1 /* 2 * Copyright (c) 1999, 2013, 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.text.Normalizer; 29 import java.util.Locale; 30 import java.util.Iterator; 31 import java.util.Map; 32 import java.util.ArrayList; 33 import java.util.HashMap; 34 import java.util.Arrays; 35 import java.util.NoSuchElementException; 36 import java.util.Spliterator; 37 import java.util.Spliterators; 38 import java.util.function.Predicate; 39 import java.util.stream.Stream; 40 import java.util.stream.StreamSupport; 41 42 43 /** 44 * A compiled representation of a regular expression. 45 * 46 * <p> A regular expression, specified as a string, must first be compiled into 47 * an instance of this class. The resulting pattern can then be used to create 48 * a {@link Matcher} object that can match arbitrary {@linkplain 49 * java.lang.CharSequence character sequences} against the regular 50 * expression. All of the state involved in performing a match resides in the 51 * matcher, so many matchers can share the same pattern. 52 * 53 * <p> A typical invocation sequence is thus 54 * 55 * <blockquote><pre> 56 * Pattern p = Pattern.{@link #compile compile}("a*b"); 57 * Matcher m = p.{@link #matcher matcher}("aaaaab"); 58 * boolean b = m.{@link Matcher#matches matches}();</pre></blockquote> 59 * 60 * <p> A {@link #matches matches} method is defined by this class as a 61 * convenience for when a regular expression is used just once. This method 62 * compiles an expression and matches an input sequence against it in a single 63 * invocation. The statement 64 * 65 * <blockquote><pre> 66 * boolean b = Pattern.matches("a*b", "aaaaab");</pre></blockquote> 67 * 68 * is equivalent to the three statements above, though for repeated matches it 69 * is less efficient since it does not allow the compiled pattern to be reused. 70 * 71 * <p> Instances of this class are immutable and are safe for use by multiple 72 * concurrent threads. Instances of the {@link Matcher} class are not safe for 73 * such use. 74 * 75 * 76 * <h3><a name="sum">Summary of regular-expression constructs</a></h3> 77 * 78 * <table border="0" cellpadding="1" cellspacing="0" 79 * summary="Regular expression constructs, and what they match"> 80 * 81 * <tr align="left"> 82 * <th align="left" id="construct">Construct</th> 83 * <th align="left" id="matches">Matches</th> 84 * </tr> 85 * 86 * <tr><th> </th></tr> 87 * <tr align="left"><th colspan="2" id="characters">Characters</th></tr> 88 * 89 * <tr><td valign="top" headers="construct characters"><i>x</i></td> 90 * <td headers="matches">The character <i>x</i></td></tr> 91 * <tr><td valign="top" headers="construct characters"><tt>\\</tt></td> 92 * <td headers="matches">The backslash character</td></tr> 93 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>n</i></td> 94 * <td headers="matches">The character with octal value <tt>0</tt><i>n</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>nn</i></td> 97 * <td headers="matches">The character with octal value <tt>0</tt><i>nn</i> 98 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 99 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>mnn</i></td> 100 * <td headers="matches">The character with octal value <tt>0</tt><i>mnn</i> 101 * (0 <tt><=</tt> <i>m</i> <tt><=</tt> 3, 102 * 0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 103 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>hh</i></td> 104 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hh</i></td></tr> 105 * <tr><td valign="top" headers="construct characters"><tt>\u</tt><i>hhhh</i></td> 106 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hhhh</i></td></tr> 107 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>{h...h}</i></td> 108 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>h...h</i> 109 * ({@link java.lang.Character#MIN_CODE_POINT Character.MIN_CODE_POINT} 110 * <= <tt>0x</tt><i>h...h</i> <= 111 * {@link java.lang.Character#MAX_CODE_POINT Character.MAX_CODE_POINT})</td></tr> 112 * <tr><td valign="top" headers="matches"><tt>\t</tt></td> 113 * <td headers="matches">The tab character (<tt>'\u0009'</tt>)</td></tr> 114 * <tr><td valign="top" headers="construct characters"><tt>\n</tt></td> 115 * <td headers="matches">The newline (line feed) character (<tt>'\u000A'</tt>)</td></tr> 116 * <tr><td valign="top" headers="construct characters"><tt>\r</tt></td> 117 * <td headers="matches">The carriage-return character (<tt>'\u000D'</tt>)</td></tr> 118 * <tr><td valign="top" headers="construct characters"><tt>\f</tt></td> 119 * <td headers="matches">The form-feed character (<tt>'\u000C'</tt>)</td></tr> 120 * <tr><td valign="top" headers="construct characters"><tt>\a</tt></td> 121 * <td headers="matches">The alert (bell) character (<tt>'\u0007'</tt>)</td></tr> 122 * <tr><td valign="top" headers="construct characters"><tt>\e</tt></td> 123 * <td headers="matches">The escape character (<tt>'\u001B'</tt>)</td></tr> 124 * <tr><td valign="top" headers="construct characters"><tt>\c</tt><i>x</i></td> 125 * <td headers="matches">The control character corresponding to <i>x</i></td></tr> 126 * 127 * <tr><th> </th></tr> 128 * <tr align="left"><th colspan="2" id="classes">Character classes</th></tr> 129 * 130 * <tr><td valign="top" headers="construct classes">{@code [abc]}</td> 131 * <td headers="matches">{@code a}, {@code b}, or {@code c} (simple class)</td></tr> 132 * <tr><td valign="top" headers="construct classes">{@code [^abc]}</td> 133 * <td headers="matches">Any character except {@code a}, {@code b}, or {@code c} (negation)</td></tr> 134 * <tr><td valign="top" headers="construct classes">{@code [a-zA-Z]}</td> 135 * <td headers="matches">{@code a} through {@code z} 136 * or {@code A} through {@code Z}, inclusive (range)</td></tr> 137 * <tr><td valign="top" headers="construct classes">{@code [a-d[m-p]]}</td> 138 * <td headers="matches">{@code a} through {@code d}, 139 * or {@code m} through {@code p}: {@code [a-dm-p]} (union)</td></tr> 140 * <tr><td valign="top" headers="construct classes">{@code [a-z&&[def]]}</td> 141 * <td headers="matches">{@code d}, {@code e}, or {@code f} (intersection)</tr> 142 * <tr><td valign="top" headers="construct classes">{@code [a-z&&[^bc]]}</td> 143 * <td headers="matches">{@code a} through {@code z}, 144 * except for {@code b} and {@code c}: {@code [ad-z]} (subtraction)</td></tr> 145 * <tr><td valign="top" headers="construct classes">{@code [a-z&&[^m-p]]}</td> 146 * <td headers="matches">{@code a} through {@code z}, 147 * and not {@code m} through {@code p}: {@code [a-lq-z]}(subtraction)</td></tr> 148 * <tr><th> </th></tr> 149 * 150 * <tr align="left"><th colspan="2" id="predef">Predefined character classes</th></tr> 151 * 152 * <tr><td valign="top" headers="construct predef"><tt>.</tt></td> 153 * <td headers="matches">Any character (may or may not match <a href="#lt">line terminators</a>)</td></tr> 154 * <tr><td valign="top" headers="construct predef"><tt>\d</tt></td> 155 * <td headers="matches">A digit: <tt>[0-9]</tt></td></tr> 156 * <tr><td valign="top" headers="construct predef"><tt>\D</tt></td> 157 * <td headers="matches">A non-digit: <tt>[^0-9]</tt></td></tr> 158 * <tr><td valign="top" headers="construct predef"><tt>\h</tt></td> 159 * <td headers="matches">A horizontal whitespace character: 160 * <tt>[ \t\xA0\u1680\u180e\u2000-\u200a\u202f\u205f\u3000]</tt></td></tr> 161 * <tr><td valign="top" headers="construct predef"><tt>\H</tt></td> 162 * <td headers="matches">A non-horizontal whitespace character: <tt>[^\h]</tt></td></tr> 163 * <tr><td valign="top" headers="construct predef"><tt>\s</tt></td> 164 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 165 * <tr><td valign="top" headers="construct predef"><tt>\S</tt></td> 166 * <td headers="matches">A non-whitespace character: <tt>[^\s]</tt></td></tr> 167 * <tr><td valign="top" headers="construct predef"><tt>\v</tt></td> 168 * <td headers="matches">A vertical whitespace character: <tt>[\n\x0B\f\r\x85\u2028\u2029]</tt> 169 * </td></tr> 170 * <tr><td valign="top" headers="construct predef"><tt>\V</tt></td> 171 * <td headers="matches">A non-vertical whitespace character: <tt>[^\v]</tt></td></tr> 172 * <tr><td valign="top" headers="construct predef"><tt>\w</tt></td> 173 * <td headers="matches">A word character: <tt>[a-zA-Z_0-9]</tt></td></tr> 174 * <tr><td valign="top" headers="construct predef"><tt>\W</tt></td> 175 * <td headers="matches">A non-word character: <tt>[^\w]</tt></td></tr> 176 * <tr><th> </th></tr> 177 * <tr align="left"><th colspan="2" id="posix"><b>POSIX character classes (US-ASCII only)</b></th></tr> 178 * 179 * <tr><td valign="top" headers="construct posix">{@code \p{Lower}}</td> 180 * <td headers="matches">A lower-case alphabetic character: {@code [a-z]}</td></tr> 181 * <tr><td valign="top" headers="construct posix">{@code \p{Upper}}</td> 182 * <td headers="matches">An upper-case alphabetic character:{@code [A-Z]}</td></tr> 183 * <tr><td valign="top" headers="construct posix">{@code \p{ASCII}}</td> 184 * <td headers="matches">All ASCII:{@code [\x00-\x7F]}</td></tr> 185 * <tr><td valign="top" headers="construct posix">{@code \p{Alpha}}</td> 186 * <td headers="matches">An alphabetic character:{@code [\p{Lower}\p{Upper}]}</td></tr> 187 * <tr><td valign="top" headers="construct posix">{@code \p{Digit}}</td> 188 * <td headers="matches">A decimal digit: {@code [0-9]}</td></tr> 189 * <tr><td valign="top" headers="construct posix">{@code \p{Alnum}}</td> 190 * <td headers="matches">An alphanumeric character:{@code [\p{Alpha}\p{Digit}]}</td></tr> 191 * <tr><td valign="top" headers="construct posix">{@code \p{Punct}}</td> 192 * <td headers="matches">Punctuation: One of {@code !"#$%&'()*+,-./:;<=>?@[\]^_`{|}~}</td></tr> 193 * <!-- {@code [\!"#\$%&'\(\)\*\+,\-\./:;\<=\>\?@\[\\\]\^_`\{\|\}~]} 194 * {@code [\X21-\X2F\X31-\X40\X5B-\X60\X7B-\X7E]} --> 195 * <tr><td valign="top" headers="construct posix">{@code \p{Graph}}</td> 196 * <td headers="matches">A visible character: {@code [\p{Alnum}\p{Punct}]}</td></tr> 197 * <tr><td valign="top" headers="construct posix">{@code \p{Print}}</td> 198 * <td headers="matches">A printable character: {@code [\p{Graph}\x20]}</td></tr> 199 * <tr><td valign="top" headers="construct posix">{@code \p{Blank}}</td> 200 * <td headers="matches">A space or a tab: {@code [ \t]}</td></tr> 201 * <tr><td valign="top" headers="construct posix">{@code \p{Cntrl}}</td> 202 * <td headers="matches">A control character: {@code [\x00-\x1F\x7F]}</td></tr> 203 * <tr><td valign="top" headers="construct posix">{@code \p{XDigit}}</td> 204 * <td headers="matches">A hexadecimal digit: {@code [0-9a-fA-F]}</td></tr> 205 * <tr><td valign="top" headers="construct posix">{@code \p{Space}}</td> 206 * <td headers="matches">A whitespace character: {@code [ \t\n\x0B\f\r]}</td></tr> 207 * 208 * <tr><th> </th></tr> 209 * <tr align="left"><th colspan="2">java.lang.Character classes (simple <a href="#jcc">java character type</a>)</th></tr> 210 * 211 * <tr><td valign="top"><tt>\p{javaLowerCase}</tt></td> 212 * <td>Equivalent to java.lang.Character.isLowerCase()</td></tr> 213 * <tr><td valign="top"><tt>\p{javaUpperCase}</tt></td> 214 * <td>Equivalent to java.lang.Character.isUpperCase()</td></tr> 215 * <tr><td valign="top"><tt>\p{javaWhitespace}</tt></td> 216 * <td>Equivalent to java.lang.Character.isWhitespace()</td></tr> 217 * <tr><td valign="top"><tt>\p{javaMirrored}</tt></td> 218 * <td>Equivalent to java.lang.Character.isMirrored()</td></tr> 219 * 220 * <tr><th> </th></tr> 221 * <tr align="left"><th colspan="2" id="unicode">Classes for Unicode scripts, blocks, categories and binary properties</th></tr> 222 * <tr><td valign="top" headers="construct unicode">{@code \p{IsLatin}}</td> 223 * <td headers="matches">A Latin script character (<a href="#usc">script</a>)</td></tr> 224 * <tr><td valign="top" headers="construct unicode">{@code \p{InGreek}}</td> 225 * <td headers="matches">A character in the Greek block (<a href="#ubc">block</a>)</td></tr> 226 * <tr><td valign="top" headers="construct unicode">{@code \p{Lu}}</td> 227 * <td headers="matches">An uppercase letter (<a href="#ucc">category</a>)</td></tr> 228 * <tr><td valign="top" headers="construct unicode">{@code \p{IsAlphabetic}}</td> 229 * <td headers="matches">An alphabetic character (<a href="#ubpc">binary property</a>)</td></tr> 230 * <tr><td valign="top" headers="construct unicode">{@code \p{Sc}}</td> 231 * <td headers="matches">A currency symbol</td></tr> 232 * <tr><td valign="top" headers="construct unicode">{@code \P{InGreek}}</td> 233 * <td headers="matches">Any character except one in the Greek block (negation)</td></tr> 234 * <tr><td valign="top" headers="construct unicode">{@code [\p{L}&&[^\p{Lu}]]}</td> 235 * <td headers="matches">Any letter except an uppercase letter (subtraction)</td></tr> 236 * 237 * <tr><th> </th></tr> 238 * <tr align="left"><th colspan="2" id="bounds">Boundary matchers</th></tr> 239 * 240 * <tr><td valign="top" headers="construct bounds"><tt>^</tt></td> 241 * <td headers="matches">The beginning of a line</td></tr> 242 * <tr><td valign="top" headers="construct bounds"><tt>$</tt></td> 243 * <td headers="matches">The end of a line</td></tr> 244 * <tr><td valign="top" headers="construct bounds"><tt>\b</tt></td> 245 * <td headers="matches">A word boundary</td></tr> 246 * <tr><td valign="top" headers="construct bounds"><tt>\B</tt></td> 247 * <td headers="matches">A non-word boundary</td></tr> 248 * <tr><td valign="top" headers="construct bounds"><tt>\A</tt></td> 249 * <td headers="matches">The beginning of the input</td></tr> 250 * <tr><td valign="top" headers="construct bounds"><tt>\G</tt></td> 251 * <td headers="matches">The end of the previous match</td></tr> 252 * <tr><td valign="top" headers="construct bounds"><tt>\Z</tt></td> 253 * <td headers="matches">The end of the input but for the final 254 * <a href="#lt">terminator</a>, if any</td></tr> 255 * <tr><td valign="top" headers="construct bounds"><tt>\z</tt></td> 256 * <td headers="matches">The end of the input</td></tr> 257 * 258 * <tr><th> </th></tr> 259 * <tr align="left"><th colspan="2" id="lineending">Linebreak matcher</th></tr> 260 * <tr><td valign="top" headers="construct lineending"><tt>\R</tt></td> 261 * <td headers="matches">Any Unicode linebreak sequence, is equivalent to 262 * <tt>\u000D\u000A|[\u000A\u000B\u000C\u000D\u0085\u2028\u2029] 263 * </tt></td></tr> 264 * 265 * <tr><th> </th></tr> 266 * <tr align="left"><th colspan="2" id="greedy">Greedy quantifiers</th></tr> 267 * 268 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>?</tt></td> 269 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 270 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>*</tt></td> 271 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 272 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>+</tt></td> 273 * <td headers="matches"><i>X</i>, one or more times</td></tr> 274 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>}</tt></td> 275 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 276 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,}</tt></td> 277 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 278 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}</tt></td> 279 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 280 * 281 * <tr><th> </th></tr> 282 * <tr align="left"><th colspan="2" id="reluc">Reluctant quantifiers</th></tr> 283 * 284 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>??</tt></td> 285 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 286 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>*?</tt></td> 287 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 288 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>+?</tt></td> 289 * <td headers="matches"><i>X</i>, one or more times</td></tr> 290 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>}?</tt></td> 291 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 292 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,}?</tt></td> 293 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 294 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}?</tt></td> 295 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 296 * 297 * <tr><th> </th></tr> 298 * <tr align="left"><th colspan="2" id="poss">Possessive quantifiers</th></tr> 299 * 300 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>?+</tt></td> 301 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 302 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>*+</tt></td> 303 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 304 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>++</tt></td> 305 * <td headers="matches"><i>X</i>, one or more times</td></tr> 306 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>}+</tt></td> 307 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 308 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,}+</tt></td> 309 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 310 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}+</tt></td> 311 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 312 * 313 * <tr><th> </th></tr> 314 * <tr align="left"><th colspan="2" id="logical">Logical operators</th></tr> 315 * 316 * <tr><td valign="top" headers="construct logical"><i>XY</i></td> 317 * <td headers="matches"><i>X</i> followed by <i>Y</i></td></tr> 318 * <tr><td valign="top" headers="construct logical"><i>X</i><tt>|</tt><i>Y</i></td> 319 * <td headers="matches">Either <i>X</i> or <i>Y</i></td></tr> 320 * <tr><td valign="top" headers="construct logical"><tt>(</tt><i>X</i><tt>)</tt></td> 321 * <td headers="matches">X, as a <a href="#cg">capturing group</a></td></tr> 322 * 323 * <tr><th> </th></tr> 324 * <tr align="left"><th colspan="2" id="backref">Back references</th></tr> 325 * 326 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>n</i></td> 327 * <td valign="bottom" headers="matches">Whatever the <i>n</i><sup>th</sup> 328 * <a href="#cg">capturing group</a> matched</td></tr> 329 * 330 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>k</i><<i>name</i>></td> 331 * <td valign="bottom" headers="matches">Whatever the 332 * <a href="#groupname">named-capturing group</a> "name" matched</td></tr> 333 * 334 * <tr><th> </th></tr> 335 * <tr align="left"><th colspan="2" id="quot">Quotation</th></tr> 336 * 337 * <tr><td valign="top" headers="construct quot"><tt>\</tt></td> 338 * <td headers="matches">Nothing, but quotes the following character</td></tr> 339 * <tr><td valign="top" headers="construct quot"><tt>\Q</tt></td> 340 * <td headers="matches">Nothing, but quotes all characters until <tt>\E</tt></td></tr> 341 * <tr><td valign="top" headers="construct quot"><tt>\E</tt></td> 342 * <td headers="matches">Nothing, but ends quoting started by <tt>\Q</tt></td></tr> 343 * <!-- Metachars: !$()*+.<>?[\]^{|} --> 344 * 345 * <tr><th> </th></tr> 346 * <tr align="left"><th colspan="2" id="special">Special constructs (named-capturing and non-capturing)</th></tr> 347 * 348 * <tr><td valign="top" headers="construct special"><tt>(?<<a href="#groupname">name</a>></tt><i>X</i><tt>)</tt></td> 349 * <td headers="matches"><i>X</i>, as a named-capturing group</td></tr> 350 * <tr><td valign="top" headers="construct special"><tt>(?:</tt><i>X</i><tt>)</tt></td> 351 * <td headers="matches"><i>X</i>, as a non-capturing group</td></tr> 352 * <tr><td valign="top" headers="construct special"><tt>(?idmsuxU-idmsuxU) </tt></td> 353 * <td headers="matches">Nothing, but turns match flags <a href="#CASE_INSENSITIVE">i</a> 354 * <a href="#UNIX_LINES">d</a> <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> 355 * <a href="#UNICODE_CASE">u</a> <a href="#COMMENTS">x</a> <a href="#UNICODE_CHARACTER_CLASS">U</a> 356 * on - off</td></tr> 357 * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux:</tt><i>X</i><tt>)</tt> </td> 358 * <td headers="matches"><i>X</i>, as a <a href="#cg">non-capturing group</a> with the 359 * given flags <a href="#CASE_INSENSITIVE">i</a> <a href="#UNIX_LINES">d</a> 360 * <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> <a href="#UNICODE_CASE">u</a > 361 * <a href="#COMMENTS">x</a> on - off</td></tr> 362 * <tr><td valign="top" headers="construct special"><tt>(?=</tt><i>X</i><tt>)</tt></td> 363 * <td headers="matches"><i>X</i>, via zero-width positive lookahead</td></tr> 364 * <tr><td valign="top" headers="construct special"><tt>(?!</tt><i>X</i><tt>)</tt></td> 365 * <td headers="matches"><i>X</i>, via zero-width negative lookahead</td></tr> 366 * <tr><td valign="top" headers="construct special"><tt>(?<=</tt><i>X</i><tt>)</tt></td> 367 * <td headers="matches"><i>X</i>, via zero-width positive lookbehind</td></tr> 368 * <tr><td valign="top" headers="construct special"><tt>(?<!</tt><i>X</i><tt>)</tt></td> 369 * <td headers="matches"><i>X</i>, via zero-width negative lookbehind</td></tr> 370 * <tr><td valign="top" headers="construct special"><tt>(?></tt><i>X</i><tt>)</tt></td> 371 * <td headers="matches"><i>X</i>, as an independent, non-capturing group</td></tr> 372 * 373 * </table> 374 * 375 * <hr> 376 * 377 * 378 * <h3><a name="bs">Backslashes, escapes, and quoting</a></h3> 379 * 380 * <p> The backslash character (<tt>'\'</tt>) serves to introduce escaped 381 * constructs, as defined in the table above, as well as to quote characters 382 * that otherwise would be interpreted as unescaped constructs. Thus the 383 * expression <tt>\\</tt> matches a single backslash and <tt>\{</tt> matches a 384 * left brace. 385 * 386 * <p> It is an error to use a backslash prior to any alphabetic character that 387 * does not denote an escaped construct; these are reserved for future 388 * extensions to the regular-expression language. A backslash may be used 389 * prior to a non-alphabetic character regardless of whether that character is 390 * part of an unescaped construct. 391 * 392 * <p> Backslashes within string literals in Java source code are interpreted 393 * as required by 394 * <cite>The Java™ Language Specification</cite> 395 * as either Unicode escapes (section 3.3) or other character escapes (section 3.10.6) 396 * It is therefore necessary to double backslashes in string 397 * literals that represent regular expressions to protect them from 398 * interpretation by the Java bytecode compiler. The string literal 399 * <tt>"\b"</tt>, for example, matches a single backspace character when 400 * interpreted as a regular expression, while <tt>"\\b"</tt> matches a 401 * word boundary. The string literal <tt>"\(hello\)"</tt> is illegal 402 * and leads to a compile-time error; in order to match the string 403 * <tt>(hello)</tt> the string literal <tt>"\\(hello\\)"</tt> 404 * must be used. 405 * 406 * <h3><a name="cc">Character Classes</a></h3> 407 * 408 * <p> Character classes may appear within other character classes, and 409 * may be composed by the union operator (implicit) and the intersection 410 * operator (<tt>&&</tt>). 411 * The union operator denotes a class that contains every character that is 412 * in at least one of its operand classes. The intersection operator 413 * denotes a class that contains every character that is in both of its 414 * operand classes. 415 * 416 * <p> The precedence of character-class operators is as follows, from 417 * highest to lowest: 418 * 419 * <blockquote><table border="0" cellpadding="1" cellspacing="0" 420 * summary="Precedence of character class operators."> 421 * <tr><th>1 </th> 422 * <td>Literal escape </td> 423 * <td><tt>\x</tt></td></tr> 424 * <tr><th>2 </th> 425 * <td>Grouping</td> 426 * <td><tt>[...]