1 /* 2 * Copyright 1999-2008 Sun Microsystems, Inc. All Rights Reserved. 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4 * 5 * This code is free software; you can redistribute it and/or modify it 6 * under the terms of the GNU General Public License version 2 only, as 7 * published by the Free Software Foundation. Sun designates this 8 * particular file as subject to the "Classpath" exception as provided 9 * by Sun in the LICENSE file that accompanied this code. 10 * 11 * This code is distributed in the hope that it will be useful, but WITHOUT 12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 14 * version 2 for more details (a copy is included in the LICENSE file that 15 * accompanied this code). 16 * 17 * You should have received a copy of the GNU General Public License version 18 * 2 along with this work; if not, write to the Free Software Foundation, 19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. 20 * 21 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, 22 * CA 95054 USA or visit www.sun.com if you need additional information or 23 * have any questions. 24 */ 25 26 package java.util.regex; 27 28 import java.security.AccessController; 29 import java.security.PrivilegedAction; 30 import java.text.CharacterIterator; 31 import java.text.Normalizer; 32 import java.util.ArrayList; 33 import java.util.HashMap; 34 import java.util.Arrays; 35 36 37 /** 38 * A compiled representation of a regular expression. 39 * 40 * <p> A regular expression, specified as a string, must first be compiled into 41 * an instance of this class. The resulting pattern can then be used to create 42 * a {@link Matcher} object that can match arbitrary {@link 43 * java.lang.CharSequence </code>character sequences<code>} against the regular 44 * expression. All of the state involved in performing a match resides in the 45 * matcher, so many matchers can share the same pattern. 46 * 47 * <p> A typical invocation sequence is thus 48 * 49 * <blockquote><pre> 50 * Pattern p = Pattern.{@link #compile compile}("a*b"); 51 * Matcher m = p.{@link #matcher matcher}("aaaaab"); 52 * boolean b = m.{@link Matcher#matches matches}();</pre></blockquote> 53 * 54 * <p> A {@link #matches matches} method is defined by this class as a 55 * convenience for when a regular expression is used just once. This method 56 * compiles an expression and matches an input sequence against it in a single 57 * invocation. The statement 58 * 59 * <blockquote><pre> 60 * boolean b = Pattern.matches("a*b", "aaaaab");</pre></blockquote> 61 * 62 * is equivalent to the three statements above, though for repeated matches it 63 * is less efficient since it does not allow the compiled pattern to be reused. 64 * 65 * <p> Instances of this class are immutable and are safe for use by multiple 66 * concurrent threads. Instances of the {@link Matcher} class are not safe for 67 * such use. 68 * 69 * 70 * <a name="sum"> 71 * <h4> Summary of regular-expression constructs </h4> 72 * 73 * <table border="0" cellpadding="1" cellspacing="0" 74 * summary="Regular expression constructs, and what they match"> 75 * 76 * <tr align="left"> 77 * <th bgcolor="#CCCCFF" align="left" id="construct">Construct</th> 78 * <th bgcolor="#CCCCFF" align="left" id="matches">Matches</th> 79 * </tr> 80 * 81 * <tr><th> </th></tr> 82 * <tr align="left"><th colspan="2" id="characters">Characters</th></tr> 83 * 84 * <tr><td valign="top" headers="construct characters"><i>x</i></td> 85 * <td headers="matches">The character <i>x</i></td></tr> 86 * <tr><td valign="top" headers="construct characters"><tt>\\</tt></td> 87 * <td headers="matches">The backslash character</td></tr> 88 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>n</i></td> 89 * <td headers="matches">The character with octal value <tt>0</tt><i>n</i> 90 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 91 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>nn</i></td> 92 * <td headers="matches">The character with octal value <tt>0</tt><i>nn</i> 93 * (0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 94 * <tr><td valign="top" headers="construct characters"><tt>\0</tt><i>mnn</i></td> 95 * <td headers="matches">The character with octal value <tt>0</tt><i>mnn</i> 96 * (0 <tt><=</tt> <i>m</i> <tt><=</tt> 3, 97 * 0 <tt><=</tt> <i>n</i> <tt><=</tt> 7)</td></tr> 98 * <tr><td valign="top" headers="construct characters"><tt>\x</tt><i>hh</i></td> 99 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hh</i></td></tr> 100 * <tr><td valign="top" headers="construct characters"><tt>\u</tt><i>hhhh</i></td> 101 * <td headers="matches">The character with hexadecimal value <tt>0x</tt><i>hhhh</i></td></tr> 102 * <tr><td valign="top" headers="matches"><tt>\t</tt></td> 103 * <td headers="matches">The tab character (<tt>'\u0009'</tt>)</td></tr> 104 * <tr><td valign="top" headers="construct characters"><tt>\n</tt></td> 105 * <td headers="matches">The newline (line feed) character (<tt>'\u000A'</tt>)</td></tr> 106 * <tr><td valign="top" headers="construct characters"><tt>\r</tt></td> 107 * <td headers="matches">The carriage-return character (<tt>'\u000D'</tt>)</td></tr> 108 * <tr><td valign="top" headers="construct characters"><tt>\f</tt></td> 109 * <td headers="matches">The form-feed character (<tt>'\u000C'</tt>)</td></tr> 110 * <tr><td valign="top" headers="construct characters"><tt>\a</tt></td> 111 * <td headers="matches">The alert (bell) character (<tt>'\u0007'</tt>)</td></tr> 112 * <tr><td valign="top" headers="construct characters"><tt>\e</tt></td> 113 * <td headers="matches">The escape character (<tt>'\u001B'</tt>)</td></tr> 114 * <tr><td valign="top" headers="construct characters"><tt>\c</tt><i>x</i></td> 115 * <td headers="matches">The control character corresponding to <i>x</i></td></tr> 116 * 117 * <tr><th> </th></tr> 118 * <tr align="left"><th colspan="2" id="classes">Character classes</th></tr> 119 * 120 * <tr><td valign="top" headers="construct classes"><tt>[abc]</tt></td> 121 * <td headers="matches"><tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (simple class)</td></tr> 122 * <tr><td valign="top" headers="construct classes"><tt>[^abc]</tt></td> 123 * <td headers="matches">Any character except <tt>a</tt>, <tt>b</tt>, or <tt>c</tt> (negation)</td></tr> 124 * <tr><td valign="top" headers="construct classes"><tt>[a-zA-Z]</tt></td> 125 * <td headers="matches"><tt>a</tt> through <tt>z</tt> 126 * or <tt>A</tt> through <tt>Z</tt>, inclusive (range)</td></tr> 127 * <tr><td valign="top" headers="construct classes"><tt>[a-d[m-p]]</tt></td> 128 * <td headers="matches"><tt>a</tt> through <tt>d</tt>, 129 * or <tt>m</tt> through <tt>p</tt>: <tt>[a-dm-p]</tt> (union)</td></tr> 130 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[def]]</tt></td> 131 * <td headers="matches"><tt>d</tt>, <tt>e</tt>, or <tt>f</tt> (intersection)</tr> 132 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^bc]]</tt></td> 133 * <td headers="matches"><tt>a</tt> through <tt>z</tt>, 134 * except for <tt>b</tt> and <tt>c</tt>: <tt>[ad-z]</tt> (subtraction)</td></tr> 135 * <tr><td valign="top" headers="construct classes"><tt>[a-z&&[^m-p]]</tt></td> 136 * <td headers="matches"><tt>a</tt> through <tt>z</tt>, 137 * and not <tt>m</tt> through <tt>p</tt>: <tt>[a-lq-z]</tt>(subtraction)</td></tr> 138 * <tr><th> </th></tr> 139 * 140 * <tr align="left"><th colspan="2" id="predef">Predefined character classes</th></tr> 141 * 142 * <tr><td valign="top" headers="construct predef"><tt>.