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 {@link 49 * java.lang.CharSequence </code>character sequences<code>} 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 * <a name="sum"> 77 * <h4> Summary of regular-expression constructs </h4> 78 * 79 * <table border="0" cellpadding="1" cellspacing="0" 80 * summary="Regular expression constructs, and what they match"> 81 * 82 * <tr align="left"> 83 * <th bgcolor="#CCCCFF" align="left" id="construct">Construct</th> 84 * <th bgcolor="#CCCCFF" align="left" id="matches">Matches</th> 85 * </tr> 86 * 87 * <tr><th> </th></tr> 88 * <tr align="left"><th colspan="2" id="characters">Characters</th></tr> 89 * 90 * <tr><td valign="top" headers="construct characters"><i>x</i></td> 91 * <td headers="matches">The character <i>x</i></td></tr> 92 * <tr><td valign="top" headers="construct characters"><tt>\\</tt></td> 93 * <td headers="matches">The backslash character</td></tr> 94 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>n</i></td> 95 * <td headers="matches">The character with octal value <tt>0</tt><i>n</i> 96 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 97 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>nn</i></td> 98 * <td headers="matches">The character with octal value <tt>0</tt><i>nn</i> 99 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 100 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>mnn</i></td> 101 * <td headers="matches">The character with octal value <tt>0</tt><i>mnn</i> 102 * (0 <tt><=</tt> <i>m</i> <tt><=</tt> 3, 103 * 0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 104 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>hh</i></td> 105 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hh</i></td></tr> 106 * <tr><td valign="top" headers="construct characters"><tt>\u</tt><i>hhhh</i></td> 107 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hhhh</i></td></tr> 108 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>{h...h}</i></td> 109 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>h...h</i> 110 * ({@link java.lang.Character#MIN_CODE_POINT Character.MIN_CODE_POINT} 111 * <= <tt>0x</tt><i>h...h</i> <=  112 * {@link java.lang.Character#MAX_CODE_POINT Character.MAX_CODE_POINT})</td></tr> 113 * <tr><td valign="top" headers="matches"><tt>\t</tt></td> 114 * <td headers="matches">The tab character (<tt>'\u0009'</tt>)</td></tr> 115 * <tr><td valign="top" headers="construct characters"><tt>\n</tt></td> 116 * <td headers="matches">The newline (line feed) character (<tt>'\u000A'</tt>)</td></tr> 117 * <tr><td valign="top" headers="construct characters"><tt>\r</tt></td> 118 * <td headers="matches">The carriage-return character (<tt>'\u000D'</tt>)</td></tr> 119 * <tr><td valign="top" headers="construct characters"><tt>\f</tt></td> 120 * <td headers="matches">The form-feed character (<tt>'\u000C'</tt>)</td></tr> 121 * <tr><td valign="top" headers="construct characters"><tt>\a</tt></td> 122 * <td headers="matches">The alert (bell) character (<tt>'\u0007'</tt>)</td></tr> 123 * <tr><td valign="top" headers="construct characters"><tt>\e</tt></td> 124 * <td headers="matches">The escape character (<tt>'\u001B'</tt>)</td></tr> 125 * <tr><td valign="top" headers="construct characters"><tt>\c</tt><i>x</i></td> 126 * <td headers="matches">The control character corresponding to <i>x</i></td></tr> 127 * 128 * <tr><th> </th></tr> 129 * <tr align="left"><th colspan="2" id="classes">Character classes</th></tr> 130 * 131 * <tr><td valign="top" headers="construct classes"><tt>[abc]</tt></td> 132 * <td headers="matches"><tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (simple class)</td></tr> 133 * <tr><td valign="top" headers="construct classes"><tt>[^abc]</tt></td> 134 * <td headers="matches">Any character except <tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (negation)</td></tr> 135 * <tr><td valign="top" headers="construct classes"><tt>[a-zA-Z]</tt></td> 136 * <td headers="matches"><tt>a</tt> through <tt>z</tt> 137 * or <tt>A</tt> through <tt>Z</tt>, inclusive (range)</td></tr> 138 * <tr><td valign="top" headers="construct classes"><tt>[a-d[m-p]]</tt></td> 139 * <td headers="matches"><tt>a</tt> through <tt>d</tt>, 140 * or <tt>m</tt> through <tt>p</tt>: <tt>[a-dm-p]</tt> (union)</td></tr> 141 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[def]]</tt></td> 142 * <td headers="matches"><tt>d</tt>, <tt>e</tt>, or <tt>f</tt> (intersection)</tr> 143 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^bc]]</tt></td> 144 * <td headers="matches"><tt>a</tt> through <tt>z</tt>, 145 * except for <tt>b</tt> and <tt>c</tt>: <tt>[ad-z]</tt> (subtraction)</td></tr> 146 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^m-p]]</tt></td> 147 * <td headers="matches"><tt>a</tt> through <tt>z</tt>, 148 * and not <tt>m</tt> through <tt>p</tt>: <tt>[a-lq-z]</tt>(subtraction)</td></tr> 149 * <tr><th> </th></tr> 150 * 151 * <tr align="left"><th colspan="2" id="predef">Predefined character classes</th></tr> 152 * 153 * <tr><td valign="top" headers="construct predef"><tt>.</tt></td> 154 * <td headers="matches">Any character (may or may not match <a href="#lt">line terminators</a>)</td></tr> 155 * <tr><td valign="top" headers="construct predef"><tt>\d</tt></td> 156 * <td headers="matches">A digit: <tt>[0-9]</tt></td></tr> 157 * <tr><td valign="top" headers="construct predef"><tt>\D</tt></td> 158 * <td headers="matches">A non-digit: <tt>[^0-9]</tt></td></tr> 159 * <tr><td valign="top" headers="construct predef"><tt>\h</tt></td> 160 * <td headers="matches">A horizontal whitespace character: 161 * <tt>[ \t\xA0\u1680\u180e\u2000-\u200a\u202f\u205f\u3000]</tt></td></tr> 162 * <tr><td valign="top" headers="construct predef"><tt>\H</tt></td> 163 * <td headers="matches">A non-horizontal whitespace character: <tt>[^\h]</tt></td></tr> 164 * <tr><td valign="top" headers="construct predef"><tt>\s</tt></td> 165 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 166 * <tr><td valign="top" headers="construct predef"><tt>\S</tt></td> 167 * <td headers="matches">A non-whitespace character: <tt>[^\s]</tt></td></tr> 168 * <tr><td valign="top" headers="construct predef"><tt>\v</tt></td> 169 * <td headers="matches">A vertical whitespace character: <tt>[\n\x0B\f\r\x85\u2028\u2029]</tt> 170 * </td></tr> 171 * <tr><td valign="top" headers="construct predef"><tt>\V</tt></td> 172 * <td headers="matches">A non-vertical whitespace character: <tt>[^\v]</tt></td></tr> 173 * <tr><td valign="top" headers="construct predef"><tt>\w</tt></td> 174 * <td headers="matches">A word character: <tt>[a-zA-Z_0-9]</tt></td></tr> 175 * <tr><td valign="top" headers="construct predef"><tt>\W</tt></td> 176 * <td headers="matches">A non-word character: <tt>[^\w]</tt></td></tr> 177 * <tr><th> </th></tr> 178 * <tr align="left"><th colspan="2" id="posix">POSIX character classes</b> (US-ASCII only)<b></th></tr> 179 * 180 * <tr><td valign="top" headers="construct posix"><tt>\p{Lower}</tt></td> 181 * <td headers="matches">A lower-case alphabetic character: <tt>[a-z]</tt></td></tr> 182 * <tr><td valign="top" headers="construct posix"><tt>\p{Upper}</tt></td> 183 * <td headers="matches">An upper-case alphabetic character:<tt>[A-Z]</tt></td></tr> 184 * <tr><td valign="top" headers="construct posix"><tt>\p{ASCII}</tt></td> 185 * <td headers="matches">All ASCII:<tt>[\x00-\x7F]</tt></td></tr> 186 * <tr><td valign="top" headers="construct posix"><tt>\p{Alpha}</tt></td> 187 * <td headers="matches">An alphabetic character:<tt>[\p{Lower}\p{Upper}]</tt></td></tr> 188 * <tr><td valign="top" headers="construct posix"><tt>\p{Digit}</tt></td> 189 * <td headers="matches">A decimal digit: <tt>[0-9]</tt></td></tr> 190 * <tr><td valign="top" headers="construct posix"><tt>\p{Alnum}</tt></td> 191 * <td headers="matches">An alphanumeric character:<tt>[\p{Alpha}\p{Digit}]</tt></td></tr> 192 * <tr><td valign="top" headers="construct posix"><tt>\p{Punct}</tt></td> 193 * <td headers="matches">Punctuation: One of <tt>!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~</tt></td></tr> 194 * <!-- <tt>[\!"#\$%&'\(\)\*\+,\-\./:;\<=\>\?@\[\\\]\^_`\{\|\}~]</tt> 195 * <tt>[\X21-\X2F\X31-\X40\X5B-\X60\X7B-\X7E]</tt> --> 196 * <tr><td valign="top" headers="construct posix"><tt>\p{Graph}</tt></td> 197 * <td headers="matches">A visible character: <tt>[\p{Alnum}\p{Punct}]</tt></td></tr> 198 * <tr><td valign="top" headers="construct posix"><tt>\p{Print}</tt></td> 199 * <td headers="matches">A printable character: <tt>[\p{Graph}\x20]</tt></td></tr> 200 * <tr><td valign="top" headers="construct posix"><tt>\p{Blank}</tt></td> 201 * <td headers="matches">A space or a tab: <tt>[ \t]</tt></td></tr> 202 * <tr><td valign="top" headers="construct posix"><tt>\p{Cntrl}</tt></td> 203 * <td headers="matches">A control character: <tt>[\x00-\x1F\x7F]</tt></td></tr> 204 * <tr><td valign="top" headers="construct posix"><tt>\p{XDigit}</tt></td> 205 * <td headers="matches">A hexadecimal digit: <tt>[0-9a-fA-F]</tt></td></tr> 206 * <tr><td valign="top" headers="construct posix"><tt>\p{Space}</tt></td> 207 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 208 * 209 * <tr><th> </th></tr> 210 * <tr align="left"><th colspan="2">java.lang.Character classes (simple <a href="#jcc">java character type</a>)</th></tr> 211 * 212 * <tr><td valign="top"><tt>\p{javaLowerCase}</tt></td> 213 * <td>Equivalent to java.lang.Character.isLowerCase()</td></tr> 214 * <tr><td valign="top"><tt>\p{javaUpperCase}</tt></td> 215 * <td>Equivalent to java.lang.Character.isUpperCase()</td></tr> 216 * <tr><td valign="top"><tt>\p{javaWhitespace}</tt></td> 217 * <td>Equivalent to java.lang.Character.isWhitespace()</td></tr> 218 * <tr><td valign="top"><tt>\p{javaMirrored}</tt></td> 219 * <td>Equivalent to java.lang.Character.isMirrored()</td></tr> 220 * 221 * <tr><th> </th></tr> 222 * <tr align="left"><th colspan="2" id="unicode">Classes for Unicode scripts, blocks, categories and binary properties</th></tr> 223 * * <tr><td valign="top" headers="construct unicode"><tt>\p{IsLatin}</tt></td> 224 * <td headers="matches">A Latin script character (<a href="#usc">script</a>)</td></tr> 225 * <tr><td valign="top" headers="construct unicode"><tt>\p{InGreek}</tt></td> 226 * <td headers="matches">A character in the Greek block (<a href="#ubc">block</a>)</td></tr> 227 * <tr><td valign="top" headers="construct unicode"><tt>\p{Lu}</tt></td> 228 * <td headers="matches">An uppercase letter (<a href="#ucc">category</a>)</td></tr> 229 * <tr><td valign="top" headers="construct unicode"><tt>\p{IsAlphabetic}</tt></td> 230 * <td headers="matches">An alphabetic character (<a href="#ubpc">binary property</a>)</td></tr> 231 * <tr><td valign="top" headers="construct unicode"><tt>\p{Sc}</tt></td> 232 * <td headers="matches">A currency symbol</td></tr> 233 * <tr><td valign="top" headers="construct unicode"><tt>\P{InGreek}</tt></td> 234 * <td headers="matches">Any character except one in the Greek block (negation)</td></tr> 235 * <tr><td valign="top" headers="construct unicode"><tt>[\p{L}&&[^\p{Lu}]] </tt></td> 236 * <td headers="matches">Any letter except an uppercase letter (subtraction)</td></tr> 237 * 238 * <tr><th> </th></tr> 239 * <tr align="left"><th colspan="2" id="bounds">Boundary matchers</th></tr> 240 * 241 * <tr><td valign="top" headers="construct bounds"><tt>^</tt></td> 242 * <td headers="matches">The beginning of a line</td></tr> 243 * <tr><td valign="top" headers="construct bounds"><tt>$</tt></td> 244 * <td headers="matches">The end of a line</td></tr> 245 * <tr><td valign="top" headers="construct bounds"><tt>\b</tt></td> 246 * <td headers="matches">A word boundary</td></tr> 247 * <tr><td valign="top" headers="construct bounds"><tt>\B</tt></td> 248 * <td headers="matches">A non-word boundary</td></tr> 249 * <tr><td valign="top" headers="construct bounds"><tt>\A</tt></td> 250 * <td headers="matches">The beginning of the input</td></tr> 251 * <tr><td valign="top" headers="construct bounds"><tt>\G</tt></td> 252 * <td headers="matches">The end of the previous match</td></tr> 253 * <tr><td valign="top" headers="construct bounds"><tt>\Z</tt></td> 254 * <td headers="matches">The end of the input but for the final 255 * <a href="#lt">terminator</a>, if any</td></tr> 256 * <tr><td valign="top" headers="construct bounds"><tt>\z</tt></td> 257 * <td headers="matches">The end of the input</td></tr> 258 * 259 * <tr><th> </th></tr> 260 * <tr align="left"><th colspan="2" id="lineending">Linebreak matcher</th></tr> 261 * <tr><td valign="top" headers="construct lineending"><tt>\R</tt></td> 262 * <td headers="matches">Any Unicode linebreak sequence, is equivalent to 263 * <tt>\u000D\u000A|[\u000A\u000B\u000C\u000D\u0085\u2028\u2029] 264 * </tt></td></tr> 265 * 266 * <tr><th> </th></tr> 267 * <tr align="left"><th colspan="2" id="greedy">Greedy quantifiers</th></tr> 268 * 269 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>?</tt></td> 270 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 271 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>*</tt></td> 272 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 273 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>+</tt></td> 274 * <td headers="matches"><i>X</i>, one or more times</td></tr> 275 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>}</tt></td> 276 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 277 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,}</tt></td> 278 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 279 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}</tt></td> 280 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 281 * 282 * <tr><th> </th></tr> 283 * <tr align="left"><th colspan="2" id="reluc">Reluctant quantifiers</th></tr> 284 * 285 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>??</tt></td> 286 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 287 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>*?</tt></td> 288 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 289 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>+?</tt></td> 290 * <td headers="matches"><i>X</i>, one or more times</td></tr> 291 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>}?</tt></td> 292 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 293 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,}?</tt></td> 294 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 295 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}?</tt></td> 296 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 297 * 298 * <tr><th> </th></tr> 299 * <tr align="left"><th colspan="2" id="poss">Possessive quantifiers</th></tr> 300 * 301 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>?+</tt></td> 302 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 303 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>*+</tt></td> 304 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 305 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>++</tt></td> 306 * <td headers="matches"><i>X</i>, one or more times</td></tr> 307 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>}+</tt></td> 308 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 309 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,}+</tt></td> 310 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 311 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}+</tt></td> 312 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 313 * 314 * <tr><th> </th></tr> 315 * <tr align="left"><th colspan="2" id="logical">Logical operators</th></tr> 316 * 317 * <tr><td valign="top" headers="construct logical"><i>XY</i></td> 318 * <td headers="matches"><i>X</i> followed by <i>Y</i></td></tr> 319 * <tr><td valign="top" headers="construct logical"><i>X</i><tt>|</tt><i>Y</i></td> 320 * <td headers="matches">Either <i>X</i> or <i>Y</i></td></tr> 321 * <tr><td valign="top" headers="construct logical"><tt>(</tt><i>X</i><tt>)</tt></td> 322 * <td headers="matches">X, as a <a href="#cg">capturing group</a></td></tr> 323 * 324 * <tr><th> </th></tr> 325 * <tr align="left"><th colspan="2" id="backref">Back references</th></tr> 326 * 327 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>n</i></td> 328 * <td valign="bottom" headers="matches">Whatever the <i>n</i><sup>th</sup> 329 * <a href="#cg">capturing group</a> matched</td></tr> 330 * 331 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>k</i><<i>name</i>></td> 332 * <td valign="bottom" headers="matches">Whatever the 333 * <a href="#groupname">named-capturing group</a> "name" matched</td></tr> 334 * 335 * <tr><th> </th></tr> 336 * <tr align="left"><th colspan="2" id="quot">Quotation</th></tr> 337 * 338 * <tr><td valign="top" headers="construct quot"><tt>\</tt></td> 339 * <td headers="matches">Nothing, but quotes the following character</td></tr> 340 * <tr><td valign="top" headers="construct quot"><tt>\Q</tt></td> 341 * <td headers="matches">Nothing, but quotes all characters until <tt>\E</tt></td></tr> 342 * <tr><td valign="top" headers="construct quot"><tt>\E</tt></td> 343 * <td headers="matches">Nothing, but ends quoting started by <tt>\Q</tt></td></tr> 344 * <!-- Metachars: !$()*+.<>?[\]^{|} --> 345 * 346 * <tr><th> </th></tr> 347 * <tr align="left"><th colspan="2" id="special">Special constructs (named-capturing and non-capturing)</th></tr> 348 * 349 * <tr><td valign="top" headers="construct special"><tt>(?<<a href="#groupname">name</a>></tt><i>X</i><tt>)</tt></td> 350 * <td headers="matches"><i>X</i>, as a named-capturing group</td></tr> 351 * <tr><td valign="top" headers="construct special"><tt>(?:</tt><i>X</i><tt>)</tt></td> 352 * <td headers="matches"><i>X</i>, as a non-capturing group</td></tr> 353 * <tr><td valign="top" headers="construct special"><tt>(?idmsuxU-idmsuxU) </tt></td> 354 * <td headers="matches">Nothing, but turns match flags <a href="#CASE_INSENSITIVE">i</a> 355 * <a href="#UNIX_LINES">d</a> <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> 356 * <a href="#UNICODE_CASE">u</a> <a href="#COMMENTS">x</a> <a href="#UNICODE_CHARACTER_CLASS">U</a> 357 * on - off</td></tr> 358 * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux:</tt><i>X</i><tt>)</tt> </td> 359 * <td headers="matches"><i>X</i>, as a <a href="#cg">non-capturing group</a> with the 360 * given flags <a href="#CASE_INSENSITIVE">i</a> <a href="#UNIX_LINES">d</a> 361 * <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> <a href="#UNICODE_CASE">u</a > 362 * <a href="#COMMENTS">x</a> on - off</td></tr> 363 * <tr><td valign="top" headers="construct special"><tt>(?=</tt><i>X</i><tt>)</tt></td> 364 * <td headers="matches"><i>X</i>, via zero-width positive lookahead</td></tr> 365 * <tr><td valign="top" headers="construct special"><tt>(?!</tt><i>X</i><tt>)</tt></td> 366 * <td headers="matches"><i>X</i>, via zero-width negative lookahead</td></tr> 367 * <tr><td valign="top" headers="construct special"><tt>(?<=</tt><i>X</i><tt>)</tt></td> 368 * <td headers="matches"><i>X</i>, via zero-width positive lookbehind</td></tr> 369 * <tr><td valign="top" headers="construct special"><tt>(?<!</tt><i>X</i><tt>)</tt></td> 370 * <td headers="matches"><i>X</i>, via zero-width negative lookbehind</td></tr> 371 * <tr><td valign="top" headers="construct special"><tt>(?></tt><i>X</i><tt>)</tt></td> 372 * <td headers="matches"><i>X</i>, as an independent, non-capturing group</td></tr> 373 * 374 * </table> 375 * 376 * <hr> 377 * 378 * 379 * <a name="bs"> 380 * <h4> Backslashes, escapes, and quoting </h4> 381 * 382 * <p> The backslash character (<tt>'\'</tt>) serves to introduce escaped 383 * constructs, as defined in the table above, as well as to quote characters 384 * that otherwise would be interpreted as unescaped constructs. Thus the 385 * expression <tt>\\</tt> matches a single backslash and <tt>\{</tt> matches a 386 * left brace. 387 * 388 * <p> It is an error to use a backslash prior to any alphabetic character that 389 * does not denote an escaped construct; these are reserved for future 390 * extensions to the regular-expression language. A backslash may be used 391 * prior to a non-alphabetic character regardless of whether that character is 392 * part of an unescaped construct. 393 * 394 * <p> Backslashes within string literals in Java source code are interpreted 395 * as required by 396 * <cite>The Java™ Language Specification</cite> 397 * as either Unicode escapes (section 3.3) or other character escapes (section 3.10.6) 398 * It is therefore necessary to double backslashes in string 399 * literals that represent regular expressions to protect them from 400 * interpretation by the Java bytecode compiler. The string literal 401 * <tt>"\b"</tt>, for example, matches a single backspace character when 402 * interpreted as a regular expression, while <tt>"\\b"</tt> matches a 403 * word boundary. The string literal <tt>"\(hello\)"</tt> is illegal 404 * and leads to a compile-time error; in order to match the string 405 * <tt>(hello)</tt> the string literal <tt>"\\(hello\\)"</tt> 406 * must be used. 407 * 408 * <a name="cc"> 409 * <h4> Character Classes </h4> 410 * 411 * <p> Character classes may appear within other character classes, and 412 * may be composed by the union operator (implicit) and the intersection 413 * operator (<tt>&&</tt>). 414 * The union operator denotes a class that contains every character that is 415 * in at least one of its operand classes. The intersection operator 416 * denotes a class that contains every character that is in both of its 417 * operand classes. 418 * 419 * <p> The precedence of character-class operators is as follows, from 420 * highest to lowest: 421 * 422 * <blockquote><table border="0" cellpadding="1" cellspacing="0" 423 * summary="Precedence of character class operators."> 424 * <tr><th>1 </th> 425 * <td>Literal escape </td> 426 * <td><tt>\x</tt></td></tr> 427 * <tr><th>2 </th> 428 * <td>Grouping</td> 429 * <td><tt>[...]</tt></td></tr> 430 * <tr><th>3 </th> 431 * <td>Range</td> 432 * <td><tt>a-z</tt></td></tr> 433 * <tr><th>4 </th> 434 * <td>Union</td> 435 * <td><tt>[a-e][i-u]</tt></td></tr> 436 * <tr><th>5 </th> 437 * <td>Intersection</td> 438 * <td><tt>[a-z&&[aeiou]]</tt></td></tr> 439 * </table></blockquote> 440 * 441 * <p> Note that a different set of metacharacters are in effect inside 442 * a character class than outside a character class. For instance, the 443 * regular expression <tt>.</tt> loses its special meaning inside a 444 * character class, while the expression <tt>-</tt> becomes a range 445 * forming metacharacter. 446 * 447 * <a name="lt"> 448 * <h4> Line terminators </h4> 449 * 450 * <p> A <i>line terminator</i> is a one- or two-character sequence that marks 451 * the end of a line of the input character sequence. The following are 452 * recognized as line terminators: 453 * 454 * <ul> 455 * 456 * <li> A newline (line feed) character (<tt>'\n'</tt>), 457 * 458 * <li> A carriage-return character followed immediately by a newline 459 * character (<tt>"\r\n"</tt>), 460 * 461 * <li> A standalone carriage-return character (<tt>'\r'</tt>), 462 * 463 * <li> A next-line character (<tt>'\u0085'</tt>), 464 * 465 * <li> A line-separator character (<tt>'\u2028'</tt>), or 466 * 467 * <li> A paragraph-separator character (<tt>'\u2029</tt>). 