |
| 1793 | prev = null; |
| 1794 | } else { |
| 1795 | // replace the "end" with "branchConn" at its tail.next |
| 1796 | // when put the "prev" into the branch as the first atom. |
| 1797 | firstTail.next = branchConn; |
| 1798 | } |
| 1799 | prev = new Branch(prev, node, branchConn); |
| 1800 | } |
| 1801 | } |
| 1802 | if (peek() != '|') { |
| 1803 | return prev; |
| 1804 | } |
| 1805 | next(); |
| 1806 | } |
| 1807 | } |
| 1808 | |
| 1809 | /** |
| 1810 | * Parsing of sequences between alternations. |
| 1811 | */ |
| 1812 | private Node sequence(Node end) { |
| 1813 | Node head = null; |
| 1814 | Node tail = null; |
| 1815 | Node node = null; |
| 1816 | LOOP: |
| 1817 | for (;;) { |
| 1818 | int ch = peek(); |
| 1819 | switch (ch) { |
| 1820 | case '(': |
| 1821 | // Because group handles its own closure, |
| 1822 | // we need to treat it differently |
| 1823 | node = group0(); |
| 1824 | // Check for comment or flag group |
| 1825 | if (node == null) |
| 1826 | continue; |
| 1827 | if (head == null) |
| 1828 | head = node; |
| 1829 | else |
| 1830 | tail.next = node; |
| 1831 | // Double return: Tail was returned in root |
| 1832 | tail = root; |
| 1833 | continue; |
| 1834 | case '[': |
| 1835 | node = clazz(true); |
| 1836 | break; |
| 1837 | case '\\': |
| 1838 | ch = nextEscaped(); |
| 1839 | if (ch == 'p' || ch == 'P') { |
| 1840 | boolean oneLetter = true; |
| 1841 | boolean comp = (ch == 'P'); |
| 1842 | ch = next(); // Consume { if present |
| 1843 | if (ch != '{') { |
| 1844 | unread(); |
| 1845 | } else { |
| 1846 | oneLetter = false; |
| 1847 | } |
| 1848 | node = family(oneLetter).maybeComplement(comp); |
| 1849 | } else { |
| 1850 | unread(); |
| 1851 | node = atom(); |
| 1852 | } |
| 1853 | break; |
| 1854 | case '^': |
| 1855 | next(); |
| 1856 | if (has(MULTILINE)) { |
| 1857 | if (has(UNIX_LINES)) |
| 1858 | node = new UnixCaret(); |
| 1859 | else |
| 1860 | node = new Caret(); |
| 1861 | } else { |
| 1862 | node = new Begin(); |
| 1863 | } |
| 1864 | break; |
| 1865 | case '$': |
| 1866 | next(); |
| 1867 | if (has(UNIX_LINES)) |
| 1868 | node = new UnixDollar(has(MULTILINE)); |
| 1869 | else |
| 1870 | node = new Dollar(has(MULTILINE)); |
| 1871 | break; |
| 1872 | case '.': |
| 1873 | next(); |
| 1874 | if (has(DOTALL)) { |
| 1875 | node = new All(); |
| 1876 | } else { |
| 1877 | if (has(UNIX_LINES)) |
| 1878 | node = new UnixDot(); |
| 1879 | else { |
| 1880 | node = new Dot(); |
| 1881 | } |
| 1882 | } |
| 1883 | break; |
| 1884 | case '|': |
| 1885 | case ')': |
| 1886 | break LOOP; |
| 1887 | case ']': // Now interpreting dangling ] and } as literals |
| 1888 | case '}': |
| 1889 | node = atom(); |
| 1890 | break; |
| 1891 | case '?': |
| 1892 | case '*': |
| 1893 | case '+': |
| 1894 | next(); |
| 1895 | throw error("Dangling meta character '" + ((char)ch) + "'"); |
| 1896 | case 0: |
| 1897 | if (cursor >= patternLength) { |
| 1898 | break LOOP; |
| 1899 | } |
| 1900 | // Fall through |
| 1901 | default: |
| 1902 | node = atom(); |
| 1903 | break; |
| 1904 | } |
| 1905 | |
| 1906 | node = closure(node); |
| 1907 | |
| 1908 | if (head == null) { |
| 1909 | head = tail = node; |
| 1910 | } else { |
| 1911 | tail.next = node; |
| 1912 | tail = node; |
| 1913 | } |
| 1914 | } |
| 1915 | if (head == null) { |
| 1916 | return end; |
| 1917 | } |
| 1918 | tail.next = end; |
| 1919 | root = tail; //double return |
| 1920 | return head; |
| 1921 | } |
| 1922 | |
| 1923 | /** |
| 1924 | * Parse and add a new Single or Slice. |
| 1925 | */ |
| 1926 | private Node atom() { |
| 1927 | int first = 0; |
| 1928 | int prev = -1; |
| 1929 | boolean hasSupplementary = false; |
| 1930 | int ch = peek(); |
| 1931 | for (;;) { |
| 1932 | switch (ch) { |
| 1933 | case '*': |
| 1934 | case '+': |
| 1935 | case '?': |
| 1936 | case '{': |
| 1937 | if (first > 1) { |
| 1938 | cursor = prev; // Unwind one character |
| 1939 | first--; |
| 1940 | } |
| 1941 | break; |
| 1942 | case '$': |
| 1943 | case '.': |
| 1944 | case '^': |
| 1945 | case '(': |
| 1946 | case '[': |
| 1947 | case '|': |
| 1948 | case ')': |
| 1949 | break; |
| 1950 | case '\\': |
| 1951 | ch = nextEscaped(); |
| 1952 | if (ch == 'p' || ch == 'P') { // Property |
| 1953 | if (first > 0) { // Slice is waiting; handle it first |
| 1954 | unread(); |
| 1955 | break; |
| 1956 | } else { // No slice; just return the family node |
| 1957 | boolean comp = (ch == 'P'); |
| 1958 | boolean oneLetter = true; |
| 1959 | ch = next(); // Consume { if present |
| 1960 | if (ch != '{') |
| 1961 | unread(); |
| 1962 | else |
| 1963 | oneLetter = false; |
| 1964 | return family(oneLetter).maybeComplement(comp); |
| 1965 | } |
| 1966 | } |
| 1967 | unread(); |
| 1968 | prev = cursor; |
| 1969 | ch = escape(false, first == 0); |
| 1970 | if (ch >= 0) { |
| 1971 | append(ch, first); |
| 1972 | first++; |
| 1973 | if (isSupplementary(ch)) { |
| 1974 | hasSupplementary = true; |
| 1975 | } |
| 1976 | ch = peek(); |
| 1977 | continue; |
| 1978 | } else if (first == 0) { |
| 1979 | return root; |
| 1980 | } |
| 1981 | // Unwind meta escape sequence |
| 1982 | cursor = prev; |
| 1983 | break; |
| 1984 | case 0: |
| 1985 | if (cursor >= patternLength) { |
| 1986 | break; |
| 1987 | } |
| 1988 | // Fall through |
| 1989 | default: |
| 1990 | prev = cursor; |
| 1991 | append(ch, first); |
| 1992 | first++; |
| 1993 | if (isSupplementary(ch)) { |
| 1994 | hasSupplementary = true; |
| 1995 | } |
| 1996 | ch = next(); |
| 1997 | continue; |
| 1998 | } |
| 1999 | break; |
| 2000 | } |
| 2001 | if (first == 1) { |
| 2002 | return newSingle(buffer[0]); |
| 2003 | } else { |
| 2004 | return newSlice(buffer, first, hasSupplementary); |
| 2005 | } |
| 2006 | } |
| 2007 | |
| 2008 | private void append(int ch, int len) { |
| 2009 | if (len >= buffer.length) { |
| 2010 | int[] tmp = new int[len+len]; |
| 2011 | System.arraycopy(buffer, 0, tmp, 0, len); |
| 2012 | buffer = tmp; |
| 2013 | } |
| 2014 | buffer[len] = ch; |
| 2015 | } |
| 2016 | |
| 2017 | /** |
| 2018 | * Parses a backref greedily, taking as many numbers as it |
| 2019 | * can. The first digit is always treated as a backref, but |
| 2020 | * multi digit numbers are only treated as a backref if at |
| 2021 | * least that many backrefs exist at this point in the regex. |
| 2022 | */ |
| 2023 | private Node ref(int refNum) { |
| 2024 | boolean done = false; |
| 2025 | while(!done) { |
| 2026 | int ch = peek(); |
| 2027 | switch(ch) { |
| 2028 | case '0': |
| 2029 | case '1': |
| 2030 | case '2': |
| 2031 | case '3': |
| 2032 | case '4': |
| 2033 | case '5': |
| 2034 | case '6': |
| 2035 | case '7': |
| 2036 | case '8': |
| 2037 | case '9': |
| 2038 | int newRefNum = (refNum * 10) + (ch - '0'); |
| 2039 | // Add another number if it doesn't make a group |
| 2040 | // that doesn't exist |
| 2041 | if (capturingGroupCount - 1 < newRefNum) { |
| 2042 | done = true; |
| 2043 | break; |
| 2044 | } |
| 2045 | refNum = newRefNum; |
| 2046 | read(); |
| 2047 | break; |
| 2048 | default: |
| 2049 | done = true; |
| 2050 | break; |
| 2051 | } |
| 2052 | } |
| 2053 | if (has(CASE_INSENSITIVE)) |
| 2054 | return new CIBackRef(refNum, has(UNICODE_CASE)); |
| 2055 | else |
| 2056 | return new BackRef(refNum); |
| 2057 | } |
| 2058 | |
| 2059 | /** |
| 2060 | * Parses an escape sequence to determine the actual value that needs |
| 2061 | * to be matched. |
| 2062 | * If -1 is returned and create was true a new object was added to the tree |
| 2063 | * to handle the escape sequence. |
| 2064 | * If the returned value is greater than zero, it is the value that |
| 2065 | * matches the escape sequence. |
| 2066 | */ |
| 2067 | private int escape(boolean inclass, boolean create) { |
| 2068 | int ch = skip(); |
| 2069 | switch (ch) { |
| 2070 | case '0': |
| 2071 | return o(); |
| 2072 | case '1': |
| 2073 | case '2': |
| 2074 | case '3': |
| 2075 | case '4': |
| 2076 | case '5': |
| 2077 | case '6': |
| 2078 | case '7': |
| 2079 | case '8': |
| 2080 | case '9': |
| 2081 | if (inclass) break; |
| 2082 | if (create) { |
| 2083 | root = ref((ch - '0')); |
| 2084 | } |
| 2085 | return -1; |
| 2086 | case 'A': |
| 2087 | if (inclass) break; |
| 2088 | if (create) root = new Begin(); |
| 2089 | return -1; |
| 2090 | case 'B': |
| 2091 | if (inclass) break; |
| 2092 | if (create) root = new Bound(Bound.NONE); |
| 2093 | return -1; |
| 2094 | case 'C': |
| 2095 | break; |
| 2096 | case 'D': |
| 2097 | if (create) root = new Ctype(ASCII.DIGIT).complement(); |
| 2098 | return -1; |
| 2099 | case 'E': |
| 2100 | case 'F': |
| 2101 | break; |
| 2102 | case 'G': |
| 2103 | if (inclass) break; |
| 2104 | if (create) root = new LastMatch(); |
| 2105 | return -1; |
| 2106 | case 'H': |
| 2107 | case 'I': |
| 2108 | case 'J': |
| 2109 | case 'K': |
| 2110 | case 'L': |
| 2111 | case 'M': |
| 2112 | case 'N': |
| 2113 | case 'O': |
| 2114 | case 'P': |
| 2115 | case 'Q': |
| 2116 | case 'R': |
| 2117 | break; |
| 2118 | case 'S': |
| 2119 | if (create) root = new Ctype(ASCII.SPACE).complement(); |
| 2120 | return -1; |
| 2121 | case 'T': |
| 2122 | case 'U': |
| 2123 | case 'V': |
| 2124 | break; |
| 2125 | case 'W': |
| 2126 | if (create) root = new Ctype(ASCII.WORD).complement(); |
| 2127 | return -1; |
| 2128 | case 'X': |
| 2129 | case 'Y': |
| 2130 | break; |
| 2131 | case 'Z': |
| 2132 | if (inclass) break; |
| 2133 | if (create) { |
| 2134 | if (has(UNIX_LINES)) |
| 2135 | root = new UnixDollar(false); |
| 2136 | else |
| 2137 | root = new Dollar(false); |
| 2138 | } |
| 2139 | return -1; |
| 2140 | case 'a': |
| 2141 | return '\007'; |
| 2142 | case 'b': |
| 2143 | if (inclass) break; |
| 2144 | if (create) root = new Bound(Bound.BOTH); |
| 2145 | return -1; |
| 2146 | case 'c': |
| 2147 | return c(); |
| 2148 | case 'd': |
| 2149 | if (create) root = new Ctype(ASCII.DIGIT); |
| 2150 | return -1; |
| 2151 | case 'e': |
| 2152 | return '\033'; |
| 2153 | case 'f': |
| 2154 | return '\f'; |
| 2155 | case 'g': |
| 2156 | case 'h': |
| 2157 | case 'i': |
| 2158 | case 'j': |
| 2159 | case 'k': |
| 2160 | case 'l': |
| 2161 | case 'm': |
| 2162 | break; |
| 2163 | case 'n': |
| 2164 | return '\n'; |
| 2165 | case 'o': |
| 2166 | case 'p': |
| 2167 | case 'q': |
| 2168 | break; |
| 2169 | case 'r': |
| 2170 | return '\r'; |
| 2171 | case 's': |
| 2172 | if (create) root = new Ctype(ASCII.SPACE); |
| 2173 | return -1; |
| 2174 | case 't': |
| 2175 | return '\t'; |
| 2176 | case 'u': |
| 2177 | return u(); |
| 2178 | case 'v': |
| 2179 | return '\013'; |
| 2180 | case 'w': |
| 2181 | if (create) root = new Ctype(ASCII.WORD); |
| 2182 | return -1; |
| 2183 | case 'x': |
| 2184 | return x(); |
| 2185 | case 'y': |
| 2186 | break; |
| 2187 | case 'z': |
| 2188 | if (inclass) break; |
| 2189 | if (create) root = new End(); |
| 2190 | return -1; |
| 2191 | default: |
| 2192 | return ch; |
| 2193 | } |
| 2194 | throw error("Illegal/unsupported escape sequence"); |
| 2195 | } |
| 2196 | |
| 2197 | /** |
| 2198 | * Parse a character class, and return the node that matches it. |
| 2199 | * |
| 2200 | * Consumes a ] on the way out if consume is true. Usually consume |
| 2201 | * is true except for the case of [abc&&def] where def is a separate |
| 2202 | * right hand node with "understood" brackets. |
| 2203 | */ |
| 2204 | private CharProperty clazz(boolean consume) { |
| 2205 | CharProperty prev = null; |
| 2206 | CharProperty node = null; |
| 2207 | BitClass bits = new BitClass(); |
| 2208 | boolean include = true; |
| 2209 | boolean firstInClass = true; |
| 2210 | int ch = next(); |
| 2211 | for (;;) { |
| 2212 | switch (ch) { |
| 2213 | case '^': |
| 2214 | // Negates if first char in a class, otherwise literal |
| 2215 | if (firstInClass) { |
| 2216 | if (temp[cursor-1] != '[') |
| 2217 | break; |
| 2218 | ch = next(); |
| 2219 | include = !include; |
| 2220 | continue; |
| 2221 | } else { |
| 2222 | // ^ not first in class, treat as literal |
| 2223 | break; |
| 2224 | } |
| 2225 | case '[': |
| 2226 | firstInClass = false; |
| 2227 | node = clazz(true); |
| 2228 | if (prev == null) |
| 2229 | prev = node; |
| 2230 | else |
| 2231 | prev = union(prev, node); |
| 2232 | ch = peek(); |
| 2233 | continue; |
| 2234 | case '&': |
| 2235 | firstInClass = false; |
| 2236 | ch = next(); |
| 2237 | if (ch == '&') { |
| 2238 | ch = next(); |
| 2239 | CharProperty rightNode = null; |
| 2240 | while (ch != ']' && ch != '&') { |
| 2241 | if (ch == '[') { |
| 2242 | if (rightNode == null) |
| 2243 | rightNode = clazz(true); |
| 2244 | else |
| 2245 | rightNode = union(rightNode, clazz(true)); |
| 2246 | } else { // abc&&def |
| 2247 | unread(); |
| 2248 | rightNode = clazz(false); |
| 2249 | } |
| 2250 | ch = peek(); |
| 2251 | } |
| 2252 | if (rightNode != null) |
| 2253 | node = rightNode; |
| 2254 | if (prev == null) { |
| 2255 | if (rightNode == null) |
| 2256 | throw error("Bad class syntax"); |
| 2257 | else |
| 2258 | prev = rightNode; |
| 2259 | } else { |
| 2260 | prev = intersection(prev, node); |
| 2261 | } |
| 2262 | } else { |
| 2263 | // treat as a literal & |
| 2264 | unread(); |
| 2265 | break; |
| 2266 | } |
| 2267 | continue; |
| 2268 | case 0: |
| 2269 | firstInClass = false; |
| 2270 | if (cursor >= patternLength) |
| 2271 | throw error("Unclosed character class"); |
| 2272 | break; |
| 2273 | case ']': |
| 2274 | firstInClass = false; |
| 2275 | if (prev != null) { |
| 2276 | if (consume) |
| 2277 | next(); |
| 2278 | return prev; |
| 2279 | } |
| 2280 | break; |
| 2281 | default: |
| 2282 | firstInClass = false; |
| 2283 | break; |
| 2284 | } |
| 2285 | node = range(bits); |
| 2286 | if (include) { |
| 2287 | if (prev == null) { |
| 2288 | prev = node; |
| 2289 | } else { |
| 2290 | if (prev != node) |
| 2291 | prev = union(prev, node); |
| 2292 | } |
| 2293 | } else { |
| 2294 | if (prev == null) { |
| 2295 | prev = node.complement(); |
| 2296 | } else { |
| 2297 | if (prev != node) |
| 2298 | prev = setDifference(prev, node); |
| 2299 | } |
| 2300 | } |
| 2301 | ch = peek(); |
| 2302 | } |
| 2303 | } |
| 2304 | |
| 2305 | private CharProperty bitsOrSingle(BitClass bits, int ch) { |
| 2306 | /* Bits can only handle codepoints in [u+0000-u+00ff] range. |
| 2307 | Use "single" node instead of bits when dealing with unicode |
| 2308 | case folding for codepoints listed below. |
| 2309 | (1)Uppercase out of range: u+00ff, u+00b5 |
| 2310 | toUpperCase(u+00ff) -> u+0178 |
| 2311 | toUpperCase(u+00b5) -> u+039c |
| 2312 | (2)LatinSmallLetterLongS u+17f |
| 2313 | toUpperCase(u+017f) -> u+0053 |
| 2314 | (3)LatinSmallLetterDotlessI u+131 |
| 2315 | toUpperCase(u+0131) -> u+0049 |
| 2316 | (4)LatinCapitalLetterIWithDotAbove u+0130 |
| 2317 | toLowerCase(u+0130) -> u+0069 |
| 2318 | (5)KelvinSign u+212a |
| 2319 | toLowerCase(u+212a) ==> u+006B |
| 2320 | (6)AngstromSign u+212b |
| 2321 | toLowerCase(u+212b) ==> u+00e5 |
| 2322 | */ |