</tt></td></tr> 427 * <tr><th>3 </th> 428 * <td>Range</td> 429 * <td><tt>a-z</tt></td></tr> 430 * <tr><th>4 </th> 431 * <td>Union</td> 432 * <td><tt>[a-e][i-u]</tt></td></tr> 433 * <tr><th>5 </th> 434 * <td>Intersection</td> 435 * <td>{@code [a-z&&[aeiou]]}</td></tr> 436 * </table></blockquote> 437 * 438 * <p> Note that a different set of metacharacters are in effect inside 439 * a character class than outside a character class. For instance, the 440 * regular expression <tt>.</tt> loses its special meaning inside a 441 * character class, while the expression <tt>-</tt> becomes a range 442 * forming metacharacter. 443 * 444 * <h3><a name="lt">Line terminators</a></h3> 445 * 446 * <p> A <i>line terminator</i> is a one- or two-character sequence that marks 447 * the end of a line of the input character sequence. The following are 448 * recognized as line terminators: 449 * 450 * <ul> 451 * 452 * <li> A newline (line feed) character (<tt>'\n'</tt>), 453 * 454 * <li> A carriage-return character followed immediately by a newline 455 * character (<tt>"\r\n"</tt>), 456 * 457 * <li> A standalone carriage-return character (<tt>'\r'</tt>), 458 * 459 * <li> A next-line character (<tt>'\u0085'</tt>), 460 * 461 * <li> A line-separator character (<tt>'\u2028'</tt>), or 462 * 463 * <li> A paragraph-separator character (<tt>'\u2029</tt>). 464 * 465 * </ul> 466 * <p>If {@link #UNIX_LINES} mode is activated, then the only line terminators 467 * recognized are newline characters. 468 * 469 * <p> The regular expression <tt>.</tt> matches any character except a line 470 * terminator unless the {@link #DOTALL} flag is specified. 471 * 472 * <p> By default, the regular expressions <tt>^</tt> and <tt>$</tt> ignore 473 * line terminators and only match at the beginning and the end, respectively, 474 * of the entire input sequence. If {@link #MULTILINE} mode is activated then 475 * <tt>^</tt> matches at the beginning of input and after any line terminator 476 * except at the end of input. When in {@link #MULTILINE} mode <tt>$</tt> 477 * matches just before a line terminator or the end of the input sequence. 478 * 479 * <h3><a name="cg">Groups and capturing</a></h3> 480 * 481 * <h4><a name="gnumber">Group number</a></h4> 482 * <p> Capturing groups are numbered by counting their opening parentheses from 483 * left to right. In the expression <tt>((A)(B(C)))</tt>, for example, there 484 * are four such groups: </p> 485 * 486 * <blockquote><table cellpadding=1 cellspacing=0 summary="Capturing group numberings"> 487 * <tr><th>1 </th> 488 * <td><tt>((A)(B(C)))</tt></td></tr> 489 * <tr><th>2 </th> 490 * <td><tt>(A)</tt></td></tr> 491 * <tr><th>3 </th> 492 * <td><tt>(B(C))</tt></td></tr> 493 * <tr><th>4 </th> 494 * <td><tt>(C)</tt></td></tr> 495 * </table></blockquote> 496 * 497 * <p> Group zero always stands for the entire expression. 498 * 499 * <p> Capturing groups are so named because, during a match, each subsequence 500 * of the input sequence that matches such a group is saved. The captured 501 * subsequence may be used later in the expression, via a back reference, and 502 * may also be retrieved from the matcher once the match operation is complete. 503 * 504 * <h4><a name="groupname">Group name</a></h4> 505 * <p>A capturing group can also be assigned a "name", a <tt>named-capturing group</tt>, 506 * and then be back-referenced later by the "name". Group names are composed of 507 * the following characters. The first character must be a <tt>letter</tt>. 508 * 509 * <ul> 510 * <li> The uppercase letters <tt>'A'</tt> through <tt>'Z'</tt> 511 * (<tt>'\u0041'</tt> through <tt>'\u005a'</tt>), 512 * <li> The lowercase letters <tt>'a'</tt> through <tt>'z'</tt> 513 * (<tt>'\u0061'</tt> through <tt>'\u007a'</tt>), 514 * <li> The digits <tt>'0'</tt> through <tt>'9'</tt> 515 * (<tt>'\u0030'</tt> through <tt>'\u0039'</tt>), 516 * </ul> 517 * 518 * <p> A <tt>named-capturing group</tt> is still numbered as described in 519 * <a href="#gnumber">Group number</a>. 520 * 521 * <p> The captured input associated with a group is always the subsequence 522 * that the group most recently matched. If a group is evaluated a second time 523 * because of quantification then its previously-captured value, if any, will 524 * be retained if the second evaluation fails. Matching the string 525 * <tt>"aba"</tt> against the expression <tt>(a(b)?)+</tt>, for example, leaves 526 * group two set to <tt>"b"</tt>. All captured input is discarded at the 527 * beginning of each match. 528 * 529 * <p> Groups beginning with <tt>(?</tt> are either pure, <i>non-capturing</i> groups 530 * that do not capture text and do not count towards the group total, or 531 * <i>named-capturing</i> group. 532 * 533 * <h3> Unicode support </h3> 534 * 535 * <p> This class is in conformance with Level 1 of <a 536 * href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical 537 * Standard #18: Unicode Regular Expression</i></a>, plus RL2.1 538 * Canonical Equivalents. 539 * <p> 540 * <b>Unicode escape sequences</b> such as <tt>\u2014</tt> in Java source code 541 * are processed as described in section 3.3 of 542 * <cite>The Java™ Language Specification</cite>. 543 * Such escape sequences are also implemented directly by the regular-expression 544 * parser so that Unicode escapes can be used in expressions that are read from 545 * files or from the keyboard. Thus the strings <tt>"\u2014"</tt> and 546 * <tt>"\\u2014"</tt>, while not equal, compile into the same pattern, which 547 * matches the character with hexadecimal value <tt>0x2014</tt>. 548 * <p> 549 * A Unicode character can also be represented in a regular-expression by 550 * using its <b>Hex notation</b>(hexadecimal code point value) directly as described in construct 551 * <tt>\x{...}</tt>, for example a supplementary character U+2011F 552 * can be specified as <tt>\x{2011F}</tt>, instead of two consecutive 553 * Unicode escape sequences of the surrogate pair 554 * <tt>\uD840</tt><tt>\uDD1F</tt>. 555 * <p> 556 * Unicode scripts, blocks, categories and binary properties are written with 557 * the <tt>\p</tt> and <tt>\P</tt> constructs as in Perl. 558 * <tt>\p{</tt><i>prop</i><tt>}</tt> matches if 559 * the input has the property <i>prop</i>, while <tt>\P{</tt><i>prop</i><tt>}</tt> 560 * does not match if the input has that property. 561 * <p> 562 * Scripts, blocks, categories and binary properties can be used both inside 563 * and outside of a character class. 564 * 565 * <p> 566 * <b><a name="usc">Scripts</a></b> are specified either with the prefix {@code Is}, as in 567 * {@code IsHiragana}, or by using the {@code script} keyword (or its short 568 * form {@code sc})as in {@code script=Hiragana} or {@code sc=Hiragana}. 569 * <p> 570 * The script names supported by <code>Pattern</code> are the valid script names 571 * accepted and defined by 572 * {@link java.lang.Character.UnicodeScript#forName(String) UnicodeScript.forName}. 573 * 574 * <p> 575 * <b><a name="ubc">Blocks</a></b> are specified with the prefix {@code In}, as in 576 * {@code InMongolian}, or by using the keyword {@code block} (or its short 577 * form {@code blk}) as in {@code block=Mongolian} or {@code blk=Mongolian}. 578 * <p> 579 * The block names supported by <code>Pattern</code> are the valid block names 580 * accepted and defined by 581 * {@link java.lang.Character.UnicodeBlock#forName(String) UnicodeBlock.forName}. 582 * <p> 583 * 584 * <b><a name="ucc">Categories</a></b> may be specified with the optional prefix {@code Is}: 585 * Both {@code \p{L}} and {@code \p{IsL}} denote the category of Unicode 586 * letters. Same as scripts and blocks, categories can also be specified 587 * by using the keyword {@code general_category} (or its short form 588 * {@code gc}) as in {@code general_category=Lu} or {@code gc=Lu}. 589 * <p> 590 * The supported categories are those of 591 * <a href="http://www.unicode.org/unicode/standard/standard.html"> 592 * <i>The Unicode Standard</i></a> in the version specified by the 593 * {@link java.lang.Character Character} class. The category names are those 594 * defined in the Standard, both normative and informative. 595 * <p> 596 * 597 * <b><a name="ubpc">Binary properties</a></b> are specified with the prefix {@code Is}, as in 598 * {@code IsAlphabetic}. The supported binary properties by <code>Pattern</code> 599 * are 600 * <ul> 601 * <li> Alphabetic 602 * <li> Ideographic 603 * <li> Letter 604 * <li> Lowercase 605 * <li> Uppercase 606 * <li> Titlecase 607 * <li> Punctuation 608 * <Li> Control 609 * <li> White_Space 610 * <li> Digit 611 * <li> Hex_Digit 612 * <li> Join_Control 613 * <li> Noncharacter_Code_Point 614 * <li> Assigned 615 * </ul> 616 * <p> 617 * The following <b>Predefined Character classes</b> and <b>POSIX character classes</b> 618 * are in conformance with the recommendation of <i>Annex C: Compatibility Properties</i> 619 * of <a href="http://www.unicode.org/reports/tr18/"><i>Unicode Regular Expression 620 * </i></a>, when {@link #UNICODE_CHARACTER_CLASS} flag is specified. 621 * 622 * <table border="0" cellpadding="1" cellspacing="0" 623 * summary="predefined and posix character classes in Unicode mode"> 624 * <tr align="left"> 625 * <th align="left" id="predef_classes">Classes</th> 626 * <th align="left" id="predef_matches">Matches</th> 627 *</tr> 628 * <tr><td><tt>\p{Lower}</tt></td> 629 * <td>A lowercase character:<tt>\p{IsLowercase}</tt></td></tr> 630 * <tr><td><tt>\p{Upper}</tt></td> 631 * <td>An uppercase character:<tt>\p{IsUppercase}</tt></td></tr> 632 * <tr><td><tt>\p{ASCII}</tt></td> 633 * <td>All ASCII:<tt>[\x00-\x7F]</tt></td></tr> 634 * <tr><td><tt>\p{Alpha}</tt></td> 635 * <td>An alphabetic character:<tt>\p{IsAlphabetic}</tt></td></tr> 636 * <tr><td><tt>\p{Digit}</tt></td> 637 * <td>A decimal digit character:<tt>p{IsDigit}</tt></td></tr> 638 * <tr><td><tt>\p{Alnum}</tt></td> 639 * <td>An alphanumeric character:<tt>[\p{IsAlphabetic}\p{IsDigit}]</tt></td></tr> 640 * <tr><td><tt>\p{Punct}</tt></td> 641 * <td>A punctuation character:<tt>p{IsPunctuation}</tt></td></tr> 642 * <tr><td><tt>\p{Graph}</tt></td> 643 * <td>A visible character: <tt>[^\p{IsWhite_Space}\p{gc=Cc}\p{gc=Cs}\p{gc=Cn}]</tt></td></tr> 644 * <tr><td><tt>\p{Print}</tt></td> 645 * <td>A printable character: {@code [\p{Graph}\p{Blank}&&[^\p{Cntrl}]]}</td></tr> 646 * <tr><td><tt>\p{Blank}</tt></td> 647 * <td>A space or a tab: {@code [\p{IsWhite_Space}&&[^\p{gc=Zl}\p{gc=Zp}\x0a\x0b\x0c\x0d\x85]]}</td></tr> 648 * <tr><td><tt>\p{Cntrl}</tt></td> 649 * <td>A control character: <tt>\p{gc=Cc}</tt></td></tr> 650 * <tr><td><tt>\p{XDigit}</tt></td> 651 * <td>A hexadecimal digit: <tt>[\p{gc=Nd}\p{IsHex_Digit}]</tt></td></tr> 652 * <tr><td><tt>\p{Space}</tt></td> 653 * <td>A whitespace character:<tt>\p{IsWhite_Space}</tt></td></tr> 654 * <tr><td><tt>\d</tt></td> 655 * <td>A digit: <tt>\p{IsDigit}</tt></td></tr> 656 * <tr><td><tt>\D</tt></td> 657 * <td>A non-digit: <tt>[^\d]</tt></td></tr> 658 * <tr><td><tt>\s</tt></td> 659 * <td>A whitespace character: <tt>\p{IsWhite_Space}</tt></td></tr> 660 * <tr><td><tt>\S</tt></td> 661 * <td>A non-whitespace character: <tt>[^\s]</tt></td></tr> 662 * <tr><td><tt>\w</tt></td> 663 * <td>A word character: <tt>[\p{Alpha}\p{gc=Mn}\p{gc=Me}\p{gc=Mc}\p{Digit}\p{gc=Pc}\p{IsJoin_Control}]</tt></td></tr> 664 * <tr><td><tt>\W</tt></td> 665 * <td>A non-word character: <tt>[^\w]</tt></td></tr> 666 * </table> 667 * <p> 668 * <a name="jcc"> 669 * Categories that behave like the java.lang.Character 670 * boolean is<i>methodname</i> methods (except for the deprecated ones) are 671 * available through the same <tt>\p{</tt><i>prop</i><tt>}</tt> syntax where 672 * the specified property has the name <tt>java<i>methodname</i></tt></a>. 673 * 674 * <h3> Comparison to Perl 5 </h3> 675 * 676 * <p>The <code>Pattern</code> engine performs traditional NFA-based matching 677 * with ordered alternation as occurs in Perl 5. 678 * 679 * <p> Perl constructs not supported by this class: </p> 680 * 681 * <ul> 682 * <li><p> Predefined character classes (Unicode character) 683 * <p><tt>\X </tt>Match Unicode 684 * <a href="http://www.unicode.org/reports/tr18/#Default_Grapheme_Clusters"> 685 * <i>extended grapheme cluster</i></a> 686 * </p></li> 687 * 688 * <li><p> The backreference constructs, <tt>\g{</tt><i>n</i><tt>}</tt> for 689 * the <i>n</i><sup>th</sup><a href="#cg">capturing group</a> and 690 * <tt>\g{</tt><i>name</i><tt>}</tt> for 691 * <a href="#groupname">named-capturing group</a>. 692 * </p></li> 693 * 694 * <li><p> The named character construct, <tt>\N{</tt><i>name</i><tt>}</tt> 695 * for a Unicode character by its name. 696 * </p></li> 697 * 698 * <li><p> The conditional constructs 699 * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>)</tt> and 700 * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>|</tt><i>Y</i><tt>)</tt>, 701 * </p></li> 702 * 703 * <li><p> The embedded code constructs <tt>(?{</tt><i>code</i><tt>})</tt> 704 * and <tt>(??{</tt><i>code</i><tt>})</tt>,</p></li> 705 * 706 * <li><p> The embedded comment syntax <tt>(?#comment)</tt>, and </p></li> 707 * 708 * <li><p> The preprocessing operations <tt>\l</tt> <tt>\u</tt>, 709 * <tt>\L</tt>, and <tt>\U</tt>. </p></li> 710 * 711 * </ul> 712 * 713 * <p> Constructs supported by this class but not by Perl: </p> 714 * 715 * <ul> 716 * 717 * <li><p> Character-class union and intersection as described 718 * <a href="#cc">above</a>.</p></li> 719 * 720 * </ul> 721 * 722 * <p> Notable differences from Perl: </p> 723 * 724 * <ul> 725 * 726 * <li><p> In Perl, <tt>\1</tt> through <tt>\9</tt> are always interpreted 727 * as back references; a backslash-escaped number greater than <tt>9</tt> is 728 * treated as a back reference if at least that many subexpressions exist, 729 * otherwise it is interpreted, if possible, as an octal escape. In this 730 * class octal escapes must always begin with a zero. In this class, 731 * <tt>\1</tt> through <tt>\9</tt> are always interpreted as back 732 * references, and a larger number is accepted as a back reference if at 733 * least that many subexpressions exist at that point in the regular 734 * expression, otherwise the parser will drop digits until the number is 735 * smaller or equal to the existing number of groups or it is one digit. 736 * </p></li> 737 * 738 * <li><p> Perl uses the <tt>g</tt> flag to request a match that resumes 739 * where the last match left off. This functionality is provided implicitly 740 * by the {@link Matcher} class: Repeated invocations of the {@link 741 * Matcher#find find} method will resume where the last match left off, 742 * unless the matcher is reset. </p></li> 743 * 744 * <li><p> In Perl, embedded flags at the top level of an expression affect 745 * the whole expression. In this class, embedded flags always take effect 746 * at the point at which they appear, whether they are at the top level or 747 * within a group; in the latter case, flags are restored at the end of the 748 * group just as in Perl. </p></li> 749 * 750 * </ul> 751 * 752 * 753 * <p> For a more precise description of the behavior of regular expression 754 * constructs, please see <a href="http://www.oreilly.com/catalog/regex3/"> 755 * <i>Mastering Regular Expressions, 3nd Edition</i>, Jeffrey E. F. Friedl, 756 * O'Reilly and Associates, 2006.</a> 757 * </p> 758 * 759 * @see java.lang.String#split(String, int) 760 * @see java.lang.String#split(String) 761 * 762 * @author Mike McCloskey 763 * @author Mark Reinhold 764 * @author JSR-51 Expert Group 765 * @since 1.4 766 * @spec JSR-51 767 */ 768 769 public final class Pattern 770 implements java.io.Serializable 771 { 772 773 /** 774 * Regular expression modifier values. Instead of being passed as 775 * arguments, they can also be passed as inline modifiers. 776 * For example, the following statements have the same effect. 777 * <pre> 778 * RegExp r1 = RegExp.compile("abc", Pattern.I|Pattern.M); 779 * RegExp r2 = RegExp.compile("(?im)abc", 0); 780 * </pre> 781 * 782 * The flags are duplicated so that the familiar Perl match flag 783 * names are available. 784 */ 785 786 /** 787 * Enables Unix lines mode. 788 * 789 * <p> In this mode, only the <tt>'\n'</tt> line terminator is recognized 790 * in the behavior of <tt>.</tt>, <tt>^</tt>, and <tt>$</tt>. 791 * 792 * <p> Unix lines mode can also be enabled via the embedded flag 793 * expression <tt>(?d)</tt>. 794 */ 795 public static final int UNIX_LINES = 0x01; 796 797 /** 798 * Enables case-insensitive matching. 799 * 800 * <p> By default, case-insensitive matching assumes that only characters 801 * in the US-ASCII charset are being matched. Unicode-aware 802 * case-insensitive matching can be enabled by specifying the {@link 803 * #UNICODE_CASE} flag in conjunction with this flag. 804 * 805 * <p> Case-insensitive matching can also be enabled via the embedded flag 806 * expression <tt>(?i)</tt>. 807 * 808 * <p> Specifying this flag may impose a slight performance penalty. </p> 809 */ 810 public static final int CASE_INSENSITIVE = 0x02; 811 812 /** 813 * Permits whitespace and comments in pattern. 814 * 815 * <p> In this mode, whitespace is ignored, and embedded comments starting 816 * with <tt>#</tt> are ignored until the end of a line. 817 * 818 * <p> Comments mode can also be enabled via the embedded flag 819 * expression <tt>(?x)</tt>. 820 */ 821 public static final int COMMENTS = 0x04; 822 823 /** 824 * Enables multiline mode. 825 * 826 * <p> In multiline mode the expressions <tt>^</tt> and <tt>$</tt> match 827 * just after or just before, respectively, a line terminator or the end of 828 * the input sequence. By default these expressions only match at the 829 * beginning and the end of the entire input sequence. 830 * 831 * <p> Multiline mode can also be enabled via the embedded flag 832 * expression <tt>(?m)</tt>. </p> 833 */ 834 public static final int MULTILINE = 0x08; 835 836 /** 837 * Enables literal parsing of the pattern. 838 * 839 * <p> When this flag is specified then the input string that specifies 840 * the pattern is treated as a sequence of literal characters. 841 * Metacharacters or escape sequences in the input sequence will be 842 * given no special meaning. 843 * 844 * <p>The flags CASE_INSENSITIVE and UNICODE_CASE retain their impact on 845 * matching when used in conjunction with this flag. The other flags 846 * become superfluous. 847 * 848 * <p> There is no embedded flag character for enabling literal parsing. 849 * @since 1.5 850 */ 851 public static final int LITERAL = 0x10; 852 853 /** 854 * Enables dotall mode. 855 * 856 * <p> In dotall mode, the expression <tt>.</tt> matches any character, 857 * including a line terminator. By default this expression does not match 858 * line terminators. 859 * 860 * <p> Dotall mode can also be enabled via the embedded flag 861 * expression <tt>(?s)</tt>. (The <tt>s</tt> is a mnemonic for 862 * "single-line" mode, which is what this is called in Perl.) </p> 863 */ 864 public static final int DOTALL = 0x20; 865 866 /** 867 * Enables Unicode-aware case folding. 868 * 869 * <p> When this flag is specified then case-insensitive matching, when 870 * enabled by the {@link #CASE_INSENSITIVE} flag, is done in a manner 871 * consistent with the Unicode Standard. By default, case-insensitive 872 * matching assumes that only characters in the US-ASCII charset are being 873 * matched. 874 * 875 * <p> Unicode-aware case folding can also be enabled via the embedded flag 876 * expression <tt>(?u)</tt>. 877 * 878 * <p> Specifying this flag may impose a performance penalty. </p> 879 */ 880 public static final int UNICODE_CASE = 0x40; 881 882 /** 883 * Enables canonical equivalence. 884 * 885 * <p> When this flag is specified then two characters will be considered 886 * to match if, and only if, their full canonical decompositions match. 887 * The expression <tt>"a\u030A"</tt>, for example, will match the 888 * string <tt>"\u00E5"</tt> when this flag is specified. By default, 889 * matching does not take canonical equivalence into account. 890 * 891 * <p> There is no embedded flag character for enabling canonical 892 * equivalence. 893 * 894 * <p> Specifying this flag may impose a performance penalty. </p> 895 */ 896 public static final int CANON_EQ = 0x80; 897 898 /** 899 * Enables the Unicode version of <i>Predefined character classes</i> and 900 * <i>POSIX character classes</i>. 901 * 902 * <p> When this flag is specified then the (US-ASCII only) 903 * <i>Predefined character classes</i> and <i>POSIX character classes</i> 904 * are in conformance with 905 * <a href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical 906 * Standard #18: Unicode Regular Expression</i></a> 907 * <i>Annex C: Compatibility Properties</i>. 908 * <p> 909 * The UNICODE_CHARACTER_CLASS mode can also be enabled via the embedded 910 * flag expression <tt>(?U)</tt>. 911 * <p> 912 * The flag implies UNICODE_CASE, that is, it enables Unicode-aware case 913 * folding. 914 * <p> 915 * Specifying this flag may impose a performance penalty. </p> 916 * @since 1.7 917 */ 918 public static final int UNICODE_CHARACTER_CLASS = 0x100; 919 920 /* Pattern has only two serialized components: The pattern string 921 * and the flags, which are all that is needed to recompile the pattern 922 * when it is deserialized. 923 */ 924 925 /** use serialVersionUID from Merlin b59 for interoperability */ 926 private static final long serialVersionUID = 5073258162644648461L; 927 928 /** 929 * The original regular-expression pattern string. 930 * 931 * @serial 932 */ 933 private String pattern; 934 935 /** 936 * The original pattern flags. 937 * 938 * @serial 939 */ 940 private int flags; 941 942 /** 943 * Boolean indicating this Pattern is compiled; this is necessary in order 944 * to lazily compile deserialized Patterns. 945 */ 946 private transient volatile boolean compiled = false; 947 948 /** 949 * The normalized pattern string. 950 */ 951 private transient String normalizedPattern; 952 953 /** 954 * The starting point of state machine for the find operation. This allows 955 * a match to start anywhere in the input. 956 */ 957 transient Node root; 958 959 /** 960 * The root of object tree for a match operation. The pattern is matched 961 * at the beginning. This may include a find that uses BnM or a First 962 * node. 963 */ 964 transient Node matchRoot; 965 966 /** 967 * Temporary storage used by parsing pattern slice. 968 */ 969 transient int[] buffer; 970 971 /** 972 * Map the "name" of the "named capturing group" to its group id 973 * node. 974 */ 975 transient volatile Map<String, Integer> namedGroups; 976 977 /** 978 * Temporary storage used while parsing group references. 979 */ 980 transient GroupHead[] groupNodes; 981 982 /** 983 * Temporary null terminated code point array used by pattern compiling. 984 */ 985 private transient int[] temp; 986 987 /** 988 * The number of capturing groups in this Pattern. Used by matchers to 989 * allocate storage needed to perform a match. 990 */ 991 transient int capturingGroupCount; 992 993 /** 994 * The local variable count used by parsing tree. Used by matchers to 995 * allocate storage needed to perform a match. 996 */ 997 transient int localCount; 998 999 /** 1000 * Index into the pattern string that keeps track of how much has been 1001 * parsed. 1002 */ 1003 private transient int cursor; 1004 1005 /** 1006 * Holds the length of the pattern string. 1007 */ 1008 private transient int patternLength; 1009 1010 /** 1011 * If the Start node might possibly match supplementary characters. 