1 /* 2 * Copyright (c) 1999, 2011, Oracle and/or its affiliates. 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. Oracle designates this 8 * particular file as subject to the "Classpath" exception as provided 9 * by Oracle 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 Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA 22 * or visit www.oracle.com if you need additional information or have any 23 * questions. 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.Locale; 33 import java.util.Map; 34 import java.util.ArrayList; 35 import java.util.HashMap; 36 import java.util.Arrays; 37 38 39 /** 40 * A compiled representation of a regular expression. 41 * 42 * <p> A regular expression, specified as a string, must first be compiled into 43 * an instance of this class. The resulting pattern can then be used to create 44 * a {@link Matcher} object that can match arbitrary {@link 45 * java.lang.CharSequence </code>character sequences<code>} against the regular 46 * expression. All of the state involved in performing a match resides in the 47 * matcher, so many matchers can share the same pattern. 48 * 49 * <p> A typical invocation sequence is thus 50 * 51 * <blockquote><pre> 52 * Pattern p = Pattern.{@link #compile compile}("a*b"); 53 * Matcher m = p.{@link #matcher matcher}("aaaaab"); 54 * boolean b = m.{@link Matcher#matches matches}();</pre></blockquote> 55 * 56 * <p> A {@link #matches matches} method is defined by this class as a 57 * convenience for when a regular expression is used just once. This method 58 * compiles an expression and matches an input sequence against it in a single 59 * invocation. The statement 60 * 61 * <blockquote><pre> 62 * boolean b = Pattern.matches("a*b", "aaaaab");</pre></blockquote> 63 * 64 * is equivalent to the three statements above, though for repeated matches it 65 * is less efficient since it does not allow the compiled pattern to be reused. 66 * 67 * <p> Instances of this class are immutable and are safe for use by multiple 68 * concurrent threads. Instances of the {@link Matcher} class are not safe for 69 * such use. 70 * 71 * 72 * <a name="sum"> 73 * <h4> Summary of regular-expression constructs </h4> 74 * 75 * <table border="0" cellpadding="1" cellspacing="0" 76 * summary="Regular expression constructs, and what they match"> 77 * 78 * <tr align="left"> 79 * <th bgcolor="#CCCCFF" align="left" id="construct">Construct</th> 80 * <th bgcolor="#CCCCFF" align="left" id="matches">Matches</th> 81 * </tr> 82 * 83 * <tr><th> </th></tr> 84 * <tr align="left"><th colspan="2" id="characters">Characters</th></tr> 85 * 86 * <tr><td valign="top" headers="construct characters"><i>x</i></td> 87 * <td headers="matches">The character <i>x</i></td></tr> 88 * <tr><td valign="top" headers="construct characters"><tt>\\</tt></td> 89 * <td headers="matches">The backslash character</td></tr> 90 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>n</i></td> 91 * <td headers="matches">The character with octal value <tt>0</tt><i>n</i> 92 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 93 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>nn</i></td> 94 * <td headers="matches">The character with octal value <tt>0</tt><i>nn</i> 95 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 96 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>mnn</i></td> 97 * <td headers="matches">The character with octal value <tt>0</tt><i>mnn</i> 98 * (0 <tt><=</tt> <i>m</i> <tt><=</tt> 3, 99 * 0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 100 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>hh</i></td> 101 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hh</i></td></tr> 102 * <tr><td valign="top" headers="construct characters"><tt>\u</tt><i>hhhh</i></td> 103 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hhhh</i></td></tr> 104 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>{h...h}</i></td> 105 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>h...h</i> 106 * ({@link java.lang.Character#MIN_CODE_POINT Character.MIN_CODE_POINT} 107 * <= <tt>0x</tt><i>h...h</i> <=  108 * {@link java.lang.Character#MAX_CODE_POINT Character.MAX_CODE_POINT})</td></tr> 109 * <tr><td valign="top" headers="matches"><tt>\t</tt></td> 110 * <td headers="matches">The tab character (<tt>'\u0009'</tt>)</td></tr> 111 * <tr><td valign="top" headers="construct characters"><tt>\n</tt></td> 112 * <td headers="matches">The newline (line feed) character (<tt>'\u000A'</tt>)</td></tr> 113 * <tr><td valign="top" headers="construct characters"><tt>\r</tt></td> 114 * <td headers="matches">The carriage-return character (<tt>'\u000D'</tt>)</td></tr> 115 * <tr><td valign="top" headers="construct characters"><tt>\f</tt></td> 116 * <td headers="matches">The form-feed character (<tt>'\u000C'</tt>)</td></tr> 117 * <tr><td valign="top" headers="construct characters"><tt>\a</tt></td> 118 * <td headers="matches">The alert (bell) character (<tt>'\u0007'</tt>)</td></tr> 119 * <tr><td valign="top" headers="construct characters"><tt>\e</tt></td> 120 * <td headers="matches">The escape character (<tt>'\u001B'</tt>)</td></tr> 121 * <tr><td valign="top" headers="construct characters"><tt>\c</tt><i>x</i></td> 122 * <td headers="matches">The control character corresponding to <i>x</i></td></tr> 123 * 124 * <tr><th> </th></tr> 125 * <tr align="left"><th colspan="2" id="classes">Character classes</th></tr> 126 * 127 * <tr><td valign="top" headers="construct classes"><tt>[abc]</tt></td> 128 * <td headers="matches"><tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (simple class)</td></tr> 129 * <tr><td valign="top" headers="construct classes"><tt>[^abc]</tt></td> 130 * <td headers="matches">Any character except <tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (negation)</td></tr> 131 * <tr><td valign="top" headers="construct classes"><tt>[a-zA-Z]</tt></td> 132 * <td headers="matches"><tt>a</tt> through <tt>z</tt> 133 * or <tt>A</tt> through <tt>Z</tt>, inclusive (range)</td></tr> 134 * <tr><td valign="top" headers="construct classes"><tt>[a-d[m-p]]</tt></td> 135 * <td headers="matches"><tt>a</tt> through <tt>d</tt>, 136 * or <tt>m</tt> through <tt>p</tt>: <tt>[a-dm-p]</tt> (union)</td></tr> 137 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[def]]</tt></td> 138 * <td headers="matches"><tt>d</tt>, <tt>e</tt>, or <tt>f</tt> (intersection)</tr> 139 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^bc]]</tt></td> 140 * <td headers="matches"><tt>a</tt> through <tt>z</tt>, 141 * except for <tt>b</tt> and <tt>c</tt>: <tt>[ad-z]</tt> (subtraction)</td></tr> 142 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^m-p]]</tt></td> 143 * <td headers="matches"><tt>a</tt> through <tt>z</tt>, 144 * and not <tt>m</tt> through <tt>p</tt>: <tt>[a-lq-z]</tt>(subtraction)</td></tr> 145 * <tr><th> </th></tr> 146 * 147 * <tr align="left"><th colspan="2" id="predef">Predefined character classes</th></tr> 148 * 149 * <tr><td valign="top" headers="construct predef"><tt>.</tt></td> 150 * <td headers="matches">Any character (may or may not match <a href="#lt">line terminators</a>)</td></tr> 151 * <tr><td valign="top" headers="construct predef"><tt>\d</tt></td> 152 * <td headers="matches">A digit: <tt>[0-9]</tt></td></tr> 153 * <tr><td valign="top" headers="construct predef"><tt>\D</tt></td> 154 * <td headers="matches">A non-digit: <tt>[^0-9]</tt></td></tr> 155 * <tr><td valign="top" headers="construct predef"><tt>\s</tt></td> 156 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 157 * <tr><td valign="top" headers="construct predef"><tt>\S</tt></td> 158 * <td headers="matches">A non-whitespace character: <tt>[^\s]</tt></td></tr> 159 * <tr><td valign="top" headers="construct predef"><tt>\w</tt></td> 160 * <td headers="matches">A word character: <tt>[a-zA-Z_0-9]</tt></td></tr> 161 * <tr><td valign="top" headers="construct predef"><tt>\W</tt></td> 162 * <td headers="matches">A non-word character: <tt>[^\w]</tt></td></tr> 163 * 164 * <tr><th> </th></tr> 165 * <tr align="left"><th colspan="2" id="posix">POSIX character classes</b> (US-ASCII only)<b></th></tr> 166 * 167 * <tr><td valign="top" headers="construct posix"><tt>\p{Lower}</tt></td> 168 * <td headers="matches">A lower-case alphabetic character: <tt>[a-z]</tt></td></tr> 169 * <tr><td valign="top" headers="construct posix"><tt>\p{Upper}</tt></td> 170 * <td headers="matches">An upper-case alphabetic character:<tt>[A-Z]</tt></td></tr> 171 * <tr><td valign="top" headers="construct posix"><tt>\p{ASCII}</tt></td> 172 * <td headers="matches">All ASCII:<tt>[\x00-\x7F]</tt></td></tr> 173 * <tr><td valign="top" headers="construct posix"><tt>\p{Alpha}</tt></td> 174 * <td headers="matches">An alphabetic character:<tt>[\p{Lower}\p{Upper}]</tt></td></tr> 175 * <tr><td valign="top" headers="construct posix"><tt>\p{Digit}</tt></td> 176 * <td headers="matches">A decimal digit: <tt>[0-9]</tt></td></tr> 177 * <tr><td valign="top" headers="construct posix"><tt>\p{Alnum}</tt></td> 178 * <td headers="matches">An alphanumeric character:<tt>[\p{Alpha}\p{Digit}]</tt></td></tr> 179 * <tr><td valign="top" headers="construct posix"><tt>\p{Punct}</tt></td> 180 * <td headers="matches">Punctuation: One of <tt>!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~</tt></td></tr> 181 * <!-- <tt>[\!"#\$%&'\(\)\*\+,\-\./:;\<=\>\?@\[\\\]\^_`\{\|\}~]</tt> 182 * <tt>[\X21-\X2F\X31-\X40\X5B-\X60\X7B-\X7E]</tt> --> 183 * <tr><td valign="top" headers="construct posix"><tt>\p{Graph}</tt></td> 184 * <td headers="matches">A visible character: <tt>[\p{Alnum}\p{Punct}]</tt></td></tr> 185 * <tr><td valign="top" headers="construct posix"><tt>\p{Print}</tt></td> 186 * <td headers="matches">A printable character: <tt>[\p{Graph}\x20]</tt></td></tr> 187 * <tr><td valign="top" headers="construct posix"><tt>\p{Blank}</tt></td> 188 * <td headers="matches">A space or a tab: <tt>[ \t]</tt></td></tr> 189 * <tr><td valign="top" headers="construct posix"><tt>\p{Cntrl}</tt></td> 190 * <td headers="matches">A control character: <tt>[\x00-\x1F\x7F]</tt></td></tr> 191 * <tr><td valign="top" headers="construct posix"><tt>\p{XDigit}</tt></td> 192 * <td headers="matches">A hexadecimal digit: <tt>[0-9a-fA-F]</tt></td></tr> 193 * <tr><td valign="top" headers="construct posix"><tt>\p{Space}</tt></td> 194 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 195 * 196 * <tr><th> </th></tr> 197 * <tr align="left"><th colspan="2">java.lang.Character classes (simple <a href="#jcc">java character type</a>)</th></tr> 198 * 199 * <tr><td valign="top"><tt>\p{javaLowerCase}</tt></td> 200 * <td>Equivalent to java.lang.Character.isLowerCase()</td></tr> 201 * <tr><td valign="top"><tt>\p{javaUpperCase}</tt></td> 202 * <td>Equivalent to java.lang.Character.isUpperCase()</td></tr> 203 * <tr><td valign="top"><tt>\p{javaWhitespace}</tt></td> 204 * <td>Equivalent to java.lang.Character.isWhitespace()</td></tr> 205 * <tr><td valign="top"><tt>\p{javaMirrored}</tt></td> 206 * <td>Equivalent to java.lang.Character.isMirrored()</td></tr> 207 * 208 * <tr><th> </th></tr> 209 * <tr align="left"><th colspan="2" id="unicode">Classes for Unicode scripts, blocks and categories</th></tr> 210 * * <tr><td valign="top" headers="construct unicode"><tt>\p{IsLatin}</tt></td> 211 * <td headers="matches">A Latin script character (simple <a href="#ubc">script</a>)</td></tr> 212 * <tr><td valign="top" headers="construct unicode"><tt>\p{InGreek}</tt></td> 213 * <td headers="matches">A character in the Greek block (simple <a href="#ubc">block</a>)</td></tr> 214 * <tr><td valign="top" headers="construct unicode"><tt>\p{Lu}</tt></td> 215 * <td headers="matches">An uppercase letter (simple <a href="#ubc">category</a>)</td></tr> 216 * <tr><td valign="top" headers="construct unicode"><tt>\p{Sc}</tt></td> 217 * <td headers="matches">A currency symbol</td></tr> 218 * <tr><td valign="top" headers="construct unicode"><tt>\P{InGreek}</tt></td> 219 * <td headers="matches">Any character except one in the Greek block (negation)</td></tr> 220 * <tr><td valign="top" headers="construct unicode"><tt>[\p{L}&&[^\p{Lu}]] </tt></td> 221 * <td headers="matches">Any letter except an uppercase letter (subtraction)</td></tr> 222 * 223 * <tr><th> </th></tr> 224 * <tr align="left"><th colspan="2" id="bounds">Boundary matchers</th></tr> 225 * 226 * <tr><td valign="top" headers="construct bounds"><tt>^</tt></td> 227 * <td headers="matches">The beginning of a line</td></tr> 228 * <tr><td valign="top" headers="construct bounds"><tt>$</tt></td> 229 * <td headers="matches">The end of a line</td></tr> 230 * <tr><td valign="top" headers="construct bounds"><tt>\b</tt></td> 231 * <td headers="matches">A word boundary</td></tr> 232 * <tr><td valign="top" headers="construct bounds"><tt>\B</tt></td> 233 * <td headers="matches">A non-word boundary</td></tr> 234 * <tr><td valign="top" headers="construct bounds"><tt>\A</tt></td> 235 * <td headers="matches">The beginning of the input</td></tr> 236 * <tr><td valign="top" headers="construct bounds"><tt>\G</tt></td> 237 * <td headers="matches">The end of the previous match</td></tr> 238 * <tr><td valign="top" headers="construct bounds"><tt>\Z</tt></td> 239 * <td headers="matches">The end of the input but for the final 240 * <a href="#lt">terminator</a>, if any</td></tr> 241 * <tr><td valign="top" headers="construct bounds"><tt>\z</tt></td> 242 * <td headers="matches">The end of the input</td></tr> 243 * 244 * <tr><th> </th></tr> 245 * <tr align="left"><th colspan="2" id="greedy">Greedy quantifiers</th></tr> 246 * 247 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>?</tt></td> 248 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 249 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>*</tt></td> 250 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 251 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>+</tt></td> 252 * <td headers="matches"><i>X</i>, one or more times</td></tr> 253 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>}</tt></td> 254 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 255 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,}</tt></td> 256 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 257 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}</tt></td> 258 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 259 * 260 * <tr><th> </th></tr> 261 * <tr align="left"><th colspan="2" id="reluc">Reluctant quantifiers</th></tr> 262 * 263 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>??</tt></td> 264 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 265 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>*?</tt></td> 266 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 267 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>+?</tt></td> 268 * <td headers="matches"><i>X</i>, one or more times</td></tr> 269 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>}?</tt></td> 270 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 271 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,}?</tt></td> 272 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 273 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}?</tt></td> 274 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 275 * 276 * <tr><th> </th></tr> 277 * <tr align="left"><th colspan="2" id="poss">Possessive quantifiers</th></tr> 278 * 279 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>?+</tt></td> 280 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 281 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>*+</tt></td> 282 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 283 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>++</tt></td> 284 * <td headers="matches"><i>X</i>, one or more times</td></tr> 285 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>}+</tt></td> 286 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 287 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,}+</tt></td> 288 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 289 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}+</tt></td> 290 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 291 * 292 * <tr><th> </th></tr> 293 * <tr align="left"><th colspan="2" id="logical">Logical operators</th></tr> 294 * 295 * <tr><td valign="top" headers="construct logical"><i>XY</i></td> 296 * <td headers="matches"><i>X</i> followed by <i>Y</i></td></tr> 297 * <tr><td valign="top" headers="construct logical"><i>X</i><tt>|</tt><i>Y</i></td> 298 * <td headers="matches">Either <i>X</i> or <i>Y</i></td></tr> 299 * <tr><td valign="top" headers="construct logical"><tt>(</tt><i>X</i><tt>)</tt></td> 300 * <td headers="matches">X, as a <a href="#cg">capturing group</a></td></tr> 301 * 302 * <tr><th> </th></tr> 303 * <tr align="left"><th colspan="2" id="backref">Back references</th></tr> 304 * 305 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>n</i></td> 306 * <td valign="bottom" headers="matches">Whatever the <i>n</i><sup>th</sup> 307 * <a href="#cg">capturing group</a> matched</td></tr> 308 * 309 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>k</i><<i>name</i>></td> 310 * <td valign="bottom" headers="matches">Whatever the 311 * <a href="#groupname">named-capturing group</a> "name" matched</td></tr> 312 * 313 * <tr><th> </th></tr> 314 * <tr align="left"><th colspan="2" id="quot">Quotation</th></tr> 315 * 316 * <tr><td valign="top" headers="construct quot"><tt>\</tt></td> 317 * <td headers="matches">Nothing, but quotes the following character</td></tr> 318 * <tr><td valign="top" headers="construct quot"><tt>\Q</tt></td> 319 * <td headers="matches">Nothing, but quotes all characters until <tt>\E</tt></td></tr> 320 * <tr><td valign="top" headers="construct quot"><tt>\E</tt></td> 321 * <td headers="matches">Nothing, but ends quoting started by <tt>\Q</tt></td></tr> 322 * <!-- Metachars: !$()*+.<>?[\]^{|} --> 323 * 324 * <tr><th> </th></tr> 325 * <tr align="left"><th colspan="2" id="special">Special constructs (named-capturing and non-capturing)</th></tr> 326 * 327 * <tr><td valign="top" headers="construct special"><tt>(?<<a href="#groupname">name</a>></tt><i>X</i><tt>)</tt></td> 328 * <td headers="matches"><i>X</i>, as a named-capturing group</td></tr> 329 * <tr><td valign="top" headers="construct special"><tt>(?:</tt><i>X</i><tt>)</tt></td> 330 * <td headers="matches"><i>X</i>, as a non-capturing group</td></tr> 331 * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux) </tt></td> 332 * <td headers="matches">Nothing, but turns match flags <a href="#CASE_INSENSITIVE">i</a> 333 * <a href="#UNIX_LINES">d</a> <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> 334 * <a href="#UNICODE_CASE">u</a> <a href="#COMMENTS">x</a> on - off</td></tr> 335 * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux:</tt><i>X</i><tt>)</tt> </td> 336 * <td headers="matches"><i>X</i>, as a <a href="#cg">non-capturing group</a> with the 337 * given flags <a href="#CASE_INSENSITIVE">i</a> <a href="#UNIX_LINES">d</a> 338 * <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> <a href="#UNICODE_CASE">u</a > 339 * <a href="#COMMENTS">x</a> on - off</td></tr> 340 * <tr><td valign="top" headers="construct special"><tt>(?=</tt><i>X</i><tt>)</tt></td> 341 * <td headers="matches"><i>X</i>, via zero-width positive lookahead</td></tr> 342 * <tr><td valign="top" headers="construct special"><tt>(?!</tt><i>X</i><tt>)</tt></td> 343 * <td headers="matches"><i>X</i>, via zero-width negative lookahead</td></tr> 344 * <tr><td valign="top" headers="construct special"><tt>(?<=</tt><i>X</i><tt>)</tt></td> 345 * <td headers="matches"><i>X</i>, via zero-width positive lookbehind</td></tr> 346 * <tr><td valign="top" headers="construct special"><tt>(?<!</tt><i>X</i><tt>)</tt></td> 347 * <td headers="matches"><i>X</i>, via zero-width negative lookbehind</td></tr> 348 * <tr><td valign="top" headers="construct special"><tt>(?></tt><i>X</i><tt>)</tt></td> 349 * <td headers="matches"><i>X</i>, as an independent, non-capturing group</td></tr> 350 * 351 * </table> 352 * 353 * <hr> 354 * 355 * 356 * <a name="bs"> 357 * <h4> Backslashes, escapes, and quoting </h4> 358 * 359 * <p> The backslash character (<tt>'\'</tt>) serves to introduce escaped 360 * constructs, as defined in the table above, as well as to quote characters 361 * that otherwise would be interpreted as unescaped constructs. Thus the 362 * expression <tt>\\</tt> matches a single backslash and <tt>\{</tt> matches a 363 * left brace. 364 * 365 * <p> It is an error to use a backslash prior to any alphabetic character that 366 * does not denote an escaped construct; these are reserved for future 367 * extensions to the regular-expression language. A backslash may be used 368 * prior to a non-alphabetic character regardless of whether that character is 369 * part of an unescaped construct. 370 * 371 * <p> Backslashes within string literals in Java source code are interpreted 372 * as required by the <a 373 * href="http://java.sun.com/docs/books/jls">Java Language 374 * Specification</a> as either <a 375 * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#100850">Unicode 376 * escapes</a> or other <a 377 * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#101089">character 378 * escapes</a>. It is therefore necessary to double backslashes in string 379 * literals that represent regular expressions to protect them from 380 * interpretation by the Java bytecode compiler. The string literal 381 * <tt>"\b"</tt>, for example, matches a single backspace character when 382 * interpreted as a regular expression, while <tt>"\\b"</tt> matches a 383 * word boundary. The string literal <tt>"\(hello\)"</tt> is illegal 384 * and leads to a compile-time error; in order to match the string 385 * <tt>(hello)</tt> the string literal <tt>"\\(hello\\)"</tt> 386 * must be used. 387 * 388 * <a name="cc"> 389 * <h4> Character Classes </h4> 390 * 391 * <p> Character classes may appear within other character classes, and 392 * may be composed by the union operator (implicit) and the intersection 393 * operator (<tt>&&</tt>). 