| 2323 | int d; |
| 2324 | if (ch < 256 && |
| 2325 | !(has(CASE_INSENSITIVE) && has(UNICODE_CASE) && |
| 2326 | (ch == 0xff || ch == 0xb5 || |
| 2327 | ch == 0x49 || ch == 0x69 || //I and i |
| 2328 | ch == 0x53 || ch == 0x73 || //S and s |
| 2329 | ch == 0x4b || ch == 0x6b || //K and k |
| 2330 | ch == 0xc5 || ch == 0xe5))) //A+ring |
| 2331 | return bits.add(ch, flags()); |
| 2332 | return newSingle(ch); |
| 2333 | } |
| 2334 | |
| 2335 | /** |
| 2336 | * Parse a single character or a character range in a character class |
| 2337 | * and return its representative node. |
| 2338 | */ |
| 2339 | private CharProperty range(BitClass bits) { |
| 2340 | int ch = peek(); |
| 2341 | if (ch == '\\') { |
| 2342 | ch = nextEscaped(); |
| 2343 | if (ch == 'p' || ch == 'P') { // A property |
| 2344 | boolean comp = (ch == 'P'); |
| 2345 | boolean oneLetter = true; |
| 2346 | // Consume { if present |
| 2347 | ch = next(); |
| 2348 | if (ch != '{') |
| 2349 | unread(); |
| 2350 | else |
| 2351 | oneLetter = false; |
| 2352 | return family(oneLetter).maybeComplement(comp); |
| 2353 | } else { // ordinary escape |
| 2354 | unread(); |
| 2355 | ch = escape(true, true); |
| 2356 | if (ch == -1) |
| 2357 | return (CharProperty) root; |
| 2358 | } |
| 2359 | } else { |
| 2360 | ch = single(); |
| 2361 | } |
| 2362 | if (ch >= 0) { |
| 2363 | if (peek() == '-') { |
| 2364 | int endRange = temp[cursor+1]; |
| 2365 | if (endRange == '[') { |
| 2366 | return bitsOrSingle(bits, ch); |
| 2367 | } |
| 2368 | if (endRange != ']') { |
| 2369 | next(); |
| 2370 | int m = single(); |
| 2371 | if (m < ch) |
| 2372 | throw error("Illegal character range"); |
| 2373 | if (has(CASE_INSENSITIVE)) |
| 2374 | return caseInsensitiveRangeFor(ch, m); |
| 2375 | else |
| 2376 | return rangeFor(ch, m); |
| 2377 | } |
| 2378 | } |
| 2379 | return bitsOrSingle(bits, ch); |
| 2380 | } |
| 2381 | throw error("Unexpected character '"+((char)ch)+"'"); |
| 2382 | } |
| 2383 | |
| 2384 | private int single() { |
| 2385 | int ch = peek(); |
| 2386 | switch (ch) { |
| 2387 | case '\\': |
| 2388 | return escape(true, false); |
| 2389 | default: |
| 2390 | next(); |
| 2391 | return ch; |
| 2392 | } |
| 2393 | } |
| 2394 | |
| 2395 | /** |
| 2396 | * Parses a Unicode character family and returns its representative node. |
| 2397 | */ |
| 2398 | private CharProperty family(boolean singleLetter) { |
| 2399 | next(); |
| 2400 | String name; |
| 2401 | |
| 2402 | if (singleLetter) { |
| 2403 | int c = temp[cursor]; |
| 2404 | if (!Character.isSupplementaryCodePoint(c)) { |
| 2405 | name = String.valueOf((char)c); |
| 2406 | } else { |
| 2407 | name = new String(temp, cursor, 1); |
| 2408 | } |
| 2409 | read(); |
| 2410 | } else { |
| 2411 | int i = cursor; |
| 2412 | mark('}'); |
| 2413 | while(read() != '}') { |
| 2414 | } |
| 2415 | mark('\000'); |
| 2416 | int j = cursor; |
| 2417 | if (j > patternLength) |
| 2418 | throw error("Unclosed character family"); |
| 2419 | if (i + 1 >= j) |
| 2420 | throw error("Empty character family"); |
| 2421 | name = new String(temp, i, j-i-1); |
| 2422 | } |
| 2423 | |
| 2424 | if (name.startsWith("In")) { |
| 2425 | return unicodeBlockPropertyFor(name.substring(2)); |
| 2426 | } else { |
| 2427 | if (name.startsWith("Is")) |
| 2428 | name = name.substring(2); |
| 2429 | return charPropertyNodeFor(name); |
| 2430 | } |
| 2431 | } |
| 2432 | |
| 2433 | /** |
| 2434 | * Returns a CharProperty matching all characters in a UnicodeBlock. |
| 2435 | */ |
| 2436 | private CharProperty unicodeBlockPropertyFor(String name) { |
| 2437 | final Character.UnicodeBlock block; |
| 2438 | try { |
| 2439 | block = Character.UnicodeBlock.forName(name); |
| 2440 | } catch (IllegalArgumentException iae) { |
| 2441 | throw error("Unknown character block name {" + name + "}"); |
| 2442 | } |
| 2443 | return new CharProperty() { |
| 2444 | boolean isSatisfiedBy(int ch) { |
| 2445 | return block == Character.UnicodeBlock.of(ch);}}; |
| 2446 | } |
| 2447 | |
| 2448 | /** |
| 2449 | * Returns a CharProperty matching all characters in a named property. |
| 2450 | */ |
| 2451 | private CharProperty charPropertyNodeFor(String name) { |
| 2452 | CharProperty p = CharPropertyNames.charPropertyFor(name); |
| 2453 | if (p == null) |
| 2454 | throw error("Unknown character property name {" + name + "}"); |
| 2455 | return p; |
| 2456 | } |
| 2457 | |
| 2458 | /** |
| 2459 | * Parses a group and returns the head node of a set of nodes that process |
| 2460 | * the group. Sometimes a double return system is used where the tail is |
| 2461 | * returned in root. |
| 2462 | */ |
| 2463 | private Node group0() { |
| 2464 | boolean capturingGroup = false; |
| 2465 | Node head = null; |
| 2466 | Node tail = null; |
| 2467 | int save = flags; |
| 2468 | root = null; |
| 2469 | int ch = next(); |
| 2470 | if (ch == '?') { |
| 2471 | ch = skip(); |
| 2472 | switch (ch) { |
| 2473 | case ':': // (?:xxx) pure group |
| 2474 | head = createGroup(true); |
| 2475 | tail = root; |
| 2476 | head.next = expr(tail); |
| 2477 | break; |
| 2478 | case '=': // (?=xxx) and (?!xxx) lookahead |
| 2479 | case '!': |
| 2480 | head = createGroup(true); |
| 2481 | tail = root; |
| 2482 | head.next = expr(tail); |
| 2483 | if (ch == '=') { |
| 2484 | head = tail = new Pos(head); |
| 2485 | } else { |
| 2486 | head = tail = new Neg(head); |
| 2487 | } |
| 2488 | break; |
| 2489 | case '>': // (?>xxx) independent group |
| 2490 | head = createGroup(true); |
| 2491 | tail = root; |
| 2492 | head.next = expr(tail); |
| 2493 | head = tail = new Ques(head, INDEPENDENT); |
| 2494 | break; |
| 2495 | case '<': // (?<xxx) look behind |
| 2496 | ch = read(); |
| 2497 | int start = cursor; |
| 2498 | head = createGroup(true); |
| 2499 | tail = root; |
| 2500 | head.next = expr(tail); |
| 2501 | tail.next = lookbehindEnd; |
| 2502 | TreeInfo info = new TreeInfo(); |
| 2503 | head.study(info); |
| 2504 | if (info.maxValid == false) { |
| 2505 | throw error("Look-behind group does not have " |
| 2506 | + "an obvious maximum length"); |
| 2507 | } |
| 2508 | boolean hasSupplementary = findSupplementary(start, patternLength); |
| 2509 | if (ch == '=') { |
| 2510 | head = tail = (hasSupplementary ? |
| 2511 | new BehindS(head, info.maxLength, |
| 2512 | info.minLength) : |
| 2513 | new Behind(head, info.maxLength, |
| 2514 | info.minLength)); |
| 2515 | } else if (ch == '!') { |
| 2516 | head = tail = (hasSupplementary ? |
| 2517 | new NotBehindS(head, info.maxLength, |
| 2518 | info.minLength) : |
| 2519 | new NotBehind(head, info.maxLength, |
| 2520 | info.minLength)); |
| 2521 | } else { |
| 2522 | throw error("Unknown look-behind group"); |
| 2523 | } |
| 2524 | break; |
| 2525 | case '$': |
| 2526 | case '@': |
| 2527 | throw error("Unknown group type"); |
| 2528 | default: // (?xxx:) inlined match flags |
| 2529 | unread(); |
| 2530 | addFlag(); |
| 2531 | ch = read(); |
| 2532 | if (ch == ')') { |
| 2533 | return null; // Inline modifier only |
| 2534 | } |
| 2535 | if (ch != ':') { |
| 2536 | throw error("Unknown inline modifier"); |
| 2537 | } |
| 2538 | head = createGroup(true); |
| 2539 | tail = root; |
| 2540 | head.next = expr(tail); |
| 2541 | break; |
| 2542 | } |
| 2543 | } else { // (xxx) a regular group |
| 2544 | capturingGroup = true; |
| 2545 | head = createGroup(false); |
| 2546 | tail = root; |
| 2547 | head.next = expr(tail); |
| 2548 | } |
| 2549 | |
| 2550 | accept(')', "Unclosed group"); |
| 2551 | flags = save; |
| 2552 | |
| 2553 | // Check for quantifiers |
| 2554 | Node node = closure(head); |
| 2555 | if (node == head) { // No closure |
| 2556 | root = tail; |
| 2557 | return node; // Dual return |
| 2558 | } |
| 2559 | if (head == tail) { // Zero length assertion |
| 2560 | root = node; |
| 2561 | return node; // Dual return |
| 2562 | } |
| 2563 | |
| 2564 | if (node instanceof Ques) { |
| 2565 | Ques ques = (Ques) node; |
| 2566 | if (ques.type == POSSESSIVE) { |
| 2567 | root = node; |
| 2568 | return node; |
| 2569 | } |
| 2570 | tail.next = new BranchConn(); |
| 2571 | tail = tail.next; |
| 2572 | if (ques.type == GREEDY) { |
| 2573 | head = new Branch(head, null, tail); |
| 2574 | } else { // Reluctant quantifier |
| 2575 | head = new Branch(null, head, tail); |
| 2576 | } |
| 2577 | root = tail; |
| 2578 | return head; |
| 2579 | } else if (node instanceof Curly) { |
| 2580 | Curly curly = (Curly) node; |
| 2581 | if (curly.type == POSSESSIVE) { |
| 2582 | root = node; |
| 2583 | return node; |
| 2584 | } |
| 2585 | // Discover if the group is deterministic |
| 2586 | TreeInfo info = new TreeInfo(); |
| 2587 | if (head.study(info)) { // Deterministic |
| 2588 | GroupTail temp = (GroupTail) tail; |
| 2589 | head = root = new GroupCurly(head.next, curly.cmin, |
| 2590 | curly.cmax, curly.type, |
| 2591 | ((GroupTail)tail).localIndex, |
| 2592 | ((GroupTail)tail).groupIndex, |
| 2593 | capturingGroup); |
| 2594 | return head; |
| 2595 | } else { // Non-deterministic |
| 2596 | int temp = ((GroupHead) head).localIndex; |
| 2597 | Loop loop; |
| 2598 | if (curly.type == GREEDY) |
| 2599 | loop = new Loop(this.localCount, temp); |
| 2600 | else // Reluctant Curly |
| 2601 | loop = new LazyLoop(this.localCount, temp); |
| 2602 | Prolog prolog = new Prolog(loop); |
| 2603 | this.localCount += 1; |
| 2604 | loop.cmin = curly.cmin; |
| 2605 | loop.cmax = curly.cmax; |
| 2606 | loop.body = head; |
| 2607 | tail.next = loop; |
| 2608 | root = loop; |
| 2609 | return prolog; // Dual return |
| 2610 | } |
| 2611 | } |
| 2612 | throw error("Internal logic error"); |
| 2613 | } |
| 2614 | |
| 2615 | /** |
| 2616 | * Create group head and tail nodes using double return. If the group is |
| 2617 | * created with anonymous true then it is a pure group and should not |
| 2618 | * affect group counting. |
| 2619 | */ |
| 2620 | private Node createGroup(boolean anonymous) { |
| 2621 | int localIndex = localCount++; |
| 2622 | int groupIndex = 0; |
| 2623 | if (!anonymous) |
| 2624 | groupIndex = capturingGroupCount++; |
| 2625 | GroupHead head = new GroupHead(localIndex); |
| 2626 | root = new GroupTail(localIndex, groupIndex); |
| 2627 | if (!anonymous && groupIndex < 10) |
| 2628 | groupNodes[groupIndex] = head; |
| 2629 | return head; |
| 2630 | } |
| 2631 | |
| 2632 | /** |
| 2633 | * Parses inlined match flags and set them appropriately. |
| 2634 | */ |
| 2635 | private void addFlag() { |
| 2636 | int ch = peek(); |
| 2637 | for (;;) { |
| 2638 | switch (ch) { |
| 2639 | case 'i': |
| 2640 | flags |= CASE_INSENSITIVE; |
| 2641 | break; |
| 2642 | case 'm': |
| 2643 | flags |= MULTILINE; |
| 2644 | break; |
| 2645 | case 's': |
| 2646 | flags |= DOTALL; |
| 2647 | break; |
| 2648 | case 'd': |
| 2649 | flags |= UNIX_LINES; |
| 2650 | break; |
| 2651 | case 'u': |
| 2652 | flags |= UNICODE_CASE; |
| 2653 | break; |
| 2654 | case 'c': |
| 2655 | flags |= CANON_EQ; |
| 2656 | break; |
| 2657 | case 'x': |
| 2658 | flags |= COMMENTS; |
| 2659 | break; |
| 2660 | case '-': // subFlag then fall through |
| 2661 | ch = next(); |
| 2662 | subFlag(); |
| 2663 | default: |
| 2664 | return; |
| 2665 | } |
| 2666 | ch = next(); |
| 2667 | } |
| 2668 | } |
| 2669 | |
| 2670 | /** |
| 2671 | * Parses the second part of inlined match flags and turns off |
| 2672 | * flags appropriately. |
| 2673 | */ |
| 2674 | private void subFlag() { |
| 2675 | int ch = peek(); |
| 2676 | for (;;) { |
| 2677 | switch (ch) { |
| 2678 | case 'i': |
| 2679 | flags &= ~CASE_INSENSITIVE; |
| 2680 | break; |
| 2681 | case 'm': |
| 2682 | flags &= ~MULTILINE; |
| 2683 | break; |
| 2684 | case 's': |
| 2685 | flags &= ~DOTALL; |
| 2686 | break; |
| 2687 | case 'd': |
| 2688 | flags &= ~UNIX_LINES; |
| 2689 | break; |
| 2690 | case 'u': |
| 2691 | flags &= ~UNICODE_CASE; |
| 2692 | break; |
| 2693 | case 'c': |
| 2694 | flags &= ~CANON_EQ; |
| 2695 | break; |
| 2696 | case 'x': |
| 2697 | flags &= ~COMMENTS; |
| 2698 | break; |
| 2699 | default: |
| 2700 | return; |
| 2701 | } |
| 2702 | ch = next(); |
| 2703 | } |
| 2704 | } |
| 2705 | |
| 2706 | static final int MAX_REPS = 0x7FFFFFFF; |
| 2707 | |
| 2708 | static final int GREEDY = 0; |
| 2709 | |
| 2710 | static final int LAZY = 1; |
| 2711 | |
| 2712 | static final int POSSESSIVE = 2; |
| 2713 | |
| 2714 | static final int INDEPENDENT = 3; |
| 2715 | |
| 2716 | /** |
| 2717 | * Processes repetition. If the next character peeked is a quantifier |
| 2718 | * then new nodes must be appended to handle the repetition. |
| 2719 | * Prev could be a single or a group, so it could be a chain of nodes. |
| 2720 | */ |
| 2721 | private Node closure(Node prev) { |
| 2722 | Node atom; |
| 2723 | int ch = peek(); |
| 2724 | switch (ch) { |
| 2725 | case '?': |
| 2726 | ch = next(); |
| 2727 | if (ch == '?') { |
| 2728 | next(); |
| 2729 | return new Ques(prev, LAZY); |
| 2730 | } else if (ch == '+') { |
| 2731 | next(); |
| 2732 | return new Ques(prev, POSSESSIVE); |
| 2733 | } |
| 2734 | return new Ques(prev, GREEDY); |
| 2735 | case '*': |
| 2736 | ch = next(); |
| 2737 | if (ch == '?') { |
| 2738 | next(); |
| 2739 | return new Curly(prev, 0, MAX_REPS, LAZY); |
| 2740 | } else if (ch == '+') { |
| 2741 | next(); |
| 2742 | return new Curly(prev, 0, MAX_REPS, POSSESSIVE); |
| 2743 | } |
| 2744 | return new Curly(prev, 0, MAX_REPS, GREEDY); |
| 2745 | case '+': |
| 2746 | ch = next(); |
| 2747 | if (ch == '?') { |
| 2748 | next(); |
| 2749 | return new Curly(prev, 1, MAX_REPS, LAZY); |
| 2750 | } else if (ch == '+') { |
| 2751 | next(); |
| 2752 | return new Curly(prev, 1, MAX_REPS, POSSESSIVE); |
| 2753 | } |
| 2754 | return new Curly(prev, 1, MAX_REPS, GREEDY); |
| 2755 | case '{': |
| 2756 | ch = temp[cursor+1]; |
| 2757 | if (ASCII.isDigit(ch)) { |
| 2758 | skip(); |
| 2759 | int cmin = 0; |
| 2760 | do { |
| 2761 | cmin = cmin * 10 + (ch - '0'); |
| 2762 | } while (ASCII.isDigit(ch = read())); |
| 2763 | int cmax = cmin; |
| 2764 | if (ch == ',') { |
| 2765 | ch = read(); |
| 2766 | cmax = MAX_REPS; |
| 2767 | if (ch != '}') { |
| 2768 | cmax = 0; |
| 2769 | while (ASCII.isDigit(ch)) { |
| 2770 | cmax = cmax * 10 + (ch - '0'); |
| 2771 | ch = read(); |
| 2772 | } |
| 2773 | } |
| 2774 | } |
| 2775 | if (ch != '}') |
| 2776 | throw error("Unclosed counted closure"); |
| 2777 | if (((cmin) | (cmax) | (cmax - cmin)) < 0) |
| 2778 | throw error("Illegal repetition range"); |
| 2779 | Curly curly; |
| 2780 | ch = peek(); |
| 2781 | if (ch == '?') { |
| 2782 | next(); |
| 2783 | curly = new Curly(prev, cmin, cmax, LAZY); |
| 2784 | } else if (ch == '+') { |
| 2785 | next(); |
| 2786 | curly = new Curly(prev, cmin, cmax, POSSESSIVE); |
| 2787 | } else { |
| 2788 | curly = new Curly(prev, cmin, cmax, GREEDY); |
| 2789 | } |
| 2790 | return curly; |
| 2791 | } else { |
| 2792 | throw error("Illegal repetition"); |
| 2793 | } |
| 2794 | default: |
| 2795 | return prev; |
| 2796 | } |