468 * 469 * </ul> 470 * <p>If {@link #UNIX_LINES} mode is activated, then the only line terminators 471 * recognized are newline characters. 472 * 473 * <p> The regular expression <tt>.</tt> matches any character except a line 474 * terminator unless the {@link #DOTALL} flag is specified. 475 * 476 * <p> By default, the regular expressions <tt>^</tt> and <tt>$</tt> ignore 477 * line terminators and only match at the beginning and the end, respectively, 478 * of the entire input sequence. If {@link #MULTILINE} mode is activated then 479 * <tt>^</tt> matches at the beginning of input and after any line terminator 480 * except at the end of input. When in {@link #MULTILINE} mode <tt>$</tt> 481 * matches just before a line terminator or the end of the input sequence. 482 * 483 * <a name="cg"> 484 * <h4> Groups and capturing </h4> 485 * 486 * <a name="gnumber"> 487 * <h5> Group number </h5> 488 * <p> Capturing groups are numbered by counting their opening parentheses from 489 * left to right. In the expression <tt>((A)(B(C)))</tt>, for example, there 490 * are four such groups: </p> 491 * 492 * <blockquote><table cellpadding=1 cellspacing=0 summary="Capturing group numberings"> 493 * <tr><th>1 </th> 494 * <td><tt>((A)(B(C)))</tt></td></tr> 495 * <tr><th>2 </th> 496 * <td><tt>(A)</tt></td></tr> 497 * <tr><th>3 </th> 498 * <td><tt>(B(C))</tt></td></tr> 499 * <tr><th>4 </th> 500 * <td><tt>(C)</tt></td></tr> 501 * </table></blockquote> 502 * 503 * <p> Group zero always stands for the entire expression. 504 * 505 * <p> Capturing groups are so named because, during a match, each subsequence 506 * of the input sequence that matches such a group is saved. The captured 507 * subsequence may be used later in the expression, via a back reference, and 508 * may also be retrieved from the matcher once the match operation is complete. 509 * 510 * <a name="groupname"> 511 * <h5> Group name </h5> 512 * <p>A capturing group can also be assigned a "name", a <tt>named-capturing group</tt>, 513 * and then be back-referenced later by the "name". Group names are composed of 514 * the following characters. The first character must be a <tt>letter</tt>. 515 * 516 * <ul> 517 * <li> The uppercase letters <tt>'A'</tt> through <tt>'Z'</tt> 518 * (<tt>'\u0041'</tt> through <tt>'\u005a'</tt>), 519 * <li> The lowercase letters <tt>'a'</tt> through <tt>'z'</tt> 520 * (<tt>'\u0061'</tt> through <tt>'\u007a'</tt>), 521 * <li> The digits <tt>'0'</tt> through <tt>'9'</tt> 522 * (<tt>'\u0030'</tt> through <tt>'\u0039'</tt>), 523 * </ul> 524 * 525 * <p> A <tt>named-capturing group</tt> is still numbered as described in 526 * <a href="#gnumber">Group number</a>. 527 * 528 * <p> The captured input associated with a group is always the subsequence 529 * that the group most recently matched. If a group is evaluated a second time 530 * because of quantification then its previously-captured value, if any, will 531 * be retained if the second evaluation fails. Matching the string 532 * <tt>"aba"</tt> against the expression <tt>(a(b)?)+</tt>, for example, leaves 533 * group two set to <tt>"b"</tt>. All captured input is discarded at the 534 * beginning of each match. 535 * 536 * <p> Groups beginning with <tt>(?</tt> are either pure, <i>non-capturing</i> groups 537 * that do not capture text and do not count towards the group total, or 538 * <i>named-capturing</i> group. 539 * 540 * <h4> Unicode support </h4> 541 * 542 * <p> This class is in conformance with Level 1 of <a 543 * href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical 544 * Standard #18: Unicode Regular Expression</i></a>, plus RL2.1 545 * Canonical Equivalents. 546 * <p> 547 * <b>Unicode escape sequences</b> such as <tt>\u2014</tt> in Java source code 548 * are processed as described in section 3.3 of 549 * <cite>The Java™ Language Specification</cite>. 550 * Such escape sequences are also implemented directly by the regular-expression 551 * parser so that Unicode escapes can be used in expressions that are read from 552 * files or from the keyboard. Thus the strings <tt>"\u2014"</tt> and 553 * <tt>"\\u2014"</tt>, while not equal, compile into the same pattern, which 554 * matches the character with hexadecimal value <tt>0x2014</tt>. 555 * <p> 556 * A Unicode character can also be represented in a regular-expression by 557 * using its <b>Hex notation</b>(hexadecimal code point value) directly as described in construct 558 * <tt>\x{...}</tt>, for example a supplementary character U+2011F 559 * can be specified as <tt>\x{2011F}</tt>, instead of two consecutive 560 * Unicode escape sequences of the surrogate pair 561 * <tt>\uD840</tt><tt>\uDD1F</tt>. 562 * <p> 563 * Unicode scripts, blocks, categories and binary properties are written with 564 * the <tt>\p</tt> and <tt>\P</tt> constructs as in Perl. 565 * <tt>\p{</tt><i>prop</i><tt>}</tt> matches if 566 * the input has the property <i>prop</i>, while <tt>\P{</tt><i>prop</i><tt>}</tt> 567 * does not match if the input has that property. 568 * <p> 569 * Scripts, blocks, categories and binary properties can be used both inside 570 * and outside of a character class. 571 * <a name="usc"> 572 * <p> 573 * <b>Scripts</b> are specified either with the prefix {@code Is}, as in 574 * {@code IsHiragana}, or by using the {@code script} keyword (or its short 575 * form {@code sc})as in {@code script=Hiragana} or {@code sc=Hiragana}. 576 * <p> 577 * The script names supported by <code>Pattern</code> are the valid script names 578 * accepted and defined by 579 * {@link java.lang.Character.UnicodeScript#forName(String) UnicodeScript.forName}. 580 * <a name="ubc"> 581 * <p> 582 * <b>Blocks</b> are specified with the prefix {@code In}, as in 583 * {@code InMongolian}, or by using the keyword {@code block} (or its short 584 * form {@code blk}) as in {@code block=Mongolian} or {@code blk=Mongolian}. 585 * <p> 586 * The block names supported by <code>Pattern</code> are the valid block names 587 * accepted and defined by 588 * {@link java.lang.Character.UnicodeBlock#forName(String) UnicodeBlock.forName}. 589 * <p> 590 * <a name="ucc"> 591 * <b>Categories</b> may be specified with the optional prefix {@code Is}: 592 * Both {@code \p{L}} and {@code \p{IsL}} denote the category of Unicode 593 * letters. Same as scripts and blocks, categories can also be specified 594 * by using the keyword {@code general_category} (or its short form 595 * {@code gc}) as in {@code general_category=Lu} or {@code gc=Lu}. 596 * <p> 597 * The supported categories are those of 598 * <a href="http://www.unicode.org/unicode/standard/standard.html"> 599 * <i>The Unicode Standard</i></a> in the version specified by the 600 * {@link java.lang.Character Character} class. The category names are those 601 * defined in the Standard, both normative and informative. 602 * <p> 603 * <a name="ubpc"> 604 * <b>Binary properties</b> are specified with the prefix {@code Is}, as in 605 * {@code IsAlphabetic}. The supported binary properties by <code>Pattern</code> 606 * are 607 * <ul> 608 * <li> Alphabetic 609 * <li> Ideographic 610 * <li> Letter 611 * <li> Lowercase 612 * <li> Uppercase 613 * <li> Titlecase 614 * <li> Punctuation 615 * <Li> Control 616 * <li> White_Space 617 * <li> Digit 618 * <li> Hex_Digit 619 * <li> Join_Control 620 * <li> Noncharacter_Code_Point 621 * <li> Assigned 622 * </ul> 623 * <p> 624 * The following <b>Predefined Character classes</b> and <b>POSIX character classes</b> 625 * are in conformance with the recommendation of <i>Annex C: Compatibility Properties</i> 626 * of <a href="http://www.unicode.org/reports/tr18/"><i>Unicode Regular Expression 627 * </i></a>, when {@link #UNICODE_CHARACTER_CLASS} flag is specified. 628 * <p> 629 * <table border="0" cellpadding="1" cellspacing="0" 630 * summary="predefined and posix character classes in Unicode mode"> 631 * <tr align="left"> 632 * <th bgcolor="#CCCCFF" align="left" id="classes">Classes</th> 633 * <th bgcolor="#CCCCFF" align="left" id="matches">Matches</th> 634 *</tr> 635 * <tr><td><tt>\p{Lower}</tt></td> 636 * <td>A lowercase character:<tt>\p{IsLowercase}</tt></td></tr> 637 * <tr><td><tt>\p{Upper}</tt></td> 638 * <td>An uppercase character:<tt>\p{IsUppercase}</tt></td></tr> 639 * <tr><td><tt>\p{ASCII}</tt></td> 640 * <td>All ASCII:<tt>[\x00-\x7F]</tt></td></tr> 641 * <tr><td><tt>\p{Alpha}</tt></td> 642 * <td>An alphabetic character:<tt>\p{IsAlphabetic}</tt></td></tr> 643 * <tr><td><tt>\p{Digit}</tt></td> 644 * <td>A decimal digit character:<tt>p{IsDigit}</tt></td></tr> 645 * <tr><td><tt>\p{Alnum}</tt></td> 646 * <td>An alphanumeric character:<tt>[\p{IsAlphabetic}\p{IsDigit}]</tt></td></tr> 647 * <tr><td><tt>\p{Punct}</tt></td> 648 * <td>A punctuation character:<tt>p{IsPunctuation}</tt></td></tr> 649 * <tr><td><tt>\p{Graph}</tt></td> 650 * <td>A visible character: <tt>[^\p{IsWhite_Space}\p{gc=Cc}\p{gc=Cs}\p{gc=Cn}]</tt></td></tr> 651 * <tr><td><tt>\p{Print}</tt></td> 652 * <td>A printable character: <tt>[\p{Graph}\p{Blank}&&[^\p{Cntrl}]]</tt></td></tr> 653 * <tr><td><tt>\p{Blank}</tt></td> 654 * <td>A space or a tab: <tt>[\p{IsWhite_Space}&&[^\p{gc=Zl}\p{gc=Zp}\x0a\x0b\x0c\x0d\x85]]</tt></td></tr> 655 * <tr><td><tt>\p{Cntrl}</tt></td> 656 * <td>A control character: <tt>\p{gc=Cc}</tt></td></tr> 657 * <tr><td><tt>\p{XDigit}</tt></td> 658 * <td>A hexadecimal digit: <tt>[\p{gc=Nd}\p{IsHex_Digit}]</tt></td></tr> 659 * <tr><td><tt>\p{Space}</tt></td> 660 * <td>A whitespace character:<tt>\p{IsWhite_Space}</tt></td></tr> 661 * <tr><td><tt>\d</tt></td> 662 * <td>A digit: <tt>\p{IsDigit}</tt></td></tr> 663 * <tr><td><tt>\D</tt></td> 664 * <td>A non-digit: <tt>[^\d]</tt></td></tr> 665 * <tr><td><tt>\s</tt></td> 666 * <td>A whitespace character: <tt>\p{IsWhite_Space}</tt></td></tr> 667 * <tr><td><tt>\S</tt></td> 668 * <td>A non-whitespace character: <tt>[^\s]</tt></td></tr> 669 * <tr><td><tt>\w</tt></td> 670 * <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> 671 * <tr><td><tt>\W</tt></td> 672 * <td>A non-word character: <tt>[^\w]</tt></td></tr> 673 * </table> 674 * <p> 675 * <a name="jcc"> 676 * Categories that behave like the java.lang.Character 677 * boolean is<i>methodname</i> methods (except for the deprecated ones) are 678 * available through the same <tt>\p{</tt><i>prop</i><tt>}</tt> syntax where 679 * the specified property has the name <tt>java<i>methodname</i></tt>. 680 * 681 * <h4> Comparison to Perl 5 </h4> 682 * 683 * <p>The <code>Pattern</code> engine performs traditional NFA-based matching 684 * with ordered alternation as occurs in Perl 5. 685 * 686 * <p> Perl constructs not supported by this class: </p> 687 * 688 * <ul> 689 * <li><p> Predefined character classes (Unicode character) 690 * <p><tt>\X </tt>Match Unicode 691 * <a href="http://www.unicode.org/reports/tr18/#Default_Grapheme_Clusters"> 692 * <i>extended grapheme cluster</i></a> 693 * </p></li> 694 * 695 * <li><p> The backreference constructs, <tt>\g{</tt><i>n</i><tt>}</tt> for 696 * the <i>n</i><sup>th</sup><a href="#cg">capturing group</a> and 697 * <tt>\g{</tt><i>name</i><tt>}</tt> for 698 * <a href="#groupname">named-capturing group</a>. 699 * </p></li> 700 * 701 * <li><p> The named character construct, <tt>\N{</tt><i>name</i><tt>}</tt> 702 * for a Unicode character by its name. 703 * </p></li> 704 * 705 * <li><p> The conditional constructs 706 * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>)</tt> and 707 * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>|</tt><i>Y</i><tt>)</tt>, 708 * </p></li> 709 * 710 * <li><p> The embedded code constructs <tt>(?{</tt><i>code</i><tt>})</tt> 711 * and <tt>(??{</tt><i>code</i><tt>})</tt>,</p></li> 712 * 713 * <li><p> The embedded comment syntax <tt>(?#comment)</tt>, and </p></li> 714 * 715 * <li><p> The preprocessing operations <tt>\l</tt> <tt>\u</tt>, 716 * <tt>\L</tt>, and <tt>\U</tt>. </p></li> 717 * 718 * </ul> 719 * 720 * <p> Constructs supported by this class but not by Perl: </p> 721 * 722 * <ul> 723 * 724 * <li><p> Character-class union and intersection as described 725 * <a href="#cc">above</a>.</p></li> 726 * 727 * </ul> 728 * 729 * <p> Notable differences from Perl: </p> 730 * 731 * <ul> 732 * 733 * <li><p> In Perl, <tt>\1</tt> through <tt>\9</tt> are always interpreted 734 * as back references; a backslash-escaped number greater than <tt>9</tt> is 735 * treated as a back reference if at least that many subexpressions exist, 736 * otherwise it is interpreted, if possible, as an octal escape. In this 737 * class octal escapes must always begin with a zero. In this class, 738 * <tt>\1</tt> through <tt>\9</tt> are always interpreted as back 739 * references, and a larger number is accepted as a back reference if at 740 * least that many subexpressions exist at that point in the regular 741 * expression, otherwise the parser will drop digits until the number is 742 * smaller or equal to the existing number of groups or it is one digit. 743 * </p></li> 744 * 745 * <li><p> Perl uses the <tt>g</tt> flag to request a match that resumes 746 * where the last match left off. This functionality is provided implicitly 747 * by the {@link Matcher} class: Repeated invocations of the {@link 748 * Matcher#find find} method will resume where the last match left off, 749 * unless the matcher is reset. </p></li> 750 * 751 * <li><p> In Perl, embedded flags at the top level of an expression affect 752 * the whole expression. In this class, embedded flags always take effect 753 * at the point at which they appear, whether they are at the top level or 754 * within a group; in the latter case, flags are restored at the end of the 755 * group just as in Perl. </p></li> 756 * 757 * </ul> 758 * 759 * 760 * <p> For a more precise description of the behavior of regular expression 761 * constructs, please see <a href="http://www.oreilly.com/catalog/regex3/"> 762 * <i>Mastering Regular Expressions, 3nd Edition</i>, Jeffrey E. F. Friedl, 763 * O'Reilly and Associates, 2006.</a> 764 * </p> 765 * 766 * @see java.lang.String#split(String, int) 767 * @see java.lang.String#split(String) 768 * 769 * @author Mike McCloskey 770 * @author Mark Reinhold 771 * @author JSR-51 Expert Group 772 * @since 1.4 773 * @spec JSR-51 774 */ 775 776 public final class Pattern 777 implements java.io.Serializable 778 { 779 780 /** 781 * Regular expression modifier values. Instead of being passed as 782 * arguments, they can also be passed as inline modifiers. 783 * For example, the following statements have the same effect. 784 * <pre> 785 * RegExp r1 = RegExp.compile("abc", Pattern.I|Pattern.M); 786 * RegExp r2 = RegExp.compile("(?im)abc", 0); 787 * </pre> 788 * 789 * The flags are duplicated so that the familiar Perl match flag 790 * names are available. 791 */ 792 793 /** 794 * Enables Unix lines mode. 795 * 796 * <p> In this mode, only the <tt>'\n'</tt> line terminator is recognized 797 * in the behavior of <tt>.</tt>, <tt>^</tt>, and <tt>$</tt>. 798 * 799 * <p> Unix lines mode can also be enabled via the embedded flag 800 * expression <tt>(?d)</tt>. 801 */ 802 public static final int UNIX_LINES = 0x01; 803 804 /** 805 * Enables case-insensitive matching. 806 * 807 * <p> By default, case-insensitive matching assumes that only characters 808 * in the US-ASCII charset are being matched. Unicode-aware 809 * case-insensitive matching can be enabled by specifying the {@link 810 * #UNICODE_CASE} flag in conjunction with this flag. 811 * 812 * <p> Case-insensitive matching can also be enabled via the embedded flag 813 * expression <tt>(?i)</tt>. 814 * 815 * <p> Specifying this flag may impose a slight performance penalty. </p> 816 */ 817 public static final int CASE_INSENSITIVE = 0x02; 818 819 /** 820 * Permits whitespace and comments in pattern. 821 * 822 * <p> In this mode, whitespace is ignored, and embedded comments starting 823 * with <tt>#</tt> are ignored until the end of a line. 824 * 825 * <p> Comments mode can also be enabled via the embedded flag 826 * expression <tt>(?x)</tt>. 827 */ 828 public static final int COMMENTS = 0x04; 829 830 /** 831 * Enables multiline mode. 832 * 833 * <p> In multiline mode the expressions <tt>^</tt> and <tt>$</tt> match 834 * just after or just before, respectively, a line terminator or the end of 835 * the input sequence. By default these expressions only match at the 836 * beginning and the end of the entire input sequence. 837 * 838 * <p> Multiline mode can also be enabled via the embedded flag 839 * expression <tt>(?m)</tt>. </p> 840 */ 841 public static final int MULTILINE = 0x08; 842 843 /** 844 * Enables literal parsing of the pattern. 845 * 846 * <p> When this flag is specified then the input string that specifies 847 * the pattern is treated as a sequence of literal characters. 848 * Metacharacters or escape sequences in the input sequence will be 849 * given no special meaning. 850 * 851 * <p>The flags CASE_INSENSITIVE and UNICODE_CASE retain their impact on 852 * matching when used in conjunction with this flag. The other flags 853 * become superfluous. 854 * 855 * <p> There is no embedded flag character for enabling literal parsing. 856 * @since 1.5 857 */ 858 public static final int LITERAL = 0x10; 859 860 /** 861 * Enables dotall mode. 862 * 863 * <p> In dotall mode, the expression <tt>.</tt> matches any character, 864 * including a line terminator. By default this expression does not match 865 * line terminators. 866 * 867 * <p> Dotall mode can also be enabled via the embedded flag 868 * expression <tt>(?s)</tt>. (The <tt>s</tt> is a mnemonic for 869 * "single-line" mode, which is what this is called in Perl.) </p> 870 */ 871 public static final int DOTALL = 0x20; 872 873 /** 874 * Enables Unicode-aware case folding. 875 * 876 * <p> When this flag is specified then case-insensitive matching, when 877 * enabled by the {@link #CASE_INSENSITIVE} flag, is done in a manner 878 * consistent with the Unicode Standard. By default, case-insensitive 879 * matching assumes that only characters in the US-ASCII charset are being 880 * matched. 881 * 882 * <p> Unicode-aware case folding can also be enabled via the embedded flag 883 * expression <tt>(?u)</tt>. 884 * 885 * <p> Specifying this flag may impose a performance penalty. </p> 886 */ 887 public static final int UNICODE_CASE = 0x40; 888 889 /** 890 * Enables canonical equivalence. 891 * 892 * <p> When this flag is specified then two characters will be considered 893 * to match if, and only if, their full canonical decompositions match. 894 * The expression <tt>"a\u030A"</tt>, for example, will match the 895 * string <tt>"\u00E5"</tt> when this flag is specified. By default, 896 * matching does not take canonical equivalence into account. 897 * 898 * <p> There is no embedded flag character for enabling canonical 899 * equivalence. 900 * 901 * <p> Specifying this flag may impose a performance penalty. </p> 902 */ 903 public static final int CANON_EQ = 0x80; 904 905 /** 906 * Enables the Unicode version of <i>Predefined character classes</i> and 907 * <i>POSIX character classes</i>. 908 * 909 * <p> When this flag is specified then the (US-ASCII only) 910 * <i>Predefined character classes</i> and <i>POSIX character classes</i> 911 * are in conformance with 912 * <a href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical 913 * Standard #18: Unicode Regular Expression</i></a> 914 * <i>Annex C: Compatibility Properties</i>. 915 * <p> 916 * The UNICODE_CHARACTER_CLASS mode can also be enabled via the embedded 917 * flag expression <tt>(?U)</tt>. 918 * <p> 919 * The flag implies UNICODE_CASE, that is, it enables Unicode-aware case 920 * folding. 921 * <p> 922 * Specifying this flag may impose a performance penalty. </p> 923 * @since 1.7 924 */ 925 public static final int UNICODE_CHARACTER_CLASS = 0x100; 926 927 /* Pattern has only two serialized components: The pattern string 928 * and the flags, which are all that is needed to recompile the pattern 929 * when it is deserialized. 930 */ 931 932 /** use serialVersionUID from Merlin b59 for interoperability */ 933 private static final long serialVersionUID = 5073258162644648461L; 934 935 /** 936 * The original regular-expression pattern string. 937 * 938 * @serial 939 */ 940 private String pattern; 941 942 /** 943 * The original pattern flags. 944 * 945 * @serial 946 */ 947 private int flags; 948 949 /** 950 * Boolean indicating this Pattern is compiled; this is necessary in order 951 * to lazily compile deserialized Patterns. 952 */ 953 private transient volatile boolean compiled = false; 954 955 /** 956 * The normalized pattern string. 957 */ 958 private transient String normalizedPattern; 959 960 /** 961 * The starting point of state machine for the find operation. This allows 962 * a match to start anywhere in the input. 963 */ 964 transient Node root; 965 966 /** 967 * The root of object tree for a match operation. The pattern is matched 968 * at the beginning. This may include a find that uses BnM or a First 969 * node. 970 */ 971 transient Node matchRoot; 972 973 /** 974 * Temporary storage used by parsing pattern slice. 975 */ 976 transient int[] buffer; 977 978 /** 979 * Map the "name" of the "named capturing group" to its group id 980 * node. 981 */ 982 transient volatile Map<String, Integer> namedGroups; 983 984 /** 985 * Temporary storage used while parsing group references. 986 */ 987 transient GroupHead[] groupNodes; 988 989 /** 990 * Temporary null terminated code point array used by pattern compiling. 991 */ 992 private transient int[] temp; 993 994 /** 995 * The number of capturing groups in this Pattern. Used by matchers to 996 * allocate storage needed to perform a match. 997 */ 998 transient int capturingGroupCount; 999 1000 /** 1001 * The local variable count used by parsing tree. Used by matchers to 1002 * allocate storage needed to perform a match. 1003 */ 1004 transient int localCount; 1005 1006 /** 1007 * Index into the pattern string that keeps track of how much has been 1008 * parsed. 1009 */ 1010 private transient int cursor; 1011 1012 /** 1013 * Holds the length of the pattern string. 1014 */ 1015 private transient int patternLength; 1016 1017 /** 1018 * If the Start node might possibly match supplementary characters. 1019 * It is set to true during compiling if 1020 * (1) There is supplementary char in pattern, or 1021 * (2) There is complement node of Category or Block 1022 */ 1023 private transient boolean hasSupplementary; 1024 1025 /** 1026 * Compiles the given regular expression into a pattern. </p> 1027 * 1028 * @param regex 1029 * The expression to be compiled 1030 * 1031 * @throws PatternSyntaxException 1032 * If the expression's syntax is invalid 1033 */ 1034 public static Pattern compile(String regex) { 1035 return new Pattern(regex, 0); 1036 } 1037 1038 /** 1039 * Compiles the given regular expression into a pattern with the given 1040 * flags. </p> 1041 * 1042 * @param regex 1043 * The expression to be compiled 1044 * 1045 * @param flags 1046 * Match flags, a bit mask that may include 1047 * {@link #CASE_INSENSITIVE}, {@link #MULTILINE}, {@link #DOTALL}, 1048 * {@link #UNICODE_CASE}, {@link #CANON_EQ}, {@link #UNIX_LINES}, 1049 * {@link #LITERAL}, {@link #UNICODE_CHARACTER_CLASS} 1050 * and {@link #COMMENTS} 1051 * 1052 * @throws IllegalArgumentException 1053 * If bit values other than those corresponding to the defined 1054 * match flags are set in <tt>flags</tt> 1055 * 1056 * @throws PatternSyntaxException 1057 * If the expression's syntax is invalid 1058 */ 1059 public static Pattern compile(String regex, int flags) { 1060 return new Pattern(regex, flags); 1061 } 1062 1063 /** 1064 * Returns the regular expression from which this pattern was compiled. 