394 * The union operator denotes a class that contains every character that is 395 * in at least one of its operand classes. The intersection operator 396 * denotes a class that contains every character that is in both of its 397 * operand classes. 398 * 399 * <p> The precedence of character-class operators is as follows, from 400 * highest to lowest: 401 * 402 * <blockquote><table border="0" cellpadding="1" cellspacing="0" 403 * summary="Precedence of character class operators."> 404 * <tr><th>1 </th> 405 * <td>Literal escape </td> 406 * <td><tt>\x</tt></td></tr> 407 * <tr><th>2 </th> 408 * <td>Grouping</td> 409 * <td><tt>[...]</tt></td></tr> 410 * <tr><th>3 </th> 411 * <td>Range</td> 412 * <td><tt>a-z</tt></td></tr> 413 * <tr><th>4 </th> 414 * <td>Union</td> 415 * <td><tt>[a-e][i-u]</tt></td></tr> 416 * <tr><th>5 </th> 417 * <td>Intersection</td> 418 * <td><tt>[a-z&&[aeiou]]</tt></td></tr> 419 * </table></blockquote> 420 * 421 * <p> Note that a different set of metacharacters are in effect inside 422 * a character class than outside a character class. For instance, the 423 * regular expression <tt>.</tt> loses its special meaning inside a 424 * character class, while the expression <tt>-</tt> becomes a range 425 * forming metacharacter. 426 * 427 * <a name="lt"> 428 * <h4> Line terminators </h4> 429 * 430 * <p> A <i>line terminator</i> is a one- or two-character sequence that marks 431 * the end of a line of the input character sequence. The following are 432 * recognized as line terminators: 433 * 434 * <ul> 435 * 436 * <li> A newline (line feed) character (<tt>'\n'</tt>), 437 * 438 * <li> A carriage-return character followed immediately by a newline 439 * character (<tt>"\r\n"</tt>), 440 * 441 * <li> A standalone carriage-return character (<tt>'\r'</tt>), 442 * 443 * <li> A next-line character (<tt>'\u0085'</tt>), 444 * 445 * <li> A line-separator character (<tt>'\u2028'</tt>), or 446 * 447 * <li> A paragraph-separator character (<tt>'\u2029</tt>). 448 * 449 * </ul> 450 * <p>If {@link #UNIX_LINES} mode is activated, then the only line terminators 451 * recognized are newline characters. 452 * 453 * <p> The regular expression <tt>.</tt> matches any character except a line 454 * terminator unless the {@link #DOTALL} flag is specified. 455 * 456 * <p> By default, the regular expressions <tt>^</tt> and <tt>$</tt> ignore 457 * line terminators and only match at the beginning and the end, respectively, 458 * of the entire input sequence. If {@link #MULTILINE} mode is activated then 459 * <tt>^</tt> matches at the beginning of input and after any line terminator 460 * except at the end of input. When in {@link #MULTILINE} mode <tt>$</tt> 461 * matches just before a line terminator or the end of the input sequence. 462 * 463 * <a name="cg"> 464 * <h4> Groups and capturing </h4> 465 * 466 * <a name="gnumber"> 467 * <h5> Group number </h5> 468 * <p> Capturing groups are numbered by counting their opening parentheses from 469 * left to right. In the expression <tt>((A)(B(C)))</tt>, for example, there 470 * are four such groups: </p> 471 * 472 * <blockquote><table cellpadding=1 cellspacing=0 summary="Capturing group numberings"> 473 * <tr><th>1 </th> 474 * <td><tt>((A)(B(C)))</tt></td></tr> 475 * <tr><th>2 </th> 476 * <td><tt>(A)</tt></td></tr> 477 * <tr><th>3 </th> 478 * <td><tt>(B(C))</tt></td></tr> 479 * <tr><th>4 </th> 480 * <td><tt>(C)</tt></td></tr> 481 * </table></blockquote> 482 * 483 * <p> Group zero always stands for the entire expression. 484 * 485 * <p> Capturing groups are so named because, during a match, each subsequence 486 * of the input sequence that matches such a group is saved. The captured 487 * subsequence may be used later in the expression, via a back reference, and 488 * may also be retrieved from the matcher once the match operation is complete. 489 * 490 * <a name="groupname"> 491 * <h5> Group name </h5> 492 * <p>A capturing group can also be assigned a "name", a <tt>named-capturing group</tt>, 493 * and then be back-referenced later by the "name". Group names are composed of 494 * the following characters. The first character must be a <tt>letter</tt>. 495 * 496 * <ul> 497 * <li> The uppercase letters <tt>'A'</tt> through <tt>'Z'</tt> 498 * (<tt>'\u0041'</tt> through <tt>'\u005a'</tt>), 499 * <li> The lowercase letters <tt>'a'</tt> through <tt>'z'</tt> 500 * (<tt>'\u0061'</tt> through <tt>'\u007a'</tt>), 501 * <li> The digits <tt>'0'</tt> through <tt>'9'</tt> 502 * (<tt>'\u0030'</tt> through <tt>'\u0039'</tt>), 503 * </ul> 504 * 505 * <p> A <tt>named-capturing group</tt> is still numbered as described in 506 * <a href="#gnumber">Group number</a>. 507 * 508 * <p> The captured input associated with a group is always the subsequence 509 * that the group most recently matched. If a group is evaluated a second time 510 * because of quantification then its previously-captured value, if any, will 511 * be retained if the second evaluation fails. Matching the string 512 * <tt>"aba"</tt> against the expression <tt>(a(b)?)+</tt>, for example, leaves 513 * group two set to <tt>"b"</tt>. All captured input is discarded at the 514 * beginning of each match. 515 * 516 * <p> Groups beginning with <tt>(?</tt> are either pure, <i>non-capturing</i> groups 517 * that do not capture text and do not count towards the group total, or 518 * <i>named-capturing</i> group. 519 * 520 * <h4> Unicode support </h4> 521 * 522 * <p> This class is in conformance with Level 1 of <a 523 * href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical 524 * Standard #18: Unicode Regular Expression Guidelines</i></a>, plus RL2.1 525 * Canonical Equivalents. 526 * 527 * <p> Unicode escape sequences such as <tt>\u2014</tt> in Java source code 528 * are processed as described in <a 529 * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#100850">\u00A73.3</a> 530 * of the Java Language Specification. Such escape sequences are also 531 * implemented directly by the regular-expression parser so that Unicode 532 * escapes can be used in expressions that are read from files or from the 533 * keyboard. Thus the strings <tt>"\u2014"</tt> and <tt>"\\u2014"</tt>, 534 * while not equal, compile into the same pattern, which matches the character 535 * with hexadecimal value <tt>0x2014</tt>. 536 * 537 * <p> A Unicode character can also be represented in a regular-expression by 538 * using its hexadecimal code point value directly as described in construct 539 * <tt>\x{...}</tt>, for example a supplementary character U+2011F 540 * can be specified as <tt>\x{2011F}</tt>, instead of two consecutive 541 * Unicode escape sequences of the surrogate pair 542 * <tt>\uD840</tt><tt>\uDD1F</tt>. 543 * 544 * <a name="ubc"> 545 * <p>Unicode scripts, blocks and categories are written with the <tt>\p</tt> and 546 * <tt>\P</tt> constructs as in Perl. <tt>\p{</tt><i>prop</i><tt>}</tt> matches if 547 * the input has the property <i>prop</i>, while <tt>\P{</tt><i>prop</i><tt>}</tt> 548 * does not match if the input has that property. 549 * <p> 550 * Scripts are specified either with the prefix {@code Is}, as in 551 * {@code IsHiragana}, or by using the {@code script} keyword (or its short 552 * form {@code sc})as in {@code script=Hiragana} or {@code sc=Hiragana}. 553 * <p> 554 * Blocks are specified with the prefix {@code In}, as in 555 * {@code InMongolian}, or by using the keyword {@code block} (or its short 556 * form {@code blk}) as in {@code block=Mongolian} or {@code blk=Mongolian}. 557 * <p> 558 * Categories may be specified with the optional prefix {@code Is}: 559 * Both {@code \p{L}} and {@code \p{IsL}} denote the category of Unicode 560 * letters. Same as scripts and blocks, categories can also be specified 561 * by using the keyword {@code general_category} (or its short form 562 * {@code gc}) as in {@code general_category=Lu} or {@code gc=Lu}. 563 * <p> 564 * Scripts, blocks and categories can be used both inside and outside of a 565 * character class. 566 * <p> The supported categories are those of 567 * <a href="http://www.unicode.org/unicode/standard/standard.html"> 568 * <i>The Unicode Standard</i></a> in the version specified by the 569 * {@link java.lang.Character Character} class. The category names are those 570 * defined in the Standard, both normative and informative. 571 * The script names supported by <code>Pattern</code> are the valid script names 572 * accepted and defined by 573 * {@link java.lang.Character.UnicodeScript#forName(String) UnicodeScript.forName}. 574 * The block names supported by <code>Pattern</code> are the valid block names 575 * accepted and defined by 576 * {@link java.lang.Character.UnicodeBlock#forName(String) UnicodeBlock.forName}. 577 * <p> 578 * <a name="jcc"> <p>Categories that behave like the java.lang.Character 579 * boolean is<i>methodname</i> methods (except for the deprecated ones) are 580 * available through the same <tt>\p{</tt><i>prop</i><tt>}</tt> syntax where 581 * the specified property has the name <tt>java<i>methodname</i></tt>. 582 * 583 * <h4> Comparison to Perl 5 </h4> 584 * 585 * <p>The <code>Pattern</code> engine performs traditional NFA-based matching 586 * with ordered alternation as occurs in Perl 5. 587 * 588 * <p> Perl constructs not supported by this class: </p> 589 * 590 * <ul> 591 * 592 * <li><p> The conditional constructs <tt>(?{</tt><i>X</i><tt>})</tt> and 593 * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>|</tt><i>Y</i><tt>)</tt>, 594 * </p></li> 595 * 596 * <li><p> The embedded code constructs <tt>(?{</tt><i>code</i><tt>})</tt> 597 * and <tt>(??{</tt><i>code</i><tt>})</tt>,</p></li> 598 * 599 * <li><p> The embedded comment syntax <tt>(?#comment)</tt>, and </p></li> 600 * 601 * <li><p> The preprocessing operations <tt>\l</tt> <tt>\u</tt>, 602 * <tt>\L</tt>, and <tt>\U</tt>. </p></li> 603 * 604 * </ul> 605 * 606 * <p> Constructs supported by this class but not by Perl: </p> 607 * 608 * <ul> 609 * 610 * <li><p> Possessive quantifiers, which greedily match as much as they can 611 * and do not back off, even when doing so would allow the overall match to 612 * succeed. </p></li> 613 * 614 * <li><p> Character-class union and intersection as described 615 * <a href="#cc">above</a>.</p></li> 616 * 617 * </ul> 618 * 619 * <p> Notable differences from Perl: </p> 620 * 621 * <ul> 622 * 623 * <li><p> In Perl, <tt>\1</tt> through <tt>\9</tt> are always interpreted 624 * as back references; a backslash-escaped number greater than <tt>9</tt> is 625 * treated as a back reference if at least that many subexpressions exist, 626 * otherwise it is interpreted, if possible, as an octal escape. In this 627 * class octal escapes must always begin with a zero. In this class, 628 * <tt>\1</tt> through <tt>\9</tt> are always interpreted as back 629 * references, and a larger number is accepted as a back reference if at 630 * least that many subexpressions exist at that point in the regular 631 * expression, otherwise the parser will drop digits until the number is 632 * smaller or equal to the existing number of groups or it is one digit. 633 * </p></li> 634 * 635 * <li><p> Perl uses the <tt>g</tt> flag to request a match that resumes 636 * where the last match left off. This functionality is provided implicitly 637 * by the {@link Matcher} class: Repeated invocations of the {@link 638 * Matcher#find find} method will resume where the last match left off, 639 * unless the matcher is reset. </p></li> 640 * 641 * <li><p> In Perl, embedded flags at the top level of an expression affect 642 * the whole expression. In this class, embedded flags always take effect 643 * at the point at which they appear, whether they are at the top level or 644 * within a group; in the latter case, flags are restored at the end of the 645 * group just as in Perl. </p></li> 646 * 647 * <li><p> Perl is forgiving about malformed matching constructs, as in the 648 * expression <tt>*a</tt>, as well as dangling brackets, as in the 649 * expression <tt>abc]</tt>, and treats them as literals. This 650 * class also accepts dangling brackets but is strict about dangling 651 * metacharacters like +, ? and *, and will throw a 652 * {@link PatternSyntaxException} if it encounters them. </p></li> 653 * 654 * </ul> 655 * 656 * 657 * <p> For a more precise description of the behavior of regular expression 658 * constructs, please see <a href="http://www.oreilly.com/catalog/regex3/"> 659 * <i>Mastering Regular Expressions, 3nd Edition</i>, Jeffrey E. F. Friedl, 660 * O'Reilly and Associates, 2006.</a> 661 * </p> 662 * 663 * @see java.lang.String#split(String, int) 664 * @see java.lang.String#split(String) 665 * 666 * @author Mike McCloskey 667 * @author Mark Reinhold 668 * @author JSR-51 Expert Group 669 * @since 1.4 670 * @spec JSR-51 671 */ 672 673 public final class Pattern 674 implements java.io.Serializable 675 { 676 677 /** 678 * Regular expression modifier values. Instead of being passed as 679 * arguments, they can also be passed as inline modifiers. 680 * For example, the following statements have the same effect. 681 * <pre> 682 * RegExp r1 = RegExp.compile("abc", Pattern.I|Pattern.M); 683 * RegExp r2 = RegExp.compile("(?im)abc", 0); 684 * </pre> 685 * 686 * The flags are duplicated so that the familiar Perl match flag 687 * names are available. 688 */ 689 690 /** 691 * Enables Unix lines mode. 692 * 693 * <p> In this mode, only the <tt>'\n'</tt> line terminator is recognized 694 * in the behavior of <tt>.</tt>, <tt>^</tt>, and <tt>$</tt>. 695 * 696 * <p> Unix lines mode can also be enabled via the embedded flag 697 * expression <tt>(?d)</tt>. 698 */ 699 public static final int UNIX_LINES = 0x01; 700 701 /** 702 * Enables case-insensitive matching. 703 * 704 * <p> By default, case-insensitive matching assumes that only characters 705 * in the US-ASCII charset are being matched. Unicode-aware 706 * case-insensitive matching can be enabled by specifying the {@link 707 * #UNICODE_CASE} flag in conjunction with this flag. 708 * 709 * <p> Case-insensitive matching can also be enabled via the embedded flag 710 * expression <tt>(?i)</tt>. 711 * 712 * <p> Specifying this flag may impose a slight performance penalty. </p> 713 */ 714 public static final int CASE_INSENSITIVE = 0x02; 715 716 /** 717 * Permits whitespace and comments in pattern. 718 * 719 * <p> In this mode, whitespace is ignored, and embedded comments starting 720 * with <tt>#</tt> are ignored until the end of a line. 721 * 722 * <p> Comments mode can also be enabled via the embedded flag 723 * expression <tt>(?x)</tt>. 724 */ 725 public static final int COMMENTS = 0x04; 726 727 /** 728 * Enables multiline mode. 729 * 730 * <p> In multiline mode the expressions <tt>^</tt> and <tt>$</tt> match 731 * just after or just before, respectively, a line terminator or the end of 732 * the input sequence. By default these expressions only match at the 733 * beginning and the end of the entire input sequence. 734 * 735 * <p> Multiline mode can also be enabled via the embedded flag 736 * expression <tt>(?m)</tt>. </p> 737 */ 738 public static final int MULTILINE = 0x08; 739 740 /** 741 * Enables literal parsing of the pattern. 742 * 743 * <p> When this flag is specified then the input string that specifies 744 * the pattern is treated as a sequence of literal characters. 745 * Metacharacters or escape sequences in the input sequence will be 746 * given no special meaning. 747 * 748 * <p>The flags CASE_INSENSITIVE and UNICODE_CASE retain their impact on 749 * matching when used in conjunction with this flag. The other flags 750 * become superfluous. 751 * 752 * <p> There is no embedded flag character for enabling literal parsing. 753 * @since 1.5 754 */ 755 public static final int LITERAL = 0x10; 756 757 /** 758 * Enables dotall mode. 759 * 760 * <p> In dotall mode, the expression <tt>.</tt> matches any character, 761 * including a line terminator. By default this expression does not match 762 * line terminators. 763 * 764 * <p> Dotall mode can also be enabled via the embedded flag 765 * expression <tt>(?s)</tt>. (The <tt>s</tt> is a mnemonic for 766 * "single-line" mode, which is what this is called in Perl.) </p> 767 */ 768 public static final int DOTALL = 0x20; 769 770 /** 771 * Enables Unicode-aware case folding. 772 * 773 * <p> When this flag is specified then case-insensitive matching, when 774 * enabled by the {@link #CASE_INSENSITIVE} flag, is done in a manner 775 * consistent with the Unicode Standard. By default, case-insensitive 776 * matching assumes that only characters in the US-ASCII charset are being 777 * matched. 778 * 779 * <p> Unicode-aware case folding can also be enabled via the embedded flag 780 * expression <tt>(?u)</tt>. 781 * 782 * <p> Specifying this flag may impose a performance penalty. </p> 783 */ 784 public static final int UNICODE_CASE = 0x40; 785 786 /** 787 * Enables canonical equivalence. 788 * 789 * <p> When this flag is specified then two characters will be considered 790 * to match if, and only if, their full canonical decompositions match. 791 * The expression <tt>"a\u030A"</tt>, for example, will match the 792 * string <tt>"\u00E5"</tt> when this flag is specified. By default, 793 * matching does not take canonical equivalence into account. 794 * 795 * <p> There is no embedded flag character for enabling canonical 796 * equivalence. 797 * 798 * <p> Specifying this flag may impose a performance penalty. </p> 799 */ 800 public static final int CANON_EQ = 0x80; 801 802 /* Pattern has only two serialized components: The pattern string 803 * and the flags, which are all that is needed to recompile the pattern 804 * when it is deserialized. 805 */ 806 807 /** use serialVersionUID from Merlin b59 for interoperability */ 808 private static final long serialVersionUID = 5073258162644648461L; 809 810 /** 811 * The original regular-expression pattern string. 812 * 813 * @serial 814 */ 815 private String pattern; 816 817 /** 818 * The original pattern flags. 819 * 820 * @serial 821 */ 822 private int flags; 823 824 /** 825 * Boolean indicating this Pattern is compiled; this is necessary in order 826 * to lazily compile deserialized Patterns. 827 */ 828 private transient volatile boolean compiled = false; 829 830 /** 831 * The normalized pattern string. 832 */ 833 private transient String normalizedPattern; 834 835 /** 836 * The starting point of state machine for the find operation. This allows 837 * a match to start anywhere in the input. 838 */ 839 transient Node root; 840 841 /** 842 * The root of object tree for a match operation. The pattern is matched 843 * at the beginning. This may include a find that uses BnM or a First 844 * node. 845 */ 846 transient Node matchRoot; 847 848 /** 849 * Temporary storage used by parsing pattern slice. 850 */ 851 transient int[] buffer; 852 853 /** 854 * Map the "name" of the "named capturing group" to its group id 855 * node. 856 */ 857 transient volatile Map<String, Integer> namedGroups; 858 859 /** 860 * Temporary storage used while parsing group references. 861 */ 862 transient GroupHead[] groupNodes; 863 864 /** 865 * Temporary null terminated code point array used by pattern compiling. 866 */ 867 private transient int[] temp; 868 869 /** 870 * The number of capturing groups in this Pattern. Used by matchers to 871 * allocate storage needed to perform a match. 872 */ 873 transient int capturingGroupCount; 874 875 /** 876 * The local variable count used by parsing tree. Used by matchers to 877 * allocate storage needed to perform a match. 878 */ 879 transient int localCount; 880 881 /** 882 * Index into the pattern string that keeps track of how much has been 883 * parsed. 884 */ 885 private transient int cursor; 886 887 /** 888 * Holds the length of the pattern string. 889 */ 890 private transient int patternLength; 891 892 /** 893 * If the Start node might possibly match supplementary characters. 894 * It is set to true during compiling if 895 * (1) There is supplementary char in pattern, or 896 * (2) There is complement node of Category or Block 897 */ 898 private transient boolean hasSupplementary; 899 900 /** 901 * Compiles the given regular expression into a pattern. </p> 902 * 903 * @param regex 904 * The expression to be compiled 905 * 906 * @throws PatternSyntaxException 907 * If the expression's syntax is invalid 908 */ 909 public static Pattern compile(String regex) { 910 return new Pattern(regex, 0); 911 } 912 913 /** 914 * Compiles the given regular expression into a pattern with the given 915 * flags. </p> 916 * 917 * @param regex 918 * The expression to be compiled 919 * 920 * @param flags 921 * Match flags, a bit mask that may include 922 * {@link #CASE_INSENSITIVE}, {@link #MULTILINE}, {@link #DOTALL}, 923 * {@link #UNICODE_CASE}, {@link #CANON_EQ}, {@link #UNIX_LINES}, 924 * {@link #LITERAL} and {@link #COMMENTS} 925 * 926 * @throws IllegalArgumentException 927 * If bit values other than those corresponding to the defined 928 * match flags are set in <tt>flags</tt> 929 * 930 * @throws PatternSyntaxException 931 * If the expression's syntax is invalid 932 */ 933 public static Pattern compile(String regex, int flags) { 934 return new Pattern(regex, flags); 935 } 936 937 /** 938 * Returns the regular expression from which this pattern was compiled. 939 * </p> 940 * 941 * @return The source of this pattern 942 */ 943 public String pattern() { 944 return pattern; 945 } 946 947 /** 948 * <p>Returns the string representation of this pattern. This 949 * is the regular expression from which this pattern was 950 * compiled.