1012 * It is set to true during compiling if 1013 * (1) There is supplementary char in pattern, or 1014 * (2) There is complement node of Category or Block 1015 */ 1016 private transient boolean hasSupplementary; 1017 1018 /** 1019 * Compiles the given regular expression into a pattern. 1020 * 1021 * @param regex 1022 * The expression to be compiled 1023 * @return the given regular expression compiled into a pattern 1024 * @throws PatternSyntaxException 1025 * If the expression's syntax is invalid 1026 */ 1027 public static Pattern compile(String regex) { 1028 return new Pattern(regex, 0); 1029 } 1030 1031 /** 1032 * Compiles the given regular expression into a pattern with the given 1033 * flags. 1034 * 1035 * @param regex 1036 * The expression to be compiled 1037 * 1038 * @param flags 1039 * Match flags, a bit mask that may include 1040 * {@link #CASE_INSENSITIVE}, {@link #MULTILINE}, {@link #DOTALL}, 1041 * {@link #UNICODE_CASE}, {@link #CANON_EQ}, {@link #UNIX_LINES}, 1042 * {@link #LITERAL}, {@link #UNICODE_CHARACTER_CLASS} 1043 * and {@link #COMMENTS} 1044 * 1045 * @return the given regular expression compiled into a pattern with the given flags 1046 * @throws IllegalArgumentException 1047 * If bit values other than those corresponding to the defined 1048 * match flags are set in <tt>flags</tt> 1049 * 1050 * @throws PatternSyntaxException 1051 * If the expression's syntax is invalid 1052 */ 1053 public static Pattern compile(String regex, int flags) { 1054 return new Pattern(regex, flags); 1055 } 1056 1057 /** 1058 * Returns the regular expression from which this pattern was compiled. 1059 * 1060 * @return The source of this pattern 1061 */ 1062 public String pattern() { 1063 return pattern; 1064 } 1065 1066 /** 1067 * <p>Returns the string representation of this pattern. This 1068 * is the regular expression from which this pattern was 1069 * compiled.</p> 1070 * 1071 * @return The string representation of this pattern 1072 * @since 1.5 1073 */ 1074 public String toString() { 1075 return pattern; 1076 } 1077 1078 /** 1079 * Creates a matcher that will match the given input against this pattern. 1080 * 1081 * @param input 1082 * The character sequence to be matched 1083 * 1084 * @return A new matcher for this pattern 1085 */ 1086 public Matcher matcher(CharSequence input) { 1087 if (!compiled) { 1088 synchronized(this) { 1089 if (!compiled) 1090 compile(); 1091 } 1092 } 1093 Matcher m = new Matcher(this, input); 1094 return m; 1095 } 1096 1097 /** 1098 * Returns this pattern's match flags. 1099 * 1100 * @return The match flags specified when this pattern was compiled 1101 */ 1102 public int flags() { 1103 return flags; 1104 } 1105 1106 /** 1107 * Compiles the given regular expression and attempts to match the given 1108 * input against it. 1109 * 1110 * <p> An invocation of this convenience method of the form 1111 * 1112 * <blockquote><pre> 1113 * Pattern.matches(regex, input);</pre></blockquote> 1114 * 1115 * behaves in exactly the same way as the expression 1116 * 1117 * <blockquote><pre> 1118 * Pattern.compile(regex).matcher(input).matches()</pre></blockquote> 1119 * 1120 * <p> If a pattern is to be used multiple times, compiling it once and reusing 1121 * it will be more efficient than invoking this method each time. </p> 1122 * 1123 * @param regex 1124 * The expression to be compiled 1125 * 1126 * @param input 1127 * The character sequence to be matched 1128 * @return whether or not the regular expression matches on the input 1129 * @throws PatternSyntaxException 1130 * If the expression's syntax is invalid 1131 */ 1132 public static boolean matches(String regex, CharSequence input) { 1133 Pattern p = Pattern.compile(regex); 1134 Matcher m = p.matcher(input); 1135 return m.matches(); 1136 } 1137 1138 /** 1139 * Splits the given input sequence around matches of this pattern. 1140 * 1141 * <p> The array returned by this method contains each substring of the 1142 * input sequence that is terminated by another subsequence that matches 1143 * this pattern or is terminated by the end of the input sequence. The 1144 * substrings in the array are in the order in which they occur in the 1145 * input. If this pattern does not match any subsequence of the input then 1146 * the resulting array has just one element, namely the input sequence in 1147 * string form. A zero-length input sequence always results zero-length 1148 * resulting array. 1149 * 1150 * <p> When there is a positive-width match at the beginning of the input 1151 * sequence then an empty leading substring is included at the beginning 1152 * of the resulting array. A zero-width match at the beginning however 1153 * never produces such empty leading substring. 1154 * 1155 * <p> The <tt>limit</tt> parameter controls the number of times the 1156 * pattern is applied and therefore affects the length of the resulting 1157 * array. If the limit <i>n</i> is greater than zero then the pattern 1158 * will be applied at most <i>n</i> - 1 times, the array's 1159 * length will be no greater than <i>n</i>, and the array's last entry 1160 * will contain all input beyond the last matched delimiter. If <i>n</i> 1161 * is non-positive then the pattern will be applied as many times as 1162 * possible and the array can have any length. If <i>n</i> is zero then 1163 * the pattern will be applied as many times as possible, the array can 1164 * have any length, and trailing empty strings will be discarded. 1165 * 1166 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1167 * results with these parameters: 1168 * 1169 * <blockquote><table cellpadding=1 cellspacing=0 1170 * summary="Split examples showing regex, limit, and result"> 1171 * <tr><th align="left"><i>Regex </i></th> 1172 * <th align="left"><i>Limit </i></th> 1173 * <th align="left"><i>Result </i></th></tr> 1174 * <tr><td align=center>:</td> 1175 * <td align=center>2</td> 1176 * <td><tt>{ "boo", "and:foo" }</tt></td></tr> 1177 * <tr><td align=center>:</td> 1178 * <td align=center>5</td> 1179 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1180 * <tr><td align=center>:</td> 1181 * <td align=center>-2</td> 1182 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1183 * <tr><td align=center>o</td> 1184 * <td align=center>5</td> 1185 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1186 * <tr><td align=center>o</td> 1187 * <td align=center>-2</td> 1188 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1189 * <tr><td align=center>o</td> 1190 * <td align=center>0</td> 1191 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1192 * </table></blockquote> 1193 * 1194 * @param input 1195 * The character sequence to be split 1196 * 1197 * @param limit 1198 * The result threshold, as described above 1199 * 1200 * @return The array of strings computed by splitting the input 1201 * around matches of this pattern 1202 */ 1203 public String[] split(CharSequence input, int limit) { 1204 if (input.length() == 0) 1205 return new String[0]; 1206 int index = 0; 1207 boolean matchLimited = limit > 0; 1208 ArrayList<String> matchList = new ArrayList<>(); 1209 Matcher m = matcher(input); 1210 1211 // Add segments before each match found 1212 while(m.find()) { 1213 if (!matchLimited || matchList.size() < limit - 1) { 1214 if (index == 0 && index == m.start() && m.start() == m.end()) { 1215 // no empty leading substring included for zero-width match 1216 // at the beginning of the input char sequence. 1217 continue; 1218 } 1219 String match = input.subSequence(index, m.start()).toString(); 1220 matchList.add(match); 1221 index = m.end(); 1222 } else if (matchList.size() == limit - 1) { // last one 1223 String match = input.subSequence(index, 1224 input.length()).toString(); 1225 matchList.add(match); 1226 index = m.end(); 1227 } 1228 } 1229 1230 // If no match was found, return this 1231 if (index == 0) 1232 return new String[] {input.toString()}; 1233 1234 // Add remaining segment 1235 if (!matchLimited || matchList.size() < limit) 1236 matchList.add(input.subSequence(index, input.length()).toString()); 1237 1238 // Construct result 1239 int resultSize = matchList.size(); 1240 if (limit == 0) 1241 while (resultSize > 0 && matchList.get(resultSize-1).equals("")) 1242 resultSize--; 1243 String[] result = new String[resultSize]; 1244 return matchList.subList(0, resultSize).toArray(result); 1245 } 1246 1247 /** 1248 * Splits the given input sequence around matches of this pattern. 1249 * 1250 * <p> This method works as if by invoking the two-argument {@link 1251 * #split(java.lang.CharSequence, int) split} method with the given input 1252 * sequence and a limit argument of zero. Trailing empty strings are 1253 * therefore not included in the resulting array. </p> 1254 * 1255 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1256 * results with these expressions: 1257 * 1258 * <blockquote><table cellpadding=1 cellspacing=0 1259 * summary="Split examples showing regex and result"> 1260 * <tr><th align="left"><i>Regex </i></th> 1261 * <th align="left"><i>Result</i></th></tr> 1262 * <tr><td align=center>:</td> 1263 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1264 * <tr><td align=center>o</td> 1265 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1266 * </table></blockquote> 1267 * 1268 * 1269 * @param input 1270 * The character sequence to be split 1271 * 1272 * @return The array of strings computed by splitting the input 1273 * around matches of this pattern 1274 */ 1275 public String[] split(CharSequence input) { 1276 return split(input, 0); 1277 } 1278 1279 /** 1280 * Returns a literal pattern <code>String</code> for the specified 1281 * <code>String</code>. 1282 * 1283 * <p>This method produces a <code>String</code> that can be used to 1284 * create a <code>Pattern</code> that would match the string 1285 * <code>s</code> as if it were a literal pattern.</p> Metacharacters 1286 * or escape sequences in the input sequence will be given no special 1287 * meaning. 1288 * 1289 * @param s The string to be literalized 1290 * @return A literal string replacement 1291 * @since 1.5 1292 */ 1293 public static String quote(String s) { 1294 int slashEIndex = s.indexOf("\\E"); 1295 if (slashEIndex == -1) 1296 return "\\Q" + s + "\\E"; 1297 1298 StringBuilder sb = new StringBuilder(s.length() * 2); 1299 sb.append("\\Q"); 1300 slashEIndex = 0; 1301 int current = 0; 1302 while ((slashEIndex = s.indexOf("\\E", current)) != -1) { 1303 sb.append(s.substring(current, slashEIndex)); 1304 current = slashEIndex + 2; 1305 sb.append("\\E\\\\E\\Q"); 1306 } 1307 sb.append(s.substring(current, s.length())); 1308 sb.append("\\E"); 1309 return sb.toString(); 1310 } 1311 1312 /** 1313 * Recompile the Pattern instance from a stream. The original pattern 1314 * string is read in and the object tree is recompiled from it. 1315 */ 1316 private void readObject(java.io.ObjectInputStream s) 1317 throws java.io.IOException, ClassNotFoundException { 1318 1319 // Read in all fields 1320 s.defaultReadObject(); 1321 1322 // Initialize counts 1323 capturingGroupCount = 1; 1324 localCount = 0; 1325 1326 // if length > 0, the Pattern is lazily compiled 1327 compiled = false; 1328 if (pattern.length() == 0) { 1329 root = new Start(lastAccept); 1330 matchRoot = lastAccept; 1331 compiled = true; 1332 } 1333 } 1334 1335 /** 1336 * This private constructor is used to create all Patterns. The pattern 1337 * string and match flags are all that is needed to completely describe 1338 * a Pattern. An empty pattern string results in an object tree with 1339 * only a Start node and a LastNode node. 1340 */ 1341 private Pattern(String p, int f) { 1342 pattern = p; 1343 flags = f; 1344 1345 // to use UNICODE_CASE if UNICODE_CHARACTER_CLASS present 1346 if ((flags & UNICODE_CHARACTER_CLASS) != 0) 1347 flags |= UNICODE_CASE; 1348 1349 // Reset group index count 1350 capturingGroupCount = 1; 1351 localCount = 0; 1352 1353 if (pattern.length() > 0) { 1354 compile(); 1355 } else { 1356 root = new Start(lastAccept); 1357 matchRoot = lastAccept; 1358 } 1359 } 1360 1361 /** 1362 * The pattern is converted to normalizedD form and then a pure group 1363 * is constructed to match canonical equivalences of the characters. 1364 */ 1365 private void normalize() { 1366 boolean inCharClass = false; 1367 int lastCodePoint = -1; 1368 1369 // Convert pattern into normalizedD form 1370 normalizedPattern = Normalizer.normalize(pattern, Normalizer.Form.NFD); 1371 patternLength = normalizedPattern.length(); 1372 1373 // Modify pattern to match canonical equivalences 1374 StringBuilder newPattern = new StringBuilder(patternLength); 1375 for(int i=0; i<patternLength; ) { 1376 int c = normalizedPattern.codePointAt(i); 1377 StringBuilder sequenceBuffer; 1378 if ((Character.getType(c) == Character.NON_SPACING_MARK) 1379 && (lastCodePoint != -1)) { 1380 sequenceBuffer = new StringBuilder(); 1381 sequenceBuffer.appendCodePoint(lastCodePoint); 1382 sequenceBuffer.appendCodePoint(c); 1383 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1384 i += Character.charCount(c); 1385 if (i >= patternLength) 1386 break; 1387 c = normalizedPattern.codePointAt(i); 1388 sequenceBuffer.appendCodePoint(c); 1389 } 1390 String ea = produceEquivalentAlternation( 1391 sequenceBuffer.toString()); 1392 newPattern.setLength(newPattern.length()-Character.charCount(lastCodePoint)); 1393 newPattern.append("(?:").append(ea).append(")"); 1394 } else if (c == '[' && lastCodePoint != '\\') { 1395 i = normalizeCharClass(newPattern, i); 1396 } else { 1397 newPattern.appendCodePoint(c); 1398 } 1399 lastCodePoint = c; 1400 i += Character.charCount(c); 1401 } 1402 normalizedPattern = newPattern.toString(); 1403 } 1404 1405 /** 1406 * Complete the character class being parsed and add a set 1407 * of alternations to it that will match the canonical equivalences 1408 * of the characters within the class. 1409 */ 1410 private int normalizeCharClass(StringBuilder newPattern, int i) { 1411 StringBuilder charClass = new StringBuilder(); 1412 StringBuilder eq = null; 1413 int lastCodePoint = -1; 1414 String result; 1415 1416 i++; 1417 charClass.append("["); 1418 while(true) { 1419 int c = normalizedPattern.codePointAt(i); 1420 StringBuilder sequenceBuffer; 1421 1422 if (c == ']' && lastCodePoint != '\\') { 1423 charClass.append((char)c); 1424 break; 1425 } else if (Character.getType(c) == Character.NON_SPACING_MARK) { 1426 sequenceBuffer = new StringBuilder(); 1427 sequenceBuffer.appendCodePoint(lastCodePoint); 1428 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1429 sequenceBuffer.appendCodePoint(c); 1430 i += Character.charCount(c); 1431 if (i >= normalizedPattern.length()) 1432 break; 1433 c = normalizedPattern.codePointAt(i); 1434 } 1435 String ea = produceEquivalentAlternation( 1436 sequenceBuffer.toString()); 1437 1438 charClass.setLength(charClass.length()-Character.charCount(lastCodePoint)); 1439 if (eq == null) 1440 eq = new StringBuilder(); 1441 eq.append('|'); 1442 eq.append(ea); 1443 } else { 1444 charClass.appendCodePoint(c); 1445 i++; 1446 } 1447 if (i == normalizedPattern.length()) 1448 throw error("Unclosed character class"); 1449 lastCodePoint = c; 1450 } 1451 1452 if (eq != null) { 1453 result = "(?:"+charClass.toString()+eq.toString()+")"; 1454 } else { 1455 result = charClass.toString(); 1456 } 1457 1458 newPattern.append(result); 1459 return i; 1460 } 1461 1462 /** 1463 * Given a specific sequence composed of a regular character and 1464 * combining marks that follow it, produce the alternation that will 1465 * match all canonical equivalences of that sequence. 1466 */ 1467 private String produceEquivalentAlternation(String source) { 1468 int len = countChars(source, 0, 1); 1469 if (source.length() == len) 1470 // source has one character. 1471 return source; 1472 1473 String base = source.substring(0,len); 1474 String combiningMarks = source.substring(len); 1475 1476 String[] perms = producePermutations(combiningMarks); 1477 StringBuilder result = new StringBuilder(source); 1478 1479 // Add combined permutations 1480 for(int x=0; x<perms.length; x++) { 1481 String next = base + perms[x]; 1482 if (x>0) 1483 result.append("|"+next); 1484 next = composeOneStep(next); 1485 if (next != null) 1486 result.append("|"+produceEquivalentAlternation(next)); 1487 } 1488 return result.toString(); 1489 } 1490 1491 /** 1492 * Returns an array of strings that have all the possible 1493 * permutations of the characters in the input string. 1494 * This is used to get a list of all possible orderings 1495 * of a set of combining marks. Note that some of the permutations 1496 * are invalid because of combining class collisions, and these 1497 * possibilities must be removed because they are not canonically 1498 * equivalent. 1499 */ 1500 private String[] producePermutations(String input) { 1501 if (input.length() == countChars(input, 0, 1)) 1502 return new String[] {input}; 1503 1504 if (input.length() == countChars(input, 0, 2)) { 1505 int c0 = Character.codePointAt(input, 0); 1506 int c1 = Character.codePointAt(input, Character.charCount(c0)); 1507 if (getClass(c1) == getClass(c0)) { 1508 return new String[] {input}; 1509 } 1510 String[] result = new String[2]; 1511 result[0] = input; 1512 StringBuilder sb = new StringBuilder(2); 1513 sb.appendCodePoint(c1); 1514 sb.appendCodePoint(c0); 1515 result[1] = sb.toString(); 1516 return result; 1517 } 1518 1519 int length = 1; 1520 int nCodePoints = countCodePoints(input); 1521 for(int x=1; x<nCodePoints; x++) 1522 length = length * (x+1); 1523 1524 String[] temp = new String[length]; 1525 1526 int combClass[] = new int[nCodePoints]; 1527 for(int x=0, i=0; x<nCodePoints; x++) { 1528 int c = Character.codePointAt(input, i); 1529 combClass[x] = getClass(c); 1530 i += Character.charCount(c); 1531 } 1532 1533 // For each char, take it out and add the permutations 1534 // of the remaining chars 1535 int index = 0; 1536 int len; 1537 // offset maintains the index in code units. 1538 loop: for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) { 1539 len = countChars(input, offset, 1); 1540 boolean skip = false; 1541 for(int y=x-1; y>=0; y--) { 1542 if (combClass[y] == combClass[x]) { 1543 continue loop; 1544 } 1545 } 1546 StringBuilder sb = new StringBuilder(input); 1547 String otherChars = sb.delete(offset, offset+len).toString(); 1548 String[] subResult = producePermutations(otherChars); 1549 1550 String prefix = input.substring(offset, offset+len); 1551 for(int y=0; y<subResult.length; y++) 1552 temp[index++] = prefix + subResult[y]; 1553 } 1554 String[] result = new String[index]; 1555 for (int x=0; x<index; x++) 1556 result[x] = temp[x]; 1557 return result; 1558 } 1559 1560 private int getClass(int c) { 1561 return sun.text.Normalizer.getCombiningClass(c); 1562 } 1563 1564 /** 1565 * Attempts to compose input by combining the first character 1566 * with the first combining mark following it. Returns a String 1567 * that is the composition of the leading character with its first 1568 * combining mark followed by the remaining combining marks. Returns 1569 * null if the first two characters cannot be further composed. 1570 */ 1571 private String composeOneStep(String input) { 1572 int len = countChars(input, 0, 2); 1573 String firstTwoCharacters = input.substring(0, len); 1574 String result = Normalizer.normalize(firstTwoCharacters, Normalizer.Form.NFC); 1575 1576 if (result.equals(firstTwoCharacters)) 1577 return null; 1578 else { 1579 String remainder = input.substring(len); 1580 return result + remainder; 1581 } 1582 } 1583 1584 /** 1585 * Preprocess any \Q...\E sequences in `temp', meta-quoting them. 1586 * See the description of `quotemeta' in perlfunc(1). 1587 */ 1588 private void RemoveQEQuoting() { 1589 final int pLen = patternLength; 1590 int i = 0; 1591 while (i < pLen-1) { 1592 if (temp[i] != '\\') 1593 i += 1; 1594 else if (temp[i + 1] != 'Q') 1595 i += 2; 1596 else 1597 break; 1598 } 1599 if (i >= pLen - 1) // No \Q sequence found 1600 return; 1601 int j = i; 1602 i += 2; 1603 int[] newtemp = new int[j + 3*(pLen-i) + 2]; 1604 System.arraycopy(temp, 0, newtemp, 0, j); 1605 1606 boolean inQuote = true; 1607 boolean beginQuote = true; 1608 while (i < pLen) { 1609 int c = temp[i++]; 1610 if (!ASCII.isAscii(c) || ASCII.isAlpha(c)) { 1611 newtemp[j++] = c; 1612 } else if (ASCII.isDigit(c)) { 1613 if (beginQuote) { 1614 /* 1615 * A unicode escape \[0xu] could be before this quote, 1616 * and we don't want this numeric char to processed as 1617 * part of the escape. 1618 */ 1619 newtemp[j++] = '\\'; 1620 newtemp[j++] = 'x'; 1621 newtemp[j++] = '3'; 1622 } 1623 newtemp[j++] = c; 1624 } else if (c != '\\') { 1625 if (inQuote) newtemp[j++] = '\\'; 1626 newtemp[j++] = c; 1627 } else if (inQuote) { 1628 if (temp[i] == 'E') { 1629 i++; 1630 inQuote = false; 1631 } else { 1632 newtemp[j++] = '\\'; 1633 newtemp[j++] = '\\'; 1634 } 1635 } else { 1636 if (temp[i] == 'Q') { 1637 i++; 1638 inQuote = true; 1639 beginQuote = true; 1640 continue; 1641 } else { 1642 newtemp[j++] = c; 1643 if (i != pLen) 1644 newtemp[j++] = temp[i++]; 1645 } 1646 } 1647 1648 beginQuote = false; 1649 } 1650 1651 patternLength = j; 1652 temp = Arrays.copyOf(newtemp, j + 2); // double zero termination 1653 } 1654 1655 /** 1656 * Copies regular expression to an int array and invokes the parsing 1657 * of the expression which will create the object tree. 1658 */ 1659 private void compile() { 1660 // Handle canonical equivalences 1661 if (has(CANON_EQ) && !has(LITERAL)) { 1662 normalize(); 1663 } else { 1664 normalizedPattern = pattern; 1665 } 1666 patternLength = normalizedPattern.length(); 1667 1668 // Copy pattern to int array for convenience 1669 // Use double zero to terminate pattern 1670 temp = new int[patternLength + 2]; 1671 1672 hasSupplementary = false; 1673 int c, count = 0; 1674 // Convert all chars into code points 1675 for (int x = 0; x < patternLength; x += Character.charCount(c)) { 1676 c = normalizedPattern.codePointAt(x); 1677 if (isSupplementary(c)) { 1678 hasSupplementary = true; 1679 } 1680 temp[count++] = c; 1681 } 1682 1683 patternLength = count; // patternLength now in code points 1684 1685 if (! has(LITERAL)) 1686 RemoveQEQuoting(); 1687 1688 // Allocate all temporary objects here. 1689 buffer = new int[32]; 1690 groupNodes = new GroupHead[10]; 1691 namedGroups = null; 1692 1693 if (has(LITERAL)) { 1694 // Literal pattern handling 1695 matchRoot = newSlice(temp, patternLength, hasSupplementary); 1696 matchRoot.next = lastAccept; 1697 } else { 1698 // Start recursive descent parsing 1699 matchRoot = expr(lastAccept); 1700 // Check extra pattern characters 1701 if (patternLength != cursor) { 1702 if (peek() == ')') { 1703 throw error("Unmatched closing ')'"); 1704 } else { 1705 throw error("Unexpected internal error"); 1706 } 1707 } 1708 } 1709 1710 // Peephole optimization 1711 if (matchRoot instanceof Slice) { 1712 root = BnM.optimize(matchRoot); 1713 if (root == matchRoot) { 1714 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1715 } 1716 } else if (matchRoot instanceof Begin || matchRoot instanceof First) { 1717 root = matchRoot; 1718 } else { 1719 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1720 } 1721 1722 // Release temporary storage 1723 temp = null; 1724 buffer = null; 1725 groupNodes = null; 1726 patternLength = 0; 1727 compiled = true; 1728 } 1729 1730 Map<String, Integer> namedGroups() { 1731 if (namedGroups == null) 1732 namedGroups = new HashMap<>(2); 1733 return namedGroups; 1734 } 1735 1736 /** 1737 * Used to print out a subtree of the Pattern to help with debugging. 