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. 1148 * 1149 * <p> When there is a positive-width match at the beginning of the input 1150 * sequence then an empty leading substring is included at the beginning 1151 * of the resulting array. A zero-width match at the beginning however 1152 * never produces such empty leading substring. 1153 * 1154 * <p> The <tt>limit</tt> parameter controls the number of times the 1155 * pattern is applied and therefore affects the length of the resulting 1156 * array. If the limit <i>n</i> is greater than zero then the pattern 1157 * will be applied at most <i>n</i> - 1 times, the array's 1158 * length will be no greater than <i>n</i>, and the array's last entry 1159 * will contain all input beyond the last matched delimiter. If <i>n</i> 1160 * is non-positive then the pattern will be applied as many times as 1161 * possible and the array can have any length. If <i>n</i> is zero then 1162 * the pattern will be applied as many times as possible, the array can 1163 * have any length, and trailing empty strings will be discarded. 1164 * 1165 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1166 * results with these parameters: 1167 * 1168 * <blockquote><table cellpadding=1 cellspacing=0 1169 * summary="Split examples showing regex, limit, and result"> 1170 * <tr><th align="left"><i>Regex </i></th> 1171 * <th align="left"><i>Limit </i></th> 1172 * <th align="left"><i>Result </i></th></tr> 1173 * <tr><td align=center>:</td> 1174 * <td align=center>2</td> 1175 * <td><tt>{ "boo", "and:foo" }</tt></td></tr> 1176 * <tr><td align=center>:</td> 1177 * <td align=center>5</td> 1178 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1179 * <tr><td align=center>:</td> 1180 * <td align=center>-2</td> 1181 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1182 * <tr><td align=center>o</td> 1183 * <td align=center>5</td> 1184 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1185 * <tr><td align=center>o</td> 1186 * <td align=center>-2</td> 1187 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1188 * <tr><td align=center>o</td> 1189 * <td align=center>0</td> 1190 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1191 * </table></blockquote> 1192 * 1193 * @param input 1194 * The character sequence to be split 1195 * 1196 * @param limit 1197 * The result threshold, as described above 1198 * 1199 * @return The array of strings computed by splitting the input 1200 * around matches of this pattern 1201 */ 1202 public String[] split(CharSequence input, int limit) { 1203 int index = 0; 1204 boolean matchLimited = limit > 0; 1205 ArrayList<String> matchList = new ArrayList<>(); 1206 Matcher m = matcher(input); 1207 1208 // Add segments before each match found 1209 while(m.find()) { 1210 if (!matchLimited || matchList.size() < limit - 1) { 1211 if (index == 0 && index == m.start() && m.start() == m.end()) { 1212 // no empty leading substring included for zero-width match 1213 // at the beginning of the input char sequence. 1214 continue; 1215 } 1216 String match = input.subSequence(index, m.start()).toString(); 1217 matchList.add(match); 1218 index = m.end(); 1219 } else if (matchList.size() == limit - 1) { // last one 1220 String match = input.subSequence(index, 1221 input.length()).toString(); 1222 matchList.add(match); 1223 index = m.end(); 1224 } 1225 } 1226 1227 // If no match was found, return this 1228 if (index == 0) 1229 return new String[] {input.toString()}; 1230 1231 // Add remaining segment 1232 if (!matchLimited || matchList.size() < limit) 1233 matchList.add(input.subSequence(index, input.length()).toString()); 1234 1235 // Construct result 1236 int resultSize = matchList.size(); 1237 if (limit == 0) 1238 while (resultSize > 0 && matchList.get(resultSize-1).equals("")) 1239 resultSize--; 1240 String[] result = new String[resultSize]; 1241 return matchList.subList(0, resultSize).toArray(result); 1242 } 1243 1244 /** 1245 * Splits the given input sequence around matches of this pattern. 1246 * 1247 * <p> This method works as if by invoking the two-argument {@link 1248 * #split(java.lang.CharSequence, int) split} method with the given input 1249 * sequence and a limit argument of zero. Trailing empty strings are 1250 * therefore not included in the resulting array. </p> 1251 * 1252 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1253 * results with these expressions: 1254 * 1255 * <blockquote><table cellpadding=1 cellspacing=0 1256 * summary="Split examples showing regex and result"> 1257 * <tr><th align="left"><i>Regex </i></th> 1258 * <th align="left"><i>Result</i></th></tr> 1259 * <tr><td align=center>:</td> 1260 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1261 * <tr><td align=center>o</td> 1262 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1263 * </table></blockquote> 1264 * 1265 * 1266 * @param input 1267 * The character sequence to be split 1268 * 1269 * @return The array of strings computed by splitting the input 1270 * around matches of this pattern 1271 */ 1272 public String[] split(CharSequence input) { 1273 return split(input, 0); 1274 } 1275 1276 /** 1277 * Returns a literal pattern <code>String</code> for the specified 1278 * <code>String</code>. 1279 * 1280 * <p>This method produces a <code>String</code> that can be used to 1281 * create a <code>Pattern</code> that would match the string 1282 * <code>s</code> as if it were a literal pattern.</p> Metacharacters 1283 * or escape sequences in the input sequence will be given no special 1284 * meaning. 1285 * 1286 * @param s The string to be literalized 1287 * @return A literal string replacement 1288 * @since 1.5 1289 */ 1290 public static String quote(String s) { 1291 int slashEIndex = s.indexOf("\\E"); 1292 if (slashEIndex == -1) 1293 return "\\Q" + s + "\\E"; 1294 1295 StringBuilder sb = new StringBuilder(s.length() * 2); 1296 sb.append("\\Q"); 1297 slashEIndex = 0; 1298 int current = 0; 1299 while ((slashEIndex = s.indexOf("\\E", current)) != -1) { 1300 sb.append(s.substring(current, slashEIndex)); 1301 current = slashEIndex + 2; 1302 sb.append("\\E\\\\E\\Q"); 1303 } 1304 sb.append(s.substring(current, s.length())); 1305 sb.append("\\E"); 1306 return sb.toString(); 1307 } 1308 1309 /** 1310 * Recompile the Pattern instance from a stream. The original pattern 1311 * string is read in and the object tree is recompiled from it. 1312 */ 1313 private void readObject(java.io.ObjectInputStream s) 1314 throws java.io.IOException, ClassNotFoundException { 1315 1316 // Read in all fields 1317 s.defaultReadObject(); 1318 1319 // Initialize counts 1320 capturingGroupCount = 1; 1321 localCount = 0; 1322 1323 // if length > 0, the Pattern is lazily compiled 1324 compiled = false; 1325 if (pattern.length() == 0) { 1326 root = new Start(lastAccept); 1327 matchRoot = lastAccept; 1328 compiled = true; 1329 } 1330 } 1331 1332 /** 1333 * This private constructor is used to create all Patterns. The pattern 1334 * string and match flags are all that is needed to completely describe 1335 * a Pattern. An empty pattern string results in an object tree with 1336 * only a Start node and a LastNode node. 1337 */ 1338 private Pattern(String p, int f) { 1339 pattern = p; 1340 flags = f; 1341 1342 // to use UNICODE_CASE if UNICODE_CHARACTER_CLASS present 1343 if ((flags & UNICODE_CHARACTER_CLASS) != 0) 1344 flags |= UNICODE_CASE; 1345 1346 // Reset group index count 1347 capturingGroupCount = 1; 1348 localCount = 0; 1349 1350 if (pattern.length() > 0) { 1351 compile(); 1352 } else { 1353 root = new Start(lastAccept); 1354 matchRoot = lastAccept; 1355 } 1356 } 1357 1358 /** 1359 * The pattern is converted to normalizedD form and then a pure group 1360 * is constructed to match canonical equivalences of the characters. 1361 */ 1362 private void normalize() { 1363 boolean inCharClass = false; 1364 int lastCodePoint = -1; 1365 1366 // Convert pattern into normalizedD form 1367 normalizedPattern = Normalizer.normalize(pattern, Normalizer.Form.NFD); 1368 patternLength = normalizedPattern.length(); 1369 1370 // Modify pattern to match canonical equivalences 1371 StringBuilder newPattern = new StringBuilder(patternLength); 1372 for(int i=0; i<patternLength; ) { 1373 int c = normalizedPattern.codePointAt(i); 1374 StringBuilder sequenceBuffer; 1375 if ((Character.getType(c) == Character.NON_SPACING_MARK) 1376 && (lastCodePoint != -1)) { 1377 sequenceBuffer = new StringBuilder(); 1378 sequenceBuffer.appendCodePoint(lastCodePoint); 1379 sequenceBuffer.appendCodePoint(c); 1380 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1381 i += Character.charCount(c); 1382 if (i >= patternLength) 1383 break; 1384 c = normalizedPattern.codePointAt(i); 1385 sequenceBuffer.appendCodePoint(c); 1386 } 1387 String ea = produceEquivalentAlternation( 1388 sequenceBuffer.toString()); 1389 newPattern.setLength(newPattern.length()-Character.charCount(lastCodePoint)); 1390 newPattern.append("(?:").append(ea).append(")"); 1391 } else if (c == '[' && lastCodePoint != '\\') { 1392 i = normalizeCharClass(newPattern, i); 1393 } else { 1394 newPattern.appendCodePoint(c); 1395 } 1396 lastCodePoint = c; 1397 i += Character.charCount(c); 1398 } 1399 normalizedPattern = newPattern.toString(); 1400 } 1401 1402 /** 1403 * Complete the character class being parsed and add a set 1404 * of alternations to it that will match the canonical equivalences 1405 * of the characters within the class. 1406 */ 1407 private int normalizeCharClass(StringBuilder newPattern, int i) { 1408 StringBuilder charClass = new StringBuilder(); 1409 StringBuilder eq = null; 1410 int lastCodePoint = -1; 1411 String result; 1412 1413 i++; 1414 charClass.append("["); 1415 while(true) { 1416 int c = normalizedPattern.codePointAt(i); 1417 StringBuilder sequenceBuffer; 1418 1419 if (c == ']' && lastCodePoint != '\\') { 1420 charClass.append((char)c); 1421 break; 1422 } else if (Character.getType(c) == Character.NON_SPACING_MARK) { 1423 sequenceBuffer = new StringBuilder(); 1424 sequenceBuffer.appendCodePoint(lastCodePoint); 1425 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1426 sequenceBuffer.appendCodePoint(c); 1427 i += Character.charCount(c); 1428 if (i >= normalizedPattern.length()) 1429 break; 1430 c = normalizedPattern.codePointAt(i); 1431 } 1432 String ea = produceEquivalentAlternation( 1433 sequenceBuffer.toString()); 1434 1435 charClass.setLength(charClass.length()-Character.charCount(lastCodePoint)); 1436 if (eq == null) 1437 eq = new StringBuilder(); 1438 eq.append('|'); 1439 eq.append(ea); 1440 } else { 1441 charClass.appendCodePoint(c); 1442 i++; 1443 } 1444 if (i == normalizedPattern.length()) 1445 throw error("Unclosed character class"); 1446 lastCodePoint = c; 1447 } 1448 1449 if (eq != null) { 1450 result = "(?:"+charClass.toString()+eq.toString()+")"; 1451 } else { 1452 result = charClass.toString(); 1453 } 1454 1455 newPattern.append(result); 1456 return i; 1457 } 1458 1459 /** 1460 * Given a specific sequence composed of a regular character and 1461 * combining marks that follow it, produce the alternation that will 1462 * match all canonical equivalences of that sequence. 1463 */ 1464 private String produceEquivalentAlternation(String source) { 1465 int len = countChars(source, 0, 1); 1466 if (source.length() == len) 1467 // source has one character. 1468 return source; 1469 1470 String base = source.substring(0,len); 1471 String combiningMarks = source.substring(len); 1472 1473 String[] perms = producePermutations(combiningMarks); 1474 StringBuilder result = new StringBuilder(source); 1475 1476 // Add combined permutations 1477 for(int x=0; x<perms.length; x++) { 1478 String next = base + perms[x]; 1479 if (x>0) 1480 result.append("|"+next); 1481 next = composeOneStep(next); 1482 if (next != null) 1483 result.append("|"+produceEquivalentAlternation(next)); 1484 } 1485 return result.toString(); 1486 } 1487 1488 /** 1489 * Returns an array of strings that have all the possible 1490 * permutations of the characters in the input string. 1491 * This is used to get a list of all possible orderings 1492 * of a set of combining marks. Note that some of the permutations 1493 * are invalid because of combining class collisions, and these 1494 * possibilities must be removed because they are not canonically 1495 * equivalent. 1496 */ 1497 private String[] producePermutations(String input) { 1498 if (input.length() == countChars(input, 0, 1)) 1499 return new String[] {input}; 1500 1501 if (input.length() == countChars(input, 0, 2)) { 1502 int c0 = Character.codePointAt(input, 0); 1503 int c1 = Character.codePointAt(input, Character.charCount(c0)); 1504 if (getClass(c1) == getClass(c0)) { 1505 return new String[] {input}; 1506 } 1507 String[] result = new String[2]; 1508 result[0] = input; 1509 StringBuilder sb = new StringBuilder(2); 1510 sb.appendCodePoint(c1); 1511 sb.appendCodePoint(c0); 1512 result[1] = sb.toString(); 1513 return result; 1514 } 1515 1516 int length = 1; 1517 int nCodePoints = countCodePoints(input); 1518 for(int x=1; x<nCodePoints; x++) 1519 length = length * (x+1); 1520 1521 String[] temp = new String[length]; 1522 1523 int combClass[] = new int[nCodePoints]; 1524 for(int x=0, i=0; x<nCodePoints; x++) { 1525 int c = Character.codePointAt(input, i); 1526 combClass[x] = getClass(c); 1527 i += Character.charCount(c); 1528 } 1529 1530 // For each char, take it out and add the permutations 1531 // of the remaining chars 1532 int index = 0; 1533 int len; 1534 // offset maintains the index in code units. 1535 loop: for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) { 1536 len = countChars(input, offset, 1); 1537 boolean skip = false; 1538 for(int y=x-1; y>=0; y--) { 1539 if (combClass[y] == combClass[x]) { 1540 continue loop; 1541 } 1542 } 1543 StringBuilder sb = new StringBuilder(input); 1544 String otherChars = sb.delete(offset, offset+len).toString(); 1545 String[] subResult = producePermutations(otherChars); 1546 1547 String prefix = input.substring(offset, offset+len); 1548 for(int y=0; y<subResult.length; y++) 1549 temp[index++] = prefix + subResult[y]; 1550 } 1551 String[] result = new String[index]; 1552 for (int x=0; x<index; x++) 1553 result[x] = temp[x]; 1554 return result; 1555 } 1556 1557 private int getClass(int c) { 1558 return sun.text.Normalizer.getCombiningClass(c); 1559 } 1560 1561 /** 1562 * Attempts to compose input by combining the first character 1563 * with the first combining mark following it. Returns a String 1564 * that is the composition of the leading character with its first 1565 * combining mark followed by the remaining combining marks. Returns 1566 * null if the first two characters cannot be further composed. 1567 */ 1568 private String composeOneStep(String input) { 1569 int len = countChars(input, 0, 2); 1570 String firstTwoCharacters = input.substring(0, len); 1571 String result = Normalizer.normalize(firstTwoCharacters, Normalizer.Form.NFC); 1572 1573 if (result.equals(firstTwoCharacters)) 1574 return null; 1575 else { 1576 String remainder = input.substring(len); 1577 return result + remainder; 1578 } 1579 } 1580 1581 /** 1582 * Preprocess any \Q...\E sequences in `temp', meta-quoting them. 1583 * See the description of `quotemeta' in perlfunc(1). 1584 */ 1585 private void RemoveQEQuoting() { 1586 final int pLen = patternLength; 1587 int i = 0; 1588 while (i < pLen-1) { 1589 if (temp[i] != '\\') 1590 i += 1; 1591 else if (temp[i + 1] != 'Q') 1592 i += 2; 1593 else 1594 break; 1595 } 1596 if (i >= pLen - 1) // No \Q sequence found 1597 return; 1598 int j = i; 1599 i += 2; 1600 int[] newtemp = new int[j + 3*(pLen-i) + 2]; 1601 System.arraycopy(temp, 0, newtemp, 0, j); 1602 1603 boolean inQuote = true; 1604 boolean beginQuote = true; 1605 while (i < pLen) { 1606 int c = temp[i++]; 1607 if (!ASCII.isAscii(c) || ASCII.isAlpha(c)) { 1608 newtemp[j++] = c; 1609 } else if (ASCII.isDigit(c)) { 1610 if (beginQuote) { 1611 /* 1612 * A unicode escape \[0xu] could be before this quote, 1613 * and we don't want this numeric char to processed as 1614 * part of the escape. 