| 2797 | } |
| 2798 | |
| 2799 | /** |
| 2800 | * Utility method for parsing control escape sequences. |
| 2801 | */ |
| 2802 | private int c() { |
| 2803 | if (cursor < patternLength) { |
| 2804 | return read() ^ 64; |
| 2805 | } |
| 2806 | throw error("Illegal control escape sequence"); |
| 2807 | } |
| 2808 | |
| 2809 | /** |
| 2810 | * Utility method for parsing octal escape sequences. |
| 2811 | */ |
| 2812 | private int o() { |
| 2813 | int n = read(); |
| 2814 | if (((n-'0')|('7'-n)) >= 0) { |
| 2815 | int m = read(); |
| 2816 | if (((m-'0')|('7'-m)) >= 0) { |
| 2817 | int o = read(); |
| 2818 | if ((((o-'0')|('7'-o)) >= 0) && (((n-'0')|('3'-n)) >= 0)) { |
| 2819 | return (n - '0') * 64 + (m - '0') * 8 + (o - '0'); |
| 2820 | } |
| 2821 | unread(); |
| 2822 | return (n - '0') * 8 + (m - '0'); |
| 2823 | } |
| 2824 | unread(); |
| 2825 | return (n - '0'); |
| 2826 | } |
| 2827 | throw error("Illegal octal escape sequence"); |
| 2828 | } |
| 2829 | |
| 2830 | /** |
| 2831 | * Utility method for parsing hexadecimal escape sequences. |
| 2832 | */ |
| 2833 | private int x() { |
| 2834 | int n = read(); |
| 2835 | if (ASCII.isHexDigit(n)) { |
| 2836 | int m = read(); |
| 2837 | if (ASCII.isHexDigit(m)) { |
| 2838 | return ASCII.toDigit(n) * 16 + ASCII.toDigit(m); |
| 2839 | } |
| 2840 | } |
| 2841 | throw error("Illegal hexadecimal escape sequence"); |
| 2842 | } |
| 2843 | |
| 2844 | /** |
| 2845 | * Utility method for parsing unicode escape sequences. |
| 2846 | */ |
| 2847 | private int u() { |
| 2848 | int n = 0; |
| 2849 | for (int i = 0; i < 4; i++) { |
| 2850 | int ch = read(); |
| 2851 | if (!ASCII.isHexDigit(ch)) { |
| 2852 | throw error("Illegal Unicode escape sequence"); |
| 2853 | } |
| 2854 | n = n * 16 + ASCII.toDigit(ch); |
| 2855 | } |
| 2856 | return n; |
| 2857 | } |
| 2858 | |
| 2859 | // |
| 2860 | // Utility methods for code point support |
| 2861 | // |
| 2862 | |
| 2863 | /** |
| 2864 | * Tests a surrogate value. |
| 2865 | */ |
| 2866 | private static final boolean isSurrogate(int c) { |
| 2867 | return c >= Character.MIN_HIGH_SURROGATE && c <= Character.MAX_LOW_SURROGATE; |
| 2868 | } |
| 2869 | |
| 2870 | private static final int countChars(CharSequence seq, int index, |
| 2871 | int lengthInCodePoints) { |
| 2872 | // optimization |
| 2873 | if (lengthInCodePoints == 1 && !Character.isHighSurrogate(seq.charAt(index))) { |
| 2874 | assert (index >= 0 && index < seq.length()); |
| 2875 | return 1; |
| 2876 | } |
| 2877 | int length = seq.length(); |
| 2878 | int x = index; |
| 2879 | if (lengthInCodePoints >= 0) { |
| 2880 | assert (index >= 0 && index < length); |
| 2881 | for (int i = 0; x < length && i < lengthInCodePoints; i++) { |
| 2882 | if (Character.isHighSurrogate(seq.charAt(x++))) { |
| 2883 | if (x < length && Character.isLowSurrogate(seq.charAt(x))) { |
| 2884 | x++; |
| 2885 | } |
| 2886 | } |
| 2887 | } |
| 2888 | return x - index; |
| 2889 | } |
| 2890 | |
| 2891 | assert (index >= 0 && index <= length); |
| 2892 | if (index == 0) { |
| 2893 | return 0; |
| 2894 | } |
| 2895 | int len = -lengthInCodePoints; |
| 2896 | for (int i = 0; x > 0 && i < len; i++) { |
| 2897 | if (Character.isLowSurrogate(seq.charAt(--x))) { |
| 2898 | if (x > 0 && Character.isHighSurrogate(seq.charAt(x-1))) { |
| 2899 | x--; |
| 2900 | } |
| 2901 | } |
| 2902 | } |
| 2903 | return index - x; |
| 2904 | } |
| 2905 | |
| 2906 | private static final int countCodePoints(CharSequence seq) { |
| 2907 | int length = seq.length(); |
| 2908 | int n = 0; |
| 2909 | for (int i = 0; i < length; ) { |
| 2910 | n++; |
| 2911 | if (Character.isHighSurrogate(seq.charAt(i++))) { |
| 2912 | if (i < length && Character.isLowSurrogate(seq.charAt(i))) { |
| 2913 | i++; |
| 2914 | } |
| 2915 | } |
| 2916 | } |
| 2917 | return n; |
| 2918 | } |
| 2919 | |
| 2920 | /** |
| 2921 | * Creates a bit vector for matching Latin-1 values. A normal BitClass |
| 2922 | * never matches values above Latin-1, and a complemented BitClass always |
| 2923 | * matches values above Latin-1. |
| 2924 | */ |
| 2925 | private static final class BitClass extends BmpCharProperty { |
| 2926 | final boolean[] bits; |
| 2927 | BitClass() { bits = new boolean[256]; } |
| 2928 | private BitClass(boolean[] bits) { this.bits = bits; } |
| 2929 | BitClass add(int c, int flags) { |
| 2930 | assert c >= 0 && c <= 255; |
| 2931 | if ((flags & CASE_INSENSITIVE) != 0) { |
| 2932 | if (ASCII.isAscii(c)) { |
| 2933 | bits[ASCII.toUpper(c)] = true; |
| 2934 | bits[ASCII.toLower(c)] = true; |
| 2935 | } else if ((flags & UNICODE_CASE) != 0) { |
| 2936 | bits[Character.toLowerCase(c)] = true; |
| 2937 | bits[Character.toUpperCase(c)] = true; |
| 2938 | } |
| 2939 | } |
| 2940 | bits[c] = true; |
| 2941 | return this; |
| 2942 | } |
| 2943 | boolean isSatisfiedBy(int ch) { |
| 2944 | return ch < 256 && bits[ch]; |
| 2945 | } |
| 2946 | } |
| 2947 | |
| 2948 | /** |
| 2949 | * Returns a suitably optimized, single character matcher. |
| 2950 | */ |
| 2951 | private CharProperty newSingle(final int ch) { |
| 2952 | if (has(CASE_INSENSITIVE)) { |
| 2953 | int lower, upper; |
| 2954 | if (has(UNICODE_CASE)) { |
| 2955 | upper = Character.toUpperCase(ch); |
| 2956 | lower = Character.toLowerCase(upper); |
| 2957 | if (upper != lower) |
| 2958 | return new SingleU(lower); |
| 2959 | } else if (ASCII.isAscii(ch)) { |
| 2960 | lower = ASCII.toLower(ch); |
| 2961 | upper = ASCII.toUpper(ch); |
| 2962 | if (lower != upper) |
| 2963 | return new SingleI(lower, upper); |
| 2964 | } |
| 2965 | } |
| 2966 | if (isSupplementary(ch)) |
| 2967 | return new SingleS(ch); // Match a given Unicode character |
| 2968 | return new Single(ch); // Match a given BMP character |
| 2969 | } |
| 2970 | |
| 2971 | /** |
| 2972 | * Utility method for creating a string slice matcher. |
| 2973 | */ |
| 2974 | private Node newSlice(int[] buf, int count, boolean hasSupplementary) { |
| 2975 | int[] tmp = new int[count]; |
| 2976 | if (has(CASE_INSENSITIVE)) { |
| 2977 | if (has(UNICODE_CASE)) { |
| 2978 | for (int i = 0; i < count; i++) { |
| 2979 | tmp[i] = Character.toLowerCase( |
| 2980 | Character.toUpperCase(buf[i])); |
| 2981 | } |
| 2982 | return hasSupplementary? new SliceUS(tmp) : new SliceU(tmp); |
| 2983 | } |
| 2984 | for (int i = 0; i < count; i++) { |
| 2985 | tmp[i] = ASCII.toLower(buf[i]); |
| 2986 | } |
| 2987 | return hasSupplementary? new SliceIS(tmp) : new SliceI(tmp); |
| 2988 | } |
| 2989 | for (int i = 0; i < count; i++) { |
| 2990 | tmp[i] = buf[i]; |
| 2991 | } |
| 2992 | return hasSupplementary ? new SliceS(tmp) : new Slice(tmp); |
| 2993 | } |
| 2994 | |
| 2995 | /** |
| 2996 | * The following classes are the building components of the object |
| 2997 | * tree that represents a compiled regular expression. The object tree |
| 2998 | * is made of individual elements that handle constructs in the Pattern. |
| 2999 | * Each type of object knows how to match its equivalent construct with |
| 3000 | * the match() method. |
| 3001 | */ |
| 3002 | |
| 3003 | /** |
| 3004 | * Base class for all node classes. Subclasses should override the match() |
| 3005 | * method as appropriate. This class is an accepting node, so its match() |
| 3006 | * always returns true. |
| 3007 | */ |
| 3008 | static class Node extends Object { |
| 3009 | Node next; |
| 3010 | Node() { |
| 3011 | next = Pattern.accept; |
| 3012 | } |
| 3013 | /** |
| 3014 | * This method implements the classic accept node. |
| 3015 | */ |
| 3016 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3017 | matcher.last = i; |
| 3018 | matcher.groups[0] = matcher.first; |
| 3019 | matcher.groups[1] = matcher.last; |
| 3020 | return true; |
| 3021 | } |
| 3022 | /** |
| 3023 | * This method is good for all zero length assertions. |
| 3024 | */ |
| 3025 | boolean study(TreeInfo info) { |
| 3026 | if (next != null) { |
| 3027 | return next.study(info); |
| 3028 | } else { |
| 3029 | return info.deterministic; |
| 3030 | } |
| 3031 | } |
| 3032 | } |
| 3033 | |
| 3034 | static class LastNode extends Node { |
| 3035 | /** |
| 3036 | * This method implements the classic accept node with |
| 3037 | * the addition of a check to see if the match occurred |
| 3038 | * using all of the input. |
| 3039 | */ |
| 3040 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3041 | if (matcher.acceptMode == Matcher.ENDANCHOR && i != matcher.to) |
| 3042 | return false; |
| 3043 | matcher.last = i; |
| 3044 | matcher.groups[0] = matcher.first; |
| 3045 | matcher.groups[1] = matcher.last; |
| 3046 | return true; |
| 3047 | } |
| 3048 | } |
| 3049 | |
| 3050 | /** |
| 3051 | * Used for REs that can start anywhere within the input string. |
| 3052 | * This basically tries to match repeatedly at each spot in the |
| 3053 | * input string, moving forward after each try. An anchored search |
| 3054 | * or a BnM will bypass this node completely. |
| 3055 | */ |
| 3056 | static class Start extends Node { |
| 3057 | int minLength; |
| 3058 | Start(Node node) { |
| 3059 | this.next = node; |
| 3060 | TreeInfo info = new TreeInfo(); |
| 3061 | next.study(info); |
| 3062 | minLength = info.minLength; |
| 3063 | } |
| 3064 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3065 | if (i > matcher.to - minLength) { |
| 3066 | matcher.hitEnd = true; |
| 3067 | return false; |
| 3068 | } |
| 3069 | boolean ret = false; |
| 3070 | int guard = matcher.to - minLength; |
| 3071 | for (; i <= guard; i++) { |
| 3072 | if (ret = next.match(matcher, i, seq)) |
| 3073 | break; |
| 3074 | if (i == guard) |
| 3075 | matcher.hitEnd = true; |
| 3076 | } |
| 3077 | if (ret) { |
| 3078 | matcher.first = i; |
| 3079 | matcher.groups[0] = matcher.first; |
| 3080 | matcher.groups[1] = matcher.last; |
| 3081 | } |
| 3082 | return ret; |
| 3083 | } |
| 3084 | boolean study(TreeInfo info) { |
| 3085 | next.study(info); |
| 3086 | info.maxValid = false; |
| 3087 | info.deterministic = false; |
| 3088 | return false; |
| 3089 | } |
| 3090 | } |
| 3091 | |
| 3092 | /* |
| 3093 | * StartS supports supplementary characters, including unpaired surrogates. |
| 3094 | */ |
| 3095 | static final class StartS extends Start { |
| 3096 | StartS(Node node) { |
| 3097 | super(node); |
| 3098 | } |
| 3099 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3100 | if (i > matcher.to - minLength) { |
| 3101 | matcher.hitEnd = true; |
| 3102 | return false; |
| 3103 | } |
| 3104 | boolean ret = false; |
| 3105 | int guard = matcher.to - minLength; |
| 3106 | while (i <= guard) { |
| 3107 | if ((ret = next.match(matcher, i, seq)) || i == guard) |
| 3108 | break; |
| 3109 | // Optimization to move to the next character. This is |
| 3110 | // faster than countChars(seq, i, 1). |
| 3111 | if (Character.isHighSurrogate(seq.charAt(i++))) { |
| 3112 | if (i < seq.length() && Character.isLowSurrogate(seq.charAt(i))) { |
| 3113 | i++; |
| 3114 | } |
| 3115 | } |
| 3116 | if (i == guard) |
| 3117 | matcher.hitEnd = true; |
| 3118 | } |
| 3119 | if (ret) { |
| 3120 | matcher.first = i; |
| 3121 | matcher.groups[0] = matcher.first; |
| 3122 | matcher.groups[1] = matcher.last; |
| 3123 | } |
| 3124 | return ret; |
| 3125 | } |
| 3126 | } |
| 3127 | |
| 3128 | /** |
| 3129 | * Node to anchor at the beginning of input. This object implements the |
| 3130 | * match for a \A sequence, and the caret anchor will use this if not in |
| 3131 | * multiline mode. |
| 3132 | */ |
| 3133 | static final class Begin extends Node { |
| 3134 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3135 | int fromIndex = (matcher.anchoringBounds) ? |
| 3136 | matcher.from : 0; |
| 3137 | if (i == fromIndex && next.match(matcher, i, seq)) { |
| 3138 | matcher.first = i; |
| 3139 | matcher.groups[0] = i; |
| 3140 | matcher.groups[1] = matcher.last; |
| 3141 | return true; |
| 3142 | } else { |
| 3143 | return false; |
| 3144 | } |
| 3145 | } |
| 3146 | } |
| 3147 | |
| 3148 | /** |
| 3149 | * Node to anchor at the end of input. This is the absolute end, so this |
| 3150 | * should not match at the last newline before the end as $ will. |
| 3151 | */ |
| 3152 | static final class End extends Node { |
| 3153 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3154 | int endIndex = (matcher.anchoringBounds) ? |
| 3155 | matcher.to : matcher.getTextLength(); |
| 3156 | if (i == endIndex) { |
| 3157 | matcher.hitEnd = true; |
| 3158 | return next.match(matcher, i, seq); |
| 3159 | } |
| 3160 | return false; |
| 3161 | } |
| 3162 | } |
| 3163 | |
| 3164 | /** |
| 3165 | * Node to anchor at the beginning of a line. This is essentially the |
| 3166 | * object to match for the multiline ^. |
| 3167 | */ |
| 3168 | static final class Caret extends Node { |
| 3169 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3170 | int startIndex = matcher.from; |
| 3171 | int endIndex = matcher.to; |
| 3172 | if (!matcher.anchoringBounds) { |
| 3173 | startIndex = 0; |
| 3174 | endIndex = matcher.getTextLength(); |
| 3175 | } |
| 3176 | // Perl does not match ^ at end of input even after newline |
| 3177 | if (i == endIndex) { |
| 3178 | matcher.hitEnd = true; |
| 3179 | return false; |
| 3180 | } |
| 3181 | if (i > startIndex) { |
| 3182 | char ch = seq.charAt(i-1); |
| 3183 | if (ch != '\n' && ch != '\r' |
| 3184 | && (ch|1) != '\u2029' |
| 3185 | && ch != '\u0085' ) { |
| 3186 | return false; |
| 3187 | } |
| 3188 | // Should treat /r/n as one newline |
| 3189 | if (ch == '\r' && seq.charAt(i) == '\n') |
| 3190 | return false; |
| 3191 | } |
| 3192 | return next.match(matcher, i, seq); |
| 3193 | } |
| 3194 | } |
| 3195 | |
| 3196 | /** |
| 3197 | * Node to anchor at the beginning of a line when in unixdot mode. |
| 3198 | */ |
| 3199 | static final class UnixCaret extends Node { |
| 3200 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3201 | int startIndex = matcher.from; |
| 3202 | int endIndex = matcher.to; |
| 3203 | if (!matcher.anchoringBounds) { |
| 3204 | startIndex = 0; |
| 3205 | endIndex = matcher.getTextLength(); |
| 3206 | } |
| 3207 | // Perl does not match ^ at end of input even after newline |
| 3208 | if (i == endIndex) { |
| 3209 | matcher.hitEnd = true; |
| 3210 | return false; |
| 3211 | } |
| 3212 | if (i > startIndex) { |
| 3213 | char ch = seq.charAt(i-1); |
| 3214 | if (ch != '\n') { |
| 3215 | return false; |
| 3216 | } |
| 3217 | } |
| 3218 | return next.match(matcher, i, seq); |
| 3219 | } |
| 3220 | } |
| 3221 | |
| 3222 | /** |
| 3223 | * Node to match the location where the last match ended. |
| 3224 | * This is used for the \G construct. |
| 3225 | */ |
| 3226 | static final class LastMatch extends Node { |
| 3227 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3228 | if (i != matcher.oldLast) |
| 3229 | return false; |
| 3230 | return next.match(matcher, i, seq); |
| 3231 | } |
| 3232 | } |
| 3233 | |
| 3234 | /** |
| 3235 | * Node to anchor at the end of a line or the end of input based on the |