</tt></td> 143 * <td headers="matches">Any character (may or may not match <a href="#lt">line terminators</a>)</td></tr> 144 * <tr><td valign="top" headers="construct predef"><tt>\d</tt></td> 145 * <td headers="matches">A digit: <tt>[0-9]</tt></td></tr> 146 * <tr><td valign="top" headers="construct predef"><tt>\D</tt></td> 147 * <td headers="matches">A non-digit: <tt>[^0-9]</tt></td></tr> 148 * <tr><td valign="top" headers="construct predef"><tt>\s</tt></td> 149 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 150 * <tr><td valign="top" headers="construct predef"><tt>\S</tt></td> 151 * <td headers="matches">A non-whitespace character: <tt>[^\s]</tt></td></tr> 152 * <tr><td valign="top" headers="construct predef"><tt>\w</tt></td> 153 * <td headers="matches">A word character: <tt>[a-zA-Z_0-9]</tt></td></tr> 154 * <tr><td valign="top" headers="construct predef"><tt>\W</tt></td> 155 * <td headers="matches">A non-word character: <tt>[^\w]</tt></td></tr> 156 * 157 * <tr><th> </th></tr> 158 * <tr align="left"><th colspan="2" id="posix">POSIX character classes</b> (US-ASCII only)<b></th></tr> 159 * 160 * <tr><td valign="top" headers="construct posix"><tt>\p{Lower}</tt></td> 161 * <td headers="matches">A lower-case alphabetic character: <tt>[a-z]</tt></td></tr> 162 * <tr><td valign="top" headers="construct posix"><tt>\p{Upper}</tt></td> 163 * <td headers="matches">An upper-case alphabetic character:<tt>[A-Z]</tt></td></tr> 164 * <tr><td valign="top" headers="construct posix"><tt>\p{ASCII}</tt></td> 165 * <td headers="matches">All ASCII:<tt>[\x00-\x7F]</tt></td></tr> 166 * <tr><td valign="top" headers="construct posix"><tt>\p{Alpha}</tt></td> 167 * <td headers="matches">An alphabetic character:<tt>[\p{Lower}\p{Upper}]</tt></td></tr> 168 * <tr><td valign="top" headers="construct posix"><tt>\p{Digit}</tt></td> 169 * <td headers="matches">A decimal digit: <tt>[0-9]</tt></td></tr> 170 * <tr><td valign="top" headers="construct posix"><tt>\p{Alnum}</tt></td> 171 * <td headers="matches">An alphanumeric character:<tt>[\p{Alpha}\p{Digit}]</tt></td></tr> 172 * <tr><td valign="top" headers="construct posix"><tt>\p{Punct}</tt></td> 173 * <td headers="matches">Punctuation: One of <tt>!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~</tt></td></tr> 174 * <!-- <tt>[\!"#\$%&'\(\)\*\+,\-\./:;\<=\>\?@\[\\\]\^_`\{\|\}~]</tt> 175 * <tt>[\X21-\X2F\X31-\X40\X5B-\X60\X7B-\X7E]</tt> --> 176 * <tr><td valign="top" headers="construct posix"><tt>\p{Graph}</tt></td> 177 * <td headers="matches">A visible character: <tt>[\p{Alnum}\p{Punct}]</tt></td></tr> 178 * <tr><td valign="top" headers="construct posix"><tt>\p{Print}</tt></td> 179 * <td headers="matches">A printable character: <tt>[\p{Graph}\x20]</tt></td></tr> 180 * <tr><td valign="top" headers="construct posix"><tt>\p{Blank}</tt></td> 181 * <td headers="matches">A space or a tab: <tt>[ \t]</tt></td></tr> 182 * <tr><td valign="top" headers="construct posix"><tt>\p{Cntrl}</tt></td> 183 * <td headers="matches">A control character: <tt>[\x00-\x1F\x7F]</tt></td></tr> 184 * <tr><td valign="top" headers="construct posix"><tt>\p{XDigit}</tt></td> 185 * <td headers="matches">A hexadecimal digit: <tt>[0-9a-fA-F]</tt></td></tr> 186 * <tr><td valign="top" headers="construct posix"><tt>\p{Space}</tt></td> 187 * <td headers="matches">A whitespace character: <tt>[ \t\n\x0B\f\r]</tt></td></tr> 188 * 189 * <tr><th> </th></tr> 190 * <tr align="left"><th colspan="2">java.lang.Character classes (simple <a href="#jcc">java character type</a>)</th></tr> 191 * 192 * <tr><td valign="top"><tt>\p{javaLowerCase}</tt></td> 193 * <td>Equivalent to java.lang.Character.isLowerCase()</td></tr> 194 * <tr><td valign="top"><tt>\p{javaUpperCase}</tt></td> 195 * <td>Equivalent to java.lang.Character.isUpperCase()</td></tr> 196 * <tr><td valign="top"><tt>\p{javaWhitespace}</tt></td> 197 * <td>Equivalent to java.lang.Character.isWhitespace()</td></tr> 198 * <tr><td valign="top"><tt>\p{javaMirrored}</tt></td> 199 * <td>Equivalent to java.lang.Character.isMirrored()</td></tr> 200 * 201 * <tr><th> </th></tr> 202 * <tr align="left"><th colspan="2" id="unicode">Classes for Unicode blocks and categories</th></tr> 203 * 204 * <tr><td valign="top" headers="construct unicode"><tt>\p{InGreek}</tt></td> 205 * <td headers="matches">A character in the Greek block (simple <a href="#ubc">block</a>)</td></tr> 206 * <tr><td valign="top" headers="construct unicode"><tt>\p{Lu}</tt></td> 207 * <td headers="matches">An uppercase letter (simple <a href="#ubc">category</a>)</td></tr> 208 * <tr><td valign="top" headers="construct unicode"><tt>\p{Sc}</tt></td> 209 * <td headers="matches">A currency symbol</td></tr> 210 * <tr><td valign="top" headers="construct unicode"><tt>\P{InGreek}</tt></td> 211 * <td headers="matches">Any character except one in the Greek block (negation)</td></tr> 212 * <tr><td valign="top" headers="construct unicode"><tt>[\p{L}&&[^\p{Lu}]] </tt></td> 213 * <td headers="matches">Any letter except an uppercase letter (subtraction)</td></tr> 214 * 215 * <tr><th> </th></tr> 216 * <tr align="left"><th colspan="2" id="bounds">Boundary matchers</th></tr> 217 * 218 * <tr><td valign="top" headers="construct bounds"><tt>^</tt></td> 219 * <td headers="matches">The beginning of a line</td></tr> 220 * <tr><td valign="top" headers="construct bounds"><tt>$</tt></td> 221 * <td headers="matches">The end of a line</td></tr> 222 * <tr><td valign="top" headers="construct bounds"><tt>\b</tt></td> 223 * <td headers="matches">A word boundary</td></tr> 224 * <tr><td valign="top" headers="construct bounds"><tt>\B</tt></td> 225 * <td headers="matches">A non-word boundary</td></tr> 226 * <tr><td valign="top" headers="construct bounds"><tt>\A</tt></td> 227 * <td headers="matches">The beginning of the input</td></tr> 228 * <tr><td valign="top" headers="construct bounds"><tt>\G</tt></td> 229 * <td headers="matches">The end of the previous match</td></tr> 230 * <tr><td valign="top" headers="construct bounds"><tt>\Z</tt></td> 231 * <td headers="matches">The end of the input but for the final 232 * <a href="#lt">terminator</a>, if any</td></tr> 233 * <tr><td valign="top" headers="construct bounds"><tt>\z</tt></td> 234 * <td headers="matches">The end of the input</td></tr> 235 * 236 * <tr><th> </th></tr> 237 * <tr align="left"><th colspan="2" id="greedy">Greedy quantifiers</th></tr> 238 * 239 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>?