1615 */ 1616 newtemp[j++] = '\\'; 1617 newtemp[j++] = 'x'; 1618 newtemp[j++] = '3'; 1619 } 1620 newtemp[j++] = c; 1621 } else if (c != '\\') { 1622 if (inQuote) newtemp[j++] = '\\'; 1623 newtemp[j++] = c; 1624 } else if (inQuote) { 1625 if (temp[i] == 'E') { 1626 i++; 1627 inQuote = false; 1628 } else { 1629 newtemp[j++] = '\\'; 1630 newtemp[j++] = '\\'; 1631 } 1632 } else { 1633 if (temp[i] == 'Q') { 1634 i++; 1635 inQuote = true; 1636 beginQuote = true; 1637 continue; 1638 } else { 1639 newtemp[j++] = c; 1640 if (i != pLen) 1641 newtemp[j++] = temp[i++]; 1642 } 1643 } 1644 1645 beginQuote = false; 1646 } 1647 1648 patternLength = j; 1649 temp = Arrays.copyOf(newtemp, j + 2); // double zero termination 1650 } 1651 1652 /** 1653 * Copies regular expression to an int array and invokes the parsing 1654 * of the expression which will create the object tree. 1655 */ 1656 private void compile() { 1657 // Handle canonical equivalences 1658 if (has(CANON_EQ) && !has(LITERAL)) { 1659 normalize(); 1660 } else { 1661 normalizedPattern = pattern; 1662 } 1663 patternLength = normalizedPattern.length(); 1664 1665 // Copy pattern to int array for convenience 1666 // Use double zero to terminate pattern 1667 temp = new int[patternLength + 2]; 1668 1669 hasSupplementary = false; 1670 int c, count = 0; 1671 // Convert all chars into code points 1672 for (int x = 0; x < patternLength; x += Character.charCount(c)) { 1673 c = normalizedPattern.codePointAt(x); 1674 if (isSupplementary(c)) { 1675 hasSupplementary = true; 1676 } 1677 temp[count++] = c; 1678 } 1679 1680 patternLength = count; // patternLength now in code points 1681 1682 if (! has(LITERAL)) 1683 RemoveQEQuoting(); 1684 1685 // Allocate all temporary objects here. 1686 buffer = new int[32]; 1687 groupNodes = new GroupHead[10]; 1688 namedGroups = null; 1689 1690 if (has(LITERAL)) { 1691 // Literal pattern handling 1692 matchRoot = newSlice(temp, patternLength, hasSupplementary); 1693 matchRoot.next = lastAccept; 1694 } else { 1695 // Start recursive descent parsing 1696 matchRoot = expr(lastAccept); 1697 // Check extra pattern characters 1698 if (patternLength != cursor) { 1699 if (peek() == ')') { 1700 throw error("Unmatched closing ')'"); 1701 } else { 1702 throw error("Unexpected internal error"); 1703 } 1704 } 1705 } 1706 1707 // Peephole optimization 1708 if (matchRoot instanceof Slice) { 1709 root = BnM.optimize(matchRoot); 1710 if (root == matchRoot) { 1711 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1712 } 1713 } else if (matchRoot instanceof Begin || matchRoot instanceof First) { 1714 root = matchRoot; 1715 } else { 1716 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1717 } 1718 1719 // Release temporary storage 1720 temp = null; 1721 buffer = null; 1722 groupNodes = null; 1723 patternLength = 0; 1724 compiled = true; 1725 } 1726 1727 Map<String, Integer> namedGroups() { 1728 if (namedGroups == null) 1729 namedGroups = new HashMap<>(2); 1730 return namedGroups; 1731 } 1732 1733 /** 1734 * Used to print out a subtree of the Pattern to help with debugging. 1735 */ 1736 private static void printObjectTree(Node node) { 1737 while(node != null) { 1738 if (node instanceof Prolog) { 1739 System.out.println(node); 1740 printObjectTree(((Prolog)node).loop); 1741 System.out.println("**** end contents prolog loop"); 1742 } else if (node instanceof Loop) { 1743 System.out.println(node); 1744 printObjectTree(((Loop)node).body); 1745 System.out.println("**** end contents Loop body"); 1746 } else if (node instanceof Curly) { 1747 System.out.println(node); 1748 printObjectTree(((Curly)node).atom); 1749 System.out.println("**** end contents Curly body"); 1750 } else if (node instanceof GroupCurly) { 1751 System.out.println(node); 1752 printObjectTree(((GroupCurly)node).atom); 1753 System.out.println("**** end contents GroupCurly body"); 1754 } else if (node instanceof GroupTail) { 1755 System.out.println(node); 1756 System.out.println("Tail next is "+node.next); 1757 return; 1758 } else { 1759 System.out.println(node); 1760 } 1761 node = node.next; 1762 if (node != null) 1763 System.out.println("->next:"); 1764 if (node == Pattern.accept) { 1765 System.out.println("Accept Node"); 1766 node = null; 1767 } 1768 } 1769 } 1770 1771 /** 1772 * Used to accumulate information about a subtree of the object graph 1773 * so that optimizations can be applied to the subtree. 1774 */ 1775 static final class TreeInfo { 1776 int minLength; 1777 int maxLength; 1778 boolean maxValid; 1779 boolean deterministic; 1780 1781 TreeInfo() { 1782 reset(); 1783 } 1784 void reset() { 1785 minLength = 0; 1786 maxLength = 0; 1787 maxValid = true; 1788 deterministic = true; 1789 } 1790 } 1791 1792 /* 1793 * The following private methods are mainly used to improve the 1794 * readability of the code. In order to let the Java compiler easily 1795 * inline them, we should not put many assertions or error checks in them. 1796 */ 1797 1798 /** 1799 * Indicates whether a particular flag is set or not. 1800 */ 1801 private boolean has(int f) { 1802 return (flags & f) != 0; 1803 } 1804 1805 /** 1806 * Match next character, signal error if failed. 1807 */ 1808 private void accept(int ch, String s) { 1809 int testChar = temp[cursor++]; 1810 if (has(COMMENTS)) 1811 testChar = parsePastWhitespace(testChar); 1812 if (ch != testChar) { 1813 throw error(s); 1814 } 1815 } 1816 1817 /** 1818 * Mark the end of pattern with a specific character. 1819 */ 1820 private void mark(int c) { 1821 temp[patternLength] = c; 1822 } 1823 1824 /** 1825 * Peek the next character, and do not advance the cursor. 1826 */ 1827 private int peek() { 1828 int ch = temp[cursor]; 1829 if (has(COMMENTS)) 1830 ch = peekPastWhitespace(ch); 1831 return ch; 1832 } 1833 1834 /** 1835 * Read the next character, and advance the cursor by one. 1836 */ 1837 private int read() { 1838 int ch = temp[cursor++]; 1839 if (has(COMMENTS)) 1840 ch = parsePastWhitespace(ch); 1841 return ch; 1842 } 1843 1844 /** 1845 * Read the next character, and advance the cursor by one, 1846 * ignoring the COMMENTS setting 1847 */ 1848 private int readEscaped() { 1849 int ch = temp[cursor++]; 1850 return ch; 1851 } 1852 1853 /** 1854 * Advance the cursor by one, and peek the next character. 1855 */ 1856 private int next() { 1857 int ch = temp[++cursor]; 1858 if (has(COMMENTS)) 1859 ch = peekPastWhitespace(ch); 1860 return ch; 1861 } 1862 1863 /** 1864 * Advance the cursor by one, and peek the next character, 1865 * ignoring the COMMENTS setting 1866 */ 1867 private int nextEscaped() { 1868 int ch = temp[++cursor]; 1869 return ch; 1870 } 1871 1872 /** 1873 * If in xmode peek past whitespace and comments. 1874 */ 1875 private int peekPastWhitespace(int ch) { 1876 while (ASCII.isSpace(ch) || ch == '#') { 1877 while (ASCII.isSpace(ch)) 1878 ch = temp[++cursor]; 1879 if (ch == '#') { 1880 ch = peekPastLine(); 1881 } 1882 } 1883 return ch; 1884 } 1885 1886 /** 1887 * If in xmode parse past whitespace and comments. 1888 */ 1889 private int parsePastWhitespace(int ch) { 1890 while (ASCII.isSpace(ch) || ch == '#') { 1891 while (ASCII.isSpace(ch)) 1892 ch = temp[cursor++]; 1893 if (ch == '#') 1894 ch = parsePastLine(); 1895 } 1896 return ch; 1897 } 1898 1899 /** 1900 * xmode parse past comment to end of line. 1901 */ 1902 private int parsePastLine() { 1903 int ch = temp[cursor++]; 1904 while (ch != 0 && !isLineSeparator(ch)) 1905 ch = temp[cursor++]; 1906 return ch; 1907 } 1908 1909 /** 1910 * xmode peek past comment to end of line. 1911 */ 1912 private int peekPastLine() { 1913 int ch = temp[++cursor]; 1914 while (ch != 0 && !isLineSeparator(ch)) 1915 ch = temp[++cursor]; 1916 return ch; 1917 } 1918 1919 /** 1920 * Determines if character is a line separator in the current mode 1921 */ 1922 private boolean isLineSeparator(int ch) { 1923 if (has(UNIX_LINES)) { 1924 return ch == '\n'; 1925 } else { 1926 return (ch == '\n' || 1927 ch == '\r' || 1928 (ch|1) == '\u2029' || 1929 ch == '\u0085'); 1930 } 1931 } 1932 1933 /** 1934 * Read the character after the next one, and advance the cursor by two. 1935 */ 1936 private int skip() { 1937 int i = cursor; 1938 int ch = temp[i+1]; 1939 cursor = i + 2; 1940 return ch; 1941 } 1942 1943 /** 1944 * Unread one next character, and retreat cursor by one. 1945 */ 1946 private void unread() { 1947 cursor--; 1948 } 1949 1950 /** 1951 * Internal method used for handling all syntax errors. The pattern is 1952 * displayed with a pointer to aid in locating the syntax error. 1953 */ 1954 private PatternSyntaxException error(String s) { 1955 return new PatternSyntaxException(s, normalizedPattern, cursor - 1); 1956 } 1957 1958 /** 1959 * Determines if there is any supplementary character or unpaired 1960 * surrogate in the specified range. 1961 */ 1962 private boolean findSupplementary(int start, int end) { 1963 for (int i = start; i < end; i++) { 1964 if (isSupplementary(temp[i])) 1965 return true; 1966 } 1967 return false; 1968 } 1969 1970 /** 1971 * Determines if the specified code point is a supplementary 1972 * character or unpaired surrogate. 1973 */ 1974 private static final boolean isSupplementary(int ch) { 1975 return ch >= Character.MIN_SUPPLEMENTARY_CODE_POINT || 1976 Character.isSurrogate((char)ch); 1977 } 1978 1979 /** 1980 * The following methods handle the main parsing. They are sorted 1981 * according to their precedence order, the lowest one first. 1982 */ 1983 1984 /** 1985 * The expression is parsed with branch nodes added for alternations. 1986 * This may be called recursively to parse sub expressions that may 1987 * contain alternations. 1988 */ 1989 private Node expr(Node end) { 1990 Node prev = null; 1991 Node firstTail = null; 1992 Branch branch = null; 1993 Node branchConn = null; 1994 1995 for (;;) { 1996 Node node = sequence(end); 1997 Node nodeTail = root; //double return 1998 if (prev == null) { 1999 prev = node; 2000 firstTail = nodeTail; 2001 } else { 2002 // Branch 2003 if (branchConn == null) { 2004 branchConn = new BranchConn(); 2005 branchConn.next = end; 2006 } 2007 if (node == end) { 2008 // if the node returned from sequence() is "end" 2009 // we have an empty expr, set a null atom into 2010 // the branch to indicate to go "next" directly. 2011 node = null; 2012 } else { 2013 // the "tail.next" of each atom goes to branchConn 2014 nodeTail.next = branchConn; 2015 } 2016 if (prev == branch) { 2017 branch.add(node); 2018 } else { 2019 if (prev == end) { 2020 prev = null; 2021 } else { 2022 // replace the "end" with "branchConn" at its tail.next 2023 // when put the "prev" into the branch as the first atom. 2024 firstTail.next = branchConn; 2025 } 2026 prev = branch = new Branch(prev, node, branchConn); 2027 } 2028 } 2029 if (peek() != '|') { 2030 return prev; 2031 } 2032 next(); 2033 } 2034 } 2035 2036 @SuppressWarnings("fallthrough") 2037 /** 2038 * Parsing of sequences between alternations. 2039 */ 2040 private Node sequence(Node end) { 2041 Node head = null; 2042 Node tail = null; 2043 Node node = null; 2044 LOOP: 2045 for (;;) { 2046 int ch = peek(); 2047 switch (ch) { 2048 case '(': 2049 // Because group handles its own closure, 2050 // we need to treat it differently 2051 node = group0(); 2052 // Check for comment or flag group 2053 if (node == null) 2054 continue; 2055 if (head == null) 2056 head = node; 2057 else 2058 tail.next = node; 2059 // Double return: Tail was returned in root 2060 tail = root; 2061 continue; 2062 case '[': 2063 node = clazz(true); 2064 break; 2065 case '\\': 2066 ch = nextEscaped(); 2067 if (ch == 'p' || ch == 'P') { 2068 boolean oneLetter = true; 2069 boolean comp = (ch == 'P'); 2070 ch = next(); // Consume { if present 2071 if (ch != '{') { 2072 unread(); 2073 } else { 2074 oneLetter = false; 2075 } 2076 node = family(oneLetter, comp); 2077 } else { 2078 unread(); 2079 node = atom(); 2080 } 2081 break; 2082 case '^': 2083 next(); 2084 if (has(MULTILINE)) { 2085 if (has(UNIX_LINES)) 2086 node = new UnixCaret(); 2087 else 2088 node = new Caret(); 2089 } else { 2090 node = new Begin(); 2091 } 2092 break; 2093 case '$': 2094 next(); 2095 if (has(UNIX_LINES)) 2096 node = new UnixDollar(has(MULTILINE)); 2097 else 2098 node = new Dollar(has(MULTILINE)); 2099 break; 2100 case '.': 2101 next(); 2102 if (has(DOTALL)) { 2103 node = new All(); 2104 } else { 2105 if (has(UNIX_LINES)) 2106 node = new UnixDot(); 2107 else { 2108 node = new Dot(); 2109 } 2110 } 2111 break; 2112 case '|': 2113 case ')': 2114 break LOOP; 2115 case ']': // Now interpreting dangling ] and } as literals 2116 case '}': 2117 node = atom(); 2118 break; 2119 case '?': 2120 case '*': 2121 case '+': 2122 next(); 2123 throw error("Dangling meta character '" + ((char)ch) + "'"); 2124 case 0: 2125 if (cursor >= patternLength) { 2126 break LOOP; 2127 } 2128 // Fall through 2129 default: 2130 node = atom(); 2131 break; 2132 } 2133 2134 node = closure(node); 2135 2136 if (head == null) { 2137 head = tail = node; 2138 } else { 2139 tail.next = node; 2140 tail = node; 2141 } 2142 } 2143 if (head == null) { 2144 return end; 2145 } 2146 tail.next = end; 2147 root = tail; //double return 2148 return head; 2149 } 2150 2151 @SuppressWarnings("fallthrough") 2152 /** 2153 * Parse and add a new Single or Slice. 2154 */ 2155 private Node atom() { 2156 int first = 0; 2157 int prev = -1; 2158 boolean hasSupplementary = false; 2159 int ch = peek(); 2160 for (;;) { 2161 switch (ch) { 2162 case '*': 2163 case '+': 2164 case '?': 2165 case '{': 2166 if (first > 1) { 2167 cursor = prev; // Unwind one character 2168 first--; 2169 } 2170 break; 2171 case '$': 2172 case '.': 2173 case '^': 2174 case '(': 2175 case '[': 2176 case '|': 2177 case ')': 2178 break; 2179 case '\\': 2180 ch = nextEscaped(); 2181 if (ch == 'p' || ch == 'P') { // Property 2182 if (first > 0) { // Slice is waiting; handle it first 2183 unread(); 2184 break; 2185 } else { // No slice; just return the family node 2186 boolean comp = (ch == 'P'); 2187 boolean oneLetter = true; 2188 ch = next(); // Consume { if present 2189 if (ch != '{') 2190 unread(); 2191 else 2192 oneLetter = false; 2193 return family(oneLetter, comp); 2194 } 2195 } 2196 unread(); 2197 prev = cursor; 2198 ch = escape(false, first == 0, false); 2199 if (ch >= 0) { 2200 append(ch, first); 2201 first++; 2202 if (isSupplementary(ch)) { 2203 hasSupplementary = true; 2204 } 2205 ch = peek(); 2206 continue; 2207 } else if (first == 0) { 2208 return root; 2209 } 2210 // Unwind meta escape sequence 2211 cursor = prev; 2212 break; 2213 case 0: 2214 if (cursor >= patternLength) { 2215 break; 2216 } 2217 // Fall through 2218 default: 2219 prev = cursor; 2220 append(ch, first); 2221 first++; 2222 if (isSupplementary(ch)) { 2223 hasSupplementary = true; 2224 } 2225 ch = next(); 2226 continue; 2227 } 2228 break; 2229 } 2230 if (first == 1) { 2231 return newSingle(buffer[0]); 2232 } else { 2233 return newSlice(buffer, first, hasSupplementary); 2234 } 2235 } 2236 2237 private void append(int ch, int len) { 2238 if (len >= buffer.length) { 2239 int[] tmp = new int[len+len]; 2240 System.arraycopy(buffer, 0, tmp, 0, len); 2241 buffer = tmp; 2242 } 2243 buffer[len] = ch; 2244 } 2245 2246 /** 2247 * Parses a backref greedily, taking as many numbers as it 2248 * can. The first digit is always treated as a backref, but 2249 * multi digit numbers are only treated as a backref if at 2250 * least that many backrefs exist at this point in the regex. 2251 */ 2252 private Node ref(int refNum) { 2253 boolean done = false; 2254 while(!done) { 2255 int ch = peek(); 2256 switch(ch) { 2257 case '0': 2258 case '1': 2259 case '2': 2260 case '3': 2261 case '4': 2262 case '5': 2263 case '6': 2264 case '7': 2265 case '8': 2266 case '9': 2267 int newRefNum = (refNum * 10) + (ch - '0'); 2268 // Add another number if it doesn't make a group 2269 // that doesn't exist 2270 if (capturingGroupCount - 1 < newRefNum) { 2271 done = true; 2272 break; 2273 } 2274 refNum = newRefNum; 2275 read(); 2276 break; 2277 default: 2278 done = true; 2279 break; 2280 } 2281 } 2282 if (has(CASE_INSENSITIVE)) 2283 return new CIBackRef(refNum, has(UNICODE_CASE)); 2284 else 2285 return new BackRef(refNum); 2286 } 2287 2288 /** 2289 * Parses an escape sequence to determine the actual value that needs 2290 * to be matched. 2291 * If -1 is returned and create was true a new object was added to the tree 2292 * to handle the escape sequence. 2293 * If the returned value is greater than zero, it is the value that 2294 * matches the escape sequence. 