</p> 951 * 952 * @return The string representation of this pattern 953 * @since 1.5 954 */ 955 public String toString() { 956 return pattern; 957 } 958 959 /** 960 * Creates a matcher that will match the given input against this pattern. 961 * </p> 962 * 963 * @param input 964 * The character sequence to be matched 965 * 966 * @return A new matcher for this pattern 967 */ 968 public Matcher matcher(CharSequence input) { 969 if (!compiled) { 970 synchronized(this) { 971 if (!compiled) 972 compile(); 973 } 974 } 975 Matcher m = new Matcher(this, input); 976 return m; 977 } 978 979 /** 980 * Returns this pattern's match flags. </p> 981 * 982 * @return The match flags specified when this pattern was compiled 983 */ 984 public int flags() { 985 return flags; 986 } 987 988 /** 989 * Compiles the given regular expression and attempts to match the given 990 * input against it. 991 * 992 * <p> An invocation of this convenience method of the form 993 * 994 * <blockquote><pre> 995 * Pattern.matches(regex, input);</pre></blockquote> 996 * 997 * behaves in exactly the same way as the expression 998 * 999 * <blockquote><pre> 1000 * Pattern.compile(regex).matcher(input).matches()</pre></blockquote> 1001 * 1002 * <p> If a pattern is to be used multiple times, compiling it once and reusing 1003 * it will be more efficient than invoking this method each time. </p> 1004 * 1005 * @param regex 1006 * The expression to be compiled 1007 * 1008 * @param input 1009 * The character sequence to be matched 1010 * 1011 * @throws PatternSyntaxException 1012 * If the expression's syntax is invalid 1013 */ 1014 public static boolean matches(String regex, CharSequence input) { 1015 Pattern p = Pattern.compile(regex); 1016 Matcher m = p.matcher(input); 1017 return m.matches(); 1018 } 1019 1020 /** 1021 * Splits the given input sequence around matches of this pattern. 1022 * 1023 * <p> The array returned by this method contains each substring of the 1024 * input sequence that is terminated by another subsequence that matches 1025 * this pattern or is terminated by the end of the input sequence. The 1026 * substrings in the array are in the order in which they occur in the 1027 * input. If this pattern does not match any subsequence of the input then 1028 * the resulting array has just one element, namely the input sequence in 1029 * string form. 1030 * 1031 * <p> The <tt>limit</tt> parameter controls the number of times the 1032 * pattern is applied and therefore affects the length of the resulting 1033 * array. If the limit <i>n</i> is greater than zero then the pattern 1034 * will be applied at most <i>n</i> - 1 times, the array's 1035 * length will be no greater than <i>n</i>, and the array's last entry 1036 * will contain all input beyond the last matched delimiter. If <i>n</i> 1037 * is non-positive then the pattern will be applied as many times as 1038 * possible and the array can have any length. If <i>n</i> is zero then 1039 * the pattern will be applied as many times as possible, the array can 1040 * have any length, and trailing empty strings will be discarded. 1041 * 1042 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1043 * results with these parameters: 1044 * 1045 * <blockquote><table cellpadding=1 cellspacing=0 1046 * summary="Split examples showing regex, limit, and result"> 1047 * <tr><th><P align="left"><i>Regex </i></th> 1048 * <th><P align="left"><i>Limit </i></th> 1049 * <th><P align="left"><i>Result </i></th></tr> 1050 * <tr><td align=center>:</td> 1051 * <td align=center>2</td> 1052 * <td><tt>{ "boo", "and:foo" }</tt></td></tr> 1053 * <tr><td align=center>:</td> 1054 * <td align=center>5</td> 1055 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1056 * <tr><td align=center>:</td> 1057 * <td align=center>-2</td> 1058 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1059 * <tr><td align=center>o</td> 1060 * <td align=center>5</td> 1061 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1062 * <tr><td align=center>o</td> 1063 * <td align=center>-2</td> 1064 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1065 * <tr><td align=center>o</td> 1066 * <td align=center>0</td> 1067 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1068 * </table></blockquote> 1069 * 1070 * 1071 * @param input 1072 * The character sequence to be split 1073 * 1074 * @param limit 1075 * The result threshold, as described above 1076 * 1077 * @return The array of strings computed by splitting the input 1078 * around matches of this pattern 1079 */ 1080 public String[] split(CharSequence input, int limit) { 1081 int index = 0; 1082 boolean matchLimited = limit > 0; 1083 ArrayList<String> matchList = new ArrayList<>(); 1084 Matcher m = matcher(input); 1085 1086 // Add segments before each match found 1087 while(m.find()) { 1088 if (!matchLimited || matchList.size() < limit - 1) { 1089 String match = input.subSequence(index, m.start()).toString(); 1090 matchList.add(match); 1091 index = m.end(); 1092 } else if (matchList.size() == limit - 1) { // last one 1093 String match = input.subSequence(index, 1094 input.length()).toString(); 1095 matchList.add(match); 1096 index = m.end(); 1097 } 1098 } 1099 1100 // If no match was found, return this 1101 if (index == 0) 1102 return new String[] {input.toString()}; 1103 1104 // Add remaining segment 1105 if (!matchLimited || matchList.size() < limit) 1106 matchList.add(input.subSequence(index, input.length()).toString()); 1107 1108 // Construct result 1109 int resultSize = matchList.size(); 1110 if (limit == 0) 1111 while (resultSize > 0 && matchList.get(resultSize-1).equals("")) 1112 resultSize--; 1113 String[] result = new String[resultSize]; 1114 return matchList.subList(0, resultSize).toArray(result); 1115 } 1116 1117 /** 1118 * Splits the given input sequence around matches of this pattern. 1119 * 1120 * <p> This method works as if by invoking the two-argument {@link 1121 * #split(java.lang.CharSequence, int) split} method with the given input 1122 * sequence and a limit argument of zero. Trailing empty strings are 1123 * therefore not included in the resulting array. </p> 1124 * 1125 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1126 * results with these expressions: 1127 * 1128 * <blockquote><table cellpadding=1 cellspacing=0 1129 * summary="Split examples showing regex and result"> 1130 * <tr><th><P align="left"><i>Regex </i></th> 1131 * <th><P align="left"><i>Result</i></th></tr> 1132 * <tr><td align=center>:</td> 1133 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1134 * <tr><td align=center>o</td> 1135 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1136 * </table></blockquote> 1137 * 1138 * 1139 * @param input 1140 * The character sequence to be split 1141 * 1142 * @return The array of strings computed by splitting the input 1143 * around matches of this pattern 1144 */ 1145 public String[] split(CharSequence input) { 1146 return split(input, 0); 1147 } 1148 1149 /** 1150 * Returns a literal pattern <code>String</code> for the specified 1151 * <code>String</code>. 1152 * 1153 * <p>This method produces a <code>String</code> that can be used to 1154 * create a <code>Pattern</code> that would match the string 1155 * <code>s</code> as if it were a literal pattern.</p> Metacharacters 1156 * or escape sequences in the input sequence will be given no special 1157 * meaning. 1158 * 1159 * @param s The string to be literalized 1160 * @return A literal string replacement 1161 * @since 1.5 1162 */ 1163 public static String quote(String s) { 1164 int slashEIndex = s.indexOf("\\E"); 1165 if (slashEIndex == -1) 1166 return "\\Q" + s + "\\E"; 1167 1168 StringBuilder sb = new StringBuilder(s.length() * 2); 1169 sb.append("\\Q"); 1170 slashEIndex = 0; 1171 int current = 0; 1172 while ((slashEIndex = s.indexOf("\\E", current)) != -1) { 1173 sb.append(s.substring(current, slashEIndex)); 1174 current = slashEIndex + 2; 1175 sb.append("\\E\\\\E\\Q"); 1176 } 1177 sb.append(s.substring(current, s.length())); 1178 sb.append("\\E"); 1179 return sb.toString(); 1180 } 1181 1182 /** 1183 * Recompile the Pattern instance from a stream. The original pattern 1184 * string is read in and the object tree is recompiled from it. 1185 */ 1186 private void readObject(java.io.ObjectInputStream s) 1187 throws java.io.IOException, ClassNotFoundException { 1188 1189 // Read in all fields 1190 s.defaultReadObject(); 1191 1192 // Initialize counts 1193 capturingGroupCount = 1; 1194 localCount = 0; 1195 1196 // if length > 0, the Pattern is lazily compiled 1197 compiled = false; 1198 if (pattern.length() == 0) { 1199 root = new Start(lastAccept); 1200 matchRoot = lastAccept; 1201 compiled = true; 1202 } 1203 } 1204 1205 /** 1206 * This private constructor is used to create all Patterns. The pattern 1207 * string and match flags are all that is needed to completely describe 1208 * a Pattern. An empty pattern string results in an object tree with 1209 * only a Start node and a LastNode node. 1210 */ 1211 private Pattern(String p, int f) { 1212 pattern = p; 1213 flags = f; 1214 1215 // Reset group index count 1216 capturingGroupCount = 1; 1217 localCount = 0; 1218 1219 if (pattern.length() > 0) { 1220 compile(); 1221 } else { 1222 root = new Start(lastAccept); 1223 matchRoot = lastAccept; 1224 } 1225 } 1226 1227 /** 1228 * The pattern is converted to normalizedD form and then a pure group 1229 * is constructed to match canonical equivalences of the characters. 1230 */ 1231 private void normalize() { 1232 boolean inCharClass = false; 1233 int lastCodePoint = -1; 1234 1235 // Convert pattern into normalizedD form 1236 normalizedPattern = Normalizer.normalize(pattern, Normalizer.Form.NFD); 1237 patternLength = normalizedPattern.length(); 1238 1239 // Modify pattern to match canonical equivalences 1240 StringBuilder newPattern = new StringBuilder(patternLength); 1241 for(int i=0; i<patternLength; ) { 1242 int c = normalizedPattern.codePointAt(i); 1243 StringBuilder sequenceBuffer; 1244 if ((Character.getType(c) == Character.NON_SPACING_MARK) 1245 && (lastCodePoint != -1)) { 1246 sequenceBuffer = new StringBuilder(); 1247 sequenceBuffer.appendCodePoint(lastCodePoint); 1248 sequenceBuffer.appendCodePoint(c); 1249 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1250 i += Character.charCount(c); 1251 if (i >= patternLength) 1252 break; 1253 c = normalizedPattern.codePointAt(i); 1254 sequenceBuffer.appendCodePoint(c); 1255 } 1256 String ea = produceEquivalentAlternation( 1257 sequenceBuffer.toString()); 1258 newPattern.setLength(newPattern.length()-Character.charCount(lastCodePoint)); 1259 newPattern.append("(?:").append(ea).append(")"); 1260 } else if (c == '[' && lastCodePoint != '\\') { 1261 i = normalizeCharClass(newPattern, i); 1262 } else { 1263 newPattern.appendCodePoint(c); 1264 } 1265 lastCodePoint = c; 1266 i += Character.charCount(c); 1267 } 1268 normalizedPattern = newPattern.toString(); 1269 } 1270 1271 /** 1272 * Complete the character class being parsed and add a set 1273 * of alternations to it that will match the canonical equivalences 1274 * of the characters within the class. 1275 */ 1276 private int normalizeCharClass(StringBuilder newPattern, int i) { 1277 StringBuilder charClass = new StringBuilder(); 1278 StringBuilder eq = null; 1279 int lastCodePoint = -1; 1280 String result; 1281 1282 i++; 1283 charClass.append("["); 1284 while(true) { 1285 int c = normalizedPattern.codePointAt(i); 1286 StringBuilder sequenceBuffer; 1287 1288 if (c == ']' && lastCodePoint != '\\') { 1289 charClass.append((char)c); 1290 break; 1291 } else if (Character.getType(c) == Character.NON_SPACING_MARK) { 1292 sequenceBuffer = new StringBuilder(); 1293 sequenceBuffer.appendCodePoint(lastCodePoint); 1294 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1295 sequenceBuffer.appendCodePoint(c); 1296 i += Character.charCount(c); 1297 if (i >= normalizedPattern.length()) 1298 break; 1299 c = normalizedPattern.codePointAt(i); 1300 } 1301 String ea = produceEquivalentAlternation( 1302 sequenceBuffer.toString()); 1303 1304 charClass.setLength(charClass.length()-Character.charCount(lastCodePoint)); 1305 if (eq == null) 1306 eq = new StringBuilder(); 1307 eq.append('|'); 1308 eq.append(ea); 1309 } else { 1310 charClass.appendCodePoint(c); 1311 i++; 1312 } 1313 if (i == normalizedPattern.length()) 1314 throw error("Unclosed character class"); 1315 lastCodePoint = c; 1316 } 1317 1318 if (eq != null) { 1319 result = "(?:"+charClass.toString()+eq.toString()+")"; 1320 } else { 1321 result = charClass.toString(); 1322 } 1323 1324 newPattern.append(result); 1325 return i; 1326 } 1327 1328 /** 1329 * Given a specific sequence composed of a regular character and 1330 * combining marks that follow it, produce the alternation that will 1331 * match all canonical equivalences of that sequence. 1332 */ 1333 private String produceEquivalentAlternation(String source) { 1334 int len = countChars(source, 0, 1); 1335 if (source.length() == len) 1336 // source has one character. 1337 return source; 1338 1339 String base = source.substring(0,len); 1340 String combiningMarks = source.substring(len); 1341 1342 String[] perms = producePermutations(combiningMarks); 1343 StringBuilder result = new StringBuilder(source); 1344 1345 // Add combined permutations 1346 for(int x=0; x<perms.length; x++) { 1347 String next = base + perms[x]; 1348 if (x>0) 1349 result.append("|"+next); 1350 next = composeOneStep(next); 1351 if (next != null) 1352 result.append("|"+produceEquivalentAlternation(next)); 1353 } 1354 return result.toString(); 1355 } 1356 1357 /** 1358 * Returns an array of strings that have all the possible 1359 * permutations of the characters in the input string. 1360 * This is used to get a list of all possible orderings 1361 * of a set of combining marks. Note that some of the permutations 1362 * are invalid because of combining class collisions, and these 1363 * possibilities must be removed because they are not canonically 1364 * equivalent. 1365 */ 1366 private String[] producePermutations(String input) { 1367 if (input.length() == countChars(input, 0, 1)) 1368 return new String[] {input}; 1369 1370 if (input.length() == countChars(input, 0, 2)) { 1371 int c0 = Character.codePointAt(input, 0); 1372 int c1 = Character.codePointAt(input, Character.charCount(c0)); 1373 if (getClass(c1) == getClass(c0)) { 1374 return new String[] {input}; 1375 } 1376 String[] result = new String[2]; 1377 result[0] = input; 1378 StringBuilder sb = new StringBuilder(2); 1379 sb.appendCodePoint(c1); 1380 sb.appendCodePoint(c0); 1381 result[1] = sb.toString(); 1382 return result; 1383 } 1384 1385 int length = 1; 1386 int nCodePoints = countCodePoints(input); 1387 for(int x=1; x<nCodePoints; x++) 1388 length = length * (x+1); 1389 1390 String[] temp = new String[length]; 1391 1392 int combClass[] = new int[nCodePoints]; 1393 for(int x=0, i=0; x<nCodePoints; x++) { 1394 int c = Character.codePointAt(input, i); 1395 combClass[x] = getClass(c); 1396 i += Character.charCount(c); 1397 } 1398 1399 // For each char, take it out and add the permutations 1400 // of the remaining chars 1401 int index = 0; 1402 int len; 1403 // offset maintains the index in code units. 1404 loop: for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) { 1405 len = countChars(input, offset, 1); 1406 boolean skip = false; 1407 for(int y=x-1; y>=0; y--) { 1408 if (combClass[y] == combClass[x]) { 1409 continue loop; 1410 } 1411 } 1412 StringBuilder sb = new StringBuilder(input); 1413 String otherChars = sb.delete(offset, offset+len).toString(); 1414 String[] subResult = producePermutations(otherChars); 1415 1416 String prefix = input.substring(offset, offset+len); 1417 for(int y=0; y<subResult.length; y++) 1418 temp[index++] = prefix + subResult[y]; 1419 } 1420 String[] result = new String[index]; 1421 for (int x=0; x<index; x++) 1422 result[x] = temp[x]; 1423 return result; 1424 } 1425 1426 private int getClass(int c) { 1427 return sun.text.Normalizer.getCombiningClass(c); 1428 } 1429 1430 /** 1431 * Attempts to compose input by combining the first character 1432 * with the first combining mark following it. Returns a String 1433 * that is the composition of the leading character with its first 1434 * combining mark followed by the remaining combining marks. Returns 1435 * null if the first two characters cannot be further composed. 1436 */ 1437 private String composeOneStep(String input) { 1438 int len = countChars(input, 0, 2); 1439 String firstTwoCharacters = input.substring(0, len); 1440 String result = Normalizer.normalize(firstTwoCharacters, Normalizer.Form.NFC); 1441 1442 if (result.equals(firstTwoCharacters)) 1443 return null; 1444 else { 1445 String remainder = input.substring(len); 1446 return result + remainder; 1447 } 1448 } 1449 1450 /** 1451 * Preprocess any \Q...\E sequences in `temp', meta-quoting them. 1452 * See the description of `quotemeta' in perlfunc(1). 1453 */ 1454 private void RemoveQEQuoting() { 1455 final int pLen = patternLength; 1456 int i = 0; 1457 while (i < pLen-1) { 1458 if (temp[i] != '\\') 1459 i += 1; 1460 else if (temp[i + 1] != 'Q') 1461 i += 2; 1462 else 1463 break; 1464 } 1465 if (i >= pLen - 1) // No \Q sequence found 1466 return; 1467 int j = i; 1468 i += 2; 1469 int[] newtemp = new int[j + 2*(pLen-i) + 2]; 1470 System.arraycopy(temp, 0, newtemp, 0, j); 1471 1472 boolean inQuote = true; 1473 while (i < pLen) { 1474 int c = temp[i++]; 1475 if (! ASCII.isAscii(c) || ASCII.isAlnum(c)) { 1476 newtemp[j++] = c; 1477 } else if (c != '\\') { 1478 if (inQuote) newtemp[j++] = '\\'; 1479 newtemp[j++] = c; 1480 } else if (inQuote) { 1481 if (temp[i] == 'E') { 1482 i++; 1483 inQuote = false; 1484 } else { 1485 newtemp[j++] = '\\'; 1486 newtemp[j++] = '\\'; 1487 } 1488 } else { 1489 if (temp[i] == 'Q') { 1490 i++; 1491 inQuote = true; 1492 } else { 1493 newtemp[j++] = c; 1494 if (i != pLen) 1495 newtemp[j++] = temp[i++]; 1496 } 1497 } 1498 } 1499 1500 patternLength = j; 1501 temp = Arrays.copyOf(newtemp, j + 2); // double zero termination 1502 } 1503 1504 /** 1505 * Copies regular expression to an int array and invokes the parsing 1506 * of the expression which will create the object tree. 1507 */ 1508 private void compile() { 1509 // Handle canonical equivalences 1510 if (has(CANON_EQ) && !has(LITERAL)) { 1511 normalize(); 1512 } else { 1513 normalizedPattern = pattern; 1514 } 1515 patternLength = normalizedPattern.length(); 1516 1517 // Copy pattern to int array for convenience 1518 // Use double zero to terminate pattern 1519 temp = new int[patternLength + 2]; 1520 1521 hasSupplementary = false; 1522 int c, count = 0; 1523 // Convert all chars into code points 1524 for (int x = 0; x < patternLength; x += Character.charCount(c)) { 1525 c = normalizedPattern.codePointAt(x); 1526 if (isSupplementary(c)) { 1527 hasSupplementary = true; 1528 } 1529 temp[count++] = c; 1530 } 1531 1532 patternLength = count; // patternLength now in code points 1533 1534 if (! has(LITERAL)) 1535 RemoveQEQuoting(); 1536 1537 // Allocate all temporary objects here. 1538 buffer = new int[32]; 1539 groupNodes = new GroupHead[10]; 1540 namedGroups = null; 1541 1542 if (has(LITERAL)) { 1543 // Literal pattern handling 1544 matchRoot = newSlice(temp, patternLength, hasSupplementary); 1545 matchRoot.next = lastAccept; 1546 } else { 1547 // Start recursive descent parsing 1548 matchRoot = expr(lastAccept); 1549 // Check extra pattern characters 1550 if (patternLength != cursor) { 1551 if (peek() == ')') { 1552 throw error("Unmatched closing ')'"); 1553 } else { 1554 throw error("Unexpected internal error"); 1555 } 1556 } 1557 } 1558 1559 // Peephole optimization 1560 if (matchRoot instanceof Slice) { 1561 root = BnM.optimize(matchRoot); 1562 if (root == matchRoot) { 1563 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1564 } 1565 } else if (matchRoot instanceof Begin || matchRoot instanceof First) { 1566 root = matchRoot; 1567 } else { 1568 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1569 } 1570 1571 // Release temporary storage 1572 temp = null; 1573 buffer = null; 1574 groupNodes = null; 1575 patternLength = 0; 1576 compiled = true; 1577 } 1578 1579 Map<String, Integer> namedGroups() { 1580 if (namedGroups == null) 1581 namedGroups = new HashMap<>(2); 1582 return namedGroups; 1583 } 1584 1585 /** 1586 * Used to print out a subtree of the Pattern to help with debugging. 1587 */ 1588 private static void printObjectTree(Node node) { 1589 while(node != null) { 1590 if (node instanceof Prolog) { 1591 System.out.println(node); 1592 printObjectTree(((Prolog)node).loop); 1593 System.out.println("**** end contents prolog loop"); 1594 } else if (node instanceof Loop) { 1595 System.out.println(node); 1596 printObjectTree(((Loop)node).body); 1597 System.out.println("**** end contents Loop body"); 1598 } else if (node instanceof Curly) { 1599 System.out.println(node); 1600 printObjectTree(((Curly)node).atom); 1601 System.out.println("**** end contents Curly body"); 1602 } else if (node instanceof GroupCurly) { 1603 System.out.println(node); 1604 printObjectTree(((GroupCurly)node).atom); 1605 System.out.println("**** end contents GroupCurly body"); 1606 } else if (node instanceof GroupTail) { 1607 System.out.println(node); 1608 System.out.println("Tail next is "+node.next); 1609 return; 1610 } else { 1611 System.out.println(node); 1612 } 1613 node = node.next; 1614 if (node != null) 1615 System.out.println("->next:"); 1616 if (node == Pattern.accept) { 1617 System.out.println("Accept Node"); 1618 node = null; 1619 } 1620 } 1621 } 1622 1623 /** 1624 * Used to accumulate information about a subtree of the object graph 1625 * so that optimizations can be applied to the subtree. 