| 3236 | * multiline mode. |
| 3237 | * |
| 3238 | * When not in multiline mode, the $ can only match at the very end |
| 3239 | * of the input, unless the input ends in a line terminator in which |
| 3240 | * it matches right before the last line terminator. |
| 3241 | * |
| 3242 | * Note that \r\n is considered an atomic line terminator. |
| 3243 | * |
| 3244 | * Like ^ the $ operator matches at a position, it does not match the |
| 3245 | * line terminators themselves. |
| 3246 | */ |
| 3247 | static final class Dollar extends Node { |
| 3248 | boolean multiline; |
| 3249 | Dollar(boolean mul) { |
| 3250 | multiline = mul; |
| 3251 | } |
| 3252 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3253 | int endIndex = (matcher.anchoringBounds) ? |
| 3254 | matcher.to : matcher.getTextLength(); |
| 3255 | if (!multiline) { |
| 3256 | if (i < endIndex - 2) |
| 3257 | return false; |
| 3258 | if (i == endIndex - 2) { |
| 3259 | char ch = seq.charAt(i); |
| 3260 | if (ch != '\r') |
| 3261 | return false; |
| 3262 | ch = seq.charAt(i + 1); |
| 3263 | if (ch != '\n') |
| 3264 | return false; |
| 3265 | } |
| 3266 | } |
| 3267 | // Matches before any line terminator; also matches at the |
| 3268 | // end of input |
| 3269 | // Before line terminator: |
| 3270 | // If multiline, we match here no matter what |
| 3271 | // If not multiline, fall through so that the end |
| 3272 | // is marked as hit; this must be a /r/n or a /n |
| 3273 | // at the very end so the end was hit; more input |
| 3274 | // could make this not match here |
| 3275 | if (i < endIndex) { |
| 3276 | char ch = seq.charAt(i); |
| 3277 | if (ch == '\n') { |
| 3278 | // No match between \r\n |
| 3279 | if (i > 0 && seq.charAt(i-1) == '\r') |
| 3280 | return false; |
| 3281 | if (multiline) |
| 3282 | return next.match(matcher, i, seq); |
| 3283 | } else if (ch == '\r' || ch == '\u0085' || |
| 3284 | (ch|1) == '\u2029') { |
| 3285 | if (multiline) |
| 3286 | return next.match(matcher, i, seq); |
| 3287 | } else { // No line terminator, no match |
| 3288 | return false; |
| 3289 | } |
| 3290 | } |
| 3291 | // Matched at current end so hit end |
| 3292 | matcher.hitEnd = true; |
| 3293 | // If a $ matches because of end of input, then more input |
| 3294 | // could cause it to fail! |
| 3295 | matcher.requireEnd = true; |
| 3296 | return next.match(matcher, i, seq); |
| 3297 | } |
| 3298 | boolean study(TreeInfo info) { |
| 3299 | next.study(info); |
| 3300 | return info.deterministic; |
| 3301 | } |
| 3302 | } |
| 3303 | |
| 3304 | /** |
| 3305 | * Node to anchor at the end of a line or the end of input based on the |
| 3306 | * multiline mode when in unix lines mode. |
| 3307 | */ |
| 3308 | static final class UnixDollar extends Node { |
| 3309 | boolean multiline; |
| 3310 | UnixDollar(boolean mul) { |
| 3311 | multiline = mul; |
| 3312 | } |
| 3313 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3314 | int endIndex = (matcher.anchoringBounds) ? |
| 3315 | matcher.to : matcher.getTextLength(); |
| 3316 | if (i < endIndex) { |
| 3317 | char ch = seq.charAt(i); |
| 3318 | if (ch == '\n') { |
| 3319 | // If not multiline, then only possible to |
| 3320 | // match at very end or one before end |
| 3321 | if (multiline == false && i != endIndex - 1) |
| 3322 | return false; |
| 3323 | // If multiline return next.match without setting |
| 3324 | // matcher.hitEnd |
| 3325 | if (multiline) |
| 3326 | return next.match(matcher, i, seq); |
| 3327 | } else { |
| 3328 | return false; |
| 3329 | } |
| 3330 | } |
| 3331 | // Matching because at the end or 1 before the end; |
| 3332 | // more input could change this so set hitEnd |
| 3333 | matcher.hitEnd = true; |
| 3334 | // If a $ matches because of end of input, then more input |
| 3335 | // could cause it to fail! |
| 3336 | matcher.requireEnd = true; |
| 3337 | return next.match(matcher, i, seq); |
| 3338 | } |
| 3339 | boolean study(TreeInfo info) { |
| 3340 | next.study(info); |
| 3341 | return info.deterministic; |
| 3342 | } |
| 3343 | } |
| 3344 | |
| 3345 | /** |
| 3346 | * Abstract node class to match one character satisfying some |
| 3347 | * boolean property. |
| 3348 | */ |
| 3349 | private static abstract class CharProperty extends Node { |
| 3350 | abstract boolean isSatisfiedBy(int ch); |
| 3351 | CharProperty complement() { |
| 3352 | return new CharProperty() { |
| 3353 | boolean isSatisfiedBy(int ch) { |
| 3354 | return ! CharProperty.this.isSatisfiedBy(ch);}}; |
| 3355 | } |
| 3356 | CharProperty maybeComplement(boolean complement) { |
| 3357 | return complement ? complement() : this; |
| 3358 | } |
| 3359 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3360 | if (i < matcher.to) { |
| 3361 | int ch = Character.codePointAt(seq, i); |
| 3362 | return isSatisfiedBy(ch) |
| 3363 | && next.match(matcher, i+Character.charCount(ch), seq); |
| 3364 | } else { |
| 3365 | matcher.hitEnd = true; |
| 3366 | return false; |
| 3367 | } |
| 3368 | } |
| 3369 | boolean study(TreeInfo info) { |
| 3370 | info.minLength++; |
| 3371 | info.maxLength++; |
| 3372 | return next.study(info); |
| 3373 | } |
| 3374 | } |
| 3375 | |
| 3376 | /** |
| 3377 | * Optimized version of CharProperty that works only for |
| 3378 | * properties never satisfied by Supplementary characters. |
| 3379 | */ |
| 3380 | private static abstract class BmpCharProperty extends CharProperty { |
| 3381 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3382 | if (i < matcher.to) { |
| 3383 | return isSatisfiedBy(seq.charAt(i)) |
| 3384 | && next.match(matcher, i+1, seq); |
| 3385 | } else { |
| 3386 | matcher.hitEnd = true; |
| 3387 | return false; |
| 3388 | } |
| 3389 | } |
| 3390 | } |
| 3391 | |
| 3392 | /** |
| 3393 | * Node class that matches a Supplementary Unicode character |
| 3394 | */ |
| 3395 | static final class SingleS extends CharProperty { |
| 3396 | final int c; |
| 3397 | SingleS(int c) { this.c = c; } |
| 3398 | boolean isSatisfiedBy(int ch) { |
| 3399 | return ch == c; |
| 3400 | } |
| 3401 | } |
| 3402 | |
| 3403 | /** |
| 3404 | * Optimization -- matches a given BMP character |
| 3405 | */ |
| 3406 | static final class Single extends BmpCharProperty { |
| 3407 | final int c; |
| 3408 | Single(int c) { this.c = c; } |
| 3409 | boolean isSatisfiedBy(int ch) { |
| 3410 | return ch == c; |
| 3411 | } |
| 3412 | } |
| 3413 | |
| 3414 | /** |
| 3415 | * Case insensitive matches a given BMP character |
| 3416 | */ |
| 3417 | static final class SingleI extends BmpCharProperty { |
| 3418 | final int lower; |
| 3419 | final int upper; |
| 3420 | SingleI(int lower, int upper) { |
| 3421 | this.lower = lower; |
| 3422 | this.upper = upper; |
| 3423 | } |
| 3424 | boolean isSatisfiedBy(int ch) { |
| 3425 | return ch == lower || ch == upper; |
| 3426 | } |
| 3427 | } |
| 3428 | |
| 3429 | /** |
| 3430 | * Unicode case insensitive matches a given Unicode character |
| 3431 | */ |
| 3432 | static final class SingleU extends CharProperty { |
| 3433 | final int lower; |
| 3434 | SingleU(int lower) { |
| 3435 | this.lower = lower; |
| 3436 | } |
| 3437 | boolean isSatisfiedBy(int ch) { |
| 3438 | return lower == ch || |
| 3439 | lower == Character.toLowerCase(Character.toUpperCase(ch)); |
| 3440 | } |
| 3441 | } |
| 3442 | |
| 3443 | /** |
| 3444 | * Node class that matches a Unicode category. |
| 3445 | */ |
| 3446 | static final class Category extends CharProperty { |
| 3447 | final int typeMask; |
| 3448 | Category(int typeMask) { this.typeMask = typeMask; } |
| 3449 | boolean isSatisfiedBy(int ch) { |
| 3450 | return (typeMask & (1 << Character.getType(ch))) != 0; |
| 3451 | } |
| 3452 | } |
| 3453 | |
| 3454 | /** |
| 3455 | * Node class that matches a POSIX type. |
| 3456 | */ |
| 3457 | static final class Ctype extends BmpCharProperty { |
| 3458 | final int ctype; |
| 3459 | Ctype(int ctype) { this.ctype = ctype; } |
| 3460 | boolean isSatisfiedBy(int ch) { |
| 3461 | return ch < 128 && ASCII.isType(ch, ctype); |
| 3462 | } |
| 3463 | } |
| 3464 | |
| 3465 | /** |
| 3466 | * Base class for all Slice nodes |
| 3467 | */ |
| 3468 | static class SliceNode extends Node { |
| 3469 | int[] buffer; |
| 3470 | SliceNode(int[] buf) { |
| 3471 | buffer = buf; |
| 3472 | } |
| 3473 | boolean study(TreeInfo info) { |
| 3474 | info.minLength += buffer.length; |
| 3475 | info.maxLength += buffer.length; |
| 3476 | return next.study(info); |
| 3477 | } |
| 3478 | } |
| 3479 | |
| 3480 | /** |
| 3481 | * Node class for a case sensitive/BMP-only sequence of literal |
| 3482 | * characters. |
| 3483 | */ |
| 3484 | static final class Slice extends SliceNode { |
| 3485 | Slice(int[] buf) { |
| 3486 | super(buf); |
| 3487 | } |
| 3488 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3489 | int[] buf = buffer; |
| 3490 | int len = buf.length; |
| 3491 | for (int j=0; j<len; j++) { |
| 3492 | if ((i+j) >= matcher.to) { |
| 3493 | matcher.hitEnd = true; |
| 3494 | return false; |
| 3495 | } |
| 3496 | if (buf[j] != seq.charAt(i+j)) |
| 3497 | return false; |
| 3498 | } |
| 3499 | return next.match(matcher, i+len, seq); |
| 3500 | } |
| 3501 | } |
| 3502 | |
| 3503 | /** |
| 3504 | * Node class for a case_insensitive/BMP-only sequence of literal |
| 3505 | * characters. |
| 3506 | */ |
| 3507 | static class SliceI extends SliceNode { |
| 3508 | SliceI(int[] buf) { |
| 3509 | super(buf); |
| 3510 | } |
| 3511 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3512 | int[] buf = buffer; |
| 3513 | int len = buf.length; |
| 3514 | for (int j=0; j<len; j++) { |
| 3515 | if ((i+j) >= matcher.to) { |
| 3516 | matcher.hitEnd = true; |
| 3517 | return false; |
| 3518 | } |
| 3519 | int c = seq.charAt(i+j); |
| 3520 | if (buf[j] != c && |
| 3521 | buf[j] != ASCII.toLower(c)) |
| 3522 | return false; |
| 3523 | } |
| 3524 | return next.match(matcher, i+len, seq); |
| 3525 | } |
| 3526 | } |
| 3527 | |
| 3528 | /** |
| 3529 | * Node class for a unicode_case_insensitive/BMP-only sequence of |
| 3530 | * literal characters. Uses unicode case folding. |
| 3531 | */ |
| 3532 | static final class SliceU extends SliceNode { |
| 3533 | SliceU(int[] buf) { |
| 3534 | super(buf); |
| 3535 | } |
| 3536 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3537 | int[] buf = buffer; |
| 3538 | int len = buf.length; |
| 3539 | for (int j=0; j<len; j++) { |
| 3540 | if ((i+j) >= matcher.to) { |
| 3541 | matcher.hitEnd = true; |
| 3542 | return false; |
| 3543 | } |
| 3544 | int c = seq.charAt(i+j); |
| 3545 | if (buf[j] != c && |
| 3546 | buf[j] != Character.toLowerCase(Character.toUpperCase(c))) |
| 3547 | return false; |
| 3548 | } |
| 3549 | return next.match(matcher, i+len, seq); |
| 3550 | } |
| 3551 | } |
| 3552 | |
| 3553 | /** |
| 3554 | * Node class for a case sensitive sequence of literal characters |
| 3555 | * including supplementary characters. |
| 3556 | */ |
| 3557 | static final class SliceS extends SliceNode { |
| 3558 | SliceS(int[] buf) { |
| 3559 | super(buf); |
| 3560 | } |
| 3561 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3562 | int[] buf = buffer; |
| 3563 | int x = i; |
| 3564 | for (int j = 0; j < buf.length; j++) { |
| 3565 | if (x >= matcher.to) { |
| 3566 | matcher.hitEnd = true; |
| 3567 | return false; |
| 3568 | } |
| 3569 | int c = Character.codePointAt(seq, x); |
| 3570 | if (buf[j] != c) |
| 3571 | return false; |
| 3572 | x += Character.charCount(c); |
| 3573 | if (x > matcher.to) { |
| 3574 | matcher.hitEnd = true; |
| 3575 | return false; |
| 3576 | } |
| 3577 | } |
| 3578 | return next.match(matcher, x, seq); |
| 3579 | } |
| 3580 | } |
| 3581 | |
| 3582 | /** |
| 3583 | * Node class for a case insensitive sequence of literal characters |
| 3584 | * including supplementary characters. |
| 3585 | */ |
| 3586 | static class SliceIS extends SliceNode { |
| 3587 | SliceIS(int[] buf) { |
| 3588 | super(buf); |
| 3589 | } |
| 3590 | int toLower(int c) { |
| 3591 | return ASCII.toLower(c); |
| 3592 | } |
| 3593 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3594 | int[] buf = buffer; |
| 3595 | int x = i; |
| 3596 | for (int j = 0; j < buf.length; j++) { |
| 3597 | if (x >= matcher.to) { |
| 3598 | matcher.hitEnd = true; |
| 3599 | return false; |
| 3600 | } |
| 3601 | int c = Character.codePointAt(seq, x); |
| 3602 | if (buf[j] != c && buf[j] != toLower(c)) |
| 3603 | return false; |
| 3604 | x += Character.charCount(c); |
| 3605 | if (x > matcher.to) { |
| 3606 | matcher.hitEnd = true; |
| 3607 | return false; |
| 3608 | } |
| 3609 | } |
| 3610 | return next.match(matcher, x, seq); |
| 3611 | } |
| 3612 | } |
| 3613 | |
| 3614 | /** |
| 3615 | * Node class for a case insensitive sequence of literal characters. |
| 3616 | * Uses unicode case folding. |
| 3617 | */ |
| 3618 | static final class SliceUS extends SliceIS { |
| 3619 | SliceUS(int[] buf) { |
| 3620 | super(buf); |
| 3621 | } |
| 3622 | int toLower(int c) { |
| 3623 | return Character.toLowerCase(Character.toUpperCase(c)); |
| 3624 | } |
| 3625 | } |
| 3626 | |
| 3627 | private static boolean inRange(int lower, int ch, int upper) { |
| 3628 | return lower <= ch && ch <= upper; |
| 3629 | } |
| 3630 | |
| 3631 | /** |
| 3632 | * Returns node for matching characters within an explicit value range. |
| 3633 | */ |
| 3634 | private static CharProperty rangeFor(final int lower, |
| 3635 | final int upper) { |
| 3636 | return new CharProperty() { |
| 3637 | boolean isSatisfiedBy(int ch) { |
| 3638 | return inRange(lower, ch, upper);}}; |
| 3639 | } |
| 3640 | |
| 3641 | /** |
| 3642 | * Returns node for matching characters within an explicit value |
| 3643 | * range in a case insensitive manner. |
| 3644 | */ |
| 3645 | private CharProperty caseInsensitiveRangeFor(final int lower, |
| 3646 | final int upper) { |
| 3647 | if (has(UNICODE_CASE)) |
| 3648 | return new CharProperty() { |
| 3649 | boolean isSatisfiedBy(int ch) { |
| 3650 | if (inRange(lower, ch, upper)) |
| 3651 | return true; |
| 3652 | int up = Character.toUpperCase(ch); |
| 3653 | return inRange(lower, up, upper) || |
| 3654 | inRange(lower, Character.toLowerCase(up), upper);}}; |
| 3655 | return new CharProperty() { |
| 3656 | boolean isSatisfiedBy(int ch) { |
| 3657 | return inRange(lower, ch, upper) || |
| 3658 | ASCII.isAscii(ch) && |
| 3659 | (inRange(lower, ASCII.toUpper(ch), upper) || |
| 3660 | inRange(lower, ASCII.toLower(ch), upper)); |
| 3661 | }}; |
| 3662 | } |
| 3663 | |
| 3664 | /** |
| 3665 | * Implements the Unicode category ALL and the dot metacharacter when |
| 3666 | * in dotall mode. |
| 3667 | */ |
| 3668 | static final class All extends CharProperty { |
| 3669 | boolean isSatisfiedBy(int ch) { |
| 3670 | return true; |
| 3671 | } |
| 3672 | } |
| 3673 | |
| 3674 | /** |
| 3675 | * Node class for the dot metacharacter when dotall is not enabled. |
| 3676 | */ |
| 3677 | static final class Dot extends CharProperty { |