2295 */ 2296 private int escape(boolean inclass, boolean create, boolean isrange) { 2297 int ch = skip(); 2298 switch (ch) { 2299 case '0': 2300 return o(); 2301 case '1': 2302 case '2': 2303 case '3': 2304 case '4': 2305 case '5': 2306 case '6': 2307 case '7': 2308 case '8': 2309 case '9': 2310 if (inclass) break; 2311 if (create) { 2312 root = ref((ch - '0')); 2313 } 2314 return -1; 2315 case 'A': 2316 if (inclass) break; 2317 if (create) root = new Begin(); 2318 return -1; 2319 case 'B': 2320 if (inclass) break; 2321 if (create) root = new Bound(Bound.NONE, has(UNICODE_CHARACTER_CLASS)); 2322 return -1; 2323 case 'C': 2324 break; 2325 case 'D': 2326 if (create) root = has(UNICODE_CHARACTER_CLASS) 2327 ? new Utype(UnicodeProp.DIGIT).complement() 2328 : new Ctype(ASCII.DIGIT).complement(); 2329 return -1; 2330 case 'E': 2331 case 'F': 2332 break; 2333 case 'G': 2334 if (inclass) break; 2335 if (create) root = new LastMatch(); 2336 return -1; 2337 case 'H': 2338 if (create) root = new HorizWS().complement(); 2339 return -1; 2340 case 'I': 2341 case 'J': 2342 case 'K': 2343 case 'L': 2344 case 'M': 2345 case 'N': 2346 case 'O': 2347 case 'P': 2348 case 'Q': 2349 break; 2350 case 'R': 2351 if (inclass) break; 2352 if (create) root = new LineEnding(); 2353 return -1; 2354 case 'S': 2355 if (create) root = has(UNICODE_CHARACTER_CLASS) 2356 ? new Utype(UnicodeProp.WHITE_SPACE).complement() 2357 : new Ctype(ASCII.SPACE).complement(); 2358 return -1; 2359 case 'T': 2360 case 'U': 2361 break; 2362 case 'V': 2363 if (create) root = new VertWS().complement(); 2364 return -1; 2365 case 'W': 2366 if (create) root = has(UNICODE_CHARACTER_CLASS) 2367 ? new Utype(UnicodeProp.WORD).complement() 2368 : new Ctype(ASCII.WORD).complement(); 2369 return -1; 2370 case 'X': 2371 case 'Y': 2372 break; 2373 case 'Z': 2374 if (inclass) break; 2375 if (create) { 2376 if (has(UNIX_LINES)) 2377 root = new UnixDollar(false); 2378 else 2379 root = new Dollar(false); 2380 } 2381 return -1; 2382 case 'a': 2383 return '\007'; 2384 case 'b': 2385 if (inclass) break; 2386 if (create) root = new Bound(Bound.BOTH, has(UNICODE_CHARACTER_CLASS)); 2387 return -1; 2388 case 'c': 2389 return c(); 2390 case 'd': 2391 if (create) root = has(UNICODE_CHARACTER_CLASS) 2392 ? new Utype(UnicodeProp.DIGIT) 2393 : new Ctype(ASCII.DIGIT); 2394 return -1; 2395 case 'e': 2396 return '\033'; 2397 case 'f': 2398 return '\f'; 2399 case 'g': 2400 break; 2401 case 'h': 2402 if (create) root = new HorizWS(); 2403 return -1; 2404 case 'i': 2405 case 'j': 2406 break; 2407 case 'k': 2408 if (inclass) 2409 break; 2410 if (read() != '<') 2411 throw error("\\k is not followed by '<' for named capturing group"); 2412 String name = groupname(read()); 2413 if (!namedGroups().containsKey(name)) 2414 throw error("(named capturing group <"+ name+"> does not exit"); 2415 if (create) { 2416 if (has(CASE_INSENSITIVE)) 2417 root = new CIBackRef(namedGroups().get(name), has(UNICODE_CASE)); 2418 else 2419 root = new BackRef(namedGroups().get(name)); 2420 } 2421 return -1; 2422 case 'l': 2423 case 'm': 2424 break; 2425 case 'n': 2426 return '\n'; 2427 case 'o': 2428 case 'p': 2429 case 'q': 2430 break; 2431 case 'r': 2432 return '\r'; 2433 case 's': 2434 if (create) root = has(UNICODE_CHARACTER_CLASS) 2435 ? new Utype(UnicodeProp.WHITE_SPACE) 2436 : new Ctype(ASCII.SPACE); 2437 return -1; 2438 case 't': 2439 return '\t'; 2440 case 'u': 2441 return u(); 2442 case 'v': 2443 // '\v' was implemented as VT/0x0B in releases < 1.8 (though 2444 // undocumented). In JDK8 '\v' is specified as a predefined 2445 // character class for all vertical whitespace characters. 2446 // So [-1, root=VertWS node] pair is returned (instead of a 2447 // single 0x0B). This breaks the range if '\v' is used as 2448 // the start or end value, such as [\v-...] or [...-\v], in 2449 // which a single definite value (0x0B) is expected. For 2450 // compatibility concern '\013'/0x0B is returned if isrange. 2451 if (isrange) 2452 return '\013'; 2453 if (create) root = new VertWS(); 2454 return -1; 2455 case 'w': 2456 if (create) root = has(UNICODE_CHARACTER_CLASS) 2457 ? new Utype(UnicodeProp.WORD) 2458 : new Ctype(ASCII.WORD); 2459 return -1; 2460 case 'x': 2461 return x(); 2462 case 'y': 2463 break; 2464 case 'z': 2465 if (inclass) break; 2466 if (create) root = new End(); 2467 return -1; 2468 default: 2469 return ch; 2470 } 2471 throw error("Illegal/unsupported escape sequence"); 2472 } 2473 2474 /** 2475 * Parse a character class, and return the node that matches it. 2476 * 2477 * Consumes a ] on the way out if consume is true. Usually consume 2478 * is true except for the case of [abc&&def] where def is a separate 2479 * right hand node with "understood" brackets. 2480 */ 2481 private CharProperty clazz(boolean consume) { 2482 CharProperty prev = null; 2483 CharProperty node = null; 2484 BitClass bits = new BitClass(); 2485 boolean include = true; 2486 boolean firstInClass = true; 2487 int ch = next(); 2488 for (;;) { 2489 switch (ch) { 2490 case '^': 2491 // Negates if first char in a class, otherwise literal 2492 if (firstInClass) { 2493 if (temp[cursor-1] != '[') 2494 break; 2495 ch = next(); 2496 include = !include; 2497 continue; 2498 } else { 2499 // ^ not first in class, treat as literal 2500 break; 2501 } 2502 case '[': 2503 firstInClass = false; 2504 node = clazz(true); 2505 if (prev == null) 2506 prev = node; 2507 else 2508 prev = union(prev, node); 2509 ch = peek(); 2510 continue; 2511 case '&': 2512 firstInClass = false; 2513 ch = next(); 2514 if (ch == '&') { 2515 ch = next(); 2516 CharProperty rightNode = null; 2517 while (ch != ']' && ch != '&') { 2518 if (ch == '[') { 2519 if (rightNode == null) 2520 rightNode = clazz(true); 2521 else 2522 rightNode = union(rightNode, clazz(true)); 2523 } else { // abc&&def 2524 unread(); 2525 rightNode = clazz(false); 2526 } 2527 ch = peek(); 2528 } 2529 if (rightNode != null) 2530 node = rightNode; 2531 if (prev == null) { 2532 if (rightNode == null) 2533 throw error("Bad class syntax"); 2534 else 2535 prev = rightNode; 2536 } else { 2537 prev = intersection(prev, node); 2538 } 2539 } else { 2540 // treat as a literal & 2541 unread(); 2542 break; 2543 } 2544 continue; 2545 case 0: 2546 firstInClass = false; 2547 if (cursor >= patternLength) 2548 throw error("Unclosed character class"); 2549 break; 2550 case ']': 2551 firstInClass = false; 2552 if (prev != null) { 2553 if (consume) 2554 next(); 2555 return prev; 2556 } 2557 break; 2558 default: 2559 firstInClass = false; 2560 break; 2561 } 2562 node = range(bits); 2563 if (include) { 2564 if (prev == null) { 2565 prev = node; 2566 } else { 2567 if (prev != node) 2568 prev = union(prev, node); 2569 } 2570 } else { 2571 if (prev == null) { 2572 prev = node.complement(); 2573 } else { 2574 if (prev != node) 2575 prev = setDifference(prev, node); 2576 } 2577 } 2578 ch = peek(); 2579 } 2580 } 2581 2582 private CharProperty bitsOrSingle(BitClass bits, int ch) { 2583 /* Bits can only handle codepoints in [u+0000-u+00ff] range. 2584 Use "single" node instead of bits when dealing with unicode 2585 case folding for codepoints listed below. 2586 (1)Uppercase out of range: u+00ff, u+00b5 2587 toUpperCase(u+00ff) -> u+0178 2588 toUpperCase(u+00b5) -> u+039c 2589 (2)LatinSmallLetterLongS u+17f 2590 toUpperCase(u+017f) -> u+0053 2591 (3)LatinSmallLetterDotlessI u+131 2592 toUpperCase(u+0131) -> u+0049 2593 (4)LatinCapitalLetterIWithDotAbove u+0130 2594 toLowerCase(u+0130) -> u+0069 2595 (5)KelvinSign u+212a 2596 toLowerCase(u+212a) ==> u+006B 2597 (6)AngstromSign u+212b 2598 toLowerCase(u+212b) ==> u+00e5 2599 */ 2600 int d; 2601 if (ch < 256 && 2602 !(has(CASE_INSENSITIVE) && has(UNICODE_CASE) && 2603 (ch == 0xff || ch == 0xb5 || 2604 ch == 0x49 || ch == 0x69 || //I and i 2605 ch == 0x53 || ch == 0x73 || //S and s 2606 ch == 0x4b || ch == 0x6b || //K and k 2607 ch == 0xc5 || ch == 0xe5))) //A+ring 2608 return bits.add(ch, flags()); 2609 return newSingle(ch); 2610 } 2611 2612 /** 2613 * Parse a single character or a character range in a character class 2614 * and return its representative node. 2615 */ 2616 private CharProperty range(BitClass bits) { 2617 int ch = peek(); 2618 if (ch == '\\') { 2619 ch = nextEscaped(); 2620 if (ch == 'p' || ch == 'P') { // A property 2621 boolean comp = (ch == 'P'); 2622 boolean oneLetter = true; 2623 // Consume { if present 2624 ch = next(); 2625 if (ch != '{') 2626 unread(); 2627 else 2628 oneLetter = false; 2629 return family(oneLetter, comp); 2630 } else { // ordinary escape 2631 boolean isrange = temp[cursor+1] == '-'; 2632 unread(); 2633 ch = escape(true, true, isrange); 2634 if (ch == -1) 2635 return (CharProperty) root; 2636 } 2637 } else { 2638 next(); 2639 } 2640 if (ch >= 0) { 2641 if (peek() == '-') { 2642 int endRange = temp[cursor+1]; 2643 if (endRange == '[') { 2644 return bitsOrSingle(bits, ch); 2645 } 2646 if (endRange != ']') { 2647 next(); 2648 int m = peek(); 2649 if (m == '\\') { 2650 m = escape(true, false, true); 2651 } else { 2652 next(); 2653 } 2654 if (m < ch) { 2655 throw error("Illegal character range"); 2656 } 2657 if (has(CASE_INSENSITIVE)) 2658 return caseInsensitiveRangeFor(ch, m); 2659 else 2660 return rangeFor(ch, m); 2661 } 2662 } 2663 return bitsOrSingle(bits, ch); 2664 } 2665 throw error("Unexpected character '"+((char)ch)+"'"); 2666 } 2667 2668 /** 2669 * Parses a Unicode character family and returns its representative node. 2670 */ 2671 private CharProperty family(boolean singleLetter, 2672 boolean maybeComplement) 2673 { 2674 next(); 2675 String name; 2676 CharProperty node = null; 2677 2678 if (singleLetter) { 2679 int c = temp[cursor]; 2680 if (!Character.isSupplementaryCodePoint(c)) { 2681 name = String.valueOf((char)c); 2682 } else { 2683 name = new String(temp, cursor, 1); 2684 } 2685 read(); 2686 } else { 2687 int i = cursor; 2688 mark('}'); 2689 while(read() != '}') { 2690 } 2691 mark('\000'); 2692 int j = cursor; 2693 if (j > patternLength) 2694 throw error("Unclosed character family"); 2695 if (i + 1 >= j) 2696 throw error("Empty character family"); 2697 name = new String(temp, i, j-i-1); 2698 } 2699 2700 int i = name.indexOf('='); 2701 if (i != -1) { 2702 // property construct \p{name=value} 2703 String value = name.substring(i + 1); 2704 name = name.substring(0, i).toLowerCase(Locale.ENGLISH); 2705 if ("sc".equals(name) || "script".equals(name)) { 2706 node = unicodeScriptPropertyFor(value); 2707 } else if ("blk".equals(name) || "block".equals(name)) { 2708 node = unicodeBlockPropertyFor(value); 2709 } else if ("gc".equals(name) || "general_category".equals(name)) { 2710 node = charPropertyNodeFor(value); 2711 } else { 2712 throw error("Unknown Unicode property {name=<" + name + ">, " 2713 + "value=<" + value + ">}"); 2714 } 2715 } else { 2716 if (name.startsWith("In")) { 2717 // \p{inBlockName} 2718 node = unicodeBlockPropertyFor(name.substring(2)); 2719 } else if (name.startsWith("Is")) { 2720 // \p{isGeneralCategory} and \p{isScriptName} 2721 name = name.substring(2); 2722 UnicodeProp uprop = UnicodeProp.forName(name); 2723 if (uprop != null) 2724 node = new Utype(uprop); 2725 if (node == null) 2726 node = CharPropertyNames.charPropertyFor(name); 2727 if (node == null) 2728 node = unicodeScriptPropertyFor(name); 2729 } else { 2730 if (has(UNICODE_CHARACTER_CLASS)) { 2731 UnicodeProp uprop = UnicodeProp.forPOSIXName(name); 2732 if (uprop != null) 2733 node = new Utype(uprop); 2734 } 2735 if (node == null) 2736 node = charPropertyNodeFor(name); 2737 } 2738 } 2739 if (maybeComplement) { 2740 if (node instanceof Category || node instanceof Block) 2741 hasSupplementary = true; 2742 node = node.complement(); 2743 } 2744 return node; 2745 } 2746 2747 2748 /** 2749 * Returns a CharProperty matching all characters belong to 2750 * a UnicodeScript. 2751 */ 2752 private CharProperty unicodeScriptPropertyFor(String name) { 2753 final Character.UnicodeScript script; 2754 try { 2755 script = Character.UnicodeScript.forName(name); 2756 } catch (IllegalArgumentException iae) { 2757 throw error("Unknown character script name {" + name + "}"); 2758 } 2759 return new Script(script); 2760 } 2761 2762 /** 2763 * Returns a CharProperty matching all characters in a UnicodeBlock. 2764 */ 2765 private CharProperty unicodeBlockPropertyFor(String name) { 2766 final Character.UnicodeBlock block; 2767 try { 2768 block = Character.UnicodeBlock.forName(name); 2769 } catch (IllegalArgumentException iae) { 2770 throw error("Unknown character block name {" + name + "}"); 2771 } 2772 return new Block(block); 2773 } 2774 2775 /** 2776 * Returns a CharProperty matching all characters in a named property. 2777 */ 2778 private CharProperty charPropertyNodeFor(String name) { 2779 CharProperty p = CharPropertyNames.charPropertyFor(name); 2780 if (p == null) 2781 throw error("Unknown character property name {" + name + "}"); 2782 return p; 2783 } 2784 2785 /** 2786 * Parses and returns the name of a "named capturing group", the trailing 2787 * ">" is consumed after parsing. 2788 */ 2789 private String groupname(int ch) { 2790 StringBuilder sb = new StringBuilder(); 2791 sb.append(Character.toChars(ch)); 2792 while (ASCII.isLower(ch=read()) || ASCII.isUpper(ch) || 2793 ASCII.isDigit(ch)) { 2794 sb.append(Character.toChars(ch)); 2795 } 2796 if (sb.length() == 0) 2797 throw error("named capturing group has 0 length name"); 2798 if (ch != '>') 2799 throw error("named capturing group is missing trailing '>'"); 2800 return sb.toString(); 2801 } 2802 2803 /** 2804 * Parses a group and returns the head node of a set of nodes that process 2805 * the group. Sometimes a double return system is used where the tail is 2806 * returned in root. 2807 */ 2808 private Node group0() { 2809 boolean capturingGroup = false; 2810 Node head = null; 2811 Node tail = null; 2812 int save = flags; 2813 root = null; 2814 int ch = next(); 2815 if (ch == '?') { 2816 ch = skip(); 2817 switch (ch) { 2818 case ':': // (?:xxx) pure group 2819 head = createGroup(true); 2820 tail = root; 2821 head.next = expr(tail); 2822 break; 2823 case '=': // (?=xxx) and (?!xxx) lookahead 2824 case '!': 2825 head = createGroup(true); 2826 tail = root; 2827 head.next = expr(tail); 2828 if (ch == '=') { 2829 head = tail = new Pos(head); 2830 } else { 2831 head = tail = new Neg(head); 2832 } 2833 break; 2834 case '>': // (?>xxx) independent group 2835 head = createGroup(true); 2836 tail = root; 2837 head.next = expr(tail); 2838 head = tail = new Ques(head, INDEPENDENT); 2839 break; 2840 case '<': // (?<xxx) look behind 2841 ch = read(); 2842 if (ASCII.isLower(ch) || ASCII.isUpper(ch)) { 2843 // named captured group 2844 String name = groupname(ch); 2845 if (namedGroups().containsKey(name)) 2846 throw error("Named capturing group <" + name 2847 + "> is already defined"); 2848 capturingGroup = true; 2849 head = createGroup(false); 2850 tail = root; 2851 namedGroups().put(name, capturingGroupCount-1); 2852 head.next = expr(tail); 2853 break; 2854 } 2855 int start = cursor; 2856 head = createGroup(true); 2857 tail = root; 2858 head.next = expr(tail); 2859 tail.next = lookbehindEnd; 2860 TreeInfo info = new TreeInfo(); 2861 head.study(info); 2862 if (info.maxValid == false) { 2863 throw error("Look-behind group does not have " 2864 + "an obvious maximum length"); 2865 } 2866 boolean hasSupplementary = findSupplementary(start, patternLength); 2867 if (ch == '=') { 2868 head = tail = (hasSupplementary ? 2869 new BehindS(head, info.maxLength, 2870 info.minLength) : 2871 new Behind(head, info.maxLength, 2872 info.minLength)); 2873 } else if (ch == '!') { 2874 head = tail = (hasSupplementary ? 2875 new NotBehindS(head, info.maxLength, 2876 info.minLength) : 2877 new NotBehind(head, info.maxLength, 2878 info.minLength)); 2879 } else { 2880 throw error("Unknown look-behind group"); 2881 } 2882 break; 2883 case '$': 2884 case '@': 2885 throw error("Unknown group type"); 2886 default: // (?xxx:) inlined match flags 2887 unread(); 2888 addFlag(); 2889 ch = read(); 2890 if (ch == ')') { 2891 return null; // Inline modifier only 2892 } 2893 if (ch != ':') { 2894 throw error("Unknown inline modifier"); 2895 } 2896 head = createGroup(true); 2897 tail = root; 2898 head.next = expr(tail); 2899 break; 2900 } 2901 } else { // (xxx) a regular group 2902 capturingGroup = true; 2903 head = createGroup(false); 2904 tail = root; 2905 head.next = expr(tail); 2906 } 2907 2908 accept(')', "Unclosed group"); 2909 flags = save; 2910 2911 // Check for quantifiers 2912 Node node = closure(head); 2913 if (node == head) { // No closure 2914 root = tail; 2915 return node; // Dual return 2916 } 2917 if (head == tail) { // Zero length assertion 2918 root = node; 2919 return node; // Dual return 2920 } 2921 2922 if (node instanceof Ques) { 2923 Ques ques = (Ques) node; 2924 if (ques.type == POSSESSIVE) { 2925 root = node; 2926 return node; 2927 } 2928 tail.next = new BranchConn(); 2929 tail = tail.next; 2930 if (ques.type == GREEDY) { 2931 head = new Branch(head, null, tail); 2932 } else { // Reluctant quantifier 2933 head = new Branch(null, head, tail); 2934 } 2935 root = tail; 2936 return head; 2937 } else if (node instanceof Curly) { 2938 Curly curly = (Curly) node; 2939 if (curly.type == POSSESSIVE) { 2940 root = node; 2941 return node; 2942 } 2943 // Discover if the group is deterministic 2944 TreeInfo info = new TreeInfo(); 2945 if (head.study(info)) { // Deterministic 2946 GroupTail temp = (GroupTail) tail; 2947 head = root = new GroupCurly(head.next, curly.cmin, 2948 curly.cmax, curly.type, 2949 ((GroupTail)tail).localIndex, 2950 ((GroupTail)tail).groupIndex, 2951 capturingGroup); 2952 return head; 2953 } else { // Non-deterministic 2954 int temp = ((GroupHead) head).localIndex; 2955 Loop loop; 2956 if (curly.type == GREEDY) 2957 loop = new Loop(this.localCount, temp); 2958 else // Reluctant Curly 2959 loop = new LazyLoop(this.localCount, temp); 2960 Prolog prolog = new Prolog(loop); 2961 this.localCount += 1; 2962 loop.cmin = curly.cmin; 2963 loop.cmax = curly.cmax; 2964 loop.body = head; 2965 tail.next = loop; 2966 root = loop; 2967 return prolog; // Dual return 2968 } 2969 } 2970 throw error("Internal logic error"); 2971 } 2972 2973 /** 2974 * Create group head and tail nodes using double return. If the group is 2975 * created with anonymous true then it is a pure group and should not 2976 * affect group counting. 2977 */ 2978 private Node createGroup(boolean anonymous) { 2979 int localIndex = localCount++; 2980 int groupIndex = 0; 2981 if (!anonymous) 2982 groupIndex = capturingGroupCount++; 2983 GroupHead head = new GroupHead(localIndex); 2984 root = new GroupTail(localIndex, groupIndex); 2985 if (!anonymous && groupIndex < 10) 2986 groupNodes[groupIndex] = head; 2987 return head; 2988 } 2989 2990 @SuppressWarnings("fallthrough") 2991 /** 2992 * Parses inlined match flags and set them appropriately. 