</tt></td> 240 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 241 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>*</tt></td> 242 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 243 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>+</tt></td> 244 * <td headers="matches"><i>X</i>, one or more times</td></tr> 245 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>}</tt></td> 246 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 247 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,}</tt></td> 248 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 249 * <tr><td valign="top" headers="construct greedy"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}</tt></td> 250 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 251 * 252 * <tr><th> </th></tr> 253 * <tr align="left"><th colspan="2" id="reluc">Reluctant quantifiers</th></tr> 254 * 255 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>??</tt></td> 256 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 257 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>*?</tt></td> 258 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 259 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>+?</tt></td> 260 * <td headers="matches"><i>X</i>, one or more times</td></tr> 261 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>}?</tt></td> 262 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 263 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,}?</tt></td> 264 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 265 * <tr><td valign="top" headers="construct reluc"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}?</tt></td> 266 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 267 * 268 * <tr><th> </th></tr> 269 * <tr align="left"><th colspan="2" id="poss">Possessive quantifiers</th></tr> 270 * 271 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>?+</tt></td> 272 * <td headers="matches"><i>X</i>, once or not at all</td></tr> 273 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>*+</tt></td> 274 * <td headers="matches"><i>X</i>, zero or more times</td></tr> 275 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>++</tt></td> 276 * <td headers="matches"><i>X</i>, one or more times</td></tr> 277 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>}+</tt></td> 278 * <td headers="matches"><i>X</i>, exactly <i>n</i> times</td></tr> 279 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,}+</tt></td> 280 * <td headers="matches"><i>X</i>, at least <i>n</i> times</td></tr> 281 * <tr><td valign="top" headers="construct poss"><i>X</i><tt>{</tt><i>n</i><tt>,</tt><i>m</i><tt>}+</tt></td> 282 * <td headers="matches"><i>X</i>, at least <i>n</i> but not more than <i>m</i> times</td></tr> 283 * 284 * <tr><th> </th></tr> 285 * <tr align="left"><th colspan="2" id="logical">Logical operators</th></tr> 286 * 287 * <tr><td valign="top" headers="construct logical"><i>XY</i></td> 288 * <td headers="matches"><i>X</i> followed by <i>Y</i></td></tr> 289 * <tr><td valign="top" headers="construct logical"><i>X</i><tt>|</tt><i>Y</i></td> 290 * <td headers="matches">Either <i>X</i> or <i>Y</i></td></tr> 291 * <tr><td valign="top" headers="construct logical"><tt>(</tt><i>X</i><tt>)</tt></td> 292 * <td headers="matches">X, as a <a href="#cg">capturing group</a></td></tr> 293 * 294 * <tr><th> </th></tr> 295 * <tr align="left"><th colspan="2" id="backref">Back references</th></tr> 296 * 297 * <tr><td valign="bottom" headers="construct backref"><tt>\</tt><i>n</i></td> 298 * <td valign="bottom" headers="matches">Whatever the <i>n</i><sup>th</sup> 299 * <a href="#cg">capturing group</a> matched</td></tr> 300 * 301 * <tr><th> </th></tr> 302 * <tr align="left"><th colspan="2" id="quot">Quotation</th></tr> 303 * 304 * <tr><td valign="top" headers="construct quot"><tt>\</tt></td> 305 * <td headers="matches">Nothing, but quotes the following character</td></tr> 306 * <tr><td valign="top" headers="construct quot"><tt>\Q</tt></td> 307 * <td headers="matches">Nothing, but quotes all characters until <tt>\E</tt></td></tr> 308 * <tr><td valign="top" headers="construct quot"><tt>\E</tt></td> 309 * <td headers="matches">Nothing, but ends quoting started by <tt>\Q</tt></td></tr> 310 * <!-- Metachars: !$()*+.<>?[\]^{|} --> 311 * 312 * <tr><th> </th></tr> 313 * <tr align="left"><th colspan="2" id="special">Special constructs (non-capturing)</th></tr> 314 * 315 * <tr><td valign="top" headers="construct special"><tt>(?:</tt><i>X</i><tt>)</tt></td> 316 * <td headers="matches"><i>X</i>, as a non-capturing group</td></tr> 317 * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux) </tt></td> 318 * <td headers="matches">Nothing, but turns match flags <a href="#CASE_INSENSITIVE">i</a> 319 * <a href="#UNIX_LINES">d</a> <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> 320 * <a href="#UNICODE_CASE">u</a> <a href="#COMMENTS">x</a> on - off</td></tr> 321 * <tr><td valign="top" headers="construct special"><tt>(?idmsux-idmsux:</tt><i>X</i><tt>)</tt> </td> 322 * <td headers="matches"><i>X</i>, as a <a href="#cg">non-capturing group</a> with the 323 * given flags <a href="#CASE_INSENSITIVE">i</a> <a href="#UNIX_LINES">d</a> 324 * <a href="#MULTILINE">m</a> <a href="#DOTALL">s</a> <a href="#UNICODE_CASE">u</a > 325 * <a href="#COMMENTS">x</a> on - off</td></tr> 326 * <tr><td valign="top" headers="construct special"><tt>(?=</tt><i>X</i><tt>)</tt></td> 327 * <td headers="matches"><i>X</i>, via zero-width positive lookahead</td></tr> 328 * <tr><td valign="top" headers="construct special"><tt>(?!</tt><i>X</i><tt>)</tt></td> 329 * <td headers="matches"><i>X</i>, via zero-width negative lookahead</td></tr> 330 * <tr><td valign="top" headers="construct special"><tt>(?<=</tt><i>X</i><tt>)</tt></td> 331 * <td headers="matches"><i>X</i>, via zero-width positive lookbehind</td></tr> 332 * <tr><td valign="top" headers="construct special"><tt>(?<!</tt><i>X</i><tt>)</tt></td> 333 * <td headers="matches"><i>X</i>, via zero-width negative lookbehind</td></tr> 334 * <tr><td valign="top" headers="construct special"><tt>(?></tt><i>X</i><tt>)</tt></td> 335 * <td headers="matches"><i>X</i>, as an independent, non-capturing group</td></tr> 336 * 337 * </table> 338 * 339 * <hr> 340 * 341 * 342 * <a name="bs"> 343 * <h4> Backslashes, escapes, and quoting </h4> 344 * 345 * <p> The backslash character (<tt>'\'</tt>) serves to introduce escaped 346 * constructs, as defined in the table above, as well as to quote characters 347 * that otherwise would be interpreted as unescaped constructs. Thus the 348 * expression <tt>\\</tt> matches a single backslash and <tt>\{</tt> matches a 349 * left brace. 350 * 351 * <p> It is an error to use a backslash prior to any alphabetic character that 352 * does not denote an escaped construct; these are reserved for future 353 * extensions to the regular-expression language. A backslash may be used 354 * prior to a non-alphabetic character regardless of whether that character is 355 * part of an unescaped construct. 