1738 */ 1739 private static void printObjectTree(Node node) { 1740 while(node != null) { 1741 if (node instanceof Prolog) { 1742 System.out.println(node); 1743 printObjectTree(((Prolog)node).loop); 1744 System.out.println("**** end contents prolog loop"); 1745 } else if (node instanceof Loop) { 1746 System.out.println(node); 1747 printObjectTree(((Loop)node).body); 1748 System.out.println("**** end contents Loop body"); 1749 } else if (node instanceof Curly) { 1750 System.out.println(node); 1751 printObjectTree(((Curly)node).atom); 1752 System.out.println("**** end contents Curly body"); 1753 } else if (node instanceof GroupCurly) { 1754 System.out.println(node); 1755 printObjectTree(((GroupCurly)node).atom); 1756 System.out.println("**** end contents GroupCurly body"); 1757 } else if (node instanceof GroupTail) { 1758 System.out.println(node); 1759 System.out.println("Tail next is "+node.next); 1760 return; 1761 } else { 1762 System.out.println(node); 1763 } 1764 node = node.next; 1765 if (node != null) 1766 System.out.println("->next:"); 1767 if (node == Pattern.accept) { 1768 System.out.println("Accept Node"); 1769 node = null; 1770 } 1771 } 1772 } 1773 1774 /** 1775 * Used to accumulate information about a subtree of the object graph 1776 * so that optimizations can be applied to the subtree. 1777 */ 1778 static final class TreeInfo { 1779 int minLength; 1780 int maxLength; 1781 boolean maxValid; 1782 boolean deterministic; 1783 1784 TreeInfo() { 1785 reset(); 1786 } 1787 void reset() { 1788 minLength = 0; 1789 maxLength = 0; 1790 maxValid = true; 1791 deterministic = true; 1792 } 1793 } 1794 1795 /* 1796 * The following private methods are mainly used to improve the 1797 * readability of the code. In order to let the Java compiler easily 1798 * inline them, we should not put many assertions or error checks in them. 1799 */ 1800 1801 /** 1802 * Indicates whether a particular flag is set or not. 1803 */ 1804 private boolean has(int f) { 1805 return (flags & f) != 0; 1806 } 1807 1808 /** 1809 * Match next character, signal error if failed. 1810 */ 1811 private void accept(int ch, String s) { 1812 int testChar = temp[cursor++]; 1813 if (has(COMMENTS)) 1814 testChar = parsePastWhitespace(testChar); 1815 if (ch != testChar) { 1816 throw error(s); 1817 } 1818 } 1819 1820 /** 1821 * Mark the end of pattern with a specific character. 1822 */ 1823 private void mark(int c) { 1824 temp[patternLength] = c; 1825 } 1826 1827 /** 1828 * Peek the next character, and do not advance the cursor. 1829 */ 1830 private int peek() { 1831 int ch = temp[cursor]; 1832 if (has(COMMENTS)) 1833 ch = peekPastWhitespace(ch); 1834 return ch; 1835 } 1836 1837 /** 1838 * Read the next character, and advance the cursor by one. 1839 */ 1840 private int read() { 1841 int ch = temp[cursor++]; 1842 if (has(COMMENTS)) 1843 ch = parsePastWhitespace(ch); 1844 return ch; 1845 } 1846 1847 /** 1848 * Read the next character, and advance the cursor by one, 1849 * ignoring the COMMENTS setting 1850 */ 1851 private int readEscaped() { 1852 int ch = temp[cursor++]; 1853 return ch; 1854 } 1855 1856 /** 1857 * Advance the cursor by one, and peek the next character. 1858 */ 1859 private int next() { 1860 int ch = temp[++cursor]; 1861 if (has(COMMENTS)) 1862 ch = peekPastWhitespace(ch); 1863 return ch; 1864 } 1865 1866 /** 1867 * Advance the cursor by one, and peek the next character, 1868 * ignoring the COMMENTS setting 1869 */ 1870 private int nextEscaped() { 1871 int ch = temp[++cursor]; 1872 return ch; 1873 } 1874 1875 /** 1876 * If in xmode peek past whitespace and comments. 1877 */ 1878 private int peekPastWhitespace(int ch) { 1879 while (ASCII.isSpace(ch) || ch == '#') { 1880 while (ASCII.isSpace(ch)) 1881 ch = temp[++cursor]; 1882 if (ch == '#') { 1883 ch = peekPastLine(); 1884 } 1885 } 1886 return ch; 1887 } 1888 1889 /** 1890 * If in xmode parse past whitespace and comments. 1891 */ 1892 private int parsePastWhitespace(int ch) { 1893 while (ASCII.isSpace(ch) || ch == '#') { 1894 while (ASCII.isSpace(ch)) 1895 ch = temp[cursor++]; 1896 if (ch == '#') 1897 ch = parsePastLine(); 1898 } 1899 return ch; 1900 } 1901 1902 /** 1903 * xmode parse past comment to end of line. 1904 */ 1905 private int parsePastLine() { 1906 int ch = temp[cursor++]; 1907 while (ch != 0 && !isLineSeparator(ch)) 1908 ch = temp[cursor++]; 1909 return ch; 1910 } 1911 1912 /** 1913 * xmode peek past comment to end of line. 1914 */ 1915 private int peekPastLine() { 1916 int ch = temp[++cursor]; 1917 while (ch != 0 && !isLineSeparator(ch)) 1918 ch = temp[++cursor]; 1919 return ch; 1920 } 1921 1922 /** 1923 * Determines if character is a line separator in the current mode 1924 */ 1925 private boolean isLineSeparator(int ch) { 1926 if (has(UNIX_LINES)) { 1927 return ch == '\n'; 1928 } else { 1929 return (ch == '\n' || 1930 ch == '\r' || 1931 (ch|1) == '\u2029' || 1932 ch == '\u0085'); 1933 } 1934 } 1935 1936 /** 1937 * Read the character after the next one, and advance the cursor by two. 1938 */ 1939 private int skip() { 1940 int i = cursor; 1941 int ch = temp[i+1]; 1942 cursor = i + 2; 1943 return ch; 1944 } 1945 1946 /** 1947 * Unread one next character, and retreat cursor by one. 1948 */ 1949 private void unread() { 1950 cursor--; 1951 } 1952 1953 /** 1954 * Internal method used for handling all syntax errors. The pattern is 1955 * displayed with a pointer to aid in locating the syntax error. 1956 */ 1957 private PatternSyntaxException error(String s) { 1958 return new PatternSyntaxException(s, normalizedPattern, cursor - 1); 1959 } 1960 1961 /** 1962 * Determines if there is any supplementary character or unpaired 1963 * surrogate in the specified range. 1964 */ 1965 private boolean findSupplementary(int start, int end) { 1966 for (int i = start; i < end; i++) { 1967 if (isSupplementary(temp[i])) 1968 return true; 1969 } 1970 return false; 1971 } 1972 1973 /** 1974 * Determines if the specified code point is a supplementary 1975 * character or unpaired surrogate. 1976 */ 1977 private static final boolean isSupplementary(int ch) { 1978 return ch >= Character.MIN_SUPPLEMENTARY_CODE_POINT || 1979 Character.isSurrogate((char)ch); 1980 } 1981 1982 /** 1983 * The following methods handle the main parsing. They are sorted 1984 * according to their precedence order, the lowest one first. 1985 */ 1986 1987 /** 1988 * The expression is parsed with branch nodes added for alternations. 1989 * This may be called recursively to parse sub expressions that may 1990 * contain alternations. 1991 */ 1992 private Node expr(Node end) { 1993 Node prev = null; 1994 Node firstTail = null; 1995 Branch branch = null; 1996 Node branchConn = null; 1997 1998 for (;;) { 1999 Node node = sequence(end); 2000 Node nodeTail = root; //double return 2001 if (prev == null) { 2002 prev = node; 2003 firstTail = nodeTail; 2004 } else { 2005 // Branch 2006 if (branchConn == null) { 2007 branchConn = new BranchConn(); 2008 branchConn.next = end; 2009 } 2010 if (node == end) { 2011 // if the node returned from sequence() is "end" 2012 // we have an empty expr, set a null atom into 2013 // the branch to indicate to go "next" directly. 2014 node = null; 2015 } else { 2016 // the "tail.next" of each atom goes to branchConn 2017 nodeTail.next = branchConn; 2018 } 2019 if (prev == branch) { 2020 branch.add(node); 2021 } else { 2022 if (prev == end) { 2023 prev = null; 2024 } else { 2025 // replace the "end" with "branchConn" at its tail.next 2026 // when put the "prev" into the branch as the first atom. 2027 firstTail.next = branchConn; 2028 } 2029 prev = branch = new Branch(prev, node, branchConn); 2030 } 2031 } 2032 if (peek() != '|') { 2033 return prev; 2034 } 2035 next(); 2036 } 2037 } 2038 2039 @SuppressWarnings("fallthrough") 2040 /** 2041 * Parsing of sequences between alternations. 2042 */ 2043 private Node sequence(Node end) { 2044 Node head = null; 2045 Node tail = null; 2046 Node node = null; 2047 LOOP: 2048 for (;;) { 2049 int ch = peek(); 2050 switch (ch) { 2051 case '(': 2052 // Because group handles its own closure, 2053 // we need to treat it differently 2054 node = group0(); 2055 // Check for comment or flag group 2056 if (node == null) 2057 continue; 2058 if (head == null) 2059 head = node; 2060 else 2061 tail.next = node; 2062 // Double return: Tail was returned in root 2063 tail = root; 2064 continue; 2065 case '[': 2066 node = clazz(true); 2067 break; 2068 case '\\': 2069 ch = nextEscaped(); 2070 if (ch == 'p' || ch == 'P') { 2071 boolean oneLetter = true; 2072 boolean comp = (ch == 'P'); 2073 ch = next(); // Consume { if present 2074 if (ch != '{') { 2075 unread(); 2076 } else { 2077 oneLetter = false; 2078 } 2079 node = family(oneLetter, comp); 2080 } else { 2081 unread(); 2082 node = atom(); 2083 } 2084 break; 2085 case '^': 2086 next(); 2087 if (has(MULTILINE)) { 2088 if (has(UNIX_LINES)) 2089 node = new UnixCaret(); 2090 else 2091 node = new Caret(); 2092 } else { 2093 node = new Begin(); 2094 } 2095 break; 2096 case '$': 2097 next(); 2098 if (has(UNIX_LINES)) 2099 node = new UnixDollar(has(MULTILINE)); 2100 else 2101 node = new Dollar(has(MULTILINE)); 2102 break; 2103 case '.': 2104 next(); 2105 if (has(DOTALL)) { 2106 node = new All(); 2107 } else { 2108 if (has(UNIX_LINES)) 2109 node = new UnixDot(); 2110 else { 2111 node = new Dot(); 2112 } 2113 } 2114 break; 2115 case '|': 2116 case ')': 2117 break LOOP; 2118 case ']': // Now interpreting dangling ] and } as literals 2119 case '}': 2120 node = atom(); 2121 break; 2122 case '?': 2123 case '*': 2124 case '+': 2125 next(); 2126 throw error("Dangling meta character '" + ((char)ch) + "'"); 2127 case 0: 2128 if (cursor >= patternLength) { 2129 break LOOP; 2130 } 2131 // Fall through 2132 default: 2133 node = atom(); 2134 break; 2135 } 2136 2137 node = closure(node); 2138 2139 if (head == null) { 2140 head = tail = node; 2141 } else { 2142 tail.next = node; 2143 tail = node; 2144 } 2145 } 2146 if (head == null) { 2147 return end; 2148 } 2149 tail.next = end; 2150 root = tail; //double return 2151 return head; 2152 } 2153 2154 @SuppressWarnings("fallthrough") 2155 /** 2156 * Parse and add a new Single or Slice. 2157 */ 2158 private Node atom() { 2159 int first = 0; 2160 int prev = -1; 2161 boolean hasSupplementary = false; 2162 int ch = peek(); 2163 for (;;) { 2164 switch (ch) { 2165 case '*': 2166 case '+': 2167 case '?': 2168 case '{': 2169 if (first > 1) { 2170 cursor = prev; // Unwind one character 2171 first--; 2172 } 2173 break; 2174 case '$': 2175 case '.': 2176 case '^': 2177 case '(': 2178 case '[': 2179 case '|': 2180 case ')': 2181 break; 2182 case '\\': 2183 ch = nextEscaped(); 2184 if (ch == 'p' || ch == 'P') { // Property 2185 if (first > 0) { // Slice is waiting; handle it first 2186 unread(); 2187 break; 2188 } else { // No slice; just return the family node 2189 boolean comp = (ch == 'P'); 2190 boolean oneLetter = true; 2191 ch = next(); // Consume { if present 2192 if (ch != '{') 2193 unread(); 2194 else 2195 oneLetter = false; 2196 return family(oneLetter, comp); 2197 } 2198 } 2199 unread(); 2200 prev = cursor; 2201 ch = escape(false, first == 0, false); 2202 if (ch >= 0) { 2203 append(ch, first); 2204 first++; 2205 if (isSupplementary(ch)) { 2206 hasSupplementary = true; 2207 } 2208 ch = peek(); 2209 continue; 2210 } else if (first == 0) { 2211 return root; 2212 } 2213 // Unwind meta escape sequence 2214 cursor = prev; 2215 break; 2216 case 0: 2217 if (cursor >= patternLength) { 2218 break; 2219 } 2220 // Fall through 2221 default: 2222 prev = cursor; 2223 append(ch, first); 2224 first++; 2225 if (isSupplementary(ch)) { 2226 hasSupplementary = true; 2227 } 2228 ch = next(); 2229 continue; 2230 } 2231 break; 2232 } 2233 if (first == 1) { 2234 return newSingle(buffer[0]); 2235 } else { 2236 return newSlice(buffer, first, hasSupplementary); 2237 } 2238 } 2239 2240 private void append(int ch, int len) { 2241 if (len >= buffer.length) { 2242 int[] tmp = new int[len+len]; 2243 System.arraycopy(buffer, 0, tmp, 0, len); 2244 buffer = tmp; 2245 } 2246 buffer[len] = ch; 2247 } 2248 2249 /** 2250 * Parses a backref greedily, taking as many numbers as it 2251 * can. The first digit is always treated as a backref, but 2252 * multi digit numbers are only treated as a backref if at 2253 * least that many backrefs exist at this point in the regex. 2254 */ 2255 private Node ref(int refNum) { 2256 boolean done = false; 2257 while(!done) { 2258 int ch = peek(); 2259 switch(ch) { 2260 case '0': 2261 case '1': 2262 case '2': 2263 case '3': 2264 case '4': 2265 case '5': 2266 case '6': 2267 case '7': 2268 case '8': 2269 case '9': 2270 int newRefNum = (refNum * 10) + (ch - '0'); 2271 // Add another number if it doesn't make a group 2272 // that doesn't exist 2273 if (capturingGroupCount - 1 < newRefNum) { 2274 done = true; 2275 break; 2276 } 2277 refNum = newRefNum; 2278 read(); 2279 break; 2280 default: 2281 done = true; 2282 break; 2283 } 2284 } 2285 if (has(CASE_INSENSITIVE)) 2286 return new CIBackRef(refNum, has(UNICODE_CASE)); 2287 else 2288 return new BackRef(refNum); 2289 } 2290 2291 /** 2292 * Parses an escape sequence to determine the actual value that needs 2293 * to be matched. 2294 * If -1 is returned and create was true a new object was added to the tree 2295 * to handle the escape sequence. 2296 * If the returned value is greater than zero, it is the value that 2297 * matches the escape sequence. 2298 */ 2299 private int escape(boolean inclass, boolean create, boolean isrange) { 2300 int ch = skip(); 2301 switch (ch) { 2302 case '0': 2303 return o(); 2304 case '1': 2305 case '2': 2306 case '3': 2307 case '4': 2308 case '5': 2309 case '6': 2310 case '7': 2311 case '8': 2312 case '9': 2313 if (inclass) break; 2314 if (create) { 2315 root = ref((ch - '0')); 2316 } 2317 return -1; 2318 case 'A': 2319 if (inclass) break; 2320 if (create) root = new Begin(); 2321 return -1; 2322 case 'B': 2323 if (inclass) break; 2324 if (create) root = new Bound(Bound.NONE, has(UNICODE_CHARACTER_CLASS)); 2325 return -1; 2326 case 'C': 2327 break; 2328 case 'D': 2329 if (create) root = has(UNICODE_CHARACTER_CLASS) 2330 ? new Utype(UnicodeProp.DIGIT).complement() 2331 : new Ctype(ASCII.DIGIT).complement(); 2332 return -1; 2333 case 'E': 2334 case 'F': 2335 break; 2336 case 'G': 2337 if (inclass) break; 2338 if (create) root = new LastMatch(); 2339 return -1; 2340 case 'H': 2341 if (create) root = new HorizWS().complement(); 2342 return -1; 2343 case 'I': 2344 case 'J': 2345 case 'K': 2346 case 'L': 2347 case 'M': 2348 case 'N': 2349 case 'O': 2350 case 'P': 2351 case 'Q': 2352 break; 2353 case 'R': 2354 if (inclass) break; 2355 if (create) root = new LineEnding(); 2356 return -1; 2357 case 'S': 2358 if (create) root = has(UNICODE_CHARACTER_CLASS) 2359 ? new Utype(UnicodeProp.WHITE_SPACE).complement() 2360 : new Ctype(ASCII.SPACE).complement(); 2361 return -1; 2362 case 'T': 2363 case 'U': 2364 break; 2365 case 'V': 2366 if (create) root = new VertWS().complement(); 2367 return -1; 2368 case 'W': 2369 if (create) root = has(UNICODE_CHARACTER_CLASS) 2370 ? new Utype(UnicodeProp.WORD).complement() 2371 : new Ctype(ASCII.WORD).complement(); 2372 return -1; 2373 case 'X': 2374 case 'Y': 2375 break; 2376 case 'Z': 2377 if (inclass) break; 2378 if (create) { 2379 if (has(UNIX_LINES)) 2380 root = new UnixDollar(false); 2381 else 2382 root = new Dollar(false); 2383 } 2384 return -1; 2385 case 'a': 2386 return '\007'; 2387 case 'b': 2388 if (inclass) break; 2389 if (create) root = new Bound(Bound.BOTH, has(UNICODE_CHARACTER_CLASS)); 2390 return -1; 2391 case 'c': 2392 return c(); 2393 case 'd': 2394 if (create) root = has(UNICODE_CHARACTER_CLASS) 2395 ? new Utype(UnicodeProp.DIGIT) 2396 : new Ctype(ASCII.DIGIT); 2397 return -1; 2398 case 'e': 2399 return '\033'; 2400 case 'f': 2401 return '\f'; 2402 case 'g': 2403 break; 2404 case 'h': 2405 if (create) root = new HorizWS(); 2406 return -1; 2407 case 'i': 2408 case 'j': 2409 break; 2410 case 'k': 2411 if (inclass) 2412 break; 2413 if (read() != '<') 2414 throw error("\\k is not followed by '<' for named capturing group"); 2415 String name = groupname(read()); 2416 if (!namedGroups().containsKey(name)) 2417 throw error("(named capturing group <"+ name+"> does not exit"); 2418 if (create) { 2419 if (has(CASE_INSENSITIVE)) 2420 root = new CIBackRef(namedGroups().get(name), has(UNICODE_CASE)); 2421 else 2422 root = new BackRef(namedGroups().get(name)); 2423 } 2424 return -1; 2425 case 'l': 2426 case 'm': 2427 break; 2428 case 'n': 2429 return '\n'; 2430 case 'o': 2431 case 'p': 2432 case 'q': 2433 break; 2434 case 'r': 2435 return '\r'; 2436 case 's': 2437 if (create) root = has(UNICODE_CHARACTER_CLASS) 2438 ? new Utype(UnicodeProp.WHITE_SPACE) 2439 : new Ctype(ASCII.SPACE); 2440 return -1; 2441 case 't': 2442 return '\t'; 2443 case 'u': 2444 return u(); 2445 case 'v': 2446 // '\v' was implemented as VT/0x0B in releases < 1.8 (though 2447 // undocumented). In JDK8 '\v' is specified as a predefined 2448 // character class for all vertical whitespace characters. 2449 // So [-1, root=VertWS node] pair is returned (instead of a 2450 // single 0x0B). This breaks the range if '\v' is used as 2451 // the start or end value, such as [\v-...] or [...-\v], in 2452 // which a single definite value (0x0B) is expected. For 2453 // compatibility concern '\013'/0x0B is returned if isrange. 2454 if (isrange) 2455 return '\013'; 2456 if (create) root = new VertWS(); 2457 return -1; 2458 case 'w': 2459 if (create) root = has(UNICODE_CHARACTER_CLASS) 2460 ? new Utype(UnicodeProp.WORD) 2461 : new Ctype(ASCII.WORD); 2462 return -1; 2463 case 'x': 2464 return x(); 2465 case 'y': 2466 break; 2467 case 'z': 2468 if (inclass) break; 2469 if (create) root = new End(); 2470 return -1; 2471 default: 2472 return ch; 2473 } 2474 throw error("Illegal/unsupported escape sequence"); 2475 } 2476 2477 /** 2478 * Parse a character class, and return the node that matches it. 2479 * 2480 * Consumes a ] on the way out if consume is true. Usually consume 2481 * is true except for the case of [abc&&def] where def is a separate 2482 * right hand node with "understood" brackets. 2483 */ 2484 private CharProperty clazz(boolean consume) { 2485 CharProperty prev = null; 2486 CharProperty node = null; 2487 BitClass bits = new BitClass(); 2488 boolean include = true; 2489 boolean firstInClass = true; 2490 int ch = next(); 2491 for (;;) { 2492 switch (ch) { 2493 case '^': 2494 // Negates if first char in a class, otherwise literal 2495 if (firstInClass) { 2496 if (temp[cursor-1] != '[') 2497 break; 2498 ch = next(); 2499 include = !include; 2500 continue; 2501 } else { 2502 // ^ not first in class, treat as literal 2503 break; 2504 } 2505 case '[': 2506 firstInClass = false; 2507 node = clazz(true); 2508 if (prev == null) 2509 prev = node; 2510 else 2511 prev = union(prev, node); 2512 ch = peek(); 2513 continue; 2514 case '&': 2515 firstInClass = false; 2516 ch = next(); 2517 if (ch == '&') { 2518 ch = next(); 2519 CharProperty rightNode = null; 2520 while (ch != ']' && ch != '&') { 2521 if (ch == '[') { 2522 if (rightNode == null) 2523 rightNode = clazz(true); 2524 else 2525 rightNode = union(rightNode, clazz(true)); 2526 } else { // abc&&def 2527 unread(); 2528 rightNode = clazz(false); 2529 } 2530 ch = peek(); 2531 } 2532 if (rightNode != null) 2533 node = rightNode; 2534 if (prev == null) { 2535 if (rightNode == null) 2536 throw error("Bad class syntax"); 2537 else 2538 prev = rightNode; 2539 } else { 2540 prev = intersection(prev, node); 2541 } 2542 } else { 2543 // treat as a literal & 2544 unread(); 2545 break; 2546 } 2547 continue; 2548 case 0: 2549 firstInClass = false; 2550 if (cursor >= patternLength) 2551 throw error("Unclosed character class"); 2552 break; 2553 case ']': 2554 firstInClass = false; 2555 if (prev != null) { 2556 if (consume) 2557 next(); 2558 return prev; 2559 } 2560 break; 2561 default: 2562 firstInClass = false; 2563 break; 2564 } 2565 node = range(bits); 2566 if (include) { 2567 if (prev == null) { 2568 prev = node; 2569 } else { 2570 if (prev != node) 2571 prev = union(prev, node); 2572 } 2573 } else { 2574 if (prev == null) { 2575 prev = node.complement(); 2576 } else { 2577 if (prev != node) 2578 prev = setDifference(prev, node); 2579 } 2580 } 2581 ch = peek(); 2582 } 2583 } 2584 2585 private CharProperty bitsOrSingle(BitClass bits, int ch) { 2586 /* Bits can only handle codepoints in [u+0000-u+00ff] range. 2587 Use "single" node instead of bits when dealing with unicode 2588 case folding for codepoints listed below. 2589 (1)Uppercase out of range: u+00ff, u+00b5 2590 toUpperCase(u+00ff) -> u+0178 2591 toUpperCase(u+00b5) -> u+039c 2592 (2)LatinSmallLetterLongS u+17f 2593 toUpperCase(u+017f) -> u+0053 2594 (3)LatinSmallLetterDotlessI u+131 2595 toUpperCase(u+0131) -> u+0049 2596 (4)LatinCapitalLetterIWithDotAbove u+0130 2597 toLowerCase(u+0130) -> u+0069 2598 (5)KelvinSign u+212a 2599 toLowerCase(u+212a) ==> u+006B 2600 (6)AngstromSign u+212b 2601 toLowerCase(u+212b) ==> u+00e5 2602 */ 2603 int d; 2604 if (ch < 256 && 2605 !(has(CASE_INSENSITIVE) && has(UNICODE_CASE) && 2606 (ch == 0xff || ch == 0xb5 || 2607 ch == 0x49 || ch == 0x69 || //I and i 2608 ch == 0x53 || ch == 0x73 || //S and s 2609 ch == 0x4b || ch == 0x6b || //K and k 2610 ch == 0xc5 || ch == 0xe5))) //A+ring 2611 return bits.add(ch, flags()); 2612 return newSingle(ch); 2613 } 2614 2615 /** 2616 * Parse a single character or a character range in a character class 2617 * and return its representative node. 2618 */ 2619 private CharProperty range(BitClass bits) { 2620 int ch = peek(); 2621 if (ch == '\\') { 2622 ch = nextEscaped(); 2623 if (ch == 'p' || ch == 'P') { // A property 2624 boolean comp = (ch == 'P'); 2625 boolean oneLetter = true; 2626 // Consume { if present 2627 ch = next(); 2628 if (ch != '{') 2629 unread(); 2630 else 2631 oneLetter = false; 2632 return family(oneLetter, comp); 2633 } else { // ordinary escape 2634 boolean isrange = temp[cursor+1] == '-'; 2635 unread(); 2636 ch = escape(true, true, isrange); 2637 if (ch == -1) 2638 return (CharProperty) root; 2639 } 2640 } else { 2641 next(); 2642 } 2643 if (ch >= 0) { 2644 if (peek() == '-') { 2645 int endRange = temp[cursor+1]; 2646 if (endRange == '[') { 2647 return bitsOrSingle(bits, ch); 2648 } 2649 if (endRange != ']') { 2650 next(); 2651 int m = peek(); 2652 if (m == '\\') { 2653 m = escape(true, false, true); 2654 } else { 2655 next(); 2656 } 2657 if (m < ch) { 2658 throw error("Illegal character range"); 2659 } 2660 if (has(CASE_INSENSITIVE)) 2661 return caseInsensitiveRangeFor(ch, m); 2662 else 2663 return rangeFor(ch, m); 2664 } 2665 } 2666 return bitsOrSingle(bits, ch); 2667 } 2668 throw error("Unexpected character '"+((char)ch)+"'"); 2669 } 2670 2671 /** 2672 * Parses a Unicode character family and returns its representative node. 2673 */ 2674 private CharProperty family(boolean singleLetter, 2675 boolean maybeComplement) 2676 { 2677 next(); 2678 String name; 2679 CharProperty node = null; 2680 2681 if (singleLetter) { 2682 int c = temp[cursor]; 2683 if (!Character.isSupplementaryCodePoint(c)) { 2684 name = String.valueOf((char)c); 2685 } else { 2686 name = new String(temp, cursor, 1); 2687 } 2688 read(); 2689 } else { 2690 int i = cursor; 2691 mark('}'); 2692 while(read() != '}') { 2693 } 2694 mark('\000'); 2695 int j = cursor; 2696 if (j > patternLength) 2697 throw error("Unclosed character family"); 2698 if (i + 1 >= j) 2699 throw error("Empty character family"); 2700 name = new String(temp, i, j-i-1); 2701 } 2702 2703 int i = name.indexOf('='); 2704 if (i != -1) { 2705 // property construct \p{name=value} 2706 String value = name.substring(i + 1); 2707 name = name.substring(0, i).toLowerCase(Locale.ENGLISH); 2708 if ("sc".equals(name) || "script".equals(name)) { 2709 node = unicodeScriptPropertyFor(value); 2710 } else if ("blk".equals(name) || "block".equals(name)) { 2711 node = unicodeBlockPropertyFor(value); 2712 } else if ("gc".equals(name) || "general_category".equals(name)) { 2713 node = charPropertyNodeFor(value); 2714 } else { 2715 throw error("Unknown Unicode property {name=<" + name + ">, " 2716 + "value=<" + value + ">}"); 2717 } 2718 } else { 2719 if (name.startsWith("In")) { 2720 // \p{inBlockName} 2721 node = unicodeBlockPropertyFor(name.substring(2)); 2722 } else if (name.startsWith("Is")) { 2723 // \p{isGeneralCategory} and \p{isScriptName} 2724 name = name.substring(2); 2725 UnicodeProp uprop = UnicodeProp.forName(name); 2726 if (uprop != null) 2727 node = new Utype(uprop); 2728 if (node == null) 2729 node = CharPropertyNames.charPropertyFor(name); 2730 if (node == null) 2731 node = unicodeScriptPropertyFor(name); 2732 } else { 2733 if (has(UNICODE_CHARACTER_CLASS)) { 2734 UnicodeProp uprop = UnicodeProp.forPOSIXName(name); 2735 if (uprop != null) 2736 node = new Utype(uprop); 2737 } 2738 if (node == null) 2739 node = charPropertyNodeFor(name); 2740 } 2741 } 2742 if (maybeComplement) { 2743 if (node instanceof Category || node instanceof Block) 2744 hasSupplementary = true; 2745 node = node.complement(); 2746 } 2747 return node; 2748 } 2749 2750 2751 /** 2752 * Returns a CharProperty matching all characters belong to 2753 * a UnicodeScript. 