2993 */ 2994 private void addFlag() { 2995 int ch = peek(); 2996 for (;;) { 2997 switch (ch) { 2998 case 'i': 2999 flags |= CASE_INSENSITIVE; 3000 break; 3001 case 'm': 3002 flags |= MULTILINE; 3003 break; 3004 case 's': 3005 flags |= DOTALL; 3006 break; 3007 case 'd': 3008 flags |= UNIX_LINES; 3009 break; 3010 case 'u': 3011 flags |= UNICODE_CASE; 3012 break; 3013 case 'c': 3014 flags |= CANON_EQ; 3015 break; 3016 case 'x': 3017 flags |= COMMENTS; 3018 break; 3019 case 'U': 3020 flags |= (UNICODE_CHARACTER_CLASS | UNICODE_CASE); 3021 break; 3022 case '-': // subFlag then fall through 3023 ch = next(); 3024 subFlag(); 3025 default: 3026 return; 3027 } 3028 ch = next(); 3029 } 3030 } 3031 3032 @SuppressWarnings("fallthrough") 3033 /** 3034 * Parses the second part of inlined match flags and turns off 3035 * flags appropriately. 3036 */ 3037 private void subFlag() { 3038 int ch = peek(); 3039 for (;;) { 3040 switch (ch) { 3041 case 'i': 3042 flags &= ~CASE_INSENSITIVE; 3043 break; 3044 case 'm': 3045 flags &= ~MULTILINE; 3046 break; 3047 case 's': 3048 flags &= ~DOTALL; 3049 break; 3050 case 'd': 3051 flags &= ~UNIX_LINES; 3052 break; 3053 case 'u': 3054 flags &= ~UNICODE_CASE; 3055 break; 3056 case 'c': 3057 flags &= ~CANON_EQ; 3058 break; 3059 case 'x': 3060 flags &= ~COMMENTS; 3061 break; 3062 case 'U': 3063 flags &= ~(UNICODE_CHARACTER_CLASS | UNICODE_CASE); 3064 default: 3065 return; 3066 } 3067 ch = next(); 3068 } 3069 } 3070 3071 static final int MAX_REPS = 0x7FFFFFFF; 3072 3073 static final int GREEDY = 0; 3074 3075 static final int LAZY = 1; 3076 3077 static final int POSSESSIVE = 2; 3078 3079 static final int INDEPENDENT = 3; 3080 3081 /** 3082 * Processes repetition. If the next character peeked is a quantifier 3083 * then new nodes must be appended to handle the repetition. 3084 * Prev could be a single or a group, so it could be a chain of nodes. 3085 */ 3086 private Node closure(Node prev) { 3087 Node atom; 3088 int ch = peek(); 3089 switch (ch) { 3090 case '?': 3091 ch = next(); 3092 if (ch == '?') { 3093 next(); 3094 return new Ques(prev, LAZY); 3095 } else if (ch == '+') { 3096 next(); 3097 return new Ques(prev, POSSESSIVE); 3098 } 3099 return new Ques(prev, GREEDY); 3100 case '*': 3101 ch = next(); 3102 if (ch == '?') { 3103 next(); 3104 return new Curly(prev, 0, MAX_REPS, LAZY); 3105 } else if (ch == '+') { 3106 next(); 3107 return new Curly(prev, 0, MAX_REPS, POSSESSIVE); 3108 } 3109 return new Curly(prev, 0, MAX_REPS, GREEDY); 3110 case '+': 3111 ch = next(); 3112 if (ch == '?') { 3113 next(); 3114 return new Curly(prev, 1, MAX_REPS, LAZY); 3115 } else if (ch == '+') { 3116 next(); 3117 return new Curly(prev, 1, MAX_REPS, POSSESSIVE); 3118 } 3119 return new Curly(prev, 1, MAX_REPS, GREEDY); 3120 case '{': 3121 ch = temp[cursor+1]; 3122 if (ASCII.isDigit(ch)) { 3123 skip(); 3124 int cmin = 0; 3125 do { 3126 cmin = cmin * 10 + (ch - '0'); 3127 } while (ASCII.isDigit(ch = read())); 3128 int cmax = cmin; 3129 if (ch == ',') { 3130 ch = read(); 3131 cmax = MAX_REPS; 3132 if (ch != '}') { 3133 cmax = 0; 3134 while (ASCII.isDigit(ch)) { 3135 cmax = cmax * 10 + (ch - '0'); 3136 ch = read(); 3137 } 3138 } 3139 } 3140 if (ch != '}') 3141 throw error("Unclosed counted closure"); 3142 if (((cmin) | (cmax) | (cmax - cmin)) < 0) 3143 throw error("Illegal repetition range"); 3144 Curly curly; 3145 ch = peek(); 3146 if (ch == '?') { 3147 next(); 3148 curly = new Curly(prev, cmin, cmax, LAZY); 3149 } else if (ch == '+') { 3150 next(); 3151 curly = new Curly(prev, cmin, cmax, POSSESSIVE); 3152 } else { 3153 curly = new Curly(prev, cmin, cmax, GREEDY); 3154 } 3155 return curly; 3156 } else { 3157 throw error("Illegal repetition"); 3158 } 3159 default: 3160 return prev; 3161 } 3162 } 3163 3164 /** 3165 * Utility method for parsing control escape sequences. 3166 */ 3167 private int c() { 3168 if (cursor < patternLength) { 3169 return read() ^ 64; 3170 } 3171 throw error("Illegal control escape sequence"); 3172 } 3173 3174 /** 3175 * Utility method for parsing octal escape sequences. 3176 */ 3177 private int o() { 3178 int n = read(); 3179 if (((n-'0')|('7'-n)) >= 0) { 3180 int m = read(); 3181 if (((m-'0')|('7'-m)) >= 0) { 3182 int o = read(); 3183 if ((((o-'0')|('7'-o)) >= 0) && (((n-'0')|('3'-n)) >= 0)) { 3184 return (n - '0') * 64 + (m - '0') * 8 + (o - '0'); 3185 } 3186 unread(); 3187 return (n - '0') * 8 + (m - '0'); 3188 } 3189 unread(); 3190 return (n - '0'); 3191 } 3192 throw error("Illegal octal escape sequence"); 3193 } 3194 3195 /** 3196 * Utility method for parsing hexadecimal escape sequences. 3197 */ 3198 private int x() { 3199 int n = read(); 3200 if (ASCII.isHexDigit(n)) { 3201 int m = read(); 3202 if (ASCII.isHexDigit(m)) { 3203 return ASCII.toDigit(n) * 16 + ASCII.toDigit(m); 3204 } 3205 } else if (n == '{' && ASCII.isHexDigit(peek())) { 3206 int ch = 0; 3207 while (ASCII.isHexDigit(n = read())) { 3208 ch = (ch << 4) + ASCII.toDigit(n); 3209 if (ch > Character.MAX_CODE_POINT) 3210 throw error("Hexadecimal codepoint is too big"); 3211 } 3212 if (n != '}') 3213 throw error("Unclosed hexadecimal escape sequence"); 3214 return ch; 3215 } 3216 throw error("Illegal hexadecimal escape sequence"); 3217 } 3218 3219 /** 3220 * Utility method for parsing unicode escape sequences. 3221 */ 3222 private int cursor() { 3223 return cursor; 3224 } 3225 3226 private void setcursor(int pos) { 3227 cursor = pos; 3228 } 3229 3230 private int uxxxx() { 3231 int n = 0; 3232 for (int i = 0; i < 4; i++) { 3233 int ch = read(); 3234 if (!ASCII.isHexDigit(ch)) { 3235 throw error("Illegal Unicode escape sequence"); 3236 } 3237 n = n * 16 + ASCII.toDigit(ch); 3238 } 3239 return n; 3240 } 3241 3242 private int u() { 3243 int n = uxxxx(); 3244 if (Character.isHighSurrogate((char)n)) { 3245 int cur = cursor(); 3246 if (read() == '\\' && read() == 'u') { 3247 int n2 = uxxxx(); 3248 if (Character.isLowSurrogate((char)n2)) 3249 return Character.toCodePoint((char)n, (char)n2); 3250 } 3251 setcursor(cur); 3252 } 3253 return n; 3254 } 3255 3256 // 3257 // Utility methods for code point support 3258 // 3259 3260 private static final int countChars(CharSequence seq, int index, 3261 int lengthInCodePoints) { 3262 // optimization 3263 if (lengthInCodePoints == 1 && !Character.isHighSurrogate(seq.charAt(index))) { 3264 assert (index >= 0 && index < seq.length()); 3265 return 1; 3266 } 3267 int length = seq.length(); 3268 int x = index; 3269 if (lengthInCodePoints >= 0) { 3270 assert (index >= 0 && index < length); 3271 for (int i = 0; x < length && i < lengthInCodePoints; i++) { 3272 if (Character.isHighSurrogate(seq.charAt(x++))) { 3273 if (x < length && Character.isLowSurrogate(seq.charAt(x))) { 3274 x++; 3275 } 3276 } 3277 } 3278 return x - index; 3279 } 3280 3281 assert (index >= 0 && index <= length); 3282 if (index == 0) { 3283 return 0; 3284 } 3285 int len = -lengthInCodePoints; 3286 for (int i = 0; x > 0 && i < len; i++) { 3287 if (Character.isLowSurrogate(seq.charAt(--x))) { 3288 if (x > 0 && Character.isHighSurrogate(seq.charAt(x-1))) { 3289 x--; 3290 } 3291 } 3292 } 3293 return index - x; 3294 } 3295 3296 private static final int countCodePoints(CharSequence seq) { 3297 int length = seq.length(); 3298 int n = 0; 3299 for (int i = 0; i < length; ) { 3300 n++; 3301 if (Character.isHighSurrogate(seq.charAt(i++))) { 3302 if (i < length && Character.isLowSurrogate(seq.charAt(i))) { 3303 i++; 3304 } 3305 } 3306 } 3307 return n; 3308 } 3309 3310 /** 3311 * Creates a bit vector for matching Latin-1 values. A normal BitClass 3312 * never matches values above Latin-1, and a complemented BitClass always 3313 * matches values above Latin-1. 3314 */ 3315 private static final class BitClass extends BmpCharProperty { 3316 final boolean[] bits; 3317 BitClass() { bits = new boolean[256]; } 3318 private BitClass(boolean[] bits) { this.bits = bits; } 3319 BitClass add(int c, int flags) { 3320 assert c >= 0 && c <= 255; 3321 if ((flags & CASE_INSENSITIVE) != 0) { 3322 if (ASCII.isAscii(c)) { 3323 bits[ASCII.toUpper(c)] = true; 3324 bits[ASCII.toLower(c)] = true; 3325 } else if ((flags & UNICODE_CASE) != 0) { 3326 bits[Character.toLowerCase(c)] = true; 3327 bits[Character.toUpperCase(c)] = true; 3328 } 3329 } 3330 bits[c] = true; 3331 return this; 3332 } 3333 boolean isSatisfiedBy(int ch) { 3334 return ch < 256 && bits[ch]; 3335 } 3336 } 3337 3338 /** 3339 * Returns a suitably optimized, single character matcher. 3340 */ 3341 private CharProperty newSingle(final int ch) { 3342 if (has(CASE_INSENSITIVE)) { 3343 int lower, upper; 3344 if (has(UNICODE_CASE)) { 3345 upper = Character.toUpperCase(ch); 3346 lower = Character.toLowerCase(upper); 3347 if (upper != lower) 3348 return new SingleU(lower); 3349 } else if (ASCII.isAscii(ch)) { 3350 lower = ASCII.toLower(ch); 3351 upper = ASCII.toUpper(ch); 3352 if (lower != upper) 3353 return new SingleI(lower, upper); 3354 } 3355 } 3356 if (isSupplementary(ch)) 3357 return new SingleS(ch); // Match a given Unicode character 3358 return new Single(ch); // Match a given BMP character 3359 } 3360 3361 /** 3362 * Utility method for creating a string slice matcher. 3363 */ 3364 private Node newSlice(int[] buf, int count, boolean hasSupplementary) { 3365 int[] tmp = new int[count]; 3366 if (has(CASE_INSENSITIVE)) { 3367 if (has(UNICODE_CASE)) { 3368 for (int i = 0; i < count; i++) { 3369 tmp[i] = Character.toLowerCase( 3370 Character.toUpperCase(buf[i])); 3371 } 3372 return hasSupplementary? new SliceUS(tmp) : new SliceU(tmp); 3373 } 3374 for (int i = 0; i < count; i++) { 3375 tmp[i] = ASCII.toLower(buf[i]); 3376 } 3377 return hasSupplementary? new SliceIS(tmp) : new SliceI(tmp); 3378 } 3379 for (int i = 0; i < count; i++) { 3380 tmp[i] = buf[i]; 3381 } 3382 return hasSupplementary ? new SliceS(tmp) : new Slice(tmp); 3383 } 3384 3385 /** 3386 * The following classes are the building components of the object 3387 * tree that represents a compiled regular expression. The object tree 3388 * is made of individual elements that handle constructs in the Pattern. 3389 * Each type of object knows how to match its equivalent construct with 3390 * the match() method. 3391 */ 3392 3393 /** 3394 * Base class for all node classes. Subclasses should override the match() 3395 * method as appropriate. This class is an accepting node, so its match() 3396 * always returns true. 3397 */ 3398 static class Node extends Object { 3399 Node next; 3400 Node() { 3401 next = Pattern.accept; 3402 } 3403 /** 3404 * This method implements the classic accept node. 3405 */ 3406 boolean match(Matcher matcher, int i, CharSequence seq) { 3407 matcher.last = i; 3408 matcher.groups[0] = matcher.first; 3409 matcher.groups[1] = matcher.last; 3410 return true; 3411 } 3412 /** 3413 * This method is good for all zero length assertions. 3414 */ 3415 boolean study(TreeInfo info) { 3416 if (next != null) { 3417 return next.study(info); 3418 } else { 3419 return info.deterministic; 3420 } 3421 } 3422 } 3423 3424 static class LastNode extends Node { 3425 /** 3426 * This method implements the classic accept node with 3427 * the addition of a check to see if the match occurred 3428 * using all of the input. 3429 */ 3430 boolean match(Matcher matcher, int i, CharSequence seq) { 3431 if (matcher.acceptMode == Matcher.ENDANCHOR && i != matcher.to) 3432 return false; 3433 matcher.last = i; 3434 matcher.groups[0] = matcher.first; 3435 matcher.groups[1] = matcher.last; 3436 return true; 3437 } 3438 } 3439 3440 /** 3441 * Used for REs that can start anywhere within the input string. 3442 * This basically tries to match repeatedly at each spot in the 3443 * input string, moving forward after each try. An anchored search 3444 * or a BnM will bypass this node completely. 3445 */ 3446 static class Start extends Node { 3447 int minLength; 3448 Start(Node node) { 3449 this.next = node; 3450 TreeInfo info = new TreeInfo(); 3451 next.study(info); 3452 minLength = info.minLength; 3453 } 3454 boolean match(Matcher matcher, int i, CharSequence seq) { 3455 if (i > matcher.to - minLength) { 3456 matcher.hitEnd = true; 3457 return false; 3458 } 3459 int guard = matcher.to - minLength; 3460 for (; i <= guard; i++) { 3461 if (next.match(matcher, i, seq)) { 3462 matcher.first = i; 3463 matcher.groups[0] = matcher.first; 3464 matcher.groups[1] = matcher.last; 3465 return true; 3466 } 3467 } 3468 matcher.hitEnd = true; 3469 return false; 3470 } 3471 boolean study(TreeInfo info) { 3472 next.study(info); 3473 info.maxValid = false; 3474 info.deterministic = false; 3475 return false; 3476 } 3477 } 3478 3479 /* 3480 * StartS supports supplementary characters, including unpaired surrogates. 3481 */ 3482 static final class StartS extends Start { 3483 StartS(Node node) { 3484 super(node); 3485 } 3486 boolean match(Matcher matcher, int i, CharSequence seq) { 3487 if (i > matcher.to - minLength) { 3488 matcher.hitEnd = true; 3489 return false; 3490 } 3491 int guard = matcher.to - minLength; 3492 while (i <= guard) { 3493 //if ((ret = next.match(matcher, i, seq)) || i == guard) 3494 if (next.match(matcher, i, seq)) { 3495 matcher.first = i; 3496 matcher.groups[0] = matcher.first; 3497 matcher.groups[1] = matcher.last; 3498 return true; 3499 } 3500 if (i == guard) 3501 break; 3502 // Optimization to move to the next character. This is 3503 // faster than countChars(seq, i, 1). 3504 if (Character.isHighSurrogate(seq.charAt(i++))) { 3505 if (i < seq.length() && 3506 Character.isLowSurrogate(seq.charAt(i))) { 3507 i++; 3508 } 3509 } 3510 } 3511 matcher.hitEnd = true; 3512 return false; 3513 } 3514 } 3515 3516 /** 3517 * Node to anchor at the beginning of input. This object implements the 3518 * match for a \A sequence, and the caret anchor will use this if not in 3519 * multiline mode. 3520 */ 3521 static final class Begin extends Node { 3522 boolean match(Matcher matcher, int i, CharSequence seq) { 3523 int fromIndex = (matcher.anchoringBounds) ? 3524 matcher.from : 0; 3525 if (i == fromIndex && next.match(matcher, i, seq)) { 3526 matcher.first = i; 3527 matcher.groups[0] = i; 3528 matcher.groups[1] = matcher.last; 3529 return true; 3530 } else { 3531 return false; 3532 } 3533 } 3534 } 3535 3536 /** 3537 * Node to anchor at the end of input. This is the absolute end, so this 3538 * should not match at the last newline before the end as $ will. 3539 */ 3540 static final class End extends Node { 3541 boolean match(Matcher matcher, int i, CharSequence seq) { 3542 int endIndex = (matcher.anchoringBounds) ? 3543 matcher.to : matcher.getTextLength(); 3544 if (i == endIndex) { 3545 matcher.hitEnd = true; 3546 return next.match(matcher, i, seq); 3547 } 3548 return false; 3549 } 3550 } 3551 3552 /** 3553 * Node to anchor at the beginning of a line. This is essentially the 3554 * object to match for the multiline ^. 3555 */ 3556 static final class Caret extends Node { 3557 boolean match(Matcher matcher, int i, CharSequence seq) { 3558 int startIndex = matcher.from; 3559 int endIndex = matcher.to; 3560 if (!matcher.anchoringBounds) { 3561 startIndex = 0; 3562 endIndex = matcher.getTextLength(); 3563 } 3564 // Perl does not match ^ at end of input even after newline 3565 if (i == endIndex) { 3566 matcher.hitEnd = true; 3567 return false; 3568 } 3569 if (i > startIndex) { 3570 char ch = seq.charAt(i-1); 3571 if (ch != '\n' && ch != '\r' 3572 && (ch|1) != '\u2029' 3573 && ch != '\u0085' ) { 3574 return false; 3575 } 3576 // Should treat /r/n as one newline 3577 if (ch == '\r' && seq.charAt(i) == '\n') 3578 return false; 3579 } 3580 return next.match(matcher, i, seq); 3581 } 3582 } 3583 3584 /** 3585 * Node to anchor at the beginning of a line when in unixdot mode. 3586 */ 3587 static final class UnixCaret extends Node { 3588 boolean match(Matcher matcher, int i, CharSequence seq) { 3589 int startIndex = matcher.from; 3590 int endIndex = matcher.to; 3591 if (!matcher.anchoringBounds) { 3592 startIndex = 0; 3593 endIndex = matcher.getTextLength(); 3594 } 3595 // Perl does not match ^ at end of input even after newline 3596 if (i == endIndex) { 3597 matcher.hitEnd = true; 3598 return false; 3599 } 3600 if (i > startIndex) { 3601 char ch = seq.charAt(i-1); 3602 if (ch != '\n') { 3603 return false; 3604 } 3605 } 3606 return next.match(matcher, i, seq); 3607 } 3608 } 3609 3610 /** 3611 * Node to match the location where the last match ended. 3612 * This is used for the \G construct. 3613 */ 3614 static final class LastMatch extends Node { 3615 boolean match(Matcher matcher, int i, CharSequence seq) { 3616 if (i != matcher.oldLast) 3617 return false; 3618 return next.match(matcher, i, seq); 3619 } 3620 } 3621 3622 /** 3623 * Node to anchor at the end of a line or the end of input based on the 3624 * multiline mode. 3625 * 3626 * When not in multiline mode, the $ can only match at the very end 3627 * of the input, unless the input ends in a line terminator in which 3628 * it matches right before the last line terminator. 3629 * 3630 * Note that \r\n is considered an atomic line terminator. 3631 * 3632 * Like ^ the $ operator matches at a position, it does not match the 3633 * line terminators themselves. 