1065 * </p> 1066 * 1067 * @return The source of this pattern 1068 */ 1069 public String pattern() { 1070 return pattern; 1071 } 1072 1073 /** 1074 * <p>Returns the string representation of this pattern. This 1075 * is the regular expression from which this pattern was 1076 * compiled.</p> 1077 * 1078 * @return The string representation of this pattern 1079 * @since 1.5 1080 */ 1081 public String toString() { 1082 return pattern; 1083 } 1084 1085 /** 1086 * Creates a matcher that will match the given input against this pattern. 1087 * </p> 1088 * 1089 * @param input 1090 * The character sequence to be matched 1091 * 1092 * @return A new matcher for this pattern 1093 */ 1094 public Matcher matcher(CharSequence input) { 1095 if (!compiled) { 1096 synchronized(this) { 1097 if (!compiled) 1098 compile(); 1099 } 1100 } 1101 Matcher m = new Matcher(this, input); 1102 return m; 1103 } 1104 1105 /** 1106 * Returns this pattern's match flags. </p> 1107 * 1108 * @return The match flags specified when this pattern was compiled 1109 */ 1110 public int flags() { 1111 return flags; 1112 } 1113 1114 /** 1115 * Compiles the given regular expression and attempts to match the given 1116 * input against it. 1117 * 1118 * <p> An invocation of this convenience method of the form 1119 * 1120 * <blockquote><pre> 1121 * Pattern.matches(regex, input);</pre></blockquote> 1122 * 1123 * behaves in exactly the same way as the expression 1124 * 1125 * <blockquote><pre> 1126 * Pattern.compile(regex).matcher(input).matches()</pre></blockquote> 1127 * 1128 * <p> If a pattern is to be used multiple times, compiling it once and reusing 1129 * it will be more efficient than invoking this method each time. </p> 1130 * 1131 * @param regex 1132 * The expression to be compiled 1133 * 1134 * @param input 1135 * The character sequence to be matched 1136 * 1137 * @throws PatternSyntaxException 1138 * If the expression's syntax is invalid 1139 */ 1140 public static boolean matches(String regex, CharSequence input) { 1141 Pattern p = Pattern.compile(regex); 1142 Matcher m = p.matcher(input); 1143 return m.matches(); 1144 } 1145 1146 /** 1147 * Splits the given input sequence around matches of this pattern. 1148 * 1149 * <p> The array returned by this method contains each substring of the 1150 * input sequence that is terminated by another subsequence that matches 1151 * this pattern or is terminated by the end of the input sequence. The 1152 * substrings in the array are in the order in which they occur in the 1153 * input. If this pattern does not match any subsequence of the input then 1154 * the resulting array has just one element, namely the input sequence in 1155 * string form. 1156 * 1157 * <p> The <tt>limit</tt> parameter controls the number of times the 1158 * pattern is applied and therefore affects the length of the resulting 1159 * array. If the limit <i>n</i> is greater than zero then the pattern 1160 * will be applied at most <i>n</i> - 1 times, the array's 1161 * length will be no greater than <i>n</i>, and the array's last entry 1162 * will contain all input beyond the last matched delimiter. If <i>n</i> 1163 * is non-positive then the pattern will be applied as many times as 1164 * possible and the array can have any length. If <i>n</i> is zero then 1165 * the pattern will be applied as many times as possible, the array can 1166 * have any length, and trailing empty strings will be discarded. 1167 * 1168 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1169 * results with these parameters: 1170 * 1171 * <blockquote><table cellpadding=1 cellspacing=0 1172 * summary="Split examples showing regex, limit, and result"> 1173 * <tr><th><P align="left"><i>Regex </i></th> 1174 * <th><P align="left"><i>Limit </i></th> 1175 * <th><P align="left"><i>Result </i></th></tr> 1176 * <tr><td align=center>:</td> 1177 * <td align=center>2</td> 1178 * <td><tt>{ "boo", "and:foo" }</tt></td></tr> 1179 * <tr><td align=center>:</td> 1180 * <td align=center>5</td> 1181 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1182 * <tr><td align=center>:</td> 1183 * <td align=center>-2</td> 1184 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1185 * <tr><td align=center>o</td> 1186 * <td align=center>5</td> 1187 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1188 * <tr><td align=center>o</td> 1189 * <td align=center>-2</td> 1190 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 1191 * <tr><td align=center>o</td> 1192 * <td align=center>0</td> 1193 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1194 * </table></blockquote> 1195 * 1196 * 1197 * @param input 1198 * The character sequence to be split 1199 * 1200 * @param limit 1201 * The result threshold, as described above 1202 * 1203 * @return The array of strings computed by splitting the input 1204 * around matches of this pattern 1205 */ 1206 public String[] split(CharSequence input, int limit) { 1207 int index = 0; 1208 boolean matchLimited = limit > 0; 1209 ArrayList<String> matchList = new ArrayList<>(); 1210 Matcher m = matcher(input); 1211 1212 // Add segments before each match found 1213 while(m.find()) { 1214 if (!matchLimited || matchList.size() < limit - 1) { 1215 String match = input.subSequence(index, m.start()).toString(); 1216 matchList.add(match); 1217 index = m.end(); 1218 } else if (matchList.size() == limit - 1) { // last one 1219 String match = input.subSequence(index, 1220 input.length()).toString(); 1221 matchList.add(match); 1222 index = m.end(); 1223 } 1224 } 1225 1226 // If no match was found, return this 1227 if (index == 0) 1228 return new String[] {input.toString()}; 1229 1230 // Add remaining segment 1231 if (!matchLimited || matchList.size() < limit) 1232 matchList.add(input.subSequence(index, input.length()).toString()); 1233 1234 // Construct result 1235 int resultSize = matchList.size(); 1236 if (limit == 0) 1237 while (resultSize > 0 && matchList.get(resultSize-1).equals("")) 1238 resultSize--; 1239 String[] result = new String[resultSize]; 1240 return matchList.subList(0, resultSize).toArray(result); 1241 } 1242 1243 /** 1244 * Splits the given input sequence around matches of this pattern. 1245 * 1246 * <p> This method works as if by invoking the two-argument {@link 1247 * #split(java.lang.CharSequence, int) split} method with the given input 1248 * sequence and a limit argument of zero. Trailing empty strings are 1249 * therefore not included in the resulting array. </p> 1250 * 1251 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1252 * results with these expressions: 1253 * 1254 * <blockquote><table cellpadding=1 cellspacing=0 1255 * summary="Split examples showing regex and result"> 1256 * <tr><th><P align="left"><i>Regex </i></th> 1257 * <th><P align="left"><i>Result</i></th></tr> 1258 * <tr><td align=center>:</td> 1259 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1260 * <tr><td align=center>o</td> 1261 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1262 * </table></blockquote> 1263 * 1264 * 1265 * @param input 1266 * The character sequence to be split 1267 * 1268 * @return The array of strings computed by splitting the input 1269 * around matches of this pattern 1270 */ 1271 public String[] split(CharSequence input) { 1272 return split(input, 0); 1273 } 1274 1275 /** 1276 * Returns a literal pattern <code>String</code> for the specified 1277 * <code>String</code>. 1278 * 1279 * <p>This method produces a <code>String</code> that can be used to 1280 * create a <code>Pattern</code> that would match the string 1281 * <code>s</code> as if it were a literal pattern.</p> Metacharacters 1282 * or escape sequences in the input sequence will be given no special 1283 * meaning. 1284 * 1285 * @param s The string to be literalized 1286 * @return A literal string replacement 1287 * @since 1.5 1288 */ 1289 public static String quote(String s) { 1290 int slashEIndex = s.indexOf("\\E"); 1291 if (slashEIndex == -1) 1292 return "\\Q" + s + "\\E"; 1293 1294 StringBuilder sb = new StringBuilder(s.length() * 2); 1295 sb.append("\\Q"); 1296 slashEIndex = 0; 1297 int current = 0; 1298 while ((slashEIndex = s.indexOf("\\E", current)) != -1) { 1299 sb.append(s.substring(current, slashEIndex)); 1300 current = slashEIndex + 2; 1301 sb.append("\\E\\\\E\\Q"); 1302 } 1303 sb.append(s.substring(current, s.length())); 1304 sb.append("\\E"); 1305 return sb.toString(); 1306 } 1307 1308 /** 1309 * Recompile the Pattern instance from a stream. The original pattern 1310 * string is read in and the object tree is recompiled from it. 1311 */ 1312 private void readObject(java.io.ObjectInputStream s) 1313 throws java.io.IOException, ClassNotFoundException { 1314 1315 // Read in all fields 1316 s.defaultReadObject(); 1317 1318 // Initialize counts 1319 capturingGroupCount = 1; 1320 localCount = 0; 1321 1322 // if length > 0, the Pattern is lazily compiled 1323 compiled = false; 1324 if (pattern.length() == 0) { 1325 root = new Start(lastAccept); 1326 matchRoot = lastAccept; 1327 compiled = true; 1328 } 1329 } 1330 1331 /** 1332 * This private constructor is used to create all Patterns. The pattern 1333 * string and match flags are all that is needed to completely describe 1334 * a Pattern. An empty pattern string results in an object tree with 1335 * only a Start node and a LastNode node. 1336 */ 1337 private Pattern(String p, int f) { 1338 pattern = p; 1339 flags = f; 1340 1341 // to use UNICODE_CASE if UNICODE_CHARACTER_CLASS present 1342 if ((flags & UNICODE_CHARACTER_CLASS) != 0) 1343 flags |= UNICODE_CASE; 1344 1345 // Reset group index count 1346 capturingGroupCount = 1; 1347 localCount = 0; 1348 1349 if (pattern.length() > 0) { 1350 compile(); 1351 } else { 1352 root = new Start(lastAccept); 1353 matchRoot = lastAccept; 1354 } 1355 } 1356 1357 /** 1358 * The pattern is converted to normalizedD form and then a pure group 1359 * is constructed to match canonical equivalences of the characters. 1360 */ 1361 private void normalize() { 1362 boolean inCharClass = false; 1363 int lastCodePoint = -1; 1364 1365 // Convert pattern into normalizedD form 1366 normalizedPattern = Normalizer.normalize(pattern, Normalizer.Form.NFD); 1367 patternLength = normalizedPattern.length(); 1368 1369 // Modify pattern to match canonical equivalences 1370 StringBuilder newPattern = new StringBuilder(patternLength); 1371 for(int i=0; i<patternLength; ) { 1372 int c = normalizedPattern.codePointAt(i); 1373 StringBuilder sequenceBuffer; 1374 if ((Character.getType(c) == Character.NON_SPACING_MARK) 1375 && (lastCodePoint != -1)) { 1376 sequenceBuffer = new StringBuilder(); 1377 sequenceBuffer.appendCodePoint(lastCodePoint); 1378 sequenceBuffer.appendCodePoint(c); 1379 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1380 i += Character.charCount(c); 1381 if (i >= patternLength) 1382 break; 1383 c = normalizedPattern.codePointAt(i); 1384 sequenceBuffer.appendCodePoint(c); 1385 } 1386 String ea = produceEquivalentAlternation( 1387 sequenceBuffer.toString()); 1388 newPattern.setLength(newPattern.length()-Character.charCount(lastCodePoint)); 1389 newPattern.append("(?:").append(ea).append(")"); 1390 } else if (c == '[' && lastCodePoint != '\\') { 1391 i = normalizeCharClass(newPattern, i); 1392 } else { 1393 newPattern.appendCodePoint(c); 1394 } 1395 lastCodePoint = c; 1396 i += Character.charCount(c); 1397 } 1398 normalizedPattern = newPattern.toString(); 1399 } 1400 1401 /** 1402 * Complete the character class being parsed and add a set 1403 * of alternations to it that will match the canonical equivalences 1404 * of the characters within the class. 1405 */ 1406 private int normalizeCharClass(StringBuilder newPattern, int i) { 1407 StringBuilder charClass = new StringBuilder(); 1408 StringBuilder eq = null; 1409 int lastCodePoint = -1; 1410 String result; 1411 1412 i++; 1413 charClass.append("["); 1414 while(true) { 1415 int c = normalizedPattern.codePointAt(i); 1416 StringBuilder sequenceBuffer; 1417 1418 if (c == ']' && lastCodePoint != '\\') { 1419 charClass.append((char)c); 1420 break; 1421 } else if (Character.getType(c) == Character.NON_SPACING_MARK) { 1422 sequenceBuffer = new StringBuilder(); 1423 sequenceBuffer.appendCodePoint(lastCodePoint); 1424 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1425 sequenceBuffer.appendCodePoint(c); 1426 i += Character.charCount(c); 1427 if (i >= normalizedPattern.length()) 1428 break; 1429 c = normalizedPattern.codePointAt(i); 1430 } 1431 String ea = produceEquivalentAlternation( 1432 sequenceBuffer.toString()); 1433 1434 charClass.setLength(charClass.length()-Character.charCount(lastCodePoint)); 1435 if (eq == null) 1436 eq = new StringBuilder(); 1437 eq.append('|'); 1438 eq.append(ea); 1439 } else { 1440 charClass.appendCodePoint(c); 1441 i++; 1442 } 1443 if (i == normalizedPattern.length()) 1444 throw error("Unclosed character class"); 1445 lastCodePoint = c; 1446 } 1447 1448 if (eq != null) { 1449 result = "(?:"+charClass.toString()+eq.toString()+")"; 1450 } else { 1451 result = charClass.toString(); 1452 } 1453 1454 newPattern.append(result); 1455 return i; 1456 } 1457 1458 /** 1459 * Given a specific sequence composed of a regular character and 1460 * combining marks that follow it, produce the alternation that will 1461 * match all canonical equivalences of that sequence. 1462 */ 1463 private String produceEquivalentAlternation(String source) { 1464 int len = countChars(source, 0, 1); 1465 if (source.length() == len) 1466 // source has one character. 1467 return source; 1468 1469 String base = source.substring(0,len); 1470 String combiningMarks = source.substring(len); 1471 1472 String[] perms = producePermutations(combiningMarks); 1473 StringBuilder result = new StringBuilder(source); 1474 1475 // Add combined permutations 1476 for(int x=0; x<perms.length; x++) { 1477 String next = base + perms[x]; 1478 if (x>0) 1479 result.append("|"+next); 1480 next = composeOneStep(next); 1481 if (next != null) 1482 result.append("|"+produceEquivalentAlternation(next)); 1483 } 1484 return result.toString(); 1485 } 1486 1487 /** 1488 * Returns an array of strings that have all the possible 1489 * permutations of the characters in the input string. 1490 * This is used to get a list of all possible orderings 1491 * of a set of combining marks. Note that some of the permutations 1492 * are invalid because of combining class collisions, and these 1493 * possibilities must be removed because they are not canonically 1494 * equivalent. 1495 */ 1496 private String[] producePermutations(String input) { 1497 if (input.length() == countChars(input, 0, 1)) 1498 return new String[] {input}; 1499 1500 if (input.length() == countChars(input, 0, 2)) { 1501 int c0 = Character.codePointAt(input, 0); 1502 int c1 = Character.codePointAt(input, Character.charCount(c0)); 1503 if (getClass(c1) == getClass(c0)) { 1504 return new String[] {input}; 1505 } 1506 String[] result = new String[2]; 1507 result[0] = input; 1508 StringBuilder sb = new StringBuilder(2); 1509 sb.appendCodePoint(c1); 1510 sb.appendCodePoint(c0); 1511 result[1] = sb.toString(); 1512 return result; 1513 } 1514 1515 int length = 1; 1516 int nCodePoints = countCodePoints(input); 1517 for(int x=1; x<nCodePoints; x++) 1518 length = length * (x+1); 1519 1520 String[] temp = new String[length]; 1521 1522 int combClass[] = new int[nCodePoints]; 1523 for(int x=0, i=0; x<nCodePoints; x++) { 1524 int c = Character.codePointAt(input, i); 1525 combClass[x] = getClass(c); 1526 i += Character.charCount(c); 1527 } 1528 1529 // For each char, take it out and add the permutations 1530 // of the remaining chars 1531 int index = 0; 1532 int len; 1533 // offset maintains the index in code units. 1534 loop: for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) { 1535 len = countChars(input, offset, 1); 1536 boolean skip = false; 1537 for(int y=x-1; y>=0; y--) { 1538 if (combClass[y] == combClass[x]) { 1539 continue loop; 1540 } 1541 } 1542 StringBuilder sb = new StringBuilder(input); 1543 String otherChars = sb.delete(offset, offset+len).toString(); 1544 String[] subResult = producePermutations(otherChars); 1545 1546 String prefix = input.substring(offset, offset+len); 1547 for(int y=0; y<subResult.length; y++) 1548 temp[index++] = prefix + subResult[y]; 1549 } 1550 String[] result = new String[index]; 1551 for (int x=0; x<index; x++) 1552 result[x] = temp[x]; 1553 return result; 1554 } 1555 1556 private int getClass(int c) { 1557 return sun.text.Normalizer.getCombiningClass(c); 1558 } 1559 1560 /** 1561 * Attempts to compose input by combining the first character 1562 * with the first combining mark following it. Returns a String 1563 * that is the composition of the leading character with its first 1564 * combining mark followed by the remaining combining marks. Returns 1565 * null if the first two characters cannot be further composed. 1566 */ 1567 private String composeOneStep(String input) { 1568 int len = countChars(input, 0, 2); 1569 String firstTwoCharacters = input.substring(0, len); 1570 String result = Normalizer.normalize(firstTwoCharacters, Normalizer.Form.NFC); 1571 1572 if (result.equals(firstTwoCharacters)) 1573 return null; 1574 else { 1575 String remainder = input.substring(len); 1576 return result + remainder; 1577 } 1578 } 1579 1580 /** 1581 * Preprocess any \Q...\E sequences in `temp', meta-quoting them. 1582 * See the description of `quotemeta' in perlfunc(1). 1583 */ 1584 private void RemoveQEQuoting() { 1585 final int pLen = patternLength; 1586 int i = 0; 1587 while (i < pLen-1) { 1588 if (temp[i] != '\\') 1589 i += 1; 1590 else if (temp[i + 1] != 'Q') 1591 i += 2; 1592 else 1593 break; 1594 } 1595 if (i >= pLen - 1) // No \Q sequence found 1596 return; 1597 int j = i; 1598 i += 2; 1599 int[] newtemp = new int[j + 3*(pLen-i) + 2]; 1600 System.arraycopy(temp, 0, newtemp, 0, j); 1601 1602 boolean inQuote = true; 1603 boolean beginQuote = true; 1604 while (i < pLen) { 1605 int c = temp[i++]; 1606 if (!ASCII.isAscii(c) || ASCII.isAlpha(c)) { 1607 newtemp[j++] = c; 1608 } else if (ASCII.isDigit(c)) { 1609 if (beginQuote) { 1610 /* 1611 * A unicode escape \[0xu] could be before this quote, 1612 * and we don't want this numeric char to processed as 1613 * part of the escape. 1614 */ 1615 newtemp[j++] = '\\'; 1616 newtemp[j++] = 'x'; 1617 newtemp[j++] = '3'; 1618 } 1619 newtemp[j++] = c; 1620 } else if (c != '\\') { 1621 if (inQuote) newtemp[j++] = '\\'; 1622 newtemp[j++] = c; 1623 } else if (inQuote) { 1624 if (temp[i] == 'E') { 1625 i++; 1626 inQuote = false; 1627 } else { 1628 newtemp[j++] = '\\'; 1629 newtemp[j++] = '\\'; 1630 } 1631 } else { 1632 if (temp[i] == 'Q') { 1633 i++; 1634 inQuote = true; 1635 beginQuote = true; 1636 continue; 1637 } else { 1638 newtemp[j++] = c; 1639 if (i != pLen) 1640 newtemp[j++] = temp[i++]; 1641 } 1642 } 1643 1644 beginQuote = false; 1645 } 1646 1647 patternLength = j; 1648 temp = Arrays.copyOf(newtemp, j + 2); // double zero termination 1649 } 1650 1651 /** 1652 * Copies regular expression to an int array and invokes the parsing 1653 * of the expression which will create the object tree. 1654 */ 1655 private void compile() { 1656 // Handle canonical equivalences 1657 if (has(CANON_EQ) && !has(LITERAL)) { 1658 normalize(); 1659 } else { 1660 normalizedPattern = pattern; 1661 } 1662 patternLength = normalizedPattern.length(); 1663 1664 // Copy pattern to int array for convenience 1665 // Use double zero to terminate pattern 1666 temp = new int[patternLength + 2]; 1667 1668 hasSupplementary = false; 1669 int c, count = 0; 1670 // Convert all chars into code points 1671 for (int x = 0; x < patternLength; x += Character.charCount(c)) { 1672 c = normalizedPattern.codePointAt(x); 1673 if (isSupplementary(c)) { 1674 hasSupplementary = true; 1675 } 1676 temp[count++] = c; 1677 } 1678 1679 patternLength = count; // patternLength now in code points 1680 1681 if (! has(LITERAL)) 1682 RemoveQEQuoting(); 1683 1684 // Allocate all temporary objects here. 1685 buffer = new int[32]; 1686 groupNodes = new GroupHead[10]; 1687 namedGroups = null; 1688 1689 if (has(LITERAL)) { 1690 // Literal pattern handling 1691 matchRoot = newSlice(temp, patternLength, hasSupplementary); 1692 matchRoot.next = lastAccept; 1693 } else { 1694 // Start recursive descent parsing 1695 matchRoot = expr(lastAccept); 1696 // Check extra pattern characters 1697 if (patternLength != cursor) { 1698 if (peek() == ')') { 1699 throw error("Unmatched closing ')'"); 1700 } else { 1701 throw error("Unexpected internal error"); 1702 } 1703 } 1704 } 1705 1706 // Peephole optimization 1707 if (matchRoot instanceof Slice) { 1708 root = BnM.optimize(matchRoot); 1709 if (root == matchRoot) { 1710 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1711 } 1712 } else if (matchRoot instanceof Begin || matchRoot instanceof First) { 1713 root = matchRoot; 1714 } else { 1715 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1716 } 1717 1718 // Release temporary storage 1719 temp = null; 1720 buffer = null; 1721 groupNodes = null; 1722 patternLength = 0; 1723 compiled = true; 1724 } 1725 1726 Map<String, Integer> namedGroups() { 1727 if (namedGroups == null) 1728 namedGroups = new HashMap<>(2); 1729 return namedGroups; 1730 } 1731 1732 /** 1733 * Used to print out a subtree of the Pattern to help with debugging. 1734 */ 1735 private static void printObjectTree(Node node) { 1736 while(node != null) { 1737 if (node instanceof Prolog) { 1738 System.out.println(node); 1739 printObjectTree(((Prolog)node).loop); 1740 System.out.println("**** end contents prolog loop"); 1741 } else if (node instanceof Loop) { 1742 System.out.println(node); 1743 printObjectTree(((Loop)node).body); 1744 System.out.println("**** end contents Loop body"); 1745 } else if (node instanceof Curly) { 1746 System.out.println(node); 1747 printObjectTree(((Curly)node).atom); 1748 System.out.println("**** end contents Curly body"); 1749 } else if (node instanceof GroupCurly) { 1750 System.out.println(node); 1751 printObjectTree(((GroupCurly)node).atom); 1752 System.out.println("**** end contents GroupCurly body"); 1753 } else if (node instanceof GroupTail) { 1754 System.out.println(node); 1755 System.out.println("Tail next is "+node.next); 1756 return; 1757 } else { 1758 System.out.println(node); 1759 } 1760 node = node.next; 1761 if (node != null) 1762 System.out.println("->next:"); 1763 if (node == Pattern.accept) { 1764 System.out.println("Accept Node"); 1765 node = null; 1766 } 1767 } 1768 } 1769 1770 /** 1771 * Used to accumulate information about a subtree of the object graph 1772 * so that optimizations can be applied to the subtree. 