356 * 357 * <p> Backslashes within string literals in Java source code are interpreted 358 * as required by the <a 359 * href="http://java.sun.com/docs/books/jls">Java Language 360 * Specification</a> as either <a 361 * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#100850">Unicode 362 * escapes</a> or other <a 363 * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#101089">character 364 * escapes</a>. It is therefore necessary to double backslashes in string 365 * literals that represent regular expressions to protect them from 366 * interpretation by the Java bytecode compiler. The string literal 367 * <tt>"\b"</tt>, for example, matches a single backspace character when 368 * interpreted as a regular expression, while <tt>"\\b"</tt> matches a 369 * word boundary. The string literal <tt>"\(hello\)"</tt> is illegal 370 * and leads to a compile-time error; in order to match the string 371 * <tt>(hello)</tt> the string literal <tt>"\\(hello\\)"</tt> 372 * must be used. 373 * 374 * <a name="cc"> 375 * <h4> Character Classes </h4> 376 * 377 * <p> Character classes may appear within other character classes, and 378 * may be composed by the union operator (implicit) and the intersection 379 * operator (<tt>&&</tt>). 380 * The union operator denotes a class that contains every character that is 381 * in at least one of its operand classes. The intersection operator 382 * denotes a class that contains every character that is in both of its 383 * operand classes. 384 * 385 * <p> The precedence of character-class operators is as follows, from 386 * highest to lowest: 387 * 388 * <blockquote><table border="0" cellpadding="1" cellspacing="0" 389 * summary="Precedence of character class operators."> 390 * <tr><th>1 </th> 391 * <td>Literal escape </td> 392 * <td><tt>\x</tt></td></tr> 393 * <tr><th>2 </th> 394 * <td>Grouping</td> 395 * <td><tt>[...]</tt></td></tr> 396 * <tr><th>3 </th> 397 * <td>Range</td> 398 * <td><tt>a-z</tt></td></tr> 399 * <tr><th>4 </th> 400 * <td>Union</td> 401 * <td><tt>[a-e][i-u]</tt></td></tr> 402 * <tr><th>5 </th> 403 * <td>Intersection</td> 404 * <td><tt>[a-z&&[aeiou]]</tt></td></tr> 405 * </table></blockquote> 406 * 407 * <p> Note that a different set of metacharacters are in effect inside 408 * a character class than outside a character class. For instance, the 409 * regular expression <tt>.</tt> loses its special meaning inside a 410 * character class, while the expression <tt>-</tt> becomes a range 411 * forming metacharacter. 412 * 413 * <a name="lt"> 414 * <h4> Line terminators </h4> 415 * 416 * <p> A <i>line terminator</i> is a one- or two-character sequence that marks 417 * the end of a line of the input character sequence. The following are 418 * recognized as line terminators: 419 * 420 * <ul> 421 * 422 * <li> A newline (line feed) character (<tt>'\n'</tt>), 423 * 424 * <li> A carriage-return character followed immediately by a newline 425 * character (<tt>"\r\n"</tt>), 426 * 427 * <li> A standalone carriage-return character (<tt>'\r'</tt>), 428 * 429 * <li> A next-line character (<tt>'\u0085'</tt>), 430 * 431 * <li> A line-separator character (<tt>'\u2028'</tt>), or 432 * 433 * <li> A paragraph-separator character (<tt>'\u2029</tt>). 434 * 435 * </ul> 436 * <p>If {@link #UNIX_LINES} mode is activated, then the only line terminators 437 * recognized are newline characters. 438 * 439 * <p> The regular expression <tt>.</tt> matches any character except a line 440 * terminator unless the {@link #DOTALL} flag is specified. 441 * 442 * <p> By default, the regular expressions <tt>^</tt> and <tt>$</tt> ignore 443 * line terminators and only match at the beginning and the end, respectively, 444 * of the entire input sequence. If {@link #MULTILINE} mode is activated then 445 * <tt>^</tt> matches at the beginning of input and after any line terminator 446 * except at the end of input. When in {@link #MULTILINE} mode <tt>$</tt> 447 * matches just before a line terminator or the end of the input sequence. 448 * 449 * <a name="cg"> 450 * <h4> Groups and capturing </h4> 451 * 452 * <p> Capturing groups are numbered by counting their opening parentheses from 453 * left to right. In the expression <tt>((A)(B(C)))</tt>, for example, there 454 * are four such groups: </p> 455 * 456 * <blockquote><table cellpadding=1 cellspacing=0 summary="Capturing group numberings"> 457 * <tr><th>1 </th> 458 * <td><tt>((A)(B(C)))</tt></td></tr> 459 * <tr><th>2 </th> 460 * <td><tt>(A)</tt></td></tr> 461 * <tr><th>3 </th> 462 * <td><tt>(B(C))</tt></td></tr> 463 * <tr><th>4 </th> 464 * <td><tt>(C)</tt></td></tr> 465 * </table></blockquote> 466 * 467 * <p> Group zero always stands for the entire expression. 468 * 469 * <p> Capturing groups are so named because, during a match, each subsequence 470 * of the input sequence that matches such a group is saved. The captured 471 * subsequence may be used later in the expression, via a back reference, and 472 * may also be retrieved from the matcher once the match operation is complete. 473 * 474 * <p> The captured input associated with a group is always the subsequence 475 * that the group most recently matched. If a group is evaluated a second time 476 * because of quantification then its previously-captured value, if any, will 477 * be retained if the second evaluation fails. Matching the string 478 * <tt>"aba"</tt> against the expression <tt>(a(b)?)+</tt>, for example, leaves 479 * group two set to <tt>"b"</tt>. All captured input is discarded at the 480 * beginning of each match. 481 * 482 * <p> Groups beginning with <tt>(?</tt> are pure, <i>non-capturing</i> groups 483 * that do not capture text and do not count towards the group total. 484 * 485 * 486 * <h4> Unicode support </h4> 487 * 488 * <p> This class is in conformance with Level 1 of <a 489 * href="http://www.unicode.org/reports/tr18/"><i>Unicode Technical 490 * Standard #18: Unicode Regular Expression Guidelines</i></a>, plus RL2.1 491 * Canonical Equivalents. 492 * 493 * <p> Unicode escape sequences such as <tt>\u2014</tt> in Java source code 494 * are processed as described in <a 495 * href="http://java.sun.com/docs/books/jls/third_edition/html/lexical.html#100850">\u00A73.3</a> 496 * of the Java Language Specification. Such escape sequences are also 497 * implemented directly by the regular-expression parser so that Unicode 498 * escapes can be used in expressions that are read from files or from the 499 * keyboard. Thus the strings <tt>"\u2014"</tt> and <tt>"\\u2014"</tt>, 500 * while not equal, compile into the same pattern, which matches the character 501 * with hexadecimal value <tt>0x2014</tt>. 502 * 503 * <a name="ubc"> <p>Unicode blocks and categories are written with the 504 * <tt>\p</tt> and <tt>\P</tt> constructs as in 505 * Perl. <tt>\p{</tt><i>prop</i><tt>}</tt> matches if the input has the 506 * property <i>prop</i>, while <tt>\P{</tt><i>prop</i><tt>}</tt> does not match if 507 * the input has that property. Blocks are specified with the prefix 508 * <tt>In</tt>, as in <tt>InMongolian</tt>. Categories may be specified with 509 * the optional prefix <tt>Is</tt>: Both <tt>\p{L}</tt> and <tt>\p{IsL}</tt> 510 * denote the category of Unicode letters. Blocks and categories can be used 511 * both inside and outside of a character class. 512 * 513 * <p> The supported categories are those of 514 * <a href="http://www.unicode.org/unicode/standard/standard.html"> 515 * <i>The Unicode Standard</i></a> in the version specified by the 516 * {@link java.lang.Character Character} class. The category names are those 517 * defined in the Standard, both normative and informative. 