3634 */ 3635 static final class Dollar extends Node { 3636 boolean multiline; 3637 Dollar(boolean mul) { 3638 multiline = mul; 3639 } 3640 boolean match(Matcher matcher, int i, CharSequence seq) { 3641 int endIndex = (matcher.anchoringBounds) ? 3642 matcher.to : matcher.getTextLength(); 3643 if (!multiline) { 3644 if (i < endIndex - 2) 3645 return false; 3646 if (i == endIndex - 2) { 3647 char ch = seq.charAt(i); 3648 if (ch != '\r') 3649 return false; 3650 ch = seq.charAt(i + 1); 3651 if (ch != '\n') 3652 return false; 3653 } 3654 } 3655 // Matches before any line terminator; also matches at the 3656 // end of input 3657 // Before line terminator: 3658 // If multiline, we match here no matter what 3659 // If not multiline, fall through so that the end 3660 // is marked as hit; this must be a /r/n or a /n 3661 // at the very end so the end was hit; more input 3662 // could make this not match here 3663 if (i < endIndex) { 3664 char ch = seq.charAt(i); 3665 if (ch == '\n') { 3666 // No match between \r\n 3667 if (i > 0 && seq.charAt(i-1) == '\r') 3668 return false; 3669 if (multiline) 3670 return next.match(matcher, i, seq); 3671 } else if (ch == '\r' || ch == '\u0085' || 3672 (ch|1) == '\u2029') { 3673 if (multiline) 3674 return next.match(matcher, i, seq); 3675 } else { // No line terminator, no match 3676 return false; 3677 } 3678 } 3679 // Matched at current end so hit end 3680 matcher.hitEnd = true; 3681 // If a $ matches because of end of input, then more input 3682 // could cause it to fail! 3683 matcher.requireEnd = true; 3684 return next.match(matcher, i, seq); 3685 } 3686 boolean study(TreeInfo info) { 3687 next.study(info); 3688 return info.deterministic; 3689 } 3690 } 3691 3692 /** 3693 * Node to anchor at the end of a line or the end of input based on the 3694 * multiline mode when in unix lines mode. 3695 */ 3696 static final class UnixDollar extends Node { 3697 boolean multiline; 3698 UnixDollar(boolean mul) { 3699 multiline = mul; 3700 } 3701 boolean match(Matcher matcher, int i, CharSequence seq) { 3702 int endIndex = (matcher.anchoringBounds) ? 3703 matcher.to : matcher.getTextLength(); 3704 if (i < endIndex) { 3705 char ch = seq.charAt(i); 3706 if (ch == '\n') { 3707 // If not multiline, then only possible to 3708 // match at very end or one before end 3709 if (multiline == false && i != endIndex - 1) 3710 return false; 3711 // If multiline return next.match without setting 3712 // matcher.hitEnd 3713 if (multiline) 3714 return next.match(matcher, i, seq); 3715 } else { 3716 return false; 3717 } 3718 } 3719 // Matching because at the end or 1 before the end; 3720 // more input could change this so set hitEnd 3721 matcher.hitEnd = true; 3722 // If a $ matches because of end of input, then more input 3723 // could cause it to fail! 3724 matcher.requireEnd = true; 3725 return next.match(matcher, i, seq); 3726 } 3727 boolean study(TreeInfo info) { 3728 next.study(info); 3729 return info.deterministic; 3730 } 3731 } 3732 3733 /** 3734 * Node class that matches a Unicode line ending '\R' 3735 */ 3736 static final class LineEnding extends Node { 3737 boolean match(Matcher matcher, int i, CharSequence seq) { 3738 // (u+000Du+000A|[u+000Au+000Bu+000Cu+000Du+0085u+2028u+2029]) 3739 if (i < matcher.to) { 3740 int ch = seq.charAt(i); 3741 if (ch == 0x0A || ch == 0x0B || ch == 0x0C || 3742 ch == 0x85 || ch == 0x2028 || ch == 0x2029) 3743 return next.match(matcher, i + 1, seq); 3744 if (ch == 0x0D) { 3745 i++; 3746 if (i < matcher.to && seq.charAt(i) == 0x0A) 3747 i++; 3748 return next.match(matcher, i, seq); 3749 } 3750 } else { 3751 matcher.hitEnd = true; 3752 } 3753 return false; 3754 } 3755 boolean study(TreeInfo info) { 3756 info.minLength++; 3757 info.maxLength += 2; 3758 return next.study(info); 3759 } 3760 } 3761 3762 /** 3763 * Abstract node class to match one character satisfying some 3764 * boolean property. 3765 */ 3766 private static abstract class CharProperty extends Node { 3767 abstract boolean isSatisfiedBy(int ch); 3768 CharProperty complement() { 3769 return new CharProperty() { 3770 boolean isSatisfiedBy(int ch) { 3771 return ! CharProperty.this.isSatisfiedBy(ch);}}; 3772 } 3773 boolean match(Matcher matcher, int i, CharSequence seq) { 3774 if (i < matcher.to) { 3775 int ch = Character.codePointAt(seq, i); 3776 return isSatisfiedBy(ch) 3777 && next.match(matcher, i+Character.charCount(ch), seq); 3778 } else { 3779 matcher.hitEnd = true; 3780 return false; 3781 } 3782 } 3783 boolean study(TreeInfo info) { 3784 info.minLength++; 3785 info.maxLength++; 3786 return next.study(info); 3787 } 3788 } 3789 3790 /** 3791 * Optimized version of CharProperty that works only for 3792 * properties never satisfied by Supplementary characters. 3793 */ 3794 private static abstract class BmpCharProperty extends CharProperty { 3795 boolean match(Matcher matcher, int i, CharSequence seq) { 3796 if (i < matcher.to) { 3797 return isSatisfiedBy(seq.charAt(i)) 3798 && next.match(matcher, i+1, seq); 3799 } else { 3800 matcher.hitEnd = true; 3801 return false; 3802 } 3803 } 3804 } 3805 3806 /** 3807 * Node class that matches a Supplementary Unicode character 3808 */ 3809 static final class SingleS extends CharProperty { 3810 final int c; 3811 SingleS(int c) { this.c = c; } 3812 boolean isSatisfiedBy(int ch) { 3813 return ch == c; 3814 } 3815 } 3816 3817 /** 3818 * Optimization -- matches a given BMP character 3819 */ 3820 static final class Single extends BmpCharProperty { 3821 final int c; 3822 Single(int c) { this.c = c; } 3823 boolean isSatisfiedBy(int ch) { 3824 return ch == c; 3825 } 3826 } 3827 3828 /** 3829 * Case insensitive matches a given BMP character 3830 */ 3831 static final class SingleI extends BmpCharProperty { 3832 final int lower; 3833 final int upper; 3834 SingleI(int lower, int upper) { 3835 this.lower = lower; 3836 this.upper = upper; 3837 } 3838 boolean isSatisfiedBy(int ch) { 3839 return ch == lower || ch == upper; 3840 } 3841 } 3842 3843 /** 3844 * Unicode case insensitive matches a given Unicode character 3845 */ 3846 static final class SingleU extends CharProperty { 3847 final int lower; 3848 SingleU(int lower) { 3849 this.lower = lower; 3850 } 3851 boolean isSatisfiedBy(int ch) { 3852 return lower == ch || 3853 lower == Character.toLowerCase(Character.toUpperCase(ch)); 3854 } 3855 } 3856 3857 /** 3858 * Node class that matches a Unicode block. 3859 */ 3860 static final class Block extends CharProperty { 3861 final Character.UnicodeBlock block; 3862 Block(Character.UnicodeBlock block) { 3863 this.block = block; 3864 } 3865 boolean isSatisfiedBy(int ch) { 3866 return block == Character.UnicodeBlock.of(ch); 3867 } 3868 } 3869 3870 /** 3871 * Node class that matches a Unicode script 3872 */ 3873 static final class Script extends CharProperty { 3874 final Character.UnicodeScript script; 3875 Script(Character.UnicodeScript script) { 3876 this.script = script; 3877 } 3878 boolean isSatisfiedBy(int ch) { 3879 return script == Character.UnicodeScript.of(ch); 3880 } 3881 } 3882 3883 /** 3884 * Node class that matches a Unicode category. 3885 */ 3886 static final class Category extends CharProperty { 3887 final int typeMask; 3888 Category(int typeMask) { this.typeMask = typeMask; } 3889 boolean isSatisfiedBy(int ch) { 3890 return (typeMask & (1 << Character.getType(ch))) != 0; 3891 } 3892 } 3893 3894 /** 3895 * Node class that matches a Unicode "type" 3896 */ 3897 static final class Utype extends CharProperty { 3898 final UnicodeProp uprop; 3899 Utype(UnicodeProp uprop) { this.uprop = uprop; } 3900 boolean isSatisfiedBy(int ch) { 3901 return uprop.is(ch); 3902 } 3903 } 3904 3905 /** 3906 * Node class that matches a POSIX type. 3907 */ 3908 static final class Ctype extends BmpCharProperty { 3909 final int ctype; 3910 Ctype(int ctype) { this.ctype = ctype; } 3911 boolean isSatisfiedBy(int ch) { 3912 return ch < 128 && ASCII.isType(ch, ctype); 3913 } 3914 } 3915 3916 /** 3917 * Node class that matches a Perl vertical whitespace 3918 */ 3919 static final class VertWS extends BmpCharProperty { 3920 boolean isSatisfiedBy(int cp) { 3921 return (cp >= 0x0A && cp <= 0x0D) || 3922 cp == 0x85 || cp == 0x2028 || cp == 0x2029; 3923 } 3924 } 3925 3926 /** 3927 * Node class that matches a Perl horizontal whitespace 3928 */ 3929 static final class HorizWS extends BmpCharProperty { 3930 boolean isSatisfiedBy(int cp) { 3931 return cp == 0x09 || cp == 0x20 || cp == 0xa0 || 3932 cp == 0x1680 || cp == 0x180e || 3933 cp >= 0x2000 && cp <= 0x200a || 3934 cp == 0x202f || cp == 0x205f || cp == 0x3000; 3935 } 3936 } 3937 3938 /** 3939 * Base class for all Slice nodes 3940 */ 3941 static class SliceNode extends Node { 3942 int[] buffer; 3943 SliceNode(int[] buf) { 3944 buffer = buf; 3945 } 3946 boolean study(TreeInfo info) { 3947 info.minLength += buffer.length; 3948 info.maxLength += buffer.length; 3949 return next.study(info); 3950 } 3951 } 3952 3953 /** 3954 * Node class for a case sensitive/BMP-only sequence of literal 3955 * characters. 3956 */ 3957 static final class Slice extends SliceNode { 3958 Slice(int[] buf) { 3959 super(buf); 3960 } 3961 boolean match(Matcher matcher, int i, CharSequence seq) { 3962 int[] buf = buffer; 3963 int len = buf.length; 3964 for (int j=0; j<len; j++) { 3965 if ((i+j) >= matcher.to) { 3966 matcher.hitEnd = true; 3967 return false; 3968 } 3969 if (buf[j] != seq.charAt(i+j)) 3970 return false; 3971 } 3972 return next.match(matcher, i+len, seq); 3973 } 3974 } 3975 3976 /** 3977 * Node class for a case_insensitive/BMP-only sequence of literal 3978 * characters. 3979 */ 3980 static class SliceI extends SliceNode { 3981 SliceI(int[] buf) { 3982 super(buf); 3983 } 3984 boolean match(Matcher matcher, int i, CharSequence seq) { 3985 int[] buf = buffer; 3986 int len = buf.length; 3987 for (int j=0; j<len; j++) { 3988 if ((i+j) >= matcher.to) { 3989 matcher.hitEnd = true; 3990 return false; 3991 } 3992 int c = seq.charAt(i+j); 3993 if (buf[j] != c && 3994 buf[j] != ASCII.toLower(c)) 3995 return false; 3996 } 3997 return next.match(matcher, i+len, seq); 3998 } 3999 } 4000 4001 /** 4002 * Node class for a unicode_case_insensitive/BMP-only sequence of 4003 * literal characters. Uses unicode case folding. 4004 */ 4005 static final class SliceU extends SliceNode { 4006 SliceU(int[] buf) { 4007 super(buf); 4008 } 4009 boolean match(Matcher matcher, int i, CharSequence seq) { 4010 int[] buf = buffer; 4011 int len = buf.length; 4012 for (int j=0; j<len; j++) { 4013 if ((i+j) >= matcher.to) { 4014 matcher.hitEnd = true; 4015 return false; 4016 } 4017 int c = seq.charAt(i+j); 4018 if (buf[j] != c && 4019 buf[j] != Character.toLowerCase(Character.toUpperCase(c))) 4020 return false; 4021 } 4022 return next.match(matcher, i+len, seq); 4023 } 4024 } 4025 4026 /** 4027 * Node class for a case sensitive sequence of literal characters 4028 * including supplementary characters. 4029 */ 4030 static final class SliceS extends SliceNode { 4031 SliceS(int[] buf) { 4032 super(buf); 4033 } 4034 boolean match(Matcher matcher, int i, CharSequence seq) { 4035 int[] buf = buffer; 4036 int x = i; 4037 for (int j = 0; j < buf.length; j++) { 4038 if (x >= matcher.to) { 4039 matcher.hitEnd = true; 4040 return false; 4041 } 4042 int c = Character.codePointAt(seq, x); 4043 if (buf[j] != c) 4044 return false; 4045 x += Character.charCount(c); 4046 if (x > matcher.to) { 4047 matcher.hitEnd = true; 4048 return false; 4049 } 4050 } 4051 return next.match(matcher, x, seq); 4052 } 4053 } 4054 4055 /** 4056 * Node class for a case insensitive sequence of literal characters 4057 * including supplementary characters. 4058 */ 4059 static class SliceIS extends SliceNode { 4060 SliceIS(int[] buf) { 4061 super(buf); 4062 } 4063 int toLower(int c) { 4064 return ASCII.toLower(c); 4065 } 4066 boolean match(Matcher matcher, int i, CharSequence seq) { 4067 int[] buf = buffer; 4068 int x = i; 4069 for (int j = 0; j < buf.length; j++) { 4070 if (x >= matcher.to) { 4071 matcher.hitEnd = true; 4072 return false; 4073 } 4074 int c = Character.codePointAt(seq, x); 4075 if (buf[j] != c && buf[j] != toLower(c)) 4076 return false; 4077 x += Character.charCount(c); 4078 if (x > matcher.to) { 4079 matcher.hitEnd = true; 4080 return false; 4081 } 4082 } 4083 return next.match(matcher, x, seq); 4084 } 4085 } 4086 4087 /** 4088 * Node class for a case insensitive sequence of literal characters. 4089 * Uses unicode case folding. 4090 */ 4091 static final class SliceUS extends SliceIS { 4092 SliceUS(int[] buf) { 4093 super(buf); 4094 } 4095 int toLower(int c) { 4096 return Character.toLowerCase(Character.toUpperCase(c)); 4097 } 4098 } 4099 4100 private static boolean inRange(int lower, int ch, int upper) { 4101 return lower <= ch && ch <= upper; 4102 } 4103 4104 /** 4105 * Returns node for matching characters within an explicit value range. 4106 */ 4107 private static CharProperty rangeFor(final int lower, 4108 final int upper) { 4109 return new CharProperty() { 4110 boolean isSatisfiedBy(int ch) { 4111 return inRange(lower, ch, upper);}}; 4112 } 4113 4114 /** 4115 * Returns node for matching characters within an explicit value 4116 * range in a case insensitive manner. 4117 */ 4118 private CharProperty caseInsensitiveRangeFor(final int lower, 4119 final int upper) { 4120 if (has(UNICODE_CASE)) 4121 return new CharProperty() { 4122 boolean isSatisfiedBy(int ch) { 4123 if (inRange(lower, ch, upper)) 4124 return true; 4125 int up = Character.toUpperCase(ch); 4126 return inRange(lower, up, upper) || 4127 inRange(lower, Character.toLowerCase(up), upper);}}; 4128 return new CharProperty() { 4129 boolean isSatisfiedBy(int ch) { 4130 return inRange(lower, ch, upper) || 4131 ASCII.isAscii(ch) && 4132 (inRange(lower, ASCII.toUpper(ch), upper) || 4133 inRange(lower, ASCII.toLower(ch), upper)); 4134 }}; 4135 } 4136 4137 /** 4138 * Implements the Unicode category ALL and the dot metacharacter when 4139 * in dotall mode. 4140 */ 4141 static final class All extends CharProperty { 4142 boolean isSatisfiedBy(int ch) { 4143 return true; 4144 } 4145 } 4146 4147 /** 4148 * Node class for the dot metacharacter when dotall is not enabled. 4149 */ 4150 static final class Dot extends CharProperty { 4151 boolean isSatisfiedBy(int ch) { 4152 return (ch != '\n' && ch != '\r' 4153 && (ch|1) != '\u2029' 4154 && ch != '\u0085'); 4155 } 4156 } 4157 4158 /** 4159 * Node class for the dot metacharacter when dotall is not enabled 4160 * but UNIX_LINES is enabled. 4161 */ 4162 static final class UnixDot extends CharProperty { 4163 boolean isSatisfiedBy(int ch) { 4164 return ch != '\n'; 4165 } 4166 } 4167 4168 /** 4169 * The 0 or 1 quantifier. This one class implements all three types. 4170 */ 4171 static final class Ques extends Node { 4172 Node atom; 4173 int type; 4174 Ques(Node node, int type) { 4175 this.atom = node; 4176 this.type = type; 4177 } 4178 boolean match(Matcher matcher, int i, CharSequence seq) { 4179 switch (type) { 4180 case GREEDY: 4181 return (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)) 4182 || next.match(matcher, i, seq); 4183 case LAZY: 4184 return next.match(matcher, i, seq) 4185 || (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)); 4186 case POSSESSIVE: 4187 if (atom.match(matcher, i, seq)) i = matcher.last; 4188 return next.match(matcher, i, seq); 4189 default: 4190 return atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq); 4191 } 4192 } 4193 boolean study(TreeInfo info) { 4194 if (type != INDEPENDENT) { 4195 int minL = info.minLength; 4196 atom.study(info); 4197 info.minLength = minL; 4198 info.deterministic = false; 4199 return next.study(info); 4200 } else { 4201 atom.study(info); 4202 return next.study(info); 4203 } 4204 } 4205 } 4206 4207 /** 4208 * Handles the curly-brace style repetition with a specified minimum and 4209 * maximum occurrences. The * quantifier is handled as a special case. 4210 * This class handles the three types. 4211 */ 4212 static final class Curly extends Node { 4213 Node atom; 4214 int type; 4215 int cmin; 4216 int cmax; 4217 4218 Curly(Node node, int cmin, int cmax, int type) { 4219 this.atom = node; 4220 this.type = type; 4221 this.cmin = cmin; 4222 this.cmax = cmax; 4223 } 4224 boolean match(Matcher matcher, int i, CharSequence seq) { 4225 int j; 4226 for (j = 0; j < cmin; j++) { 4227 if (atom.match(matcher, i, seq)) { 4228 i = matcher.last; 4229 continue; 4230 } 4231 return false; 4232 } 4233 if (type == GREEDY) 4234 return match0(matcher, i, j, seq); 4235 else if (type == LAZY) 4236 return match1(matcher, i, j, seq); 4237 else 4238 return match2(matcher, i, j, seq); 4239 } 4240 // Greedy match. 