1773 */ 1774 static final class TreeInfo { 1775 int minLength; 1776 int maxLength; 1777 boolean maxValid; 1778 boolean deterministic; 1779 1780 TreeInfo() { 1781 reset(); 1782 } 1783 void reset() { 1784 minLength = 0; 1785 maxLength = 0; 1786 maxValid = true; 1787 deterministic = true; 1788 } 1789 } 1790 1791 /* 1792 * The following private methods are mainly used to improve the 1793 * readability of the code. In order to let the Java compiler easily 1794 * inline them, we should not put many assertions or error checks in them. 1795 */ 1796 1797 /** 1798 * Indicates whether a particular flag is set or not. 1799 */ 1800 private boolean has(int f) { 1801 return (flags & f) != 0; 1802 } 1803 1804 /** 1805 * Match next character, signal error if failed. 1806 */ 1807 private void accept(int ch, String s) { 1808 int testChar = temp[cursor++]; 1809 if (has(COMMENTS)) 1810 testChar = parsePastWhitespace(testChar); 1811 if (ch != testChar) { 1812 throw error(s); 1813 } 1814 } 1815 1816 /** 1817 * Mark the end of pattern with a specific character. 1818 */ 1819 private void mark(int c) { 1820 temp[patternLength] = c; 1821 } 1822 1823 /** 1824 * Peek the next character, and do not advance the cursor. 1825 */ 1826 private int peek() { 1827 int ch = temp[cursor]; 1828 if (has(COMMENTS)) 1829 ch = peekPastWhitespace(ch); 1830 return ch; 1831 } 1832 1833 /** 1834 * Read the next character, and advance the cursor by one. 1835 */ 1836 private int read() { 1837 int ch = temp[cursor++]; 1838 if (has(COMMENTS)) 1839 ch = parsePastWhitespace(ch); 1840 return ch; 1841 } 1842 1843 /** 1844 * Read the next character, and advance the cursor by one, 1845 * ignoring the COMMENTS setting 1846 */ 1847 private int readEscaped() { 1848 int ch = temp[cursor++]; 1849 return ch; 1850 } 1851 1852 /** 1853 * Advance the cursor by one, and peek the next character. 1854 */ 1855 private int next() { 1856 int ch = temp[++cursor]; 1857 if (has(COMMENTS)) 1858 ch = peekPastWhitespace(ch); 1859 return ch; 1860 } 1861 1862 /** 1863 * Advance the cursor by one, and peek the next character, 1864 * ignoring the COMMENTS setting 1865 */ 1866 private int nextEscaped() { 1867 int ch = temp[++cursor]; 1868 return ch; 1869 } 1870 1871 /** 1872 * If in xmode peek past whitespace and comments. 1873 */ 1874 private int peekPastWhitespace(int ch) { 1875 while (ASCII.isSpace(ch) || ch == '#') { 1876 while (ASCII.isSpace(ch)) 1877 ch = temp[++cursor]; 1878 if (ch == '#') { 1879 ch = peekPastLine(); 1880 } 1881 } 1882 return ch; 1883 } 1884 1885 /** 1886 * If in xmode parse past whitespace and comments. 1887 */ 1888 private int parsePastWhitespace(int ch) { 1889 while (ASCII.isSpace(ch) || ch == '#') { 1890 while (ASCII.isSpace(ch)) 1891 ch = temp[cursor++]; 1892 if (ch == '#') 1893 ch = parsePastLine(); 1894 } 1895 return ch; 1896 } 1897 1898 /** 1899 * xmode parse past comment to end of line. 1900 */ 1901 private int parsePastLine() { 1902 int ch = temp[cursor++]; 1903 while (ch != 0 && !isLineSeparator(ch)) 1904 ch = temp[cursor++]; 1905 return ch; 1906 } 1907 1908 /** 1909 * xmode peek past comment to end of line. 1910 */ 1911 private int peekPastLine() { 1912 int ch = temp[++cursor]; 1913 while (ch != 0 && !isLineSeparator(ch)) 1914 ch = temp[++cursor]; 1915 return ch; 1916 } 1917 1918 /** 1919 * Determines if character is a line separator in the current mode 1920 */ 1921 private boolean isLineSeparator(int ch) { 1922 if (has(UNIX_LINES)) { 1923 return ch == '\n'; 1924 } else { 1925 return (ch == '\n' || 1926 ch == '\r' || 1927 (ch|1) == '\u2029' || 1928 ch == '\u0085'); 1929 } 1930 } 1931 1932 /** 1933 * Read the character after the next one, and advance the cursor by two. 1934 */ 1935 private int skip() { 1936 int i = cursor; 1937 int ch = temp[i+1]; 1938 cursor = i + 2; 1939 return ch; 1940 } 1941 1942 /** 1943 * Unread one next character, and retreat cursor by one. 1944 */ 1945 private void unread() { 1946 cursor--; 1947 } 1948 1949 /** 1950 * Internal method used for handling all syntax errors. The pattern is 1951 * displayed with a pointer to aid in locating the syntax error. 1952 */ 1953 private PatternSyntaxException error(String s) { 1954 return new PatternSyntaxException(s, normalizedPattern, cursor - 1); 1955 } 1956 1957 /** 1958 * Determines if there is any supplementary character or unpaired 1959 * surrogate in the specified range. 1960 */ 1961 private boolean findSupplementary(int start, int end) { 1962 for (int i = start; i < end; i++) { 1963 if (isSupplementary(temp[i])) 1964 return true; 1965 } 1966 return false; 1967 } 1968 1969 /** 1970 * Determines if the specified code point is a supplementary 1971 * character or unpaired surrogate. 1972 */ 1973 private static final boolean isSupplementary(int ch) { 1974 return ch >= Character.MIN_SUPPLEMENTARY_CODE_POINT || 1975 Character.isSurrogate((char)ch); 1976 } 1977 1978 /** 1979 * The following methods handle the main parsing. They are sorted 1980 * according to their precedence order, the lowest one first. 1981 */ 1982 1983 /** 1984 * The expression is parsed with branch nodes added for alternations. 1985 * This may be called recursively to parse sub expressions that may 1986 * contain alternations. 1987 */ 1988 private Node expr(Node end) { 1989 Node prev = null; 1990 Node firstTail = null; 1991 Branch branch = null; 1992 Node branchConn = null; 1993 1994 for (;;) { 1995 Node node = sequence(end); 1996 Node nodeTail = root; //double return 1997 if (prev == null) { 1998 prev = node; 1999 firstTail = nodeTail; 2000 } else { 2001 // Branch 2002 if (branchConn == null) { 2003 branchConn = new BranchConn(); 2004 branchConn.next = end; 2005 } 2006 if (node == end) { 2007 // if the node returned from sequence() is "end" 2008 // we have an empty expr, set a null atom into 2009 // the branch to indicate to go "next" directly. 2010 node = null; 2011 } else { 2012 // the "tail.next" of each atom goes to branchConn 2013 nodeTail.next = branchConn; 2014 } 2015 if (prev == branch) { 2016 branch.add(node); 2017 } else { 2018 if (prev == end) { 2019 prev = null; 2020 } else { 2021 // replace the "end" with "branchConn" at its tail.next 2022 // when put the "prev" into the branch as the first atom. 2023 firstTail.next = branchConn; 2024 } 2025 prev = branch = new Branch(prev, node, branchConn); 2026 } 2027 } 2028 if (peek() != '|') { 2029 return prev; 2030 } 2031 next(); 2032 } 2033 } 2034 2035 @SuppressWarnings("fallthrough") 2036 /** 2037 * Parsing of sequences between alternations. 2038 */ 2039 private Node sequence(Node end) { 2040 Node head = null; 2041 Node tail = null; 2042 Node node = null; 2043 LOOP: 2044 for (;;) { 2045 int ch = peek(); 2046 switch (ch) { 2047 case '(': 2048 // Because group handles its own closure, 2049 // we need to treat it differently 2050 node = group0(); 2051 // Check for comment or flag group 2052 if (node == null) 2053 continue; 2054 if (head == null) 2055 head = node; 2056 else 2057 tail.next = node; 2058 // Double return: Tail was returned in root 2059 tail = root; 2060 continue; 2061 case '[': 2062 node = clazz(true); 2063 break; 2064 case '\\': 2065 ch = nextEscaped(); 2066 if (ch == 'p' || ch == 'P') { 2067 boolean oneLetter = true; 2068 boolean comp = (ch == 'P'); 2069 ch = next(); // Consume { if present 2070 if (ch != '{') { 2071 unread(); 2072 } else { 2073 oneLetter = false; 2074 } 2075 node = family(oneLetter, comp); 2076 } else { 2077 unread(); 2078 node = atom(); 2079 } 2080 break; 2081 case '^': 2082 next(); 2083 if (has(MULTILINE)) { 2084 if (has(UNIX_LINES)) 2085 node = new UnixCaret(); 2086 else 2087 node = new Caret(); 2088 } else { 2089 node = new Begin(); 2090 } 2091 break; 2092 case '$': 2093 next(); 2094 if (has(UNIX_LINES)) 2095 node = new UnixDollar(has(MULTILINE)); 2096 else 2097 node = new Dollar(has(MULTILINE)); 2098 break; 2099 case '.': 2100 next(); 2101 if (has(DOTALL)) { 2102 node = new All(); 2103 } else { 2104 if (has(UNIX_LINES)) 2105 node = new UnixDot(); 2106 else { 2107 node = new Dot(); 2108 } 2109 } 2110 break; 2111 case '|': 2112 case ')': 2113 break LOOP; 2114 case ']': // Now interpreting dangling ] and } as literals 2115 case '}': 2116 node = atom(); 2117 break; 2118 case '?': 2119 case '*': 2120 case '+': 2121 next(); 2122 throw error("Dangling meta character '" + ((char)ch) + "'"); 2123 case 0: 2124 if (cursor >= patternLength) { 2125 break LOOP; 2126 } 2127 // Fall through 2128 default: 2129 node = atom(); 2130 break; 2131 } 2132 2133 node = closure(node); 2134 2135 if (head == null) { 2136 head = tail = node; 2137 } else { 2138 tail.next = node; 2139 tail = node; 2140 } 2141 } 2142 if (head == null) { 2143 return end; 2144 } 2145 tail.next = end; 2146 root = tail; //double return 2147 return head; 2148 } 2149 2150 @SuppressWarnings("fallthrough") 2151 /** 2152 * Parse and add a new Single or Slice. 2153 */ 2154 private Node atom() { 2155 int first = 0; 2156 int prev = -1; 2157 boolean hasSupplementary = false; 2158 int ch = peek(); 2159 for (;;) { 2160 switch (ch) { 2161 case '*': 2162 case '+': 2163 case '?': 2164 case '{': 2165 if (first > 1) { 2166 cursor = prev; // Unwind one character 2167 first--; 2168 } 2169 break; 2170 case '$': 2171 case '.': 2172 case '^': 2173 case '(': 2174 case '[': 2175 case '|': 2176 case ')': 2177 break; 2178 case '\\': 2179 ch = nextEscaped(); 2180 if (ch == 'p' || ch == 'P') { // Property 2181 if (first > 0) { // Slice is waiting; handle it first 2182 unread(); 2183 break; 2184 } else { // No slice; just return the family node 2185 boolean comp = (ch == 'P'); 2186 boolean oneLetter = true; 2187 ch = next(); // Consume { if present 2188 if (ch != '{') 2189 unread(); 2190 else 2191 oneLetter = false; 2192 return family(oneLetter, comp); 2193 } 2194 } 2195 unread(); 2196 prev = cursor; 2197 ch = escape(false, first == 0, false); 2198 if (ch >= 0) { 2199 append(ch, first); 2200 first++; 2201 if (isSupplementary(ch)) { 2202 hasSupplementary = true; 2203 } 2204 ch = peek(); 2205 continue; 2206 } else if (first == 0) { 2207 return root; 2208 } 2209 // Unwind meta escape sequence 2210 cursor = prev; 2211 break; 2212 case 0: 2213 if (cursor >= patternLength) { 2214 break; 2215 } 2216 // Fall through 2217 default: 2218 prev = cursor; 2219 append(ch, first); 2220 first++; 2221 if (isSupplementary(ch)) { 2222 hasSupplementary = true; 2223 } 2224 ch = next(); 2225 continue; 2226 } 2227 break; 2228 } 2229 if (first == 1) { 2230 return newSingle(buffer[0]); 2231 } else { 2232 return newSlice(buffer, first, hasSupplementary); 2233 } 2234 } 2235 2236 private void append(int ch, int len) { 2237 if (len >= buffer.length) { 2238 int[] tmp = new int[len+len]; 2239 System.arraycopy(buffer, 0, tmp, 0, len); 2240 buffer = tmp; 2241 } 2242 buffer[len] = ch; 2243 } 2244 2245 /** 2246 * Parses a backref greedily, taking as many numbers as it 2247 * can. The first digit is always treated as a backref, but 2248 * multi digit numbers are only treated as a backref if at 2249 * least that many backrefs exist at this point in the regex. 2250 */ 2251 private Node ref(int refNum) { 2252 boolean done = false; 2253 while(!done) { 2254 int ch = peek(); 2255 switch(ch) { 2256 case '0': 2257 case '1': 2258 case '2': 2259 case '3': 2260 case '4': 2261 case '5': 2262 case '6': 2263 case '7': 2264 case '8': 2265 case '9': 2266 int newRefNum = (refNum * 10) + (ch - '0'); 2267 // Add another number if it doesn't make a group 2268 // that doesn't exist 2269 if (capturingGroupCount - 1 < newRefNum) { 2270 done = true; 2271 break; 2272 } 2273 refNum = newRefNum; 2274 read(); 2275 break; 2276 default: 2277 done = true; 2278 break; 2279 } 2280 } 2281 if (has(CASE_INSENSITIVE)) 2282 return new CIBackRef(refNum, has(UNICODE_CASE)); 2283 else 2284 return new BackRef(refNum); 2285 } 2286 2287 /** 2288 * Parses an escape sequence to determine the actual value that needs 2289 * to be matched. 2290 * If -1 is returned and create was true a new object was added to the tree 2291 * to handle the escape sequence. 2292 * If the returned value is greater than zero, it is the value that 2293 * matches the escape sequence. 2294 */ 2295 private int escape(boolean inclass, boolean create, boolean isrange) { 2296 int ch = skip(); 2297 switch (ch) { 2298 case '0': 2299 return o(); 2300 case '1': 2301 case '2': 2302 case '3': 2303 case '4': 2304 case '5': 2305 case '6': 2306 case '7': 2307 case '8': 2308 case '9': 2309 if (inclass) break; 2310 if (create) { 2311 root = ref((ch - '0')); 2312 } 2313 return -1; 2314 case 'A': 2315 if (inclass) break; 2316 if (create) root = new Begin(); 2317 return -1; 2318 case 'B': 2319 if (inclass) break; 2320 if (create) root = new Bound(Bound.NONE, has(UNICODE_CHARACTER_CLASS)); 2321 return -1; 2322 case 'C': 2323 break; 2324 case 'D': 2325 if (create) root = has(UNICODE_CHARACTER_CLASS) 2326 ? new Utype(UnicodeProp.DIGIT).complement() 2327 : new Ctype(ASCII.DIGIT).complement(); 2328 return -1; 2329 case 'E': 2330 case 'F': 2331 break; 2332 case 'G': 2333 if (inclass) break; 2334 if (create) root = new LastMatch(); 2335 return -1; 2336 case 'H': 2337 if (create) root = new HorizWS().complement(); 2338 return -1; 2339 case 'I': 2340 case 'J': 2341 case 'K': 2342 case 'L': 2343 case 'M': 2344 case 'N': 2345 case 'O': 2346 case 'P': 2347 case 'Q': 2348 break; 2349 case 'R': 2350 if (inclass) break; 2351 if (create) root = new LineEnding(); 2352 return -1; 2353 case 'S': 2354 if (create) root = has(UNICODE_CHARACTER_CLASS) 2355 ? new Utype(UnicodeProp.WHITE_SPACE).complement() 2356 : new Ctype(ASCII.SPACE).complement(); 2357 return -1; 2358 case 'T': 2359 case 'U': 2360 break; 2361 case 'V': 2362 if (create) root = new VertWS().complement(); 2363 return -1; 2364 case 'W': 2365 if (create) root = has(UNICODE_CHARACTER_CLASS) 2366 ? new Utype(UnicodeProp.WORD).complement() 2367 : new Ctype(ASCII.WORD).complement(); 2368 return -1; 2369 case 'X': 2370 case 'Y': 2371 break; 2372 case 'Z': 2373 if (inclass) break; 2374 if (create) { 2375 if (has(UNIX_LINES)) 2376 root = new UnixDollar(false); 2377 else 2378 root = new Dollar(false); 2379 } 2380 return -1; 2381 case 'a': 2382 return '\007'; 2383 case 'b': 2384 if (inclass) break; 2385 if (create) root = new Bound(Bound.BOTH, has(UNICODE_CHARACTER_CLASS)); 2386 return -1; 2387 case 'c': 2388 return c(); 2389 case 'd': 2390 if (create) root = has(UNICODE_CHARACTER_CLASS) 2391 ? new Utype(UnicodeProp.DIGIT) 2392 : new Ctype(ASCII.DIGIT); 2393 return -1; 2394 case 'e': 2395 return '\033'; 2396 case 'f': 2397 return '\f'; 2398 case 'g': 2399 break; 2400 case 'h': 2401 if (create) root = new HorizWS(); 2402 return -1; 2403 case 'i': 2404 case 'j': 2405 break; 2406 case 'k': 2407 if (inclass) 2408 break; 2409 if (read() != '<') 2410 throw error("\\k is not followed by '<' for named capturing group"); 2411 String name = groupname(read()); 2412 if (!namedGroups().containsKey(name)) 2413 throw error("(named capturing group <"+ name+"> does not exit"); 2414 if (create) { 2415 if (has(CASE_INSENSITIVE)) 2416 root = new CIBackRef(namedGroups().get(name), has(UNICODE_CASE)); 2417 else 2418 root = new BackRef(namedGroups().get(name)); 2419 } 2420 return -1; 2421 case 'l': 2422 case 'm': 2423 break; 2424 case 'n': 2425 return '\n'; 2426 case 'o': 2427 case 'p': 2428 case 'q': 2429 break; 2430 case 'r': 2431 return '\r'; 2432 case 's': 2433 if (create) root = has(UNICODE_CHARACTER_CLASS) 2434 ? new Utype(UnicodeProp.WHITE_SPACE) 2435 : new Ctype(ASCII.SPACE); 2436 return -1; 2437 case 't': 2438 return '\t'; 2439 case 'u': 2440 return u(); 2441 case 'v': 2442 // '\v' was implemented as VT/0x0B in releases < 1.8 (though 2443 // undocumented). In JDK8 '\v' is specified as a predefined 2444 // character class for all vertical whitespace characters. 2445 // So [-1, root=VertWS node] pair is returned (instead of a 2446 // single 0x0B). This breaks the range if '\v' is used as 2447 // the start or end value, such as [\v-...] or [...-\v], in 2448 // which a single definite value (0x0B) is expected. For 2449 // compatiblity concern '\013'/0x0B is returned if isrange. 2450 if (isrange) 2451 return '\013'; 2452 if (create) root = new VertWS(); 2453 return -1; 2454 case 'w': 2455 if (create) root = has(UNICODE_CHARACTER_CLASS) 2456 ? new Utype(UnicodeProp.WORD) 2457 : new Ctype(ASCII.WORD); 2458 return -1; 2459 case 'x': 2460 return x(); 2461 case 'y': 2462 break; 2463 case 'z': 2464 if (inclass) break; 2465 if (create) root = new End(); 2466 return -1; 2467 default: 2468 return ch; 2469 } 2470 throw error("Illegal/unsupported escape sequence"); 2471 } 2472 2473 /** 2474 * Parse a character class, and return the node that matches it. 2475 * 2476 * Consumes a ] on the way out if consume is true. Usually consume 2477 * is true except for the case of [abc&&def] where def is a separate 2478 * right hand node with "understood" brackets. 2479 */ 2480 private CharProperty clazz(boolean consume) { 2481 CharProperty prev = null; 2482 CharProperty node = null; 2483 BitClass bits = new BitClass(); 2484 boolean include = true; 2485 boolean firstInClass = true; 2486 int ch = next(); 2487 for (;;) { 2488 switch (ch) { 2489 case '^': 2490 // Negates if first char in a class, otherwise literal 2491 if (firstInClass) { 2492 if (temp[cursor-1] != '[') 2493 break; 2494 ch = next(); 2495 include = !include; 2496 continue; 2497 } else { 2498 // ^ not first in class, treat as literal 2499 break; 2500 } 2501 case '[': 2502 firstInClass = false; 2503 node = clazz(true); 2504 if (prev == null) 2505 prev = node; 2506 else 2507 prev = union(prev, node); 2508 ch = peek(); 2509 continue; 2510 case '&': 2511 firstInClass = false; 2512 ch = next(); 2513 if (ch == '&') { 2514 ch = next(); 2515 CharProperty rightNode = null; 2516 while (ch != ']' && ch != '&') { 2517 if (ch == '[') { 2518 if (rightNode == null) 2519 rightNode = clazz(true); 2520 else 2521 rightNode = union(rightNode, clazz(true)); 2522 } else { // abc&&def 2523 unread(); 2524 rightNode = clazz(false); 2525 } 2526 ch = peek(); 2527 } 2528 if (rightNode != null) 2529 node = rightNode; 2530 if (prev == null) { 2531 if (rightNode == null) 2532 throw error("Bad class syntax"); 2533 else 2534 prev = rightNode; 2535 } else { 2536 prev = intersection(prev, node); 2537 } 2538 } else { 2539 // treat as a literal & 2540 unread(); 2541 break; 2542 } 2543 continue; 2544 case 0: 2545 firstInClass = false; 2546 if (cursor >= patternLength) 2547 throw error("Unclosed character class"); 2548 break; 2549 case ']': 2550 firstInClass = false; 2551 if (prev != null) { 2552 if (consume) 2553 next(); 2554 return prev; 2555 } 2556 break; 2557 default: 2558 firstInClass = false; 2559 break; 2560 } 2561 node = range(bits); 2562 if (include) { 2563 if (prev == null) { 2564 prev = node; 2565 } else { 2566 if (prev != node) 2567 prev = union(prev, node); 2568 } 2569 } else { 2570 if (prev == null) { 2571 prev = node.complement(); 2572 } else { 2573 if (prev != node) 2574 prev = setDifference(prev, node); 2575 } 2576 } 2577 ch = peek(); 2578 } 2579 } 2580 2581 private CharProperty bitsOrSingle(BitClass bits, int ch) { 2582 /* Bits can only handle codepoints in [u+0000-u+00ff] range. 2583 Use "single" node instead of bits when dealing with unicode 2584 case folding for codepoints listed below. 2585 (1)Uppercase out of range: u+00ff, u+00b5 2586 toUpperCase(u+00ff) -> u+0178 2587 toUpperCase(u+00b5) -> u+039c 2588 (2)LatinSmallLetterLongS u+17f 2589 toUpperCase(u+017f) -> u+0053 2590 (3)LatinSmallLetterDotlessI u+131 2591 toUpperCase(u+0131) -> u+0049 2592 (4)LatinCapitalLetterIWithDotAbove u+0130 2593 toLowerCase(u+0130) -> u+0069 2594 (5)KelvinSign u+212a 2595 toLowerCase(u+212a) ==> u+006B 2596 (6)AngstromSign u+212b 2597 toLowerCase(u+212b) ==> u+00e5 2598 */ 2599 int d; 2600 if (ch < 256 && 2601 !(has(CASE_INSENSITIVE) && has(UNICODE_CASE) && 2602 (ch == 0xff || ch == 0xb5 || 2603 ch == 0x49 || ch == 0x69 || //I and i 2604 ch == 0x53 || ch == 0x73 || //S and s 2605 ch == 0x4b || ch == 0x6b || //K and k 2606 ch == 0xc5 || ch == 0xe5))) //A+ring 2607 return bits.add(ch, flags()); 2608 return newSingle(ch); 2609 } 2610 2611 /** 2612 * Parse a single character or a character range in a character class 2613 * and return its representative node. 2614 */ 2615 private CharProperty range(BitClass bits) { 2616 int ch = peek(); 2617 if (ch == '\\') { 2618 ch = nextEscaped(); 2619 if (ch == 'p' || ch == 'P') { // A property 2620 boolean comp = (ch == 'P'); 2621 boolean oneLetter = true; 2622 // Consume { if present 2623 ch = next(); 2624 if (ch != '{') 2625 unread(); 2626 else 2627 oneLetter = false; 2628 return family(oneLetter, comp); 2629 } else { // ordinary escape 2630 boolean isrange = temp[cursor+1] == '-'; 2631 unread(); 2632 ch = escape(true, true, isrange); 2633 if (ch == -1) 2634 return (CharProperty) root; 2635 } 2636 } else { 2637 next(); 2638 } 2639 if (ch >= 0) { 2640 if (peek() == '-') { 2641 int endRange = temp[cursor+1]; 2642 if (endRange == '[') { 2643 return bitsOrSingle(bits, ch); 2644 } 2645 if (endRange != ']') { 2646 next(); 2647 int m = peek(); 2648 if (m == '\\') { 2649 m = escape(true, false, true); 2650 } else { 2651 next(); 2652 } 2653 if (m < ch) { 2654 throw error("Illegal character range"); 2655 } 2656 if (has(CASE_INSENSITIVE)) 2657 return caseInsensitiveRangeFor(ch, m); 2658 else 2659 return rangeFor(ch, m); 2660 } 2661 } 2662 return bitsOrSingle(bits, ch); 2663 } 2664 throw error("Unexpected character '"+((char)ch)+"'"); 2665 } 2666 2667 /** 2668 * Parses a Unicode character family and returns its representative node. 2669 */ 2670 private CharProperty family(boolean singleLetter, 2671 boolean maybeComplement) 2672 { 2673 next(); 2674 String name; 2675 CharProperty node = null; 2676 2677 if (singleLetter) { 2678 int c = temp[cursor]; 2679 if (!Character.isSupplementaryCodePoint(c)) { 2680 name = String.valueOf((char)c); 2681 } else { 2682 name = new String(temp, cursor, 1); 2683 } 2684 read(); 2685 } else { 2686 int i = cursor; 2687 mark('}'); 2688 while(read() != '}') { 2689 } 2690 mark('\000'); 2691 int j = cursor; 2692 if (j > patternLength) 2693 throw error("Unclosed character family"); 2694 if (i + 1 >= j) 2695 throw error("Empty character family"); 2696 name = new String(temp, i, j-i-1); 2697 } 2698 2699 int i = name.indexOf('='); 2700 if (i != -1) { 2701 // property construct \p{name=value} 2702 String value = name.substring(i + 1); 2703 name = name.substring(0, i).toLowerCase(Locale.ENGLISH); 2704 if ("sc".equals(name) || "script".equals(name)) { 2705 node = unicodeScriptPropertyFor(value); 2706 } else if ("blk".equals(name) || "block".equals(name)) { 2707 node = unicodeBlockPropertyFor(value); 2708 } else if ("gc".equals(name) || "general_category".equals(name)) { 2709 node = charPropertyNodeFor(value); 2710 } else { 2711 throw error("Unknown Unicode property {name=<" + name + ">, " 2712 + "value=<" + value + ">}"); 2713 } 2714 } else { 2715 if (name.startsWith("In")) { 2716 // \p{inBlockName} 2717 node = unicodeBlockPropertyFor(name.substring(2)); 2718 } else if (name.startsWith("Is")) { 2719 // \p{isGeneralCategory} and \p{isScriptName} 2720 name = name.substring(2); 2721 UnicodeProp uprop = UnicodeProp.forName(name); 2722 if (uprop != null) 2723 node = new Utype(uprop); 2724 if (node == null) 2725 node = CharPropertyNames.charPropertyFor(name); 2726 if (node == null) 2727 node = unicodeScriptPropertyFor(name); 2728 } else { 2729 if (has(UNICODE_CHARACTER_CLASS)) { 2730 UnicodeProp uprop = UnicodeProp.forPOSIXName(name); 2731 if (uprop != null) 2732 node = new Utype(uprop); 2733 } 2734 if (node == null) 2735 node = charPropertyNodeFor(name); 2736 } 2737 } 2738 if (maybeComplement) { 2739 if (node instanceof Category || node instanceof Block) 2740 hasSupplementary = true; 2741 node = node.complement(); 2742 } 2743 return node; 2744 } 2745 2746 2747 /** 2748 * Returns a CharProperty matching all characters belong to 2749 * a UnicodeScript. 