518 * The block names supported by <code>Pattern</code> are the valid block names 519 * accepted and defined by 520 * {@link java.lang.Character.UnicodeBlock#forName(String) UnicodeBlock.forName}. 521 * 522 * <a name="jcc"> <p>Categories that behave like the java.lang.Character 523 * boolean is<i>methodname</i> methods (except for the deprecated ones) are 524 * available through the same <tt>\p{</tt><i>prop</i><tt>}</tt> syntax where 525 * the specified property has the name <tt>java<i>methodname</i></tt>. 526 * 527 * <h4> Comparison to Perl 5 </h4> 528 * 529 * <p>The <code>Pattern</code> engine performs traditional NFA-based matching 530 * with ordered alternation as occurs in Perl 5. 531 * 532 * <p> Perl constructs not supported by this class: </p> 533 * 534 * <ul> 535 * 536 * <li><p> The conditional constructs <tt>(?{</tt><i>X</i><tt>})</tt> and 537 * <tt>(?(</tt><i>condition</i><tt>)</tt><i>X</i><tt>|</tt><i>Y</i><tt>)</tt>, 538 * </p></li> 539 * 540 * <li><p> The embedded code constructs <tt>(?{</tt><i>code</i><tt>})</tt> 541 * and <tt>(??{</tt><i>code</i><tt>})</tt>,</p></li> 542 * 543 * <li><p> The embedded comment syntax <tt>(?#comment)</tt>, and </p></li> 544 * 545 * <li><p> The preprocessing operations <tt>\l</tt> <tt>\u</tt>, 546 * <tt>\L</tt>, and <tt>\U</tt>. </p></li> 547 * 548 * </ul> 549 * 550 * <p> Constructs supported by this class but not by Perl: </p> 551 * 552 * <ul> 553 * 554 * <li><p> Possessive quantifiers, which greedily match as much as they can 555 * and do not back off, even when doing so would allow the overall match to 556 * succeed. </p></li> 557 * 558 * <li><p> Character-class union and intersection as described 559 * <a href="#cc">above</a>.</p></li> 560 * 561 * </ul> 562 * 563 * <p> Notable differences from Perl: </p> 564 * 565 * <ul> 566 * 567 * <li><p> In Perl, <tt>\1</tt> through <tt>\9</tt> are always interpreted 568 * as back references; a backslash-escaped number greater than <tt>9</tt> is 569 * treated as a back reference if at least that many subexpressions exist, 570 * otherwise it is interpreted, if possible, as an octal escape. In this 571 * class octal escapes must always begin with a zero. In this class, 572 * <tt>\1</tt> through <tt>\9</tt> are always interpreted as back 573 * references, and a larger number is accepted as a back reference if at 574 * least that many subexpressions exist at that point in the regular 575 * expression, otherwise the parser will drop digits until the number is 576 * smaller or equal to the existing number of groups or it is one digit. 577 * </p></li> 578 * 579 * <li><p> Perl uses the <tt>g</tt> flag to request a match that resumes 580 * where the last match left off. This functionality is provided implicitly 581 * by the {@link Matcher} class: Repeated invocations of the {@link 582 * Matcher#find find} method will resume where the last match left off, 583 * unless the matcher is reset. </p></li> 584 * 585 * <li><p> In Perl, embedded flags at the top level of an expression affect 586 * the whole expression. In this class, embedded flags always take effect 587 * at the point at which they appear, whether they are at the top level or 588 * within a group; in the latter case, flags are restored at the end of the 589 * group just as in Perl. </p></li> 590 * 591 * <li><p> Perl is forgiving about malformed matching constructs, as in the 592 * expression <tt>*a</tt>, as well as dangling brackets, as in the 593 * expression <tt>abc]</tt>, and treats them as literals. This 594 * class also accepts dangling brackets but is strict about dangling 595 * metacharacters like +, ? and *, and will throw a 596 * {@link PatternSyntaxException} if it encounters them. </p></li> 597 * 598 * </ul> 599 * 600 * 601 * <p> For a more precise description of the behavior of regular expression 602 * constructs, please see <a href="http://www.oreilly.com/catalog/regex3/"> 603 * <i>Mastering Regular Expressions, 3nd Edition</i>, Jeffrey E. F. Friedl, 604 * O'Reilly and Associates, 2006.</a> 605 * </p> 606 * 607 * @see java.lang.String#split(String, int) 608 * @see java.lang.String#split(String) 609 * 610 * @author Mike McCloskey 611 * @author Mark Reinhold 612 * @author JSR-51 Expert Group 613 * @since 1.4 614 * @spec JSR-51 615 */ 616 617 public final class Pattern 618 implements java.io.Serializable 619 { 620 621 /** 622 * Regular expression modifier values. Instead of being passed as 623 * arguments, they can also be passed as inline modifiers. 624 * For example, the following statements have the same effect. 625 * <pre> 626 * RegExp r1 = RegExp.compile("abc", Pattern.I|Pattern.M); 627 * RegExp r2 = RegExp.compile("(?im)abc", 0); 628 * </pre> 629 * 630 * The flags are duplicated so that the familiar Perl match flag 631 * names are available. 632 */ 633 634 /** 635 * Enables Unix lines mode. 636 * 637 * <p> In this mode, only the <tt>'\n'</tt> line terminator is recognized 638 * in the behavior of <tt>.</tt>, <tt>^</tt>, and <tt>$</tt>. 639 * 640 * <p> Unix lines mode can also be enabled via the embedded flag 641 * expression <tt>(?d)</tt>. 642 */ 643 public static final int UNIX_LINES = 0x01; 644 645 /** 646 * Enables case-insensitive matching. 647 * 648 * <p> By default, case-insensitive matching assumes that only characters 649 * in the US-ASCII charset are being matched. Unicode-aware 650 * case-insensitive matching can be enabled by specifying the {@link 651 * #UNICODE_CASE} flag in conjunction with this flag. 652 * 653 * <p> Case-insensitive matching can also be enabled via the embedded flag 654 * expression <tt>(?i)</tt>. 655 * 656 * <p> Specifying this flag may impose a slight performance penalty. </p> 657 */ 658 public static final int CASE_INSENSITIVE = 0x02; 659 660 /** 661 * Permits whitespace and comments in pattern. 662 * 663 * <p> In this mode, whitespace is ignored, and embedded comments starting 664 * with <tt>#</tt> are ignored until the end of a line. 665 * 666 * <p> Comments mode can also be enabled via the embedded flag 667 * expression <tt>(?x)</tt>. 668 */ 669 public static final int COMMENTS = 0x04; 670 671 /** 672 * Enables multiline mode. 673 * 674 * <p> In multiline mode the expressions <tt>^</tt> and <tt>$</tt> match 675 * just after or just before, respectively, a line terminator or the end of 676 * the input sequence. By default these expressions only match at the 677 * beginning and the end of the entire input sequence. 678 * 679 * <p> Multiline mode can also be enabled via the embedded flag 680 * expression <tt>(?m)</tt>. </p> 681 */ 682 public static final int MULTILINE = 0x08; 683 684 /** 685 * Enables literal parsing of the pattern. 686 * 687 * <p> When this flag is specified then the input string that specifies 688 * the pattern is treated as a sequence of literal characters. 689 * Metacharacters or escape sequences in the input sequence will be 690 * given no special meaning. 