4241 // i is the index to start matching at 4242 // j is the number of atoms that have matched 4243 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4244 if (j >= cmax) { 4245 // We have matched the maximum... continue with the rest of 4246 // the regular expression 4247 return next.match(matcher, i, seq); 4248 } 4249 int backLimit = j; 4250 while (atom.match(matcher, i, seq)) { 4251 // k is the length of this match 4252 int k = matcher.last - i; 4253 if (k == 0) // Zero length match 4254 break; 4255 // Move up index and number matched 4256 i = matcher.last; 4257 j++; 4258 // We are greedy so match as many as we can 4259 while (j < cmax) { 4260 if (!atom.match(matcher, i, seq)) 4261 break; 4262 if (i + k != matcher.last) { 4263 if (match0(matcher, matcher.last, j+1, seq)) 4264 return true; 4265 break; 4266 } 4267 i += k; 4268 j++; 4269 } 4270 // Handle backing off if match fails 4271 while (j >= backLimit) { 4272 if (next.match(matcher, i, seq)) 4273 return true; 4274 i -= k; 4275 j--; 4276 } 4277 return false; 4278 } 4279 return next.match(matcher, i, seq); 4280 } 4281 // Reluctant match. At this point, the minimum has been satisfied. 4282 // i is the index to start matching at 4283 // j is the number of atoms that have matched 4284 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4285 for (;;) { 4286 // Try finishing match without consuming any more 4287 if (next.match(matcher, i, seq)) 4288 return true; 4289 // At the maximum, no match found 4290 if (j >= cmax) 4291 return false; 4292 // Okay, must try one more atom 4293 if (!atom.match(matcher, i, seq)) 4294 return false; 4295 // If we haven't moved forward then must break out 4296 if (i == matcher.last) 4297 return false; 4298 // Move up index and number matched 4299 i = matcher.last; 4300 j++; 4301 } 4302 } 4303 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4304 for (; j < cmax; j++) { 4305 if (!atom.match(matcher, i, seq)) 4306 break; 4307 if (i == matcher.last) 4308 break; 4309 i = matcher.last; 4310 } 4311 return next.match(matcher, i, seq); 4312 } 4313 boolean study(TreeInfo info) { 4314 // Save original info 4315 int minL = info.minLength; 4316 int maxL = info.maxLength; 4317 boolean maxV = info.maxValid; 4318 boolean detm = info.deterministic; 4319 info.reset(); 4320 4321 atom.study(info); 4322 4323 int temp = info.minLength * cmin + minL; 4324 if (temp < minL) { 4325 temp = 0xFFFFFFF; // arbitrary large number 4326 } 4327 info.minLength = temp; 4328 4329 if (maxV & info.maxValid) { 4330 temp = info.maxLength * cmax + maxL; 4331 info.maxLength = temp; 4332 if (temp < maxL) { 4333 info.maxValid = false; 4334 } 4335 } else { 4336 info.maxValid = false; 4337 } 4338 4339 if (info.deterministic && cmin == cmax) 4340 info.deterministic = detm; 4341 else 4342 info.deterministic = false; 4343 return next.study(info); 4344 } 4345 } 4346 4347 /** 4348 * Handles the curly-brace style repetition with a specified minimum and 4349 * maximum occurrences in deterministic cases. This is an iterative 4350 * optimization over the Prolog and Loop system which would handle this 4351 * in a recursive way. The * quantifier is handled as a special case. 4352 * If capture is true then this class saves group settings and ensures 4353 * that groups are unset when backing off of a group match. 4354 */ 4355 static final class GroupCurly extends Node { 4356 Node atom; 4357 int type; 4358 int cmin; 4359 int cmax; 4360 int localIndex; 4361 int groupIndex; 4362 boolean capture; 4363 4364 GroupCurly(Node node, int cmin, int cmax, int type, int local, 4365 int group, boolean capture) { 4366 this.atom = node; 4367 this.type = type; 4368 this.cmin = cmin; 4369 this.cmax = cmax; 4370 this.localIndex = local; 4371 this.groupIndex = group; 4372 this.capture = capture; 4373 } 4374 boolean match(Matcher matcher, int i, CharSequence seq) { 4375 int[] groups = matcher.groups; 4376 int[] locals = matcher.locals; 4377 int save0 = locals[localIndex]; 4378 int save1 = 0; 4379 int save2 = 0; 4380 4381 if (capture) { 4382 save1 = groups[groupIndex]; 4383 save2 = groups[groupIndex+1]; 4384 } 4385 4386 // Notify GroupTail there is no need to setup group info 4387 // because it will be set here 4388 locals[localIndex] = -1; 4389 4390 boolean ret = true; 4391 for (int j = 0; j < cmin; j++) { 4392 if (atom.match(matcher, i, seq)) { 4393 if (capture) { 4394 groups[groupIndex] = i; 4395 groups[groupIndex+1] = matcher.last; 4396 } 4397 i = matcher.last; 4398 } else { 4399 ret = false; 4400 break; 4401 } 4402 } 4403 if (ret) { 4404 if (type == GREEDY) { 4405 ret = match0(matcher, i, cmin, seq); 4406 } else if (type == LAZY) { 4407 ret = match1(matcher, i, cmin, seq); 4408 } else { 4409 ret = match2(matcher, i, cmin, seq); 4410 } 4411 } 4412 if (!ret) { 4413 locals[localIndex] = save0; 4414 if (capture) { 4415 groups[groupIndex] = save1; 4416 groups[groupIndex+1] = save2; 4417 } 4418 } 4419 return ret; 4420 } 4421 // Aggressive group match 4422 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4423 // don't back off passing the starting "j" 4424 int min = j; 4425 int[] groups = matcher.groups; 4426 int save0 = 0; 4427 int save1 = 0; 4428 if (capture) { 4429 save0 = groups[groupIndex]; 4430 save1 = groups[groupIndex+1]; 4431 } 4432 for (;;) { 4433 if (j >= cmax) 4434 break; 4435 if (!atom.match(matcher, i, seq)) 4436 break; 4437 int k = matcher.last - i; 4438 if (k <= 0) { 4439 if (capture) { 4440 groups[groupIndex] = i; 4441 groups[groupIndex+1] = i + k; 4442 } 4443 i = i + k; 4444 break; 4445 } 4446 for (;;) { 4447 if (capture) { 4448 groups[groupIndex] = i; 4449 groups[groupIndex+1] = i + k; 4450 } 4451 i = i + k; 4452 if (++j >= cmax) 4453 break; 4454 if (!atom.match(matcher, i, seq)) 4455 break; 4456 if (i + k != matcher.last) { 4457 if (match0(matcher, i, j, seq)) 4458 return true; 4459 break; 4460 } 4461 } 4462 while (j > min) { 4463 if (next.match(matcher, i, seq)) { 4464 if (capture) { 4465 groups[groupIndex+1] = i; 4466 groups[groupIndex] = i - k; 4467 } 4468 return true; 4469 } 4470 // backing off 4471 i = i - k; 4472 if (capture) { 4473 groups[groupIndex+1] = i; 4474 groups[groupIndex] = i - k; 4475 } 4476 j--; 4477 4478 } 4479 break; 4480 } 4481 if (capture) { 4482 groups[groupIndex] = save0; 4483 groups[groupIndex+1] = save1; 4484 } 4485 return next.match(matcher, i, seq); 4486 } 4487 // Reluctant matching 4488 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4489 for (;;) { 4490 if (next.match(matcher, i, seq)) 4491 return true; 4492 if (j >= cmax) 4493 return false; 4494 if (!atom.match(matcher, i, seq)) 4495 return false; 4496 if (i == matcher.last) 4497 return false; 4498 if (capture) { 4499 matcher.groups[groupIndex] = i; 4500 matcher.groups[groupIndex+1] = matcher.last; 4501 } 4502 i = matcher.last; 4503 j++; 4504 } 4505 } 4506 // Possessive matching 4507 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4508 for (; j < cmax; j++) { 4509 if (!atom.match(matcher, i, seq)) { 4510 break; 4511 } 4512 if (capture) { 4513 matcher.groups[groupIndex] = i; 4514 matcher.groups[groupIndex+1] = matcher.last; 4515 } 4516 if (i == matcher.last) { 4517 break; 4518 } 4519 i = matcher.last; 4520 } 4521 return next.match(matcher, i, seq); 4522 } 4523 boolean study(TreeInfo info) { 4524 // Save original info 4525 int minL = info.minLength; 4526 int maxL = info.maxLength; 4527 boolean maxV = info.maxValid; 4528 boolean detm = info.deterministic; 4529 info.reset(); 4530 4531 atom.study(info); 4532 4533 int temp = info.minLength * cmin + minL; 4534 if (temp < minL) { 4535 temp = 0xFFFFFFF; // Arbitrary large number 4536 } 4537 info.minLength = temp; 4538 4539 if (maxV & info.maxValid) { 4540 temp = info.maxLength * cmax + maxL; 4541 info.maxLength = temp; 4542 if (temp < maxL) { 4543 info.maxValid = false; 4544 } 4545 } else { 4546 info.maxValid = false; 4547 } 4548 4549 if (info.deterministic && cmin == cmax) { 4550 info.deterministic = detm; 4551 } else { 4552 info.deterministic = false; 4553 } 4554 return next.study(info); 4555 } 4556 } 4557 4558 /** 4559 * A Guard node at the end of each atom node in a Branch. It 4560 * serves the purpose of chaining the "match" operation to 4561 * "next" but not the "study", so we can collect the TreeInfo 4562 * of each atom node without including the TreeInfo of the 4563 * "next". 4564 */ 4565 static final class BranchConn extends Node { 4566 BranchConn() {}; 4567 boolean match(Matcher matcher, int i, CharSequence seq) { 4568 return next.match(matcher, i, seq); 4569 } 4570 boolean study(TreeInfo info) { 4571 return info.deterministic; 4572 } 4573 } 4574 4575 /** 4576 * Handles the branching of alternations. Note this is also used for 4577 * the ? quantifier to branch between the case where it matches once 4578 * and where it does not occur. 4579 */ 4580 static final class Branch extends Node { 4581 Node[] atoms = new Node[2]; 4582 int size = 2; 4583 Node conn; 4584 Branch(Node first, Node second, Node branchConn) { 4585 conn = branchConn; 4586 atoms[0] = first; 4587 atoms[1] = second; 4588 } 4589 4590 void add(Node node) { 4591 if (size >= atoms.length) { 4592 Node[] tmp = new Node[atoms.length*2]; 4593 System.arraycopy(atoms, 0, tmp, 0, atoms.length); 4594 atoms = tmp; 4595 } 4596 atoms[size++] = node; 4597 } 4598 4599 boolean match(Matcher matcher, int i, CharSequence seq) { 4600 for (int n = 0; n < size; n++) { 4601 if (atoms[n] == null) { 4602 if (conn.next.match(matcher, i, seq)) 4603 return true; 4604 } else if (atoms[n].match(matcher, i, seq)) { 4605 return true; 4606 } 4607 } 4608 return false; 4609 } 4610 4611 boolean study(TreeInfo info) { 4612 int minL = info.minLength; 4613 int maxL = info.maxLength; 4614 boolean maxV = info.maxValid; 4615 4616 int minL2 = Integer.MAX_VALUE; //arbitrary large enough num 4617 int maxL2 = -1; 4618 for (int n = 0; n < size; n++) { 4619 info.reset(); 4620 if (atoms[n] != null) 4621 atoms[n].study(info); 4622 minL2 = Math.min(minL2, info.minLength); 4623 maxL2 = Math.max(maxL2, info.maxLength); 4624 maxV = (maxV & info.maxValid); 4625 } 4626 4627 minL += minL2; 4628 maxL += maxL2; 4629 4630 info.reset(); 4631 conn.next.study(info); 4632 4633 info.minLength += minL; 4634 info.maxLength += maxL; 4635 info.maxValid &= maxV; 4636 info.deterministic = false; 4637 return false; 4638 } 4639 } 4640 4641 /** 4642 * The GroupHead saves the location where the group begins in the locals 4643 * and restores them when the match is done. 4644 * 4645 * The matchRef is used when a reference to this group is accessed later 4646 * in the expression. The locals will have a negative value in them to 4647 * indicate that we do not want to unset the group if the reference 4648 * doesn't match. 4649 */ 4650 static final class GroupHead extends Node { 4651 int localIndex; 4652 GroupHead(int localCount) { 4653 localIndex = localCount; 4654 } 4655 boolean match(Matcher matcher, int i, CharSequence seq) { 4656 int save = matcher.locals[localIndex]; 4657 matcher.locals[localIndex] = i; 4658 boolean ret = next.match(matcher, i, seq); 4659 matcher.locals[localIndex] = save; 4660 return ret; 4661 } 4662 boolean matchRef(Matcher matcher, int i, CharSequence seq) { 4663 int save = matcher.locals[localIndex]; 4664 matcher.locals[localIndex] = ~i; // HACK 4665 boolean ret = next.match(matcher, i, seq); 4666 matcher.locals[localIndex] = save; 4667 return ret; 4668 } 4669 } 4670 4671 /** 4672 * Recursive reference to a group in the regular expression. It calls 4673 * matchRef because if the reference fails to match we would not unset 4674 * the group. 4675 */ 4676 static final class GroupRef extends Node { 4677 GroupHead head; 4678 GroupRef(GroupHead head) { 4679 this.head = head; 4680 } 4681 boolean match(Matcher matcher, int i, CharSequence seq) { 4682 return head.matchRef(matcher, i, seq) 4683 && next.match(matcher, matcher.last, seq); 4684 } 4685 boolean study(TreeInfo info) { 4686 info.maxValid = false; 4687 info.deterministic = false; 4688 return next.study(info); 4689 } 4690 } 4691 4692 /** 4693 * The GroupTail handles the setting of group beginning and ending 4694 * locations when groups are successfully matched. It must also be able to 4695 * unset groups that have to be backed off of. 4696 * 4697 * The GroupTail node is also used when a previous group is referenced, 4698 * and in that case no group information needs to be set. 4699 */ 4700 static final class GroupTail extends Node { 4701 int localIndex; 4702 int groupIndex; 4703 GroupTail(int localCount, int groupCount) { 4704 localIndex = localCount; 4705 groupIndex = groupCount + groupCount; 4706 } 4707 boolean match(Matcher matcher, int i, CharSequence seq) { 4708 int tmp = matcher.locals[localIndex]; 4709 if (tmp >= 0) { // This is the normal group case. 4710 // Save the group so we can unset it if it 4711 // backs off of a match. 4712 int groupStart = matcher.groups[groupIndex]; 4713 int groupEnd = matcher.groups[groupIndex+1]; 4714 4715 matcher.groups[groupIndex] = tmp; 4716 matcher.groups[groupIndex+1] = i; 4717 if (next.match(matcher, i, seq)) { 4718 return true; 4719 } 4720 matcher.groups[groupIndex] = groupStart; 4721 matcher.groups[groupIndex+1] = groupEnd; 4722 return false; 4723 } else { 4724 // This is a group reference case. We don't need to save any 4725 // group info because it isn't really a group. 4726 matcher.last = i; 4727 return true; 4728 } 4729 } 4730 } 4731 4732 /** 4733 * This sets up a loop to handle a recursive quantifier structure. 4734 */ 4735 static final class Prolog extends Node { 4736 Loop loop; 4737 Prolog(Loop loop) { 4738 this.loop = loop; 4739 } 4740 boolean match(Matcher matcher, int i, CharSequence seq) { 4741 return loop.matchInit(matcher, i, seq); 4742 } 4743 boolean study(TreeInfo info) { 4744 return loop.study(info); 4745 } 4746 } 4747 4748 /** 4749 * Handles the repetition count for a greedy Curly. The matchInit 4750 * is called from the Prolog to save the index of where the group 4751 * beginning is stored. A zero length group check occurs in the 4752 * normal match but is skipped in the matchInit. 4753 */ 4754 static class Loop extends Node { 4755 Node body; 4756 int countIndex; // local count index in matcher locals 4757 int beginIndex; // group beginning index 4758 int cmin, cmax; 4759 Loop(int countIndex, int beginIndex) { 4760 this.countIndex = countIndex; 4761 this.beginIndex = beginIndex; 4762 } 4763 boolean match(Matcher matcher, int i, CharSequence seq) { 4764 // Avoid infinite loop in zero-length case. 4765 if (i > matcher.locals[beginIndex]) { 4766 int count = matcher.locals[countIndex]; 4767 4768 // This block is for before we reach the minimum 4769 // iterations required for the loop to match 4770 if (count < cmin) { 4771 matcher.locals[countIndex] = count + 1; 4772 boolean b = body.match(matcher, i, seq); 4773 // If match failed we must backtrack, so 4774 // the loop count should NOT be incremented 4775 if (!b) 4776 matcher.locals[countIndex] = count; 4777 // Return success or failure since we are under 4778 // minimum 4779 return b; 4780 } 4781 // This block is for after we have the minimum 4782 // iterations required for the loop to match 4783 if (count < cmax) { 4784 matcher.locals[countIndex] = count + 1; 4785 boolean b = body.match(matcher, i, seq); 4786 // If match failed we must backtrack, so 4787 // the loop count should NOT be incremented 4788 if (!b) 4789 matcher.locals[countIndex] = count; 4790 else 4791 return true; 4792 } 4793 } 4794 return next.match(matcher, i, seq); 4795 } 4796 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4797 int save = matcher.locals[countIndex]; 4798 boolean ret = false; 4799 if (0 < cmin) { 4800 matcher.locals[countIndex] = 1; 4801 ret = body.match(matcher, i, seq); 4802 } else if (0 < cmax) { 4803 matcher.locals[countIndex] = 1; 4804 ret = body.match(matcher, i, seq); 4805 if (ret == false) 4806 ret = next.match(matcher, i, seq); 4807 } else { 4808 ret = next.match(matcher, i, seq); 4809 } 4810 matcher.locals[countIndex] = save; 4811 return ret; 4812 } 4813 boolean study(TreeInfo info) { 4814 info.maxValid = false; 4815 info.deterministic = false; 4816 return false; 4817 } 4818 } 4819 4820 /** 4821 * Handles the repetition count for a reluctant Curly. The matchInit 4822 * is called from the Prolog to save the index of where the group 4823 * beginning is stored. A zero length group check occurs in the 4824 * normal match but is skipped in the matchInit. 4825 */ 4826 static final class LazyLoop extends Loop { 4827 LazyLoop(int countIndex, int beginIndex) { 4828 super(countIndex, beginIndex); 4829 } 4830 boolean match(Matcher matcher, int i, CharSequence seq) { 4831 // Check for zero length group 4832 if (i > matcher.locals[beginIndex]) { 4833 int count = matcher.locals[countIndex]; 4834 if (count < cmin) { 4835 matcher.locals[countIndex] = count + 1; 4836 boolean result = body.match(matcher, i, seq); 4837 // If match failed we must backtrack, so 4838 // the loop count should NOT be incremented 4839 if (!result) 4840 matcher.locals[countIndex] = count; 4841 return result; 4842 } 4843 if (next.match(matcher, i, seq)) 4844 return true; 4845 if (count < cmax) { 4846 matcher.locals[countIndex] = count + 1; 4847 boolean result = body.match(matcher, i, seq); 4848 // If match failed we must backtrack, so 4849 // the loop count should NOT be incremented 4850 if (!result) 4851 matcher.locals[countIndex] = count; 4852 return result; 4853 } 4854 return false; 4855 } 4856 return next.match(matcher, i, seq); 4857 } 4858 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4859 int save = matcher.locals[countIndex]; 4860 boolean ret = false; 4861 if (0 < cmin) { 4862 matcher.locals[countIndex] = 1; 4863 ret = body.match(matcher, i, seq); 4864 } else if (next.match(matcher, i, seq)) { 4865 ret = true; 4866 } else if (0 < cmax) { 4867 matcher.locals[countIndex] = 1; 4868 ret = body.match(matcher, i, seq); 4869 } 4870 matcher.locals[countIndex] = save; 4871 return ret; 4872 } 4873 boolean study(TreeInfo info) { 4874 info.maxValid = false; 4875 info.deterministic = false; 4876 return false; 4877 } 4878 } 4879 4880 /** 4881 * Refers to a group in the regular expression. Attempts to match 4882 * whatever the group referred to last matched. 