| 3678 | boolean isSatisfiedBy(int ch) { |
| 3679 | return (ch != '\n' && ch != '\r' |
| 3680 | && (ch|1) != '\u2029' |
| 3681 | && ch != '\u0085'); |
| 3682 | } |
| 3683 | } |
| 3684 | |
| 3685 | /** |
| 3686 | * Node class for the dot metacharacter when dotall is not enabled |
| 3687 | * but UNIX_LINES is enabled. |
| 3688 | */ |
| 3689 | static final class UnixDot extends CharProperty { |
| 3690 | boolean isSatisfiedBy(int ch) { |
| 3691 | return ch != '\n'; |
| 3692 | } |
| 3693 | } |
| 3694 | |
| 3695 | /** |
| 3696 | * The 0 or 1 quantifier. This one class implements all three types. |
| 3697 | */ |
| 3698 | static final class Ques extends Node { |
| 3699 | Node atom; |
| 3700 | int type; |
| 3701 | Ques(Node node, int type) { |
| 3702 | this.atom = node; |
| 3703 | this.type = type; |
| 3704 | } |
| 3705 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3706 | switch (type) { |
| 3707 | case GREEDY: |
| 3708 | return (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)) |
| 3709 | || next.match(matcher, i, seq); |
| 3710 | case LAZY: |
| 3711 | return next.match(matcher, i, seq) |
| 3712 | || (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)); |
| 3713 | case POSSESSIVE: |
| 3714 | if (atom.match(matcher, i, seq)) i = matcher.last; |
| 3715 | return next.match(matcher, i, seq); |
| 3716 | default: |
| 3717 | return atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq); |
| 3718 | } |
| 3719 | } |
| 3720 | boolean study(TreeInfo info) { |
| 3721 | if (type != INDEPENDENT) { |
| 3722 | int minL = info.minLength; |
| 3723 | atom.study(info); |
| 3724 | info.minLength = minL; |
| 3725 | info.deterministic = false; |
| 3726 | return next.study(info); |
| 3727 | } else { |
| 3728 | atom.study(info); |
| 3729 | return next.study(info); |
| 3730 | } |
| 3731 | } |
| 3732 | } |
| 3733 | |
| 3734 | /** |
| 3735 | * Handles the curly-brace style repetition with a specified minimum and |
| 3736 | * maximum occurrences. The * quantifier is handled as a special case. |
| 3737 | * This class handles the three types. |
| 3738 | */ |
| 3739 | static final class Curly extends Node { |
| 3740 | Node atom; |
| 3741 | int type; |
| 3742 | int cmin; |
| 3743 | int cmax; |
| 3744 | |
| 3745 | Curly(Node node, int cmin, int cmax, int type) { |
| 3746 | this.atom = node; |
| 3747 | this.type = type; |
| 3748 | this.cmin = cmin; |
| 3749 | this.cmax = cmax; |
| 3750 | } |
| 3751 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3752 | int j; |
| 3753 | for (j = 0; j < cmin; j++) { |
| 3754 | if (atom.match(matcher, i, seq)) { |
| 3755 | i = matcher.last; |
| 3756 | continue; |
| 3757 | } |
| 3758 | return false; |
| 3759 | } |
| 3760 | if (type == GREEDY) |
| 3761 | return match0(matcher, i, j, seq); |
| 3762 | else if (type == LAZY) |
| 3763 | return match1(matcher, i, j, seq); |
| 3764 | else |
| 3765 | return match2(matcher, i, j, seq); |
| 3766 | } |
| 3767 | // Greedy match. |
| 3768 | // i is the index to start matching at |
| 3769 | // j is the number of atoms that have matched |
| 3770 | boolean match0(Matcher matcher, int i, int j, CharSequence seq) { |
| 3771 | if (j >= cmax) { |
| 3772 | // We have matched the maximum... continue with the rest of |
| 3773 | // the regular expression |
| 3774 | return next.match(matcher, i, seq); |
| 3775 | } |
| 3776 | int backLimit = j; |
| 3777 | while (atom.match(matcher, i, seq)) { |
| 3778 | // k is the length of this match |
| 3779 | int k = matcher.last - i; |
| 3780 | if (k == 0) // Zero length match |
| 3781 | break; |
| 3782 | // Move up index and number matched |
| 3783 | i = matcher.last; |
| 3784 | j++; |
| 3785 | // We are greedy so match as many as we can |
| 3786 | while (j < cmax) { |
| 3787 | if (!atom.match(matcher, i, seq)) |
| 3788 | break; |
| 3789 | if (i + k != matcher.last) { |
| 3790 | if (match0(matcher, matcher.last, j+1, seq)) |
| 3791 | return true; |
| 3792 | break; |
| 3793 | } |
| 3794 | i += k; |
| 3795 | j++; |
| 3796 | } |
| 3797 | // Handle backing off if match fails |
| 3798 | while (j >= backLimit) { |
| 3799 | if (next.match(matcher, i, seq)) |
| 3800 | return true; |
| 3801 | i -= k; |
| 3802 | j--; |
| 3803 | } |
| 3804 | return false; |
| 3805 | } |
| 3806 | return next.match(matcher, i, seq); |
| 3807 | } |
| 3808 | // Reluctant match. At this point, the minimum has been satisfied. |
| 3809 | // i is the index to start matching at |
| 3810 | // j is the number of atoms that have matched |
| 3811 | boolean match1(Matcher matcher, int i, int j, CharSequence seq) { |
| 3812 | for (;;) { |
| 3813 | // Try finishing match without consuming any more |
| 3814 | if (next.match(matcher, i, seq)) |
| 3815 | return true; |
| 3816 | // At the maximum, no match found |
| 3817 | if (j >= cmax) |
| 3818 | return false; |
| 3819 | // Okay, must try one more atom |
| 3820 | if (!atom.match(matcher, i, seq)) |
| 3821 | return false; |
| 3822 | // If we haven't moved forward then must break out |
| 3823 | if (i == matcher.last) |
| 3824 | return false; |
| 3825 | // Move up index and number matched |
| 3826 | i = matcher.last; |
| 3827 | j++; |
| 3828 | } |
| 3829 | } |
| 3830 | boolean match2(Matcher matcher, int i, int j, CharSequence seq) { |
| 3831 | for (; j < cmax; j++) { |
| 3832 | if (!atom.match(matcher, i, seq)) |
| 3833 | break; |
| 3834 | if (i == matcher.last) |
| 3835 | break; |
| 3836 | i = matcher.last; |
| 3837 | } |
| 3838 | return next.match(matcher, i, seq); |
| 3839 | } |
| 3840 | boolean study(TreeInfo info) { |
| 3841 | // Save original info |
| 3842 | int minL = info.minLength; |
| 3843 | int maxL = info.maxLength; |
| 3844 | boolean maxV = info.maxValid; |
| 3845 | boolean detm = info.deterministic; |
| 3846 | info.reset(); |
| 3847 | |
| 3848 | atom.study(info); |
| 3849 | |
| 3850 | int temp = info.minLength * cmin + minL; |
| 3851 | if (temp < minL) { |
| 3852 | temp = 0xFFFFFFF; // arbitrary large number |
| 3853 | } |
| 3854 | info.minLength = temp; |
| 3855 | |
| 3856 | if (maxV & info.maxValid) { |
| 3857 | temp = info.maxLength * cmax + maxL; |
| 3858 | info.maxLength = temp; |
| 3859 | if (temp < maxL) { |
| 3860 | info.maxValid = false; |
| 3861 | } |
| 3862 | } else { |
| 3863 | info.maxValid = false; |
| 3864 | } |
| 3865 | |
| 3866 | if (info.deterministic && cmin == cmax) |
| 3867 | info.deterministic = detm; |
| 3868 | else |
| 3869 | info.deterministic = false; |
| 3870 | |
| 3871 | return next.study(info); |
| 3872 | } |
| 3873 | } |
| 3874 | |
| 3875 | /** |
| 3876 | * Handles the curly-brace style repetition with a specified minimum and |
| 3877 | * maximum occurrences in deterministic cases. This is an iterative |
| 3878 | * optimization over the Prolog and Loop system which would handle this |
| 3879 | * in a recursive way. The * quantifier is handled as a special case. |
| 3880 | * If capture is true then this class saves group settings and ensures |
| 3881 | * that groups are unset when backing off of a group match. |
| 3882 | */ |
| 3883 | static final class GroupCurly extends Node { |
| 3884 | Node atom; |
| 3885 | int type; |
| 3886 | int cmin; |
| 3887 | int cmax; |
| 3888 | int localIndex; |
| 3889 | int groupIndex; |
| 3890 | boolean capture; |
| 3891 | |
| 3892 | GroupCurly(Node node, int cmin, int cmax, int type, int local, |
| 3893 | int group, boolean capture) { |
| 3894 | this.atom = node; |
| 3895 | this.type = type; |
| 3896 | this.cmin = cmin; |
| 3897 | this.cmax = cmax; |
| 3898 | this.localIndex = local; |
| 3899 | this.groupIndex = group; |
| 3900 | this.capture = capture; |
| 3901 | } |
| 3902 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 3903 | int[] groups = matcher.groups; |
| 3904 | int[] locals = matcher.locals; |
| 3905 | int save0 = locals[localIndex]; |
| 3906 | int save1 = 0; |
| 3907 | int save2 = 0; |
| 3908 | |
| 3909 | if (capture) { |
| 3910 | save1 = groups[groupIndex]; |
| 3911 | save2 = groups[groupIndex+1]; |
| 3912 | } |
| 3913 | |
| 3914 | // Notify GroupTail there is no need to setup group info |
| 3915 | // because it will be set here |
| 3916 | locals[localIndex] = -1; |
| 3917 | |
| 3918 | boolean ret = true; |
| 3919 | for (int j = 0; j < cmin; j++) { |
| 3920 | if (atom.match(matcher, i, seq)) { |
| 3921 | if (capture) { |
| 3922 | groups[groupIndex] = i; |
| 3923 | groups[groupIndex+1] = matcher.last; |
| 3924 | } |
| 3925 | i = matcher.last; |
| 3926 | } else { |
| 3927 | ret = false; |
| 3928 | break; |
| 3929 | } |
| 3930 | } |
| 3931 | if (ret) { |
| 3932 | if (type == GREEDY) { |
| 3933 | ret = match0(matcher, i, cmin, seq); |
| 3934 | } else if (type == LAZY) { |
| 3935 | ret = match1(matcher, i, cmin, seq); |
| 3936 | } else { |
| 3937 | ret = match2(matcher, i, cmin, seq); |
| 3938 | } |
| 3939 | } |
| 3940 | if (!ret) { |
| 3941 | locals[localIndex] = save0; |
| 3942 | if (capture) { |
| 3943 | groups[groupIndex] = save1; |
| 3944 | groups[groupIndex+1] = save2; |
| 3945 | } |
| 3946 | } |
| 3947 | return ret; |
| 3948 | } |
| 3949 | // Aggressive group match |
| 3950 | boolean match0(Matcher matcher, int i, int j, CharSequence seq) { |
| 3951 | int[] groups = matcher.groups; |
| 3952 | int save0 = 0; |
| 3953 | int save1 = 0; |
| 3954 | if (capture) { |
| 3955 | save0 = groups[groupIndex]; |
| 3956 | save1 = groups[groupIndex+1]; |
| 3957 | } |
| 3958 | for (;;) { |
| 3959 | if (j >= cmax) |
| 3960 | break; |
| 3961 | if (!atom.match(matcher, i, seq)) |
| 3962 | break; |
| 3963 | int k = matcher.last - i; |
| 3964 | if (k <= 0) { |
| 3965 | if (capture) { |
| 3966 | groups[groupIndex] = i; |
| 3967 | groups[groupIndex+1] = i + k; |
| 3968 | } |
| 3969 | i = i + k; |
| 3970 | break; |
| 3971 | } |
| 3972 | for (;;) { |
| 3973 | if (capture) { |
| 3974 | groups[groupIndex] = i; |
| 3975 | groups[groupIndex+1] = i + k; |
| 3976 | } |
| 3977 | i = i + k; |
| 3978 | if (++j >= cmax) |
| 3979 | break; |
| 3980 | if (!atom.match(matcher, i, seq)) |
| 3981 | break; |
| 3982 | if (i + k != matcher.last) { |
| 3983 | if (match0(matcher, i, j, seq)) |
| 3984 | return true; |
| 3985 | break; |
| 3986 | } |
| 3987 | } |
| 3988 | while (j > cmin) { |
| 3989 | if (next.match(matcher, i, seq)) { |
| 3990 | if (capture) { |
| 3991 | groups[groupIndex+1] = i; |
| 3992 | groups[groupIndex] = i - k; |
| 3993 | } |
| 3994 | i = i - k; |
| 3995 | return true; |
| 3996 | } |
| 3997 | // backing off |
| 3998 | if (capture) { |
| 3999 | groups[groupIndex+1] = i; |
| 4000 | groups[groupIndex] = i - k; |
| 4001 | } |
| 4002 | i = i - k; |
| 4003 | j--; |
| 4004 | } |
| 4005 | break; |
| 4006 | } |
| 4007 | if (capture) { |
| 4008 | groups[groupIndex] = save0; |
| 4009 | groups[groupIndex+1] = save1; |
| 4010 | } |
| 4011 | return next.match(matcher, i, seq); |
| 4012 | } |
| 4013 | // Reluctant matching |
| 4014 | boolean match1(Matcher matcher, int i, int j, CharSequence seq) { |
| 4015 | for (;;) { |
| 4016 | if (next.match(matcher, i, seq)) |
| 4017 | return true; |
| 4018 | if (j >= cmax) |
| 4019 | return false; |
| 4020 | if (!atom.match(matcher, i, seq)) |
| 4021 | return false; |
| 4022 | if (i == matcher.last) |
| 4023 | return false; |
| 4024 | if (capture) { |
| 4025 | matcher.groups[groupIndex] = i; |
| 4026 | matcher.groups[groupIndex+1] = matcher.last; |
| 4027 | } |
| 4028 | i = matcher.last; |
| 4029 | j++; |
| 4030 | } |
| 4031 | } |
| 4032 | // Possessive matching |
| 4033 | boolean match2(Matcher matcher, int i, int j, CharSequence seq) { |
| 4034 | for (; j < cmax; j++) { |
| 4035 | if (!atom.match(matcher, i, seq)) { |
| 4036 | break; |
| 4037 | } |
| 4038 | if (capture) { |
| 4039 | matcher.groups[groupIndex] = i; |
| 4040 | matcher.groups[groupIndex+1] = matcher.last; |
| 4041 | } |
| 4042 | if (i == matcher.last) { |
| 4043 | break; |
| 4044 | } |
| 4045 | i = matcher.last; |
| 4046 | } |
| 4047 | return next.match(matcher, i, seq); |
| 4048 | } |
| 4049 | boolean study(TreeInfo info) { |
| 4050 | // Save original info |
| 4051 | int minL = info.minLength; |
| 4052 | int maxL = info.maxLength; |
| 4053 | boolean maxV = info.maxValid; |
| 4054 | boolean detm = info.deterministic; |
| 4055 | info.reset(); |
| 4056 | |
| 4057 | atom.study(info); |
| 4058 | |
| 4059 | int temp = info.minLength * cmin + minL; |
| 4060 | if (temp < minL) { |
| 4061 | temp = 0xFFFFFFF; // Arbitrary large number |
| 4062 | } |
| 4063 | info.minLength = temp; |
| 4064 | |
| 4065 | if (maxV & info.maxValid) { |
| 4066 | temp = info.maxLength * cmax + maxL; |
| 4067 | info.maxLength = temp; |
| 4068 | if (temp < maxL) { |
| 4069 | info.maxValid = false; |
| 4070 | } |
| 4071 | } else { |
| 4072 | info.maxValid = false; |
| 4073 | } |
| 4074 | |
| 4075 | if (info.deterministic && cmin == cmax) { |
| 4076 | info.deterministic = detm; |
| 4077 | } else { |
| 4078 | info.deterministic = false; |
| 4079 | } |
| 4080 | |
| 4081 | return next.study(info); |
| 4082 | } |
| 4083 | } |
| 4084 | |
| 4085 | /** |
| 4086 | * A Guard node at the end of each atom node in a Branch. It |
| 4087 | * serves the purpose of chaining the "match" operation to |
| 4088 | * "next" but not the "study", so we can collect the TreeInfo |
| 4089 | * of each atom node without including the TreeInfo of the |
| 4090 | * "next". |
| 4091 | */ |
| 4092 | static final class BranchConn extends Node { |
| 4093 | BranchConn() {}; |
| 4094 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4095 | return next.match(matcher, i, seq); |
| 4096 | } |
| 4097 | boolean study(TreeInfo info) { |
| 4098 | return info.deterministic; |
| 4099 | } |
| 4100 | } |
| 4101 | |
| 4102 | /** |
| 4103 | * Handles the branching of alternations. Note this is also used for |
| 4104 | * the ? quantifier to branch between the case where it matches once |
| 4105 | * and where it does not occur. |
| 4106 | */ |
| 4107 | static final class Branch extends Node { |
| 4108 | Node[] atoms = new Node[2]; |
| 4109 | int size = 2; |
| 4110 | Node conn; |
| 4111 | Branch(Node first, Node second, Node branchConn) { |
| 4112 | conn = branchConn; |
| 4113 | atoms[0] = first; |
| 4114 | atoms[1] = second; |
| 4115 | } |
| 4116 | |
| 4117 | void add(Node node) { |
| 4118 | if (size >= atoms.length) { |
| 4119 | Node[] tmp = new Node[atoms.length*2]; |
| 4120 | System.arraycopy(atoms, 0, tmp, 0, atoms.length); |
| 4121 | atoms = tmp; |
| 4122 | } |
| 4123 | atoms[size++] = node; |
| 4124 | } |
| 4125 | |
| 4126 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4127 | for (int n = 0; n < size; n++) { |
| 4128 | if (atoms[n] == null) { |
| 4129 | if (conn.next.match(matcher, i, seq)) |
| 4130 | return true; |
| 4131 | } else if (atoms[n].match(matcher, i, seq)) { |
| 4132 | return true; |
| 4133 | } |
| 4134 | } |
| 4135 | return false; |
| 4136 | } |
| 4137 | |
| 4138 | boolean study(TreeInfo info) { |
| 4139 | int minL = info.minLength; |
| 4140 | int maxL = info.maxLength; |
| 4141 | boolean maxV = info.maxValid; |
| 4142 | |