691 * 692 * <p>The flags CASE_INSENSITIVE and UNICODE_CASE retain their impact on 693 * matching when used in conjunction with this flag. The other flags 694 * become superfluous. 695 * 696 * <p> There is no embedded flag character for enabling literal parsing. 697 * @since 1.5 698 */ 699 public static final int LITERAL = 0x10; 700 701 /** 702 * Enables dotall mode. 703 * 704 * <p> In dotall mode, the expression <tt>.</tt> matches any character, 705 * including a line terminator. By default this expression does not match 706 * line terminators. 707 * 708 * <p> Dotall mode can also be enabled via the embedded flag 709 * expression <tt>(?s)</tt>. (The <tt>s</tt> is a mnemonic for 710 * "single-line" mode, which is what this is called in Perl.) </p> 711 */ 712 public static final int DOTALL = 0x20; 713 714 /** 715 * Enables Unicode-aware case folding. 716 * 717 * <p> When this flag is specified then case-insensitive matching, when 718 * enabled by the {@link #CASE_INSENSITIVE} flag, is done in a manner 719 * consistent with the Unicode Standard. By default, case-insensitive 720 * matching assumes that only characters in the US-ASCII charset are being 721 * matched. 722 * 723 * <p> Unicode-aware case folding can also be enabled via the embedded flag 724 * expression <tt>(?u)</tt>. 725 * 726 * <p> Specifying this flag may impose a performance penalty. </p> 727 */ 728 public static final int UNICODE_CASE = 0x40; 729 730 /** 731 * Enables canonical equivalence. 732 * 733 * <p> When this flag is specified then two characters will be considered 734 * to match if, and only if, their full canonical decompositions match. 735 * The expression <tt>"a\u030A"</tt>, for example, will match the 736 * string <tt>"\u00E5"</tt> when this flag is specified. By default, 737 * matching does not take canonical equivalence into account. 738 * 739 * <p> There is no embedded flag character for enabling canonical 740 * equivalence. 741 * 742 * <p> Specifying this flag may impose a performance penalty. </p> 743 */ 744 public static final int CANON_EQ = 0x80; 745 746 /* Pattern has only two serialized components: The pattern string 747 * and the flags, which are all that is needed to recompile the pattern 748 * when it is deserialized. 749 */ 750 751 /** use serialVersionUID from Merlin b59 for interoperability */ 752 private static final long serialVersionUID = 5073258162644648461L; 753 754 /** 755 * The original regular-expression pattern string. 756 * 757 * @serial 758 */ 759 private String pattern; 760 761 /** 762 * The original pattern flags. 763 * 764 * @serial 765 */ 766 private int flags; 767 768 /** 769 * Boolean indicating this Pattern is compiled; this is necessary in order 770 * to lazily compile deserialized Patterns. 771 */ 772 private transient volatile boolean compiled = false; 773 774 /** 775 * The normalized pattern string. 776 */ 777 private transient String normalizedPattern; 778 779 /** 780 * The starting point of state machine for the find operation. This allows 781 * a match to start anywhere in the input. 782 */ 783 transient Node root; 784 785 /** 786 * The root of object tree for a match operation. The pattern is matched 787 * at the beginning. This may include a find that uses BnM or a First 788 * node. 789 */ 790 transient Node matchRoot; 791 792 /** 793 * Temporary storage used by parsing pattern slice. 794 */ 795 transient int[] buffer; 796 797 /** 798 * Temporary storage used while parsing group references. 799 */ 800 transient GroupHead[] groupNodes; 801 802 /** 803 * Temporary null terminated code point array used by pattern compiling. 804 */ 805 private transient int[] temp; 806 807 /** 808 * The number of capturing groups in this Pattern. Used by matchers to 809 * allocate storage needed to perform a match. 810 */ 811 transient int capturingGroupCount; 812 813 /** 814 * The local variable count used by parsing tree. Used by matchers to 815 * allocate storage needed to perform a match. 816 */ 817 transient int localCount; 818 819 /** 820 * Index into the pattern string that keeps track of how much has been 821 * parsed. 822 */ 823 private transient int cursor; 824 825 /** 826 * Holds the length of the pattern string. 827 */ 828 private transient int patternLength; 829 830 /** 831 * Compiles the given regular expression into a pattern. </p> 832 * 833 * @param regex 834 * The expression to be compiled 835 * 836 * @throws PatternSyntaxException 837 * If the expression's syntax is invalid 838 */ 839 public static Pattern compile(String regex) { 840 return new Pattern(regex, 0); 841 } 842 843 /** 844 * Compiles the given regular expression into a pattern with the given 845 * flags. </p> 846 * 847 * @param regex 848 * The expression to be compiled 849 * 850 * @param flags 851 * Match flags, a bit mask that may include 852 * {@link #CASE_INSENSITIVE}, {@link #MULTILINE}, {@link #DOTALL}, 853 * {@link #UNICODE_CASE}, {@link #CANON_EQ}, {@link #UNIX_LINES}, 854 * {@link #LITERAL} and {@link #COMMENTS} 855 * 856 * @throws IllegalArgumentException 857 * If bit values other than those corresponding to the defined 858 * match flags are set in <tt>flags</tt> 859 * 860 * @throws PatternSyntaxException 861 * If the expression's syntax is invalid 862 */ 863 public static Pattern compile(String regex, int flags) { 864 return new Pattern(regex, flags); 865 } 866 867 /** 868 * Returns the regular expression from which this pattern was compiled. 869 * </p> 870 * 871 * @return The source of this pattern 872 */ 873 public String pattern() { 874 return pattern; 875 } 876 877 /** 878 * <p>Returns the string representation of this pattern. This 879 * is the regular expression from which this pattern was 880 * compiled.</p> 881 * 882 * @return The string representation of this pattern 883 * @since 1.5 884 */ 885 public String toString() { 886 return pattern; 887 } 888 889 /** 890 * Creates a matcher that will match the given input against this pattern. 891 * </p> 892 * 893 * @param input 894 * The character sequence to be matched 895 * 896 * @return A new matcher for this pattern 897 */ 898 public Matcher matcher(CharSequence input) { 899 if (!compiled) { 900 synchronized(this) { 901 if (!compiled) 902 compile(); 903 } 904 } 905 Matcher m = new Matcher(this, input); 906 return m; 907 } 908 909 /** 910 * Returns this pattern's match flags. </p> 911 * 912 * @return The match flags specified when this pattern was compiled 913 */ 914 public int flags() { 915 return flags; 916 } 917 918 /** 919 * Compiles the given regular expression and attempts to match the given 920 * input against it. 921 * 922 * <p> An invocation of this convenience method of the form 923 * 924 * <blockquote><pre> 925 * Pattern.matches(regex, input);</pre></blockquote> 926 * 927 * behaves in exactly the same way as the expression 928 * 929 * <blockquote><pre> 930 * Pattern.compile(regex).matcher(input).matches()</pre></blockquote> 931 * 932 * <p> If a pattern is to be used multiple times, compiling it once and reusing 933 * it will be more efficient than invoking this method each time. </p> 934 * 935 * @param regex 936 * The expression to be compiled 937 * 938 * @param input 939 * The character sequence to be matched 940 * 941 * @throws PatternSyntaxException 942 * If the expression's syntax is invalid 943 */ 944 public static boolean matches(String regex, CharSequence input) { 945 Pattern p = Pattern.compile(regex); 946 Matcher m = p.matcher(input); 947 return m.matches(); 948 } 949 950 /** 951 * Splits the given input sequence around matches of this pattern. 