4883 */ 4884 static class BackRef extends Node { 4885 int groupIndex; 4886 BackRef(int groupCount) { 4887 super(); 4888 groupIndex = groupCount + groupCount; 4889 } 4890 boolean match(Matcher matcher, int i, CharSequence seq) { 4891 int j = matcher.groups[groupIndex]; 4892 int k = matcher.groups[groupIndex+1]; 4893 4894 int groupSize = k - j; 4895 // If the referenced group didn't match, neither can this 4896 if (j < 0) 4897 return false; 4898 4899 // If there isn't enough input left no match 4900 if (i + groupSize > matcher.to) { 4901 matcher.hitEnd = true; 4902 return false; 4903 } 4904 // Check each new char to make sure it matches what the group 4905 // referenced matched last time around 4906 for (int index=0; index<groupSize; index++) 4907 if (seq.charAt(i+index) != seq.charAt(j+index)) 4908 return false; 4909 4910 return next.match(matcher, i+groupSize, seq); 4911 } 4912 boolean study(TreeInfo info) { 4913 info.maxValid = false; 4914 return next.study(info); 4915 } 4916 } 4917 4918 static class CIBackRef extends Node { 4919 int groupIndex; 4920 boolean doUnicodeCase; 4921 CIBackRef(int groupCount, boolean doUnicodeCase) { 4922 super(); 4923 groupIndex = groupCount + groupCount; 4924 this.doUnicodeCase = doUnicodeCase; 4925 } 4926 boolean match(Matcher matcher, int i, CharSequence seq) { 4927 int j = matcher.groups[groupIndex]; 4928 int k = matcher.groups[groupIndex+1]; 4929 4930 int groupSize = k - j; 4931 4932 // If the referenced group didn't match, neither can this 4933 if (j < 0) 4934 return false; 4935 4936 // If there isn't enough input left no match 4937 if (i + groupSize > matcher.to) { 4938 matcher.hitEnd = true; 4939 return false; 4940 } 4941 4942 // Check each new char to make sure it matches what the group 4943 // referenced matched last time around 4944 int x = i; 4945 for (int index=0; index<groupSize; index++) { 4946 int c1 = Character.codePointAt(seq, x); 4947 int c2 = Character.codePointAt(seq, j); 4948 if (c1 != c2) { 4949 if (doUnicodeCase) { 4950 int cc1 = Character.toUpperCase(c1); 4951 int cc2 = Character.toUpperCase(c2); 4952 if (cc1 != cc2 && 4953 Character.toLowerCase(cc1) != 4954 Character.toLowerCase(cc2)) 4955 return false; 4956 } else { 4957 if (ASCII.toLower(c1) != ASCII.toLower(c2)) 4958 return false; 4959 } 4960 } 4961 x += Character.charCount(c1); 4962 j += Character.charCount(c2); 4963 } 4964 4965 return next.match(matcher, i+groupSize, seq); 4966 } 4967 boolean study(TreeInfo info) { 4968 info.maxValid = false; 4969 return next.study(info); 4970 } 4971 } 4972 4973 /** 4974 * Searches until the next instance of its atom. This is useful for 4975 * finding the atom efficiently without passing an instance of it 4976 * (greedy problem) and without a lot of wasted search time (reluctant 4977 * problem). 4978 */ 4979 static final class First extends Node { 4980 Node atom; 4981 First(Node node) { 4982 this.atom = BnM.optimize(node); 4983 } 4984 boolean match(Matcher matcher, int i, CharSequence seq) { 4985 if (atom instanceof BnM) { 4986 return atom.match(matcher, i, seq) 4987 && next.match(matcher, matcher.last, seq); 4988 } 4989 for (;;) { 4990 if (i > matcher.to) { 4991 matcher.hitEnd = true; 4992 return false; 4993 } 4994 if (atom.match(matcher, i, seq)) { 4995 return next.match(matcher, matcher.last, seq); 4996 } 4997 i += countChars(seq, i, 1); 4998 matcher.first++; 4999 } 5000 } 5001 boolean study(TreeInfo info) { 5002 atom.study(info); 5003 info.maxValid = false; 5004 info.deterministic = false; 5005 return next.study(info); 5006 } 5007 } 5008 5009 static final class Conditional extends Node { 5010 Node cond, yes, not; 5011 Conditional(Node cond, Node yes, Node not) { 5012 this.cond = cond; 5013 this.yes = yes; 5014 this.not = not; 5015 } 5016 boolean match(Matcher matcher, int i, CharSequence seq) { 5017 if (cond.match(matcher, i, seq)) { 5018 return yes.match(matcher, i, seq); 5019 } else { 5020 return not.match(matcher, i, seq); 5021 } 5022 } 5023 boolean study(TreeInfo info) { 5024 int minL = info.minLength; 5025 int maxL = info.maxLength; 5026 boolean maxV = info.maxValid; 5027 info.reset(); 5028 yes.study(info); 5029 5030 int minL2 = info.minLength; 5031 int maxL2 = info.maxLength; 5032 boolean maxV2 = info.maxValid; 5033 info.reset(); 5034 not.study(info); 5035 5036 info.minLength = minL + Math.min(minL2, info.minLength); 5037 info.maxLength = maxL + Math.max(maxL2, info.maxLength); 5038 info.maxValid = (maxV & maxV2 & info.maxValid); 5039 info.deterministic = false; 5040 return next.study(info); 5041 } 5042 } 5043 5044 /** 5045 * Zero width positive lookahead. 5046 */ 5047 static final class Pos extends Node { 5048 Node cond; 5049 Pos(Node cond) { 5050 this.cond = cond; 5051 } 5052 boolean match(Matcher matcher, int i, CharSequence seq) { 5053 int savedTo = matcher.to; 5054 boolean conditionMatched = false; 5055 5056 // Relax transparent region boundaries for lookahead 5057 if (matcher.transparentBounds) 5058 matcher.to = matcher.getTextLength(); 5059 try { 5060 conditionMatched = cond.match(matcher, i, seq); 5061 } finally { 5062 // Reinstate region boundaries 5063 matcher.to = savedTo; 5064 } 5065 return conditionMatched && next.match(matcher, i, seq); 5066 } 5067 } 5068 5069 /** 5070 * Zero width negative lookahead. 5071 */ 5072 static final class Neg extends Node { 5073 Node cond; 5074 Neg(Node cond) { 5075 this.cond = cond; 5076 } 5077 boolean match(Matcher matcher, int i, CharSequence seq) { 5078 int savedTo = matcher.to; 5079 boolean conditionMatched = false; 5080 5081 // Relax transparent region boundaries for lookahead 5082 if (matcher.transparentBounds) 5083 matcher.to = matcher.getTextLength(); 5084 try { 5085 if (i < matcher.to) { 5086 conditionMatched = !cond.match(matcher, i, seq); 5087 } else { 5088 // If a negative lookahead succeeds then more input 5089 // could cause it to fail! 5090 matcher.requireEnd = true; 5091 conditionMatched = !cond.match(matcher, i, seq); 5092 } 5093 } finally { 5094 // Reinstate region boundaries 5095 matcher.to = savedTo; 5096 } 5097 return conditionMatched && next.match(matcher, i, seq); 5098 } 5099 } 5100 5101 /** 5102 * For use with lookbehinds; matches the position where the lookbehind 5103 * was encountered. 5104 */ 5105 static Node lookbehindEnd = new Node() { 5106 boolean match(Matcher matcher, int i, CharSequence seq) { 5107 return i == matcher.lookbehindTo; 5108 } 5109 }; 5110 5111 /** 5112 * Zero width positive lookbehind. 5113 */ 5114 static class Behind extends Node { 5115 Node cond; 5116 int rmax, rmin; 5117 Behind(Node cond, int rmax, int rmin) { 5118 this.cond = cond; 5119 this.rmax = rmax; 5120 this.rmin = rmin; 5121 } 5122 5123 boolean match(Matcher matcher, int i, CharSequence seq) { 5124 int savedFrom = matcher.from; 5125 boolean conditionMatched = false; 5126 int startIndex = (!matcher.transparentBounds) ? 5127 matcher.from : 0; 5128 int from = Math.max(i - rmax, startIndex); 5129 // Set end boundary 5130 int savedLBT = matcher.lookbehindTo; 5131 matcher.lookbehindTo = i; 5132 // Relax transparent region boundaries for lookbehind 5133 if (matcher.transparentBounds) 5134 matcher.from = 0; 5135 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 5136 conditionMatched = cond.match(matcher, j, seq); 5137 } 5138 matcher.from = savedFrom; 5139 matcher.lookbehindTo = savedLBT; 5140 return conditionMatched && next.match(matcher, i, seq); 5141 } 5142 } 5143 5144 /** 5145 * Zero width positive lookbehind, including supplementary 5146 * characters or unpaired surrogates. 5147 */ 5148 static final class BehindS extends Behind { 5149 BehindS(Node cond, int rmax, int rmin) { 5150 super(cond, rmax, rmin); 5151 } 5152 boolean match(Matcher matcher, int i, CharSequence seq) { 5153 int rmaxChars = countChars(seq, i, -rmax); 5154 int rminChars = countChars(seq, i, -rmin); 5155 int savedFrom = matcher.from; 5156 int startIndex = (!matcher.transparentBounds) ? 5157 matcher.from : 0; 5158 boolean conditionMatched = false; 5159 int from = Math.max(i - rmaxChars, startIndex); 5160 // Set end boundary 5161 int savedLBT = matcher.lookbehindTo; 5162 matcher.lookbehindTo = i; 5163 // Relax transparent region boundaries for lookbehind 5164 if (matcher.transparentBounds) 5165 matcher.from = 0; 5166 5167 for (int j = i - rminChars; 5168 !conditionMatched && j >= from; 5169 j -= j>from ? countChars(seq, j, -1) : 1) { 5170 conditionMatched = cond.match(matcher, j, seq); 5171 } 5172 matcher.from = savedFrom; 5173 matcher.lookbehindTo = savedLBT; 5174 return conditionMatched && next.match(matcher, i, seq); 5175 } 5176 } 5177 5178 /** 5179 * Zero width negative lookbehind. 5180 */ 5181 static class NotBehind extends Node { 5182 Node cond; 5183 int rmax, rmin; 5184 NotBehind(Node cond, int rmax, int rmin) { 5185 this.cond = cond; 5186 this.rmax = rmax; 5187 this.rmin = rmin; 5188 } 5189 5190 boolean match(Matcher matcher, int i, CharSequence seq) { 5191 int savedLBT = matcher.lookbehindTo; 5192 int savedFrom = matcher.from; 5193 boolean conditionMatched = false; 5194 int startIndex = (!matcher.transparentBounds) ? 5195 matcher.from : 0; 5196 int from = Math.max(i - rmax, startIndex); 5197 matcher.lookbehindTo = i; 5198 // Relax transparent region boundaries for lookbehind 5199 if (matcher.transparentBounds) 5200 matcher.from = 0; 5201 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 5202 conditionMatched = cond.match(matcher, j, seq); 5203 } 5204 // Reinstate region boundaries 5205 matcher.from = savedFrom; 5206 matcher.lookbehindTo = savedLBT; 5207 return !conditionMatched && next.match(matcher, i, seq); 5208 } 5209 } 5210 5211 /** 5212 * Zero width negative lookbehind, including supplementary 5213 * characters or unpaired surrogates. 5214 */ 5215 static final class NotBehindS extends NotBehind { 5216 NotBehindS(Node cond, int rmax, int rmin) { 5217 super(cond, rmax, rmin); 5218 } 5219 boolean match(Matcher matcher, int i, CharSequence seq) { 5220 int rmaxChars = countChars(seq, i, -rmax); 5221 int rminChars = countChars(seq, i, -rmin); 5222 int savedFrom = matcher.from; 5223 int savedLBT = matcher.lookbehindTo; 5224 boolean conditionMatched = false; 5225 int startIndex = (!matcher.transparentBounds) ? 5226 matcher.from : 0; 5227 int from = Math.max(i - rmaxChars, startIndex); 5228 matcher.lookbehindTo = i; 5229 // Relax transparent region boundaries for lookbehind 5230 if (matcher.transparentBounds) 5231 matcher.from = 0; 5232 for (int j = i - rminChars; 5233 !conditionMatched && j >= from; 5234 j -= j>from ? countChars(seq, j, -1) : 1) { 5235 conditionMatched = cond.match(matcher, j, seq); 5236 } 5237 //Reinstate region boundaries 5238 matcher.from = savedFrom; 5239 matcher.lookbehindTo = savedLBT; 5240 return !conditionMatched && next.match(matcher, i, seq); 5241 } 5242 } 5243 5244 /** 5245 * Returns the set union of two CharProperty nodes. 5246 */ 5247 private static CharProperty union(final CharProperty lhs, 5248 final CharProperty rhs) { 5249 return new CharProperty() { 5250 boolean isSatisfiedBy(int ch) { 5251 return lhs.isSatisfiedBy(ch) || rhs.isSatisfiedBy(ch);}}; 5252 } 5253 5254 /** 5255 * Returns the set intersection of two CharProperty nodes. 5256 */ 5257 private static CharProperty intersection(final CharProperty lhs, 5258 final CharProperty rhs) { 5259 return new CharProperty() { 5260 boolean isSatisfiedBy(int ch) { 5261 return lhs.isSatisfiedBy(ch) && rhs.isSatisfiedBy(ch);}}; 5262 } 5263 5264 /** 5265 * Returns the set difference of two CharProperty nodes. 5266 */ 5267 private static CharProperty setDifference(final CharProperty lhs, 5268 final CharProperty rhs) { 5269 return new CharProperty() { 5270 boolean isSatisfiedBy(int ch) { 5271 return ! rhs.isSatisfiedBy(ch) && lhs.isSatisfiedBy(ch);}}; 5272 } 5273 5274 /** 5275 * Handles word boundaries. Includes a field to allow this one class to 5276 * deal with the different types of word boundaries we can match. The word 5277 * characters include underscores, letters, and digits. Non spacing marks 5278 * can are also part of a word if they have a base character, otherwise 5279 * they are ignored for purposes of finding word boundaries. 5280 */ 5281 static final class Bound extends Node { 5282 static int LEFT = 0x1; 5283 static int RIGHT= 0x2; 5284 static int BOTH = 0x3; 5285 static int NONE = 0x4; 5286 int type; 5287 boolean useUWORD; 5288 Bound(int n, boolean useUWORD) { 5289 type = n; 5290 this.useUWORD = useUWORD; 5291 } 5292 5293 boolean isWord(int ch) { 5294 return useUWORD ? UnicodeProp.WORD.is(ch) 5295 : (ch == '_' || Character.isLetterOrDigit(ch)); 5296 } 5297 5298 int check(Matcher matcher, int i, CharSequence seq) { 5299 int ch; 5300 boolean left = false; 5301 int startIndex = matcher.from; 5302 int endIndex = matcher.to; 5303 if (matcher.transparentBounds) { 5304 startIndex = 0; 5305 endIndex = matcher.getTextLength(); 5306 } 5307 if (i > startIndex) { 5308 ch = Character.codePointBefore(seq, i); 5309 left = (isWord(ch) || 5310 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5311 && hasBaseCharacter(matcher, i-1, seq))); 5312 } 5313 boolean right = false; 5314 if (i < endIndex) { 5315 ch = Character.codePointAt(seq, i); 5316 right = (isWord(ch) || 5317 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5318 && hasBaseCharacter(matcher, i, seq))); 5319 } else { 5320 // Tried to access char past the end 5321 matcher.hitEnd = true; 5322 // The addition of another char could wreck a boundary 5323 matcher.requireEnd = true; 5324 } 5325 return ((left ^ right) ? (right ? LEFT : RIGHT) : NONE); 5326 } 5327 boolean match(Matcher matcher, int i, CharSequence seq) { 5328 return (check(matcher, i, seq) & type) > 0 5329 && next.match(matcher, i, seq); 5330 } 5331 } 5332 5333 /** 5334 * Non spacing marks only count as word characters in bounds calculations 5335 * if they have a base character. 5336 */ 5337 private static boolean hasBaseCharacter(Matcher matcher, int i, 5338 CharSequence seq) 5339 { 5340 int start = (!matcher.transparentBounds) ? 5341 matcher.from : 0; 5342 for (int x=i; x >= start; x--) { 5343 int ch = Character.codePointAt(seq, x); 5344 if (Character.isLetterOrDigit(ch)) 5345 return true; 5346 if (Character.getType(ch) == Character.NON_SPACING_MARK) 5347 continue; 5348 return false; 5349 } 5350 return false; 5351 } 5352 5353 /** 5354 * Attempts to match a slice in the input using the Boyer-Moore string 5355 * matching algorithm. The algorithm is based on the idea that the 5356 * pattern can be shifted farther ahead in the search text if it is 5357 * matched right to left. 5358 * <p> 5359 * The pattern is compared to the input one character at a time, from 5360 * the rightmost character in the pattern to the left. If the characters 5361 * all match the pattern has been found. If a character does not match, 5362 * the pattern is shifted right a distance that is the maximum of two 5363 * functions, the bad character shift and the good suffix shift. This 5364 * shift moves the attempted match position through the input more 5365 * quickly than a naive one position at a time check. 5366 * <p> 5367 * The bad character shift is based on the character from the text that 5368 * did not match. If the character does not appear in the pattern, the 5369 * pattern can be shifted completely beyond the bad character. If the 5370 * character does occur in the pattern, the pattern can be shifted to 5371 * line the pattern up with the next occurrence of that character. 5372 * <p> 5373 * The good suffix shift is based on the idea that some subset on the right 5374 * side of the pattern has matched. When a bad character is found, the 5375 * pattern can be shifted right by the pattern length if the subset does 5376 * not occur again in pattern, or by the amount of distance to the 5377 * next occurrence of the subset in the pattern. 5378 * 5379 * Boyer-Moore search methods adapted from code by Amy Yu. 5380 */ 5381 static class BnM extends Node { 5382 int[] buffer; 5383 int[] lastOcc; 5384 int[] optoSft; 5385 5386 /** 5387 * Pre calculates arrays needed to generate the bad character 5388 * shift and the good suffix shift. Only the last seven bits 5389 * are used to see if chars match; This keeps the tables small 5390 * and covers the heavily used ASCII range, but occasionally 5391 * results in an aliased match for the bad character shift. 5392 */ 5393 static Node optimize(Node node) { 5394 if (!(node instanceof Slice)) { 5395 return node; 5396 } 5397 5398 int[] src = ((Slice) node).buffer; 5399 int patternLength = src.length; 5400 // The BM algorithm requires a bit of overhead; 5401 // If the pattern is short don't use it, since 5402 // a shift larger than the pattern length cannot 5403 // be used anyway. 