2750 */ 2751 private CharProperty unicodeScriptPropertyFor(String name) { 2752 final Character.UnicodeScript script; 2753 try { 2754 script = Character.UnicodeScript.forName(name); 2755 } catch (IllegalArgumentException iae) { 2756 throw error("Unknown character script name {" + name + "}"); 2757 } 2758 return new Script(script); 2759 } 2760 2761 /** 2762 * Returns a CharProperty matching all characters in a UnicodeBlock. 2763 */ 2764 private CharProperty unicodeBlockPropertyFor(String name) { 2765 final Character.UnicodeBlock block; 2766 try { 2767 block = Character.UnicodeBlock.forName(name); 2768 } catch (IllegalArgumentException iae) { 2769 throw error("Unknown character block name {" + name + "}"); 2770 } 2771 return new Block(block); 2772 } 2773 2774 /** 2775 * Returns a CharProperty matching all characters in a named property. 2776 */ 2777 private CharProperty charPropertyNodeFor(String name) { 2778 CharProperty p = CharPropertyNames.charPropertyFor(name); 2779 if (p == null) 2780 throw error("Unknown character property name {" + name + "}"); 2781 return p; 2782 } 2783 2784 /** 2785 * Parses and returns the name of a "named capturing group", the trailing 2786 * ">" is consumed after parsing. 2787 */ 2788 private String groupname(int ch) { 2789 StringBuilder sb = new StringBuilder(); 2790 sb.append(Character.toChars(ch)); 2791 while (ASCII.isLower(ch=read()) || ASCII.isUpper(ch) || 2792 ASCII.isDigit(ch)) { 2793 sb.append(Character.toChars(ch)); 2794 } 2795 if (sb.length() == 0) 2796 throw error("named capturing group has 0 length name"); 2797 if (ch != '>') 2798 throw error("named capturing group is missing trailing '>'"); 2799 return sb.toString(); 2800 } 2801 2802 /** 2803 * Parses a group and returns the head node of a set of nodes that process 2804 * the group. Sometimes a double return system is used where the tail is 2805 * returned in root. 2806 */ 2807 private Node group0() { 2808 boolean capturingGroup = false; 2809 Node head = null; 2810 Node tail = null; 2811 int save = flags; 2812 root = null; 2813 int ch = next(); 2814 if (ch == '?') { 2815 ch = skip(); 2816 switch (ch) { 2817 case ':': // (?:xxx) pure group 2818 head = createGroup(true); 2819 tail = root; 2820 head.next = expr(tail); 2821 break; 2822 case '=': // (?=xxx) and (?!xxx) lookahead 2823 case '!': 2824 head = createGroup(true); 2825 tail = root; 2826 head.next = expr(tail); 2827 if (ch == '=') { 2828 head = tail = new Pos(head); 2829 } else { 2830 head = tail = new Neg(head); 2831 } 2832 break; 2833 case '>': // (?>xxx) independent group 2834 head = createGroup(true); 2835 tail = root; 2836 head.next = expr(tail); 2837 head = tail = new Ques(head, INDEPENDENT); 2838 break; 2839 case '<': // (?<xxx) look behind 2840 ch = read(); 2841 if (ASCII.isLower(ch) || ASCII.isUpper(ch)) { 2842 // named captured group 2843 String name = groupname(ch); 2844 if (namedGroups().containsKey(name)) 2845 throw error("Named capturing group <" + name 2846 + "> is already defined"); 2847 capturingGroup = true; 2848 head = createGroup(false); 2849 tail = root; 2850 namedGroups().put(name, capturingGroupCount-1); 2851 head.next = expr(tail); 2852 break; 2853 } 2854 int start = cursor; 2855 head = createGroup(true); 2856 tail = root; 2857 head.next = expr(tail); 2858 tail.next = lookbehindEnd; 2859 TreeInfo info = new TreeInfo(); 2860 head.study(info); 2861 if (info.maxValid == false) { 2862 throw error("Look-behind group does not have " 2863 + "an obvious maximum length"); 2864 } 2865 boolean hasSupplementary = findSupplementary(start, patternLength); 2866 if (ch == '=') { 2867 head = tail = (hasSupplementary ? 2868 new BehindS(head, info.maxLength, 2869 info.minLength) : 2870 new Behind(head, info.maxLength, 2871 info.minLength)); 2872 } else if (ch == '!') { 2873 head = tail = (hasSupplementary ? 2874 new NotBehindS(head, info.maxLength, 2875 info.minLength) : 2876 new NotBehind(head, info.maxLength, 2877 info.minLength)); 2878 } else { 2879 throw error("Unknown look-behind group"); 2880 } 2881 break; 2882 case '$': 2883 case '@': 2884 throw error("Unknown group type"); 2885 default: // (?xxx:) inlined match flags 2886 unread(); 2887 addFlag(); 2888 ch = read(); 2889 if (ch == ')') { 2890 return null; // Inline modifier only 2891 } 2892 if (ch != ':') { 2893 throw error("Unknown inline modifier"); 2894 } 2895 head = createGroup(true); 2896 tail = root; 2897 head.next = expr(tail); 2898 break; 2899 } 2900 } else { // (xxx) a regular group 2901 capturingGroup = true; 2902 head = createGroup(false); 2903 tail = root; 2904 head.next = expr(tail); 2905 } 2906 2907 accept(')', "Unclosed group"); 2908 flags = save; 2909 2910 // Check for quantifiers 2911 Node node = closure(head); 2912 if (node == head) { // No closure 2913 root = tail; 2914 return node; // Dual return 2915 } 2916 if (head == tail) { // Zero length assertion 2917 root = node; 2918 return node; // Dual return 2919 } 2920 2921 if (node instanceof Ques) { 2922 Ques ques = (Ques) node; 2923 if (ques.type == POSSESSIVE) { 2924 root = node; 2925 return node; 2926 } 2927 tail.next = new BranchConn(); 2928 tail = tail.next; 2929 if (ques.type == GREEDY) { 2930 head = new Branch(head, null, tail); 2931 } else { // Reluctant quantifier 2932 head = new Branch(null, head, tail); 2933 } 2934 root = tail; 2935 return head; 2936 } else if (node instanceof Curly) { 2937 Curly curly = (Curly) node; 2938 if (curly.type == POSSESSIVE) { 2939 root = node; 2940 return node; 2941 } 2942 // Discover if the group is deterministic 2943 TreeInfo info = new TreeInfo(); 2944 if (head.study(info)) { // Deterministic 2945 GroupTail temp = (GroupTail) tail; 2946 head = root = new GroupCurly(head.next, curly.cmin, 2947 curly.cmax, curly.type, 2948 ((GroupTail)tail).localIndex, 2949 ((GroupTail)tail).groupIndex, 2950 capturingGroup); 2951 return head; 2952 } else { // Non-deterministic 2953 int temp = ((GroupHead) head).localIndex; 2954 Loop loop; 2955 if (curly.type == GREEDY) 2956 loop = new Loop(this.localCount, temp); 2957 else // Reluctant Curly 2958 loop = new LazyLoop(this.localCount, temp); 2959 Prolog prolog = new Prolog(loop); 2960 this.localCount += 1; 2961 loop.cmin = curly.cmin; 2962 loop.cmax = curly.cmax; 2963 loop.body = head; 2964 tail.next = loop; 2965 root = loop; 2966 return prolog; // Dual return 2967 } 2968 } 2969 throw error("Internal logic error"); 2970 } 2971 2972 /** 2973 * Create group head and tail nodes using double return. If the group is 2974 * created with anonymous true then it is a pure group and should not 2975 * affect group counting. 2976 */ 2977 private Node createGroup(boolean anonymous) { 2978 int localIndex = localCount++; 2979 int groupIndex = 0; 2980 if (!anonymous) 2981 groupIndex = capturingGroupCount++; 2982 GroupHead head = new GroupHead(localIndex); 2983 root = new GroupTail(localIndex, groupIndex); 2984 if (!anonymous && groupIndex < 10) 2985 groupNodes[groupIndex] = head; 2986 return head; 2987 } 2988 2989 @SuppressWarnings("fallthrough") 2990 /** 2991 * Parses inlined match flags and set them appropriately. 2992 */ 2993 private void addFlag() { 2994 int ch = peek(); 2995 for (;;) { 2996 switch (ch) { 2997 case 'i': 2998 flags |= CASE_INSENSITIVE; 2999 break; 3000 case 'm': 3001 flags |= MULTILINE; 3002 break; 3003 case 's': 3004 flags |= DOTALL; 3005 break; 3006 case 'd': 3007 flags |= UNIX_LINES; 3008 break; 3009 case 'u': 3010 flags |= UNICODE_CASE; 3011 break; 3012 case 'c': 3013 flags |= CANON_EQ; 3014 break; 3015 case 'x': 3016 flags |= COMMENTS; 3017 break; 3018 case 'U': 3019 flags |= (UNICODE_CHARACTER_CLASS | UNICODE_CASE); 3020 break; 3021 case '-': // subFlag then fall through 3022 ch = next(); 3023 subFlag(); 3024 default: 3025 return; 3026 } 3027 ch = next(); 3028 } 3029 } 3030 3031 @SuppressWarnings("fallthrough") 3032 /** 3033 * Parses the second part of inlined match flags and turns off 3034 * flags appropriately. 3035 */ 3036 private void subFlag() { 3037 int ch = peek(); 3038 for (;;) { 3039 switch (ch) { 3040 case 'i': 3041 flags &= ~CASE_INSENSITIVE; 3042 break; 3043 case 'm': 3044 flags &= ~MULTILINE; 3045 break; 3046 case 's': 3047 flags &= ~DOTALL; 3048 break; 3049 case 'd': 3050 flags &= ~UNIX_LINES; 3051 break; 3052 case 'u': 3053 flags &= ~UNICODE_CASE; 3054 break; 3055 case 'c': 3056 flags &= ~CANON_EQ; 3057 break; 3058 case 'x': 3059 flags &= ~COMMENTS; 3060 break; 3061 case 'U': 3062 flags &= ~(UNICODE_CHARACTER_CLASS | UNICODE_CASE); 3063 default: 3064 return; 3065 } 3066 ch = next(); 3067 } 3068 } 3069 3070 static final int MAX_REPS = 0x7FFFFFFF; 3071 3072 static final int GREEDY = 0; 3073 3074 static final int LAZY = 1; 3075 3076 static final int POSSESSIVE = 2; 3077 3078 static final int INDEPENDENT = 3; 3079 3080 /** 3081 * Processes repetition. If the next character peeked is a quantifier 3082 * then new nodes must be appended to handle the repetition. 3083 * Prev could be a single or a group, so it could be a chain of nodes. 3084 */ 3085 private Node closure(Node prev) { 3086 Node atom; 3087 int ch = peek(); 3088 switch (ch) { 3089 case '?': 3090 ch = next(); 3091 if (ch == '?') { 3092 next(); 3093 return new Ques(prev, LAZY); 3094 } else if (ch == '+') { 3095 next(); 3096 return new Ques(prev, POSSESSIVE); 3097 } 3098 return new Ques(prev, GREEDY); 3099 case '*': 3100 ch = next(); 3101 if (ch == '?') { 3102 next(); 3103 return new Curly(prev, 0, MAX_REPS, LAZY); 3104 } else if (ch == '+') { 3105 next(); 3106 return new Curly(prev, 0, MAX_REPS, POSSESSIVE); 3107 } 3108 return new Curly(prev, 0, MAX_REPS, GREEDY); 3109 case '+': 3110 ch = next(); 3111 if (ch == '?') { 3112 next(); 3113 return new Curly(prev, 1, MAX_REPS, LAZY); 3114 } else if (ch == '+') { 3115 next(); 3116 return new Curly(prev, 1, MAX_REPS, POSSESSIVE); 3117 } 3118 return new Curly(prev, 1, MAX_REPS, GREEDY); 3119 case '{': 3120 ch = temp[cursor+1]; 3121 if (ASCII.isDigit(ch)) { 3122 skip(); 3123 int cmin = 0; 3124 do { 3125 cmin = cmin * 10 + (ch - '0'); 3126 } while (ASCII.isDigit(ch = read())); 3127 int cmax = cmin; 3128 if (ch == ',') { 3129 ch = read(); 3130 cmax = MAX_REPS; 3131 if (ch != '}') { 3132 cmax = 0; 3133 while (ASCII.isDigit(ch)) { 3134 cmax = cmax * 10 + (ch - '0'); 3135 ch = read(); 3136 } 3137 } 3138 } 3139 if (ch != '}') 3140 throw error("Unclosed counted closure"); 3141 if (((cmin) | (cmax) | (cmax - cmin)) < 0) 3142 throw error("Illegal repetition range"); 3143 Curly curly; 3144 ch = peek(); 3145 if (ch == '?') { 3146 next(); 3147 curly = new Curly(prev, cmin, cmax, LAZY); 3148 } else if (ch == '+') { 3149 next(); 3150 curly = new Curly(prev, cmin, cmax, POSSESSIVE); 3151 } else { 3152 curly = new Curly(prev, cmin, cmax, GREEDY); 3153 } 3154 return curly; 3155 } else { 3156 throw error("Illegal repetition"); 3157 } 3158 default: 3159 return prev; 3160 } 3161 } 3162 3163 /** 3164 * Utility method for parsing control escape sequences. 3165 */ 3166 private int c() { 3167 if (cursor < patternLength) { 3168 return read() ^ 64; 3169 } 3170 throw error("Illegal control escape sequence"); 3171 } 3172 3173 /** 3174 * Utility method for parsing octal escape sequences. 3175 */ 3176 private int o() { 3177 int n = read(); 3178 if (((n-'0')|('7'-n)) >= 0) { 3179 int m = read(); 3180 if (((m-'0')|('7'-m)) >= 0) { 3181 int o = read(); 3182 if ((((o-'0')|('7'-o)) >= 0) && (((n-'0')|('3'-n)) >= 0)) { 3183 return (n - '0') * 64 + (m - '0') * 8 + (o - '0'); 3184 } 3185 unread(); 3186 return (n - '0') * 8 + (m - '0'); 3187 } 3188 unread(); 3189 return (n - '0'); 3190 } 3191 throw error("Illegal octal escape sequence"); 3192 } 3193 3194 /** 3195 * Utility method for parsing hexadecimal escape sequences. 3196 */ 3197 private int x() { 3198 int n = read(); 3199 if (ASCII.isHexDigit(n)) { 3200 int m = read(); 3201 if (ASCII.isHexDigit(m)) { 3202 return ASCII.toDigit(n) * 16 + ASCII.toDigit(m); 3203 } 3204 } else if (n == '{' && ASCII.isHexDigit(peek())) { 3205 int ch = 0; 3206 while (ASCII.isHexDigit(n = read())) { 3207 ch = (ch << 4) + ASCII.toDigit(n); 3208 if (ch > Character.MAX_CODE_POINT) 3209 throw error("Hexadecimal codepoint is too big"); 3210 } 3211 if (n != '}') 3212 throw error("Unclosed hexadecimal escape sequence"); 3213 return ch; 3214 } 3215 throw error("Illegal hexadecimal escape sequence"); 3216 } 3217 3218 /** 3219 * Utility method for parsing unicode escape sequences. 3220 */ 3221 private int cursor() { 3222 return cursor; 3223 } 3224 3225 private void setcursor(int pos) { 3226 cursor = pos; 3227 } 3228 3229 private int uxxxx() { 3230 int n = 0; 3231 for (int i = 0; i < 4; i++) { 3232 int ch = read(); 3233 if (!ASCII.isHexDigit(ch)) { 3234 throw error("Illegal Unicode escape sequence"); 3235 } 3236 n = n * 16 + ASCII.toDigit(ch); 3237 } 3238 return n; 3239 } 3240 3241 private int u() { 3242 int n = uxxxx(); 3243 if (Character.isHighSurrogate((char)n)) { 3244 int cur = cursor(); 3245 if (read() == '\\' && read() == 'u') { 3246 int n2 = uxxxx(); 3247 if (Character.isLowSurrogate((char)n2)) 3248 return Character.toCodePoint((char)n, (char)n2); 3249 } 3250 setcursor(cur); 3251 } 3252 return n; 3253 } 3254 3255 // 3256 // Utility methods for code point support 3257 // 3258 3259 private static final int countChars(CharSequence seq, int index, 3260 int lengthInCodePoints) { 3261 // optimization 3262 if (lengthInCodePoints == 1 && !Character.isHighSurrogate(seq.charAt(index))) { 3263 assert (index >= 0 && index < seq.length()); 3264 return 1; 3265 } 3266 int length = seq.length(); 3267 int x = index; 3268 if (lengthInCodePoints >= 0) { 3269 assert (index >= 0 && index < length); 3270 for (int i = 0; x < length && i < lengthInCodePoints; i++) { 3271 if (Character.isHighSurrogate(seq.charAt(x++))) { 3272 if (x < length && Character.isLowSurrogate(seq.charAt(x))) { 3273 x++; 3274 } 3275 } 3276 } 3277 return x - index; 3278 } 3279 3280 assert (index >= 0 && index <= length); 3281 if (index == 0) { 3282 return 0; 3283 } 3284 int len = -lengthInCodePoints; 3285 for (int i = 0; x > 0 && i < len; i++) { 3286 if (Character.isLowSurrogate(seq.charAt(--x))) { 3287 if (x > 0 && Character.isHighSurrogate(seq.charAt(x-1))) { 3288 x--; 3289 } 3290 } 3291 } 3292 return index - x; 3293 } 3294 3295 private static final int countCodePoints(CharSequence seq) { 3296 int length = seq.length(); 3297 int n = 0; 3298 for (int i = 0; i < length; ) { 3299 n++; 3300 if (Character.isHighSurrogate(seq.charAt(i++))) { 3301 if (i < length && Character.isLowSurrogate(seq.charAt(i))) { 3302 i++; 3303 } 3304 } 3305 } 3306 return n; 3307 } 3308 3309 /** 3310 * Creates a bit vector for matching Latin-1 values. A normal BitClass 3311 * never matches values above Latin-1, and a complemented BitClass always 3312 * matches values above Latin-1. 3313 */ 3314 private static final class BitClass extends BmpCharProperty { 3315 final boolean[] bits; 3316 BitClass() { bits = new boolean[256]; } 3317 private BitClass(boolean[] bits) { this.bits = bits; } 3318 BitClass add(int c, int flags) { 3319 assert c >= 0 && c <= 255; 3320 if ((flags & CASE_INSENSITIVE) != 0) { 3321 if (ASCII.isAscii(c)) { 3322 bits[ASCII.toUpper(c)] = true; 3323 bits[ASCII.toLower(c)] = true; 3324 } else if ((flags & UNICODE_CASE) != 0) { 3325 bits[Character.toLowerCase(c)] = true; 3326 bits[Character.toUpperCase(c)] = true; 3327 } 3328 } 3329 bits[c] = true; 3330 return this; 3331 } 3332 boolean isSatisfiedBy(int ch) { 3333 return ch < 256 && bits[ch]; 3334 } 3335 } 3336 3337 /** 3338 * Returns a suitably optimized, single character matcher. 3339 */ 3340 private CharProperty newSingle(final int ch) { 3341 if (has(CASE_INSENSITIVE)) { 3342 int lower, upper; 3343 if (has(UNICODE_CASE)) { 3344 upper = Character.toUpperCase(ch); 3345 lower = Character.toLowerCase(upper); 3346 if (upper != lower) 3347 return new SingleU(lower); 3348 } else if (ASCII.isAscii(ch)) { 3349 lower = ASCII.toLower(ch); 3350 upper = ASCII.toUpper(ch); 3351 if (lower != upper) 3352 return new SingleI(lower, upper); 3353 } 3354 } 3355 if (isSupplementary(ch)) 3356 return new SingleS(ch); // Match a given Unicode character 3357 return new Single(ch); // Match a given BMP character 3358 } 3359 3360 /** 3361 * Utility method for creating a string slice matcher. 3362 */ 3363 private Node newSlice(int[] buf, int count, boolean hasSupplementary) { 3364 int[] tmp = new int[count]; 3365 if (has(CASE_INSENSITIVE)) { 3366 if (has(UNICODE_CASE)) { 3367 for (int i = 0; i < count; i++) { 3368 tmp[i] = Character.toLowerCase( 3369 Character.toUpperCase(buf[i])); 3370 } 3371 return hasSupplementary? new SliceUS(tmp) : new SliceU(tmp); 3372 } 3373 for (int i = 0; i < count; i++) { 3374 tmp[i] = ASCII.toLower(buf[i]); 3375 } 3376 return hasSupplementary? new SliceIS(tmp) : new SliceI(tmp); 3377 } 3378 for (int i = 0; i < count; i++) { 3379 tmp[i] = buf[i]; 3380 } 3381 return hasSupplementary ? new SliceS(tmp) : new Slice(tmp); 3382 } 3383 3384 /** 3385 * The following classes are the building components of the object 3386 * tree that represents a compiled regular expression. The object tree 3387 * is made of individual elements that handle constructs in the Pattern. 3388 * Each type of object knows how to match its equivalent construct with 3389 * the match() method. 3390 */ 3391 3392 /** 3393 * Base class for all node classes. Subclasses should override the match() 3394 * method as appropriate. This class is an accepting node, so its match() 3395 * always returns true. 3396 */ 3397 static class Node extends Object { 3398 Node next; 3399 Node() { 3400 next = Pattern.accept; 3401 } 3402 /** 3403 * This method implements the classic accept node. 3404 */ 3405 boolean match(Matcher matcher, int i, CharSequence seq) { 3406 matcher.last = i; 3407 matcher.groups[0] = matcher.first; 3408 matcher.groups[1] = matcher.last; 3409 return true; 3410 } 3411 /** 3412 * This method is good for all zero length assertions. 3413 */ 3414 boolean study(TreeInfo info) { 3415 if (next != null) { 3416 return next.study(info); 3417 } else { 3418 return info.deterministic; 3419 } 3420 } 3421 } 3422 3423 static class LastNode extends Node { 3424 /** 3425 * This method implements the classic accept node with 3426 * the addition of a check to see if the match occurred 3427 * using all of the input. 3428 */ 3429 boolean match(Matcher matcher, int i, CharSequence seq) { 3430 if (matcher.acceptMode == Matcher.ENDANCHOR && i != matcher.to) 3431 return false; 3432 matcher.last = i; 3433 matcher.groups[0] = matcher.first; 3434 matcher.groups[1] = matcher.last; 3435 return true; 3436 } 3437 } 3438 3439 /** 3440 * Used for REs that can start anywhere within the input string. 3441 * This basically tries to match repeatedly at each spot in the 3442 * input string, moving forward after each try. An anchored search 3443 * or a BnM will bypass this node completely. 3444 */ 3445 static class Start extends Node { 3446 int minLength; 3447 Start(Node node) { 3448 this.next = node; 3449 TreeInfo info = new TreeInfo(); 3450 next.study(info); 3451 minLength = info.minLength; 3452 } 3453 boolean match(Matcher matcher, int i, CharSequence seq) { 3454 if (i > matcher.to - minLength) { 3455 matcher.hitEnd = true; 3456 return false; 3457 } 3458 int guard = matcher.to - minLength; 3459 for (; i <= guard; i++) { 3460 if (next.match(matcher, i, seq)) { 3461 matcher.first = i; 3462 matcher.groups[0] = matcher.first; 3463 matcher.groups[1] = matcher.last; 3464 return true; 3465 } 3466 } 3467 matcher.hitEnd = true; 3468 return false; 3469 } 3470 boolean study(TreeInfo info) { 3471 next.study(info); 3472 info.maxValid = false; 3473 info.deterministic = false; 3474 return false; 3475 } 3476 } 3477 3478 /* 3479 * StartS supports supplementary characters, including unpaired surrogates. 3480 */ 3481 static final class StartS extends Start { 3482 StartS(Node node) { 3483 super(node); 3484 } 3485 boolean match(Matcher matcher, int i, CharSequence seq) { 3486 if (i > matcher.to - minLength) { 3487 matcher.hitEnd = true; 3488 return false; 3489 } 3490 int guard = matcher.to - minLength; 3491 while (i <= guard) { 3492 //if ((ret = next.match(matcher, i, seq)) || i == guard) 3493 if (next.match(matcher, i, seq)) { 3494 matcher.first = i; 3495 matcher.groups[0] = matcher.first; 3496 matcher.groups[1] = matcher.last; 3497 return true; 3498 } 3499 if (i == guard) 3500 break; 3501 // Optimization to move to the next character. This is 3502 // faster than countChars(seq, i, 1). 3503 if (Character.isHighSurrogate(seq.charAt(i++))) { 3504 if (i < seq.length() && 3505 Character.isLowSurrogate(seq.charAt(i))) { 3506 i++; 3507 } 3508 } 3509 } 3510 matcher.hitEnd = true; 3511 return false; 3512 } 3513 } 3514 3515 /** 3516 * Node to anchor at the beginning of input. This object implements the 3517 * match for a \A sequence, and the caret anchor will use this if not in 3518 * multiline mode. 3519 */ 3520 static final class Begin extends Node { 3521 boolean match(Matcher matcher, int i, CharSequence seq) { 3522 int fromIndex = (matcher.anchoringBounds) ? 3523 matcher.from : 0; 3524 if (i == fromIndex && next.match(matcher, i, seq)) { 3525 matcher.first = i; 3526 matcher.groups[0] = i; 3527 matcher.groups[1] = matcher.last; 3528 return true; 3529 } else { 3530 return false; 3531 } 3532 } 3533 } 3534 3535 /** 3536 * Node to anchor at the end of input. This is the absolute end, so this 3537 * should not match at the last newline before the end as $ will. 3538 */ 3539 static final class End extends Node { 3540 boolean match(Matcher matcher, int i, CharSequence seq) { 3541 int endIndex = (matcher.anchoringBounds) ? 3542 matcher.to : matcher.getTextLength(); 3543 if (i == endIndex) { 3544 matcher.hitEnd = true; 3545 return next.match(matcher, i, seq); 3546 } 3547 return false; 3548 } 3549 } 3550 3551 /** 3552 * Node to anchor at the beginning of a line. This is essentially the 3553 * object to match for the multiline ^. 