1626 */ 1627 static final class TreeInfo { 1628 int minLength; 1629 int maxLength; 1630 boolean maxValid; 1631 boolean deterministic; 1632 1633 TreeInfo() { 1634 reset(); 1635 } 1636 void reset() { 1637 minLength = 0; 1638 maxLength = 0; 1639 maxValid = true; 1640 deterministic = true; 1641 } 1642 } 1643 1644 /* 1645 * The following private methods are mainly used to improve the 1646 * readability of the code. In order to let the Java compiler easily 1647 * inline them, we should not put many assertions or error checks in them. 1648 */ 1649 1650 /** 1651 * Indicates whether a particular flag is set or not. 1652 */ 1653 private boolean has(int f) { 1654 return (flags & f) != 0; 1655 } 1656 1657 /** 1658 * Match next character, signal error if failed. 1659 */ 1660 private void accept(int ch, String s) { 1661 int testChar = temp[cursor++]; 1662 if (has(COMMENTS)) 1663 testChar = parsePastWhitespace(testChar); 1664 if (ch != testChar) { 1665 throw error(s); 1666 } 1667 } 1668 1669 /** 1670 * Mark the end of pattern with a specific character. 1671 */ 1672 private void mark(int c) { 1673 temp[patternLength] = c; 1674 } 1675 1676 /** 1677 * Peek the next character, and do not advance the cursor. 1678 */ 1679 private int peek() { 1680 int ch = temp[cursor]; 1681 if (has(COMMENTS)) 1682 ch = peekPastWhitespace(ch); 1683 return ch; 1684 } 1685 1686 /** 1687 * Read the next character, and advance the cursor by one. 1688 */ 1689 private int read() { 1690 int ch = temp[cursor++]; 1691 if (has(COMMENTS)) 1692 ch = parsePastWhitespace(ch); 1693 return ch; 1694 } 1695 1696 /** 1697 * Read the next character, and advance the cursor by one, 1698 * ignoring the COMMENTS setting 1699 */ 1700 private int readEscaped() { 1701 int ch = temp[cursor++]; 1702 return ch; 1703 } 1704 1705 /** 1706 * Advance the cursor by one, and peek the next character. 1707 */ 1708 private int next() { 1709 int ch = temp[++cursor]; 1710 if (has(COMMENTS)) 1711 ch = peekPastWhitespace(ch); 1712 return ch; 1713 } 1714 1715 /** 1716 * Advance the cursor by one, and peek the next character, 1717 * ignoring the COMMENTS setting 1718 */ 1719 private int nextEscaped() { 1720 int ch = temp[++cursor]; 1721 return ch; 1722 } 1723 1724 /** 1725 * If in xmode peek past whitespace and comments. 1726 */ 1727 private int peekPastWhitespace(int ch) { 1728 while (ASCII.isSpace(ch) || ch == '#') { 1729 while (ASCII.isSpace(ch)) 1730 ch = temp[++cursor]; 1731 if (ch == '#') { 1732 ch = peekPastLine(); 1733 } 1734 } 1735 return ch; 1736 } 1737 1738 /** 1739 * If in xmode parse past whitespace and comments. 1740 */ 1741 private int parsePastWhitespace(int ch) { 1742 while (ASCII.isSpace(ch) || ch == '#') { 1743 while (ASCII.isSpace(ch)) 1744 ch = temp[cursor++]; 1745 if (ch == '#') 1746 ch = parsePastLine(); 1747 } 1748 return ch; 1749 } 1750 1751 /** 1752 * xmode parse past comment to end of line. 1753 */ 1754 private int parsePastLine() { 1755 int ch = temp[cursor++]; 1756 while (ch != 0 && !isLineSeparator(ch)) 1757 ch = temp[cursor++]; 1758 return ch; 1759 } 1760 1761 /** 1762 * xmode peek past comment to end of line. 1763 */ 1764 private int peekPastLine() { 1765 int ch = temp[++cursor]; 1766 while (ch != 0 && !isLineSeparator(ch)) 1767 ch = temp[++cursor]; 1768 return ch; 1769 } 1770 1771 /** 1772 * Determines if character is a line separator in the current mode 1773 */ 1774 private boolean isLineSeparator(int ch) { 1775 if (has(UNIX_LINES)) { 1776 return ch == '\n'; 1777 } else { 1778 return (ch == '\n' || 1779 ch == '\r' || 1780 (ch|1) == '\u2029' || 1781 ch == '\u0085'); 1782 } 1783 } 1784 1785 /** 1786 * Read the character after the next one, and advance the cursor by two. 1787 */ 1788 private int skip() { 1789 int i = cursor; 1790 int ch = temp[i+1]; 1791 cursor = i + 2; 1792 return ch; 1793 } 1794 1795 /** 1796 * Unread one next character, and retreat cursor by one. 1797 */ 1798 private void unread() { 1799 cursor--; 1800 } 1801 1802 /** 1803 * Internal method used for handling all syntax errors. The pattern is 1804 * displayed with a pointer to aid in locating the syntax error. 1805 */ 1806 private PatternSyntaxException error(String s) { 1807 return new PatternSyntaxException(s, normalizedPattern, cursor - 1); 1808 } 1809 1810 /** 1811 * Determines if there is any supplementary character or unpaired 1812 * surrogate in the specified range. 1813 */ 1814 private boolean findSupplementary(int start, int end) { 1815 for (int i = start; i < end; i++) { 1816 if (isSupplementary(temp[i])) 1817 return true; 1818 } 1819 return false; 1820 } 1821 1822 /** 1823 * Determines if the specified code point is a supplementary 1824 * character or unpaired surrogate. 1825 */ 1826 private static final boolean isSupplementary(int ch) { 1827 return ch >= Character.MIN_SUPPLEMENTARY_CODE_POINT || 1828 Character.isSurrogate((char)ch); 1829 } 1830 1831 /** 1832 * The following methods handle the main parsing. They are sorted 1833 * according to their precedence order, the lowest one first. 1834 */ 1835 1836 /** 1837 * The expression is parsed with branch nodes added for alternations. 1838 * This may be called recursively to parse sub expressions that may 1839 * contain alternations. 1840 */ 1841 private Node expr(Node end) { 1842 Node prev = null; 1843 Node firstTail = null; 1844 Node branchConn = null; 1845 1846 for (;;) { 1847 Node node = sequence(end); 1848 Node nodeTail = root; //double return 1849 if (prev == null) { 1850 prev = node; 1851 firstTail = nodeTail; 1852 } else { 1853 // Branch 1854 if (branchConn == null) { 1855 branchConn = new BranchConn(); 1856 branchConn.next = end; 1857 } 1858 if (node == end) { 1859 // if the node returned from sequence() is "end" 1860 // we have an empty expr, set a null atom into 1861 // the branch to indicate to go "next" directly. 1862 node = null; 1863 } else { 1864 // the "tail.next" of each atom goes to branchConn 1865 nodeTail.next = branchConn; 1866 } 1867 if (prev instanceof Branch) { 1868 ((Branch)prev).add(node); 1869 } else { 1870 if (prev == end) { 1871 prev = null; 1872 } else { 1873 // replace the "end" with "branchConn" at its tail.next 1874 // when put the "prev" into the branch as the first atom. 1875 firstTail.next = branchConn; 1876 } 1877 prev = new Branch(prev, node, branchConn); 1878 } 1879 } 1880 if (peek() != '|') { 1881 return prev; 1882 } 1883 next(); 1884 } 1885 } 1886 1887 /** 1888 * Parsing of sequences between alternations. 1889 */ 1890 private Node sequence(Node end) { 1891 Node head = null; 1892 Node tail = null; 1893 Node node = null; 1894 LOOP: 1895 for (;;) { 1896 int ch = peek(); 1897 switch (ch) { 1898 case '(': 1899 // Because group handles its own closure, 1900 // we need to treat it differently 1901 node = group0(); 1902 // Check for comment or flag group 1903 if (node == null) 1904 continue; 1905 if (head == null) 1906 head = node; 1907 else 1908 tail.next = node; 1909 // Double return: Tail was returned in root 1910 tail = root; 1911 continue; 1912 case '[': 1913 node = clazz(true); 1914 break; 1915 case '\\': 1916 ch = nextEscaped(); 1917 if (ch == 'p' || ch == 'P') { 1918 boolean oneLetter = true; 1919 boolean comp = (ch == 'P'); 1920 ch = next(); // Consume { if present 1921 if (ch != '{') { 1922 unread(); 1923 } else { 1924 oneLetter = false; 1925 } 1926 node = family(oneLetter, comp); 1927 } else { 1928 unread(); 1929 node = atom(); 1930 } 1931 break; 1932 case '^': 1933 next(); 1934 if (has(MULTILINE)) { 1935 if (has(UNIX_LINES)) 1936 node = new UnixCaret(); 1937 else 1938 node = new Caret(); 1939 } else { 1940 node = new Begin(); 1941 } 1942 break; 1943 case '$': 1944 next(); 1945 if (has(UNIX_LINES)) 1946 node = new UnixDollar(has(MULTILINE)); 1947 else 1948 node = new Dollar(has(MULTILINE)); 1949 break; 1950 case '.': 1951 next(); 1952 if (has(DOTALL)) { 1953 node = new All(); 1954 } else { 1955 if (has(UNIX_LINES)) 1956 node = new UnixDot(); 1957 else { 1958 node = new Dot(); 1959 } 1960 } 1961 break; 1962 case '|': 1963 case ')': 1964 break LOOP; 1965 case ']': // Now interpreting dangling ] and } as literals 1966 case '}': 1967 node = atom(); 1968 break; 1969 case '?': 1970 case '*': 1971 case '+': 1972 next(); 1973 throw error("Dangling meta character '" + ((char)ch) + "'"); 1974 case 0: 1975 if (cursor >= patternLength) { 1976 break LOOP; 1977 } 1978 // Fall through 1979 default: 1980 node = atom(); 1981 break; 1982 } 1983 1984 node = closure(node); 1985 1986 if (head == null) { 1987 head = tail = node; 1988 } else { 1989 tail.next = node; 1990 tail = node; 1991 } 1992 } 1993 if (head == null) { 1994 return end; 1995 } 1996 tail.next = end; 1997 root = tail; //double return 1998 return head; 1999 } 2000 2001 /** 2002 * Parse and add a new Single or Slice. 2003 */ 2004 private Node atom() { 2005 int first = 0; 2006 int prev = -1; 2007 boolean hasSupplementary = false; 2008 int ch = peek(); 2009 for (;;) { 2010 switch (ch) { 2011 case '*': 2012 case '+': 2013 case '?': 2014 case '{': 2015 if (first > 1) { 2016 cursor = prev; // Unwind one character 2017 first--; 2018 } 2019 break; 2020 case '$': 2021 case '.': 2022 case '^': 2023 case '(': 2024 case '[': 2025 case '|': 2026 case ')': 2027 break; 2028 case '\\': 2029 ch = nextEscaped(); 2030 if (ch == 'p' || ch == 'P') { // Property 2031 if (first > 0) { // Slice is waiting; handle it first 2032 unread(); 2033 break; 2034 } else { // No slice; just return the family node 2035 boolean comp = (ch == 'P'); 2036 boolean oneLetter = true; 2037 ch = next(); // Consume { if present 2038 if (ch != '{') 2039 unread(); 2040 else 2041 oneLetter = false; 2042 return family(oneLetter, comp); 2043 } 2044 } 2045 unread(); 2046 prev = cursor; 2047 ch = escape(false, first == 0); 2048 if (ch >= 0) { 2049 append(ch, first); 2050 first++; 2051 if (isSupplementary(ch)) { 2052 hasSupplementary = true; 2053 } 2054 ch = peek(); 2055 continue; 2056 } else if (first == 0) { 2057 return root; 2058 } 2059 // Unwind meta escape sequence 2060 cursor = prev; 2061 break; 2062 case 0: 2063 if (cursor >= patternLength) { 2064 break; 2065 } 2066 // Fall through 2067 default: 2068 prev = cursor; 2069 append(ch, first); 2070 first++; 2071 if (isSupplementary(ch)) { 2072 hasSupplementary = true; 2073 } 2074 ch = next(); 2075 continue; 2076 } 2077 break; 2078 } 2079 if (first == 1) { 2080 return newSingle(buffer[0]); 2081 } else { 2082 return newSlice(buffer, first, hasSupplementary); 2083 } 2084 } 2085 2086 private void append(int ch, int len) { 2087 if (len >= buffer.length) { 2088 int[] tmp = new int[len+len]; 2089 System.arraycopy(buffer, 0, tmp, 0, len); 2090 buffer = tmp; 2091 } 2092 buffer[len] = ch; 2093 } 2094 2095 /** 2096 * Parses a backref greedily, taking as many numbers as it 2097 * can. The first digit is always treated as a backref, but 2098 * multi digit numbers are only treated as a backref if at 2099 * least that many backrefs exist at this point in the regex. 2100 */ 2101 private Node ref(int refNum) { 2102 boolean done = false; 2103 while(!done) { 2104 int ch = peek(); 2105 switch(ch) { 2106 case '0': 2107 case '1': 2108 case '2': 2109 case '3': 2110 case '4': 2111 case '5': 2112 case '6': 2113 case '7': 2114 case '8': 2115 case '9': 2116 int newRefNum = (refNum * 10) + (ch - '0'); 2117 // Add another number if it doesn't make a group 2118 // that doesn't exist 2119 if (capturingGroupCount - 1 < newRefNum) { 2120 done = true; 2121 break; 2122 } 2123 refNum = newRefNum; 2124 read(); 2125 break; 2126 default: 2127 done = true; 2128 break; 2129 } 2130 } 2131 if (has(CASE_INSENSITIVE)) 2132 return new CIBackRef(refNum, has(UNICODE_CASE)); 2133 else 2134 return new BackRef(refNum); 2135 } 2136 2137 /** 2138 * Parses an escape sequence to determine the actual value that needs 2139 * to be matched. 2140 * If -1 is returned and create was true a new object was added to the tree 2141 * to handle the escape sequence. 2142 * If the returned value is greater than zero, it is the value that 2143 * matches the escape sequence. 2144 */ 2145 private int escape(boolean inclass, boolean create) { 2146 int ch = skip(); 2147 switch (ch) { 2148 case '0': 2149 return o(); 2150 case '1': 2151 case '2': 2152 case '3': 2153 case '4': 2154 case '5': 2155 case '6': 2156 case '7': 2157 case '8': 2158 case '9': 2159 if (inclass) break; 2160 if (create) { 2161 root = ref((ch - '0')); 2162 } 2163 return -1; 2164 case 'A': 2165 if (inclass) break; 2166 if (create) root = new Begin(); 2167 return -1; 2168 case 'B': 2169 if (inclass) break; 2170 if (create) root = new Bound(Bound.NONE); 2171 return -1; 2172 case 'C': 2173 break; 2174 case 'D': 2175 if (create) root = new Ctype(ASCII.DIGIT).complement(); 2176 return -1; 2177 case 'E': 2178 case 'F': 2179 break; 2180 case 'G': 2181 if (inclass) break; 2182 if (create) root = new LastMatch(); 2183 return -1; 2184 case 'H': 2185 case 'I': 2186 case 'J': 2187 case 'K': 2188 case 'L': 2189 case 'M': 2190 case 'N': 2191 case 'O': 2192 case 'P': 2193 case 'Q': 2194 case 'R': 2195 break; 2196 case 'S': 2197 if (create) root = new Ctype(ASCII.SPACE).complement(); 2198 return -1; 2199 case 'T': 2200 case 'U': 2201 case 'V': 2202 break; 2203 case 'W': 2204 if (create) root = new Ctype(ASCII.WORD).complement(); 2205 return -1; 2206 case 'X': 2207 case 'Y': 2208 break; 2209 case 'Z': 2210 if (inclass) break; 2211 if (create) { 2212 if (has(UNIX_LINES)) 2213 root = new UnixDollar(false); 2214 else 2215 root = new Dollar(false); 2216 } 2217 return -1; 2218 case 'a': 2219 return '\007'; 2220 case 'b': 2221 if (inclass) break; 2222 if (create) root = new Bound(Bound.BOTH); 2223 return -1; 2224 case 'c': 2225 return c(); 2226 case 'd': 2227 if (create) root = new Ctype(ASCII.DIGIT); 2228 return -1; 2229 case 'e': 2230 return '\033'; 2231 case 'f': 2232 return '\f'; 2233 case 'g': 2234 case 'h': 2235 case 'i': 2236 case 'j': 2237 break; 2238 case 'k': 2239 if (inclass) 2240 break; 2241 if (read() != '<') 2242 throw error("\\k is not followed by '<' for named capturing group"); 2243 String name = groupname(read()); 2244 if (!namedGroups().containsKey(name)) 2245 throw error("(named capturing group <"+ name+"> does not exit"); 2246 if (create) { 2247 if (has(CASE_INSENSITIVE)) 2248 root = new CIBackRef(namedGroups().get(name), has(UNICODE_CASE)); 2249 else 2250 root = new BackRef(namedGroups().get(name)); 2251 } 2252 return -1; 2253 case 'l': 2254 case 'm': 2255 break; 2256 case 'n': 2257 return '\n'; 2258 case 'o': 2259 case 'p': 2260 case 'q': 2261 break; 2262 case 'r': 2263 return '\r'; 2264 case 's': 2265 if (create) root = new Ctype(ASCII.SPACE); 2266 return -1; 2267 case 't': 2268 return '\t'; 2269 case 'u': 2270 return u(); 2271 case 'v': 2272 return '\013'; 2273 case 'w': 2274 if (create) root = new Ctype(ASCII.WORD); 2275 return -1; 2276 case 'x': 2277 return x(); 2278 case 'y': 2279 break; 2280 case 'z': 2281 if (inclass) break; 2282 if (create) root = new End(); 2283 return -1; 2284 default: 2285 return ch; 2286 } 2287 throw error("Illegal/unsupported escape sequence"); 2288 } 2289 2290 /** 2291 * Parse a character class, and return the node that matches it. 2292 * 2293 * Consumes a ] on the way out if consume is true. Usually consume 2294 * is true except for the case of [abc&&def] where def is a separate 2295 * right hand node with "understood" brackets. 2296 */ 2297 private CharProperty clazz(boolean consume) { 2298 CharProperty prev = null; 2299 CharProperty node = null; 2300 BitClass bits = new BitClass(); 2301 boolean include = true; 2302 boolean firstInClass = true; 2303 int ch = next(); 2304 for (;;) { 2305 switch (ch) { 2306 case '^': 2307 // Negates if first char in a class, otherwise literal 2308 if (firstInClass) { 2309 if (temp[cursor-1] != '[') 2310 break; 2311 ch = next(); 2312 include = !include; 2313 continue; 2314 } else { 2315 // ^ not first in class, treat as literal 2316 break; 2317 } 2318 case '[': 2319 firstInClass = false; 2320 node = clazz(true); 2321 if (prev == null) 2322 prev = node; 2323 else 2324 prev = union(prev, node); 2325 ch = peek(); 2326 continue; 2327 case '&': 2328 firstInClass = false; 2329 ch = next(); 2330 if (ch == '&') { 2331 ch = next(); 2332 CharProperty rightNode = null; 2333 while (ch != ']' && ch != '&') { 2334 if (ch == '[') { 2335 if (rightNode == null) 2336 rightNode = clazz(true); 2337 else 2338 rightNode = union(rightNode, clazz(true)); 2339 } else { // abc&&def 2340 unread(); 2341 rightNode = clazz(false); 2342 } 2343 ch = peek(); 2344 } 2345 if (rightNode != null) 2346 node = rightNode; 2347 if (prev == null) { 2348 if (rightNode == null) 2349 throw error("Bad class syntax"); 2350 else 2351 prev = rightNode; 2352 } else { 2353 prev = intersection(prev, node); 2354 } 2355 } else { 2356 // treat as a literal & 2357 unread(); 2358 break; 2359 } 2360 continue; 2361 case 0: 2362 firstInClass = false; 2363 if (cursor >= patternLength) 2364 throw error("Unclosed character class"); 2365 break; 2366 case ']': 2367 firstInClass = false; 2368 if (prev != null) { 2369 if (consume) 2370 next(); 2371 return prev; 2372 } 2373 break; 2374 default: 2375 firstInClass = false; 2376 break; 2377 } 2378 node = range(bits); 2379 if (include) { 2380 if (prev == null) { 2381 prev = node; 2382 } else { 2383 if (prev != node) 2384 prev = union(prev, node); 2385 } 2386 } else { 2387 if (prev == null) { 2388 prev = node.complement(); 2389 } else { 2390 if (prev != node) 2391 prev = setDifference(prev, node); 2392 } 2393 } 2394 ch = peek(); 2395 } 2396 } 2397 2398 private CharProperty bitsOrSingle(BitClass bits, int ch) { 2399 /* Bits can only handle codepoints in [u+0000-u+00ff] range. 2400 Use "single" node instead of bits when dealing with unicode 2401 case folding for codepoints listed below. 2402 (1)Uppercase out of range: u+00ff, u+00b5 2403 toUpperCase(u+00ff) -> u+0178 2404 toUpperCase(u+00b5) -> u+039c 2405 (2)LatinSmallLetterLongS u+17f 2406 toUpperCase(u+017f) -> u+0053 2407 (3)LatinSmallLetterDotlessI u+131 2408 toUpperCase(u+0131) -> u+0049 2409 (4)LatinCapitalLetterIWithDotAbove u+0130 2410 toLowerCase(u+0130) -> u+0069 2411 (5)KelvinSign u+212a 2412 toLowerCase(u+212a) ==> u+006B 2413 (6)AngstromSign u+212b 2414 toLowerCase(u+212b) ==> u+00e5 2415 */ 2416 int d; 2417 if (ch < 256 && 2418 !(has(CASE_INSENSITIVE) && has(UNICODE_CASE) && 2419 (ch == 0xff || ch == 0xb5 || 2420 ch == 0x49 || ch == 0x69 || //I and i 2421 ch == 0x53 || ch == 0x73 || //S and s 2422 ch == 0x4b || ch == 0x6b || //K and k 2423 ch == 0xc5 || ch == 0xe5))) //A+ring 2424 return bits.add(ch, flags()); 2425 return newSingle(ch); 2426 } 2427 2428 /** 2429 * Parse a single character or a character range in a character class 2430 * and return its representative node. 2431 */ 2432 private CharProperty range(BitClass bits) { 2433 int ch = peek(); 2434 if (ch == '\\') { 2435 ch = nextEscaped(); 2436 if (ch == 'p' || ch == 'P') { // A property 2437 boolean comp = (ch == 'P'); 2438 boolean oneLetter = true; 2439 // Consume { if present 2440 ch = next(); 2441 if (ch != '{') 2442 unread(); 2443 else 2444 oneLetter = false; 2445 return family(oneLetter, comp); 2446 } else { // ordinary escape 2447 unread(); 2448 ch = escape(true, true); 2449 if (ch == -1) 2450 return (CharProperty) root; 2451 } 2452 } else { 2453 ch = single(); 2454 } 2455 if (ch >= 0) { 2456 if (peek() == '-') { 2457 int endRange = temp[cursor+1]; 2458 if (endRange == '[') { 2459 return bitsOrSingle(bits, ch); 2460 } 2461 if (endRange != ']') { 2462 next(); 2463 int m = single(); 2464 if (m < ch) 2465 throw error("Illegal character range"); 2466 if (has(CASE_INSENSITIVE)) 2467 return caseInsensitiveRangeFor(ch, m); 2468 else 2469 return rangeFor(ch, m); 2470 } 2471 } 2472 return bitsOrSingle(bits, ch); 2473 } 2474 throw error("Unexpected character '"+((char)ch)+"'"); 2475 } 2476 2477 private int single() { 2478 int ch = peek(); 2479 switch (ch) { 2480 case '\\': 2481 return escape(true, false); 2482 default: 2483 next(); 2484 return ch; 2485 } 2486 } 2487 2488 /** 2489 * Parses a Unicode character family and returns its representative node. 2490 */ 2491 private CharProperty family(boolean singleLetter, 2492 boolean maybeComplement) 2493 { 2494 next(); 2495 String name; 2496 CharProperty node; 2497 2498 if (singleLetter) { 2499 int c = temp[cursor]; 2500 if (!Character.isSupplementaryCodePoint(c)) { 2501 name = String.valueOf((char)c); 2502 } else { 2503 name = new String(temp, cursor, 1); 2504 } 2505 read(); 2506 } else { 2507 int i = cursor; 2508 mark('}'); 2509 while(read() != '}') { 2510 } 2511 mark('\000'); 2512 int j = cursor; 2513 if (j > patternLength) 2514 throw error("Unclosed character family"); 2515 if (i + 1 >= j) 2516 throw error("Empty character family"); 2517 name = new String(temp, i, j-i-1); 2518 } 2519 2520 int i = name.indexOf('='); 2521 if (i != -1) { 2522 // property construct \p{name=value} 2523 String value = name.substring(i + 1); 2524 name = name.substring(0, i).toLowerCase(Locale.ENGLISH); 2525 if ("sc".equals(name) || "script".equals(name)) { 2526 node = unicodeScriptPropertyFor(value); 2527 } else if ("blk".equals(name) || "block".equals(name)) { 2528 node = unicodeBlockPropertyFor(value); 2529 } else if ("gc".equals(name) || "general_category".equals(name)) { 2530 node = charPropertyNodeFor(value); 2531 } else { 2532 throw error("Unknown Unicode property {name=<" + name + ">, " 2533 + "value=<" + value + ">}"); 2534 } 2535 } else { 2536 if (name.startsWith("In")) { 2537 // \p{inBlockName} 2538 node = unicodeBlockPropertyFor(name.substring(2)); 2539 } else if (name.startsWith("Is")) { 2540 // \p{isGeneralCategory} and \p{isScriptName} 2541 name = name.substring(2); 2542 node = CharPropertyNames.charPropertyFor(name); 2543 if (node == null) 2544 node = unicodeScriptPropertyFor(name); 2545 } else { 2546 node = charPropertyNodeFor(name); 2547 } 2548 } 2549 if (maybeComplement) { 2550 if (node instanceof Category || node instanceof Block) 2551 hasSupplementary = true; 2552 node = node.complement(); 2553 } 2554 return node; 2555 } 2556 2557 2558 /** 2559 * Returns a CharProperty matching all characters belong to 2560 * a UnicodeScript. 