952 * 953 * <p> The array returned by this method contains each substring of the 954 * input sequence that is terminated by another subsequence that matches 955 * this pattern or is terminated by the end of the input sequence. The 956 * substrings in the array are in the order in which they occur in the 957 * input. If this pattern does not match any subsequence of the input then 958 * the resulting array has just one element, namely the input sequence in 959 * string form. 960 * 961 * <p> The <tt>limit</tt> parameter controls the number of times the 962 * pattern is applied and therefore affects the length of the resulting 963 * array. If the limit <i>n</i> is greater than zero then the pattern 964 * will be applied at most <i>n</i> - 1 times, the array's 965 * length will be no greater than <i>n</i>, and the array's last entry 966 * will contain all input beyond the last matched delimiter. If <i>n</i> 967 * is non-positive then the pattern will be applied as many times as 968 * possible and the array can have any length. If <i>n</i> is zero then 969 * the pattern will be applied as many times as possible, the array can 970 * have any length, and trailing empty strings will be discarded. 971 * 972 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 973 * results with these parameters: 974 * 975 * <blockquote><table cellpadding=1 cellspacing=0 976 * summary="Split examples showing regex, limit, and result"> 977 * <tr><th><P align="left"><i>Regex </i></th> 978 * <th><P align="left"><i>Limit </i></th> 979 * <th><P align="left"><i>Result </i></th></tr> 980 * <tr><td align=center>:</td> 981 * <td align=center>2</td> 982 * <td><tt>{ "boo", "and:foo" }</tt></td></tr> 983 * <tr><td align=center>:</td> 984 * <td align=center>5</td> 985 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 986 * <tr><td align=center>:</td> 987 * <td align=center>-2</td> 988 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 989 * <tr><td align=center>o</td> 990 * <td align=center>5</td> 991 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 992 * <tr><td align=center>o</td> 993 * <td align=center>-2</td> 994 * <td><tt>{ "b", "", ":and:f", "", "" }</tt></td></tr> 995 * <tr><td align=center>o</td> 996 * <td align=center>0</td> 997 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 998 * </table></blockquote> 999 * 1000 * 1001 * @param input 1002 * The character sequence to be split 1003 * 1004 * @param limit 1005 * The result threshold, as described above 1006 * 1007 * @return The array of strings computed by splitting the input 1008 * around matches of this pattern 1009 */ 1010 public String[] split(CharSequence input, int limit) { 1011 int index = 0; 1012 boolean matchLimited = limit > 0; 1013 ArrayList<String> matchList = new ArrayList<String>(); 1014 Matcher m = matcher(input); 1015 1016 // Add segments before each match found 1017 while(m.find()) { 1018 if (!matchLimited || matchList.size() < limit - 1) { 1019 String match = input.subSequence(index, m.start()).toString(); 1020 matchList.add(match); 1021 index = m.end(); 1022 } else if (matchList.size() == limit - 1) { // last one 1023 String match = input.subSequence(index, 1024 input.length()).toString(); 1025 matchList.add(match); 1026 index = m.end(); 1027 } 1028 } 1029 1030 // If no match was found, return this 1031 if (index == 0) 1032 return new String[] {input.toString()}; 1033 1034 // Add remaining segment 1035 if (!matchLimited || matchList.size() < limit) 1036 matchList.add(input.subSequence(index, input.length()).toString()); 1037 1038 // Construct result 1039 int resultSize = matchList.size(); 1040 if (limit == 0) 1041 while (resultSize > 0 && matchList.get(resultSize-1).equals("")) 1042 resultSize--; 1043 String[] result = new String[resultSize]; 1044 return matchList.subList(0, resultSize).toArray(result); 1045 } 1046 1047 /** 1048 * Splits the given input sequence around matches of this pattern. 1049 * 1050 * <p> This method works as if by invoking the two-argument {@link 1051 * #split(java.lang.CharSequence, int) split} method with the given input 1052 * sequence and a limit argument of zero. Trailing empty strings are 1053 * therefore not included in the resulting array. </p> 1054 * 1055 * <p> The input <tt>"boo:and:foo"</tt>, for example, yields the following 1056 * results with these expressions: 1057 * 1058 * <blockquote><table cellpadding=1 cellspacing=0 1059 * summary="Split examples showing regex and result"> 1060 * <tr><th><P align="left"><i>Regex </i></th> 1061 * <th><P align="left"><i>Result</i></th></tr> 1062 * <tr><td align=center>:</td> 1063 * <td><tt>{ "boo", "and", "foo" }</tt></td></tr> 1064 * <tr><td align=center>o</td> 1065 * <td><tt>{ "b", "", ":and:f" }</tt></td></tr> 1066 * </table></blockquote> 1067 * 1068 * 1069 * @param input 1070 * The character sequence to be split 1071 * 1072 * @return The array of strings computed by splitting the input 1073 * around matches of this pattern 1074 */ 1075 public String[] split(CharSequence input) { 1076 return split(input, 0); 1077 } 1078 1079 /** 1080 * Returns a literal pattern <code>String</code> for the specified 1081 * <code>String</code>. 1082 * 1083 * <p>This method produces a <code>String</code> that can be used to 1084 * create a <code>Pattern</code> that would match the string 1085 * <code>s</code> as if it were a literal pattern.</p> Metacharacters 1086 * or escape sequences in the input sequence will be given no special 1087 * meaning. 1088 * 1089 * @param s The string to be literalized 1090 * @return A literal string replacement 1091 * @since 1.5 1092 */ 1093 public static String quote(String s) { 1094 int slashEIndex = s.indexOf("\\E"); 1095 if (slashEIndex == -1) 1096 return "\\Q" + s + "\\E"; 1097 1098 StringBuilder sb = new StringBuilder(s.length() * 2); 1099 sb.append("\\Q"); 1100 slashEIndex = 0; 1101 int current = 0; 1102 while ((slashEIndex = s.indexOf("\\E", current)) != -1) { 1103 sb.append(s.substring(current, slashEIndex)); 1104 current = slashEIndex + 2; 1105 sb.append("\\E\\\\E\\Q"); 1106 } 1107 sb.append(s.substring(current, s.length())); 1108 sb.append("\\E"); 1109 return sb.toString(); 1110 } 1111 1112 /** 1113 * Recompile the Pattern instance from a stream. The original pattern 1114 * string is read in and the object tree is recompiled from it. 1115 */ 1116 private void readObject(java.io.ObjectInputStream s) 1117 throws java.io.IOException, ClassNotFoundException { 1118 1119 // Read in all fields 1120 s.defaultReadObject(); 1121 1122 // Initialize counts 1123 capturingGroupCount = 1; 1124 localCount = 0; 1125 1126 // if length > 0, the Pattern is lazily compiled 1127 compiled = false; 1128 if (pattern.length() == 0) { 1129 root = new Start(lastAccept); 1130 matchRoot = lastAccept; 1131 compiled = true; 1132 } 1133 } 1134 1135 /** 1136 * This private constructor is used to create all Patterns. The pattern 1137 * string and match flags are all that is needed to completely describe 1138 * a Pattern. An empty pattern string results in an object tree with 1139 * only a Start node and a LastNode node. 1140 */ 1141 private Pattern(String p, int f) { 1142 pattern = p; 1143 flags = f; 1144 1145 // Reset group index count 1146 capturingGroupCount = 1; 1147 localCount = 0; 1148 1149 if (pattern.length() > 0) { 1150 compile(); 1151 } else { 1152 root = new Start(lastAccept); 1153 matchRoot = lastAccept; 1154 } 1155 } 1156 1157 /** 1158 * The pattern is converted to normalizedD form and then a pure group 1159 * is constructed to match canonical equivalences of the characters. 