2754 */ 2755 private CharProperty unicodeScriptPropertyFor(String name) { 2756 final Character.UnicodeScript script; 2757 try { 2758 script = Character.UnicodeScript.forName(name); 2759 } catch (IllegalArgumentException iae) { 2760 throw error("Unknown character script name {" + name + "}"); 2761 } 2762 return new Script(script); 2763 } 2764 2765 /** 2766 * Returns a CharProperty matching all characters in a UnicodeBlock. 2767 */ 2768 private CharProperty unicodeBlockPropertyFor(String name) { 2769 final Character.UnicodeBlock block; 2770 try { 2771 block = Character.UnicodeBlock.forName(name); 2772 } catch (IllegalArgumentException iae) { 2773 throw error("Unknown character block name {" + name + "}"); 2774 } 2775 return new Block(block); 2776 } 2777 2778 /** 2779 * Returns a CharProperty matching all characters in a named property. 2780 */ 2781 private CharProperty charPropertyNodeFor(String name) { 2782 CharProperty p = CharPropertyNames.charPropertyFor(name); 2783 if (p == null) 2784 throw error("Unknown character property name {" + name + "}"); 2785 return p; 2786 } 2787 2788 /** 2789 * Parses and returns the name of a "named capturing group", the trailing 2790 * ">" is consumed after parsing. 2791 */ 2792 private String groupname(int ch) { 2793 StringBuilder sb = new StringBuilder(); 2794 sb.append(Character.toChars(ch)); 2795 while (ASCII.isLower(ch=read()) || ASCII.isUpper(ch) || 2796 ASCII.isDigit(ch)) { 2797 sb.append(Character.toChars(ch)); 2798 } 2799 if (sb.length() == 0) 2800 throw error("named capturing group has 0 length name"); 2801 if (ch != '>') 2802 throw error("named capturing group is missing trailing '>'"); 2803 return sb.toString(); 2804 } 2805 2806 /** 2807 * Parses a group and returns the head node of a set of nodes that process 2808 * the group. Sometimes a double return system is used where the tail is 2809 * returned in root. 2810 */ 2811 private Node group0() { 2812 boolean capturingGroup = false; 2813 Node head = null; 2814 Node tail = null; 2815 int save = flags; 2816 root = null; 2817 int ch = next(); 2818 if (ch == '?') { 2819 ch = skip(); 2820 switch (ch) { 2821 case ':': // (?:xxx) pure group 2822 head = createGroup(true); 2823 tail = root; 2824 head.next = expr(tail); 2825 break; 2826 case '=': // (?=xxx) and (?!xxx) lookahead 2827 case '!': 2828 head = createGroup(true); 2829 tail = root; 2830 head.next = expr(tail); 2831 if (ch == '=') { 2832 head = tail = new Pos(head); 2833 } else { 2834 head = tail = new Neg(head); 2835 } 2836 break; 2837 case '>': // (?>xxx) independent group 2838 head = createGroup(true); 2839 tail = root; 2840 head.next = expr(tail); 2841 head = tail = new Ques(head, INDEPENDENT); 2842 break; 2843 case '<': // (?<xxx) look behind 2844 ch = read(); 2845 if (ASCII.isLower(ch) || ASCII.isUpper(ch)) { 2846 // named captured group 2847 String name = groupname(ch); 2848 if (namedGroups().containsKey(name)) 2849 throw error("Named capturing group <" + name 2850 + "> is already defined"); 2851 capturingGroup = true; 2852 head = createGroup(false); 2853 tail = root; 2854 namedGroups().put(name, capturingGroupCount-1); 2855 head.next = expr(tail); 2856 break; 2857 } 2858 int start = cursor; 2859 head = createGroup(true); 2860 tail = root; 2861 head.next = expr(tail); 2862 tail.next = lookbehindEnd; 2863 TreeInfo info = new TreeInfo(); 2864 head.study(info); 2865 if (info.maxValid == false) { 2866 throw error("Look-behind group does not have " 2867 + "an obvious maximum length"); 2868 } 2869 boolean hasSupplementary = findSupplementary(start, patternLength); 2870 if (ch == '=') { 2871 head = tail = (hasSupplementary ? 2872 new BehindS(head, info.maxLength, 2873 info.minLength) : 2874 new Behind(head, info.maxLength, 2875 info.minLength)); 2876 } else if (ch == '!') { 2877 head = tail = (hasSupplementary ? 2878 new NotBehindS(head, info.maxLength, 2879 info.minLength) : 2880 new NotBehind(head, info.maxLength, 2881 info.minLength)); 2882 } else { 2883 throw error("Unknown look-behind group"); 2884 } 2885 break; 2886 case '$': 2887 case '@': 2888 throw error("Unknown group type"); 2889 default: // (?xxx:) inlined match flags 2890 unread(); 2891 addFlag(); 2892 ch = read(); 2893 if (ch == ')') { 2894 return null; // Inline modifier only 2895 } 2896 if (ch != ':') { 2897 throw error("Unknown inline modifier"); 2898 } 2899 head = createGroup(true); 2900 tail = root; 2901 head.next = expr(tail); 2902 break; 2903 } 2904 } else { // (xxx) a regular group 2905 capturingGroup = true; 2906 head = createGroup(false); 2907 tail = root; 2908 head.next = expr(tail); 2909 } 2910 2911 accept(')', "Unclosed group"); 2912 flags = save; 2913 2914 // Check for quantifiers 2915 Node node = closure(head); 2916 if (node == head) { // No closure 2917 root = tail; 2918 return node; // Dual return 2919 } 2920 if (head == tail) { // Zero length assertion 2921 root = node; 2922 return node; // Dual return 2923 } 2924 2925 if (node instanceof Ques) { 2926 Ques ques = (Ques) node; 2927 if (ques.type == POSSESSIVE) { 2928 root = node; 2929 return node; 2930 } 2931 tail.next = new BranchConn(); 2932 tail = tail.next; 2933 if (ques.type == GREEDY) { 2934 head = new Branch(head, null, tail); 2935 } else { // Reluctant quantifier 2936 head = new Branch(null, head, tail); 2937 } 2938 root = tail; 2939 return head; 2940 } else if (node instanceof Curly) { 2941 Curly curly = (Curly) node; 2942 if (curly.type == POSSESSIVE) { 2943 root = node; 2944 return node; 2945 } 2946 // Discover if the group is deterministic 2947 TreeInfo info = new TreeInfo(); 2948 if (head.study(info)) { // Deterministic 2949 GroupTail temp = (GroupTail) tail; 2950 head = root = new GroupCurly(head.next, curly.cmin, 2951 curly.cmax, curly.type, 2952 ((GroupTail)tail).localIndex, 2953 ((GroupTail)tail).groupIndex, 2954 capturingGroup); 2955 return head; 2956 } else { // Non-deterministic 2957 int temp = ((GroupHead) head).localIndex; 2958 Loop loop; 2959 if (curly.type == GREEDY) 2960 loop = new Loop(this.localCount, temp); 2961 else // Reluctant Curly 2962 loop = new LazyLoop(this.localCount, temp); 2963 Prolog prolog = new Prolog(loop); 2964 this.localCount += 1; 2965 loop.cmin = curly.cmin; 2966 loop.cmax = curly.cmax; 2967 loop.body = head; 2968 tail.next = loop; 2969 root = loop; 2970 return prolog; // Dual return 2971 } 2972 } 2973 throw error("Internal logic error"); 2974 } 2975 2976 /** 2977 * Create group head and tail nodes using double return. If the group is 2978 * created with anonymous true then it is a pure group and should not 2979 * affect group counting. 2980 */ 2981 private Node createGroup(boolean anonymous) { 2982 int localIndex = localCount++; 2983 int groupIndex = 0; 2984 if (!anonymous) 2985 groupIndex = capturingGroupCount++; 2986 GroupHead head = new GroupHead(localIndex); 2987 root = new GroupTail(localIndex, groupIndex); 2988 if (!anonymous && groupIndex < 10) 2989 groupNodes[groupIndex] = head; 2990 return head; 2991 } 2992 2993 @SuppressWarnings("fallthrough") 2994 /** 2995 * Parses inlined match flags and set them appropriately. 2996 */ 2997 private void addFlag() { 2998 int ch = peek(); 2999 for (;;) { 3000 switch (ch) { 3001 case 'i': 3002 flags |= CASE_INSENSITIVE; 3003 break; 3004 case 'm': 3005 flags |= MULTILINE; 3006 break; 3007 case 's': 3008 flags |= DOTALL; 3009 break; 3010 case 'd': 3011 flags |= UNIX_LINES; 3012 break; 3013 case 'u': 3014 flags |= UNICODE_CASE; 3015 break; 3016 case 'c': 3017 flags |= CANON_EQ; 3018 break; 3019 case 'x': 3020 flags |= COMMENTS; 3021 break; 3022 case 'U': 3023 flags |= (UNICODE_CHARACTER_CLASS | UNICODE_CASE); 3024 break; 3025 case '-': // subFlag then fall through 3026 ch = next(); 3027 subFlag(); 3028 default: 3029 return; 3030 } 3031 ch = next(); 3032 } 3033 } 3034 3035 @SuppressWarnings("fallthrough") 3036 /** 3037 * Parses the second part of inlined match flags and turns off 3038 * flags appropriately. 3039 */ 3040 private void subFlag() { 3041 int ch = peek(); 3042 for (;;) { 3043 switch (ch) { 3044 case 'i': 3045 flags &= ~CASE_INSENSITIVE; 3046 break; 3047 case 'm': 3048 flags &= ~MULTILINE; 3049 break; 3050 case 's': 3051 flags &= ~DOTALL; 3052 break; 3053 case 'd': 3054 flags &= ~UNIX_LINES; 3055 break; 3056 case 'u': 3057 flags &= ~UNICODE_CASE; 3058 break; 3059 case 'c': 3060 flags &= ~CANON_EQ; 3061 break; 3062 case 'x': 3063 flags &= ~COMMENTS; 3064 break; 3065 case 'U': 3066 flags &= ~(UNICODE_CHARACTER_CLASS | UNICODE_CASE); 3067 default: 3068 return; 3069 } 3070 ch = next(); 3071 } 3072 } 3073 3074 static final int MAX_REPS = 0x7FFFFFFF; 3075 3076 static final int GREEDY = 0; 3077 3078 static final int LAZY = 1; 3079 3080 static final int POSSESSIVE = 2; 3081 3082 static final int INDEPENDENT = 3; 3083 3084 /** 3085 * Processes repetition. If the next character peeked is a quantifier 3086 * then new nodes must be appended to handle the repetition. 3087 * Prev could be a single or a group, so it could be a chain of nodes. 3088 */ 3089 private Node closure(Node prev) { 3090 Node atom; 3091 int ch = peek(); 3092 switch (ch) { 3093 case '?': 3094 ch = next(); 3095 if (ch == '?') { 3096 next(); 3097 return new Ques(prev, LAZY); 3098 } else if (ch == '+') { 3099 next(); 3100 return new Ques(prev, POSSESSIVE); 3101 } 3102 return new Ques(prev, GREEDY); 3103 case '*': 3104 ch = next(); 3105 if (ch == '?') { 3106 next(); 3107 return new Curly(prev, 0, MAX_REPS, LAZY); 3108 } else if (ch == '+') { 3109 next(); 3110 return new Curly(prev, 0, MAX_REPS, POSSESSIVE); 3111 } 3112 return new Curly(prev, 0, MAX_REPS, GREEDY); 3113 case '+': 3114 ch = next(); 3115 if (ch == '?') { 3116 next(); 3117 return new Curly(prev, 1, MAX_REPS, LAZY); 3118 } else if (ch == '+') { 3119 next(); 3120 return new Curly(prev, 1, MAX_REPS, POSSESSIVE); 3121 } 3122 return new Curly(prev, 1, MAX_REPS, GREEDY); 3123 case '{': 3124 ch = temp[cursor+1]; 3125 if (ASCII.isDigit(ch)) { 3126 skip(); 3127 int cmin = 0; 3128 do { 3129 cmin = cmin * 10 + (ch - '0'); 3130 } while (ASCII.isDigit(ch = read())); 3131 int cmax = cmin; 3132 if (ch == ',') { 3133 ch = read(); 3134 cmax = MAX_REPS; 3135 if (ch != '}') { 3136 cmax = 0; 3137 while (ASCII.isDigit(ch)) { 3138 cmax = cmax * 10 + (ch - '0'); 3139 ch = read(); 3140 } 3141 } 3142 } 3143 if (ch != '}') 3144 throw error("Unclosed counted closure"); 3145 if (((cmin) | (cmax) | (cmax - cmin)) < 0) 3146 throw error("Illegal repetition range"); 3147 Curly curly; 3148 ch = peek(); 3149 if (ch == '?') { 3150 next(); 3151 curly = new Curly(prev, cmin, cmax, LAZY); 3152 } else if (ch == '+') { 3153 next(); 3154 curly = new Curly(prev, cmin, cmax, POSSESSIVE); 3155 } else { 3156 curly = new Curly(prev, cmin, cmax, GREEDY); 3157 } 3158 return curly; 3159 } else { 3160 throw error("Illegal repetition"); 3161 } 3162 default: 3163 return prev; 3164 } 3165 } 3166 3167 /** 3168 * Utility method for parsing control escape sequences. 3169 */ 3170 private int c() { 3171 if (cursor < patternLength) { 3172 return read() ^ 64; 3173 } 3174 throw error("Illegal control escape sequence"); 3175 } 3176 3177 /** 3178 * Utility method for parsing octal escape sequences. 3179 */ 3180 private int o() { 3181 int n = read(); 3182 if (((n-'0')|('7'-n)) >= 0) { 3183 int m = read(); 3184 if (((m-'0')|('7'-m)) >= 0) { 3185 int o = read(); 3186 if ((((o-'0')|('7'-o)) >= 0) && (((n-'0')|('3'-n)) >= 0)) { 3187 return (n - '0') * 64 + (m - '0') * 8 + (o - '0'); 3188 } 3189 unread(); 3190 return (n - '0') * 8 + (m - '0'); 3191 } 3192 unread(); 3193 return (n - '0'); 3194 } 3195 throw error("Illegal octal escape sequence"); 3196 } 3197 3198 /** 3199 * Utility method for parsing hexadecimal escape sequences. 3200 */ 3201 private int x() { 3202 int n = read(); 3203 if (ASCII.isHexDigit(n)) { 3204 int m = read(); 3205 if (ASCII.isHexDigit(m)) { 3206 return ASCII.toDigit(n) * 16 + ASCII.toDigit(m); 3207 } 3208 } else if (n == '{' && ASCII.isHexDigit(peek())) { 3209 int ch = 0; 3210 while (ASCII.isHexDigit(n = read())) { 3211 ch = (ch << 4) + ASCII.toDigit(n); 3212 if (ch > Character.MAX_CODE_POINT) 3213 throw error("Hexadecimal codepoint is too big"); 3214 } 3215 if (n != '}') 3216 throw error("Unclosed hexadecimal escape sequence"); 3217 return ch; 3218 } 3219 throw error("Illegal hexadecimal escape sequence"); 3220 } 3221 3222 /** 3223 * Utility method for parsing unicode escape sequences. 3224 */ 3225 private int cursor() { 3226 return cursor; 3227 } 3228 3229 private void setcursor(int pos) { 3230 cursor = pos; 3231 } 3232 3233 private int uxxxx() { 3234 int n = 0; 3235 for (int i = 0; i < 4; i++) { 3236 int ch = read(); 3237 if (!ASCII.isHexDigit(ch)) { 3238 throw error("Illegal Unicode escape sequence"); 3239 } 3240 n = n * 16 + ASCII.toDigit(ch); 3241 } 3242 return n; 3243 } 3244 3245 private int u() { 3246 int n = uxxxx(); 3247 if (Character.isHighSurrogate((char)n)) { 3248 int cur = cursor(); 3249 if (read() == '\\' && read() == 'u') { 3250 int n2 = uxxxx(); 3251 if (Character.isLowSurrogate((char)n2)) 3252 return Character.toCodePoint((char)n, (char)n2); 3253 } 3254 setcursor(cur); 3255 } 3256 return n; 3257 } 3258 3259 // 3260 // Utility methods for code point support 3261 // 3262 3263 private static final int countChars(CharSequence seq, int index, 3264 int lengthInCodePoints) { 3265 // optimization 3266 if (lengthInCodePoints == 1 && !Character.isHighSurrogate(seq.charAt(index))) { 3267 assert (index >= 0 && index < seq.length()); 3268 return 1; 3269 } 3270 int length = seq.length(); 3271 int x = index; 3272 if (lengthInCodePoints >= 0) { 3273 assert (index >= 0 && index < length); 3274 for (int i = 0; x < length && i < lengthInCodePoints; i++) { 3275 if (Character.isHighSurrogate(seq.charAt(x++))) { 3276 if (x < length && Character.isLowSurrogate(seq.charAt(x))) { 3277 x++; 3278 } 3279 } 3280 } 3281 return x - index; 3282 } 3283 3284 assert (index >= 0 && index <= length); 3285 if (index == 0) { 3286 return 0; 3287 } 3288 int len = -lengthInCodePoints; 3289 for (int i = 0; x > 0 && i < len; i++) { 3290 if (Character.isLowSurrogate(seq.charAt(--x))) { 3291 if (x > 0 && Character.isHighSurrogate(seq.charAt(x-1))) { 3292 x--; 3293 } 3294 } 3295 } 3296 return index - x; 3297 } 3298 3299 private static final int countCodePoints(CharSequence seq) { 3300 int length = seq.length(); 3301 int n = 0; 3302 for (int i = 0; i < length; ) { 3303 n++; 3304 if (Character.isHighSurrogate(seq.charAt(i++))) { 3305 if (i < length && Character.isLowSurrogate(seq.charAt(i))) { 3306 i++; 3307 } 3308 } 3309 } 3310 return n; 3311 } 3312 3313 /** 3314 * Creates a bit vector for matching Latin-1 values. A normal BitClass 3315 * never matches values above Latin-1, and a complemented BitClass always 3316 * matches values above Latin-1. 3317 */ 3318 private static final class BitClass extends BmpCharProperty { 3319 final boolean[] bits; 3320 BitClass() { bits = new boolean[256]; } 3321 private BitClass(boolean[] bits) { this.bits = bits; } 3322 BitClass add(int c, int flags) { 3323 assert c >= 0 && c <= 255; 3324 if ((flags & CASE_INSENSITIVE) != 0) { 3325 if (ASCII.isAscii(c)) { 3326 bits[ASCII.toUpper(c)] = true; 3327 bits[ASCII.toLower(c)] = true; 3328 } else if ((flags & UNICODE_CASE) != 0) { 3329 bits[Character.toLowerCase(c)] = true; 3330 bits[Character.toUpperCase(c)] = true; 3331 } 3332 } 3333 bits[c] = true; 3334 return this; 3335 } 3336 boolean isSatisfiedBy(int ch) { 3337 return ch < 256 && bits[ch]; 3338 } 3339 } 3340 3341 /** 3342 * Returns a suitably optimized, single character matcher. 3343 */ 3344 private CharProperty newSingle(final int ch) { 3345 if (has(CASE_INSENSITIVE)) { 3346 int lower, upper; 3347 if (has(UNICODE_CASE)) { 3348 upper = Character.toUpperCase(ch); 3349 lower = Character.toLowerCase(upper); 3350 if (upper != lower) 3351 return new SingleU(lower); 3352 } else if (ASCII.isAscii(ch)) { 3353 lower = ASCII.toLower(ch); 3354 upper = ASCII.toUpper(ch); 3355 if (lower != upper) 3356 return new SingleI(lower, upper); 3357 } 3358 } 3359 if (isSupplementary(ch)) 3360 return new SingleS(ch); // Match a given Unicode character 3361 return new Single(ch); // Match a given BMP character 3362 } 3363 3364 /** 3365 * Utility method for creating a string slice matcher. 3366 */ 3367 private Node newSlice(int[] buf, int count, boolean hasSupplementary) { 3368 int[] tmp = new int[count]; 3369 if (has(CASE_INSENSITIVE)) { 3370 if (has(UNICODE_CASE)) { 3371 for (int i = 0; i < count; i++) { 3372 tmp[i] = Character.toLowerCase( 3373 Character.toUpperCase(buf[i])); 3374 } 3375 return hasSupplementary? new SliceUS(tmp) : new SliceU(tmp); 3376 } 3377 for (int i = 0; i < count; i++) { 3378 tmp[i] = ASCII.toLower(buf[i]); 3379 } 3380 return hasSupplementary? new SliceIS(tmp) : new SliceI(tmp); 3381 } 3382 for (int i = 0; i < count; i++) { 3383 tmp[i] = buf[i]; 3384 } 3385 return hasSupplementary ? new SliceS(tmp) : new Slice(tmp); 3386 } 3387 3388 /** 3389 * The following classes are the building components of the object 3390 * tree that represents a compiled regular expression. The object tree 3391 * is made of individual elements that handle constructs in the Pattern. 3392 * Each type of object knows how to match its equivalent construct with 3393 * the match() method. 3394 */ 3395 3396 /** 3397 * Base class for all node classes. Subclasses should override the match() 3398 * method as appropriate. This class is an accepting node, so its match() 3399 * always returns true. 3400 */ 3401 static class Node extends Object { 3402 Node next; 3403 Node() { 3404 next = Pattern.accept; 3405 } 3406 /** 3407 * This method implements the classic accept node. 3408 */ 3409 boolean match(Matcher matcher, int i, CharSequence seq) { 3410 matcher.last = i; 3411 matcher.groups[0] = matcher.first; 3412 matcher.groups[1] = matcher.last; 3413 return true; 3414 } 3415 /** 3416 * This method is good for all zero length assertions. 3417 */ 3418 boolean study(TreeInfo info) { 3419 if (next != null) { 3420 return next.study(info); 3421 } else { 3422 return info.deterministic; 3423 } 3424 } 3425 } 3426 3427 static class LastNode extends Node { 3428 /** 3429 * This method implements the classic accept node with 3430 * the addition of a check to see if the match occurred 3431 * using all of the input. 3432 */ 3433 boolean match(Matcher matcher, int i, CharSequence seq) { 3434 if (matcher.acceptMode == Matcher.ENDANCHOR && i != matcher.to) 3435 return false; 3436 matcher.last = i; 3437 matcher.groups[0] = matcher.first; 3438 matcher.groups[1] = matcher.last; 3439 return true; 3440 } 3441 } 3442 3443 /** 3444 * Used for REs that can start anywhere within the input string. 3445 * This basically tries to match repeatedly at each spot in the 3446 * input string, moving forward after each try. An anchored search 3447 * or a BnM will bypass this node completely. 3448 */ 3449 static class Start extends Node { 3450 int minLength; 3451 Start(Node node) { 3452 this.next = node; 3453 TreeInfo info = new TreeInfo(); 3454 next.study(info); 3455 minLength = info.minLength; 3456 } 3457 boolean match(Matcher matcher, int i, CharSequence seq) { 3458 if (i > matcher.to - minLength) { 3459 matcher.hitEnd = true; 3460 return false; 3461 } 3462 int guard = matcher.to - minLength; 3463 for (; i <= guard; i++) { 3464 if (next.match(matcher, i, seq)) { 3465 matcher.first = i; 3466 matcher.groups[0] = matcher.first; 3467 matcher.groups[1] = matcher.last; 3468 return true; 3469 } 3470 } 3471 matcher.hitEnd = true; 3472 return false; 3473 } 3474 boolean study(TreeInfo info) { 3475 next.study(info); 3476 info.maxValid = false; 3477 info.deterministic = false; 3478 return false; 3479 } 3480 } 3481 3482 /* 3483 * StartS supports supplementary characters, including unpaired surrogates. 3484 */ 3485 static final class StartS extends Start { 3486 StartS(Node node) { 3487 super(node); 3488 } 3489 boolean match(Matcher matcher, int i, CharSequence seq) { 3490 if (i > matcher.to - minLength) { 3491 matcher.hitEnd = true; 3492 return false; 3493 } 3494 int guard = matcher.to - minLength; 3495 while (i <= guard) { 3496 //if ((ret = next.match(matcher, i, seq)) || i == guard) 3497 if (next.match(matcher, i, seq)) { 3498 matcher.first = i; 3499 matcher.groups[0] = matcher.first; 3500 matcher.groups[1] = matcher.last; 3501 return true; 3502 } 3503 if (i == guard) 3504 break; 3505 // Optimization to move to the next character. This is 3506 // faster than countChars(seq, i, 1). 3507 if (Character.isHighSurrogate(seq.charAt(i++))) { 3508 if (i < seq.length() && 3509 Character.isLowSurrogate(seq.charAt(i))) { 3510 i++; 3511 } 3512 } 3513 } 3514 matcher.hitEnd = true; 3515 return false; 3516 } 3517 } 3518 3519 /** 3520 * Node to anchor at the beginning of input. This object implements the 3521 * match for a \A sequence, and the caret anchor will use this if not in 3522 * multiline mode. 3523 */ 3524 static final class Begin extends Node { 3525 boolean match(Matcher matcher, int i, CharSequence seq) { 3526 int fromIndex = (matcher.anchoringBounds) ? 3527 matcher.from : 0; 3528 if (i == fromIndex && next.match(matcher, i, seq)) { 3529 matcher.first = i; 3530 matcher.groups[0] = i; 3531 matcher.groups[1] = matcher.last; 3532 return true; 3533 } else { 3534 return false; 3535 } 3536 } 3537 } 3538 3539 /** 3540 * Node to anchor at the end of input. This is the absolute end, so this 3541 * should not match at the last newline before the end as $ will. 3542 */ 3543 static final class End extends Node { 3544 boolean match(Matcher matcher, int i, CharSequence seq) { 3545 int endIndex = (matcher.anchoringBounds) ? 3546 matcher.to : matcher.getTextLength(); 3547 if (i == endIndex) { 3548 matcher.hitEnd = true; 3549 return next.match(matcher, i, seq); 3550 } 3551 return false; 3552 } 3553 } 3554 3555 /** 3556 * Node to anchor at the beginning of a line. This is essentially the 3557 * object to match for the multiline ^. 