2561 */ 2562 private CharProperty unicodeScriptPropertyFor(String name) { 2563 final Character.UnicodeScript script; 2564 try { 2565 script = Character.UnicodeScript.forName(name); 2566 } catch (IllegalArgumentException iae) { 2567 throw error("Unknown character script name {" + name + "}"); 2568 } 2569 return new Script(script); 2570 } 2571 2572 /** 2573 * Returns a CharProperty matching all characters in a UnicodeBlock. 2574 */ 2575 private CharProperty unicodeBlockPropertyFor(String name) { 2576 final Character.UnicodeBlock block; 2577 try { 2578 block = Character.UnicodeBlock.forName(name); 2579 } catch (IllegalArgumentException iae) { 2580 throw error("Unknown character block name {" + name + "}"); 2581 } 2582 return new Block(block); 2583 } 2584 2585 /** 2586 * Returns a CharProperty matching all characters in a named property. 2587 */ 2588 private CharProperty charPropertyNodeFor(String name) { 2589 CharProperty p = CharPropertyNames.charPropertyFor(name); 2590 if (p == null) 2591 throw error("Unknown character property name {" + name + "}"); 2592 return p; 2593 } 2594 2595 /** 2596 * Parses and returns the name of a "named capturing group", the trailing 2597 * ">" is consumed after parsing. 2598 */ 2599 private String groupname(int ch) { 2600 StringBuilder sb = new StringBuilder(); 2601 sb.append(Character.toChars(ch)); 2602 while (ASCII.isLower(ch=read()) || ASCII.isUpper(ch) || 2603 ASCII.isDigit(ch)) { 2604 sb.append(Character.toChars(ch)); 2605 } 2606 if (sb.length() == 0) 2607 throw error("named capturing group has 0 length name"); 2608 if (ch != '>') 2609 throw error("named capturing group is missing trailing '>'"); 2610 return sb.toString(); 2611 } 2612 2613 /** 2614 * Parses a group and returns the head node of a set of nodes that process 2615 * the group. Sometimes a double return system is used where the tail is 2616 * returned in root. 2617 */ 2618 private Node group0() { 2619 boolean capturingGroup = false; 2620 Node head = null; 2621 Node tail = null; 2622 int save = flags; 2623 root = null; 2624 int ch = next(); 2625 if (ch == '?') { 2626 ch = skip(); 2627 switch (ch) { 2628 case ':': // (?:xxx) pure group 2629 head = createGroup(true); 2630 tail = root; 2631 head.next = expr(tail); 2632 break; 2633 case '=': // (?=xxx) and (?!xxx) lookahead 2634 case '!': 2635 head = createGroup(true); 2636 tail = root; 2637 head.next = expr(tail); 2638 if (ch == '=') { 2639 head = tail = new Pos(head); 2640 } else { 2641 head = tail = new Neg(head); 2642 } 2643 break; 2644 case '>': // (?>xxx) independent group 2645 head = createGroup(true); 2646 tail = root; 2647 head.next = expr(tail); 2648 head = tail = new Ques(head, INDEPENDENT); 2649 break; 2650 case '<': // (?<xxx) look behind 2651 ch = read(); 2652 if (ASCII.isLower(ch) || ASCII.isUpper(ch)) { 2653 // named captured group 2654 String name = groupname(ch); 2655 if (namedGroups().containsKey(name)) 2656 throw error("Named capturing group <" + name 2657 + "> is already defined"); 2658 capturingGroup = true; 2659 head = createGroup(false); 2660 tail = root; 2661 namedGroups().put(name, capturingGroupCount-1); 2662 head.next = expr(tail); 2663 break; 2664 } 2665 int start = cursor; 2666 head = createGroup(true); 2667 tail = root; 2668 head.next = expr(tail); 2669 tail.next = lookbehindEnd; 2670 TreeInfo info = new TreeInfo(); 2671 head.study(info); 2672 if (info.maxValid == false) { 2673 throw error("Look-behind group does not have " 2674 + "an obvious maximum length"); 2675 } 2676 boolean hasSupplementary = findSupplementary(start, patternLength); 2677 if (ch == '=') { 2678 head = tail = (hasSupplementary ? 2679 new BehindS(head, info.maxLength, 2680 info.minLength) : 2681 new Behind(head, info.maxLength, 2682 info.minLength)); 2683 } else if (ch == '!') { 2684 head = tail = (hasSupplementary ? 2685 new NotBehindS(head, info.maxLength, 2686 info.minLength) : 2687 new NotBehind(head, info.maxLength, 2688 info.minLength)); 2689 } else { 2690 throw error("Unknown look-behind group"); 2691 } 2692 break; 2693 case '$': 2694 case '@': 2695 throw error("Unknown group type"); 2696 default: // (?xxx:) inlined match flags 2697 unread(); 2698 addFlag(); 2699 ch = read(); 2700 if (ch == ')') { 2701 return null; // Inline modifier only 2702 } 2703 if (ch != ':') { 2704 throw error("Unknown inline modifier"); 2705 } 2706 head = createGroup(true); 2707 tail = root; 2708 head.next = expr(tail); 2709 break; 2710 } 2711 } else { // (xxx) a regular group 2712 capturingGroup = true; 2713 head = createGroup(false); 2714 tail = root; 2715 head.next = expr(tail); 2716 } 2717 2718 accept(')', "Unclosed group"); 2719 flags = save; 2720 2721 // Check for quantifiers 2722 Node node = closure(head); 2723 if (node == head) { // No closure 2724 root = tail; 2725 return node; // Dual return 2726 } 2727 if (head == tail) { // Zero length assertion 2728 root = node; 2729 return node; // Dual return 2730 } 2731 2732 if (node instanceof Ques) { 2733 Ques ques = (Ques) node; 2734 if (ques.type == POSSESSIVE) { 2735 root = node; 2736 return node; 2737 } 2738 tail.next = new BranchConn(); 2739 tail = tail.next; 2740 if (ques.type == GREEDY) { 2741 head = new Branch(head, null, tail); 2742 } else { // Reluctant quantifier 2743 head = new Branch(null, head, tail); 2744 } 2745 root = tail; 2746 return head; 2747 } else if (node instanceof Curly) { 2748 Curly curly = (Curly) node; 2749 if (curly.type == POSSESSIVE) { 2750 root = node; 2751 return node; 2752 } 2753 // Discover if the group is deterministic 2754 TreeInfo info = new TreeInfo(); 2755 if (head.study(info)) { // Deterministic 2756 GroupTail temp = (GroupTail) tail; 2757 head = root = new GroupCurly(head.next, curly.cmin, 2758 curly.cmax, curly.type, 2759 ((GroupTail)tail).localIndex, 2760 ((GroupTail)tail).groupIndex, 2761 capturingGroup); 2762 return head; 2763 } else { // Non-deterministic 2764 int temp = ((GroupHead) head).localIndex; 2765 Loop loop; 2766 if (curly.type == GREEDY) 2767 loop = new Loop(this.localCount, temp); 2768 else // Reluctant Curly 2769 loop = new LazyLoop(this.localCount, temp); 2770 Prolog prolog = new Prolog(loop); 2771 this.localCount += 1; 2772 loop.cmin = curly.cmin; 2773 loop.cmax = curly.cmax; 2774 loop.body = head; 2775 tail.next = loop; 2776 root = loop; 2777 return prolog; // Dual return 2778 } 2779 } 2780 throw error("Internal logic error"); 2781 } 2782 2783 /** 2784 * Create group head and tail nodes using double return. If the group is 2785 * created with anonymous true then it is a pure group and should not 2786 * affect group counting. 2787 */ 2788 private Node createGroup(boolean anonymous) { 2789 int localIndex = localCount++; 2790 int groupIndex = 0; 2791 if (!anonymous) 2792 groupIndex = capturingGroupCount++; 2793 GroupHead head = new GroupHead(localIndex); 2794 root = new GroupTail(localIndex, groupIndex); 2795 if (!anonymous && groupIndex < 10) 2796 groupNodes[groupIndex] = head; 2797 return head; 2798 } 2799 2800 /** 2801 * Parses inlined match flags and set them appropriately. 2802 */ 2803 private void addFlag() { 2804 int ch = peek(); 2805 for (;;) { 2806 switch (ch) { 2807 case 'i': 2808 flags |= CASE_INSENSITIVE; 2809 break; 2810 case 'm': 2811 flags |= MULTILINE; 2812 break; 2813 case 's': 2814 flags |= DOTALL; 2815 break; 2816 case 'd': 2817 flags |= UNIX_LINES; 2818 break; 2819 case 'u': 2820 flags |= UNICODE_CASE; 2821 break; 2822 case 'c': 2823 flags |= CANON_EQ; 2824 break; 2825 case 'x': 2826 flags |= COMMENTS; 2827 break; 2828 case '-': // subFlag then fall through 2829 ch = next(); 2830 subFlag(); 2831 default: 2832 return; 2833 } 2834 ch = next(); 2835 } 2836 } 2837 2838 /** 2839 * Parses the second part of inlined match flags and turns off 2840 * flags appropriately. 2841 */ 2842 private void subFlag() { 2843 int ch = peek(); 2844 for (;;) { 2845 switch (ch) { 2846 case 'i': 2847 flags &= ~CASE_INSENSITIVE; 2848 break; 2849 case 'm': 2850 flags &= ~MULTILINE; 2851 break; 2852 case 's': 2853 flags &= ~DOTALL; 2854 break; 2855 case 'd': 2856 flags &= ~UNIX_LINES; 2857 break; 2858 case 'u': 2859 flags &= ~UNICODE_CASE; 2860 break; 2861 case 'c': 2862 flags &= ~CANON_EQ; 2863 break; 2864 case 'x': 2865 flags &= ~COMMENTS; 2866 break; 2867 default: 2868 return; 2869 } 2870 ch = next(); 2871 } 2872 } 2873 2874 static final int MAX_REPS = 0x7FFFFFFF; 2875 2876 static final int GREEDY = 0; 2877 2878 static final int LAZY = 1; 2879 2880 static final int POSSESSIVE = 2; 2881 2882 static final int INDEPENDENT = 3; 2883 2884 /** 2885 * Processes repetition. If the next character peeked is a quantifier 2886 * then new nodes must be appended to handle the repetition. 2887 * Prev could be a single or a group, so it could be a chain of nodes. 2888 */ 2889 private Node closure(Node prev) { 2890 Node atom; 2891 int ch = peek(); 2892 switch (ch) { 2893 case '?': 2894 ch = next(); 2895 if (ch == '?') { 2896 next(); 2897 return new Ques(prev, LAZY); 2898 } else if (ch == '+') { 2899 next(); 2900 return new Ques(prev, POSSESSIVE); 2901 } 2902 return new Ques(prev, GREEDY); 2903 case '*': 2904 ch = next(); 2905 if (ch == '?') { 2906 next(); 2907 return new Curly(prev, 0, MAX_REPS, LAZY); 2908 } else if (ch == '+') { 2909 next(); 2910 return new Curly(prev, 0, MAX_REPS, POSSESSIVE); 2911 } 2912 return new Curly(prev, 0, MAX_REPS, GREEDY); 2913 case '+': 2914 ch = next(); 2915 if (ch == '?') { 2916 next(); 2917 return new Curly(prev, 1, MAX_REPS, LAZY); 2918 } else if (ch == '+') { 2919 next(); 2920 return new Curly(prev, 1, MAX_REPS, POSSESSIVE); 2921 } 2922 return new Curly(prev, 1, MAX_REPS, GREEDY); 2923 case '{': 2924 ch = temp[cursor+1]; 2925 if (ASCII.isDigit(ch)) { 2926 skip(); 2927 int cmin = 0; 2928 do { 2929 cmin = cmin * 10 + (ch - '0'); 2930 } while (ASCII.isDigit(ch = read())); 2931 int cmax = cmin; 2932 if (ch == ',') { 2933 ch = read(); 2934 cmax = MAX_REPS; 2935 if (ch != '}') { 2936 cmax = 0; 2937 while (ASCII.isDigit(ch)) { 2938 cmax = cmax * 10 + (ch - '0'); 2939 ch = read(); 2940 } 2941 } 2942 } 2943 if (ch != '}') 2944 throw error("Unclosed counted closure"); 2945 if (((cmin) | (cmax) | (cmax - cmin)) < 0) 2946 throw error("Illegal repetition range"); 2947 Curly curly; 2948 ch = peek(); 2949 if (ch == '?') { 2950 next(); 2951 curly = new Curly(prev, cmin, cmax, LAZY); 2952 } else if (ch == '+') { 2953 next(); 2954 curly = new Curly(prev, cmin, cmax, POSSESSIVE); 2955 } else { 2956 curly = new Curly(prev, cmin, cmax, GREEDY); 2957 } 2958 return curly; 2959 } else { 2960 throw error("Illegal repetition"); 2961 } 2962 default: 2963 return prev; 2964 } 2965 } 2966 2967 /** 2968 * Utility method for parsing control escape sequences. 2969 */ 2970 private int c() { 2971 if (cursor < patternLength) { 2972 return read() ^ 64; 2973 } 2974 throw error("Illegal control escape sequence"); 2975 } 2976 2977 /** 2978 * Utility method for parsing octal escape sequences. 2979 */ 2980 private int o() { 2981 int n = read(); 2982 if (((n-'0')|('7'-n)) >= 0) { 2983 int m = read(); 2984 if (((m-'0')|('7'-m)) >= 0) { 2985 int o = read(); 2986 if ((((o-'0')|('7'-o)) >= 0) && (((n-'0')|('3'-n)) >= 0)) { 2987 return (n - '0') * 64 + (m - '0') * 8 + (o - '0'); 2988 } 2989 unread(); 2990 return (n - '0') * 8 + (m - '0'); 2991 } 2992 unread(); 2993 return (n - '0'); 2994 } 2995 throw error("Illegal octal escape sequence"); 2996 } 2997 2998 /** 2999 * Utility method for parsing hexadecimal escape sequences. 3000 */ 3001 private int x() { 3002 int n = read(); 3003 if (ASCII.isHexDigit(n)) { 3004 int m = read(); 3005 if (ASCII.isHexDigit(m)) { 3006 return ASCII.toDigit(n) * 16 + ASCII.toDigit(m); 3007 } 3008 } else if (n == '{' && ASCII.isHexDigit(peek())) { 3009 int ch = 0; 3010 while (ASCII.isHexDigit(n = read())) { 3011 ch = (ch << 4) + ASCII.toDigit(n); 3012 if (ch > Character.MAX_CODE_POINT) 3013 throw error("Hexadecimal codepoint is too big"); 3014 } 3015 if (n != '}') 3016 throw error("Unclosed hexadecimal escape sequence"); 3017 return ch; 3018 } 3019 throw error("Illegal hexadecimal escape sequence"); 3020 } 3021 3022 /** 3023 * Utility method for parsing unicode escape sequences. 3024 */ 3025 private int cursor() { 3026 return cursor; 3027 } 3028 3029 private void setcursor(int pos) { 3030 cursor = pos; 3031 } 3032 3033 private int uxxxx() { 3034 int n = 0; 3035 for (int i = 0; i < 4; i++) { 3036 int ch = read(); 3037 if (!ASCII.isHexDigit(ch)) { 3038 throw error("Illegal Unicode escape sequence"); 3039 } 3040 n = n * 16 + ASCII.toDigit(ch); 3041 } 3042 return n; 3043 } 3044 3045 private int u() { 3046 int n = uxxxx(); 3047 if (Character.isHighSurrogate((char)n)) { 3048 int cur = cursor(); 3049 if (read() == '\\' && read() == 'u') { 3050 int n2 = uxxxx(); 3051 if (Character.isLowSurrogate((char)n2)) 3052 return Character.toCodePoint((char)n, (char)n2); 3053 } 3054 setcursor(cur); 3055 } 3056 return n; 3057 } 3058 3059 // 3060 // Utility methods for code point support 3061 // 3062 3063 private static final int countChars(CharSequence seq, int index, 3064 int lengthInCodePoints) { 3065 // optimization 3066 if (lengthInCodePoints == 1 && !Character.isHighSurrogate(seq.charAt(index))) { 3067 assert (index >= 0 && index < seq.length()); 3068 return 1; 3069 } 3070 int length = seq.length(); 3071 int x = index; 3072 if (lengthInCodePoints >= 0) { 3073 assert (index >= 0 && index < length); 3074 for (int i = 0; x < length && i < lengthInCodePoints; i++) { 3075 if (Character.isHighSurrogate(seq.charAt(x++))) { 3076 if (x < length && Character.isLowSurrogate(seq.charAt(x))) { 3077 x++; 3078 } 3079 } 3080 } 3081 return x - index; 3082 } 3083 3084 assert (index >= 0 && index <= length); 3085 if (index == 0) { 3086 return 0; 3087 } 3088 int len = -lengthInCodePoints; 3089 for (int i = 0; x > 0 && i < len; i++) { 3090 if (Character.isLowSurrogate(seq.charAt(--x))) { 3091 if (x > 0 && Character.isHighSurrogate(seq.charAt(x-1))) { 3092 x--; 3093 } 3094 } 3095 } 3096 return index - x; 3097 } 3098 3099 private static final int countCodePoints(CharSequence seq) { 3100 int length = seq.length(); 3101 int n = 0; 3102 for (int i = 0; i < length; ) { 3103 n++; 3104 if (Character.isHighSurrogate(seq.charAt(i++))) { 3105 if (i < length && Character.isLowSurrogate(seq.charAt(i))) { 3106 i++; 3107 } 3108 } 3109 } 3110 return n; 3111 } 3112 3113 /** 3114 * Creates a bit vector for matching Latin-1 values. A normal BitClass 3115 * never matches values above Latin-1, and a complemented BitClass always 3116 * matches values above Latin-1. 3117 */ 3118 private static final class BitClass extends BmpCharProperty { 3119 final boolean[] bits; 3120 BitClass() { bits = new boolean[256]; } 3121 private BitClass(boolean[] bits) { this.bits = bits; } 3122 BitClass add(int c, int flags) { 3123 assert c >= 0 && c <= 255; 3124 if ((flags & CASE_INSENSITIVE) != 0) { 3125 if (ASCII.isAscii(c)) { 3126 bits[ASCII.toUpper(c)] = true; 3127 bits[ASCII.toLower(c)] = true; 3128 } else if ((flags & UNICODE_CASE) != 0) { 3129 bits[Character.toLowerCase(c)] = true; 3130 bits[Character.toUpperCase(c)] = true; 3131 } 3132 } 3133 bits[c] = true; 3134 return this; 3135 } 3136 boolean isSatisfiedBy(int ch) { 3137 return ch < 256 && bits[ch]; 3138 } 3139 } 3140 3141 /** 3142 * Returns a suitably optimized, single character matcher. 3143 */ 3144 private CharProperty newSingle(final int ch) { 3145 if (has(CASE_INSENSITIVE)) { 3146 int lower, upper; 3147 if (has(UNICODE_CASE)) { 3148 upper = Character.toUpperCase(ch); 3149 lower = Character.toLowerCase(upper); 3150 if (upper != lower) 3151 return new SingleU(lower); 3152 } else if (ASCII.isAscii(ch)) { 3153 lower = ASCII.toLower(ch); 3154 upper = ASCII.toUpper(ch); 3155 if (lower != upper) 3156 return new SingleI(lower, upper); 3157 } 3158 } 3159 if (isSupplementary(ch)) 3160 return new SingleS(ch); // Match a given Unicode character 3161 return new Single(ch); // Match a given BMP character 3162 } 3163 3164 /** 3165 * Utility method for creating a string slice matcher. 3166 */ 3167 private Node newSlice(int[] buf, int count, boolean hasSupplementary) { 3168 int[] tmp = new int[count]; 3169 if (has(CASE_INSENSITIVE)) { 3170 if (has(UNICODE_CASE)) { 3171 for (int i = 0; i < count; i++) { 3172 tmp[i] = Character.toLowerCase( 3173 Character.toUpperCase(buf[i])); 3174 } 3175 return hasSupplementary? new SliceUS(tmp) : new SliceU(tmp); 3176 } 3177 for (int i = 0; i < count; i++) { 3178 tmp[i] = ASCII.toLower(buf[i]); 3179 } 3180 return hasSupplementary? new SliceIS(tmp) : new SliceI(tmp); 3181 } 3182 for (int i = 0; i < count; i++) { 3183 tmp[i] = buf[i]; 3184 } 3185 return hasSupplementary ? new SliceS(tmp) : new Slice(tmp); 3186 } 3187 3188 /** 3189 * The following classes are the building components of the object 3190 * tree that represents a compiled regular expression. The object tree 3191 * is made of individual elements that handle constructs in the Pattern. 3192 * Each type of object knows how to match its equivalent construct with 3193 * the match() method. 3194 */ 3195 3196 /** 3197 * Base class for all node classes. Subclasses should override the match() 3198 * method as appropriate. This class is an accepting node, so its match() 3199 * always returns true. 3200 */ 3201 static class Node extends Object { 3202 Node next; 3203 Node() { 3204 next = Pattern.accept; 3205 } 3206 /** 3207 * This method implements the classic accept node. 3208 */ 3209 boolean match(Matcher matcher, int i, CharSequence seq) { 3210 matcher.last = i; 3211 matcher.groups[0] = matcher.first; 3212 matcher.groups[1] = matcher.last; 3213 return true; 3214 } 3215 /** 3216 * This method is good for all zero length assertions. 3217 */ 3218 boolean study(TreeInfo info) { 3219 if (next != null) { 3220 return next.study(info); 3221 } else { 3222 return info.deterministic; 3223 } 3224 } 3225 } 3226 3227 static class LastNode extends Node { 3228 /** 3229 * This method implements the classic accept node with 3230 * the addition of a check to see if the match occurred 3231 * using all of the input. 3232 */ 3233 boolean match(Matcher matcher, int i, CharSequence seq) { 3234 if (matcher.acceptMode == Matcher.ENDANCHOR && i != matcher.to) 3235 return false; 3236 matcher.last = i; 3237 matcher.groups[0] = matcher.first; 3238 matcher.groups[1] = matcher.last; 3239 return true; 3240 } 3241 } 3242 3243 /** 3244 * Used for REs that can start anywhere within the input string. 3245 * This basically tries to match repeatedly at each spot in the 3246 * input string, moving forward after each try. An anchored search 3247 * or a BnM will bypass this node completely. 3248 */ 3249 static class Start extends Node { 3250 int minLength; 3251 Start(Node node) { 3252 this.next = node; 3253 TreeInfo info = new TreeInfo(); 3254 next.study(info); 3255 minLength = info.minLength; 3256 } 3257 boolean match(Matcher matcher, int i, CharSequence seq) { 3258 if (i > matcher.to - minLength) { 3259 matcher.hitEnd = true; 3260 return false; 3261 } 3262 int guard = matcher.to - minLength; 3263 for (; i <= guard; i++) { 3264 if (next.match(matcher, i, seq)) { 3265 matcher.first = i; 3266 matcher.groups[0] = matcher.first; 3267 matcher.groups[1] = matcher.last; 3268 return true; 3269 } 3270 } 3271 matcher.hitEnd = true; 3272 return false; 3273 } 3274 boolean study(TreeInfo info) { 3275 next.study(info); 3276 info.maxValid = false; 3277 info.deterministic = false; 3278 return false; 3279 } 3280 } 3281 3282 /* 3283 * StartS supports supplementary characters, including unpaired surrogates. 3284 */ 3285 static final class StartS extends Start { 3286 StartS(Node node) { 3287 super(node); 3288 } 3289 boolean match(Matcher matcher, int i, CharSequence seq) { 3290 if (i > matcher.to - minLength) { 3291 matcher.hitEnd = true; 3292 return false; 3293 } 3294 int guard = matcher.to - minLength; 3295 while (i <= guard) { 3296 //if ((ret = next.match(matcher, i, seq)) || i == guard) 3297 if (next.match(matcher, i, seq)) { 3298 matcher.first = i; 3299 matcher.groups[0] = matcher.first; 3300 matcher.groups[1] = matcher.last; 3301 return true; 3302 } 3303 if (i == guard) 3304 break; 3305 // Optimization to move to the next character. This is 3306 // faster than countChars(seq, i, 1). 3307 if (Character.isHighSurrogate(seq.charAt(i++))) { 3308 if (i < seq.length() && 3309 Character.isLowSurrogate(seq.charAt(i))) { 3310 i++; 3311 } 3312 } 3313 } 3314 matcher.hitEnd = true; 3315 return false; 3316 } 3317 } 3318 3319 /** 3320 * Node to anchor at the beginning of input. This object implements the 3321 * match for a \A sequence, and the caret anchor will use this if not in 3322 * multiline mode. 3323 */ 3324 static final class Begin extends Node { 3325 boolean match(Matcher matcher, int i, CharSequence seq) { 3326 int fromIndex = (matcher.anchoringBounds) ? 3327 matcher.from : 0; 3328 if (i == fromIndex && next.match(matcher, i, seq)) { 3329 matcher.first = i; 3330 matcher.groups[0] = i; 3331 matcher.groups[1] = matcher.last; 3332 return true; 3333 } else { 3334 return false; 3335 } 3336 } 3337 } 3338 3339 /** 3340 * Node to anchor at the end of input. This is the absolute end, so this 3341 * should not match at the last newline before the end as $ will. 3342 */ 3343 static final class End extends Node { 3344 boolean match(Matcher matcher, int i, CharSequence seq) { 3345 int endIndex = (matcher.anchoringBounds) ? 