| 4143 | int minL2 = Integer.MAX_VALUE; //arbitrary large enough num |
| 4144 | int maxL2 = -1; |
| 4145 | for (int n = 0; n < size; n++) { |
| 4146 | info.reset(); |
| 4147 | if (atoms[n] != null) |
| 4148 | atoms[n].study(info); |
| 4149 | minL2 = Math.min(minL2, info.minLength); |
| 4150 | maxL2 = Math.max(maxL2, info.maxLength); |
| 4151 | maxV = (maxV & info.maxValid); |
| 4152 | } |
| 4153 | |
| 4154 | minL += minL2; |
| 4155 | maxL += maxL2; |
| 4156 | |
| 4157 | info.reset(); |
| 4158 | conn.next.study(info); |
| 4159 | |
| 4160 | info.minLength += minL; |
| 4161 | info.maxLength += maxL; |
| 4162 | info.maxValid &= maxV; |
| 4163 | info.deterministic = false; |
| 4164 | return false; |
| 4165 | } |
| 4166 | } |
| 4167 | |
| 4168 | /** |
| 4169 | * The GroupHead saves the location where the group begins in the locals |
| 4170 | * and restores them when the match is done. |
| 4171 | * |
| 4172 | * The matchRef is used when a reference to this group is accessed later |
| 4173 | * in the expression. The locals will have a negative value in them to |
| 4174 | * indicate that we do not want to unset the group if the reference |
| 4175 | * doesn't match. |
| 4176 | */ |
| 4177 | static final class GroupHead extends Node { |
| 4178 | int localIndex; |
| 4179 | GroupHead(int localCount) { |
| 4180 | localIndex = localCount; |
| 4181 | } |
| 4182 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4183 | int save = matcher.locals[localIndex]; |
| 4184 | matcher.locals[localIndex] = i; |
| 4185 | boolean ret = next.match(matcher, i, seq); |
| 4186 | matcher.locals[localIndex] = save; |
| 4187 | return ret; |
| 4188 | } |
| 4189 | boolean matchRef(Matcher matcher, int i, CharSequence seq) { |
| 4190 | int save = matcher.locals[localIndex]; |
| 4191 | matcher.locals[localIndex] = ~i; // HACK |
| 4192 | boolean ret = next.match(matcher, i, seq); |
| 4193 | matcher.locals[localIndex] = save; |
| 4194 | return ret; |
| 4195 | } |
| 4196 | } |
| 4197 | |
| 4198 | /** |
| 4199 | * Recursive reference to a group in the regular expression. It calls |
| 4200 | * matchRef because if the reference fails to match we would not unset |
| 4201 | * the group. |
| 4202 | */ |
| 4203 | static final class GroupRef extends Node { |
| 4204 | GroupHead head; |
| 4205 | GroupRef(GroupHead head) { |
| 4206 | this.head = head; |
| 4207 | } |
| 4208 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4209 | return head.matchRef(matcher, i, seq) |
| 4210 | && next.match(matcher, matcher.last, seq); |
| 4211 | } |
| 4212 | boolean study(TreeInfo info) { |
| 4213 | info.maxValid = false; |
| 4214 | info.deterministic = false; |
| 4215 | return next.study(info); |
| 4216 | } |
| 4217 | } |
| 4218 | |
| 4219 | /** |
| 4220 | * The GroupTail handles the setting of group beginning and ending |
| 4221 | * locations when groups are successfully matched. It must also be able to |
| 4222 | * unset groups that have to be backed off of. |
| 4223 | * |
| 4224 | * The GroupTail node is also used when a previous group is referenced, |
| 4225 | * and in that case no group information needs to be set. |
| 4226 | */ |
| 4227 | static final class GroupTail extends Node { |
| 4228 | int localIndex; |
| 4229 | int groupIndex; |
| 4230 | GroupTail(int localCount, int groupCount) { |
| 4231 | localIndex = localCount; |
| 4232 | groupIndex = groupCount + groupCount; |
| 4233 | } |
| 4234 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4235 | int tmp = matcher.locals[localIndex]; |
| 4236 | if (tmp >= 0) { // This is the normal group case. |
| 4237 | // Save the group so we can unset it if it |
| 4238 | // backs off of a match. |
| 4239 | int groupStart = matcher.groups[groupIndex]; |
| 4240 | int groupEnd = matcher.groups[groupIndex+1]; |
| 4241 | |
| 4242 | matcher.groups[groupIndex] = tmp; |
| 4243 | matcher.groups[groupIndex+1] = i; |
| 4244 | if (next.match(matcher, i, seq)) { |
| 4245 | return true; |
| 4246 | } |
| 4247 | matcher.groups[groupIndex] = groupStart; |
| 4248 | matcher.groups[groupIndex+1] = groupEnd; |
| 4249 | return false; |
| 4250 | } else { |
| 4251 | // This is a group reference case. We don't need to save any |
| 4252 | // group info because it isn't really a group. |
| 4253 | matcher.last = i; |
| 4254 | return true; |
| 4255 | } |
| 4256 | } |
| 4257 | } |
| 4258 | |
| 4259 | /** |
| 4260 | * This sets up a loop to handle a recursive quantifier structure. |
| 4261 | */ |
| 4262 | static final class Prolog extends Node { |
| 4263 | Loop loop; |
| 4264 | Prolog(Loop loop) { |
| 4265 | this.loop = loop; |
| 4266 | } |
| 4267 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4268 | return loop.matchInit(matcher, i, seq); |
| 4269 | } |
| 4270 | boolean study(TreeInfo info) { |
| 4271 | return loop.study(info); |
| 4272 | } |
| 4273 | } |
| 4274 | |
| 4275 | /** |
| 4276 | * Handles the repetition count for a greedy Curly. The matchInit |
| 4277 | * is called from the Prolog to save the index of where the group |
| 4278 | * beginning is stored. A zero length group check occurs in the |
| 4279 | * normal match but is skipped in the matchInit. |
| 4280 | */ |
| 4281 | static class Loop extends Node { |
| 4282 | Node body; |
| 4283 | int countIndex; // local count index in matcher locals |
| 4284 | int beginIndex; // group beginning index |
| 4285 | int cmin, cmax; |
| 4286 | Loop(int countIndex, int beginIndex) { |
| 4287 | this.countIndex = countIndex; |
| 4288 | this.beginIndex = beginIndex; |
| 4289 | } |
| 4290 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4291 | // Avoid infinite loop in zero-length case. |
| 4292 | if (i > matcher.locals[beginIndex]) { |
| 4293 | int count = matcher.locals[countIndex]; |
| 4294 | |
| 4295 | // This block is for before we reach the minimum |
| 4296 | // iterations required for the loop to match |
| 4297 | if (count < cmin) { |
| 4298 | matcher.locals[countIndex] = count + 1; |
| 4299 | boolean b = body.match(matcher, i, seq); |
| 4300 | // If match failed we must backtrack, so |
| 4301 | // the loop count should NOT be incremented |
| 4302 | if (!b) |
| 4303 | matcher.locals[countIndex] = count; |
| 4304 | // Return success or failure since we are under |
| 4305 | // minimum |
| 4306 | return b; |
| 4307 | } |
| 4308 | // This block is for after we have the minimum |
| 4309 | // iterations required for the loop to match |
| 4310 | if (count < cmax) { |
| 4311 | matcher.locals[countIndex] = count + 1; |
| 4312 | boolean b = body.match(matcher, i, seq); |
| 4313 | // If match failed we must backtrack, so |
| 4314 | // the loop count should NOT be incremented |
| 4315 | if (!b) |
| 4316 | matcher.locals[countIndex] = count; |
| 4317 | else |
| 4318 | return true; |
| 4319 | } |
| 4320 | } |
| 4321 | return next.match(matcher, i, seq); |
| 4322 | } |
| 4323 | boolean matchInit(Matcher matcher, int i, CharSequence seq) { |
| 4324 | int save = matcher.locals[countIndex]; |
| 4325 | boolean ret = false; |
| 4326 | if (0 < cmin) { |
| 4327 | matcher.locals[countIndex] = 1; |
| 4328 | ret = body.match(matcher, i, seq); |
| 4329 | } else if (0 < cmax) { |
| 4330 | matcher.locals[countIndex] = 1; |
| 4331 | ret = body.match(matcher, i, seq); |
| 4332 | if (ret == false) |
| 4333 | ret = next.match(matcher, i, seq); |
| 4334 | } else { |
| 4335 | ret = next.match(matcher, i, seq); |
| 4336 | } |
| 4337 | matcher.locals[countIndex] = save; |
| 4338 | return ret; |
| 4339 | } |
| 4340 | boolean study(TreeInfo info) { |
| 4341 | info.maxValid = false; |
| 4342 | info.deterministic = false; |
| 4343 | return false; |
| 4344 | } |
| 4345 | } |
| 4346 | |
| 4347 | /** |
| 4348 | * Handles the repetition count for a reluctant Curly. The matchInit |
| 4349 | * is called from the Prolog to save the index of where the group |
| 4350 | * beginning is stored. A zero length group check occurs in the |
| 4351 | * normal match but is skipped in the matchInit. |
| 4352 | */ |
| 4353 | static final class LazyLoop extends Loop { |
| 4354 | LazyLoop(int countIndex, int beginIndex) { |
| 4355 | super(countIndex, beginIndex); |
| 4356 | } |
| 4357 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4358 | // Check for zero length group |
| 4359 | if (i > matcher.locals[beginIndex]) { |
| 4360 | int count = matcher.locals[countIndex]; |
| 4361 | if (count < cmin) { |
| 4362 | matcher.locals[countIndex] = count + 1; |
| 4363 | boolean result = body.match(matcher, i, seq); |
| 4364 | // If match failed we must backtrack, so |
| 4365 | // the loop count should NOT be incremented |
| 4366 | if (!result) |
| 4367 | matcher.locals[countIndex] = count; |
| 4368 | return result; |
| 4369 | } |
| 4370 | if (next.match(matcher, i, seq)) |
| 4371 | return true; |
| 4372 | if (count < cmax) { |
| 4373 | matcher.locals[countIndex] = count + 1; |
| 4374 | boolean result = body.match(matcher, i, seq); |
| 4375 | // If match failed we must backtrack, so |
| 4376 | // the loop count should NOT be incremented |
| 4377 | if (!result) |
| 4378 | matcher.locals[countIndex] = count; |
| 4379 | return result; |
| 4380 | } |
| 4381 | return false; |
| 4382 | } |
| 4383 | return next.match(matcher, i, seq); |
| 4384 | } |
| 4385 | boolean matchInit(Matcher matcher, int i, CharSequence seq) { |
| 4386 | int save = matcher.locals[countIndex]; |
| 4387 | boolean ret = false; |
| 4388 | if (0 < cmin) { |
| 4389 | matcher.locals[countIndex] = 1; |
| 4390 | ret = body.match(matcher, i, seq); |
| 4391 | } else if (next.match(matcher, i, seq)) { |
| 4392 | ret = true; |
| 4393 | } else if (0 < cmax) { |
| 4394 | matcher.locals[countIndex] = 1; |
| 4395 | ret = body.match(matcher, i, seq); |
| 4396 | } |
| 4397 | matcher.locals[countIndex] = save; |
| 4398 | return ret; |
| 4399 | } |
| 4400 | boolean study(TreeInfo info) { |
| 4401 | info.maxValid = false; |
| 4402 | info.deterministic = false; |
| 4403 | return false; |
| 4404 | } |
| 4405 | } |
| 4406 | |
| 4407 | /** |
| 4408 | * Refers to a group in the regular expression. Attempts to match |
| 4409 | * whatever the group referred to last matched. |
| 4410 | */ |
| 4411 | static class BackRef extends Node { |
| 4412 | int groupIndex; |
| 4413 | BackRef(int groupCount) { |
| 4414 | super(); |
| 4415 | groupIndex = groupCount + groupCount; |
| 4416 | } |
| 4417 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4418 | int j = matcher.groups[groupIndex]; |
| 4419 | int k = matcher.groups[groupIndex+1]; |
| 4420 | |
| 4421 | int groupSize = k - j; |
| 4422 | |
| 4423 | // If the referenced group didn't match, neither can this |
| 4424 | if (j < 0) |
| 4425 | return false; |
| 4426 | |
| 4427 | // If there isn't enough input left no match |
| 4428 | if (i + groupSize > matcher.to) { |
| 4429 | matcher.hitEnd = true; |
| 4430 | return false; |
| 4431 | } |
| 4432 | |
| 4433 | // Check each new char to make sure it matches what the group |
| 4434 | // referenced matched last time around |
| 4435 | for (int index=0; index<groupSize; index++) |
| 4436 | if (seq.charAt(i+index) != seq.charAt(j+index)) |
| 4437 | return false; |
| 4438 | |
| 4439 | return next.match(matcher, i+groupSize, seq); |
| 4440 | } |
| 4441 | boolean study(TreeInfo info) { |
| 4442 | info.maxValid = false; |
| 4443 | return next.study(info); |
| 4444 | } |
| 4445 | } |
| 4446 | |
| 4447 | static class CIBackRef extends Node { |
| 4448 | int groupIndex; |
| 4449 | boolean doUnicodeCase; |
| 4450 | CIBackRef(int groupCount, boolean doUnicodeCase) { |
| 4451 | super(); |
| 4452 | groupIndex = groupCount + groupCount; |
| 4453 | this.doUnicodeCase = doUnicodeCase; |
| 4454 | } |
| 4455 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4456 | int j = matcher.groups[groupIndex]; |
| 4457 | int k = matcher.groups[groupIndex+1]; |
| 4458 | |
| 4459 | int groupSize = k - j; |
| 4460 | |
| 4461 | // If the referenced group didn't match, neither can this |
| 4462 | if (j < 0) |
| 4463 | return false; |
| 4464 | |
| 4465 | // If there isn't enough input left no match |
| 4466 | if (i + groupSize > matcher.to) { |
| 4467 | matcher.hitEnd = true; |
| 4468 | return false; |
| 4469 | } |
| 4470 | |
| 4471 | // Check each new char to make sure it matches what the group |
| 4472 | // referenced matched last time around |
| 4473 | int x = i; |
| 4474 | for (int index=0; index<groupSize; index++) { |
| 4475 | int c1 = Character.codePointAt(seq, x); |
| 4476 | int c2 = Character.codePointAt(seq, j); |
| 4477 | if (c1 != c2) { |
| 4478 | if (doUnicodeCase) { |
| 4479 | int cc1 = Character.toUpperCase(c1); |
| 4480 | int cc2 = Character.toUpperCase(c2); |
| 4481 | if (cc1 != cc2 && |
| 4482 | Character.toLowerCase(cc1) != |
| 4483 | Character.toLowerCase(cc2)) |
| 4484 | return false; |
| 4485 | } else { |
| 4486 | if (ASCII.toLower(c1) != ASCII.toLower(c2)) |
| 4487 | return false; |
| 4488 | } |
| 4489 | } |
| 4490 | x += Character.charCount(c1); |
| 4491 | j += Character.charCount(c2); |
| 4492 | } |
| 4493 | |
| 4494 | return next.match(matcher, i+groupSize, seq); |
| 4495 | } |
| 4496 | boolean study(TreeInfo info) { |
| 4497 | info.maxValid = false; |
| 4498 | return next.study(info); |
| 4499 | } |
| 4500 | } |
| 4501 | |
| 4502 | /** |
| 4503 | * Searches until the next instance of its atom. This is useful for |
| 4504 | * finding the atom efficiently without passing an instance of it |
| 4505 | * (greedy problem) and without a lot of wasted search time (reluctant |
| 4506 | * problem). |
| 4507 | */ |
| 4508 | static final class First extends Node { |
| 4509 | Node atom; |
| 4510 | First(Node node) { |
| 4511 | this.atom = BnM.optimize(node); |
| 4512 | } |
| 4513 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4514 | if (atom instanceof BnM) { |
| 4515 | return atom.match(matcher, i, seq) |
| 4516 | && next.match(matcher, matcher.last, seq); |
| 4517 | } |
| 4518 | for (;;) { |
| 4519 | if (i > matcher.to) { |
| 4520 | matcher.hitEnd = true; |
| 4521 | return false; |
| 4522 | } |
| 4523 | if (atom.match(matcher, i, seq)) { |
| 4524 | return next.match(matcher, matcher.last, seq); |
| 4525 | } |
| 4526 | i += countChars(seq, i, 1); |
| 4527 | matcher.first++; |
| 4528 | } |
| 4529 | } |
| 4530 | boolean study(TreeInfo info) { |
| 4531 | atom.study(info); |
| 4532 | info.maxValid = false; |
| 4533 | info.deterministic = false; |
| 4534 | return next.study(info); |
| 4535 | } |
| 4536 | } |
| 4537 | |
| 4538 | static final class Conditional extends Node { |
| 4539 | Node cond, yes, not; |
| 4540 | Conditional(Node cond, Node yes, Node not) { |
| 4541 | this.cond = cond; |
| 4542 | this.yes = yes; |
| 4543 | this.not = not; |
| 4544 | } |
| 4545 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4546 | if (cond.match(matcher, i, seq)) { |
| 4547 | return yes.match(matcher, i, seq); |
| 4548 | } else { |
| 4549 | return not.match(matcher, i, seq); |
| 4550 | } |
| 4551 | } |
| 4552 | boolean study(TreeInfo info) { |
| 4553 | int minL = info.minLength; |