3558 */ 3559 static final class Caret extends Node { 3560 boolean match(Matcher matcher, int i, CharSequence seq) { 3561 int startIndex = matcher.from; 3562 int endIndex = matcher.to; 3563 if (!matcher.anchoringBounds) { 3564 startIndex = 0; 3565 endIndex = matcher.getTextLength(); 3566 } 3567 // Perl does not match ^ at end of input even after newline 3568 if (i == endIndex) { 3569 matcher.hitEnd = true; 3570 return false; 3571 } 3572 if (i > startIndex) { 3573 char ch = seq.charAt(i-1); 3574 if (ch != '\n' && ch != '\r' 3575 && (ch|1) != '\u2029' 3576 && ch != '\u0085' ) { 3577 return false; 3578 } 3579 // Should treat /r/n as one newline 3580 if (ch == '\r' && seq.charAt(i) == '\n') 3581 return false; 3582 } 3583 return next.match(matcher, i, seq); 3584 } 3585 } 3586 3587 /** 3588 * Node to anchor at the beginning of a line when in unixdot mode. 3589 */ 3590 static final class UnixCaret extends Node { 3591 boolean match(Matcher matcher, int i, CharSequence seq) { 3592 int startIndex = matcher.from; 3593 int endIndex = matcher.to; 3594 if (!matcher.anchoringBounds) { 3595 startIndex = 0; 3596 endIndex = matcher.getTextLength(); 3597 } 3598 // Perl does not match ^ at end of input even after newline 3599 if (i == endIndex) { 3600 matcher.hitEnd = true; 3601 return false; 3602 } 3603 if (i > startIndex) { 3604 char ch = seq.charAt(i-1); 3605 if (ch != '\n') { 3606 return false; 3607 } 3608 } 3609 return next.match(matcher, i, seq); 3610 } 3611 } 3612 3613 /** 3614 * Node to match the location where the last match ended. 3615 * This is used for the \G construct. 3616 */ 3617 static final class LastMatch extends Node { 3618 boolean match(Matcher matcher, int i, CharSequence seq) { 3619 if (i != matcher.oldLast) 3620 return false; 3621 return next.match(matcher, i, seq); 3622 } 3623 } 3624 3625 /** 3626 * Node to anchor at the end of a line or the end of input based on the 3627 * multiline mode. 3628 * 3629 * When not in multiline mode, the $ can only match at the very end 3630 * of the input, unless the input ends in a line terminator in which 3631 * it matches right before the last line terminator. 3632 * 3633 * Note that \r\n is considered an atomic line terminator. 3634 * 3635 * Like ^ the $ operator matches at a position, it does not match the 3636 * line terminators themselves. 3637 */ 3638 static final class Dollar extends Node { 3639 boolean multiline; 3640 Dollar(boolean mul) { 3641 multiline = mul; 3642 } 3643 boolean match(Matcher matcher, int i, CharSequence seq) { 3644 int endIndex = (matcher.anchoringBounds) ? 3645 matcher.to : matcher.getTextLength(); 3646 if (!multiline) { 3647 if (i < endIndex - 2) 3648 return false; 3649 if (i == endIndex - 2) { 3650 char ch = seq.charAt(i); 3651 if (ch != '\r') 3652 return false; 3653 ch = seq.charAt(i + 1); 3654 if (ch != '\n') 3655 return false; 3656 } 3657 } 3658 // Matches before any line terminator; also matches at the 3659 // end of input 3660 // Before line terminator: 3661 // If multiline, we match here no matter what 3662 // If not multiline, fall through so that the end 3663 // is marked as hit; this must be a /r/n or a /n 3664 // at the very end so the end was hit; more input 3665 // could make this not match here 3666 if (i < endIndex) { 3667 char ch = seq.charAt(i); 3668 if (ch == '\n') { 3669 // No match between \r\n 3670 if (i > 0 && seq.charAt(i-1) == '\r') 3671 return false; 3672 if (multiline) 3673 return next.match(matcher, i, seq); 3674 } else if (ch == '\r' || ch == '\u0085' || 3675 (ch|1) == '\u2029') { 3676 if (multiline) 3677 return next.match(matcher, i, seq); 3678 } else { // No line terminator, no match 3679 return false; 3680 } 3681 } 3682 // Matched at current end so hit end 3683 matcher.hitEnd = true; 3684 // If a $ matches because of end of input, then more input 3685 // could cause it to fail! 3686 matcher.requireEnd = true; 3687 return next.match(matcher, i, seq); 3688 } 3689 boolean study(TreeInfo info) { 3690 next.study(info); 3691 return info.deterministic; 3692 } 3693 } 3694 3695 /** 3696 * Node to anchor at the end of a line or the end of input based on the 3697 * multiline mode when in unix lines mode. 3698 */ 3699 static final class UnixDollar extends Node { 3700 boolean multiline; 3701 UnixDollar(boolean mul) { 3702 multiline = mul; 3703 } 3704 boolean match(Matcher matcher, int i, CharSequence seq) { 3705 int endIndex = (matcher.anchoringBounds) ? 3706 matcher.to : matcher.getTextLength(); 3707 if (i < endIndex) { 3708 char ch = seq.charAt(i); 3709 if (ch == '\n') { 3710 // If not multiline, then only possible to 3711 // match at very end or one before end 3712 if (multiline == false && i != endIndex - 1) 3713 return false; 3714 // If multiline return next.match without setting 3715 // matcher.hitEnd 3716 if (multiline) 3717 return next.match(matcher, i, seq); 3718 } else { 3719 return false; 3720 } 3721 } 3722 // Matching because at the end or 1 before the end; 3723 // more input could change this so set hitEnd 3724 matcher.hitEnd = true; 3725 // If a $ matches because of end of input, then more input 3726 // could cause it to fail! 3727 matcher.requireEnd = true; 3728 return next.match(matcher, i, seq); 3729 } 3730 boolean study(TreeInfo info) { 3731 next.study(info); 3732 return info.deterministic; 3733 } 3734 } 3735 3736 /** 3737 * Node class that matches a Unicode line ending '\R' 3738 */ 3739 static final class LineEnding extends Node { 3740 boolean match(Matcher matcher, int i, CharSequence seq) { 3741 // (u+000Du+000A|[u+000Au+000Bu+000Cu+000Du+0085u+2028u+2029]) 3742 if (i < matcher.to) { 3743 int ch = seq.charAt(i); 3744 if (ch == 0x0A || ch == 0x0B || ch == 0x0C || 3745 ch == 0x85 || ch == 0x2028 || ch == 0x2029) 3746 return next.match(matcher, i + 1, seq); 3747 if (ch == 0x0D) { 3748 i++; 3749 if (i < matcher.to && seq.charAt(i) == 0x0A) 3750 i++; 3751 return next.match(matcher, i, seq); 3752 } 3753 } else { 3754 matcher.hitEnd = true; 3755 } 3756 return false; 3757 } 3758 boolean study(TreeInfo info) { 3759 info.minLength++; 3760 info.maxLength += 2; 3761 return next.study(info); 3762 } 3763 } 3764 3765 /** 3766 * Abstract node class to match one character satisfying some 3767 * boolean property. 3768 */ 3769 private static abstract class CharProperty extends Node { 3770 abstract boolean isSatisfiedBy(int ch); 3771 CharProperty complement() { 3772 return new CharProperty() { 3773 boolean isSatisfiedBy(int ch) { 3774 return ! CharProperty.this.isSatisfiedBy(ch);}}; 3775 } 3776 boolean match(Matcher matcher, int i, CharSequence seq) { 3777 if (i < matcher.to) { 3778 int ch = Character.codePointAt(seq, i); 3779 return isSatisfiedBy(ch) 3780 && next.match(matcher, i+Character.charCount(ch), seq); 3781 } else { 3782 matcher.hitEnd = true; 3783 return false; 3784 } 3785 } 3786 boolean study(TreeInfo info) { 3787 info.minLength++; 3788 info.maxLength++; 3789 return next.study(info); 3790 } 3791 } 3792 3793 /** 3794 * Optimized version of CharProperty that works only for 3795 * properties never satisfied by Supplementary characters. 3796 */ 3797 private static abstract class BmpCharProperty extends CharProperty { 3798 boolean match(Matcher matcher, int i, CharSequence seq) { 3799 if (i < matcher.to) { 3800 return isSatisfiedBy(seq.charAt(i)) 3801 && next.match(matcher, i+1, seq); 3802 } else { 3803 matcher.hitEnd = true; 3804 return false; 3805 } 3806 } 3807 } 3808 3809 /** 3810 * Node class that matches a Supplementary Unicode character 3811 */ 3812 static final class SingleS extends CharProperty { 3813 final int c; 3814 SingleS(int c) { this.c = c; } 3815 boolean isSatisfiedBy(int ch) { 3816 return ch == c; 3817 } 3818 } 3819 3820 /** 3821 * Optimization -- matches a given BMP character 3822 */ 3823 static final class Single extends BmpCharProperty { 3824 final int c; 3825 Single(int c) { this.c = c; } 3826 boolean isSatisfiedBy(int ch) { 3827 return ch == c; 3828 } 3829 } 3830 3831 /** 3832 * Case insensitive matches a given BMP character 3833 */ 3834 static final class SingleI extends BmpCharProperty { 3835 final int lower; 3836 final int upper; 3837 SingleI(int lower, int upper) { 3838 this.lower = lower; 3839 this.upper = upper; 3840 } 3841 boolean isSatisfiedBy(int ch) { 3842 return ch == lower || ch == upper; 3843 } 3844 } 3845 3846 /** 3847 * Unicode case insensitive matches a given Unicode character 3848 */ 3849 static final class SingleU extends CharProperty { 3850 final int lower; 3851 SingleU(int lower) { 3852 this.lower = lower; 3853 } 3854 boolean isSatisfiedBy(int ch) { 3855 return lower == ch || 3856 lower == Character.toLowerCase(Character.toUpperCase(ch)); 3857 } 3858 } 3859 3860 /** 3861 * Node class that matches a Unicode block. 3862 */ 3863 static final class Block extends CharProperty { 3864 final Character.UnicodeBlock block; 3865 Block(Character.UnicodeBlock block) { 3866 this.block = block; 3867 } 3868 boolean isSatisfiedBy(int ch) { 3869 return block == Character.UnicodeBlock.of(ch); 3870 } 3871 } 3872 3873 /** 3874 * Node class that matches a Unicode script 3875 */ 3876 static final class Script extends CharProperty { 3877 final Character.UnicodeScript script; 3878 Script(Character.UnicodeScript script) { 3879 this.script = script; 3880 } 3881 boolean isSatisfiedBy(int ch) { 3882 return script == Character.UnicodeScript.of(ch); 3883 } 3884 } 3885 3886 /** 3887 * Node class that matches a Unicode category. 3888 */ 3889 static final class Category extends CharProperty { 3890 final int typeMask; 3891 Category(int typeMask) { this.typeMask = typeMask; } 3892 boolean isSatisfiedBy(int ch) { 3893 return (typeMask & (1 << Character.getType(ch))) != 0; 3894 } 3895 } 3896 3897 /** 3898 * Node class that matches a Unicode "type" 3899 */ 3900 static final class Utype extends CharProperty { 3901 final UnicodeProp uprop; 3902 Utype(UnicodeProp uprop) { this.uprop = uprop; } 3903 boolean isSatisfiedBy(int ch) { 3904 return uprop.is(ch); 3905 } 3906 } 3907 3908 /** 3909 * Node class that matches a POSIX type. 3910 */ 3911 static final class Ctype extends BmpCharProperty { 3912 final int ctype; 3913 Ctype(int ctype) { this.ctype = ctype; } 3914 boolean isSatisfiedBy(int ch) { 3915 return ch < 128 && ASCII.isType(ch, ctype); 3916 } 3917 } 3918 3919 /** 3920 * Node class that matches a Perl vertical whitespace 3921 */ 3922 static final class VertWS extends BmpCharProperty { 3923 boolean isSatisfiedBy(int cp) { 3924 return (cp >= 0x0A && cp <= 0x0D) || 3925 cp == 0x85 || cp == 0x2028 || cp == 0x2029; 3926 } 3927 } 3928 3929 /** 3930 * Node class that matches a Perl horizontal whitespace 3931 */ 3932 static final class HorizWS extends BmpCharProperty { 3933 boolean isSatisfiedBy(int cp) { 3934 return cp == 0x09 || cp == 0x20 || cp == 0xa0 || 3935 cp == 0x1680 || cp == 0x180e || 3936 cp >= 0x2000 && cp <= 0x200a || 3937 cp == 0x202f || cp == 0x205f || cp == 0x3000; 3938 } 3939 } 3940 3941 /** 3942 * Base class for all Slice nodes 3943 */ 3944 static class SliceNode extends Node { 3945 int[] buffer; 3946 SliceNode(int[] buf) { 3947 buffer = buf; 3948 } 3949 boolean study(TreeInfo info) { 3950 info.minLength += buffer.length; 3951 info.maxLength += buffer.length; 3952 return next.study(info); 3953 } 3954 } 3955 3956 /** 3957 * Node class for a case sensitive/BMP-only sequence of literal 3958 * characters. 3959 */ 3960 static final class Slice extends SliceNode { 3961 Slice(int[] buf) { 3962 super(buf); 3963 } 3964 boolean match(Matcher matcher, int i, CharSequence seq) { 3965 int[] buf = buffer; 3966 int len = buf.length; 3967 for (int j=0; j<len; j++) { 3968 if ((i+j) >= matcher.to) { 3969 matcher.hitEnd = true; 3970 return false; 3971 } 3972 if (buf[j] != seq.charAt(i+j)) 3973 return false; 3974 } 3975 return next.match(matcher, i+len, seq); 3976 } 3977 } 3978 3979 /** 3980 * Node class for a case_insensitive/BMP-only sequence of literal 3981 * characters. 3982 */ 3983 static class SliceI extends SliceNode { 3984 SliceI(int[] buf) { 3985 super(buf); 3986 } 3987 boolean match(Matcher matcher, int i, CharSequence seq) { 3988 int[] buf = buffer; 3989 int len = buf.length; 3990 for (int j=0; j<len; j++) { 3991 if ((i+j) >= matcher.to) { 3992 matcher.hitEnd = true; 3993 return false; 3994 } 3995 int c = seq.charAt(i+j); 3996 if (buf[j] != c && 3997 buf[j] != ASCII.toLower(c)) 3998 return false; 3999 } 4000 return next.match(matcher, i+len, seq); 4001 } 4002 } 4003 4004 /** 4005 * Node class for a unicode_case_insensitive/BMP-only sequence of 4006 * literal characters. Uses unicode case folding. 4007 */ 4008 static final class SliceU extends SliceNode { 4009 SliceU(int[] buf) { 4010 super(buf); 4011 } 4012 boolean match(Matcher matcher, int i, CharSequence seq) { 4013 int[] buf = buffer; 4014 int len = buf.length; 4015 for (int j=0; j<len; j++) { 4016 if ((i+j) >= matcher.to) { 4017 matcher.hitEnd = true; 4018 return false; 4019 } 4020 int c = seq.charAt(i+j); 4021 if (buf[j] != c && 4022 buf[j] != Character.toLowerCase(Character.toUpperCase(c))) 4023 return false; 4024 } 4025 return next.match(matcher, i+len, seq); 4026 } 4027 } 4028 4029 /** 4030 * Node class for a case sensitive sequence of literal characters 4031 * including supplementary characters. 4032 */ 4033 static final class SliceS extends SliceNode { 4034 SliceS(int[] buf) { 4035 super(buf); 4036 } 4037 boolean match(Matcher matcher, int i, CharSequence seq) { 4038 int[] buf = buffer; 4039 int x = i; 4040 for (int j = 0; j < buf.length; j++) { 4041 if (x >= matcher.to) { 4042 matcher.hitEnd = true; 4043 return false; 4044 } 4045 int c = Character.codePointAt(seq, x); 4046 if (buf[j] != c) 4047 return false; 4048 x += Character.charCount(c); 4049 if (x > matcher.to) { 4050 matcher.hitEnd = true; 4051 return false; 4052 } 4053 } 4054 return next.match(matcher, x, seq); 4055 } 4056 } 4057 4058 /** 4059 * Node class for a case insensitive sequence of literal characters 4060 * including supplementary characters. 4061 */ 4062 static class SliceIS extends SliceNode { 4063 SliceIS(int[] buf) { 4064 super(buf); 4065 } 4066 int toLower(int c) { 4067 return ASCII.toLower(c); 4068 } 4069 boolean match(Matcher matcher, int i, CharSequence seq) { 4070 int[] buf = buffer; 4071 int x = i; 4072 for (int j = 0; j < buf.length; j++) { 4073 if (x >= matcher.to) { 4074 matcher.hitEnd = true; 4075 return false; 4076 } 4077 int c = Character.codePointAt(seq, x); 4078 if (buf[j] != c && buf[j] != toLower(c)) 4079 return false; 4080 x += Character.charCount(c); 4081 if (x > matcher.to) { 4082 matcher.hitEnd = true; 4083 return false; 4084 } 4085 } 4086 return next.match(matcher, x, seq); 4087 } 4088 } 4089 4090 /** 4091 * Node class for a case insensitive sequence of literal characters. 4092 * Uses unicode case folding. 4093 */ 4094 static final class SliceUS extends SliceIS { 4095 SliceUS(int[] buf) { 4096 super(buf); 4097 } 4098 int toLower(int c) { 4099 return Character.toLowerCase(Character.toUpperCase(c)); 4100 } 4101 } 4102 4103 private static boolean inRange(int lower, int ch, int upper) { 4104 return lower <= ch && ch <= upper; 4105 } 4106 4107 /** 4108 * Returns node for matching characters within an explicit value range. 4109 */ 4110 private static CharProperty rangeFor(final int lower, 4111 final int upper) { 4112 return new CharProperty() { 4113 boolean isSatisfiedBy(int ch) { 4114 return inRange(lower, ch, upper);}}; 4115 } 4116 4117 /** 4118 * Returns node for matching characters within an explicit value 4119 * range in a case insensitive manner. 4120 */ 4121 private CharProperty caseInsensitiveRangeFor(final int lower, 4122 final int upper) { 4123 if (has(UNICODE_CASE)) 4124 return new CharProperty() { 4125 boolean isSatisfiedBy(int ch) { 4126 if (inRange(lower, ch, upper)) 4127 return true; 4128 int up = Character.toUpperCase(ch); 4129 return inRange(lower, up, upper) || 4130 inRange(lower, Character.toLowerCase(up), upper);}}; 4131 return new CharProperty() { 4132 boolean isSatisfiedBy(int ch) { 4133 return inRange(lower, ch, upper) || 4134 ASCII.isAscii(ch) && 4135 (inRange(lower, ASCII.toUpper(ch), upper) || 4136 inRange(lower, ASCII.toLower(ch), upper)); 4137 }}; 4138 } 4139 4140 /** 4141 * Implements the Unicode category ALL and the dot metacharacter when 4142 * in dotall mode. 4143 */ 4144 static final class All extends CharProperty { 4145 boolean isSatisfiedBy(int ch) { 4146 return true; 4147 } 4148 } 4149 4150 /** 4151 * Node class for the dot metacharacter when dotall is not enabled. 4152 */ 4153 static final class Dot extends CharProperty { 4154 boolean isSatisfiedBy(int ch) { 4155 return (ch != '\n' && ch != '\r' 4156 && (ch|1) != '\u2029' 4157 && ch != '\u0085'); 4158 } 4159 } 4160 4161 /** 4162 * Node class for the dot metacharacter when dotall is not enabled 4163 * but UNIX_LINES is enabled. 4164 */ 4165 static final class UnixDot extends CharProperty { 4166 boolean isSatisfiedBy(int ch) { 4167 return ch != '\n'; 4168 } 4169 } 4170 4171 /** 4172 * The 0 or 1 quantifier. This one class implements all three types. 4173 */ 4174 static final class Ques extends Node { 4175 Node atom; 4176 int type; 4177 Ques(Node node, int type) { 4178 this.atom = node; 4179 this.type = type; 4180 } 4181 boolean match(Matcher matcher, int i, CharSequence seq) { 4182 switch (type) { 4183 case GREEDY: 4184 return (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)) 4185 || next.match(matcher, i, seq); 4186 case LAZY: 4187 return next.match(matcher, i, seq) 4188 || (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)); 4189 case POSSESSIVE: 4190 if (atom.match(matcher, i, seq)) i = matcher.last; 4191 return next.match(matcher, i, seq); 4192 default: 4193 return atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq); 4194 } 4195 } 4196 boolean study(TreeInfo info) { 4197 if (type != INDEPENDENT) { 4198 int minL = info.minLength; 4199 atom.study(info); 4200 info.minLength = minL; 4201 info.deterministic = false; 4202 return next.study(info); 4203 } else { 4204 atom.study(info); 4205 return next.study(info); 4206 } 4207 } 4208 } 4209 4210 /** 4211 * Handles the curly-brace style repetition with a specified minimum and 4212 * maximum occurrences. The * quantifier is handled as a special case. 4213 * This class handles the three types. 4214 */ 4215 static final class Curly extends Node { 4216 Node atom; 4217 int type; 4218 int cmin; 4219 int cmax; 4220 4221 Curly(Node node, int cmin, int cmax, int type) { 4222 this.atom = node; 4223 this.type = type; 4224 this.cmin = cmin; 4225 this.cmax = cmax; 4226 } 4227 boolean match(Matcher matcher, int i, CharSequence seq) { 4228 int j; 4229 for (j = 0; j < cmin; j++) { 4230 if (atom.match(matcher, i, seq)) { 4231 i = matcher.last; 4232 continue; 4233 } 4234 return false; 4235 } 4236 if (type == GREEDY) 4237 return match0(matcher, i, j, seq); 4238 else if (type == LAZY) 4239 return match1(matcher, i, j, seq); 4240 else 4241 return match2(matcher, i, j, seq); 4242 } 4243 // Greedy match. 4244 // i is the index to start matching at 4245 // j is the number of atoms that have matched 4246 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4247 if (j >= cmax) { 4248 // We have matched the maximum... continue with the rest of 4249 // the regular expression 4250 return next.match(matcher, i, seq); 4251 } 4252 int backLimit = j; 4253 while (atom.match(matcher, i, seq)) { 4254 // k is the length of this match 4255 int k = matcher.last - i; 4256 if (k == 0) // Zero length match 4257 break; 4258 // Move up index and number matched 4259 i = matcher.last; 4260 j++; 4261 // We are greedy so match as many as we can 4262 while (j < cmax) { 4263 if (!atom.match(matcher, i, seq)) 4264 break; 4265 if (i + k != matcher.last) { 4266 if (match0(matcher, matcher.last, j+1, seq)) 4267 return true; 4268 break; 4269 } 4270 i += k; 4271 j++; 4272 } 4273 // Handle backing off if match fails 4274 while (j >= backLimit) { 4275 if (next.match(matcher, i, seq)) 4276 return true; 4277 i -= k; 4278 j--; 4279 } 4280 return false; 4281 } 4282 return next.match(matcher, i, seq); 4283 } 4284 // Reluctant match. At this point, the minimum has been satisfied. 4285 // i is the index to start matching at 4286 // j is the number of atoms that have matched 4287 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4288 for (;;) { 4289 // Try finishing match without consuming any more 4290 if (next.match(matcher, i, seq)) 4291 return true; 4292 // At the maximum, no match found 4293 if (j >= cmax) 4294 return false; 4295 // Okay, must try one more atom 4296 if (!atom.match(matcher, i, seq)) 4297 return false; 4298 // If we haven't moved forward then must break out 4299 if (i == matcher.last) 4300 return false; 4301 // Move up index and number matched 4302 i = matcher.last; 4303 j++; 4304 } 4305 } 4306 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4307 for (; j < cmax; j++) { 4308 if (!atom.match(matcher, i, seq)) 4309 break; 4310 if (i == matcher.last) 4311 break; 4312 i = matcher.last; 4313 } 4314 return next.match(matcher, i, seq); 4315 } 4316 boolean study(TreeInfo info) { 4317 // Save original info 4318 int minL = info.minLength; 4319 int maxL = info.maxLength; 4320 boolean maxV = info.maxValid; 4321 boolean detm = info.deterministic; 4322 info.reset(); 4323 4324 atom.study(info); 4325 4326 int temp = info.minLength * cmin + minL; 4327 if (temp < minL) { 4328 temp = 0xFFFFFFF; // arbitrary large number 4329 } 4330 info.minLength = temp; 4331 4332 if (maxV & info.maxValid) { 4333 temp = info.maxLength * cmax + maxL; 4334 info.maxLength = temp; 4335 if (temp < maxL) { 4336 info.maxValid = false; 4337 } 4338 } else { 4339 info.maxValid = false; 4340 } 4341 4342 if (info.deterministic && cmin == cmax) 4343 info.deterministic = detm; 4344 else 4345 info.deterministic = false; 4346 return next.study(info); 4347 } 4348 } 4349 4350 /** 4351 * Handles the curly-brace style repetition with a specified minimum and 4352 * maximum occurrences in deterministic cases. This is an iterative 4353 * optimization over the Prolog and Loop system which would handle this 4354 * in a recursive way. The * quantifier is handled as a special case. 4355 * If capture is true then this class saves group settings and ensures 4356 * that groups are unset when backing off of a group match. 