3346 matcher.to : matcher.getTextLength(); 3347 if (i == endIndex) { 3348 matcher.hitEnd = true; 3349 return next.match(matcher, i, seq); 3350 } 3351 return false; 3352 } 3353 } 3354 3355 /** 3356 * Node to anchor at the beginning of a line. This is essentially the 3357 * object to match for the multiline ^. 3358 */ 3359 static final class Caret extends Node { 3360 boolean match(Matcher matcher, int i, CharSequence seq) { 3361 int startIndex = matcher.from; 3362 int endIndex = matcher.to; 3363 if (!matcher.anchoringBounds) { 3364 startIndex = 0; 3365 endIndex = matcher.getTextLength(); 3366 } 3367 // Perl does not match ^ at end of input even after newline 3368 if (i == endIndex) { 3369 matcher.hitEnd = true; 3370 return false; 3371 } 3372 if (i > startIndex) { 3373 char ch = seq.charAt(i-1); 3374 if (ch != '\n' && ch != '\r' 3375 && (ch|1) != '\u2029' 3376 && ch != '\u0085' ) { 3377 return false; 3378 } 3379 // Should treat /r/n as one newline 3380 if (ch == '\r' && seq.charAt(i) == '\n') 3381 return false; 3382 } 3383 return next.match(matcher, i, seq); 3384 } 3385 } 3386 3387 /** 3388 * Node to anchor at the beginning of a line when in unixdot mode. 3389 */ 3390 static final class UnixCaret extends Node { 3391 boolean match(Matcher matcher, int i, CharSequence seq) { 3392 int startIndex = matcher.from; 3393 int endIndex = matcher.to; 3394 if (!matcher.anchoringBounds) { 3395 startIndex = 0; 3396 endIndex = matcher.getTextLength(); 3397 } 3398 // Perl does not match ^ at end of input even after newline 3399 if (i == endIndex) { 3400 matcher.hitEnd = true; 3401 return false; 3402 } 3403 if (i > startIndex) { 3404 char ch = seq.charAt(i-1); 3405 if (ch != '\n') { 3406 return false; 3407 } 3408 } 3409 return next.match(matcher, i, seq); 3410 } 3411 } 3412 3413 /** 3414 * Node to match the location where the last match ended. 3415 * This is used for the \G construct. 3416 */ 3417 static final class LastMatch extends Node { 3418 boolean match(Matcher matcher, int i, CharSequence seq) { 3419 if (i != matcher.oldLast) 3420 return false; 3421 return next.match(matcher, i, seq); 3422 } 3423 } 3424 3425 /** 3426 * Node to anchor at the end of a line or the end of input based on the 3427 * multiline mode. 3428 * 3429 * When not in multiline mode, the $ can only match at the very end 3430 * of the input, unless the input ends in a line terminator in which 3431 * it matches right before the last line terminator. 3432 * 3433 * Note that \r\n is considered an atomic line terminator. 3434 * 3435 * Like ^ the $ operator matches at a position, it does not match the 3436 * line terminators themselves. 3437 */ 3438 static final class Dollar extends Node { 3439 boolean multiline; 3440 Dollar(boolean mul) { 3441 multiline = mul; 3442 } 3443 boolean match(Matcher matcher, int i, CharSequence seq) { 3444 int endIndex = (matcher.anchoringBounds) ? 3445 matcher.to : matcher.getTextLength(); 3446 if (!multiline) { 3447 if (i < endIndex - 2) 3448 return false; 3449 if (i == endIndex - 2) { 3450 char ch = seq.charAt(i); 3451 if (ch != '\r') 3452 return false; 3453 ch = seq.charAt(i + 1); 3454 if (ch != '\n') 3455 return false; 3456 } 3457 } 3458 // Matches before any line terminator; also matches at the 3459 // end of input 3460 // Before line terminator: 3461 // If multiline, we match here no matter what 3462 // If not multiline, fall through so that the end 3463 // is marked as hit; this must be a /r/n or a /n 3464 // at the very end so the end was hit; more input 3465 // could make this not match here 3466 if (i < endIndex) { 3467 char ch = seq.charAt(i); 3468 if (ch == '\n') { 3469 // No match between \r\n 3470 if (i > 0 && seq.charAt(i-1) == '\r') 3471 return false; 3472 if (multiline) 3473 return next.match(matcher, i, seq); 3474 } else if (ch == '\r' || ch == '\u0085' || 3475 (ch|1) == '\u2029') { 3476 if (multiline) 3477 return next.match(matcher, i, seq); 3478 } else { // No line terminator, no match 3479 return false; 3480 } 3481 } 3482 // Matched at current end so hit end 3483 matcher.hitEnd = true; 3484 // If a $ matches because of end of input, then more input 3485 // could cause it to fail! 3486 matcher.requireEnd = true; 3487 return next.match(matcher, i, seq); 3488 } 3489 boolean study(TreeInfo info) { 3490 next.study(info); 3491 return info.deterministic; 3492 } 3493 } 3494 3495 /** 3496 * Node to anchor at the end of a line or the end of input based on the 3497 * multiline mode when in unix lines mode. 3498 */ 3499 static final class UnixDollar extends Node { 3500 boolean multiline; 3501 UnixDollar(boolean mul) { 3502 multiline = mul; 3503 } 3504 boolean match(Matcher matcher, int i, CharSequence seq) { 3505 int endIndex = (matcher.anchoringBounds) ? 3506 matcher.to : matcher.getTextLength(); 3507 if (i < endIndex) { 3508 char ch = seq.charAt(i); 3509 if (ch == '\n') { 3510 // If not multiline, then only possible to 3511 // match at very end or one before end 3512 if (multiline == false && i != endIndex - 1) 3513 return false; 3514 // If multiline return next.match without setting 3515 // matcher.hitEnd 3516 if (multiline) 3517 return next.match(matcher, i, seq); 3518 } else { 3519 return false; 3520 } 3521 } 3522 // Matching because at the end or 1 before the end; 3523 // more input could change this so set hitEnd 3524 matcher.hitEnd = true; 3525 // If a $ matches because of end of input, then more input 3526 // could cause it to fail! 3527 matcher.requireEnd = true; 3528 return next.match(matcher, i, seq); 3529 } 3530 boolean study(TreeInfo info) { 3531 next.study(info); 3532 return info.deterministic; 3533 } 3534 } 3535 3536 /** 3537 * Abstract node class to match one character satisfying some 3538 * boolean property. 3539 */ 3540 private static abstract class CharProperty extends Node { 3541 abstract boolean isSatisfiedBy(int ch); 3542 CharProperty complement() { 3543 return new CharProperty() { 3544 boolean isSatisfiedBy(int ch) { 3545 return ! CharProperty.this.isSatisfiedBy(ch);}}; 3546 } 3547 boolean match(Matcher matcher, int i, CharSequence seq) { 3548 if (i < matcher.to) { 3549 int ch = Character.codePointAt(seq, i); 3550 return isSatisfiedBy(ch) 3551 && next.match(matcher, i+Character.charCount(ch), seq); 3552 } else { 3553 matcher.hitEnd = true; 3554 return false; 3555 } 3556 } 3557 boolean study(TreeInfo info) { 3558 info.minLength++; 3559 info.maxLength++; 3560 return next.study(info); 3561 } 3562 } 3563 3564 /** 3565 * Optimized version of CharProperty that works only for 3566 * properties never satisfied by Supplementary characters. 3567 */ 3568 private static abstract class BmpCharProperty extends CharProperty { 3569 boolean match(Matcher matcher, int i, CharSequence seq) { 3570 if (i < matcher.to) { 3571 return isSatisfiedBy(seq.charAt(i)) 3572 && next.match(matcher, i+1, seq); 3573 } else { 3574 matcher.hitEnd = true; 3575 return false; 3576 } 3577 } 3578 } 3579 3580 /** 3581 * Node class that matches a Supplementary Unicode character 3582 */ 3583 static final class SingleS extends CharProperty { 3584 final int c; 3585 SingleS(int c) { this.c = c; } 3586 boolean isSatisfiedBy(int ch) { 3587 return ch == c; 3588 } 3589 } 3590 3591 /** 3592 * Optimization -- matches a given BMP character 3593 */ 3594 static final class Single extends BmpCharProperty { 3595 final int c; 3596 Single(int c) { this.c = c; } 3597 boolean isSatisfiedBy(int ch) { 3598 return ch == c; 3599 } 3600 } 3601 3602 /** 3603 * Case insensitive matches a given BMP character 3604 */ 3605 static final class SingleI extends BmpCharProperty { 3606 final int lower; 3607 final int upper; 3608 SingleI(int lower, int upper) { 3609 this.lower = lower; 3610 this.upper = upper; 3611 } 3612 boolean isSatisfiedBy(int ch) { 3613 return ch == lower || ch == upper; 3614 } 3615 } 3616 3617 /** 3618 * Unicode case insensitive matches a given Unicode character 3619 */ 3620 static final class SingleU extends CharProperty { 3621 final int lower; 3622 SingleU(int lower) { 3623 this.lower = lower; 3624 } 3625 boolean isSatisfiedBy(int ch) { 3626 return lower == ch || 3627 lower == Character.toLowerCase(Character.toUpperCase(ch)); 3628 } 3629 } 3630 3631 3632 /** 3633 * Node class that matches a Unicode block. 3634 */ 3635 static final class Block extends CharProperty { 3636 final Character.UnicodeBlock block; 3637 Block(Character.UnicodeBlock block) { 3638 this.block = block; 3639 } 3640 boolean isSatisfiedBy(int ch) { 3641 return block == Character.UnicodeBlock.of(ch); 3642 } 3643 } 3644 3645 /** 3646 * Node class that matches a Unicode script 3647 */ 3648 static final class Script extends CharProperty { 3649 final Character.UnicodeScript script; 3650 Script(Character.UnicodeScript script) { 3651 this.script = script; 3652 } 3653 boolean isSatisfiedBy(int ch) { 3654 return script == Character.UnicodeScript.of(ch); 3655 } 3656 } 3657 3658 /** 3659 * Node class that matches a Unicode category. 3660 */ 3661 static final class Category extends CharProperty { 3662 final int typeMask; 3663 Category(int typeMask) { this.typeMask = typeMask; } 3664 boolean isSatisfiedBy(int ch) { 3665 return (typeMask & (1 << Character.getType(ch))) != 0; 3666 } 3667 } 3668 3669 /** 3670 * Node class that matches a POSIX type. 3671 */ 3672 static final class Ctype extends BmpCharProperty { 3673 final int ctype; 3674 Ctype(int ctype) { this.ctype = ctype; } 3675 boolean isSatisfiedBy(int ch) { 3676 return ch < 128 && ASCII.isType(ch, ctype); 3677 } 3678 } 3679 3680 /** 3681 * Base class for all Slice nodes 3682 */ 3683 static class SliceNode extends Node { 3684 int[] buffer; 3685 SliceNode(int[] buf) { 3686 buffer = buf; 3687 } 3688 boolean study(TreeInfo info) { 3689 info.minLength += buffer.length; 3690 info.maxLength += buffer.length; 3691 return next.study(info); 3692 } 3693 } 3694 3695 /** 3696 * Node class for a case sensitive/BMP-only sequence of literal 3697 * characters. 3698 */ 3699 static final class Slice extends SliceNode { 3700 Slice(int[] buf) { 3701 super(buf); 3702 } 3703 boolean match(Matcher matcher, int i, CharSequence seq) { 3704 int[] buf = buffer; 3705 int len = buf.length; 3706 for (int j=0; j<len; j++) { 3707 if ((i+j) >= matcher.to) { 3708 matcher.hitEnd = true; 3709 return false; 3710 } 3711 if (buf[j] != seq.charAt(i+j)) 3712 return false; 3713 } 3714 return next.match(matcher, i+len, seq); 3715 } 3716 } 3717 3718 /** 3719 * Node class for a case_insensitive/BMP-only sequence of literal 3720 * characters. 3721 */ 3722 static class SliceI extends SliceNode { 3723 SliceI(int[] buf) { 3724 super(buf); 3725 } 3726 boolean match(Matcher matcher, int i, CharSequence seq) { 3727 int[] buf = buffer; 3728 int len = buf.length; 3729 for (int j=0; j<len; j++) { 3730 if ((i+j) >= matcher.to) { 3731 matcher.hitEnd = true; 3732 return false; 3733 } 3734 int c = seq.charAt(i+j); 3735 if (buf[j] != c && 3736 buf[j] != ASCII.toLower(c)) 3737 return false; 3738 } 3739 return next.match(matcher, i+len, seq); 3740 } 3741 } 3742 3743 /** 3744 * Node class for a unicode_case_insensitive/BMP-only sequence of 3745 * literal characters. Uses unicode case folding. 3746 */ 3747 static final class SliceU extends SliceNode { 3748 SliceU(int[] buf) { 3749 super(buf); 3750 } 3751 boolean match(Matcher matcher, int i, CharSequence seq) { 3752 int[] buf = buffer; 3753 int len = buf.length; 3754 for (int j=0; j<len; j++) { 3755 if ((i+j) >= matcher.to) { 3756 matcher.hitEnd = true; 3757 return false; 3758 } 3759 int c = seq.charAt(i+j); 3760 if (buf[j] != c && 3761 buf[j] != Character.toLowerCase(Character.toUpperCase(c))) 3762 return false; 3763 } 3764 return next.match(matcher, i+len, seq); 3765 } 3766 } 3767 3768 /** 3769 * Node class for a case sensitive sequence of literal characters 3770 * including supplementary characters. 3771 */ 3772 static final class SliceS extends SliceNode { 3773 SliceS(int[] buf) { 3774 super(buf); 3775 } 3776 boolean match(Matcher matcher, int i, CharSequence seq) { 3777 int[] buf = buffer; 3778 int x = i; 3779 for (int j = 0; j < buf.length; j++) { 3780 if (x >= matcher.to) { 3781 matcher.hitEnd = true; 3782 return false; 3783 } 3784 int c = Character.codePointAt(seq, x); 3785 if (buf[j] != c) 3786 return false; 3787 x += Character.charCount(c); 3788 if (x > matcher.to) { 3789 matcher.hitEnd = true; 3790 return false; 3791 } 3792 } 3793 return next.match(matcher, x, seq); 3794 } 3795 } 3796 3797 /** 3798 * Node class for a case insensitive sequence of literal characters 3799 * including supplementary characters. 3800 */ 3801 static class SliceIS extends SliceNode { 3802 SliceIS(int[] buf) { 3803 super(buf); 3804 } 3805 int toLower(int c) { 3806 return ASCII.toLower(c); 3807 } 3808 boolean match(Matcher matcher, int i, CharSequence seq) { 3809 int[] buf = buffer; 3810 int x = i; 3811 for (int j = 0; j < buf.length; j++) { 3812 if (x >= matcher.to) { 3813 matcher.hitEnd = true; 3814 return false; 3815 } 3816 int c = Character.codePointAt(seq, x); 3817 if (buf[j] != c && buf[j] != toLower(c)) 3818 return false; 3819 x += Character.charCount(c); 3820 if (x > matcher.to) { 3821 matcher.hitEnd = true; 3822 return false; 3823 } 3824 } 3825 return next.match(matcher, x, seq); 3826 } 3827 } 3828 3829 /** 3830 * Node class for a case insensitive sequence of literal characters. 3831 * Uses unicode case folding. 3832 */ 3833 static final class SliceUS extends SliceIS { 3834 SliceUS(int[] buf) { 3835 super(buf); 3836 } 3837 int toLower(int c) { 3838 return Character.toLowerCase(Character.toUpperCase(c)); 3839 } 3840 } 3841 3842 private static boolean inRange(int lower, int ch, int upper) { 3843 return lower <= ch && ch <= upper; 3844 } 3845 3846 /** 3847 * Returns node for matching characters within an explicit value range. 3848 */ 3849 private static CharProperty rangeFor(final int lower, 3850 final int upper) { 3851 return new CharProperty() { 3852 boolean isSatisfiedBy(int ch) { 3853 return inRange(lower, ch, upper);}}; 3854 } 3855 3856 /** 3857 * Returns node for matching characters within an explicit value 3858 * range in a case insensitive manner. 3859 */ 3860 private CharProperty caseInsensitiveRangeFor(final int lower, 3861 final int upper) { 3862 if (has(UNICODE_CASE)) 3863 return new CharProperty() { 3864 boolean isSatisfiedBy(int ch) { 3865 if (inRange(lower, ch, upper)) 3866 return true; 3867 int up = Character.toUpperCase(ch); 3868 return inRange(lower, up, upper) || 3869 inRange(lower, Character.toLowerCase(up), upper);}}; 3870 return new CharProperty() { 3871 boolean isSatisfiedBy(int ch) { 3872 return inRange(lower, ch, upper) || 3873 ASCII.isAscii(ch) && 3874 (inRange(lower, ASCII.toUpper(ch), upper) || 3875 inRange(lower, ASCII.toLower(ch), upper)); 3876 }}; 3877 } 3878 3879 /** 3880 * Implements the Unicode category ALL and the dot metacharacter when 3881 * in dotall mode. 3882 */ 3883 static final class All extends CharProperty { 3884 boolean isSatisfiedBy(int ch) { 3885 return true; 3886 } 3887 } 3888 3889 /** 3890 * Node class for the dot metacharacter when dotall is not enabled. 3891 */ 3892 static final class Dot extends CharProperty { 3893 boolean isSatisfiedBy(int ch) { 3894 return (ch != '\n' && ch != '\r' 3895 && (ch|1) != '\u2029' 3896 && ch != '\u0085'); 3897 } 3898 } 3899 3900 /** 3901 * Node class for the dot metacharacter when dotall is not enabled 3902 * but UNIX_LINES is enabled. 3903 */ 3904 static final class UnixDot extends CharProperty { 3905 boolean isSatisfiedBy(int ch) { 3906 return ch != '\n'; 3907 } 3908 } 3909 3910 /** 3911 * The 0 or 1 quantifier. This one class implements all three types. 3912 */ 3913 static final class Ques extends Node { 3914 Node atom; 3915 int type; 3916 Ques(Node node, int type) { 3917 this.atom = node; 3918 this.type = type; 3919 } 3920 boolean match(Matcher matcher, int i, CharSequence seq) { 3921 switch (type) { 3922 case GREEDY: 3923 return (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)) 3924 || next.match(matcher, i, seq); 3925 case LAZY: 3926 return next.match(matcher, i, seq) 3927 || (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)); 3928 case POSSESSIVE: 3929 if (atom.match(matcher, i, seq)) i = matcher.last; 3930 return next.match(matcher, i, seq); 3931 default: 3932 return atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq); 3933 } 3934 } 3935 boolean study(TreeInfo info) { 3936 if (type != INDEPENDENT) { 3937 int minL = info.minLength; 3938 atom.study(info); 3939 info.minLength = minL; 3940 info.deterministic = false; 3941 return next.study(info); 3942 } else { 3943 atom.study(info); 3944 return next.study(info); 3945 } 3946 } 3947 } 3948 3949 /** 3950 * Handles the curly-brace style repetition with a specified minimum and 3951 * maximum occurrences. The * quantifier is handled as a special case. 3952 * This class handles the three types. 3953 */ 3954 static final class Curly extends Node { 3955 Node atom; 3956 int type; 3957 int cmin; 3958 int cmax; 3959 3960 Curly(Node node, int cmin, int cmax, int type) { 3961 this.atom = node; 3962 this.type = type; 3963 this.cmin = cmin; 3964 this.cmax = cmax; 3965 } 3966 boolean match(Matcher matcher, int i, CharSequence seq) { 3967 int j; 3968 for (j = 0; j < cmin; j++) { 3969 if (atom.match(matcher, i, seq)) { 3970 i = matcher.last; 3971 continue; 3972 } 3973 return false; 3974 } 3975 if (type == GREEDY) 3976 return match0(matcher, i, j, seq); 3977 else if (type == LAZY) 3978 return match1(matcher, i, j, seq); 3979 else 3980 return match2(matcher, i, j, seq); 3981 } 3982 // Greedy match. 3983 // i is the index to start matching at 3984 // j is the number of atoms that have matched 3985 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 3986 if (j >= cmax) { 3987 // We have matched the maximum... continue with the rest of 3988 // the regular expression 3989 return next.match(matcher, i, seq); 3990 } 3991 int backLimit = j; 3992 while (atom.match(matcher, i, seq)) { 3993 // k is the length of this match 3994 int k = matcher.last - i; 3995 if (k == 0) // Zero length match 3996 break; 3997 // Move up index and number matched 3998 i = matcher.last; 3999 j++; 4000 // We are greedy so match as many as we can 4001 while (j < cmax) { 4002 if (!atom.match(matcher, i, seq)) 4003 break; 4004 if (i + k != matcher.last) { 4005 if (match0(matcher, matcher.last, j+1, seq)) 4006 return true; 4007 break; 4008 } 4009 i += k; 4010 j++; 4011 } 4012 // Handle backing off if match fails 4013 while (j >= backLimit) { 4014 if (next.match(matcher, i, seq)) 4015 return true; 4016 i -= k; 4017 j--; 4018 } 4019 return false; 4020 } 4021 return next.match(matcher, i, seq); 4022 } 4023 // Reluctant match. At this point, the minimum has been satisfied. 4024 // i is the index to start matching at 4025 // j is the number of atoms that have matched 4026 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4027 for (;;) { 4028 // Try finishing match without consuming any more 4029 if (next.match(matcher, i, seq)) 4030 return true; 4031 // At the maximum, no match found 4032 if (j >= cmax) 4033 return false; 4034 // Okay, must try one more atom 4035 if (!atom.match(matcher, i, seq)) 4036 return false; 4037 // If we haven't moved forward then must break out 4038 if (i == matcher.last) 4039 return false; 4040 // Move up index and number matched 4041 i = matcher.last; 4042 j++; 4043 } 4044 } 4045 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4046 for (; j < cmax; j++) { 4047 if (!atom.match(matcher, i, seq)) 4048 break; 4049 if (i == matcher.last) 4050 break; 4051 i = matcher.last; 4052 } 4053 return next.match(matcher, i, seq); 4054 } 4055 boolean study(TreeInfo info) { 4056 // Save original info 4057 int minL = info.minLength; 4058 int maxL = info.maxLength; 4059 boolean maxV = info.maxValid; 4060 boolean detm = info.deterministic; 4061 info.reset(); 4062 4063 atom.study(info); 4064 4065 int temp = info.minLength * cmin + minL; 4066 if (temp < minL) { 4067 temp = 0xFFFFFFF; // arbitrary large number 4068 } 4069 info.minLength = temp; 4070 4071 if (maxV & info.maxValid) { 4072 temp = info.maxLength * cmax + maxL; 4073 info.maxLength = temp; 4074 if (temp < maxL) { 4075 info.maxValid = false; 4076 } 4077 } else { 4078 info.maxValid = false; 4079 } 4080 4081 if (info.deterministic && cmin == cmax) 4082 info.deterministic = detm; 4083 else 4084 info.deterministic = false; 4085 4086 return next.study(info); 4087 } 4088 } 4089 4090 /** 4091 * Handles the curly-brace style repetition with a specified minimum and 4092 * maximum occurrences in deterministic cases. This is an iterative 4093 * optimization over the Prolog and Loop system which would handle this 4094 * in a recursive way. The * quantifier is handled as a special case. 4095 * If capture is true then this class saves group settings and ensures 4096 * that groups are unset when backing off of a group match. 