| 4554 | int maxL = info.maxLength; |
| 4555 | boolean maxV = info.maxValid; |
| 4556 | info.reset(); |
| 4557 | yes.study(info); |
| 4558 | |
| 4559 | int minL2 = info.minLength; |
| 4560 | int maxL2 = info.maxLength; |
| 4561 | boolean maxV2 = info.maxValid; |
| 4562 | info.reset(); |
| 4563 | not.study(info); |
| 4564 | |
| 4565 | info.minLength = minL + Math.min(minL2, info.minLength); |
| 4566 | info.maxLength = maxL + Math.max(maxL2, info.maxLength); |
| 4567 | info.maxValid = (maxV & maxV2 & info.maxValid); |
| 4568 | info.deterministic = false; |
| 4569 | return next.study(info); |
| 4570 | } |
| 4571 | } |
| 4572 | |
| 4573 | /** |
| 4574 | * Zero width positive lookahead. |
| 4575 | */ |
| 4576 | static final class Pos extends Node { |
| 4577 | Node cond; |
| 4578 | Pos(Node cond) { |
| 4579 | this.cond = cond; |
| 4580 | } |
| 4581 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4582 | int savedTo = matcher.to; |
| 4583 | boolean conditionMatched = false; |
| 4584 | |
| 4585 | // Relax transparent region boundaries for lookahead |
| 4586 | if (matcher.transparentBounds) |
| 4587 | matcher.to = matcher.getTextLength(); |
| 4588 | try { |
| 4589 | conditionMatched = cond.match(matcher, i, seq); |
| 4590 | } finally { |
| 4591 | // Reinstate region boundaries |
| 4592 | matcher.to = savedTo; |
| 4593 | } |
| 4594 | return conditionMatched && next.match(matcher, i, seq); |
| 4595 | } |
| 4596 | } |
| 4597 | |
| 4598 | /** |
| 4599 | * Zero width negative lookahead. |
| 4600 | */ |
| 4601 | static final class Neg extends Node { |
| 4602 | Node cond; |
| 4603 | Neg(Node cond) { |
| 4604 | this.cond = cond; |
| 4605 | } |
| 4606 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4607 | int savedTo = matcher.to; |
| 4608 | boolean conditionMatched = false; |
| 4609 | |
| 4610 | // Relax transparent region boundaries for lookahead |
| 4611 | if (matcher.transparentBounds) |
| 4612 | matcher.to = matcher.getTextLength(); |
| 4613 | try { |
| 4614 | if (i < matcher.to) { |
| 4615 | conditionMatched = !cond.match(matcher, i, seq); |
| 4616 | } else { |
| 4617 | // If a negative lookahead succeeds then more input |
| 4618 | // could cause it to fail! |
| 4619 | matcher.requireEnd = true; |
| 4620 | conditionMatched = !cond.match(matcher, i, seq); |
| 4621 | } |
| 4622 | } finally { |
| 4623 | // Reinstate region boundaries |
| 4624 | matcher.to = savedTo; |
| 4625 | } |
| 4626 | return conditionMatched && next.match(matcher, i, seq); |
| 4627 | } |
| 4628 | } |
| 4629 | |
| 4630 | /** |
| 4631 | * For use with lookbehinds; matches the position where the lookbehind |
| 4632 | * was encountered. |
| 4633 | */ |
| 4634 | static Node lookbehindEnd = new Node() { |
| 4635 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4636 | return i == matcher.lookbehindTo; |
| 4637 | } |
| 4638 | }; |
| 4639 | |
| 4640 | /** |
| 4641 | * Zero width positive lookbehind. |
| 4642 | */ |
| 4643 | static class Behind extends Node { |
| 4644 | Node cond; |
| 4645 | int rmax, rmin; |
| 4646 | Behind(Node cond, int rmax, int rmin) { |
| 4647 | this.cond = cond; |
| 4648 | this.rmax = rmax; |
| 4649 | this.rmin = rmin; |
| 4650 | } |
| 4651 | |
| 4652 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4653 | int savedFrom = matcher.from; |
| 4654 | boolean conditionMatched = false; |
| 4655 | int startIndex = (!matcher.transparentBounds) ? |
| 4656 | matcher.from : 0; |
| 4657 | int from = Math.max(i - rmax, startIndex); |
| 4658 | // Set end boundary |
| 4659 | int savedLBT = matcher.lookbehindTo; |
| 4660 | matcher.lookbehindTo = i; |
| 4661 | // Relax transparent region boundaries for lookbehind |
| 4662 | if (matcher.transparentBounds) |
| 4663 | matcher.from = 0; |
| 4664 | for (int j = i - rmin; !conditionMatched && j >= from; j--) { |
| 4665 | conditionMatched = cond.match(matcher, j, seq); |
| 4666 | } |
| 4667 | matcher.from = savedFrom; |
| 4668 | matcher.lookbehindTo = savedLBT; |
| 4669 | return conditionMatched && next.match(matcher, i, seq); |
| 4670 | } |
| 4671 | } |
| 4672 | |
| 4673 | /** |
| 4674 | * Zero width positive lookbehind, including supplementary |
| 4675 | * characters or unpaired surrogates. |
| 4676 | */ |
| 4677 | static final class BehindS extends Behind { |
| 4678 | BehindS(Node cond, int rmax, int rmin) { |
| 4679 | super(cond, rmax, rmin); |
| 4680 | } |
| 4681 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4682 | int rmaxChars = countChars(seq, i, -rmax); |
| 4683 | int rminChars = countChars(seq, i, -rmin); |
| 4684 | int savedFrom = matcher.from; |
| 4685 | int startIndex = (!matcher.transparentBounds) ? |
| 4686 | matcher.from : 0; |
| 4687 | boolean conditionMatched = false; |
| 4688 | int from = Math.max(i - rmaxChars, startIndex); |
| 4689 | // Set end boundary |
| 4690 | int savedLBT = matcher.lookbehindTo; |
| 4691 | matcher.lookbehindTo = i; |
| 4692 | // Relax transparent region boundaries for lookbehind |
| 4693 | if (matcher.transparentBounds) |
| 4694 | matcher.from = 0; |
| 4695 | |
| 4696 | for (int j = i - rminChars; |
| 4697 | !conditionMatched && j >= from; |
| 4698 | j -= j>from ? countChars(seq, j, -1) : 1) { |
| 4699 | conditionMatched = cond.match(matcher, j, seq); |
| 4700 | } |
| 4701 | matcher.from = savedFrom; |
| 4702 | matcher.lookbehindTo = savedLBT; |
| 4703 | return conditionMatched && next.match(matcher, i, seq); |
| 4704 | } |
| 4705 | } |
| 4706 | |
| 4707 | /** |
| 4708 | * Zero width negative lookbehind. |
| 4709 | */ |
| 4710 | static class NotBehind extends Node { |
| 4711 | Node cond; |
| 4712 | int rmax, rmin; |
| 4713 | NotBehind(Node cond, int rmax, int rmin) { |
| 4714 | this.cond = cond; |
| 4715 | this.rmax = rmax; |
| 4716 | this.rmin = rmin; |
| 4717 | } |
| 4718 | |
| 4719 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4720 | int savedLBT = matcher.lookbehindTo; |
| 4721 | int savedFrom = matcher.from; |
| 4722 | boolean conditionMatched = false; |
| 4723 | int startIndex = (!matcher.transparentBounds) ? |
| 4724 | matcher.from : 0; |
| 4725 | int from = Math.max(i - rmax, startIndex); |
| 4726 | matcher.lookbehindTo = i; |
| 4727 | // Relax transparent region boundaries for lookbehind |
| 4728 | if (matcher.transparentBounds) |
| 4729 | matcher.from = 0; |
| 4730 | for (int j = i - rmin; !conditionMatched && j >= from; j--) { |
| 4731 | conditionMatched = cond.match(matcher, j, seq); |
| 4732 | } |
| 4733 | // Reinstate region boundaries |
| 4734 | matcher.from = savedFrom; |
| 4735 | matcher.lookbehindTo = savedLBT; |
| 4736 | return !conditionMatched && next.match(matcher, i, seq); |
| 4737 | } |
| 4738 | } |
| 4739 | |
| 4740 | /** |
| 4741 | * Zero width negative lookbehind, including supplementary |
| 4742 | * characters or unpaired surrogates. |
| 4743 | */ |
| 4744 | static final class NotBehindS extends NotBehind { |
| 4745 | NotBehindS(Node cond, int rmax, int rmin) { |
| 4746 | super(cond, rmax, rmin); |
| 4747 | } |
| 4748 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4749 | int rmaxChars = countChars(seq, i, -rmax); |
| 4750 | int rminChars = countChars(seq, i, -rmin); |
| 4751 | int savedFrom = matcher.from; |
| 4752 | int savedLBT = matcher.lookbehindTo; |
| 4753 | boolean conditionMatched = false; |
| 4754 | int startIndex = (!matcher.transparentBounds) ? |
| 4755 | matcher.from : 0; |
| 4756 | int from = Math.max(i - rmaxChars, startIndex); |
| 4757 | matcher.lookbehindTo = i; |
| 4758 | // Relax transparent region boundaries for lookbehind |
| 4759 | if (matcher.transparentBounds) |
| 4760 | matcher.from = 0; |
| 4761 | for (int j = i - rminChars; |
| 4762 | !conditionMatched && j >= from; |
| 4763 | j -= j>from ? countChars(seq, j, -1) : 1) { |
| 4764 | conditionMatched = cond.match(matcher, j, seq); |
| 4765 | } |
| 4766 | //Reinstate region boundaries |
| 4767 | matcher.from = savedFrom; |
| 4768 | matcher.lookbehindTo = savedLBT; |
| 4769 | return !conditionMatched && next.match(matcher, i, seq); |
| 4770 | } |
| 4771 | } |
| 4772 | |
| 4773 | /** |
| 4774 | * Returns the set union of two CharProperty nodes. |
| 4775 | */ |
| 4776 | private static CharProperty union(final CharProperty lhs, |
| 4777 | final CharProperty rhs) { |
| 4778 | return new CharProperty() { |
| 4779 | boolean isSatisfiedBy(int ch) { |
| 4780 | return lhs.isSatisfiedBy(ch) || rhs.isSatisfiedBy(ch);}}; |
| 4781 | } |
| 4782 | |
| 4783 | /** |
| 4784 | * Returns the set intersection of two CharProperty nodes. |
| 4785 | */ |
| 4786 | private static CharProperty intersection(final CharProperty lhs, |
| 4787 | final CharProperty rhs) { |
| 4788 | return new CharProperty() { |
| 4789 | boolean isSatisfiedBy(int ch) { |
| 4790 | return lhs.isSatisfiedBy(ch) && rhs.isSatisfiedBy(ch);}}; |
| 4791 | } |
| 4792 | |
| 4793 | /** |
| 4794 | * Returns the set difference of two CharProperty nodes. |
| 4795 | */ |
| 4796 | private static CharProperty setDifference(final CharProperty lhs, |
| 4797 | final CharProperty rhs) { |
| 4798 | return new CharProperty() { |
| 4799 | boolean isSatisfiedBy(int ch) { |
| 4800 | return ! rhs.isSatisfiedBy(ch) && lhs.isSatisfiedBy(ch);}}; |
| 4801 | } |
| 4802 | |
| 4803 | /** |
| 4804 | * Handles word boundaries. Includes a field to allow this one class to |
| 4805 | * deal with the different types of word boundaries we can match. The word |
| 4806 | * characters include underscores, letters, and digits. Non spacing marks |
| 4807 | * can are also part of a word if they have a base character, otherwise |
| 4808 | * they are ignored for purposes of finding word boundaries. |
| 4809 | */ |
| 4810 | static final class Bound extends Node { |
| 4811 | static int LEFT = 0x1; |
| 4812 | static int RIGHT= 0x2; |
| 4813 | static int BOTH = 0x3; |
| 4814 | static int NONE = 0x4; |
| 4815 | int type; |
| 4816 | Bound(int n) { |
| 4817 | type = n; |
| 4818 | } |
| 4819 | int check(Matcher matcher, int i, CharSequence seq) { |
| 4820 | int ch; |
| 4821 | boolean left = false; |
| 4822 | int startIndex = matcher.from; |
| 4823 | int endIndex = matcher.to; |
| 4824 | if (matcher.transparentBounds) { |
| 4825 | startIndex = 0; |
| 4826 | endIndex = matcher.getTextLength(); |
| 4827 | } |
| 4828 | if (i > startIndex) { |
| 4829 | ch = Character.codePointBefore(seq, i); |
| 4830 | left = (ch == '_' || Character.isLetterOrDigit(ch) || |
| 4831 | ((Character.getType(ch) == Character.NON_SPACING_MARK) |
| 4832 | && hasBaseCharacter(matcher, i-1, seq))); |
| 4833 | } |
| 4834 | boolean right = false; |
| 4835 | if (i < endIndex) { |
| 4836 | ch = Character.codePointAt(seq, i); |
| 4837 | right = (ch == '_' || Character.isLetterOrDigit(ch) || |
| 4838 | ((Character.getType(ch) == Character.NON_SPACING_MARK) |
| 4839 | && hasBaseCharacter(matcher, i, seq))); |
| 4840 | } else { |
| 4841 | // Tried to access char past the end |
| 4842 | matcher.hitEnd = true; |
| 4843 | // The addition of another char could wreck a boundary |
| 4844 | matcher.requireEnd = true; |
| 4845 | } |
| 4846 | return ((left ^ right) ? (right ? LEFT : RIGHT) : NONE); |
| 4847 | } |
| 4848 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4849 | return (check(matcher, i, seq) & type) > 0 |
| 4850 | && next.match(matcher, i, seq); |
| 4851 | } |
| 4852 | } |
| 4853 | |
| 4854 | /** |
| 4855 | * Non spacing marks only count as word characters in bounds calculations |
| 4856 | * if they have a base character. |
| 4857 | */ |
| 4858 | private static boolean hasBaseCharacter(Matcher matcher, int i, |
| 4859 | CharSequence seq) |
| 4860 | { |
| 4861 | int start = (!matcher.transparentBounds) ? |
| 4862 | matcher.from : 0; |
| 4863 | for (int x=i; x >= start; x--) { |
| 4864 | int ch = Character.codePointAt(seq, x); |
| 4865 | if (Character.isLetterOrDigit(ch)) |
| 4866 | return true; |
| 4867 | if (Character.getType(ch) == Character.NON_SPACING_MARK) |
| 4868 | continue; |
| 4869 | return false; |
| 4870 | } |
| 4871 | return false; |
| 4872 | } |
| 4873 | |
| 4874 | /** |
| 4875 | * Attempts to match a slice in the input using the Boyer-Moore string |
| 4876 | * matching algorithm. The algorithm is based on the idea that the |
| 4877 | * pattern can be shifted farther ahead in the search text if it is |
| 4878 | * matched right to left. |
| 4879 | * <p> |
| 4880 | * The pattern is compared to the input one character at a time, from |
| 4881 | * the rightmost character in the pattern to the left. If the characters |
| 4882 | * all match the pattern has been found. If a character does not match, |
| 4883 | * the pattern is shifted right a distance that is the maximum of two |
| 4884 | * functions, the bad character shift and the good suffix shift. This |
| 4885 | * shift moves the attempted match position through the input more |
| 4886 | * quickly than a naive one position at a time check. |
| 4887 | * <p> |
| 4888 | * The bad character shift is based on the character from the text that |
| 4889 | * did not match. If the character does not appear in the pattern, the |
| 4890 | * pattern can be shifted completely beyond the bad character. If the |
| 4891 | * character does occur in the pattern, the pattern can be shifted to |
| 4892 | * line the pattern up with the next occurrence of that character. |
| 4893 | * <p> |
| 4894 | * The good suffix shift is based on the idea that some subset on the right |
| 4895 | * side of the pattern has matched. When a bad character is found, the |
| 4896 | * pattern can be shifted right by the pattern length if the subset does |
| 4897 | * not occur again in pattern, or by the amount of distance to the |
| 4898 | * next occurrence of the subset in the pattern. |
| 4899 | * |
| 4900 | * Boyer-Moore search methods adapted from code by Amy Yu. |
| 4901 | */ |
| 4902 | static class BnM extends Node { |
| 4903 | int[] buffer; |
| 4904 | int[] lastOcc; |
| 4905 | int[] optoSft; |
| 4906 | |
| 4907 | /** |
| 4908 | * Pre calculates arrays needed to generate the bad character |
| 4909 | * shift and the good suffix shift. Only the last seven bits |
| 4910 | * are used to see if chars match; This keeps the tables small |
| 4911 | * and covers the heavily used ASCII range, but occasionally |
| 4912 | * results in an aliased match for the bad character shift. |
| 4913 | */ |
| 4914 | static Node optimize(Node node) { |
| 4915 | if (!(node instanceof Slice)) { |
| 4916 | return node; |
| 4917 | } |
| 4918 | |
| 4919 | int[] src = ((Slice) node).buffer; |
| 4920 | int patternLength = src.length; |
| 4921 | // The BM algorithm requires a bit of overhead; |
| 4922 | // If the pattern is short don't use it, since |
| 4923 | // a shift larger than the pattern length cannot |
| 4924 | // be used anyway. |
| 4925 | if (patternLength < 4) { |
| 4926 | return node; |
| 4927 | } |
| 4928 | int i, j, k; |
| 4929 | int[] lastOcc = new int[128]; |
| 4930 | int[] optoSft = new int[patternLength]; |