3554 */ 3555 static final class Caret extends Node { 3556 boolean match(Matcher matcher, int i, CharSequence seq) { 3557 int startIndex = matcher.from; 3558 int endIndex = matcher.to; 3559 if (!matcher.anchoringBounds) { 3560 startIndex = 0; 3561 endIndex = matcher.getTextLength(); 3562 } 3563 // Perl does not match ^ at end of input even after newline 3564 if (i == endIndex) { 3565 matcher.hitEnd = true; 3566 return false; 3567 } 3568 if (i > startIndex) { 3569 char ch = seq.charAt(i-1); 3570 if (ch != '\n' && ch != '\r' 3571 && (ch|1) != '\u2029' 3572 && ch != '\u0085' ) { 3573 return false; 3574 } 3575 // Should treat /r/n as one newline 3576 if (ch == '\r' && seq.charAt(i) == '\n') 3577 return false; 3578 } 3579 return next.match(matcher, i, seq); 3580 } 3581 } 3582 3583 /** 3584 * Node to anchor at the beginning of a line when in unixdot mode. 3585 */ 3586 static final class UnixCaret extends Node { 3587 boolean match(Matcher matcher, int i, CharSequence seq) { 3588 int startIndex = matcher.from; 3589 int endIndex = matcher.to; 3590 if (!matcher.anchoringBounds) { 3591 startIndex = 0; 3592 endIndex = matcher.getTextLength(); 3593 } 3594 // Perl does not match ^ at end of input even after newline 3595 if (i == endIndex) { 3596 matcher.hitEnd = true; 3597 return false; 3598 } 3599 if (i > startIndex) { 3600 char ch = seq.charAt(i-1); 3601 if (ch != '\n') { 3602 return false; 3603 } 3604 } 3605 return next.match(matcher, i, seq); 3606 } 3607 } 3608 3609 /** 3610 * Node to match the location where the last match ended. 3611 * This is used for the \G construct. 3612 */ 3613 static final class LastMatch extends Node { 3614 boolean match(Matcher matcher, int i, CharSequence seq) { 3615 if (i != matcher.oldLast) 3616 return false; 3617 return next.match(matcher, i, seq); 3618 } 3619 } 3620 3621 /** 3622 * Node to anchor at the end of a line or the end of input based on the 3623 * multiline mode. 3624 * 3625 * When not in multiline mode, the $ can only match at the very end 3626 * of the input, unless the input ends in a line terminator in which 3627 * it matches right before the last line terminator. 3628 * 3629 * Note that \r\n is considered an atomic line terminator. 3630 * 3631 * Like ^ the $ operator matches at a position, it does not match the 3632 * line terminators themselves. 3633 */ 3634 static final class Dollar extends Node { 3635 boolean multiline; 3636 Dollar(boolean mul) { 3637 multiline = mul; 3638 } 3639 boolean match(Matcher matcher, int i, CharSequence seq) { 3640 int endIndex = (matcher.anchoringBounds) ? 3641 matcher.to : matcher.getTextLength(); 3642 if (!multiline) { 3643 if (i < endIndex - 2) 3644 return false; 3645 if (i == endIndex - 2) { 3646 char ch = seq.charAt(i); 3647 if (ch != '\r') 3648 return false; 3649 ch = seq.charAt(i + 1); 3650 if (ch != '\n') 3651 return false; 3652 } 3653 } 3654 // Matches before any line terminator; also matches at the 3655 // end of input 3656 // Before line terminator: 3657 // If multiline, we match here no matter what 3658 // If not multiline, fall through so that the end 3659 // is marked as hit; this must be a /r/n or a /n 3660 // at the very end so the end was hit; more input 3661 // could make this not match here 3662 if (i < endIndex) { 3663 char ch = seq.charAt(i); 3664 if (ch == '\n') { 3665 // No match between \r\n 3666 if (i > 0 && seq.charAt(i-1) == '\r') 3667 return false; 3668 if (multiline) 3669 return next.match(matcher, i, seq); 3670 } else if (ch == '\r' || ch == '\u0085' || 3671 (ch|1) == '\u2029') { 3672 if (multiline) 3673 return next.match(matcher, i, seq); 3674 } else { // No line terminator, no match 3675 return false; 3676 } 3677 } 3678 // Matched at current end so hit end 3679 matcher.hitEnd = true; 3680 // If a $ matches because of end of input, then more input 3681 // could cause it to fail! 3682 matcher.requireEnd = true; 3683 return next.match(matcher, i, seq); 3684 } 3685 boolean study(TreeInfo info) { 3686 next.study(info); 3687 return info.deterministic; 3688 } 3689 } 3690 3691 /** 3692 * Node to anchor at the end of a line or the end of input based on the 3693 * multiline mode when in unix lines mode. 3694 */ 3695 static final class UnixDollar extends Node { 3696 boolean multiline; 3697 UnixDollar(boolean mul) { 3698 multiline = mul; 3699 } 3700 boolean match(Matcher matcher, int i, CharSequence seq) { 3701 int endIndex = (matcher.anchoringBounds) ? 3702 matcher.to : matcher.getTextLength(); 3703 if (i < endIndex) { 3704 char ch = seq.charAt(i); 3705 if (ch == '\n') { 3706 // If not multiline, then only possible to 3707 // match at very end or one before end 3708 if (multiline == false && i != endIndex - 1) 3709 return false; 3710 // If multiline return next.match without setting 3711 // matcher.hitEnd 3712 if (multiline) 3713 return next.match(matcher, i, seq); 3714 } else { 3715 return false; 3716 } 3717 } 3718 // Matching because at the end or 1 before the end; 3719 // more input could change this so set hitEnd 3720 matcher.hitEnd = true; 3721 // If a $ matches because of end of input, then more input 3722 // could cause it to fail! 3723 matcher.requireEnd = true; 3724 return next.match(matcher, i, seq); 3725 } 3726 boolean study(TreeInfo info) { 3727 next.study(info); 3728 return info.deterministic; 3729 } 3730 } 3731 3732 /** 3733 * Node class that matches a Unicode line ending '\R' 3734 */ 3735 static final class LineEnding extends Node { 3736 boolean match(Matcher matcher, int i, CharSequence seq) { 3737 // (u+000Du+000A|[u+000Au+000Bu+000Cu+000Du+0085u+2028u+2029]) 3738 if (i < matcher.to) { 3739 int ch = seq.charAt(i); 3740 if (ch == 0x0A || ch == 0x0B || ch == 0x0C || 3741 ch == 0x85 || ch == 0x2028 || ch == 0x2029) 3742 return next.match(matcher, i + 1, seq); 3743 if (ch == 0x0D) { 3744 i++; 3745 if (i < matcher.to && seq.charAt(i) == 0x0A) 3746 i++; 3747 return next.match(matcher, i, seq); 3748 } 3749 } else { 3750 matcher.hitEnd = true; 3751 } 3752 return false; 3753 } 3754 boolean study(TreeInfo info) { 3755 info.minLength++; 3756 info.maxLength += 2; 3757 return next.study(info); 3758 } 3759 } 3760 3761 /** 3762 * Abstract node class to match one character satisfying some 3763 * boolean property. 3764 */ 3765 private static abstract class CharProperty extends Node { 3766 abstract boolean isSatisfiedBy(int ch); 3767 CharProperty complement() { 3768 return new CharProperty() { 3769 boolean isSatisfiedBy(int ch) { 3770 return ! CharProperty.this.isSatisfiedBy(ch);}}; 3771 } 3772 boolean match(Matcher matcher, int i, CharSequence seq) { 3773 if (i < matcher.to) { 3774 int ch = Character.codePointAt(seq, i); 3775 return isSatisfiedBy(ch) 3776 && next.match(matcher, i+Character.charCount(ch), seq); 3777 } else { 3778 matcher.hitEnd = true; 3779 return false; 3780 } 3781 } 3782 boolean study(TreeInfo info) { 3783 info.minLength++; 3784 info.maxLength++; 3785 return next.study(info); 3786 } 3787 } 3788 3789 /** 3790 * Optimized version of CharProperty that works only for 3791 * properties never satisfied by Supplementary characters. 3792 */ 3793 private static abstract class BmpCharProperty extends CharProperty { 3794 boolean match(Matcher matcher, int i, CharSequence seq) { 3795 if (i < matcher.to) { 3796 return isSatisfiedBy(seq.charAt(i)) 3797 && next.match(matcher, i+1, seq); 3798 } else { 3799 matcher.hitEnd = true; 3800 return false; 3801 } 3802 } 3803 } 3804 3805 /** 3806 * Node class that matches a Supplementary Unicode character 3807 */ 3808 static final class SingleS extends CharProperty { 3809 final int c; 3810 SingleS(int c) { this.c = c; } 3811 boolean isSatisfiedBy(int ch) { 3812 return ch == c; 3813 } 3814 } 3815 3816 /** 3817 * Optimization -- matches a given BMP character 3818 */ 3819 static final class Single extends BmpCharProperty { 3820 final int c; 3821 Single(int c) { this.c = c; } 3822 boolean isSatisfiedBy(int ch) { 3823 return ch == c; 3824 } 3825 } 3826 3827 /** 3828 * Case insensitive matches a given BMP character 3829 */ 3830 static final class SingleI extends BmpCharProperty { 3831 final int lower; 3832 final int upper; 3833 SingleI(int lower, int upper) { 3834 this.lower = lower; 3835 this.upper = upper; 3836 } 3837 boolean isSatisfiedBy(int ch) { 3838 return ch == lower || ch == upper; 3839 } 3840 } 3841 3842 /** 3843 * Unicode case insensitive matches a given Unicode character 3844 */ 3845 static final class SingleU extends CharProperty { 3846 final int lower; 3847 SingleU(int lower) { 3848 this.lower = lower; 3849 } 3850 boolean isSatisfiedBy(int ch) { 3851 return lower == ch || 3852 lower == Character.toLowerCase(Character.toUpperCase(ch)); 3853 } 3854 } 3855 3856 /** 3857 * Node class that matches a Unicode block. 3858 */ 3859 static final class Block extends CharProperty { 3860 final Character.UnicodeBlock block; 3861 Block(Character.UnicodeBlock block) { 3862 this.block = block; 3863 } 3864 boolean isSatisfiedBy(int ch) { 3865 return block == Character.UnicodeBlock.of(ch); 3866 } 3867 } 3868 3869 /** 3870 * Node class that matches a Unicode script 3871 */ 3872 static final class Script extends CharProperty { 3873 final Character.UnicodeScript script; 3874 Script(Character.UnicodeScript script) { 3875 this.script = script; 3876 } 3877 boolean isSatisfiedBy(int ch) { 3878 return script == Character.UnicodeScript.of(ch); 3879 } 3880 } 3881 3882 /** 3883 * Node class that matches a Unicode category. 3884 */ 3885 static final class Category extends CharProperty { 3886 final int typeMask; 3887 Category(int typeMask) { this.typeMask = typeMask; } 3888 boolean isSatisfiedBy(int ch) { 3889 return (typeMask & (1 << Character.getType(ch))) != 0; 3890 } 3891 } 3892 3893 /** 3894 * Node class that matches a Unicode "type" 3895 */ 3896 static final class Utype extends CharProperty { 3897 final UnicodeProp uprop; 3898 Utype(UnicodeProp uprop) { this.uprop = uprop; } 3899 boolean isSatisfiedBy(int ch) { 3900 return uprop.is(ch); 3901 } 3902 } 3903 3904 /** 3905 * Node class that matches a POSIX type. 3906 */ 3907 static final class Ctype extends BmpCharProperty { 3908 final int ctype; 3909 Ctype(int ctype) { this.ctype = ctype; } 3910 boolean isSatisfiedBy(int ch) { 3911 return ch < 128 && ASCII.isType(ch, ctype); 3912 } 3913 } 3914 3915 /** 3916 * Node class that matches a Perl vertical whitespace 3917 */ 3918 static final class VertWS extends BmpCharProperty { 3919 boolean isSatisfiedBy(int cp) { 3920 return (cp >= 0x0A && cp <= 0x0D) || 3921 cp == 0x85 || cp == 0x2028 || cp == 0x2029; 3922 } 3923 } 3924 3925 /** 3926 * Node class that matches a Perl horizontal whitespace 3927 */ 3928 static final class HorizWS extends BmpCharProperty { 3929 boolean isSatisfiedBy(int cp) { 3930 return cp == 0x09 || cp == 0x20 || cp == 0xa0 || 3931 cp == 0x1680 || cp == 0x180e || 3932 cp >= 0x2000 && cp <= 0x200a || 3933 cp == 0x202f || cp == 0x205f || cp == 0x3000; 3934 } 3935 } 3936 3937 /** 3938 * Base class for all Slice nodes 3939 */ 3940 static class SliceNode extends Node { 3941 int[] buffer; 3942 SliceNode(int[] buf) { 3943 buffer = buf; 3944 } 3945 boolean study(TreeInfo info) { 3946 info.minLength += buffer.length; 3947 info.maxLength += buffer.length; 3948 return next.study(info); 3949 } 3950 } 3951 3952 /** 3953 * Node class for a case sensitive/BMP-only sequence of literal 3954 * characters. 3955 */ 3956 static final class Slice extends SliceNode { 3957 Slice(int[] buf) { 3958 super(buf); 3959 } 3960 boolean match(Matcher matcher, int i, CharSequence seq) { 3961 int[] buf = buffer; 3962 int len = buf.length; 3963 for (int j=0; j<len; j++) { 3964 if ((i+j) >= matcher.to) { 3965 matcher.hitEnd = true; 3966 return false; 3967 } 3968 if (buf[j] != seq.charAt(i+j)) 3969 return false; 3970 } 3971 return next.match(matcher, i+len, seq); 3972 } 3973 } 3974 3975 /** 3976 * Node class for a case_insensitive/BMP-only sequence of literal 3977 * characters. 3978 */ 3979 static class SliceI extends SliceNode { 3980 SliceI(int[] buf) { 3981 super(buf); 3982 } 3983 boolean match(Matcher matcher, int i, CharSequence seq) { 3984 int[] buf = buffer; 3985 int len = buf.length; 3986 for (int j=0; j<len; j++) { 3987 if ((i+j) >= matcher.to) { 3988 matcher.hitEnd = true; 3989 return false; 3990 } 3991 int c = seq.charAt(i+j); 3992 if (buf[j] != c && 3993 buf[j] != ASCII.toLower(c)) 3994 return false; 3995 } 3996 return next.match(matcher, i+len, seq); 3997 } 3998 } 3999 4000 /** 4001 * Node class for a unicode_case_insensitive/BMP-only sequence of 4002 * literal characters. Uses unicode case folding. 4003 */ 4004 static final class SliceU extends SliceNode { 4005 SliceU(int[] buf) { 4006 super(buf); 4007 } 4008 boolean match(Matcher matcher, int i, CharSequence seq) { 4009 int[] buf = buffer; 4010 int len = buf.length; 4011 for (int j=0; j<len; j++) { 4012 if ((i+j) >= matcher.to) { 4013 matcher.hitEnd = true; 4014 return false; 4015 } 4016 int c = seq.charAt(i+j); 4017 if (buf[j] != c && 4018 buf[j] != Character.toLowerCase(Character.toUpperCase(c))) 4019 return false; 4020 } 4021 return next.match(matcher, i+len, seq); 4022 } 4023 } 4024 4025 /** 4026 * Node class for a case sensitive sequence of literal characters 4027 * including supplementary characters. 4028 */ 4029 static final class SliceS extends SliceNode { 4030 SliceS(int[] buf) { 4031 super(buf); 4032 } 4033 boolean match(Matcher matcher, int i, CharSequence seq) { 4034 int[] buf = buffer; 4035 int x = i; 4036 for (int j = 0; j < buf.length; j++) { 4037 if (x >= matcher.to) { 4038 matcher.hitEnd = true; 4039 return false; 4040 } 4041 int c = Character.codePointAt(seq, x); 4042 if (buf[j] != c) 4043 return false; 4044 x += Character.charCount(c); 4045 if (x > matcher.to) { 4046 matcher.hitEnd = true; 4047 return false; 4048 } 4049 } 4050 return next.match(matcher, x, seq); 4051 } 4052 } 4053 4054 /** 4055 * Node class for a case insensitive sequence of literal characters 4056 * including supplementary characters. 4057 */ 4058 static class SliceIS extends SliceNode { 4059 SliceIS(int[] buf) { 4060 super(buf); 4061 } 4062 int toLower(int c) { 4063 return ASCII.toLower(c); 4064 } 4065 boolean match(Matcher matcher, int i, CharSequence seq) { 4066 int[] buf = buffer; 4067 int x = i; 4068 for (int j = 0; j < buf.length; j++) { 4069 if (x >= matcher.to) { 4070 matcher.hitEnd = true; 4071 return false; 4072 } 4073 int c = Character.codePointAt(seq, x); 4074 if (buf[j] != c && buf[j] != toLower(c)) 4075 return false; 4076 x += Character.charCount(c); 4077 if (x > matcher.to) { 4078 matcher.hitEnd = true; 4079 return false; 4080 } 4081 } 4082 return next.match(matcher, x, seq); 4083 } 4084 } 4085 4086 /** 4087 * Node class for a case insensitive sequence of literal characters. 4088 * Uses unicode case folding. 4089 */ 4090 static final class SliceUS extends SliceIS { 4091 SliceUS(int[] buf) { 4092 super(buf); 4093 } 4094 int toLower(int c) { 4095 return Character.toLowerCase(Character.toUpperCase(c)); 4096 } 4097 } 4098 4099 private static boolean inRange(int lower, int ch, int upper) { 4100 return lower <= ch && ch <= upper; 4101 } 4102 4103 /** 4104 * Returns node for matching characters within an explicit value range. 4105 */ 4106 private static CharProperty rangeFor(final int lower, 4107 final int upper) { 4108 return new CharProperty() { 4109 boolean isSatisfiedBy(int ch) { 4110 return inRange(lower, ch, upper);}}; 4111 } 4112 4113 /** 4114 * Returns node for matching characters within an explicit value 4115 * range in a case insensitive manner. 4116 */ 4117 private CharProperty caseInsensitiveRangeFor(final int lower, 4118 final int upper) { 4119 if (has(UNICODE_CASE)) 4120 return new CharProperty() { 4121 boolean isSatisfiedBy(int ch) { 4122 if (inRange(lower, ch, upper)) 4123 return true; 4124 int up = Character.toUpperCase(ch); 4125 return inRange(lower, up, upper) || 4126 inRange(lower, Character.toLowerCase(up), upper);}}; 4127 return new CharProperty() { 4128 boolean isSatisfiedBy(int ch) { 4129 return inRange(lower, ch, upper) || 4130 ASCII.isAscii(ch) && 4131 (inRange(lower, ASCII.toUpper(ch), upper) || 4132 inRange(lower, ASCII.toLower(ch), upper)); 4133 }}; 4134 } 4135 4136 /** 4137 * Implements the Unicode category ALL and the dot metacharacter when 4138 * in dotall mode. 4139 */ 4140 static final class All extends CharProperty { 4141 boolean isSatisfiedBy(int ch) { 4142 return true; 4143 } 4144 } 4145 4146 /** 4147 * Node class for the dot metacharacter when dotall is not enabled. 4148 */ 4149 static final class Dot extends CharProperty { 4150 boolean isSatisfiedBy(int ch) { 4151 return (ch != '\n' && ch != '\r' 4152 && (ch|1) != '\u2029' 4153 && ch != '\u0085'); 4154 } 4155 } 4156 4157 /** 4158 * Node class for the dot metacharacter when dotall is not enabled 4159 * but UNIX_LINES is enabled. 4160 */ 4161 static final class UnixDot extends CharProperty { 4162 boolean isSatisfiedBy(int ch) { 4163 return ch != '\n'; 4164 } 4165 } 4166 4167 /** 4168 * The 0 or 1 quantifier. This one class implements all three types. 4169 */ 4170 static final class Ques extends Node { 4171 Node atom; 4172 int type; 4173 Ques(Node node, int type) { 4174 this.atom = node; 4175 this.type = type; 4176 } 4177 boolean match(Matcher matcher, int i, CharSequence seq) { 4178 switch (type) { 4179 case GREEDY: 4180 return (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)) 4181 || next.match(matcher, i, seq); 4182 case LAZY: 4183 return next.match(matcher, i, seq) 4184 || (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)); 4185 case POSSESSIVE: 4186 if (atom.match(matcher, i, seq)) i = matcher.last; 4187 return next.match(matcher, i, seq); 4188 default: 4189 return atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq); 4190 } 4191 } 4192 boolean study(TreeInfo info) { 4193 if (type != INDEPENDENT) { 4194 int minL = info.minLength; 4195 atom.study(info); 4196 info.minLength = minL; 4197 info.deterministic = false; 4198 return next.study(info); 4199 } else { 4200 atom.study(info); 4201 return next.study(info); 4202 } 4203 } 4204 } 4205 4206 /** 4207 * Handles the curly-brace style repetition with a specified minimum and 4208 * maximum occurrences. The * quantifier is handled as a special case. 4209 * This class handles the three types. 4210 */ 4211 static final class Curly extends Node { 4212 Node atom; 4213 int type; 4214 int cmin; 4215 int cmax; 4216 4217 Curly(Node node, int cmin, int cmax, int type) { 4218 this.atom = node; 4219 this.type = type; 4220 this.cmin = cmin; 4221 this.cmax = cmax; 4222 } 4223 boolean match(Matcher matcher, int i, CharSequence seq) { 4224 int j; 4225 for (j = 0; j < cmin; j++) { 4226 if (atom.match(matcher, i, seq)) { 4227 i = matcher.last; 4228 continue; 4229 } 4230 return false; 4231 } 4232 if (type == GREEDY) 4233 return match0(matcher, i, j, seq); 4234 else if (type == LAZY) 4235 return match1(matcher, i, j, seq); 4236 else 4237 return match2(matcher, i, j, seq); 4238 } 4239 // Greedy match. 4240 // i is the index to start matching at 4241 // j is the number of atoms that have matched 4242 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4243 if (j >= cmax) { 4244 // We have matched the maximum... continue with the rest of 4245 // the regular expression 4246 return next.match(matcher, i, seq); 4247 } 4248 int backLimit = j; 4249 while (atom.match(matcher, i, seq)) { 4250 // k is the length of this match 4251 int k = matcher.last - i; 4252 if (k == 0) // Zero length match 4253 break; 4254 // Move up index and number matched 4255 i = matcher.last; 4256 j++; 4257 // We are greedy so match as many as we can 4258 while (j < cmax) { 4259 if (!atom.match(matcher, i, seq)) 4260 break; 4261 if (i + k != matcher.last) { 4262 if (match0(matcher, matcher.last, j+1, seq)) 4263 return true; 4264 break; 4265 } 4266 i += k; 4267 j++; 4268 } 4269 // Handle backing off if match fails 4270 while (j >= backLimit) { 4271 if (next.match(matcher, i, seq)) 4272 return true; 4273 i -= k; 4274 j--; 4275 } 4276 return false; 4277 } 4278 return next.match(matcher, i, seq); 4279 } 4280 // Reluctant match. At this point, the minimum has been satisfied. 4281 // i is the index to start matching at 4282 // j is the number of atoms that have matched 4283 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4284 for (;;) { 4285 // Try finishing match without consuming any more 4286 if (next.match(matcher, i, seq)) 4287 return true; 4288 // At the maximum, no match found 4289 if (j >= cmax) 4290 return false; 4291 // Okay, must try one more atom 4292 if (!atom.match(matcher, i, seq)) 4293 return false; 4294 // If we haven't moved forward then must break out 4295 if (i == matcher.last) 4296 return false; 4297 // Move up index and number matched 4298 i = matcher.last; 4299 j++; 4300 } 4301 } 4302 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4303 for (; j < cmax; j++) { 4304 if (!atom.match(matcher, i, seq)) 4305 break; 4306 if (i == matcher.last) 4307 break; 4308 i = matcher.last; 4309 } 4310 return next.match(matcher, i, seq); 4311 } 4312 boolean study(TreeInfo info) { 4313 // Save original info 4314 int minL = info.minLength; 4315 int maxL = info.maxLength; 4316 boolean maxV = info.maxValid; 4317 boolean detm = info.deterministic; 4318 info.reset(); 4319 4320 atom.study(info); 4321 4322 int temp = info.minLength * cmin + minL; 4323 if (temp < minL) { 4324 temp = 0xFFFFFFF; // arbitrary large number 4325 } 4326 info.minLength = temp; 4327 4328 if (maxV & info.maxValid) { 4329 temp = info.maxLength * cmax + maxL; 4330 info.maxLength = temp; 4331 if (temp < maxL) { 4332 info.maxValid = false; 4333 } 4334 } else { 4335 info.maxValid = false; 4336 } 4337 4338 if (info.deterministic && cmin == cmax) 4339 info.deterministic = detm; 4340 else 4341 info.deterministic = false; 4342 return next.study(info); 4343 } 4344 } 4345 4346 /** 4347 * Handles the curly-brace style repetition with a specified minimum and 4348 * maximum occurrences in deterministic cases. This is an iterative 4349 * optimization over the Prolog and Loop system which would handle this 4350 * in a recursive way. The * quantifier is handled as a special case. 4351 * If capture is true then this class saves group settings and ensures 4352 * that groups are unset when backing off of a group match. 