4097 */ 4098 static final class GroupCurly extends Node { 4099 Node atom; 4100 int type; 4101 int cmin; 4102 int cmax; 4103 int localIndex; 4104 int groupIndex; 4105 boolean capture; 4106 4107 GroupCurly(Node node, int cmin, int cmax, int type, int local, 4108 int group, boolean capture) { 4109 this.atom = node; 4110 this.type = type; 4111 this.cmin = cmin; 4112 this.cmax = cmax; 4113 this.localIndex = local; 4114 this.groupIndex = group; 4115 this.capture = capture; 4116 } 4117 boolean match(Matcher matcher, int i, CharSequence seq) { 4118 int[] groups = matcher.groups; 4119 int[] locals = matcher.locals; 4120 int save0 = locals[localIndex]; 4121 int save1 = 0; 4122 int save2 = 0; 4123 4124 if (capture) { 4125 save1 = groups[groupIndex]; 4126 save2 = groups[groupIndex+1]; 4127 } 4128 4129 // Notify GroupTail there is no need to setup group info 4130 // because it will be set here 4131 locals[localIndex] = -1; 4132 4133 boolean ret = true; 4134 for (int j = 0; j < cmin; j++) { 4135 if (atom.match(matcher, i, seq)) { 4136 if (capture) { 4137 groups[groupIndex] = i; 4138 groups[groupIndex+1] = matcher.last; 4139 } 4140 i = matcher.last; 4141 } else { 4142 ret = false; 4143 break; 4144 } 4145 } 4146 if (ret) { 4147 if (type == GREEDY) { 4148 ret = match0(matcher, i, cmin, seq); 4149 } else if (type == LAZY) { 4150 ret = match1(matcher, i, cmin, seq); 4151 } else { 4152 ret = match2(matcher, i, cmin, seq); 4153 } 4154 } 4155 if (!ret) { 4156 locals[localIndex] = save0; 4157 if (capture) { 4158 groups[groupIndex] = save1; 4159 groups[groupIndex+1] = save2; 4160 } 4161 } 4162 return ret; 4163 } 4164 // Aggressive group match 4165 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4166 int[] groups = matcher.groups; 4167 int save0 = 0; 4168 int save1 = 0; 4169 if (capture) { 4170 save0 = groups[groupIndex]; 4171 save1 = groups[groupIndex+1]; 4172 } 4173 for (;;) { 4174 if (j >= cmax) 4175 break; 4176 if (!atom.match(matcher, i, seq)) 4177 break; 4178 int k = matcher.last - i; 4179 if (k <= 0) { 4180 if (capture) { 4181 groups[groupIndex] = i; 4182 groups[groupIndex+1] = i + k; 4183 } 4184 i = i + k; 4185 break; 4186 } 4187 for (;;) { 4188 if (capture) { 4189 groups[groupIndex] = i; 4190 groups[groupIndex+1] = i + k; 4191 } 4192 i = i + k; 4193 if (++j >= cmax) 4194 break; 4195 if (!atom.match(matcher, i, seq)) 4196 break; 4197 if (i + k != matcher.last) { 4198 if (match0(matcher, i, j, seq)) 4199 return true; 4200 break; 4201 } 4202 } 4203 while (j > cmin) { 4204 if (next.match(matcher, i, seq)) { 4205 if (capture) { 4206 groups[groupIndex+1] = i; 4207 groups[groupIndex] = i - k; 4208 } 4209 i = i - k; 4210 return true; 4211 } 4212 // backing off 4213 if (capture) { 4214 groups[groupIndex+1] = i; 4215 groups[groupIndex] = i - k; 4216 } 4217 i = i - k; 4218 j--; 4219 } 4220 break; 4221 } 4222 if (capture) { 4223 groups[groupIndex] = save0; 4224 groups[groupIndex+1] = save1; 4225 } 4226 return next.match(matcher, i, seq); 4227 } 4228 // Reluctant matching 4229 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4230 for (;;) { 4231 if (next.match(matcher, i, seq)) 4232 return true; 4233 if (j >= cmax) 4234 return false; 4235 if (!atom.match(matcher, i, seq)) 4236 return false; 4237 if (i == matcher.last) 4238 return false; 4239 if (capture) { 4240 matcher.groups[groupIndex] = i; 4241 matcher.groups[groupIndex+1] = matcher.last; 4242 } 4243 i = matcher.last; 4244 j++; 4245 } 4246 } 4247 // Possessive matching 4248 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4249 for (; j < cmax; j++) { 4250 if (!atom.match(matcher, i, seq)) { 4251 break; 4252 } 4253 if (capture) { 4254 matcher.groups[groupIndex] = i; 4255 matcher.groups[groupIndex+1] = matcher.last; 4256 } 4257 if (i == matcher.last) { 4258 break; 4259 } 4260 i = matcher.last; 4261 } 4262 return next.match(matcher, i, seq); 4263 } 4264 boolean study(TreeInfo info) { 4265 // Save original info 4266 int minL = info.minLength; 4267 int maxL = info.maxLength; 4268 boolean maxV = info.maxValid; 4269 boolean detm = info.deterministic; 4270 info.reset(); 4271 4272 atom.study(info); 4273 4274 int temp = info.minLength * cmin + minL; 4275 if (temp < minL) { 4276 temp = 0xFFFFFFF; // Arbitrary large number 4277 } 4278 info.minLength = temp; 4279 4280 if (maxV & info.maxValid) { 4281 temp = info.maxLength * cmax + maxL; 4282 info.maxLength = temp; 4283 if (temp < maxL) { 4284 info.maxValid = false; 4285 } 4286 } else { 4287 info.maxValid = false; 4288 } 4289 4290 if (info.deterministic && cmin == cmax) { 4291 info.deterministic = detm; 4292 } else { 4293 info.deterministic = false; 4294 } 4295 4296 return next.study(info); 4297 } 4298 } 4299 4300 /** 4301 * A Guard node at the end of each atom node in a Branch. It 4302 * serves the purpose of chaining the "match" operation to 4303 * "next" but not the "study", so we can collect the TreeInfo 4304 * of each atom node without including the TreeInfo of the 4305 * "next". 4306 */ 4307 static final class BranchConn extends Node { 4308 BranchConn() {}; 4309 boolean match(Matcher matcher, int i, CharSequence seq) { 4310 return next.match(matcher, i, seq); 4311 } 4312 boolean study(TreeInfo info) { 4313 return info.deterministic; 4314 } 4315 } 4316 4317 /** 4318 * Handles the branching of alternations. Note this is also used for 4319 * the ? quantifier to branch between the case where it matches once 4320 * and where it does not occur. 4321 */ 4322 static final class Branch extends Node { 4323 Node[] atoms = new Node[2]; 4324 int size = 2; 4325 Node conn; 4326 Branch(Node first, Node second, Node branchConn) { 4327 conn = branchConn; 4328 atoms[0] = first; 4329 atoms[1] = second; 4330 } 4331 4332 void add(Node node) { 4333 if (size >= atoms.length) { 4334 Node[] tmp = new Node[atoms.length*2]; 4335 System.arraycopy(atoms, 0, tmp, 0, atoms.length); 4336 atoms = tmp; 4337 } 4338 atoms[size++] = node; 4339 } 4340 4341 boolean match(Matcher matcher, int i, CharSequence seq) { 4342 for (int n = 0; n < size; n++) { 4343 if (atoms[n] == null) { 4344 if (conn.next.match(matcher, i, seq)) 4345 return true; 4346 } else if (atoms[n].match(matcher, i, seq)) { 4347 return true; 4348 } 4349 } 4350 return false; 4351 } 4352 4353 boolean study(TreeInfo info) { 4354 int minL = info.minLength; 4355 int maxL = info.maxLength; 4356 boolean maxV = info.maxValid; 4357 4358 int minL2 = Integer.MAX_VALUE; //arbitrary large enough num 4359 int maxL2 = -1; 4360 for (int n = 0; n < size; n++) { 4361 info.reset(); 4362 if (atoms[n] != null) 4363 atoms[n].study(info); 4364 minL2 = Math.min(minL2, info.minLength); 4365 maxL2 = Math.max(maxL2, info.maxLength); 4366 maxV = (maxV & info.maxValid); 4367 } 4368 4369 minL += minL2; 4370 maxL += maxL2; 4371 4372 info.reset(); 4373 conn.next.study(info); 4374 4375 info.minLength += minL; 4376 info.maxLength += maxL; 4377 info.maxValid &= maxV; 4378 info.deterministic = false; 4379 return false; 4380 } 4381 } 4382 4383 /** 4384 * The GroupHead saves the location where the group begins in the locals 4385 * and restores them when the match is done. 4386 * 4387 * The matchRef is used when a reference to this group is accessed later 4388 * in the expression. The locals will have a negative value in them to 4389 * indicate that we do not want to unset the group if the reference 4390 * doesn't match. 4391 */ 4392 static final class GroupHead extends Node { 4393 int localIndex; 4394 GroupHead(int localCount) { 4395 localIndex = localCount; 4396 } 4397 boolean match(Matcher matcher, int i, CharSequence seq) { 4398 int save = matcher.locals[localIndex]; 4399 matcher.locals[localIndex] = i; 4400 boolean ret = next.match(matcher, i, seq); 4401 matcher.locals[localIndex] = save; 4402 return ret; 4403 } 4404 boolean matchRef(Matcher matcher, int i, CharSequence seq) { 4405 int save = matcher.locals[localIndex]; 4406 matcher.locals[localIndex] = ~i; // HACK 4407 boolean ret = next.match(matcher, i, seq); 4408 matcher.locals[localIndex] = save; 4409 return ret; 4410 } 4411 } 4412 4413 /** 4414 * Recursive reference to a group in the regular expression. It calls 4415 * matchRef because if the reference fails to match we would not unset 4416 * the group. 4417 */ 4418 static final class GroupRef extends Node { 4419 GroupHead head; 4420 GroupRef(GroupHead head) { 4421 this.head = head; 4422 } 4423 boolean match(Matcher matcher, int i, CharSequence seq) { 4424 return head.matchRef(matcher, i, seq) 4425 && next.match(matcher, matcher.last, seq); 4426 } 4427 boolean study(TreeInfo info) { 4428 info.maxValid = false; 4429 info.deterministic = false; 4430 return next.study(info); 4431 } 4432 } 4433 4434 /** 4435 * The GroupTail handles the setting of group beginning and ending 4436 * locations when groups are successfully matched. It must also be able to 4437 * unset groups that have to be backed off of. 4438 * 4439 * The GroupTail node is also used when a previous group is referenced, 4440 * and in that case no group information needs to be set. 4441 */ 4442 static final class GroupTail extends Node { 4443 int localIndex; 4444 int groupIndex; 4445 GroupTail(int localCount, int groupCount) { 4446 localIndex = localCount; 4447 groupIndex = groupCount + groupCount; 4448 } 4449 boolean match(Matcher matcher, int i, CharSequence seq) { 4450 int tmp = matcher.locals[localIndex]; 4451 if (tmp >= 0) { // This is the normal group case. 4452 // Save the group so we can unset it if it 4453 // backs off of a match. 4454 int groupStart = matcher.groups[groupIndex]; 4455 int groupEnd = matcher.groups[groupIndex+1]; 4456 4457 matcher.groups[groupIndex] = tmp; 4458 matcher.groups[groupIndex+1] = i; 4459 if (next.match(matcher, i, seq)) { 4460 return true; 4461 } 4462 matcher.groups[groupIndex] = groupStart; 4463 matcher.groups[groupIndex+1] = groupEnd; 4464 return false; 4465 } else { 4466 // This is a group reference case. We don't need to save any 4467 // group info because it isn't really a group. 4468 matcher.last = i; 4469 return true; 4470 } 4471 } 4472 } 4473 4474 /** 4475 * This sets up a loop to handle a recursive quantifier structure. 4476 */ 4477 static final class Prolog extends Node { 4478 Loop loop; 4479 Prolog(Loop loop) { 4480 this.loop = loop; 4481 } 4482 boolean match(Matcher matcher, int i, CharSequence seq) { 4483 return loop.matchInit(matcher, i, seq); 4484 } 4485 boolean study(TreeInfo info) { 4486 return loop.study(info); 4487 } 4488 } 4489 4490 /** 4491 * Handles the repetition count for a greedy Curly. The matchInit 4492 * is called from the Prolog to save the index of where the group 4493 * beginning is stored. A zero length group check occurs in the 4494 * normal match but is skipped in the matchInit. 4495 */ 4496 static class Loop extends Node { 4497 Node body; 4498 int countIndex; // local count index in matcher locals 4499 int beginIndex; // group beginning index 4500 int cmin, cmax; 4501 Loop(int countIndex, int beginIndex) { 4502 this.countIndex = countIndex; 4503 this.beginIndex = beginIndex; 4504 } 4505 boolean match(Matcher matcher, int i, CharSequence seq) { 4506 // Avoid infinite loop in zero-length case. 4507 if (i > matcher.locals[beginIndex]) { 4508 int count = matcher.locals[countIndex]; 4509 4510 // This block is for before we reach the minimum 4511 // iterations required for the loop to match 4512 if (count < cmin) { 4513 matcher.locals[countIndex] = count + 1; 4514 boolean b = body.match(matcher, i, seq); 4515 // If match failed we must backtrack, so 4516 // the loop count should NOT be incremented 4517 if (!b) 4518 matcher.locals[countIndex] = count; 4519 // Return success or failure since we are under 4520 // minimum 4521 return b; 4522 } 4523 // This block is for after we have the minimum 4524 // iterations required for the loop to match 4525 if (count < cmax) { 4526 matcher.locals[countIndex] = count + 1; 4527 boolean b = body.match(matcher, i, seq); 4528 // If match failed we must backtrack, so 4529 // the loop count should NOT be incremented 4530 if (!b) 4531 matcher.locals[countIndex] = count; 4532 else 4533 return true; 4534 } 4535 } 4536 return next.match(matcher, i, seq); 4537 } 4538 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4539 int save = matcher.locals[countIndex]; 4540 boolean ret = false; 4541 if (0 < cmin) { 4542 matcher.locals[countIndex] = 1; 4543 ret = body.match(matcher, i, seq); 4544 } else if (0 < cmax) { 4545 matcher.locals[countIndex] = 1; 4546 ret = body.match(matcher, i, seq); 4547 if (ret == false) 4548 ret = next.match(matcher, i, seq); 4549 } else { 4550 ret = next.match(matcher, i, seq); 4551 } 4552 matcher.locals[countIndex] = save; 4553 return ret; 4554 } 4555 boolean study(TreeInfo info) { 4556 info.maxValid = false; 4557 info.deterministic = false; 4558 return false; 4559 } 4560 } 4561 4562 /** 4563 * Handles the repetition count for a reluctant Curly. The matchInit 4564 * is called from the Prolog to save the index of where the group 4565 * beginning is stored. A zero length group check occurs in the 4566 * normal match but is skipped in the matchInit. 4567 */ 4568 static final class LazyLoop extends Loop { 4569 LazyLoop(int countIndex, int beginIndex) { 4570 super(countIndex, beginIndex); 4571 } 4572 boolean match(Matcher matcher, int i, CharSequence seq) { 4573 // Check for zero length group 4574 if (i > matcher.locals[beginIndex]) { 4575 int count = matcher.locals[countIndex]; 4576 if (count < cmin) { 4577 matcher.locals[countIndex] = count + 1; 4578 boolean result = body.match(matcher, i, seq); 4579 // If match failed we must backtrack, so 4580 // the loop count should NOT be incremented 4581 if (!result) 4582 matcher.locals[countIndex] = count; 4583 return result; 4584 } 4585 if (next.match(matcher, i, seq)) 4586 return true; 4587 if (count < cmax) { 4588 matcher.locals[countIndex] = count + 1; 4589 boolean result = body.match(matcher, i, seq); 4590 // If match failed we must backtrack, so 4591 // the loop count should NOT be incremented 4592 if (!result) 4593 matcher.locals[countIndex] = count; 4594 return result; 4595 } 4596 return false; 4597 } 4598 return next.match(matcher, i, seq); 4599 } 4600 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4601 int save = matcher.locals[countIndex]; 4602 boolean ret = false; 4603 if (0 < cmin) { 4604 matcher.locals[countIndex] = 1; 4605 ret = body.match(matcher, i, seq); 4606 } else if (next.match(matcher, i, seq)) { 4607 ret = true; 4608 } else if (0 < cmax) { 4609 matcher.locals[countIndex] = 1; 4610 ret = body.match(matcher, i, seq); 4611 } 4612 matcher.locals[countIndex] = save; 4613 return ret; 4614 } 4615 boolean study(TreeInfo info) { 4616 info.maxValid = false; 4617 info.deterministic = false; 4618 return false; 4619 } 4620 } 4621 4622 /** 4623 * Refers to a group in the regular expression. Attempts to match 4624 * whatever the group referred to last matched. 4625 */ 4626 static class BackRef extends Node { 4627 int groupIndex; 4628 BackRef(int groupCount) { 4629 super(); 4630 groupIndex = groupCount + groupCount; 4631 } 4632 boolean match(Matcher matcher, int i, CharSequence seq) { 4633 int j = matcher.groups[groupIndex]; 4634 int k = matcher.groups[groupIndex+1]; 4635 4636 int groupSize = k - j; 4637 4638 // If the referenced group didn't match, neither can this 4639 if (j < 0) 4640 return false; 4641 4642 // If there isn't enough input left no match 4643 if (i + groupSize > matcher.to) { 4644 matcher.hitEnd = true; 4645 return false; 4646 } 4647 4648 // Check each new char to make sure it matches what the group 4649 // referenced matched last time around 4650 for (int index=0; index<groupSize; index++) 4651 if (seq.charAt(i+index) != seq.charAt(j+index)) 4652 return false; 4653 4654 return next.match(matcher, i+groupSize, seq); 4655 } 4656 boolean study(TreeInfo info) { 4657 info.maxValid = false; 4658 return next.study(info); 4659 } 4660 } 4661 4662 static class CIBackRef extends Node { 4663 int groupIndex; 4664 boolean doUnicodeCase; 4665 CIBackRef(int groupCount, boolean doUnicodeCase) { 4666 super(); 4667 groupIndex = groupCount + groupCount; 4668 this.doUnicodeCase = doUnicodeCase; 4669 } 4670 boolean match(Matcher matcher, int i, CharSequence seq) { 4671 int j = matcher.groups[groupIndex]; 4672 int k = matcher.groups[groupIndex+1]; 4673 4674 int groupSize = k - j; 4675 4676 // If the referenced group didn't match, neither can this 4677 if (j < 0) 4678 return false; 4679 4680 // If there isn't enough input left no match 4681 if (i + groupSize > matcher.to) { 4682 matcher.hitEnd = true; 4683 return false; 4684 } 4685 4686 // Check each new char to make sure it matches what the group 4687 // referenced matched last time around 4688 int x = i; 4689 for (int index=0; index<groupSize; index++) { 4690 int c1 = Character.codePointAt(seq, x); 4691 int c2 = Character.codePointAt(seq, j); 4692 if (c1 != c2) { 4693 if (doUnicodeCase) { 4694 int cc1 = Character.toUpperCase(c1); 4695 int cc2 = Character.toUpperCase(c2); 4696 if (cc1 != cc2 && 4697 Character.toLowerCase(cc1) != 4698 Character.toLowerCase(cc2)) 4699 return false; 4700 } else { 4701 if (ASCII.toLower(c1) != ASCII.toLower(c2)) 4702 return false; 4703 } 4704 } 4705 x += Character.charCount(c1); 4706 j += Character.charCount(c2); 4707 } 4708 4709 return next.match(matcher, i+groupSize, seq); 4710 } 4711 boolean study(TreeInfo info) { 4712 info.maxValid = false; 4713 return next.study(info); 4714 } 4715 } 4716 4717 /** 4718 * Searches until the next instance of its atom. This is useful for 4719 * finding the atom efficiently without passing an instance of it 4720 * (greedy problem) and without a lot of wasted search time (reluctant 4721 * problem). 4722 */ 4723 static final class First extends Node { 4724 Node atom; 4725 First(Node node) { 4726 this.atom = BnM.optimize(node); 4727 } 4728 boolean match(Matcher matcher, int i, CharSequence seq) { 4729 if (atom instanceof BnM) { 4730 return atom.match(matcher, i, seq) 4731 && next.match(matcher, matcher.last, seq); 4732 } 4733 for (;;) { 4734 if (i > matcher.to) { 4735 matcher.hitEnd = true; 4736 return false; 4737 } 4738 if (atom.match(matcher, i, seq)) { 4739 return next.match(matcher, matcher.last, seq); 4740 } 4741 i += countChars(seq, i, 1); 4742 matcher.first++; 4743 } 4744 } 4745 boolean study(TreeInfo info) { 4746 atom.study(info); 4747 info.maxValid = false; 4748 info.deterministic = false; 4749 return next.study(info); 4750 } 4751 } 4752 4753 static final class Conditional extends Node { 4754 Node cond, yes, not; 4755 Conditional(Node cond, Node yes, Node not) { 4756 this.cond = cond; 4757 this.yes = yes; 4758 this.not = not; 4759 } 4760 boolean match(Matcher matcher, int i, CharSequence seq) { 4761 if (cond.match(matcher, i, seq)) { 4762 return yes.match(matcher, i, seq); 4763 } else { 4764 return not.match(matcher, i, seq); 4765 } 4766 } 4767 boolean study(TreeInfo info) { 4768 int minL = info.minLength; 4769 int maxL = info.maxLength; 4770 boolean maxV = info.maxValid; 4771 info.reset(); 4772 yes.study(info); 4773 4774 int minL2 = info.minLength; 4775 int maxL2 = info.maxLength; 4776 boolean maxV2 = info.maxValid; 4777 info.reset(); 4778 not.study(info); 4779 4780 info.minLength = minL + Math.min(minL2, info.minLength); 4781 info.maxLength = maxL + Math.max(maxL2, info.maxLength); 4782 info.maxValid = (maxV & maxV2 & info.maxValid); 4783 info.deterministic = false; 4784 return next.study(info); 4785 } 4786 } 4787 4788 /** 4789 * Zero width positive lookahead. 4790 */ 4791 static final class Pos extends Node { 4792 Node cond; 4793 Pos(Node cond) { 4794 this.cond = cond; 4795 } 4796 boolean match(Matcher matcher, int i, CharSequence seq) { 4797 int savedTo = matcher.to; 4798 boolean conditionMatched = false; 4799 4800 // Relax transparent region boundaries for lookahead 4801 if (matcher.transparentBounds) 4802 matcher.to = matcher.getTextLength(); 4803 try { 4804 conditionMatched = cond.match(matcher, i, seq); 4805 } finally { 4806 // Reinstate region boundaries 4807 matcher.to = savedTo; 4808 } 4809 return conditionMatched && next.match(matcher, i, seq); 4810 } 4811 } 4812 4813 /** 4814 * Zero width negative lookahead. 4815 */ 4816 static final class Neg extends Node { 4817 Node cond; 4818 Neg(Node cond) { 4819 this.cond = cond; 4820 } 4821 boolean match(Matcher matcher, int i, CharSequence seq) { 4822 int savedTo = matcher.to; 4823 boolean conditionMatched = false; 4824 4825 // Relax transparent region boundaries for lookahead 4826 if (matcher.transparentBounds) 4827 matcher.to = matcher.getTextLength(); 4828 try { 4829 if (i < matcher.to) { 4830 conditionMatched = !cond.match(matcher, i, seq); 4831 } else { 4832 // If a negative lookahead succeeds then more input 4833 // could cause it to fail! 4834 matcher.requireEnd = true; 4835 conditionMatched = !cond.match(matcher, i, seq); 4836 } 4837 } finally { 4838 // Reinstate region boundaries 4839 matcher.to = savedTo; 4840 } 4841 return conditionMatched && next.match(matcher, i, seq); 4842 } 4843 } 4844 4845 /** 4846 * For use with lookbehinds; matches the position where the lookbehind 4847 * was encountered. 4848 */ 4849 static Node lookbehindEnd = new Node() { 4850 boolean match(Matcher matcher, int i, CharSequence seq) { 4851 return i == matcher.lookbehindTo; 4852 } 4853 }; 4854 4855 /** 4856 * Zero width positive lookbehind. 4857 */ 4858 static class Behind extends Node { 4859 Node cond; 4860 int rmax, rmin; 4861 Behind(Node cond, int rmax, int rmin) { 4862 this.cond = cond; 4863 this.rmax = rmax; 4864 this.rmin = rmin; 4865 } 4866 4867 boolean match(Matcher matcher, int i, CharSequence seq) { 4868 int savedFrom = matcher.from; 4869 boolean conditionMatched = false; 4870 int startIndex = (!matcher.transparentBounds) ? 4871 matcher.from : 0; 4872 int from = Math.max(i - rmax, startIndex); 4873 // Set end boundary 4874 int savedLBT = matcher.lookbehindTo; 4875 matcher.lookbehindTo = i; 4876 // Relax transparent region boundaries for lookbehind 4877 if (matcher.transparentBounds) 4878 matcher.from = 0; 4879 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 4880 conditionMatched = cond.match(matcher, j, seq); 4881 } 4882 matcher.from = savedFrom; 4883 matcher.lookbehindTo = savedLBT; 4884 return conditionMatched && next.match(matcher, i, seq); 4885 } 4886 } 4887 4888 /** 4889 * Zero width positive lookbehind, including supplementary 4890 * characters or unpaired surrogates. 4891 */ 4892 static final class BehindS extends Behind { 4893 BehindS(Node cond, int rmax, int rmin) { 4894 super(cond, rmax, rmin); 4895 } 4896 boolean match(Matcher matcher, int i, CharSequence seq) { 4897 int rmaxChars = countChars(seq, i, -rmax); 4898 int rminChars = countChars(seq, i, -rmin); 4899 int savedFrom = matcher.from; 4900 int startIndex = (!matcher.transparentBounds) ? 