| 4931 | // Precalculate part of the bad character shift |
| 4932 | // It is a table for where in the pattern each |
| 4933 | // lower 7-bit value occurs |
| 4934 | for (i = 0; i < patternLength; i++) { |
| 4935 | lastOcc[src[i]&0x7F] = i + 1; |
| 4936 | } |
| 4937 | // Precalculate the good suffix shift |
| 4938 | // i is the shift amount being considered |
| 4939 | NEXT: for (i = patternLength; i > 0; i--) { |
| 4940 | // j is the beginning index of suffix being considered |
| 4941 | for (j = patternLength - 1; j >= i; j--) { |
| 4942 | // Testing for good suffix |
| 4943 | if (src[j] == src[j-i]) { |
| 4944 | // src[j..len] is a good suffix |
| 4945 | optoSft[j-1] = i; |
| 4946 | } else { |
| 4947 | // No match. The array has already been |
| 4948 | // filled up with correct values before. |
| 4949 | continue NEXT; |
| 4950 | } |
| 4951 | } |
| 4952 | // This fills up the remaining of optoSft |
| 4953 | // any suffix can not have larger shift amount |
| 4954 | // then its sub-suffix. Why??? |
| 4955 | while (j > 0) { |
| 4956 | optoSft[--j] = i; |
| 4957 | } |
| 4958 | } |
| 4959 | // Set the guard value because of unicode compression |
| 4960 | optoSft[patternLength-1] = 1; |
| 4961 | if (node instanceof SliceS) |
| 4962 | return new BnMS(src, lastOcc, optoSft, node.next); |
| 4963 | return new BnM(src, lastOcc, optoSft, node.next); |
| 4964 | } |
| 4965 | BnM(int[] src, int[] lastOcc, int[] optoSft, Node next) { |
| 4966 | this.buffer = src; |
| 4967 | this.lastOcc = lastOcc; |
| 4968 | this.optoSft = optoSft; |
| 4969 | this.next = next; |
| 4970 | } |
| 4971 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 4972 | int[] src = buffer; |
| 4973 | int patternLength = src.length; |
| 4974 | int last = matcher.to - patternLength; |
| 4975 | |
| 4976 | // Loop over all possible match positions in text |
| 4977 | NEXT: while (i <= last) { |
| 4978 | // Loop over pattern from right to left |
| 4979 | for (int j = patternLength - 1; j >= 0; j--) { |
| 4980 | int ch = seq.charAt(i+j); |
| 4981 | if (ch != src[j]) { |
| 4982 | // Shift search to the right by the maximum of the |
| 4983 | // bad character shift and the good suffix shift |
| 4984 | i += Math.max(j + 1 - lastOcc[ch&0x7F], optoSft[j]); |
| 4985 | continue NEXT; |
| 4986 | } |
| 4987 | } |
| 4988 | // Entire pattern matched starting at i |
| 4989 | matcher.first = i; |
| 4990 | boolean ret = next.match(matcher, i + patternLength, seq); |
| 4991 | if (ret) { |
| 4992 | matcher.first = i; |
| 4993 | matcher.groups[0] = matcher.first; |
| 4994 | matcher.groups[1] = matcher.last; |
| 4995 | return true; |
| 4996 | } |
| 4997 | i++; |
| 4998 | } |
| 4999 | // BnM is only used as the leading node in the unanchored case, |
| 5000 | // and it replaced its Start() which always searches to the end |
| 5001 | // if it doesn't find what it's looking for, so hitEnd is true. |
| 5002 | matcher.hitEnd = true; |
| 5003 | return false; |
| 5004 | } |
| 5005 | boolean study(TreeInfo info) { |
| 5006 | info.minLength += buffer.length; |
| 5007 | info.maxValid = false; |
| 5008 | return next.study(info); |
| 5009 | } |
| 5010 | } |
| 5011 | |
| 5012 | /** |
| 5013 | * Supplementary support version of BnM(). Unpaired surrogates are |
| 5014 | * also handled by this class. |
| 5015 | */ |
| 5016 | static final class BnMS extends BnM { |
| 5017 | int lengthInChars; |
| 5018 | |
| 5019 | BnMS(int[] src, int[] lastOcc, int[] optoSft, Node next) { |
| 5020 | super(src, lastOcc, optoSft, next); |
| 5021 | for (int x = 0; x < buffer.length; x++) { |
| 5022 | lengthInChars += Character.charCount(buffer[x]); |
| 5023 | } |
| 5024 | } |
| 5025 | boolean match(Matcher matcher, int i, CharSequence seq) { |
| 5026 | int[] src = buffer; |
| 5027 | int patternLength = src.length; |
| 5028 | int last = matcher.to - lengthInChars; |
| 5029 | |
| 5030 | // Loop over all possible match positions in text |
| 5031 | NEXT: while (i <= last) { |
| 5032 | // Loop over pattern from right to left |
| 5033 | int ch; |
| 5034 | for (int j = countChars(seq, i, patternLength), x = patternLength - 1; |
| 5035 | j > 0; j -= Character.charCount(ch), x--) { |
| 5036 | ch = Character.codePointBefore(seq, i+j); |
| 5037 | if (ch != src[x]) { |
| 5038 | // Shift search to the right by the maximum of the |
| 5039 | // bad character shift and the good suffix shift |
| 5040 | int n = Math.max(x + 1 - lastOcc[ch&0x7F], optoSft[x]); |
| 5041 | i += countChars(seq, i, n); |
| 5042 | continue NEXT; |
| 5043 | } |
| 5044 | } |
| 5045 | // Entire pattern matched starting at i |
| 5046 | matcher.first = i; |
| 5047 | boolean ret = next.match(matcher, i + lengthInChars, seq); |
| 5048 | if (ret) { |
| 5049 | matcher.first = i; |
| 5050 | matcher.groups[0] = matcher.first; |
| 5051 | matcher.groups[1] = matcher.last; |
| 5052 | return true; |
| 5053 | } |
| 5054 | i += countChars(seq, i, 1); |
| 5055 | } |
| 5056 | matcher.hitEnd = true; |
| 5057 | return false; |
| 5058 | } |
| 5059 | } |
| 5060 | |
| 5061 | /////////////////////////////////////////////////////////////////////////////// |
| 5062 | /////////////////////////////////////////////////////////////////////////////// |
| 5063 | |
| 5064 | /** |
| 5065 | * This must be the very first initializer. |
| 5066 | */ |
| 5067 | static Node accept = new Node(); |
| 5068 | |
| 5069 | static Node lastAccept = new LastNode(); |
| 5070 | |
| 5071 | private static class CharPropertyNames { |
| 5072 | |
| 5073 | static CharProperty charPropertyFor(String name) { |
| 5074 | CharPropertyFactory m = map.get(name); |
| 5075 | return m == null ? null : m.make(); |
| 5076 | } |
| 5077 | |
| 5078 | private static abstract class CharPropertyFactory { |
| 5079 | abstract CharProperty make(); |
| 5080 | } |
| 5081 | |
| 5082 | private static void defCategory(String name, |
| 5083 | final int typeMask) { |
| 5084 | map.put(name, new CharPropertyFactory() { |
| 5085 | CharProperty make() { return new Category(typeMask);}}); |
| 5086 | } |
| 5087 | |
| 5088 | private static void defRange(String name, |
| 5089 | final int lower, final int upper) { |
| 5090 | map.put(name, new CharPropertyFactory() { |
| 5091 | CharProperty make() { return rangeFor(lower, upper);}}); |
| 5092 | } |
| 5093 | |
| 5094 | private static void defCtype(String name, |
| 5095 | final int ctype) { |
| 5096 | map.put(name, new CharPropertyFactory() { |
| 5097 | CharProperty make() { return new Ctype(ctype);}}); |
| 5098 | } |
| 5099 | |
| 5100 | private static abstract class CloneableProperty |
| 5101 | extends CharProperty implements Cloneable |
| 5102 | { |
| 5103 | public CloneableProperty clone() { |
| 5104 | try { |
| 5105 | return (CloneableProperty) super.clone(); |
| 5106 | } catch (CloneNotSupportedException e) { |
| 5107 | throw new AssertionError(e); |
| 5108 | } |
| 5109 | } |
| 5110 | } |
| 5111 | |
| 5112 | private static void defClone(String name, |
| 5113 | final CloneableProperty p) { |
| 5114 | map.put(name, new CharPropertyFactory() { |
| 5115 | CharProperty make() { return p.clone();}}); |
| 5116 | } |
| 5117 | |
| 5118 | private static final HashMap<String, CharPropertyFactory> map |
| 5119 | = new HashMap<String, CharPropertyFactory>(); |
| 5120 | |
| 5121 | static { |
| 5122 | // Unicode character property aliases, defined in |
| 5123 | // http://www.unicode.org/Public/UNIDATA/PropertyValueAliases.txt |
| 5124 | defCategory("Cn", 1<<Character.UNASSIGNED); |
| 5125 | defCategory("Lu", 1<<Character.UPPERCASE_LETTER); |
| 5126 | defCategory("Ll", 1<<Character.LOWERCASE_LETTER); |
| 5127 | defCategory("Lt", 1<<Character.TITLECASE_LETTER); |
| 5128 | defCategory("Lm", 1<<Character.MODIFIER_LETTER); |
| 5129 | defCategory("Lo", 1<<Character.OTHER_LETTER); |
| 5130 | defCategory("Mn", 1<<Character.NON_SPACING_MARK); |
| 5131 | defCategory("Me", 1<<Character.ENCLOSING_MARK); |
| 5132 | defCategory("Mc", 1<<Character.COMBINING_SPACING_MARK); |
| 5133 | defCategory("Nd", 1<<Character.DECIMAL_DIGIT_NUMBER); |
| 5134 | defCategory("Nl", 1<<Character.LETTER_NUMBER); |
| 5135 | defCategory("No", 1<<Character.OTHER_NUMBER); |
| 5136 | defCategory("Zs", 1<<Character.SPACE_SEPARATOR); |
| 5137 | defCategory("Zl", 1<<Character.LINE_SEPARATOR); |
| 5138 | defCategory("Zp", 1<<Character.PARAGRAPH_SEPARATOR); |
| 5139 | defCategory("Cc", 1<<Character.CONTROL); |
| 5140 | defCategory("Cf", 1<<Character.FORMAT); |
| 5141 | defCategory("Co", 1<<Character.PRIVATE_USE); |
| 5142 | defCategory("Cs", 1<<Character.SURROGATE); |
| 5143 | defCategory("Pd", 1<<Character.DASH_PUNCTUATION); |
| 5144 | defCategory("Ps", 1<<Character.START_PUNCTUATION); |
| 5145 | defCategory("Pe", 1<<Character.END_PUNCTUATION); |
| 5146 | defCategory("Pc", 1<<Character.CONNECTOR_PUNCTUATION); |
| 5147 | defCategory("Po", 1<<Character.OTHER_PUNCTUATION); |
| 5148 | defCategory("Sm", 1<<Character.MATH_SYMBOL); |
| 5149 | defCategory("Sc", 1<<Character.CURRENCY_SYMBOL); |
| 5150 | defCategory("Sk", 1<<Character.MODIFIER_SYMBOL); |
| 5151 | defCategory("So", 1<<Character.OTHER_SYMBOL); |
| 5152 | defCategory("Pi", 1<<Character.INITIAL_QUOTE_PUNCTUATION); |
| 5153 | defCategory("Pf", 1<<Character.FINAL_QUOTE_PUNCTUATION); |
| 5154 | defCategory("L", ((1<<Character.UPPERCASE_LETTER) | |
| 5155 | (1<<Character.LOWERCASE_LETTER) | |
| 5156 | (1<<Character.TITLECASE_LETTER) | |
| 5157 | (1<<Character.MODIFIER_LETTER) | |
| 5158 | (1<<Character.OTHER_LETTER))); |
| 5159 | defCategory("M", ((1<<Character.NON_SPACING_MARK) | |
| 5160 | (1<<Character.ENCLOSING_MARK) | |
| 5161 | (1<<Character.COMBINING_SPACING_MARK))); |
| 5162 | defCategory("N", ((1<<Character.DECIMAL_DIGIT_NUMBER) | |
| 5163 | (1<<Character.LETTER_NUMBER) | |
| 5164 | (1<<Character.OTHER_NUMBER))); |
| 5165 | defCategory("Z", ((1<<Character.SPACE_SEPARATOR) | |
| 5166 | (1<<Character.LINE_SEPARATOR) | |
| 5167 | (1<<Character.PARAGRAPH_SEPARATOR))); |
| 5168 | defCategory("C", ((1<<Character.CONTROL) | |
| 5169 | (1<<Character.FORMAT) | |
| 5170 | (1<<Character.PRIVATE_USE) | |
| 5171 | (1<<Character.SURROGATE))); // Other |
| 5172 | defCategory("P", ((1<<Character.DASH_PUNCTUATION) | |
| 5173 | (1<<Character.START_PUNCTUATION) | |
| 5174 | (1<<Character.END_PUNCTUATION) | |
| 5175 | (1<<Character.CONNECTOR_PUNCTUATION) | |
| 5176 | (1<<Character.OTHER_PUNCTUATION) | |
| 5177 | (1<<Character.INITIAL_QUOTE_PUNCTUATION) | |
| 5178 | (1<<Character.FINAL_QUOTE_PUNCTUATION))); |
| 5179 | defCategory("S", ((1<<Character.MATH_SYMBOL) | |
| 5180 | (1<<Character.CURRENCY_SYMBOL) | |
| 5181 | (1<<Character.MODIFIER_SYMBOL) | |
| 5182 | (1<<Character.OTHER_SYMBOL))); |
| 5183 | defCategory("LC", ((1<<Character.UPPERCASE_LETTER) | |
| 5184 | (1<<Character.LOWERCASE_LETTER) | |
| 5185 | (1<<Character.TITLECASE_LETTER))); |
| 5186 | defCategory("LD", ((1<<Character.UPPERCASE_LETTER) | |
| 5187 | (1<<Character.LOWERCASE_LETTER) | |
| 5188 | (1<<Character.TITLECASE_LETTER) | |
| 5189 | (1<<Character.MODIFIER_LETTER) | |
| 5190 | (1<<Character.OTHER_LETTER) | |
| 5191 | (1<<Character.DECIMAL_DIGIT_NUMBER))); |
| 5192 | defRange("L1", 0x00, 0xFF); // Latin-1 |
| 5193 | map.put("all", new CharPropertyFactory() { |
| 5194 | CharProperty make() { return new All(); }}); |
| 5195 | |
| 5196 | // Posix regular expression character classes, defined in |
| 5197 | // http://www.unix.org/onlinepubs/009695399/basedefs/xbd_chap09.html |
| 5198 | defRange("ASCII", 0x00, 0x7F); // ASCII |
| 5199 | defCtype("Alnum", ASCII.ALNUM); // Alphanumeric characters |
| 5200 | defCtype("Alpha", ASCII.ALPHA); // Alphabetic characters |
| 5201 | defCtype("Blank", ASCII.BLANK); // Space and tab characters |
| 5202 | defCtype("Cntrl", ASCII.CNTRL); // Control characters |
| 5203 | defRange("Digit", '0', '9'); // Numeric characters |
| 5204 | defCtype("Graph", ASCII.GRAPH); // printable and visible |
| 5205 | defRange("Lower", 'a', 'z'); // Lower-case alphabetic |
| 5206 | defRange("Print", 0x20, 0x7E); // Printable characters |
| 5207 | defCtype("Punct", ASCII.PUNCT); // Punctuation characters |
| 5208 | defCtype("Space", ASCII.SPACE); // Space characters |
| 5209 | defRange("Upper", 'A', 'Z'); // Upper-case alphabetic |
| 5210 | defCtype("XDigit",ASCII.XDIGIT); // hexadecimal digits |
| 5211 | |
| 5212 | // Java character properties, defined by methods in Character.java |
| 5213 | defClone("javaLowerCase", new CloneableProperty() { |
| 5214 | boolean isSatisfiedBy(int ch) { |
| 5215 | return Character.isLowerCase(ch);}}); |
| 5216 | defClone("javaUpperCase", new CloneableProperty() { |
| 5217 | boolean isSatisfiedBy(int ch) { |
| 5218 | return Character.isUpperCase(ch);}}); |
| 5219 | defClone("javaTitleCase", new CloneableProperty() { |
| 5220 | boolean isSatisfiedBy(int ch) { |
| 5221 | return Character.isTitleCase(ch);}}); |
| 5222 | defClone("javaDigit", new CloneableProperty() { |
| 5223 | boolean isSatisfiedBy(int ch) { |
| 5224 | return Character.isDigit(ch);}}); |
| 5225 | defClone("javaDefined", new CloneableProperty() { |
| 5226 | boolean isSatisfiedBy(int ch) { |
| 5227 | return Character.isDefined(ch);}}); |
| 5228 | defClone("javaLetter", new CloneableProperty() { |
| 5229 | boolean isSatisfiedBy(int ch) { |
| 5230 | return Character.isLetter(ch);}}); |
| 5231 | defClone("javaLetterOrDigit", new CloneableProperty() { |
| 5232 | boolean isSatisfiedBy(int ch) { |
| 5233 | return Character.isLetterOrDigit(ch);}}); |
| 5234 | defClone("javaJavaIdentifierStart", new CloneableProperty() { |
| 5235 | boolean isSatisfiedBy(int ch) { |
| 5236 | return Character.isJavaIdentifierStart(ch);}}); |
| 5237 | defClone("javaJavaIdentifierPart", new CloneableProperty() { |
| 5238 | boolean isSatisfiedBy(int ch) { |
| 5239 | return Character.isJavaIdentifierPart(ch);}}); |
| 5240 | defClone("javaUnicodeIdentifierStart", new CloneableProperty() { |
| 5241 | boolean isSatisfiedBy(int ch) { |
| 5242 | return Character.isUnicodeIdentifierStart(ch);}}); |
| 5243 | defClone("javaUnicodeIdentifierPart", new CloneableProperty() { |
| 5244 | boolean isSatisfiedBy(int ch) { |
| 5245 | return Character.isUnicodeIdentifierPart(ch);}}); |
| 5246 | defClone("javaIdentifierIgnorable", new CloneableProperty() { |
| 5247 | boolean isSatisfiedBy(int ch) { |
| 5248 | return Character.isIdentifierIgnorable(ch);}}); |
| 5249 | defClone("javaSpaceChar", new CloneableProperty() { |
| 5250 | boolean isSatisfiedBy(int ch) { |
| 5251 | return Character.isSpaceChar(ch);}}); |
| 5252 | defClone("javaWhitespace", new CloneableProperty() { |
| 5253 | boolean isSatisfiedBy(int ch) { |
| 5254 | return Character.isWhitespace(ch);}}); |
| 5255 | defClone("javaISOControl", new CloneableProperty() { |
| 5256 | boolean isSatisfiedBy(int ch) { |
| 5257 | return Character.isISOControl(ch);}}); |
| 5258 | defClone("javaMirrored", new CloneableProperty() { |
| 5259 | boolean isSatisfiedBy(int ch) { |
| 5260 | return Character.isMirrored(ch);}}); |
| 5261 | } |
| 5262 | } |
| 5263 | } |