5404 if (patternLength < 4) { 5405 return node; 5406 } 5407 int i, j, k; 5408 int[] lastOcc = new int[128]; 5409 int[] optoSft = new int[patternLength]; 5410 // Precalculate part of the bad character shift 5411 // It is a table for where in the pattern each 5412 // lower 7-bit value occurs 5413 for (i = 0; i < patternLength; i++) { 5414 lastOcc[src[i]&0x7F] = i + 1; 5415 } 5416 // Precalculate the good suffix shift 5417 // i is the shift amount being considered 5418 NEXT: for (i = patternLength; i > 0; i--) { 5419 // j is the beginning index of suffix being considered 5420 for (j = patternLength - 1; j >= i; j--) { 5421 // Testing for good suffix 5422 if (src[j] == src[j-i]) { 5423 // src[j..len] is a good suffix 5424 optoSft[j-1] = i; 5425 } else { 5426 // No match. The array has already been 5427 // filled up with correct values before. 5428 continue NEXT; 5429 } 5430 } 5431 // This fills up the remaining of optoSft 5432 // any suffix can not have larger shift amount 5433 // then its sub-suffix. Why??? 5434 while (j > 0) { 5435 optoSft[--j] = i; 5436 } 5437 } 5438 // Set the guard value because of unicode compression 5439 optoSft[patternLength-1] = 1; 5440 if (node instanceof SliceS) 5441 return new BnMS(src, lastOcc, optoSft, node.next); 5442 return new BnM(src, lastOcc, optoSft, node.next); 5443 } 5444 BnM(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5445 this.buffer = src; 5446 this.lastOcc = lastOcc; 5447 this.optoSft = optoSft; 5448 this.next = next; 5449 } 5450 boolean match(Matcher matcher, int i, CharSequence seq) { 5451 int[] src = buffer; 5452 int patternLength = src.length; 5453 int last = matcher.to - patternLength; 5454 5455 // Loop over all possible match positions in text 5456 NEXT: while (i <= last) { 5457 // Loop over pattern from right to left 5458 for (int j = patternLength - 1; j >= 0; j--) { 5459 int ch = seq.charAt(i+j); 5460 if (ch != src[j]) { 5461 // Shift search to the right by the maximum of the 5462 // bad character shift and the good suffix shift 5463 i += Math.max(j + 1 - lastOcc[ch&0x7F], optoSft[j]); 5464 continue NEXT; 5465 } 5466 } 5467 // Entire pattern matched starting at i 5468 matcher.first = i; 5469 boolean ret = next.match(matcher, i + patternLength, seq); 5470 if (ret) { 5471 matcher.first = i; 5472 matcher.groups[0] = matcher.first; 5473 matcher.groups[1] = matcher.last; 5474 return true; 5475 } 5476 i++; 5477 } 5478 // BnM is only used as the leading node in the unanchored case, 5479 // and it replaced its Start() which always searches to the end 5480 // if it doesn't find what it's looking for, so hitEnd is true. 5481 matcher.hitEnd = true; 5482 return false; 5483 } 5484 boolean study(TreeInfo info) { 5485 info.minLength += buffer.length; 5486 info.maxValid = false; 5487 return next.study(info); 5488 } 5489 } 5490 5491 /** 5492 * Supplementary support version of BnM(). Unpaired surrogates are 5493 * also handled by this class. 5494 */ 5495 static final class BnMS extends BnM { 5496 int lengthInChars; 5497 5498 BnMS(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5499 super(src, lastOcc, optoSft, next); 5500 for (int x = 0; x < buffer.length; x++) { 5501 lengthInChars += Character.charCount(buffer[x]); 5502 } 5503 } 5504 boolean match(Matcher matcher, int i, CharSequence seq) { 5505 int[] src = buffer; 5506 int patternLength = src.length; 5507 int last = matcher.to - lengthInChars; 5508 5509 // Loop over all possible match positions in text 5510 NEXT: while (i <= last) { 5511 // Loop over pattern from right to left 5512 int ch; 5513 for (int j = countChars(seq, i, patternLength), x = patternLength - 1; 5514 j > 0; j -= Character.charCount(ch), x--) { 5515 ch = Character.codePointBefore(seq, i+j); 5516 if (ch != src[x]) { 5517 // Shift search to the right by the maximum of the 5518 // bad character shift and the good suffix shift 5519 int n = Math.max(x + 1 - lastOcc[ch&0x7F], optoSft[x]); 5520 i += countChars(seq, i, n); 5521 continue NEXT; 5522 } 5523 } 5524 // Entire pattern matched starting at i 5525 matcher.first = i; 5526 boolean ret = next.match(matcher, i + lengthInChars, seq); 5527 if (ret) { 5528 matcher.first = i; 5529 matcher.groups[0] = matcher.first; 5530 matcher.groups[1] = matcher.last; 5531 return true; 5532 } 5533 i += countChars(seq, i, 1); 5534 } 5535 matcher.hitEnd = true; 5536 return false; 5537 } 5538 } 5539 5540 /////////////////////////////////////////////////////////////////////////////// 5541 /////////////////////////////////////////////////////////////////////////////// 5542 5543 /** 5544 * This must be the very first initializer. 5545 */ 5546 static Node accept = new Node(); 5547 5548 static Node lastAccept = new LastNode(); 5549 5550 private static class CharPropertyNames { 5551 5552 static CharProperty charPropertyFor(String name) { 5553 CharPropertyFactory m = map.get(name); 5554 return m == null ? null : m.make(); 5555 } 5556 5557 private static abstract class CharPropertyFactory { 5558 abstract CharProperty make(); 5559 } 5560 5561 private static void defCategory(String name, 5562 final int typeMask) { 5563 map.put(name, new CharPropertyFactory() { 5564 CharProperty make() { return new Category(typeMask);}}); 5565 } 5566 5567 private static void defRange(String name, 5568 final int lower, final int upper) { 5569 map.put(name, new CharPropertyFactory() { 5570 CharProperty make() { return rangeFor(lower, upper);}}); 5571 } 5572 5573 private static void defCtype(String name, 5574 final int ctype) { 5575 map.put(name, new CharPropertyFactory() { 5576 CharProperty make() { return new Ctype(ctype);}}); 5577 } 5578 5579 private static abstract class CloneableProperty 5580 extends CharProperty implements Cloneable 5581 { 5582 public CloneableProperty clone() { 5583 try { 5584 return (CloneableProperty) super.clone(); 5585 } catch (CloneNotSupportedException e) { 5586 throw new AssertionError(e); 5587 } 5588 } 5589 } 5590 5591 private static void defClone(String name, 5592 final CloneableProperty p) { 5593 map.put(name, new CharPropertyFactory() { 5594 CharProperty make() { return p.clone();}}); 5595 } 5596 5597 private static final HashMap<String, CharPropertyFactory> map 5598 = new HashMap<>(); 5599 5600 static { 5601 // Unicode character property aliases, defined in 5602 // http://www.unicode.org/Public/UNIDATA/PropertyValueAliases.txt 5603 defCategory("Cn", 1<<Character.UNASSIGNED); 5604 defCategory("Lu", 1<<Character.UPPERCASE_LETTER); 5605 defCategory("Ll", 1<<Character.LOWERCASE_LETTER); 5606 defCategory("Lt", 1<<Character.TITLECASE_LETTER); 5607 defCategory("Lm", 1<<Character.MODIFIER_LETTER); 5608 defCategory("Lo", 1<<Character.OTHER_LETTER); 5609 defCategory("Mn", 1<<Character.NON_SPACING_MARK); 5610 defCategory("Me", 1<<Character.ENCLOSING_MARK); 5611 defCategory("Mc", 1<<Character.COMBINING_SPACING_MARK); 5612 defCategory("Nd", 1<<Character.DECIMAL_DIGIT_NUMBER); 5613 defCategory("Nl", 1<<Character.LETTER_NUMBER); 5614 defCategory("No", 1<<Character.OTHER_NUMBER); 5615 defCategory("Zs", 1<<Character.SPACE_SEPARATOR); 5616 defCategory("Zl", 1<<Character.LINE_SEPARATOR); 5617 defCategory("Zp", 1<<Character.PARAGRAPH_SEPARATOR); 5618 defCategory("Cc", 1<<Character.CONTROL); 5619 defCategory("Cf", 1<<Character.FORMAT); 5620 defCategory("Co", 1<<Character.PRIVATE_USE); 5621 defCategory("Cs", 1<<Character.SURROGATE); 5622 defCategory("Pd", 1<<Character.DASH_PUNCTUATION); 5623 defCategory("Ps", 1<<Character.START_PUNCTUATION); 5624 defCategory("Pe", 1<<Character.END_PUNCTUATION); 5625 defCategory("Pc", 1<<Character.CONNECTOR_PUNCTUATION); 5626 defCategory("Po", 1<<Character.OTHER_PUNCTUATION); 5627 defCategory("Sm", 1<<Character.MATH_SYMBOL); 5628 defCategory("Sc", 1<<Character.CURRENCY_SYMBOL); 5629 defCategory("Sk", 1<<Character.MODIFIER_SYMBOL); 5630 defCategory("So", 1<<Character.OTHER_SYMBOL); 5631 defCategory("Pi", 1<<Character.INITIAL_QUOTE_PUNCTUATION); 5632 defCategory("Pf", 1<<Character.FINAL_QUOTE_PUNCTUATION); 5633 defCategory("L", ((1<<Character.UPPERCASE_LETTER) | 5634 (1<<Character.LOWERCASE_LETTER) | 5635 (1<<Character.TITLECASE_LETTER) | 5636 (1<<Character.MODIFIER_LETTER) | 5637 (1<<Character.OTHER_LETTER))); 5638 defCategory("M", ((1<<Character.NON_SPACING_MARK) | 5639 (1<<Character.ENCLOSING_MARK) | 5640 (1<<Character.COMBINING_SPACING_MARK))); 5641 defCategory("N", ((1<<Character.DECIMAL_DIGIT_NUMBER) | 5642 (1<<Character.LETTER_NUMBER) | 5643 (1<<Character.OTHER_NUMBER))); 5644 defCategory("Z", ((1<<Character.SPACE_SEPARATOR) | 5645 (1<<Character.LINE_SEPARATOR) | 5646 (1<<Character.PARAGRAPH_SEPARATOR))); 5647 defCategory("C", ((1<<Character.CONTROL) | 5648 (1<<Character.FORMAT) | 5649 (1<<Character.PRIVATE_USE) | 5650 (1<<Character.SURROGATE))); // Other 5651 defCategory("P", ((1<<Character.DASH_PUNCTUATION) | 5652 (1<<Character.START_PUNCTUATION) | 5653 (1<<Character.END_PUNCTUATION) | 5654 (1<<Character.CONNECTOR_PUNCTUATION) | 5655 (1<<Character.OTHER_PUNCTUATION) | 5656 (1<<Character.INITIAL_QUOTE_PUNCTUATION) | 5657 (1<<Character.FINAL_QUOTE_PUNCTUATION))); 5658 defCategory("S", ((1<<Character.MATH_SYMBOL) | 5659 (1<<Character.CURRENCY_SYMBOL) | 5660 (1<<Character.MODIFIER_SYMBOL) | 5661 (1<<Character.OTHER_SYMBOL))); 5662 defCategory("LC", ((1<<Character.UPPERCASE_LETTER) | 5663 (1<<Character.LOWERCASE_LETTER) | 5664 (1<<Character.TITLECASE_LETTER))); 5665 defCategory("LD", ((1<<Character.UPPERCASE_LETTER) | 5666 (1<<Character.LOWERCASE_LETTER) | 5667 (1<<Character.TITLECASE_LETTER) | 5668 (1<<Character.MODIFIER_LETTER) | 5669 (1<<Character.OTHER_LETTER) | 5670 (1<<Character.DECIMAL_DIGIT_NUMBER))); 5671 defRange("L1", 0x00, 0xFF); // Latin-1 5672 map.put("all", new CharPropertyFactory() { 5673 CharProperty make() { return new All(); }}); 5674 5675 // Posix regular expression character classes, defined in 5676 // http://www.unix.org/onlinepubs/009695399/basedefs/xbd_chap09.html 5677 defRange("ASCII", 0x00, 0x7F); // ASCII 5678 defCtype("Alnum", ASCII.ALNUM); // Alphanumeric characters 5679 defCtype("Alpha", ASCII.ALPHA); // Alphabetic characters 5680 defCtype("Blank", ASCII.BLANK); // Space and tab characters 5681 defCtype("Cntrl", ASCII.CNTRL); // Control characters 5682 defRange("Digit", '0', '9'); // Numeric characters 5683 defCtype("Graph", ASCII.GRAPH); // printable and visible 5684 defRange("Lower", 'a', 'z'); // Lower-case alphabetic 5685 defRange("Print", 0x20, 0x7E); // Printable characters 5686 defCtype("Punct", ASCII.PUNCT); // Punctuation characters 5687 defCtype("Space", ASCII.SPACE); // Space characters 5688 defRange("Upper", 'A', 'Z'); // Upper-case alphabetic 5689 defCtype("XDigit",ASCII.XDIGIT); // hexadecimal digits 5690 5691 // Java character properties, defined by methods in Character.java 5692 defClone("javaLowerCase", new CloneableProperty() { 5693 boolean isSatisfiedBy(int ch) { 5694 return Character.isLowerCase(ch);}}); 5695 defClone("javaUpperCase", new CloneableProperty() { 5696 boolean isSatisfiedBy(int ch) { 5697 return Character.isUpperCase(ch);}}); 5698 defClone("javaAlphabetic", new CloneableProperty() { 5699 boolean isSatisfiedBy(int ch) { 5700 return Character.isAlphabetic(ch);}}); 5701 defClone("javaIdeographic", new CloneableProperty() { 5702 boolean isSatisfiedBy(int ch) { 5703 return Character.isIdeographic(ch);}}); 5704 defClone("javaTitleCase", new CloneableProperty() { 5705 boolean isSatisfiedBy(int ch) { 5706 return Character.isTitleCase(ch);}}); 5707 defClone("javaDigit", new CloneableProperty() { 5708 boolean isSatisfiedBy(int ch) { 5709 return Character.isDigit(ch);}}); 5710 defClone("javaDefined", new CloneableProperty() { 5711 boolean isSatisfiedBy(int ch) { 5712 return Character.isDefined(ch);}}); 5713 defClone("javaLetter", new CloneableProperty() { 5714 boolean isSatisfiedBy(int ch) { 5715 return Character.isLetter(ch);}}); 5716 defClone("javaLetterOrDigit", new CloneableProperty() { 5717 boolean isSatisfiedBy(int ch) { 5718 return Character.isLetterOrDigit(ch);}}); 5719 defClone("javaJavaIdentifierStart", new CloneableProperty() { 5720 boolean isSatisfiedBy(int ch) { 5721 return Character.isJavaIdentifierStart(ch);}}); 5722 defClone("javaJavaIdentifierPart", new CloneableProperty() { 5723 boolean isSatisfiedBy(int ch) { 5724 return Character.isJavaIdentifierPart(ch);}}); 5725 defClone("javaUnicodeIdentifierStart", new CloneableProperty() { 5726 boolean isSatisfiedBy(int ch) { 5727 return Character.isUnicodeIdentifierStart(ch);}}); 5728 defClone("javaUnicodeIdentifierPart", new CloneableProperty() { 5729 boolean isSatisfiedBy(int ch) { 5730 return Character.isUnicodeIdentifierPart(ch);}}); 5731 defClone("javaIdentifierIgnorable", new CloneableProperty() { 5732 boolean isSatisfiedBy(int ch) { 5733 return Character.isIdentifierIgnorable(ch);}}); 5734 defClone("javaSpaceChar", new CloneableProperty() { 5735 boolean isSatisfiedBy(int ch) { 5736 return Character.isSpaceChar(ch);}}); 5737 defClone("javaWhitespace", new CloneableProperty() { 5738 boolean isSatisfiedBy(int ch) { 5739 return Character.isWhitespace(ch);}}); 5740 defClone("javaISOControl", new CloneableProperty() { 5741 boolean isSatisfiedBy(int ch) { 5742 return Character.isISOControl(ch);}}); 5743 defClone("javaMirrored", new CloneableProperty() { 5744 boolean isSatisfiedBy(int ch) { 5745 return Character.isMirrored(ch);}}); 5746 } 5747 } 5748 5749 /** 5750 * Creates a predicate which can be used to match a string. 5751 * 5752 * @return The predicate which can be used for matching on a string 5753 * @since 1.8 5754 */ 5755 public Predicate<String> asPredicate() { 5756 return s -> matcher(s).find(); 5757 } 5758 5759 /** 5760 * Creates a stream from the given input sequence around matches of this 5761 * pattern. 5762 * 5763 * <p> The stream returned by this method contains each substring of the 5764 * input sequence that is terminated by another subsequence that matches 5765 * this pattern or is terminated by the end of the input sequence. The 5766 * substrings in the stream are in the order in which they occur in the 5767 * input. Trailing empty strings will be discarded and not encountered in 5768 * the stream. 5769 * 5770 * <p> If this pattern does not match any subsequence of the input then 5771 * the resulting stream has just one element, namely the input sequence in 5772 * string form. 5773 * 5774 * <p> When there is a positive-width match at the beginning of the input 5775 * sequence then an empty leading substring is included at the beginning 5776 * of the stream. A zero-width match at the beginning however never produces 5777 * such empty leading substring. 5778 * 5779 * <p> If the input sequence is mutable, it must remain constant during the 5780 * execution of the terminal stream operation. Otherwise, the result of the 5781 * terminal stream operation is undefined. 5782 * 5783 * @param input 5784 * The character sequence to be split 5785 * 5786 * @return The stream of strings computed by splitting the input 5787 * around matches of this pattern 5788 * @see #split(CharSequence) 5789 * @since 1.8 5790 */ 5791 public Stream<String> splitAsStream(final CharSequence input) { 5792 class MatcherIterator implements Iterator<String> { 5793 private final Matcher matcher; 5794 // The start position of the next sub-sequence of input 5795 // when current == input.length there are no more elements 5796 private int current; 5797 // null if the next element, if any, needs to obtained 5798 private String nextElement; 5799 // > 0 if there are N next empty elements 5800 private int emptyElementCount; 5801 5802 MatcherIterator() { 5803 this.matcher = matcher(input); 5804 } 5805 5806 public String next() { 5807 if (!hasNext()) 5808 throw new NoSuchElementException(); 5809 5810 if (emptyElementCount == 0) { 5811 String n = nextElement; 5812 nextElement = null; 5813 return n; 5814 } else { 5815 emptyElementCount--; 5816 return ""; 5817 } 5818 } 5819 5820 public boolean hasNext() { 5821 if (nextElement != null || emptyElementCount > 0) 5822 return true; 5823 5824 if (current == input.length()) 5825 return false; 5826 5827 // Consume the next matching element 5828 // Count sequence of matching empty elements 5829 while (matcher.find()) { 5830 nextElement = input.subSequence(current, matcher.start()).toString(); 5831 current = matcher.end(); 5832 if (!nextElement.isEmpty()) { 5833 return true; 5834 } else if (current > 0) { // no empty leading substring for zero-width 5835 // match at the beginning of the input 5836 emptyElementCount++; 5837 } 5838 } 5839 5840 // Consume last matching element 5841 nextElement = input.subSequence(current, input.length()).toString(); 5842 current = input.length(); 5843 if (!nextElement.isEmpty()) { 5844 return true; 5845 } else { 5846 // Ignore a terminal sequence of matching empty elements 5847 emptyElementCount = 0; 5848 nextElement = null; 5849 return false; 5850 } 5851 } 5852 } 5853 return StreamSupport.stream(Spliterators.spliteratorUnknownSize( 5854 new MatcherIterator(), Spliterator.ORDERED | Spliterator.NONNULL), false); 5855 } 5856 }