4353 */ 4354 static final class GroupCurly extends Node { 4355 Node atom; 4356 int type; 4357 int cmin; 4358 int cmax; 4359 int localIndex; 4360 int groupIndex; 4361 boolean capture; 4362 4363 GroupCurly(Node node, int cmin, int cmax, int type, int local, 4364 int group, boolean capture) { 4365 this.atom = node; 4366 this.type = type; 4367 this.cmin = cmin; 4368 this.cmax = cmax; 4369 this.localIndex = local; 4370 this.groupIndex = group; 4371 this.capture = capture; 4372 } 4373 boolean match(Matcher matcher, int i, CharSequence seq) { 4374 int[] groups = matcher.groups; 4375 int[] locals = matcher.locals; 4376 int save0 = locals[localIndex]; 4377 int save1 = 0; 4378 int save2 = 0; 4379 4380 if (capture) { 4381 save1 = groups[groupIndex]; 4382 save2 = groups[groupIndex+1]; 4383 } 4384 4385 // Notify GroupTail there is no need to setup group info 4386 // because it will be set here 4387 locals[localIndex] = -1; 4388 4389 boolean ret = true; 4390 for (int j = 0; j < cmin; j++) { 4391 if (atom.match(matcher, i, seq)) { 4392 if (capture) { 4393 groups[groupIndex] = i; 4394 groups[groupIndex+1] = matcher.last; 4395 } 4396 i = matcher.last; 4397 } else { 4398 ret = false; 4399 break; 4400 } 4401 } 4402 if (ret) { 4403 if (type == GREEDY) { 4404 ret = match0(matcher, i, cmin, seq); 4405 } else if (type == LAZY) { 4406 ret = match1(matcher, i, cmin, seq); 4407 } else { 4408 ret = match2(matcher, i, cmin, seq); 4409 } 4410 } 4411 if (!ret) { 4412 locals[localIndex] = save0; 4413 if (capture) { 4414 groups[groupIndex] = save1; 4415 groups[groupIndex+1] = save2; 4416 } 4417 } 4418 return ret; 4419 } 4420 // Aggressive group match 4421 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 4422 // don't back off passing the starting "j" 4423 int min = j; 4424 int[] groups = matcher.groups; 4425 int save0 = 0; 4426 int save1 = 0; 4427 if (capture) { 4428 save0 = groups[groupIndex]; 4429 save1 = groups[groupIndex+1]; 4430 } 4431 for (;;) { 4432 if (j >= cmax) 4433 break; 4434 if (!atom.match(matcher, i, seq)) 4435 break; 4436 int k = matcher.last - i; 4437 if (k <= 0) { 4438 if (capture) { 4439 groups[groupIndex] = i; 4440 groups[groupIndex+1] = i + k; 4441 } 4442 i = i + k; 4443 break; 4444 } 4445 for (;;) { 4446 if (capture) { 4447 groups[groupIndex] = i; 4448 groups[groupIndex+1] = i + k; 4449 } 4450 i = i + k; 4451 if (++j >= cmax) 4452 break; 4453 if (!atom.match(matcher, i, seq)) 4454 break; 4455 if (i + k != matcher.last) { 4456 if (match0(matcher, i, j, seq)) 4457 return true; 4458 break; 4459 } 4460 } 4461 while (j > min) { 4462 if (next.match(matcher, i, seq)) { 4463 if (capture) { 4464 groups[groupIndex+1] = i; 4465 groups[groupIndex] = i - k; 4466 } 4467 i = i - k; 4468 return true; 4469 } 4470 // backing off 4471 if (capture) { 4472 groups[groupIndex+1] = i; 4473 groups[groupIndex] = i - k; 4474 } 4475 i = i - k; 4476 j--; 4477 } 4478 break; 4479 } 4480 if (capture) { 4481 groups[groupIndex] = save0; 4482 groups[groupIndex+1] = save1; 4483 } 4484 return next.match(matcher, i, seq); 4485 } 4486 // Reluctant matching 4487 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4488 for (;;) { 4489 if (next.match(matcher, i, seq)) 4490 return true; 4491 if (j >= cmax) 4492 return false; 4493 if (!atom.match(matcher, i, seq)) 4494 return false; 4495 if (i == matcher.last) 4496 return false; 4497 if (capture) { 4498 matcher.groups[groupIndex] = i; 4499 matcher.groups[groupIndex+1] = matcher.last; 4500 } 4501 i = matcher.last; 4502 j++; 4503 } 4504 } 4505 // Possessive matching 4506 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4507 for (; j < cmax; j++) { 4508 if (!atom.match(matcher, i, seq)) { 4509 break; 4510 } 4511 if (capture) { 4512 matcher.groups[groupIndex] = i; 4513 matcher.groups[groupIndex+1] = matcher.last; 4514 } 4515 if (i == matcher.last) { 4516 break; 4517 } 4518 i = matcher.last; 4519 } 4520 return next.match(matcher, i, seq); 4521 } 4522 boolean study(TreeInfo info) { 4523 // Save original info 4524 int minL = info.minLength; 4525 int maxL = info.maxLength; 4526 boolean maxV = info.maxValid; 4527 boolean detm = info.deterministic; 4528 info.reset(); 4529 4530 atom.study(info); 4531 4532 int temp = info.minLength * cmin + minL; 4533 if (temp < minL) { 4534 temp = 0xFFFFFFF; // Arbitrary large number 4535 } 4536 info.minLength = temp; 4537 4538 if (maxV & info.maxValid) { 4539 temp = info.maxLength * cmax + maxL; 4540 info.maxLength = temp; 4541 if (temp < maxL) { 4542 info.maxValid = false; 4543 } 4544 } else { 4545 info.maxValid = false; 4546 } 4547 4548 if (info.deterministic && cmin == cmax) { 4549 info.deterministic = detm; 4550 } else { 4551 info.deterministic = false; 4552 } 4553 return next.study(info); 4554 } 4555 } 4556 4557 /** 4558 * A Guard node at the end of each atom node in a Branch. It 4559 * serves the purpose of chaining the "match" operation to 4560 * "next" but not the "study", so we can collect the TreeInfo 4561 * of each atom node without including the TreeInfo of the 4562 * "next". 4563 */ 4564 static final class BranchConn extends Node { 4565 BranchConn() {}; 4566 boolean match(Matcher matcher, int i, CharSequence seq) { 4567 return next.match(matcher, i, seq); 4568 } 4569 boolean study(TreeInfo info) { 4570 return info.deterministic; 4571 } 4572 } 4573 4574 /** 4575 * Handles the branching of alternations. Note this is also used for 4576 * the ? quantifier to branch between the case where it matches once 4577 * and where it does not occur. 4578 */ 4579 static final class Branch extends Node { 4580 Node[] atoms = new Node[2]; 4581 int size = 2; 4582 Node conn; 4583 Branch(Node first, Node second, Node branchConn) { 4584 conn = branchConn; 4585 atoms[0] = first; 4586 atoms[1] = second; 4587 } 4588 4589 void add(Node node) { 4590 if (size >= atoms.length) { 4591 Node[] tmp = new Node[atoms.length*2]; 4592 System.arraycopy(atoms, 0, tmp, 0, atoms.length); 4593 atoms = tmp; 4594 } 4595 atoms[size++] = node; 4596 } 4597 4598 boolean match(Matcher matcher, int i, CharSequence seq) { 4599 for (int n = 0; n < size; n++) { 4600 if (atoms[n] == null) { 4601 if (conn.next.match(matcher, i, seq)) 4602 return true; 4603 } else if (atoms[n].match(matcher, i, seq)) { 4604 return true; 4605 } 4606 } 4607 return false; 4608 } 4609 4610 boolean study(TreeInfo info) { 4611 int minL = info.minLength; 4612 int maxL = info.maxLength; 4613 boolean maxV = info.maxValid; 4614 4615 int minL2 = Integer.MAX_VALUE; //arbitrary large enough num 4616 int maxL2 = -1; 4617 for (int n = 0; n < size; n++) { 4618 info.reset(); 4619 if (atoms[n] != null) 4620 atoms[n].study(info); 4621 minL2 = Math.min(minL2, info.minLength); 4622 maxL2 = Math.max(maxL2, info.maxLength); 4623 maxV = (maxV & info.maxValid); 4624 } 4625 4626 minL += minL2; 4627 maxL += maxL2; 4628 4629 info.reset(); 4630 conn.next.study(info); 4631 4632 info.minLength += minL; 4633 info.maxLength += maxL; 4634 info.maxValid &= maxV; 4635 info.deterministic = false; 4636 return false; 4637 } 4638 } 4639 4640 /** 4641 * The GroupHead saves the location where the group begins in the locals 4642 * and restores them when the match is done. 4643 * 4644 * The matchRef is used when a reference to this group is accessed later 4645 * in the expression. The locals will have a negative value in them to 4646 * indicate that we do not want to unset the group if the reference 4647 * doesn't match. 4648 */ 4649 static final class GroupHead extends Node { 4650 int localIndex; 4651 GroupHead(int localCount) { 4652 localIndex = localCount; 4653 } 4654 boolean match(Matcher matcher, int i, CharSequence seq) { 4655 int save = matcher.locals[localIndex]; 4656 matcher.locals[localIndex] = i; 4657 boolean ret = next.match(matcher, i, seq); 4658 matcher.locals[localIndex] = save; 4659 return ret; 4660 } 4661 boolean matchRef(Matcher matcher, int i, CharSequence seq) { 4662 int save = matcher.locals[localIndex]; 4663 matcher.locals[localIndex] = ~i; // HACK 4664 boolean ret = next.match(matcher, i, seq); 4665 matcher.locals[localIndex] = save; 4666 return ret; 4667 } 4668 } 4669 4670 /** 4671 * Recursive reference to a group in the regular expression. It calls 4672 * matchRef because if the reference fails to match we would not unset 4673 * the group. 4674 */ 4675 static final class GroupRef extends Node { 4676 GroupHead head; 4677 GroupRef(GroupHead head) { 4678 this.head = head; 4679 } 4680 boolean match(Matcher matcher, int i, CharSequence seq) { 4681 return head.matchRef(matcher, i, seq) 4682 && next.match(matcher, matcher.last, seq); 4683 } 4684 boolean study(TreeInfo info) { 4685 info.maxValid = false; 4686 info.deterministic = false; 4687 return next.study(info); 4688 } 4689 } 4690 4691 /** 4692 * The GroupTail handles the setting of group beginning and ending 4693 * locations when groups are successfully matched. It must also be able to 4694 * unset groups that have to be backed off of. 4695 * 4696 * The GroupTail node is also used when a previous group is referenced, 4697 * and in that case no group information needs to be set. 4698 */ 4699 static final class GroupTail extends Node { 4700 int localIndex; 4701 int groupIndex; 4702 GroupTail(int localCount, int groupCount) { 4703 localIndex = localCount; 4704 groupIndex = groupCount + groupCount; 4705 } 4706 boolean match(Matcher matcher, int i, CharSequence seq) { 4707 int tmp = matcher.locals[localIndex]; 4708 if (tmp >= 0) { // This is the normal group case. 4709 // Save the group so we can unset it if it 4710 // backs off of a match. 4711 int groupStart = matcher.groups[groupIndex]; 4712 int groupEnd = matcher.groups[groupIndex+1]; 4713 4714 matcher.groups[groupIndex] = tmp; 4715 matcher.groups[groupIndex+1] = i; 4716 if (next.match(matcher, i, seq)) { 4717 return true; 4718 } 4719 matcher.groups[groupIndex] = groupStart; 4720 matcher.groups[groupIndex+1] = groupEnd; 4721 return false; 4722 } else { 4723 // This is a group reference case. We don't need to save any 4724 // group info because it isn't really a group. 4725 matcher.last = i; 4726 return true; 4727 } 4728 } 4729 } 4730 4731 /** 4732 * This sets up a loop to handle a recursive quantifier structure. 4733 */ 4734 static final class Prolog extends Node { 4735 Loop loop; 4736 Prolog(Loop loop) { 4737 this.loop = loop; 4738 } 4739 boolean match(Matcher matcher, int i, CharSequence seq) { 4740 return loop.matchInit(matcher, i, seq); 4741 } 4742 boolean study(TreeInfo info) { 4743 return loop.study(info); 4744 } 4745 } 4746 4747 /** 4748 * Handles the repetition count for a greedy Curly. The matchInit 4749 * is called from the Prolog to save the index of where the group 4750 * beginning is stored. A zero length group check occurs in the 4751 * normal match but is skipped in the matchInit. 4752 */ 4753 static class Loop extends Node { 4754 Node body; 4755 int countIndex; // local count index in matcher locals 4756 int beginIndex; // group beginning index 4757 int cmin, cmax; 4758 Loop(int countIndex, int beginIndex) { 4759 this.countIndex = countIndex; 4760 this.beginIndex = beginIndex; 4761 } 4762 boolean match(Matcher matcher, int i, CharSequence seq) { 4763 // Avoid infinite loop in zero-length case. 4764 if (i > matcher.locals[beginIndex]) { 4765 int count = matcher.locals[countIndex]; 4766 4767 // This block is for before we reach the minimum 4768 // iterations required for the loop to match 4769 if (count < cmin) { 4770 matcher.locals[countIndex] = count + 1; 4771 boolean b = body.match(matcher, i, seq); 4772 // If match failed we must backtrack, so 4773 // the loop count should NOT be incremented 4774 if (!b) 4775 matcher.locals[countIndex] = count; 4776 // Return success or failure since we are under 4777 // minimum 4778 return b; 4779 } 4780 // This block is for after we have the minimum 4781 // iterations required for the loop to match 4782 if (count < cmax) { 4783 matcher.locals[countIndex] = count + 1; 4784 boolean b = body.match(matcher, i, seq); 4785 // If match failed we must backtrack, so 4786 // the loop count should NOT be incremented 4787 if (!b) 4788 matcher.locals[countIndex] = count; 4789 else 4790 return true; 4791 } 4792 } 4793 return next.match(matcher, i, seq); 4794 } 4795 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4796 int save = matcher.locals[countIndex]; 4797 boolean ret = false; 4798 if (0 < cmin) { 4799 matcher.locals[countIndex] = 1; 4800 ret = body.match(matcher, i, seq); 4801 } else if (0 < cmax) { 4802 matcher.locals[countIndex] = 1; 4803 ret = body.match(matcher, i, seq); 4804 if (ret == false) 4805 ret = next.match(matcher, i, seq); 4806 } else { 4807 ret = next.match(matcher, i, seq); 4808 } 4809 matcher.locals[countIndex] = save; 4810 return ret; 4811 } 4812 boolean study(TreeInfo info) { 4813 info.maxValid = false; 4814 info.deterministic = false; 4815 return false; 4816 } 4817 } 4818 4819 /** 4820 * Handles the repetition count for a reluctant Curly. The matchInit 4821 * is called from the Prolog to save the index of where the group 4822 * beginning is stored. A zero length group check occurs in the 4823 * normal match but is skipped in the matchInit. 4824 */ 4825 static final class LazyLoop extends Loop { 4826 LazyLoop(int countIndex, int beginIndex) { 4827 super(countIndex, beginIndex); 4828 } 4829 boolean match(Matcher matcher, int i, CharSequence seq) { 4830 // Check for zero length group 4831 if (i > matcher.locals[beginIndex]) { 4832 int count = matcher.locals[countIndex]; 4833 if (count < cmin) { 4834 matcher.locals[countIndex] = count + 1; 4835 boolean result = body.match(matcher, i, seq); 4836 // If match failed we must backtrack, so 4837 // the loop count should NOT be incremented 4838 if (!result) 4839 matcher.locals[countIndex] = count; 4840 return result; 4841 } 4842 if (next.match(matcher, i, seq)) 4843 return true; 4844 if (count < cmax) { 4845 matcher.locals[countIndex] = count + 1; 4846 boolean result = body.match(matcher, i, seq); 4847 // If match failed we must backtrack, so 4848 // the loop count should NOT be incremented 4849 if (!result) 4850 matcher.locals[countIndex] = count; 4851 return result; 4852 } 4853 return false; 4854 } 4855 return next.match(matcher, i, seq); 4856 } 4857 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4858 int save = matcher.locals[countIndex]; 4859 boolean ret = false; 4860 if (0 < cmin) { 4861 matcher.locals[countIndex] = 1; 4862 ret = body.match(matcher, i, seq); 4863 } else if (next.match(matcher, i, seq)) { 4864 ret = true; 4865 } else if (0 < cmax) { 4866 matcher.locals[countIndex] = 1; 4867 ret = body.match(matcher, i, seq); 4868 } 4869 matcher.locals[countIndex] = save; 4870 return ret; 4871 } 4872 boolean study(TreeInfo info) { 4873 info.maxValid = false; 4874 info.deterministic = false; 4875 return false; 4876 } 4877 } 4878 4879 /** 4880 * Refers to a group in the regular expression. Attempts to match 4881 * whatever the group referred to last matched. 4882 */ 4883 static class BackRef extends Node { 4884 int groupIndex; 4885 BackRef(int groupCount) { 4886 super(); 4887 groupIndex = groupCount + groupCount; 4888 } 4889 boolean match(Matcher matcher, int i, CharSequence seq) { 4890 int j = matcher.groups[groupIndex]; 4891 int k = matcher.groups[groupIndex+1]; 4892 4893 int groupSize = k - j; 4894 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 4905 // Check each new char to make sure it matches what the group 4906 // referenced matched last time around 4907 for (int index=0; index<groupSize; index++) 4908 if (seq.charAt(i+index) != seq.charAt(j+index)) 4909 return false; 4910 4911 return next.match(matcher, i+groupSize, seq); 4912 } 4913 boolean study(TreeInfo info) { 4914 info.maxValid = false; 4915 return next.study(info); 4916 } 4917 } 4918 4919 static class CIBackRef extends Node { 4920 int groupIndex; 4921 boolean doUnicodeCase; 4922 CIBackRef(int groupCount, boolean doUnicodeCase) { 4923 super(); 4924 groupIndex = groupCount + groupCount; 4925 this.doUnicodeCase = doUnicodeCase; 4926 } 4927 boolean match(Matcher matcher, int i, CharSequence seq) { 4928 int j = matcher.groups[groupIndex]; 4929 int k = matcher.groups[groupIndex+1]; 4930 4931 int groupSize = k - j; 4932 4933 // If the referenced group didn't match, neither can this 4934 if (j < 0) 4935 return false; 4936 4937 // If there isn't enough input left no match 4938 if (i + groupSize > matcher.to) { 4939 matcher.hitEnd = true; 4940 return false; 4941 } 4942 4943 // Check each new char to make sure it matches what the group 4944 // referenced matched last time around 4945 int x = i; 4946 for (int index=0; index<groupSize; index++) { 4947 int c1 = Character.codePointAt(seq, x); 4948 int c2 = Character.codePointAt(seq, j); 4949 if (c1 != c2) { 4950 if (doUnicodeCase) { 4951 int cc1 = Character.toUpperCase(c1); 4952 int cc2 = Character.toUpperCase(c2); 4953 if (cc1 != cc2 && 4954 Character.toLowerCase(cc1) != 4955 Character.toLowerCase(cc2)) 4956 return false; 4957 } else { 4958 if (ASCII.toLower(c1) != ASCII.toLower(c2)) 4959 return false; 4960 } 4961 } 4962 x += Character.charCount(c1); 4963 j += Character.charCount(c2); 4964 } 4965 4966 return next.match(matcher, i+groupSize, seq); 4967 } 4968 boolean study(TreeInfo info) { 4969 info.maxValid = false; 4970 return next.study(info); 4971 } 4972 } 4973 4974 /** 4975 * Searches until the next instance of its atom. This is useful for 4976 * finding the atom efficiently without passing an instance of it 4977 * (greedy problem) and without a lot of wasted search time (reluctant 4978 * problem). 4979 */ 4980 static final class First extends Node { 4981 Node atom; 4982 First(Node node) { 4983 this.atom = BnM.optimize(node); 4984 } 4985 boolean match(Matcher matcher, int i, CharSequence seq) { 4986 if (atom instanceof BnM) { 4987 return atom.match(matcher, i, seq) 4988 && next.match(matcher, matcher.last, seq); 4989 } 4990 for (;;) { 4991 if (i > matcher.to) { 4992 matcher.hitEnd = true; 4993 return false; 4994 } 4995 if (atom.match(matcher, i, seq)) { 4996 return next.match(matcher, matcher.last, seq); 4997 } 4998 i += countChars(seq, i, 1); 4999 matcher.first++; 5000 } 5001 } 5002 boolean study(TreeInfo info) { 5003 atom.study(info); 5004 info.maxValid = false; 5005 info.deterministic = false; 5006 return next.study(info); 5007 } 5008 } 5009 5010 static final class Conditional extends Node { 5011 Node cond, yes, not; 5012 Conditional(Node cond, Node yes, Node not) { 5013 this.cond = cond; 5014 this.yes = yes; 5015 this.not = not; 5016 } 5017 boolean match(Matcher matcher, int i, CharSequence seq) { 5018 if (cond.match(matcher, i, seq)) { 5019 return yes.match(matcher, i, seq); 5020 } else { 5021 return not.match(matcher, i, seq); 5022 } 5023 } 5024 boolean study(TreeInfo info) { 5025 int minL = info.minLength; 5026 int maxL = info.maxLength; 5027 boolean maxV = info.maxValid; 5028 info.reset(); 5029 yes.study(info); 5030 5031 int minL2 = info.minLength; 5032 int maxL2 = info.maxLength; 5033 boolean maxV2 = info.maxValid; 5034 info.reset(); 5035 not.study(info); 5036 5037 info.minLength = minL + Math.min(minL2, info.minLength); 5038 info.maxLength = maxL + Math.max(maxL2, info.maxLength); 5039 info.maxValid = (maxV & maxV2 & info.maxValid); 5040 info.deterministic = false; 5041 return next.study(info); 5042 } 5043 } 5044 5045 /** 5046 * Zero width positive lookahead. 5047 */ 5048 static final class Pos extends Node { 5049 Node cond; 5050 Pos(Node cond) { 5051 this.cond = cond; 5052 } 5053 boolean match(Matcher matcher, int i, CharSequence seq) { 5054 int savedTo = matcher.to; 5055 boolean conditionMatched = false; 5056 5057 // Relax transparent region boundaries for lookahead 5058 if (matcher.transparentBounds) 5059 matcher.to = matcher.getTextLength(); 5060 try { 5061 conditionMatched = cond.match(matcher, i, seq); 5062 } finally { 5063 // Reinstate region boundaries 5064 matcher.to = savedTo; 5065 } 5066 return conditionMatched && next.match(matcher, i, seq); 5067 } 5068 } 5069 5070 /** 5071 * Zero width negative lookahead. 5072 */ 5073 static final class Neg extends Node { 5074 Node cond; 5075 Neg(Node cond) { 5076 this.cond = cond; 5077 } 5078 boolean match(Matcher matcher, int i, CharSequence seq) { 5079 int savedTo = matcher.to; 5080 boolean conditionMatched = false; 5081 5082 // Relax transparent region boundaries for lookahead 5083 if (matcher.transparentBounds) 5084 matcher.to = matcher.getTextLength(); 5085 try { 5086 if (i < matcher.to) { 5087 conditionMatched = !cond.match(matcher, i, seq); 5088 } else { 5089 // If a negative lookahead succeeds then more input 5090 // could cause it to fail! 5091 matcher.requireEnd = true; 5092 conditionMatched = !cond.match(matcher, i, seq); 5093 } 5094 } finally { 5095 // Reinstate region boundaries 5096 matcher.to = savedTo; 5097 } 5098 return conditionMatched && next.match(matcher, i, seq); 5099 } 5100 } 5101 5102 /** 5103 * For use with lookbehinds; matches the position where the lookbehind 5104 * was encountered. 5105 */ 5106 static Node lookbehindEnd = new Node() { 5107 boolean match(Matcher matcher, int i, CharSequence seq) { 5108 return i == matcher.lookbehindTo; 5109 } 5110 }; 5111 5112 /** 5113 * Zero width positive lookbehind. 5114 */ 5115 static class Behind extends Node { 5116 Node cond; 5117 int rmax, rmin; 5118 Behind(Node cond, int rmax, int rmin) { 5119 this.cond = cond; 5120 this.rmax = rmax; 5121 this.rmin = rmin; 5122 } 5123 5124 boolean match(Matcher matcher, int i, CharSequence seq) { 5125 int savedFrom = matcher.from; 5126 boolean conditionMatched = false; 5127 int startIndex = (!matcher.transparentBounds) ? 5128 matcher.from : 0; 5129 int from = Math.max(i - rmax, startIndex); 5130 // Set end boundary 5131 int savedLBT = matcher.lookbehindTo; 5132 matcher.lookbehindTo = i; 5133 // Relax transparent region boundaries for lookbehind 5134 if (matcher.transparentBounds) 5135 matcher.from = 0; 5136 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 5137 conditionMatched = cond.match(matcher, j, seq); 5138 } 5139 matcher.from = savedFrom; 5140 matcher.lookbehindTo = savedLBT; 5141 return conditionMatched && next.match(matcher, i, seq); 5142 } 5143 } 5144 5145 /** 5146 * Zero width positive lookbehind, including supplementary 5147 * characters or unpaired surrogates. 5148 */ 5149 static final class BehindS extends Behind { 5150 BehindS(Node cond, int rmax, int rmin) { 5151 super(cond, rmax, rmin); 5152 } 5153 boolean match(Matcher matcher, int i, CharSequence seq) { 5154 int rmaxChars = countChars(seq, i, -rmax); 5155 int rminChars = countChars(seq, i, -rmin); 5156 int savedFrom = matcher.from; 5157 int startIndex = (!matcher.transparentBounds) ? 5158 matcher.from : 0; 5159 boolean conditionMatched = false; 5160 int from = Math.max(i - rmaxChars, startIndex); 5161 // Set end boundary 5162 int savedLBT = matcher.lookbehindTo; 5163 matcher.lookbehindTo = i; 5164 // Relax transparent region boundaries for lookbehind 5165 if (matcher.transparentBounds) 5166 matcher.from = 0; 5167 5168 for (int j = i - rminChars; 5169 !conditionMatched && j >= from; 5170 j -= j>from ? countChars(seq, j, -1) : 1) { 5171 conditionMatched = cond.match(matcher, j, seq); 5172 } 5173 matcher.from = savedFrom; 5174 matcher.lookbehindTo = savedLBT; 5175 return conditionMatched && next.match(matcher, i, seq); 5176 } 5177 } 5178 5179 /** 5180 * Zero width negative lookbehind. 5181 */ 5182 static class NotBehind extends Node { 5183 Node cond; 5184 int rmax, rmin; 5185 NotBehind(Node cond, int rmax, int rmin) { 5186 this.cond = cond; 5187 this.rmax = rmax; 5188 this.rmin = rmin; 5189 } 5190 5191 boolean match(Matcher matcher, int i, CharSequence seq) { 5192 int savedLBT = matcher.lookbehindTo; 5193 int savedFrom = matcher.from; 5194 boolean conditionMatched = false; 5195 int startIndex = (!matcher.transparentBounds) ? 