4357 */ 4358 static final class GroupCurly extends Node { 4359 Node atom; 4360 int type; 4361 int cmin; 4362 int cmax; 4363 int localIndex; 4364 int groupIndex; 4365 boolean capture; 4366 4367 GroupCurly(Node node, int cmin, int cmax, int type, int local, 4368 int group, boolean capture) { 4369 this.atom = node; 4370 this.type = type; 4371 this.cmin = cmin; 4372 this.cmax = cmax; 4373 this.localIndex = local; 4374 this.groupIndex = group; 4375 this.capture = capture; 4376 } 4377 boolean match(Matcher matcher, int i, CharSequence seq) { 4378 int[] groups = matcher.groups; 4379 int[] locals = matcher.locals; 4380 int save0 = locals[localIndex]; 4381 int save1 = 0; 4382 int save2 = 0; 4383 4384 if (capture) { 4385 save1 = groups[groupIndex]; 4386 save2 = groups[groupIndex+1]; 4387 } 4388 4389 // Notify GroupTail there is no need to setup group info 4390 // because it will be set here 4391 locals[localIndex] = -1; 4392 4393 boolean ret = true; 4394 for (int j = 0; j < cmin; j++) { 4395 if (atom.match(matcher, i, seq)) { 4396 if (capture) { 4397 groups[groupIndex] = i; 4398 groups[groupIndex+1] = matcher.last; 4399 } 4400 i = matcher.last; 4401 } else { 4402 ret = false; 4403 break; 4404 } 4405 } 4406 if (ret) { 4407 if (type == GREEDY) { 4408 ret = match0(matcher, i, cmin, seq); 4409 } else if (type == LAZY) { 4410 ret = match1(matcher, i, cmin, seq); 4411 } else { 4412 ret = match2(matcher, i, cmin, seq); 4413 } 4414 } 4415 if (!ret) { 4416 locals[localIndex] = save0; 4417 if (capture) { 4418 groups[groupIndex] = save1; 4419 groups[groupIndex+1] = save2; 4420 } 4421 } 4422 return ret; 4423 } 4424 // Aggressive group match 4425 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4426 // don't back off passing the starting "j" 4427 int min = j; 4428 int[] groups = matcher.groups; 4429 int save0 = 0; 4430 int save1 = 0; 4431 if (capture) { 4432 save0 = groups[groupIndex]; 4433 save1 = groups[groupIndex+1]; 4434 } 4435 for (;;) { 4436 if (j >= cmax) 4437 break; 4438 if (!atom.match(matcher, i, seq)) 4439 break; 4440 int k = matcher.last - i; 4441 if (k <= 0) { 4442 if (capture) { 4443 groups[groupIndex] = i; 4444 groups[groupIndex+1] = i + k; 4445 } 4446 i = i + k; 4447 break; 4448 } 4449 for (;;) { 4450 if (capture) { 4451 groups[groupIndex] = i; 4452 groups[groupIndex+1] = i + k; 4453 } 4454 i = i + k; 4455 if (++j >= cmax) 4456 break; 4457 if (!atom.match(matcher, i, seq)) 4458 break; 4459 if (i + k != matcher.last) { 4460 if (match0(matcher, i, j, seq)) 4461 return true; 4462 break; 4463 } 4464 } 4465 while (j > min) { 4466 if (next.match(matcher, i, seq)) { 4467 if (capture) { 4468 groups[groupIndex+1] = i; 4469 groups[groupIndex] = i - k; 4470 } 4471 return true; 4472 } 4473 // backing off 4474 i = i - k; 4475 if (capture) { 4476 groups[groupIndex+1] = i; 4477 groups[groupIndex] = i - k; 4478 } 4479 j--; 4480 4481 } 4482 break; 4483 } 4484 if (capture) { 4485 groups[groupIndex] = save0; 4486 groups[groupIndex+1] = save1; 4487 } 4488 return next.match(matcher, i, seq); 4489 } 4490 // Reluctant matching 4491 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4492 for (;;) { 4493 if (next.match(matcher, i, seq)) 4494 return true; 4495 if (j >= cmax) 4496 return false; 4497 if (!atom.match(matcher, i, seq)) 4498 return false; 4499 if (i == matcher.last) 4500 return false; 4501 if (capture) { 4502 matcher.groups[groupIndex] = i; 4503 matcher.groups[groupIndex+1] = matcher.last; 4504 } 4505 i = matcher.last; 4506 j++; 4507 } 4508 } 4509 // Possessive matching 4510 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4511 for (; j < cmax; j++) { 4512 if (!atom.match(matcher, i, seq)) { 4513 break; 4514 } 4515 if (capture) { 4516 matcher.groups[groupIndex] = i; 4517 matcher.groups[groupIndex+1] = matcher.last; 4518 } 4519 if (i == matcher.last) { 4520 break; 4521 } 4522 i = matcher.last; 4523 } 4524 return next.match(matcher, i, seq); 4525 } 4526 boolean study(TreeInfo info) { 4527 // Save original info 4528 int minL = info.minLength; 4529 int maxL = info.maxLength; 4530 boolean maxV = info.maxValid; 4531 boolean detm = info.deterministic; 4532 info.reset(); 4533 4534 atom.study(info); 4535 4536 int temp = info.minLength * cmin + minL; 4537 if (temp < minL) { 4538 temp = 0xFFFFFFF; // Arbitrary large number 4539 } 4540 info.minLength = temp; 4541 4542 if (maxV & info.maxValid) { 4543 temp = info.maxLength * cmax + maxL; 4544 info.maxLength = temp; 4545 if (temp < maxL) { 4546 info.maxValid = false; 4547 } 4548 } else { 4549 info.maxValid = false; 4550 } 4551 4552 if (info.deterministic && cmin == cmax) { 4553 info.deterministic = detm; 4554 } else { 4555 info.deterministic = false; 4556 } 4557 return next.study(info); 4558 } 4559 } 4560 4561 /** 4562 * A Guard node at the end of each atom node in a Branch. It 4563 * serves the purpose of chaining the "match" operation to 4564 * "next" but not the "study", so we can collect the TreeInfo 4565 * of each atom node without including the TreeInfo of the 4566 * "next". 4567 */ 4568 static final class BranchConn extends Node { 4569 BranchConn() {}; 4570 boolean match(Matcher matcher, int i, CharSequence seq) { 4571 return next.match(matcher, i, seq); 4572 } 4573 boolean study(TreeInfo info) { 4574 return info.deterministic; 4575 } 4576 } 4577 4578 /** 4579 * Handles the branching of alternations. Note this is also used for 4580 * the ? quantifier to branch between the case where it matches once 4581 * and where it does not occur. 4582 */ 4583 static final class Branch extends Node { 4584 Node[] atoms = new Node[2]; 4585 int size = 2; 4586 Node conn; 4587 Branch(Node first, Node second, Node branchConn) { 4588 conn = branchConn; 4589 atoms[0] = first; 4590 atoms[1] = second; 4591 } 4592 4593 void add(Node node) { 4594 if (size >= atoms.length) { 4595 Node[] tmp = new Node[atoms.length*2]; 4596 System.arraycopy(atoms, 0, tmp, 0, atoms.length); 4597 atoms = tmp; 4598 } 4599 atoms[size++] = node; 4600 } 4601 4602 boolean match(Matcher matcher, int i, CharSequence seq) { 4603 for (int n = 0; n < size; n++) { 4604 if (atoms[n] == null) { 4605 if (conn.next.match(matcher, i, seq)) 4606 return true; 4607 } else if (atoms[n].match(matcher, i, seq)) { 4608 return true; 4609 } 4610 } 4611 return false; 4612 } 4613 4614 boolean study(TreeInfo info) { 4615 int minL = info.minLength; 4616 int maxL = info.maxLength; 4617 boolean maxV = info.maxValid; 4618 4619 int minL2 = Integer.MAX_VALUE; //arbitrary large enough num 4620 int maxL2 = -1; 4621 for (int n = 0; n < size; n++) { 4622 info.reset(); 4623 if (atoms[n] != null) 4624 atoms[n].study(info); 4625 minL2 = Math.min(minL2, info.minLength); 4626 maxL2 = Math.max(maxL2, info.maxLength); 4627 maxV = (maxV & info.maxValid); 4628 } 4629 4630 minL += minL2; 4631 maxL += maxL2; 4632 4633 info.reset(); 4634 conn.next.study(info); 4635 4636 info.minLength += minL; 4637 info.maxLength += maxL; 4638 info.maxValid &= maxV; 4639 info.deterministic = false; 4640 return false; 4641 } 4642 } 4643 4644 /** 4645 * The GroupHead saves the location where the group begins in the locals 4646 * and restores them when the match is done. 4647 * 4648 * The matchRef is used when a reference to this group is accessed later 4649 * in the expression. The locals will have a negative value in them to 4650 * indicate that we do not want to unset the group if the reference 4651 * doesn't match. 4652 */ 4653 static final class GroupHead extends Node { 4654 int localIndex; 4655 GroupHead(int localCount) { 4656 localIndex = localCount; 4657 } 4658 boolean match(Matcher matcher, int i, CharSequence seq) { 4659 int save = matcher.locals[localIndex]; 4660 matcher.locals[localIndex] = i; 4661 boolean ret = next.match(matcher, i, seq); 4662 matcher.locals[localIndex] = save; 4663 return ret; 4664 } 4665 boolean matchRef(Matcher matcher, int i, CharSequence seq) { 4666 int save = matcher.locals[localIndex]; 4667 matcher.locals[localIndex] = ~i; // HACK 4668 boolean ret = next.match(matcher, i, seq); 4669 matcher.locals[localIndex] = save; 4670 return ret; 4671 } 4672 } 4673 4674 /** 4675 * Recursive reference to a group in the regular expression. It calls 4676 * matchRef because if the reference fails to match we would not unset 4677 * the group. 4678 */ 4679 static final class GroupRef extends Node { 4680 GroupHead head; 4681 GroupRef(GroupHead head) { 4682 this.head = head; 4683 } 4684 boolean match(Matcher matcher, int i, CharSequence seq) { 4685 return head.matchRef(matcher, i, seq) 4686 && next.match(matcher, matcher.last, seq); 4687 } 4688 boolean study(TreeInfo info) { 4689 info.maxValid = false; 4690 info.deterministic = false; 4691 return next.study(info); 4692 } 4693 } 4694 4695 /** 4696 * The GroupTail handles the setting of group beginning and ending 4697 * locations when groups are successfully matched. It must also be able to 4698 * unset groups that have to be backed off of. 4699 * 4700 * The GroupTail node is also used when a previous group is referenced, 4701 * and in that case no group information needs to be set. 4702 */ 4703 static final class GroupTail extends Node { 4704 int localIndex; 4705 int groupIndex; 4706 GroupTail(int localCount, int groupCount) { 4707 localIndex = localCount; 4708 groupIndex = groupCount + groupCount; 4709 } 4710 boolean match(Matcher matcher, int i, CharSequence seq) { 4711 int tmp = matcher.locals[localIndex]; 4712 if (tmp >= 0) { // This is the normal group case. 4713 // Save the group so we can unset it if it 4714 // backs off of a match. 4715 int groupStart = matcher.groups[groupIndex]; 4716 int groupEnd = matcher.groups[groupIndex+1]; 4717 4718 matcher.groups[groupIndex] = tmp; 4719 matcher.groups[groupIndex+1] = i; 4720 if (next.match(matcher, i, seq)) { 4721 return true; 4722 } 4723 matcher.groups[groupIndex] = groupStart; 4724 matcher.groups[groupIndex+1] = groupEnd; 4725 return false; 4726 } else { 4727 // This is a group reference case. We don't need to save any 4728 // group info because it isn't really a group. 4729 matcher.last = i; 4730 return true; 4731 } 4732 } 4733 } 4734 4735 /** 4736 * This sets up a loop to handle a recursive quantifier structure. 4737 */ 4738 static final class Prolog extends Node { 4739 Loop loop; 4740 Prolog(Loop loop) { 4741 this.loop = loop; 4742 } 4743 boolean match(Matcher matcher, int i, CharSequence seq) { 4744 return loop.matchInit(matcher, i, seq); 4745 } 4746 boolean study(TreeInfo info) { 4747 return loop.study(info); 4748 } 4749 } 4750 4751 /** 4752 * Handles the repetition count for a greedy Curly. The matchInit 4753 * is called from the Prolog to save the index of where the group 4754 * beginning is stored. A zero length group check occurs in the 4755 * normal match but is skipped in the matchInit. 4756 */ 4757 static class Loop extends Node { 4758 Node body; 4759 int countIndex; // local count index in matcher locals 4760 int beginIndex; // group beginning index 4761 int cmin, cmax; 4762 Loop(int countIndex, int beginIndex) { 4763 this.countIndex = countIndex; 4764 this.beginIndex = beginIndex; 4765 } 4766 boolean match(Matcher matcher, int i, CharSequence seq) { 4767 // Avoid infinite loop in zero-length case. 4768 if (i > matcher.locals[beginIndex]) { 4769 int count = matcher.locals[countIndex]; 4770 4771 // This block is for before we reach the minimum 4772 // iterations required for the loop to match 4773 if (count < cmin) { 4774 matcher.locals[countIndex] = count + 1; 4775 boolean b = body.match(matcher, i, seq); 4776 // If match failed we must backtrack, so 4777 // the loop count should NOT be incremented 4778 if (!b) 4779 matcher.locals[countIndex] = count; 4780 // Return success or failure since we are under 4781 // minimum 4782 return b; 4783 } 4784 // This block is for after we have the minimum 4785 // iterations required for the loop to match 4786 if (count < cmax) { 4787 matcher.locals[countIndex] = count + 1; 4788 boolean b = body.match(matcher, i, seq); 4789 // If match failed we must backtrack, so 4790 // the loop count should NOT be incremented 4791 if (!b) 4792 matcher.locals[countIndex] = count; 4793 else 4794 return true; 4795 } 4796 } 4797 return next.match(matcher, i, seq); 4798 } 4799 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4800 int save = matcher.locals[countIndex]; 4801 boolean ret = false; 4802 if (0 < cmin) { 4803 matcher.locals[countIndex] = 1; 4804 ret = body.match(matcher, i, seq); 4805 } else if (0 < cmax) { 4806 matcher.locals[countIndex] = 1; 4807 ret = body.match(matcher, i, seq); 4808 if (ret == false) 4809 ret = next.match(matcher, i, seq); 4810 } else { 4811 ret = next.match(matcher, i, seq); 4812 } 4813 matcher.locals[countIndex] = save; 4814 return ret; 4815 } 4816 boolean study(TreeInfo info) { 4817 info.maxValid = false; 4818 info.deterministic = false; 4819 return false; 4820 } 4821 } 4822 4823 /** 4824 * Handles the repetition count for a reluctant Curly. The matchInit 4825 * is called from the Prolog to save the index of where the group 4826 * beginning is stored. A zero length group check occurs in the 4827 * normal match but is skipped in the matchInit. 4828 */ 4829 static final class LazyLoop extends Loop { 4830 LazyLoop(int countIndex, int beginIndex) { 4831 super(countIndex, beginIndex); 4832 } 4833 boolean match(Matcher matcher, int i, CharSequence seq) { 4834 // Check for zero length group 4835 if (i > matcher.locals[beginIndex]) { 4836 int count = matcher.locals[countIndex]; 4837 if (count < cmin) { 4838 matcher.locals[countIndex] = count + 1; 4839 boolean result = body.match(matcher, i, seq); 4840 // If match failed we must backtrack, so 4841 // the loop count should NOT be incremented 4842 if (!result) 4843 matcher.locals[countIndex] = count; 4844 return result; 4845 } 4846 if (next.match(matcher, i, seq)) 4847 return true; 4848 if (count < cmax) { 4849 matcher.locals[countIndex] = count + 1; 4850 boolean result = body.match(matcher, i, seq); 4851 // If match failed we must backtrack, so 4852 // the loop count should NOT be incremented 4853 if (!result) 4854 matcher.locals[countIndex] = count; 4855 return result; 4856 } 4857 return false; 4858 } 4859 return next.match(matcher, i, seq); 4860 } 4861 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4862 int save = matcher.locals[countIndex]; 4863 boolean ret = false; 4864 if (0 < cmin) { 4865 matcher.locals[countIndex] = 1; 4866 ret = body.match(matcher, i, seq); 4867 } else if (next.match(matcher, i, seq)) { 4868 ret = true; 4869 } else if (0 < cmax) { 4870 matcher.locals[countIndex] = 1; 4871 ret = body.match(matcher, i, seq); 4872 } 4873 matcher.locals[countIndex] = save; 4874 return ret; 4875 } 4876 boolean study(TreeInfo info) { 4877 info.maxValid = false; 4878 info.deterministic = false; 4879 return false; 4880 } 4881 } 4882 4883 /** 4884 * Refers to a group in the regular expression. Attempts to match 4885 * whatever the group referred to last matched. 4886 */ 4887 static class BackRef extends Node { 4888 int groupIndex; 4889 BackRef(int groupCount) { 4890 super(); 4891 groupIndex = groupCount + groupCount; 4892 } 4893 boolean match(Matcher matcher, int i, CharSequence seq) { 4894 int j = matcher.groups[groupIndex]; 4895 int k = matcher.groups[groupIndex+1]; 4896 4897 int groupSize = k - j; 4898 // If the referenced group didn't match, neither can this 4899 if (j < 0) 4900 return false; 4901 4902 // If there isn't enough input left no match 4903 if (i + groupSize > matcher.to) { 4904 matcher.hitEnd = true; 4905 return false; 4906 } 4907 // Check each new char to make sure it matches what the group 4908 // referenced matched last time around 4909 for (int index=0; index<groupSize; index++) 4910 if (seq.charAt(i+index) != seq.charAt(j+index)) 4911 return false; 4912 4913 return next.match(matcher, i+groupSize, seq); 4914 } 4915 boolean study(TreeInfo info) { 4916 info.maxValid = false; 4917 return next.study(info); 4918 } 4919 } 4920 4921 static class CIBackRef extends Node { 4922 int groupIndex; 4923 boolean doUnicodeCase; 4924 CIBackRef(int groupCount, boolean doUnicodeCase) { 4925 super(); 4926 groupIndex = groupCount + groupCount; 4927 this.doUnicodeCase = doUnicodeCase; 4928 } 4929 boolean match(Matcher matcher, int i, CharSequence seq) { 4930 int j = matcher.groups[groupIndex]; 4931 int k = matcher.groups[groupIndex+1]; 4932 4933 int groupSize = k - j; 4934 4935 // If the referenced group didn't match, neither can this 4936 if (j < 0) 4937 return false; 4938 4939 // If there isn't enough input left no match 4940 if (i + groupSize > matcher.to) { 4941 matcher.hitEnd = true; 4942 return false; 4943 } 4944 4945 // Check each new char to make sure it matches what the group 4946 // referenced matched last time around 4947 int x = i; 4948 for (int index=0; index<groupSize; index++) { 4949 int c1 = Character.codePointAt(seq, x); 4950 int c2 = Character.codePointAt(seq, j); 4951 if (c1 != c2) { 4952 if (doUnicodeCase) { 4953 int cc1 = Character.toUpperCase(c1); 4954 int cc2 = Character.toUpperCase(c2); 4955 if (cc1 != cc2 && 4956 Character.toLowerCase(cc1) != 4957 Character.toLowerCase(cc2)) 4958 return false; 4959 } else { 4960 if (ASCII.toLower(c1) != ASCII.toLower(c2)) 4961 return false; 4962 } 4963 } 4964 x += Character.charCount(c1); 4965 j += Character.charCount(c2); 4966 } 4967 4968 return next.match(matcher, i+groupSize, seq); 4969 } 4970 boolean study(TreeInfo info) { 4971 info.maxValid = false; 4972 return next.study(info); 4973 } 4974 } 4975 4976 /** 4977 * Searches until the next instance of its atom. This is useful for 4978 * finding the atom efficiently without passing an instance of it 4979 * (greedy problem) and without a lot of wasted search time (reluctant 4980 * problem). 4981 */ 4982 static final class First extends Node { 4983 Node atom; 4984 First(Node node) { 4985 this.atom = BnM.optimize(node); 4986 } 4987 boolean match(Matcher matcher, int i, CharSequence seq) { 4988 if (atom instanceof BnM) { 4989 return atom.match(matcher, i, seq) 4990 && next.match(matcher, matcher.last, seq); 4991 } 4992 for (;;) { 4993 if (i > matcher.to) { 4994 matcher.hitEnd = true; 4995 return false; 4996 } 4997 if (atom.match(matcher, i, seq)) { 4998 return next.match(matcher, matcher.last, seq); 4999 } 5000 i += countChars(seq, i, 1); 5001 matcher.first++; 5002 } 5003 } 5004 boolean study(TreeInfo info) { 5005 atom.study(info); 5006 info.maxValid = false; 5007 info.deterministic = false; 5008 return next.study(info); 5009 } 5010 } 5011 5012 static final class Conditional extends Node { 5013 Node cond, yes, not; 5014 Conditional(Node cond, Node yes, Node not) { 5015 this.cond = cond; 5016 this.yes = yes; 5017 this.not = not; 5018 } 5019 boolean match(Matcher matcher, int i, CharSequence seq) { 5020 if (cond.match(matcher, i, seq)) { 5021 return yes.match(matcher, i, seq); 5022 } else { 5023 return not.match(matcher, i, seq); 5024 } 5025 } 5026 boolean study(TreeInfo info) { 5027 int minL = info.minLength; 5028 int maxL = info.maxLength; 5029 boolean maxV = info.maxValid; 5030 info.reset(); 5031 yes.study(info); 5032 5033 int minL2 = info.minLength; 5034 int maxL2 = info.maxLength; 5035 boolean maxV2 = info.maxValid; 5036 info.reset(); 5037 not.study(info); 5038 5039 info.minLength = minL + Math.min(minL2, info.minLength); 5040 info.maxLength = maxL + Math.max(maxL2, info.maxLength); 5041 info.maxValid = (maxV & maxV2 & info.maxValid); 5042 info.deterministic = false; 5043 return next.study(info); 5044 } 5045 } 5046 5047 /** 5048 * Zero width positive lookahead. 5049 */ 5050 static final class Pos extends Node { 5051 Node cond; 5052 Pos(Node cond) { 5053 this.cond = cond; 5054 } 5055 boolean match(Matcher matcher, int i, CharSequence seq) { 5056 int savedTo = matcher.to; 5057 boolean conditionMatched = false; 5058 5059 // Relax transparent region boundaries for lookahead 5060 if (matcher.transparentBounds) 5061 matcher.to = matcher.getTextLength(); 5062 try { 5063 conditionMatched = cond.match(matcher, i, seq); 5064 } finally { 5065 // Reinstate region boundaries 5066 matcher.to = savedTo; 5067 } 5068 return conditionMatched && next.match(matcher, i, seq); 5069 } 5070 } 5071 5072 /** 5073 * Zero width negative lookahead. 5074 */ 5075 static final class Neg extends Node { 5076 Node cond; 5077 Neg(Node cond) { 5078 this.cond = cond; 5079 } 5080 boolean match(Matcher matcher, int i, CharSequence seq) { 5081 int savedTo = matcher.to; 5082 boolean conditionMatched = false; 5083 5084 // Relax transparent region boundaries for lookahead 5085 if (matcher.transparentBounds) 5086 matcher.to = matcher.getTextLength(); 5087 try { 5088 if (i < matcher.to) { 5089 conditionMatched = !cond.match(matcher, i, seq); 5090 } else { 5091 // If a negative lookahead succeeds then more input 5092 // could cause it to fail! 5093 matcher.requireEnd = true; 5094 conditionMatched = !cond.match(matcher, i, seq); 5095 } 5096 } finally { 5097 // Reinstate region boundaries 5098 matcher.to = savedTo; 5099 } 5100 return conditionMatched && next.match(matcher, i, seq); 5101 } 5102 } 5103 5104 /** 5105 * For use with lookbehinds; matches the position where the lookbehind 5106 * was encountered. 5107 */ 5108 static Node lookbehindEnd = new Node() { 5109 boolean match(Matcher matcher, int i, CharSequence seq) { 5110 return i == matcher.lookbehindTo; 5111 } 5112 }; 5113 5114 /** 5115 * Zero width positive lookbehind. 5116 */ 5117 static class Behind extends Node { 5118 Node cond; 5119 int rmax, rmin; 5120 Behind(Node cond, int rmax, int rmin) { 5121 this.cond = cond; 5122 this.rmax = rmax; 5123 this.rmin = rmin; 5124 } 5125 5126 boolean match(Matcher matcher, int i, CharSequence seq) { 5127 int savedFrom = matcher.from; 5128 boolean conditionMatched = false; 5129 int startIndex = (!matcher.transparentBounds) ? 5130 matcher.from : 0; 5131 int from = Math.max(i - rmax, startIndex); 5132 // Set end boundary 5133 int savedLBT = matcher.lookbehindTo; 5134 matcher.lookbehindTo = i; 5135 // Relax transparent region boundaries for lookbehind 5136 if (matcher.transparentBounds) 5137 matcher.from = 0; 5138 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 5139 conditionMatched = cond.match(matcher, j, seq); 5140 } 5141 matcher.from = savedFrom; 5142 matcher.lookbehindTo = savedLBT; 5143 return conditionMatched && next.match(matcher, i, seq); 5144 } 5145 } 5146 5147 /** 5148 * Zero width positive lookbehind, including supplementary 5149 * characters or unpaired surrogates. 5150 */ 5151 static final class BehindS extends Behind { 5152 BehindS(Node cond, int rmax, int rmin) { 5153 super(cond, rmax, rmin); 5154 } 5155 boolean match(Matcher matcher, int i, CharSequence seq) { 5156 int rmaxChars = countChars(seq, i, -rmax); 5157 int rminChars = countChars(seq, i, -rmin); 5158 int savedFrom = matcher.from; 5159 int startIndex = (!matcher.transparentBounds) ? 5160 matcher.from : 0; 5161 boolean conditionMatched = false; 5162 int from = Math.max(i - rmaxChars, startIndex); 5163 // Set end boundary 5164 int savedLBT = matcher.lookbehindTo; 5165 matcher.lookbehindTo = i; 5166 // Relax transparent region boundaries for lookbehind 5167 if (matcher.transparentBounds) 5168 matcher.from = 0; 5169 5170 for (int j = i - rminChars; 5171 !conditionMatched && j >= from; 5172 j -= j>from ? countChars(seq, j, -1) : 1) { 5173 conditionMatched = cond.match(matcher, j, seq); 5174 } 5175 matcher.from = savedFrom; 5176 matcher.lookbehindTo = savedLBT; 5177 return conditionMatched && next.match(matcher, i, seq); 5178 } 5179 } 5180 5181 /** 5182 * Zero width negative lookbehind. 5183 */ 5184 static class NotBehind extends Node { 5185 Node cond; 5186 int rmax, rmin; 5187 NotBehind(Node cond, int rmax, int rmin) { 5188 this.cond = cond; 5189 this.rmax = rmax; 5190 this.rmin = rmin; 5191 } 5192 5193 boolean match(Matcher matcher, int i, CharSequence seq) { 5194 int savedLBT = matcher.lookbehindTo; 5195 int savedFrom = matcher.from; 5196 boolean conditionMatched = false; 5197 int startIndex = (!matcher.transparentBounds) ? 5198 matcher.from : 0; 5199 int from = Math.max(i - rmax, startIndex); 5200 matcher.lookbehindTo = i; 5201 // Relax transparent region boundaries for lookbehind 5202 if (matcher.transparentBounds) 5203 matcher.from = 0; 5204 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 5205 conditionMatched = cond.match(matcher, j, seq); 5206 } 5207 // Reinstate region boundaries 5208 matcher.from = savedFrom; 5209 matcher.lookbehindTo = savedLBT; 5210 return !conditionMatched && next.match(matcher, i, seq); 5211 } 5212 } 5213 5214 /** 5215 * Zero width negative lookbehind, including supplementary 5216 * characters or unpaired surrogates. 5217 */ 5218 static final class NotBehindS extends NotBehind { 5219 NotBehindS(Node cond, int rmax, int rmin) { 5220 super(cond, rmax, rmin); 5221 } 5222 boolean match(Matcher matcher, int i, CharSequence seq) { 5223 int rmaxChars = countChars(seq, i, -rmax); 5224 int rminChars = countChars(seq, i, -rmin); 5225 int savedFrom = matcher.from; 5226 int savedLBT = matcher.lookbehindTo; 5227 boolean conditionMatched = false; 5228 int startIndex = (!matcher.transparentBounds) ? 