4901 matcher.from : 0; 4902 boolean conditionMatched = false; 4903 int from = Math.max(i - rmaxChars, startIndex); 4904 // Set end boundary 4905 int savedLBT = matcher.lookbehindTo; 4906 matcher.lookbehindTo = i; 4907 // Relax transparent region boundaries for lookbehind 4908 if (matcher.transparentBounds) 4909 matcher.from = 0; 4910 4911 for (int j = i - rminChars; 4912 !conditionMatched && j >= from; 4913 j -= j>from ? countChars(seq, j, -1) : 1) { 4914 conditionMatched = cond.match(matcher, j, seq); 4915 } 4916 matcher.from = savedFrom; 4917 matcher.lookbehindTo = savedLBT; 4918 return conditionMatched && next.match(matcher, i, seq); 4919 } 4920 } 4921 4922 /** 4923 * Zero width negative lookbehind. 4924 */ 4925 static class NotBehind extends Node { 4926 Node cond; 4927 int rmax, rmin; 4928 NotBehind(Node cond, int rmax, int rmin) { 4929 this.cond = cond; 4930 this.rmax = rmax; 4931 this.rmin = rmin; 4932 } 4933 4934 boolean match(Matcher matcher, int i, CharSequence seq) { 4935 int savedLBT = matcher.lookbehindTo; 4936 int savedFrom = matcher.from; 4937 boolean conditionMatched = false; 4938 int startIndex = (!matcher.transparentBounds) ? 4939 matcher.from : 0; 4940 int from = Math.max(i - rmax, startIndex); 4941 matcher.lookbehindTo = i; 4942 // Relax transparent region boundaries for lookbehind 4943 if (matcher.transparentBounds) 4944 matcher.from = 0; 4945 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 4946 conditionMatched = cond.match(matcher, j, seq); 4947 } 4948 // Reinstate region boundaries 4949 matcher.from = savedFrom; 4950 matcher.lookbehindTo = savedLBT; 4951 return !conditionMatched && next.match(matcher, i, seq); 4952 } 4953 } 4954 4955 /** 4956 * Zero width negative lookbehind, including supplementary 4957 * characters or unpaired surrogates. 4958 */ 4959 static final class NotBehindS extends NotBehind { 4960 NotBehindS(Node cond, int rmax, int rmin) { 4961 super(cond, rmax, rmin); 4962 } 4963 boolean match(Matcher matcher, int i, CharSequence seq) { 4964 int rmaxChars = countChars(seq, i, -rmax); 4965 int rminChars = countChars(seq, i, -rmin); 4966 int savedFrom = matcher.from; 4967 int savedLBT = matcher.lookbehindTo; 4968 boolean conditionMatched = false; 4969 int startIndex = (!matcher.transparentBounds) ? 4970 matcher.from : 0; 4971 int from = Math.max(i - rmaxChars, startIndex); 4972 matcher.lookbehindTo = i; 4973 // Relax transparent region boundaries for lookbehind 4974 if (matcher.transparentBounds) 4975 matcher.from = 0; 4976 for (int j = i - rminChars; 4977 !conditionMatched && j >= from; 4978 j -= j>from ? countChars(seq, j, -1) : 1) { 4979 conditionMatched = cond.match(matcher, j, seq); 4980 } 4981 //Reinstate region boundaries 4982 matcher.from = savedFrom; 4983 matcher.lookbehindTo = savedLBT; 4984 return !conditionMatched && next.match(matcher, i, seq); 4985 } 4986 } 4987 4988 /** 4989 * Returns the set union of two CharProperty nodes. 4990 */ 4991 private static CharProperty union(final CharProperty lhs, 4992 final CharProperty rhs) { 4993 return new CharProperty() { 4994 boolean isSatisfiedBy(int ch) { 4995 return lhs.isSatisfiedBy(ch) || rhs.isSatisfiedBy(ch);}}; 4996 } 4997 4998 /** 4999 * Returns the set intersection of two CharProperty nodes. 5000 */ 5001 private static CharProperty intersection(final CharProperty lhs, 5002 final CharProperty rhs) { 5003 return new CharProperty() { 5004 boolean isSatisfiedBy(int ch) { 5005 return lhs.isSatisfiedBy(ch) && rhs.isSatisfiedBy(ch);}}; 5006 } 5007 5008 /** 5009 * Returns the set difference of two CharProperty nodes. 5010 */ 5011 private static CharProperty setDifference(final CharProperty lhs, 5012 final CharProperty rhs) { 5013 return new CharProperty() { 5014 boolean isSatisfiedBy(int ch) { 5015 return ! rhs.isSatisfiedBy(ch) && lhs.isSatisfiedBy(ch);}}; 5016 } 5017 5018 /** 5019 * Handles word boundaries. Includes a field to allow this one class to 5020 * deal with the different types of word boundaries we can match. The word 5021 * characters include underscores, letters, and digits. Non spacing marks 5022 * can are also part of a word if they have a base character, otherwise 5023 * they are ignored for purposes of finding word boundaries. 5024 */ 5025 static final class Bound extends Node { 5026 static int LEFT = 0x1; 5027 static int RIGHT= 0x2; 5028 static int BOTH = 0x3; 5029 static int NONE = 0x4; 5030 int type; 5031 Bound(int n) { 5032 type = n; 5033 } 5034 int check(Matcher matcher, int i, CharSequence seq) { 5035 int ch; 5036 boolean left = false; 5037 int startIndex = matcher.from; 5038 int endIndex = matcher.to; 5039 if (matcher.transparentBounds) { 5040 startIndex = 0; 5041 endIndex = matcher.getTextLength(); 5042 } 5043 if (i > startIndex) { 5044 ch = Character.codePointBefore(seq, i); 5045 left = (ch == '_' || Character.isLetterOrDigit(ch) || 5046 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5047 && hasBaseCharacter(matcher, i-1, seq))); 5048 } 5049 boolean right = false; 5050 if (i < endIndex) { 5051 ch = Character.codePointAt(seq, i); 5052 right = (ch == '_' || Character.isLetterOrDigit(ch) || 5053 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5054 && hasBaseCharacter(matcher, i, seq))); 5055 } else { 5056 // Tried to access char past the end 5057 matcher.hitEnd = true; 5058 // The addition of another char could wreck a boundary 5059 matcher.requireEnd = true; 5060 } 5061 return ((left ^ right) ? (right ? LEFT : RIGHT) : NONE); 5062 } 5063 boolean match(Matcher matcher, int i, CharSequence seq) { 5064 return (check(matcher, i, seq) & type) > 0 5065 && next.match(matcher, i, seq); 5066 } 5067 } 5068 5069 /** 5070 * Non spacing marks only count as word characters in bounds calculations 5071 * if they have a base character. 5072 */ 5073 private static boolean hasBaseCharacter(Matcher matcher, int i, 5074 CharSequence seq) 5075 { 5076 int start = (!matcher.transparentBounds) ? 5077 matcher.from : 0; 5078 for (int x=i; x >= start; x--) { 5079 int ch = Character.codePointAt(seq, x); 5080 if (Character.isLetterOrDigit(ch)) 5081 return true; 5082 if (Character.getType(ch) == Character.NON_SPACING_MARK) 5083 continue; 5084 return false; 5085 } 5086 return false; 5087 } 5088 5089 /** 5090 * Attempts to match a slice in the input using the Boyer-Moore string 5091 * matching algorithm. The algorithm is based on the idea that the 5092 * pattern can be shifted farther ahead in the search text if it is 5093 * matched right to left. 5094 * <p> 5095 * The pattern is compared to the input one character at a time, from 5096 * the rightmost character in the pattern to the left. If the characters 5097 * all match the pattern has been found. If a character does not match, 5098 * the pattern is shifted right a distance that is the maximum of two 5099 * functions, the bad character shift and the good suffix shift. This 5100 * shift moves the attempted match position through the input more 5101 * quickly than a naive one position at a time check. 5102 * <p> 5103 * The bad character shift is based on the character from the text that 5104 * did not match. If the character does not appear in the pattern, the 5105 * pattern can be shifted completely beyond the bad character. If the 5106 * character does occur in the pattern, the pattern can be shifted to 5107 * line the pattern up with the next occurrence of that character. 5108 * <p> 5109 * The good suffix shift is based on the idea that some subset on the right 5110 * side of the pattern has matched. When a bad character is found, the 5111 * pattern can be shifted right by the pattern length if the subset does 5112 * not occur again in pattern, or by the amount of distance to the 5113 * next occurrence of the subset in the pattern. 5114 * 5115 * Boyer-Moore search methods adapted from code by Amy Yu. 5116 */ 5117 static class BnM extends Node { 5118 int[] buffer; 5119 int[] lastOcc; 5120 int[] optoSft; 5121 5122 /** 5123 * Pre calculates arrays needed to generate the bad character 5124 * shift and the good suffix shift. Only the last seven bits 5125 * are used to see if chars match; This keeps the tables small 5126 * and covers the heavily used ASCII range, but occasionally 5127 * results in an aliased match for the bad character shift. 5128 */ 5129 static Node optimize(Node node) { 5130 if (!(node instanceof Slice)) { 5131 return node; 5132 } 5133 5134 int[] src = ((Slice) node).buffer; 5135 int patternLength = src.length; 5136 // The BM algorithm requires a bit of overhead; 5137 // If the pattern is short don't use it, since 5138 // a shift larger than the pattern length cannot 5139 // be used anyway. 5140 if (patternLength < 4) { 5141 return node; 5142 } 5143 int i, j, k; 5144 int[] lastOcc = new int[128]; 5145 int[] optoSft = new int[patternLength]; 5146 // Precalculate part of the bad character shift 5147 // It is a table for where in the pattern each 5148 // lower 7-bit value occurs 5149 for (i = 0; i < patternLength; i++) { 5150 lastOcc[src[i]&0x7F] = i + 1; 5151 } 5152 // Precalculate the good suffix shift 5153 // i is the shift amount being considered 5154 NEXT: for (i = patternLength; i > 0; i--) { 5155 // j is the beginning index of suffix being considered 5156 for (j = patternLength - 1; j >= i; j--) { 5157 // Testing for good suffix 5158 if (src[j] == src[j-i]) { 5159 // src[j..len] is a good suffix 5160 optoSft[j-1] = i; 5161 } else { 5162 // No match. The array has already been 5163 // filled up with correct values before. 5164 continue NEXT; 5165 } 5166 } 5167 // This fills up the remaining of optoSft 5168 // any suffix can not have larger shift amount 5169 // then its sub-suffix. Why??? 5170 while (j > 0) { 5171 optoSft[--j] = i; 5172 } 5173 } 5174 // Set the guard value because of unicode compression 5175 optoSft[patternLength-1] = 1; 5176 if (node instanceof SliceS) 5177 return new BnMS(src, lastOcc, optoSft, node.next); 5178 return new BnM(src, lastOcc, optoSft, node.next); 5179 } 5180 BnM(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5181 this.buffer = src; 5182 this.lastOcc = lastOcc; 5183 this.optoSft = optoSft; 5184 this.next = next; 5185 } 5186 boolean match(Matcher matcher, int i, CharSequence seq) { 5187 int[] src = buffer; 5188 int patternLength = src.length; 5189 int last = matcher.to - patternLength; 5190 5191 // Loop over all possible match positions in text 5192 NEXT: while (i <= last) { 5193 // Loop over pattern from right to left 5194 for (int j = patternLength - 1; j >= 0; j--) { 5195 int ch = seq.charAt(i+j); 5196 if (ch != src[j]) { 5197 // Shift search to the right by the maximum of the 5198 // bad character shift and the good suffix shift 5199 i += Math.max(j + 1 - lastOcc[ch&0x7F], optoSft[j]); 5200 continue NEXT; 5201 } 5202 } 5203 // Entire pattern matched starting at i 5204 matcher.first = i; 5205 boolean ret = next.match(matcher, i + patternLength, seq); 5206 if (ret) { 5207 matcher.first = i; 5208 matcher.groups[0] = matcher.first; 5209 matcher.groups[1] = matcher.last; 5210 return true; 5211 } 5212 i++; 5213 } 5214 // BnM is only used as the leading node in the unanchored case, 5215 // and it replaced its Start() which always searches to the end 5216 // if it doesn't find what it's looking for, so hitEnd is true. 5217 matcher.hitEnd = true; 5218 return false; 5219 } 5220 boolean study(TreeInfo info) { 5221 info.minLength += buffer.length; 5222 info.maxValid = false; 5223 return next.study(info); 5224 } 5225 } 5226 5227 /** 5228 * Supplementary support version of BnM(). Unpaired surrogates are 5229 * also handled by this class. 5230 */ 5231 static final class BnMS extends BnM { 5232 int lengthInChars; 5233 5234 BnMS(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5235 super(src, lastOcc, optoSft, next); 5236 for (int x = 0; x < buffer.length; x++) { 5237 lengthInChars += Character.charCount(buffer[x]); 5238 } 5239 } 5240 boolean match(Matcher matcher, int i, CharSequence seq) { 5241 int[] src = buffer; 5242 int patternLength = src.length; 5243 int last = matcher.to - lengthInChars; 5244 5245 // Loop over all possible match positions in text 5246 NEXT: while (i <= last) { 5247 // Loop over pattern from right to left 5248 int ch; 5249 for (int j = countChars(seq, i, patternLength), x = patternLength - 1; 5250 j > 0; j -= Character.charCount(ch), x--) { 5251 ch = Character.codePointBefore(seq, i+j); 5252 if (ch != src[x]) { 5253 // Shift search to the right by the maximum of the 5254 // bad character shift and the good suffix shift 5255 int n = Math.max(x + 1 - lastOcc[ch&0x7F], optoSft[x]); 5256 i += countChars(seq, i, n); 5257 continue NEXT; 5258 } 5259 } 5260 // Entire pattern matched starting at i 5261 matcher.first = i; 5262 boolean ret = next.match(matcher, i + lengthInChars, seq); 5263 if (ret) { 5264 matcher.first = i; 5265 matcher.groups[0] = matcher.first; 5266 matcher.groups[1] = matcher.last; 5267 return true; 5268 } 5269 i += countChars(seq, i, 1); 5270 } 5271 matcher.hitEnd = true; 5272 return false; 5273 } 5274 } 5275 5276 /////////////////////////////////////////////////////////////////////////////// 5277 /////////////////////////////////////////////////////////////////////////////// 5278 5279 /** 5280 * This must be the very first initializer. 5281 */ 5282 static Node accept = new Node(); 5283 5284 static Node lastAccept = new LastNode(); 5285 5286 private static class CharPropertyNames { 5287 5288 static CharProperty charPropertyFor(String name) { 5289 CharPropertyFactory m = map.get(name); 5290 return m == null ? null : m.make(); 5291 } 5292 5293 private static abstract class CharPropertyFactory { 5294 abstract CharProperty make(); 5295 } 5296 5297 private static void defCategory(String name, 5298 final int typeMask) { 5299 map.put(name, new CharPropertyFactory() { 5300 CharProperty make() { return new Category(typeMask);}}); 5301 } 5302 5303 private static void defRange(String name, 5304 final int lower, final int upper) { 5305 map.put(name, new CharPropertyFactory() { 5306 CharProperty make() { return rangeFor(lower, upper);}}); 5307 } 5308 5309 private static void defCtype(String name, 5310 final int ctype) { 5311 map.put(name, new CharPropertyFactory() { 5312 CharProperty make() { return new Ctype(ctype);}}); 5313 } 5314 5315 private static abstract class CloneableProperty 5316 extends CharProperty implements Cloneable 5317 { 5318 public CloneableProperty clone() { 5319 try { 5320 return (CloneableProperty) super.clone(); 5321 } catch (CloneNotSupportedException e) { 5322 throw new AssertionError(e); 5323 } 5324 } 5325 } 5326 5327 private static void defClone(String name, 5328 final CloneableProperty p) { 5329 map.put(name, new CharPropertyFactory() { 5330 CharProperty make() { return p.clone();}}); 5331 } 5332 5333 private static final HashMap<String, CharPropertyFactory> map 5334 = new HashMap<>(); 5335 5336 static { 5337 // Unicode character property aliases, defined in 5338 // http://www.unicode.org/Public/UNIDATA/PropertyValueAliases.txt 5339 defCategory("Cn", 1<<Character.UNASSIGNED); 5340 defCategory("Lu", 1<<Character.UPPERCASE_LETTER); 5341 defCategory("Ll", 1<<Character.LOWERCASE_LETTER); 5342 defCategory("Lt", 1<<Character.TITLECASE_LETTER); 5343 defCategory("Lm", 1<<Character.MODIFIER_LETTER); 5344 defCategory("Lo", 1<<Character.OTHER_LETTER); 5345 defCategory("Mn", 1<<Character.NON_SPACING_MARK); 5346 defCategory("Me", 1<<Character.ENCLOSING_MARK); 5347 defCategory("Mc", 1<<Character.COMBINING_SPACING_MARK); 5348 defCategory("Nd", 1<<Character.DECIMAL_DIGIT_NUMBER); 5349 defCategory("Nl", 1<<Character.LETTER_NUMBER); 5350 defCategory("No", 1<<Character.OTHER_NUMBER); 5351 defCategory("Zs", 1<<Character.SPACE_SEPARATOR); 5352 defCategory("Zl", 1<<Character.LINE_SEPARATOR); 5353 defCategory("Zp", 1<<Character.PARAGRAPH_SEPARATOR); 5354 defCategory("Cc", 1<<Character.CONTROL); 5355 defCategory("Cf", 1<<Character.FORMAT); 5356 defCategory("Co", 1<<Character.PRIVATE_USE); 5357 defCategory("Cs", 1<<Character.SURROGATE); 5358 defCategory("Pd", 1<<Character.DASH_PUNCTUATION); 5359 defCategory("Ps", 1<<Character.START_PUNCTUATION); 5360 defCategory("Pe", 1<<Character.END_PUNCTUATION); 5361 defCategory("Pc", 1<<Character.CONNECTOR_PUNCTUATION); 5362 defCategory("Po", 1<<Character.OTHER_PUNCTUATION); 5363 defCategory("Sm", 1<<Character.MATH_SYMBOL); 5364 defCategory("Sc", 1<<Character.CURRENCY_SYMBOL); 5365 defCategory("Sk", 1<<Character.MODIFIER_SYMBOL); 5366 defCategory("So", 1<<Character.OTHER_SYMBOL); 5367 defCategory("Pi", 1<<Character.INITIAL_QUOTE_PUNCTUATION); 5368 defCategory("Pf", 1<<Character.FINAL_QUOTE_PUNCTUATION); 5369 defCategory("L", ((1<<Character.UPPERCASE_LETTER) | 5370 (1<<Character.LOWERCASE_LETTER) | 5371 (1<<Character.TITLECASE_LETTER) | 5372 (1<<Character.MODIFIER_LETTER) | 5373 (1<<Character.OTHER_LETTER))); 5374 defCategory("M", ((1<<Character.NON_SPACING_MARK) | 5375 (1<<Character.ENCLOSING_MARK) | 5376 (1<<Character.COMBINING_SPACING_MARK))); 5377 defCategory("N", ((1<<Character.DECIMAL_DIGIT_NUMBER) | 5378 (1<<Character.LETTER_NUMBER) | 5379 (1<<Character.OTHER_NUMBER))); 5380 defCategory("Z", ((1<<Character.SPACE_SEPARATOR) | 5381 (1<<Character.LINE_SEPARATOR) | 5382 (1<<Character.PARAGRAPH_SEPARATOR))); 5383 defCategory("C", ((1<<Character.CONTROL) | 5384 (1<<Character.FORMAT) | 5385 (1<<Character.PRIVATE_USE) | 5386 (1<<Character.SURROGATE))); // Other 5387 defCategory("P", ((1<<Character.DASH_PUNCTUATION) | 5388 (1<<Character.START_PUNCTUATION) | 5389 (1<<Character.END_PUNCTUATION) | 5390 (1<<Character.CONNECTOR_PUNCTUATION) | 5391 (1<<Character.OTHER_PUNCTUATION) | 5392 (1<<Character.INITIAL_QUOTE_PUNCTUATION) | 5393 (1<<Character.FINAL_QUOTE_PUNCTUATION))); 5394 defCategory("S", ((1<<Character.MATH_SYMBOL) | 5395 (1<<Character.CURRENCY_SYMBOL) | 5396 (1<<Character.MODIFIER_SYMBOL) | 5397 (1<<Character.OTHER_SYMBOL))); 5398 defCategory("LC", ((1<<Character.UPPERCASE_LETTER) | 5399 (1<<Character.LOWERCASE_LETTER) | 5400 (1<<Character.TITLECASE_LETTER))); 5401 defCategory("LD", ((1<<Character.UPPERCASE_LETTER) | 5402 (1<<Character.LOWERCASE_LETTER) | 5403 (1<<Character.TITLECASE_LETTER) | 5404 (1<<Character.MODIFIER_LETTER) | 5405 (1<<Character.OTHER_LETTER) | 5406 (1<<Character.DECIMAL_DIGIT_NUMBER))); 5407 defRange("L1", 0x00, 0xFF); // Latin-1 5408 map.put("all", new CharPropertyFactory() { 5409 CharProperty make() { return new All(); }}); 5410 5411 // Posix regular expression character classes, defined in 5412 // http://www.unix.org/onlinepubs/009695399/basedefs/xbd_chap09.html 5413 defRange("ASCII", 0x00, 0x7F); // ASCII 5414 defCtype("Alnum", ASCII.ALNUM); // Alphanumeric characters 5415 defCtype("Alpha", ASCII.ALPHA); // Alphabetic characters 5416 defCtype("Blank", ASCII.BLANK); // Space and tab characters 5417 defCtype("Cntrl", ASCII.CNTRL); // Control characters 5418 defRange("Digit", '0', '9'); // Numeric characters 5419 defCtype("Graph", ASCII.GRAPH); // printable and visible 5420 defRange("Lower", 'a', 'z'); // Lower-case alphabetic 5421 defRange("Print", 0x20, 0x7E); // Printable characters 5422 defCtype("Punct", ASCII.PUNCT); // Punctuation characters 5423 defCtype("Space", ASCII.SPACE); // Space characters 5424 defRange("Upper", 'A', 'Z'); // Upper-case alphabetic 5425 defCtype("XDigit",ASCII.XDIGIT); // hexadecimal digits 5426 5427 // Java character properties, defined by methods in Character.java 5428 defClone("javaLowerCase", new CloneableProperty() { 5429 boolean isSatisfiedBy(int ch) { 5430 return Character.isLowerCase(ch);}}); 5431 defClone("javaUpperCase", new CloneableProperty() { 5432 boolean isSatisfiedBy(int ch) { 5433 return Character.isUpperCase(ch);}}); 5434 defClone("javaTitleCase", new CloneableProperty() { 5435 boolean isSatisfiedBy(int ch) { 5436 return Character.isTitleCase(ch);}}); 5437 defClone("javaDigit", new CloneableProperty() { 5438 boolean isSatisfiedBy(int ch) { 5439 return Character.isDigit(ch);}}); 5440 defClone("javaDefined", new CloneableProperty() { 5441 boolean isSatisfiedBy(int ch) { 5442 return Character.isDefined(ch);}}); 5443 defClone("javaLetter", new CloneableProperty() { 5444 boolean isSatisfiedBy(int ch) { 5445 return Character.isLetter(ch);}}); 5446 defClone("javaLetterOrDigit", new CloneableProperty() { 5447 boolean isSatisfiedBy(int ch) { 5448 return Character.isLetterOrDigit(ch);}}); 5449 defClone("javaJavaIdentifierStart", new CloneableProperty() { 5450 boolean isSatisfiedBy(int ch) { 5451 return Character.isJavaIdentifierStart(ch);}}); 5452 defClone("javaJavaIdentifierPart", new CloneableProperty() { 5453 boolean isSatisfiedBy(int ch) { 5454 return Character.isJavaIdentifierPart(ch);}}); 5455 defClone("javaUnicodeIdentifierStart", new CloneableProperty() { 5456 boolean isSatisfiedBy(int ch) { 5457 return Character.isUnicodeIdentifierStart(ch);}}); 5458 defClone("javaUnicodeIdentifierPart", new CloneableProperty() { 5459 boolean isSatisfiedBy(int ch) { 5460 return Character.isUnicodeIdentifierPart(ch);}}); 5461 defClone("javaIdentifierIgnorable", new CloneableProperty() { 5462 boolean isSatisfiedBy(int ch) { 5463 return Character.isIdentifierIgnorable(ch);}}); 5464 defClone("javaSpaceChar", new CloneableProperty() { 5465 boolean isSatisfiedBy(int ch) { 5466 return Character.isSpaceChar(ch);}}); 5467 defClone("javaWhitespace", new CloneableProperty() { 5468 boolean isSatisfiedBy(int ch) { 5469 return Character.isWhitespace(ch);}}); 5470 defClone("javaISOControl", new CloneableProperty() { 5471 boolean isSatisfiedBy(int ch) { 5472 return Character.isISOControl(ch);}}); 5473 defClone("javaMirrored", new CloneableProperty() { 5474 boolean isSatisfiedBy(int ch) { 5475 return Character.isMirrored(ch);}}); 5476 } 5477 } 5478 }