5196 matcher.from : 0; 5197 int from = Math.max(i - rmax, startIndex); 5198 matcher.lookbehindTo = i; 5199 // Relax transparent region boundaries for lookbehind 5200 if (matcher.transparentBounds) 5201 matcher.from = 0; 5202 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 5203 conditionMatched = cond.match(matcher, j, seq); 5204 } 5205 // Reinstate region boundaries 5206 matcher.from = savedFrom; 5207 matcher.lookbehindTo = savedLBT; 5208 return !conditionMatched && next.match(matcher, i, seq); 5209 } 5210 } 5211 5212 /** 5213 * Zero width negative lookbehind, including supplementary 5214 * characters or unpaired surrogates. 5215 */ 5216 static final class NotBehindS extends NotBehind { 5217 NotBehindS(Node cond, int rmax, int rmin) { 5218 super(cond, rmax, rmin); 5219 } 5220 boolean match(Matcher matcher, int i, CharSequence seq) { 5221 int rmaxChars = countChars(seq, i, -rmax); 5222 int rminChars = countChars(seq, i, -rmin); 5223 int savedFrom = matcher.from; 5224 int savedLBT = matcher.lookbehindTo; 5225 boolean conditionMatched = false; 5226 int startIndex = (!matcher.transparentBounds) ? 5227 matcher.from : 0; 5228 int from = Math.max(i - rmaxChars, startIndex); 5229 matcher.lookbehindTo = i; 5230 // Relax transparent region boundaries for lookbehind 5231 if (matcher.transparentBounds) 5232 matcher.from = 0; 5233 for (int j = i - rminChars; 5234 !conditionMatched && j >= from; 5235 j -= j>from ? countChars(seq, j, -1) : 1) { 5236 conditionMatched = cond.match(matcher, j, seq); 5237 } 5238 //Reinstate region boundaries 5239 matcher.from = savedFrom; 5240 matcher.lookbehindTo = savedLBT; 5241 return !conditionMatched && next.match(matcher, i, seq); 5242 } 5243 } 5244 5245 /** 5246 * Returns the set union of two CharProperty nodes. 5247 */ 5248 private static CharProperty union(final CharProperty lhs, 5249 final CharProperty rhs) { 5250 return new CharProperty() { 5251 boolean isSatisfiedBy(int ch) { 5252 return lhs.isSatisfiedBy(ch) || rhs.isSatisfiedBy(ch);}}; 5253 } 5254 5255 /** 5256 * Returns the set intersection of two CharProperty nodes. 5257 */ 5258 private static CharProperty intersection(final CharProperty lhs, 5259 final CharProperty rhs) { 5260 return new CharProperty() { 5261 boolean isSatisfiedBy(int ch) { 5262 return lhs.isSatisfiedBy(ch) && rhs.isSatisfiedBy(ch);}}; 5263 } 5264 5265 /** 5266 * Returns the set difference of two CharProperty nodes. 5267 */ 5268 private static CharProperty setDifference(final CharProperty lhs, 5269 final CharProperty rhs) { 5270 return new CharProperty() { 5271 boolean isSatisfiedBy(int ch) { 5272 return ! rhs.isSatisfiedBy(ch) && lhs.isSatisfiedBy(ch);}}; 5273 } 5274 5275 /** 5276 * Handles word boundaries. Includes a field to allow this one class to 5277 * deal with the different types of word boundaries we can match. The word 5278 * characters include underscores, letters, and digits. Non spacing marks 5279 * can are also part of a word if they have a base character, otherwise 5280 * they are ignored for purposes of finding word boundaries. 5281 */ 5282 static final class Bound extends Node { 5283 static int LEFT = 0x1; 5284 static int RIGHT= 0x2; 5285 static int BOTH = 0x3; 5286 static int NONE = 0x4; 5287 int type; 5288 boolean useUWORD; 5289 Bound(int n, boolean useUWORD) { 5290 type = n; 5291 this.useUWORD = useUWORD; 5292 } 5293 5294 boolean isWord(int ch) { 5295 return useUWORD ? UnicodeProp.WORD.is(ch) 5296 : (ch == '_' || Character.isLetterOrDigit(ch)); 5297 } 5298 5299 int check(Matcher matcher, int i, CharSequence seq) { 5300 int ch; 5301 boolean left = false; 5302 int startIndex = matcher.from; 5303 int endIndex = matcher.to; 5304 if (matcher.transparentBounds) { 5305 startIndex = 0; 5306 endIndex = matcher.getTextLength(); 5307 } 5308 if (i > startIndex) { 5309 ch = Character.codePointBefore(seq, i); 5310 left = (isWord(ch) || 5311 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5312 && hasBaseCharacter(matcher, i-1, seq))); 5313 } 5314 boolean right = false; 5315 if (i < endIndex) { 5316 ch = Character.codePointAt(seq, i); 5317 right = (isWord(ch) || 5318 ((Character.getType(ch) == Character.NON_SPACING_MARK) 5319 && hasBaseCharacter(matcher, i, seq))); 5320 } else { 5321 // Tried to access char past the end 5322 matcher.hitEnd = true; 5323 // The addition of another char could wreck a boundary 5324 matcher.requireEnd = true; 5325 } 5326 return ((left ^ right) ? (right ? LEFT : RIGHT) : NONE); 5327 } 5328 boolean match(Matcher matcher, int i, CharSequence seq) { 5329 return (check(matcher, i, seq) & type) > 0 5330 && next.match(matcher, i, seq); 5331 } 5332 } 5333 5334 /** 5335 * Non spacing marks only count as word characters in bounds calculations 5336 * if they have a base character. 5337 */ 5338 private static boolean hasBaseCharacter(Matcher matcher, int i, 5339 CharSequence seq) 5340 { 5341 int start = (!matcher.transparentBounds) ? 5342 matcher.from : 0; 5343 for (int x=i; x >= start; x--) { 5344 int ch = Character.codePointAt(seq, x); 5345 if (Character.isLetterOrDigit(ch)) 5346 return true; 5347 if (Character.getType(ch) == Character.NON_SPACING_MARK) 5348 continue; 5349 return false; 5350 } 5351 return false; 5352 } 5353 5354 /** 5355 * Attempts to match a slice in the input using the Boyer-Moore string 5356 * matching algorithm. The algorithm is based on the idea that the 5357 * pattern can be shifted farther ahead in the search text if it is 5358 * matched right to left. 5359 * <p> 5360 * The pattern is compared to the input one character at a time, from 5361 * the rightmost character in the pattern to the left. If the characters 5362 * all match the pattern has been found. If a character does not match, 5363 * the pattern is shifted right a distance that is the maximum of two 5364 * functions, the bad character shift and the good suffix shift. This 5365 * shift moves the attempted match position through the input more 5366 * quickly than a naive one position at a time check. 5367 * <p> 5368 * The bad character shift is based on the character from the text that 5369 * did not match. If the character does not appear in the pattern, the 5370 * pattern can be shifted completely beyond the bad character. If the 5371 * character does occur in the pattern, the pattern can be shifted to 5372 * line the pattern up with the next occurrence of that character. 5373 * <p> 5374 * The good suffix shift is based on the idea that some subset on the right 5375 * side of the pattern has matched. When a bad character is found, the 5376 * pattern can be shifted right by the pattern length if the subset does 5377 * not occur again in pattern, or by the amount of distance to the 5378 * next occurrence of the subset in the pattern. 5379 * 5380 * Boyer-Moore search methods adapted from code by Amy Yu. 5381 */ 5382 static class BnM extends Node { 5383 int[] buffer; 5384 int[] lastOcc; 5385 int[] optoSft; 5386 5387 /** 5388 * Pre calculates arrays needed to generate the bad character 5389 * shift and the good suffix shift. Only the last seven bits 5390 * are used to see if chars match; This keeps the tables small 5391 * and covers the heavily used ASCII range, but occasionally 5392 * results in an aliased match for the bad character shift. 5393 */ 5394 static Node optimize(Node node) { 5395 if (!(node instanceof Slice)) { 5396 return node; 5397 } 5398 5399 int[] src = ((Slice) node).buffer; 5400 int patternLength = src.length; 5401 // The BM algorithm requires a bit of overhead; 5402 // If the pattern is short don't use it, since 5403 // a shift larger than the pattern length cannot 5404 // be used anyway. 5405 if (patternLength < 4) { 5406 return node; 5407 } 5408 int i, j, k; 5409 int[] lastOcc = new int[128]; 5410 int[] optoSft = new int[patternLength]; 5411 // Precalculate part of the bad character shift 5412 // It is a table for where in the pattern each 5413 // lower 7-bit value occurs 5414 for (i = 0; i < patternLength; i++) { 5415 lastOcc[src[i]&0x7F] = i + 1; 5416 } 5417 // Precalculate the good suffix shift 5418 // i is the shift amount being considered 5419 NEXT: for (i = patternLength; i > 0; i--) { 5420 // j is the beginning index of suffix being considered 5421 for (j = patternLength - 1; j >= i; j--) { 5422 // Testing for good suffix 5423 if (src[j] == src[j-i]) { 5424 // src[j..len] is a good suffix 5425 optoSft[j-1] = i; 5426 } else { 5427 // No match. The array has already been 5428 // filled up with correct values before. 5429 continue NEXT; 5430 } 5431 } 5432 // This fills up the remaining of optoSft 5433 // any suffix can not have larger shift amount 5434 // then its sub-suffix. Why??? 5435 while (j > 0) { 5436 optoSft[--j] = i; 5437 } 5438 } 5439 // Set the guard value because of unicode compression 5440 optoSft[patternLength-1] = 1; 5441 if (node instanceof SliceS) 5442 return new BnMS(src, lastOcc, optoSft, node.next); 5443 return new BnM(src, lastOcc, optoSft, node.next); 5444 } 5445 BnM(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5446 this.buffer = src; 5447 this.lastOcc = lastOcc; 5448 this.optoSft = optoSft; 5449 this.next = next; 5450 } 5451 boolean match(Matcher matcher, int i, CharSequence seq) { 5452 int[] src = buffer; 5453 int patternLength = src.length; 5454 int last = matcher.to - patternLength; 5455 5456 // Loop over all possible match positions in text 5457 NEXT: while (i <= last) { 5458 // Loop over pattern from right to left 5459 for (int j = patternLength - 1; j >= 0; j--) { 5460 int ch = seq.charAt(i+j); 5461 if (ch != src[j]) { 5462 // Shift search to the right by the maximum of the 5463 // bad character shift and the good suffix shift 5464 i += Math.max(j + 1 - lastOcc[ch&0x7F], optoSft[j]); 5465 continue NEXT; 5466 } 5467 } 5468 // Entire pattern matched starting at i 5469 matcher.first = i; 5470 boolean ret = next.match(matcher, i + patternLength, seq); 5471 if (ret) { 5472 matcher.first = i; 5473 matcher.groups[0] = matcher.first; 5474 matcher.groups[1] = matcher.last; 5475 return true; 5476 } 5477 i++; 5478 } 5479 // BnM is only used as the leading node in the unanchored case, 5480 // and it replaced its Start() which always searches to the end 5481 // if it doesn't find what it's looking for, so hitEnd is true. 5482 matcher.hitEnd = true; 5483 return false; 5484 } 5485 boolean study(TreeInfo info) { 5486 info.minLength += buffer.length; 5487 info.maxValid = false; 5488 return next.study(info); 5489 } 5490 } 5491 5492 /** 5493 * Supplementary support version of BnM(). Unpaired surrogates are 5494 * also handled by this class. 5495 */ 5496 static final class BnMS extends BnM { 5497 int lengthInChars; 5498 5499 BnMS(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5500 super(src, lastOcc, optoSft, next); 5501 for (int x = 0; x < buffer.length; x++) { 5502 lengthInChars += Character.charCount(buffer[x]); 5503 } 5504 } 5505 boolean match(Matcher matcher, int i, CharSequence seq) { 5506 int[] src = buffer; 5507 int patternLength = src.length; 5508 int last = matcher.to - lengthInChars; 5509 5510 // Loop over all possible match positions in text 5511 NEXT: while (i <= last) { 5512 // Loop over pattern from right to left 5513 int ch; 5514 for (int j = countChars(seq, i, patternLength), x = patternLength - 1; 5515 j > 0; j -= Character.charCount(ch), x--) { 5516 ch = Character.codePointBefore(seq, i+j); 5517 if (ch != src[x]) { 5518 // Shift search to the right by the maximum of the 5519 // bad character shift and the good suffix shift 5520 int n = Math.max(x + 1 - lastOcc[ch&0x7F], optoSft[x]); 5521 i += countChars(seq, i, n); 5522 continue NEXT; 5523 } 5524 } 5525 // Entire pattern matched starting at i 5526 matcher.first = i; 5527 boolean ret = next.match(matcher, i + lengthInChars, seq); 5528 if (ret) { 5529 matcher.first = i; 5530 matcher.groups[0] = matcher.first; 5531 matcher.groups[1] = matcher.last; 5532 return true; 5533 } 5534 i += countChars(seq, i, 1); 5535 } 5536 matcher.hitEnd = true; 5537 return false; 5538 } 5539 } 5540 5541 /////////////////////////////////////////////////////////////////////////////// 5542 /////////////////////////////////////////////////////////////////////////////// 5543 5544 /** 5545 * This must be the very first initializer. 5546 */ 5547 static Node accept = new Node(); 5548 5549 static Node lastAccept = new LastNode(); 5550 5551 private static class CharPropertyNames { 5552 5553 static CharProperty charPropertyFor(String name) { 5554 CharPropertyFactory m = map.get(name); 5555 return m == null ? null : m.make(); 5556 } 5557 5558 private static abstract class CharPropertyFactory { 5559 abstract CharProperty make(); 5560 } 5561 5562 private static void defCategory(String name, 5563 final int typeMask) { 5564 map.put(name, new CharPropertyFactory() { 5565 CharProperty make() { return new Category(typeMask);}}); 5566 } 5567 5568 private static void defRange(String name, 5569 final int lower, final int upper) { 5570 map.put(name, new CharPropertyFactory() { 5571 CharProperty make() { return rangeFor(lower, upper);}}); 5572 } 5573 5574 private static void defCtype(String name, 5575 final int ctype) { 5576 map.put(name, new CharPropertyFactory() { 5577 CharProperty make() { return new Ctype(ctype);}}); 5578 } 5579 5580 private static abstract class CloneableProperty 5581 extends CharProperty implements Cloneable 5582 { 5583 public CloneableProperty clone() { 5584 try { 5585 return (CloneableProperty) super.clone(); 5586 } catch (CloneNotSupportedException e) { 5587 throw new AssertionError(e); 5588 } 5589 } 5590 } 5591 5592 private static void defClone(String name, 5593 final CloneableProperty p) { 5594 map.put(name, new CharPropertyFactory() { 5595 CharProperty make() { return p.clone();}}); 5596 } 5597 5598 private static final HashMap<String, CharPropertyFactory> map 5599 = new HashMap<>(); 5600 5601 static { 5602 // Unicode character property aliases, defined in 5603 // http://www.unicode.org/Public/UNIDATA/PropertyValueAliases.txt 5604 defCategory("Cn", 1<<Character.UNASSIGNED); 5605 defCategory("Lu", 1<<Character.UPPERCASE_LETTER); 5606 defCategory("Ll", 1<<Character.LOWERCASE_LETTER); 5607 defCategory("Lt", 1<<Character.TITLECASE_LETTER); 5608 defCategory("Lm", 1<<Character.MODIFIER_LETTER); 5609 defCategory("Lo", 1<<Character.OTHER_LETTER); 5610 defCategory("Mn", 1<<Character.NON_SPACING_MARK); 5611 defCategory("Me", 1<<Character.ENCLOSING_MARK); 5612 defCategory("Mc", 1<<Character.COMBINING_SPACING_MARK); 5613 defCategory("Nd", 1<<Character.DECIMAL_DIGIT_NUMBER); 5614 defCategory("Nl", 1<<Character.LETTER_NUMBER); 5615 defCategory("No", 1<<Character.OTHER_NUMBER); 5616 defCategory("Zs", 1<<Character.SPACE_SEPARATOR); 5617 defCategory("Zl", 1<<Character.LINE_SEPARATOR); 5618 defCategory("Zp", 1<<Character.PARAGRAPH_SEPARATOR); 5619 defCategory("Cc", 1<<Character.CONTROL); 5620 defCategory("Cf", 1<<Character.FORMAT); 5621 defCategory("Co", 1<<Character.PRIVATE_USE); 5622 defCategory("Cs", 1<<Character.SURROGATE); 5623 defCategory("Pd", 1<<Character.DASH_PUNCTUATION); 5624 defCategory("Ps", 1<<Character.START_PUNCTUATION); 5625 defCategory("Pe", 1<<Character.END_PUNCTUATION); 5626 defCategory("Pc", 1<<Character.CONNECTOR_PUNCTUATION); 5627 defCategory("Po", 1<<Character.OTHER_PUNCTUATION); 5628 defCategory("Sm", 1<<Character.MATH_SYMBOL); 5629 defCategory("Sc", 1<<Character.CURRENCY_SYMBOL); 5630 defCategory("Sk", 1<<Character.MODIFIER_SYMBOL); 5631 defCategory("So", 1<<Character.OTHER_SYMBOL); 5632 defCategory("Pi", 1<<Character.INITIAL_QUOTE_PUNCTUATION); 5633 defCategory("Pf", 1<<Character.FINAL_QUOTE_PUNCTUATION); 5634 defCategory("L", ((1<<Character.UPPERCASE_LETTER) | 5635 (1<<Character.LOWERCASE_LETTER) | 5636 (1<<Character.TITLECASE_LETTER) | 5637 (1<<Character.MODIFIER_LETTER) | 5638 (1<<Character.OTHER_LETTER))); 5639 defCategory("M", ((1<<Character.NON_SPACING_MARK) | 5640 (1<<Character.ENCLOSING_MARK) | 5641 (1<<Character.COMBINING_SPACING_MARK))); 5642 defCategory("N", ((1<<Character.DECIMAL_DIGIT_NUMBER) | 5643 (1<<Character.LETTER_NUMBER) | 5644 (1<<Character.OTHER_NUMBER))); 5645 defCategory("Z", ((1<<Character.SPACE_SEPARATOR) | 5646 (1<<Character.LINE_SEPARATOR) | 5647 (1<<Character.PARAGRAPH_SEPARATOR))); 5648 defCategory("C", ((1<<Character.CONTROL) | 5649 (1<<Character.FORMAT) | 5650 (1<<Character.PRIVATE_USE) | 5651 (1<<Character.SURROGATE))); // Other 5652 defCategory("P", ((1<<Character.DASH_PUNCTUATION) | 5653 (1<<Character.START_PUNCTUATION) | 5654 (1<<Character.END_PUNCTUATION) | 5655 (1<<Character.CONNECTOR_PUNCTUATION) | 5656 (1<<Character.OTHER_PUNCTUATION) | 5657 (1<<Character.INITIAL_QUOTE_PUNCTUATION) | 5658 (1<<Character.FINAL_QUOTE_PUNCTUATION))); 5659 defCategory("S", ((1<<Character.MATH_SYMBOL) | 5660 (1<<Character.CURRENCY_SYMBOL) | 5661 (1<<Character.MODIFIER_SYMBOL) | 5662 (1<<Character.OTHER_SYMBOL))); 5663 defCategory("LC", ((1<<Character.UPPERCASE_LETTER) | 5664 (1<<Character.LOWERCASE_LETTER) | 5665 (1<<Character.TITLECASE_LETTER))); 5666 defCategory("LD", ((1<<Character.UPPERCASE_LETTER) | 5667 (1<<Character.LOWERCASE_LETTER) | 5668 (1<<Character.TITLECASE_LETTER) | 5669 (1<<Character.MODIFIER_LETTER) | 5670 (1<<Character.OTHER_LETTER) | 5671 (1<<Character.DECIMAL_DIGIT_NUMBER))); 5672 defRange("L1", 0x00, 0xFF); // Latin-1 5673 map.put("all", new CharPropertyFactory() { 5674 CharProperty make() { return new All(); }}); 5675 5676 // Posix regular expression character classes, defined in 5677 // http://www.unix.org/onlinepubs/009695399/basedefs/xbd_chap09.html 5678 defRange("ASCII", 0x00, 0x7F); // ASCII 5679 defCtype("Alnum", ASCII.ALNUM); // Alphanumeric characters 5680 defCtype("Alpha", ASCII.ALPHA); // Alphabetic characters 5681 defCtype("Blank", ASCII.BLANK); // Space and tab characters 5682 defCtype("Cntrl", ASCII.CNTRL); // Control characters 5683 defRange("Digit", '0', '9'); // Numeric characters 5684 defCtype("Graph", ASCII.GRAPH); // printable and visible 5685 defRange("Lower", 'a', 'z'); // Lower-case alphabetic 5686 defRange("Print", 0x20, 0x7E); // Printable characters 5687 defCtype("Punct", ASCII.PUNCT); // Punctuation characters 5688 defCtype("Space", ASCII.SPACE); // Space characters 5689 defRange("Upper", 'A', 'Z'); // Upper-case alphabetic 5690 defCtype("XDigit",ASCII.XDIGIT); // hexadecimal digits 5691 5692 // Java character properties, defined by methods in Character.java 5693 defClone("javaLowerCase", new CloneableProperty() { 5694 boolean isSatisfiedBy(int ch) { 5695 return Character.isLowerCase(ch);}}); 5696 defClone("javaUpperCase", new CloneableProperty() { 5697 boolean isSatisfiedBy(int ch) { 5698 return Character.isUpperCase(ch);}}); 5699 defClone("javaAlphabetic", new CloneableProperty() { 5700 boolean isSatisfiedBy(int ch) { 5701 return Character.isAlphabetic(ch);}}); 5702 defClone("javaIdeographic", new CloneableProperty() { 5703 boolean isSatisfiedBy(int ch) { 5704 return Character.isIdeographic(ch);}}); 5705 defClone("javaTitleCase", new CloneableProperty() { 5706 boolean isSatisfiedBy(int ch) { 5707 return Character.isTitleCase(ch);}}); 5708 defClone("javaDigit", new CloneableProperty() { 5709 boolean isSatisfiedBy(int ch) { 5710 return Character.isDigit(ch);}}); 5711 defClone("javaDefined", new CloneableProperty() { 5712 boolean isSatisfiedBy(int ch) { 5713 return Character.isDefined(ch);}}); 5714 defClone("javaLetter", new CloneableProperty() { 5715 boolean isSatisfiedBy(int ch) { 5716 return Character.isLetter(ch);}}); 5717 defClone("javaLetterOrDigit", new CloneableProperty() { 5718 boolean isSatisfiedBy(int ch) { 5719 return Character.isLetterOrDigit(ch);}}); 5720 defClone("javaJavaIdentifierStart", new CloneableProperty() { 5721 boolean isSatisfiedBy(int ch) { 5722 return Character.isJavaIdentifierStart(ch);}}); 5723 defClone("javaJavaIdentifierPart", new CloneableProperty() { 5724 boolean isSatisfiedBy(int ch) { 5725 return Character.isJavaIdentifierPart(ch);}}); 5726 defClone("javaUnicodeIdentifierStart", new CloneableProperty() { 5727 boolean isSatisfiedBy(int ch) { 5728 return Character.isUnicodeIdentifierStart(ch);}}); 5729 defClone("javaUnicodeIdentifierPart", new CloneableProperty() { 5730 boolean isSatisfiedBy(int ch) { 5731 return Character.isUnicodeIdentifierPart(ch);}}); 5732 defClone("javaIdentifierIgnorable", new CloneableProperty() { 5733 boolean isSatisfiedBy(int ch) { 5734 return Character.isIdentifierIgnorable(ch);}}); 5735 defClone("javaSpaceChar", new CloneableProperty() { 5736 boolean isSatisfiedBy(int ch) { 5737 return Character.isSpaceChar(ch);}}); 5738 defClone("javaWhitespace", new CloneableProperty() { 5739 boolean isSatisfiedBy(int ch) { 5740 return Character.isWhitespace(ch);}}); 5741 defClone("javaISOControl", new CloneableProperty() { 5742 boolean isSatisfiedBy(int ch) { 5743 return Character.isISOControl(ch);}}); 5744 defClone("javaMirrored", new CloneableProperty() { 5745 boolean isSatisfiedBy(int ch) { 5746 return Character.isMirrored(ch);}}); 5747 } 5748 } 5749 5750 /** 5751 * Creates a predicate which can be used to match a string. 5752 * 5753 * @return The predicate which can be used for matching on a string 5754 * @since 1.8 5755 */ 5756 public Predicate<String> asPredicate() { 5757 return s -> matcher(s).find(); 5758 } 5759 5760 /** 5761 * Creates a stream from the given input sequence around matches of this 5762 * pattern. 5763 * 5764 * <p> The stream returned by this method contains each substring of the 5765 * input sequence that is terminated by another subsequence that matches 5766 * this pattern or is terminated by the end of the input sequence. The 5767 * substrings in the stream are in the order in which they occur in the 5768 * input. 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> If the input sequence is mutable, it must remain constant during the 5775 * execution of the terminal stream operation. Otherwise, the result of the 5776 * terminal stream operation is undefined. 5777 * 5778 * @param input 5779 * The character sequence to be split 5780 * 5781 * @return The stream of strings computed by splitting the input 5782 * around matches of this pattern 5783 * @see #split(CharSequence) 5784 * @since 1.8 5785 */ 5786 public Stream<String> splitAsStream(final CharSequence input) { 5787 class MatcherIterator implements Iterator<String> { 5788 private final Matcher matcher; 5789 // The start position of the next sub-sequence of input 5790 // when current == input.length there are no more elements 5791 private int current; 5792 // null if the next element, if any, needs to obtained 5793 private String nextElement; 5794 5795 MatcherIterator() { 5796 this.matcher = matcher(input); 5797 } 5798 5799 public String next() { 5800 if (!hasNext()) 5801 throw new NoSuchElementException(); 5802 5803 String n = nextElement; 5804 nextElement = null; 5805 return n; 5806 } 5807 5808 public boolean hasNext() { 5809 if (nextElement != null) 5810 return true; 5811 5812 if (current == input.length()) 5813 return false; 5814 5815 if (matcher.find()) { 5816 nextElement = input.subSequence(current, matcher.start()).toString(); 5817 current = matcher.end(); 5818 } else { 5819 nextElement = input.subSequence(current, input.length()).toString(); 5820 current = input.length(); 5821 } 5822 return true; 5823 } 5824 } 5825 return StreamSupport.stream(Spliterators.spliteratorUnknownSize( 5826 new MatcherIterator(), Spliterator.ORDERED | Spliterator.NONNULL)); 5827 } 5828 }