5229 matcher.from : 0; 5230 int from = Math.max(i - rmaxChars, startIndex); 5231 matcher.lookbehindTo = i; 5232 // Relax transparent region boundaries for lookbehind 5233 if (matcher.transparentBounds) 5234 matcher.from = 0; 5235 for (int j = i - rminChars; 5236 !conditionMatched && j >= from; 5237 j -= j>from ? countChars(seq, j, -1) : 1) { 5238 conditionMatched = cond.match(matcher, j, seq); 5239 } 5240 //Reinstate region boundaries 5241 matcher.from = savedFrom; 5242 matcher.lookbehindTo = savedLBT; 5243 return !conditionMatched && next.match(matcher, i, seq); 5244 } 5245 } 5246 5247 /** 5248 * Returns the set union of two CharProperty nodes. 5249 */ 5250 private static CharProperty union(final CharProperty lhs, 5251 final CharProperty rhs) { 5252 return new CharProperty() { 5253 boolean isSatisfiedBy(int ch) { 5254 return lhs.isSatisfiedBy(ch) || rhs.isSatisfiedBy(ch);}}; 5255 } 5256 5257 /** 5258 * Returns the set intersection of two CharProperty nodes. 5259 */ 5260 private static CharProperty intersection(final CharProperty lhs, 5261 final CharProperty rhs) { 5262 return new CharProperty() { 5263 boolean isSatisfiedBy(int ch) { 5264 return lhs.isSatisfiedBy(ch) && rhs.isSatisfiedBy(ch);}}; 5265 } 5266 5267 /** 5268 * Returns the set difference of two CharProperty nodes. 5269 */ 5270 private static CharProperty setDifference(final CharProperty lhs, 5271 final CharProperty rhs) { 5272 return new CharProperty() { 5273 boolean isSatisfiedBy(int ch) { 5274 return ! rhs.isSatisfiedBy(ch) && lhs.isSatisfiedBy(ch);}}; 5275 } 5276 5277 /** 5278 * Handles word boundaries. Includes a field to allow this one class to 5279 * deal with the different types of word boundaries we can match. The word 5280 * characters include underscores, letters, and digits. Non spacing marks 5281 * can are also part of a word if they have a base character, otherwise 5282 * they are ignored for purposes of finding word boundaries. 5283 */ 5284 static final class Bound extends Node { 5285 static int LEFT = 0x1; 5286 static int RIGHT= 0x2; 5287 static int BOTH = 0x3; 5288 static int NONE = 0x4; 5289 int type; 5290 boolean useUWORD; 5291 Bound(int n, boolean useUWORD) { 5292 type = n; 5293 this.useUWORD = useUWORD; 5294 } 5295 5296 boolean isWord(int ch) { 5297 return useUWORD ? UnicodeProp.WORD.is(ch) 5298 : (ch == '_' || Character.isLetterOrDigit(ch)); 5299 } 5300 5301 int check(Matcher matcher, int i, CharSequence seq) { 5302 int ch; 5303 boolean left = false; 5304 int startIndex = matcher.from; 5305 int endIndex = matcher.to; 5306 if (matcher.transparentBounds) { 5307 startIndex = 0; 5308 endIndex = matcher.getTextLength(); 5309 } 5310 if (i > startIndex) { 5311 ch = Character.codePointBefore(seq, i); 5312 left = (isWord(ch) || 5313 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5314 && hasBaseCharacter(matcher, i-1, seq))); 5315 } 5316 boolean right = false; 5317 if (i < endIndex) { 5318 ch = Character.codePointAt(seq, i); 5319 right = (isWord(ch) || 5320 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5321 && hasBaseCharacter(matcher, i, seq))); 5322 } else { 5323 // Tried to access char past the end 5324 matcher.hitEnd = true; 5325 // The addition of another char could wreck a boundary 5326 matcher.requireEnd = true; 5327 } 5328 return ((left ^ right) ? (right ? LEFT : RIGHT) : NONE); 5329 } 5330 boolean match(Matcher matcher, int i, CharSequence seq) { 5331 return (check(matcher, i, seq) & type) > 0 5332 && next.match(matcher, i, seq); 5333 } 5334 } 5335 5336 /** 5337 * Non spacing marks only count as word characters in bounds calculations 5338 * if they have a base character. 5339 */ 5340 private static boolean hasBaseCharacter(Matcher matcher, int i, 5341 CharSequence seq) 5342 { 5343 int start = (!matcher.transparentBounds) ? 5344 matcher.from : 0; 5345 for (int x=i; x >= start; x--) { 5346 int ch = Character.codePointAt(seq, x); 5347 if (Character.isLetterOrDigit(ch)) 5348 return true; 5349 if (Character.getType(ch) == Character.NON_SPACING_MARK) 5350 continue; 5351 return false; 5352 } 5353 return false; 5354 } 5355 5356 /** 5357 * Attempts to match a slice in the input using the Boyer-Moore string 5358 * matching algorithm. The algorithm is based on the idea that the 5359 * pattern can be shifted farther ahead in the search text if it is 5360 * matched right to left. 5361 * <p> 5362 * The pattern is compared to the input one character at a time, from 5363 * the rightmost character in the pattern to the left. If the characters 5364 * all match the pattern has been found. If a character does not match, 5365 * the pattern is shifted right a distance that is the maximum of two 5366 * functions, the bad character shift and the good suffix shift. This 5367 * shift moves the attempted match position through the input more 5368 * quickly than a naive one position at a time check. 5369 * <p> 5370 * The bad character shift is based on the character from the text that 5371 * did not match. If the character does not appear in the pattern, the 5372 * pattern can be shifted completely beyond the bad character. If the 5373 * character does occur in the pattern, the pattern can be shifted to 5374 * line the pattern up with the next occurrence of that character. 5375 * <p> 5376 * The good suffix shift is based on the idea that some subset on the right 5377 * side of the pattern has matched. When a bad character is found, the 5378 * pattern can be shifted right by the pattern length if the subset does 5379 * not occur again in pattern, or by the amount of distance to the 5380 * next occurrence of the subset in the pattern. 5381 * 5382 * Boyer-Moore search methods adapted from code by Amy Yu. 5383 */ 5384 static class BnM extends Node { 5385 int[] buffer; 5386 int[] lastOcc; 5387 int[] optoSft; 5388 5389 /** 5390 * Pre calculates arrays needed to generate the bad character 5391 * shift and the good suffix shift. Only the last seven bits 5392 * are used to see if chars match; This keeps the tables small 5393 * and covers the heavily used ASCII range, but occasionally 5394 * results in an aliased match for the bad character shift. 5395 */ 5396 static Node optimize(Node node) { 5397 if (!(node instanceof Slice)) { 5398 return node; 5399 } 5400 5401 int[] src = ((Slice) node).buffer; 5402 int patternLength = src.length; 5403 // The BM algorithm requires a bit of overhead; 5404 // If the pattern is short don't use it, since 5405 // a shift larger than the pattern length cannot 5406 // be used anyway. 5407 if (patternLength < 4) { 5408 return node; 5409 } 5410 int i, j, k; 5411 int[] lastOcc = new int[128]; 5412 int[] optoSft = new int[patternLength]; 5413 // Precalculate part of the bad character shift 5414 // It is a table for where in the pattern each 5415 // lower 7-bit value occurs 5416 for (i = 0; i < patternLength; i++) { 5417 lastOcc[src[i]&0x7F] = i + 1; 5418 } 5419 // Precalculate the good suffix shift 5420 // i is the shift amount being considered 5421 NEXT: for (i = patternLength; i > 0; i--) { 5422 // j is the beginning index of suffix being considered 5423 for (j = patternLength - 1; j >= i; j--) { 5424 // Testing for good suffix 5425 if (src[j] == src[j-i]) { 5426 // src[j..len] is a good suffix 5427 optoSft[j-1] = i; 5428 } else { 5429 // No match. The array has already been 5430 // filled up with correct values before. 5431 continue NEXT; 5432 } 5433 } 5434 // This fills up the remaining of optoSft 5435 // any suffix can not have larger shift amount 5436 // then its sub-suffix. Why??? 5437 while (j > 0) { 5438 optoSft[--j] = i; 5439 } 5440 } 5441 // Set the guard value because of unicode compression 5442 optoSft[patternLength-1] = 1; 5443 if (node instanceof SliceS) 5444 return new BnMS(src, lastOcc, optoSft, node.next); 5445 return new BnM(src, lastOcc, optoSft, node.next); 5446 } 5447 BnM(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5448 this.buffer = src; 5449 this.lastOcc = lastOcc; 5450 this.optoSft = optoSft; 5451 this.next = next; 5452 } 5453 boolean match(Matcher matcher, int i, CharSequence seq) { 5454 int[] src = buffer; 5455 int patternLength = src.length; 5456 int last = matcher.to - patternLength; 5457 5458 // Loop over all possible match positions in text 5459 NEXT: while (i <= last) { 5460 // Loop over pattern from right to left 5461 for (int j = patternLength - 1; j >= 0; j--) { 5462 int ch = seq.charAt(i+j); 5463 if (ch != src[j]) { 5464 // Shift search to the right by the maximum of the 5465 // bad character shift and the good suffix shift 5466 i += Math.max(j + 1 - lastOcc[ch&0x7F], optoSft[j]); 5467 continue NEXT; 5468 } 5469 } 5470 // Entire pattern matched starting at i 5471 matcher.first = i; 5472 boolean ret = next.match(matcher, i + patternLength, seq); 5473 if (ret) { 5474 matcher.first = i; 5475 matcher.groups[0] = matcher.first; 5476 matcher.groups[1] = matcher.last; 5477 return true; 5478 } 5479 i++; 5480 } 5481 // BnM is only used as the leading node in the unanchored case, 5482 // and it replaced its Start() which always searches to the end 5483 // if it doesn't find what it's looking for, so hitEnd is true. 5484 matcher.hitEnd = true; 5485 return false; 5486 } 5487 boolean study(TreeInfo info) { 5488 info.minLength += buffer.length; 5489 info.maxValid = false; 5490 return next.study(info); 5491 } 5492 } 5493 5494 /** 5495 * Supplementary support version of BnM(). Unpaired surrogates are 5496 * also handled by this class. 5497 */ 5498 static final class BnMS extends BnM { 5499 int lengthInChars; 5500 5501 BnMS(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5502 super(src, lastOcc, optoSft, next); 5503 for (int x = 0; x < buffer.length; x++) { 5504 lengthInChars += Character.charCount(buffer[x]); 5505 } 5506 } 5507 boolean match(Matcher matcher, int i, CharSequence seq) { 5508 int[] src = buffer; 5509 int patternLength = src.length; 5510 int last = matcher.to - lengthInChars; 5511 5512 // Loop over all possible match positions in text 5513 NEXT: while (i <= last) { 5514 // Loop over pattern from right to left 5515 int ch; 5516 for (int j = countChars(seq, i, patternLength), x = patternLength - 1; 5517 j > 0; j -= Character.charCount(ch), x--) { 5518 ch = Character.codePointBefore(seq, i+j); 5519 if (ch != src[x]) { 5520 // Shift search to the right by the maximum of the 5521 // bad character shift and the good suffix shift 5522 int n = Math.max(x + 1 - lastOcc[ch&0x7F], optoSft[x]); 5523 i += countChars(seq, i, n); 5524 continue NEXT; 5525 } 5526 } 5527 // Entire pattern matched starting at i 5528 matcher.first = i; 5529 boolean ret = next.match(matcher, i + lengthInChars, seq); 5530 if (ret) { 5531 matcher.first = i; 5532 matcher.groups[0] = matcher.first; 5533 matcher.groups[1] = matcher.last; 5534 return true; 5535 } 5536 i += countChars(seq, i, 1); 5537 } 5538 matcher.hitEnd = true; 5539 return false; 5540 } 5541 } 5542 5543 /////////////////////////////////////////////////////////////////////////////// 5544 /////////////////////////////////////////////////////////////////////////////// 5545 5546 /** 5547 * This must be the very first initializer. 5548 */ 5549 static Node accept = new Node(); 5550 5551 static Node lastAccept = new LastNode(); 5552 5553 private static class CharPropertyNames { 5554 5555 static CharProperty charPropertyFor(String name) { 5556 CharPropertyFactory m = map.get(name); 5557 return m == null ? null : m.make(); 5558 } 5559 5560 private static abstract class CharPropertyFactory { 5561 abstract CharProperty make(); 5562 } 5563 5564 private static void defCategory(String name, 5565 final int typeMask) { 5566 map.put(name, new CharPropertyFactory() { 5567 CharProperty make() { return new Category(typeMask);}}); 5568 } 5569 5570 private static void defRange(String name, 5571 final int lower, final int upper) { 5572 map.put(name, new CharPropertyFactory() { 5573 CharProperty make() { return rangeFor(lower, upper);}}); 5574 } 5575 5576 private static void defCtype(String name, 5577 final int ctype) { 5578 map.put(name, new CharPropertyFactory() { 5579 CharProperty make() { return new Ctype(ctype);}}); 5580 } 5581 5582 private static abstract class CloneableProperty 5583 extends CharProperty implements Cloneable 5584 { 5585 public CloneableProperty clone() { 5586 try { 5587 return (CloneableProperty) super.clone(); 5588 } catch (CloneNotSupportedException e) { 5589 throw new AssertionError(e); 5590 } 5591 } 5592 } 5593 5594 private static void defClone(String name, 5595 final CloneableProperty p) { 5596 map.put(name, new CharPropertyFactory() { 5597 CharProperty make() { return p.clone();}}); 5598 } 5599 5600 private static final HashMap<String, CharPropertyFactory> map 5601 = new HashMap<>(); 5602 5603 static { 5604 // Unicode character property aliases, defined in 5605 // http://www.unicode.org/Public/UNIDATA/PropertyValueAliases.txt 5606 defCategory("Cn", 1<<Character.UNASSIGNED); 5607 defCategory("Lu", 1<<Character.UPPERCASE_LETTER); 5608 defCategory("Ll", 1<<Character.LOWERCASE_LETTER); 5609 defCategory("Lt", 1<<Character.TITLECASE_LETTER); 5610 defCategory("Lm", 1<<Character.MODIFIER_LETTER); 5611 defCategory("Lo", 1<<Character.OTHER_LETTER); 5612 defCategory("Mn", 1<<Character.NON_SPACING_MARK); 5613 defCategory("Me", 1<<Character.ENCLOSING_MARK); 5614 defCategory("Mc", 1<<Character.COMBINING_SPACING_MARK); 5615 defCategory("Nd", 1<<Character.DECIMAL_DIGIT_NUMBER); 5616 defCategory("Nl", 1<<Character.LETTER_NUMBER); 5617 defCategory("No", 1<<Character.OTHER_NUMBER); 5618 defCategory("Zs", 1<<Character.SPACE_SEPARATOR); 5619 defCategory("Zl", 1<<Character.LINE_SEPARATOR); 5620 defCategory("Zp", 1<<Character.PARAGRAPH_SEPARATOR); 5621 defCategory("Cc", 1<<Character.CONTROL); 5622 defCategory("Cf", 1<<Character.FORMAT); 5623 defCategory("Co", 1<<Character.PRIVATE_USE); 5624 defCategory("Cs", 1<<Character.SURROGATE); 5625 defCategory("Pd", 1<<Character.DASH_PUNCTUATION); 5626 defCategory("Ps", 1<<Character.START_PUNCTUATION); 5627 defCategory("Pe", 1<<Character.END_PUNCTUATION); 5628 defCategory("Pc", 1<<Character.CONNECTOR_PUNCTUATION); 5629 defCategory("Po", 1<<Character.OTHER_PUNCTUATION); 5630 defCategory("Sm", 1<<Character.MATH_SYMBOL); 5631 defCategory("Sc", 1<<Character.CURRENCY_SYMBOL); 5632 defCategory("Sk", 1<<Character.MODIFIER_SYMBOL); 5633 defCategory("So", 1<<Character.OTHER_SYMBOL); 5634 defCategory("Pi", 1<<Character.INITIAL_QUOTE_PUNCTUATION); 5635 defCategory("Pf", 1<<Character.FINAL_QUOTE_PUNCTUATION); 5636 defCategory("L", ((1<<Character.UPPERCASE_LETTER) | 5637 (1<<Character.LOWERCASE_LETTER) | 5638 (1<<Character.TITLECASE_LETTER) | 5639 (1<<Character.MODIFIER_LETTER) | 5640 (1<<Character.OTHER_LETTER))); 5641 defCategory("M", ((1<<Character.NON_SPACING_MARK) | 5642 (1<<Character.ENCLOSING_MARK) | 5643 (1<<Character.COMBINING_SPACING_MARK))); 5644 defCategory("N", ((1<<Character.DECIMAL_DIGIT_NUMBER) | 5645 (1<<Character.LETTER_NUMBER) | 5646 (1<<Character.OTHER_NUMBER))); 5647 defCategory("Z", ((1<<Character.SPACE_SEPARATOR) | 5648 (1<<Character.LINE_SEPARATOR) | 5649 (1<<Character.PARAGRAPH_SEPARATOR))); 5650 defCategory("C", ((1<<Character.CONTROL) | 5651 (1<<Character.FORMAT) | 5652 (1<<Character.PRIVATE_USE) | 5653 (1<<Character.SURROGATE))); // Other 5654 defCategory("P", ((1<<Character.DASH_PUNCTUATION) | 5655 (1<<Character.START_PUNCTUATION) | 5656 (1<<Character.END_PUNCTUATION) | 5657 (1<<Character.CONNECTOR_PUNCTUATION) | 5658 (1<<Character.OTHER_PUNCTUATION) | 5659 (1<<Character.INITIAL_QUOTE_PUNCTUATION) | 5660 (1<<Character.FINAL_QUOTE_PUNCTUATION))); 5661 defCategory("S", ((1<<Character.MATH_SYMBOL) | 5662 (1<<Character.CURRENCY_SYMBOL) | 5663 (1<<Character.MODIFIER_SYMBOL) | 5664 (1<<Character.OTHER_SYMBOL))); 5665 defCategory("LC", ((1<<Character.UPPERCASE_LETTER) | 5666 (1<<Character.LOWERCASE_LETTER) | 5667 (1<<Character.TITLECASE_LETTER))); 5668 defCategory("LD", ((1<<Character.UPPERCASE_LETTER) | 5669 (1<<Character.LOWERCASE_LETTER) | 5670 (1<<Character.TITLECASE_LETTER) | 5671 (1<<Character.MODIFIER_LETTER) | 5672 (1<<Character.OTHER_LETTER) | 5673 (1<<Character.DECIMAL_DIGIT_NUMBER))); 5674 defRange("L1", 0x00, 0xFF); // Latin-1 5675 map.put("all", new CharPropertyFactory() { 5676 CharProperty make() { return new All(); }}); 5677 5678 // Posix regular expression character classes, defined in 5679 // http://www.unix.org/onlinepubs/009695399/basedefs/xbd_chap09.html 5680 defRange("ASCII", 0x00, 0x7F); // ASCII 5681 defCtype("Alnum", ASCII.ALNUM); // Alphanumeric characters 5682 defCtype("Alpha", ASCII.ALPHA); // Alphabetic characters 5683 defCtype("Blank", ASCII.BLANK); // Space and tab characters 5684 defCtype("Cntrl", ASCII.CNTRL); // Control characters 5685 defRange("Digit", '0', '9'); // Numeric characters 5686 defCtype("Graph", ASCII.GRAPH); // printable and visible 5687 defRange("Lower", 'a', 'z'); // Lower-case alphabetic 5688 defRange("Print", 0x20, 0x7E); // Printable characters 5689 defCtype("Punct", ASCII.PUNCT); // Punctuation characters 5690 defCtype("Space", ASCII.SPACE); // Space characters 5691 defRange("Upper", 'A', 'Z'); // Upper-case alphabetic 5692 defCtype("XDigit",ASCII.XDIGIT); // hexadecimal digits 5693 5694 // Java character properties, defined by methods in Character.java 5695 defClone("javaLowerCase", new CloneableProperty() { 5696 boolean isSatisfiedBy(int ch) { 5697 return Character.isLowerCase(ch);}}); 5698 defClone("javaUpperCase", new CloneableProperty() { 5699 boolean isSatisfiedBy(int ch) { 5700 return Character.isUpperCase(ch);}}); 5701 defClone("javaAlphabetic", new CloneableProperty() { 5702 boolean isSatisfiedBy(int ch) { 5703 return Character.isAlphabetic(ch);}}); 5704 defClone("javaIdeographic", new CloneableProperty() { 5705 boolean isSatisfiedBy(int ch) { 5706 return Character.isIdeographic(ch);}}); 5707 defClone("javaTitleCase", new CloneableProperty() { 5708 boolean isSatisfiedBy(int ch) { 5709 return Character.isTitleCase(ch);}}); 5710 defClone("javaDigit", new CloneableProperty() { 5711 boolean isSatisfiedBy(int ch) { 5712 return Character.isDigit(ch);}}); 5713 defClone("javaDefined", new CloneableProperty() { 5714 boolean isSatisfiedBy(int ch) { 5715 return Character.isDefined(ch);}}); 5716 defClone("javaLetter", new CloneableProperty() { 5717 boolean isSatisfiedBy(int ch) { 5718 return Character.isLetter(ch);}}); 5719 defClone("javaLetterOrDigit", new CloneableProperty() { 5720 boolean isSatisfiedBy(int ch) { 5721 return Character.isLetterOrDigit(ch);}}); 5722 defClone("javaJavaIdentifierStart", new CloneableProperty() { 5723 boolean isSatisfiedBy(int ch) { 5724 return Character.isJavaIdentifierStart(ch);}}); 5725 defClone("javaJavaIdentifierPart", new CloneableProperty() { 5726 boolean isSatisfiedBy(int ch) { 5727 return Character.isJavaIdentifierPart(ch);}}); 5728 defClone("javaUnicodeIdentifierStart", new CloneableProperty() { 5729 boolean isSatisfiedBy(int ch) { 5730 return Character.isUnicodeIdentifierStart(ch);}}); 5731 defClone("javaUnicodeIdentifierPart", new CloneableProperty() { 5732 boolean isSatisfiedBy(int ch) { 5733 return Character.isUnicodeIdentifierPart(ch);}}); 5734 defClone("javaIdentifierIgnorable", new CloneableProperty() { 5735 boolean isSatisfiedBy(int ch) { 5736 return Character.isIdentifierIgnorable(ch);}}); 5737 defClone("javaSpaceChar", new CloneableProperty() { 5738 boolean isSatisfiedBy(int ch) { 5739 return Character.isSpaceChar(ch);}}); 5740 defClone("javaWhitespace", new CloneableProperty() { 5741 boolean isSatisfiedBy(int ch) { 5742 return Character.isWhitespace(ch);}}); 5743 defClone("javaISOControl", new CloneableProperty() { 5744 boolean isSatisfiedBy(int ch) { 5745 return Character.isISOControl(ch);}}); 5746 defClone("javaMirrored", new CloneableProperty() { 5747 boolean isSatisfiedBy(int ch) { 5748 return Character.isMirrored(ch);}}); 5749 } 5750 } 5751 5752 /** 5753 * Creates a predicate which can be used to match a string. 5754 * 5755 * @return The predicate which can be used for matching on a string 5756 * @since 1.8 5757 */ 5758 public Predicate<String> asPredicate() { 5759 return s -> matcher(s).find(); 5760 } 5761 5762 /** 5763 * Creates a stream from the given input sequence around matches of this 5764 * pattern. 5765 * 5766 * <p> The stream returned by this method contains each substring of the 5767 * input sequence that is terminated by another subsequence that matches 5768 * this pattern or is terminated by the end of the input sequence. The 5769 * substrings in the stream are in the order in which they occur in the 5770 * input. Trailing empty strings will be discarded and not encountered in 5771 * the stream. 5772 * 5773 * <p> If this pattern does not match any subsequence of the input then 5774 * the resulting stream has just one element, namely the input sequence in 5775 * string form. 5776 * 5777 * <p> A zero-length input sequence always results an empty stream. 5778 * 5779 * <p> When there is a positive-width match at the beginning of the input 5780 * sequence then an empty leading substring is included at the beginning 5781 * of the stream. A zero-width match at the beginning however never produces 5782 * such empty leading substring. 5783 * 5784 * <p> If the input sequence is mutable, it must remain constant during the 5785 * execution of the terminal stream operation. Otherwise, the result of the 5786 * terminal stream operation is undefined. 5787 * 5788 * @param input 5789 * The character sequence to be split 5790 * 5791 * @return The stream of strings computed by splitting the input 5792 * around matches of this pattern 5793 * @see #split(CharSequence) 5794 * @since 1.8 5795 */ 5796 public Stream<String> splitAsStream(final CharSequence input) { 5797 class MatcherIterator implements Iterator<String> { 5798 private final Matcher matcher; 5799 // The start position of the next sub-sequence of input 5800 // when current == input.length there are no more elements 5801 private int current; 5802 // null if the next element, if any, needs to obtained 5803 private String nextElement; 5804 // > 0 if there are N next empty elements 5805 private int emptyElementCount; 5806 5807 MatcherIterator() { 5808 this.matcher = matcher(input); 5809 } 5810 5811 public String next() { 5812 if (!hasNext()) 5813 throw new NoSuchElementException(); 5814 5815 if (emptyElementCount == 0) { 5816 String n = nextElement; 5817 nextElement = null; 5818 return n; 5819 } else { 5820 emptyElementCount--; 5821 return ""; 5822 } 5823 } 5824 5825 public boolean hasNext() { 5826 if (nextElement != null || emptyElementCount > 0) 5827 return true; 5828 5829 if (current == input.length()) 5830 return false; 5831 5832 // Consume the next matching element 5833 // Count sequence of matching empty elements 5834 while (matcher.find()) { 5835 nextElement = input.subSequence(current, matcher.start()).toString(); 5836 current = matcher.end(); 5837 if (!nextElement.isEmpty()) { 5838 return true; 5839 } else if (current > 0) { // no empty leading substring for zero-width 5840 // match at the beginning of the input 5841 emptyElementCount++; 5842 } 5843 } 5844 5845 // Consume last matching element 5846 nextElement = input.subSequence(current, input.length()).toString(); 5847 current = input.length(); 5848 if (!nextElement.isEmpty()) { 5849 return true; 5850 } else { 5851 // Ignore a terminal sequence of matching empty elements 5852 emptyElementCount = 0; 5853 nextElement = null; 5854 return false; 5855 } 5856 } 5857 } 5858 return StreamSupport.stream(Spliterators.spliteratorUnknownSize( 5859 new MatcherIterator(), Spliterator.ORDERED | Spliterator.NONNULL), false); 5860 } 5861 }