1160 */ 1161 private void normalize() { 1162 boolean inCharClass = false; 1163 int lastCodePoint = -1; 1164 1165 // Convert pattern into normalizedD form 1166 normalizedPattern = Normalizer.normalize(pattern, Normalizer.Form.NFD); 1167 patternLength = normalizedPattern.length(); 1168 1169 // Modify pattern to match canonical equivalences 1170 StringBuilder newPattern = new StringBuilder(patternLength); 1171 for(int i=0; i<patternLength; ) { 1172 int c = normalizedPattern.codePointAt(i); 1173 StringBuilder sequenceBuffer; 1174 if ((Character.getType(c) == Character.NON_SPACING_MARK) 1175 && (lastCodePoint != -1)) { 1176 sequenceBuffer = new StringBuilder(); 1177 sequenceBuffer.appendCodePoint(lastCodePoint); 1178 sequenceBuffer.appendCodePoint(c); 1179 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1180 i += Character.charCount(c); 1181 if (i >= patternLength) 1182 break; 1183 c = normalizedPattern.codePointAt(i); 1184 sequenceBuffer.appendCodePoint(c); 1185 } 1186 String ea = produceEquivalentAlternation( 1187 sequenceBuffer.toString()); 1188 newPattern.setLength(newPattern.length()-Character.charCount(lastCodePoint)); 1189 newPattern.append("(?:").append(ea).append(")"); 1190 } else if (c == '[' && lastCodePoint != '\\') { 1191 i = normalizeCharClass(newPattern, i); 1192 } else { 1193 newPattern.appendCodePoint(c); 1194 } 1195 lastCodePoint = c; 1196 i += Character.charCount(c); 1197 } 1198 normalizedPattern = newPattern.toString(); 1199 } 1200 1201 /** 1202 * Complete the character class being parsed and add a set 1203 * of alternations to it that will match the canonical equivalences 1204 * of the characters within the class. 1205 */ 1206 private int normalizeCharClass(StringBuilder newPattern, int i) { 1207 StringBuilder charClass = new StringBuilder(); 1208 StringBuilder eq = null; 1209 int lastCodePoint = -1; 1210 String result; 1211 1212 i++; 1213 charClass.append("["); 1214 while(true) { 1215 int c = normalizedPattern.codePointAt(i); 1216 StringBuilder sequenceBuffer; 1217 1218 if (c == ']' && lastCodePoint != '\\') { 1219 charClass.append((char)c); 1220 break; 1221 } else if (Character.getType(c) == Character.NON_SPACING_MARK) { 1222 sequenceBuffer = new StringBuilder(); 1223 sequenceBuffer.appendCodePoint(lastCodePoint); 1224 while(Character.getType(c) == Character.NON_SPACING_MARK) { 1225 sequenceBuffer.appendCodePoint(c); 1226 i += Character.charCount(c); 1227 if (i >= normalizedPattern.length()) 1228 break; 1229 c = normalizedPattern.codePointAt(i); 1230 } 1231 String ea = produceEquivalentAlternation( 1232 sequenceBuffer.toString()); 1233 1234 charClass.setLength(charClass.length()-Character.charCount(lastCodePoint)); 1235 if (eq == null) 1236 eq = new StringBuilder(); 1237 eq.append('|'); 1238 eq.append(ea); 1239 } else { 1240 charClass.appendCodePoint(c); 1241 i++; 1242 } 1243 if (i == normalizedPattern.length()) 1244 throw error("Unclosed character class"); 1245 lastCodePoint = c; 1246 } 1247 1248 if (eq != null) { 1249 result = "(?:"+charClass.toString()+eq.toString()+")"; 1250 } else { 1251 result = charClass.toString(); 1252 } 1253 1254 newPattern.append(result); 1255 return i; 1256 } 1257 1258 /** 1259 * Given a specific sequence composed of a regular character and 1260 * combining marks that follow it, produce the alternation that will 1261 * match all canonical equivalences of that sequence. 1262 */ 1263 private String produceEquivalentAlternation(String source) { 1264 int len = countChars(source, 0, 1); 1265 if (source.length() == len) 1266 // source has one character. 1267 return source; 1268 1269 String base = source.substring(0,len); 1270 String combiningMarks = source.substring(len); 1271 1272 String[] perms = producePermutations(combiningMarks); 1273 StringBuilder result = new StringBuilder(source); 1274 1275 // Add combined permutations 1276 for(int x=0; x<perms.length; x++) { 1277 String next = base + perms[x]; 1278 if (x>0) 1279 result.append("|"+next); 1280 next = composeOneStep(next); 1281 if (next != null) 1282 result.append("|"+produceEquivalentAlternation(next)); 1283 } 1284 return result.toString(); 1285 } 1286 1287 /** 1288 * Returns an array of strings that have all the possible 1289 * permutations of the characters in the input string. 1290 * This is used to get a list of all possible orderings 1291 * of a set of combining marks. Note that some of the permutations 1292 * are invalid because of combining class collisions, and these 1293 * possibilities must be removed because they are not canonically 1294 * equivalent. 1295 */ 1296 private String[] producePermutations(String input) { 1297 if (input.length() == countChars(input, 0, 1)) 1298 return new String[] {input}; 1299 1300 if (input.length() == countChars(input, 0, 2)) { 1301 int c0 = Character.codePointAt(input, 0); 1302 int c1 = Character.codePointAt(input, Character.charCount(c0)); 1303 if (getClass(c1) == getClass(c0)) { 1304 return new String[] {input}; 1305 } 1306 String[] result = new String[2]; 1307 result[0] = input; 1308 StringBuilder sb = new StringBuilder(2); 1309 sb.appendCodePoint(c1); 1310 sb.appendCodePoint(c0); 1311 result[1] = sb.toString(); 1312 return result; 1313 } 1314 1315 int length = 1; 1316 int nCodePoints = countCodePoints(input); 1317 for(int x=1; x<nCodePoints; x++) 1318 length = length * (x+1); 1319 1320 String[] temp = new String[length]; 1321 1322 int combClass[] = new int[nCodePoints]; 1323 for(int x=0, i=0; x<nCodePoints; x++) { 1324 int c = Character.codePointAt(input, i); 1325 combClass[x] = getClass(c); 1326 i += Character.charCount(c); 1327 } 1328 1329 // For each char, take it out and add the permutations 1330 // of the remaining chars 1331 int index = 0; 1332 int len; 1333 // offset maintains the index in code units. 1334 loop: for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) { 1335 len = countChars(input, offset, 1); 1336 boolean skip = false; 1337 for(int y=x-1; y>=0; y--) { 1338 if (combClass[y] == combClass[x]) { 1339 continue loop; 1340 } 1341 } 1342 StringBuilder sb = new StringBuilder(input); 1343 String otherChars = sb.delete(offset, offset+len).toString(); 1344 String[] subResult = producePermutations(otherChars); 1345 1346 String prefix = input.substring(offset, offset+len); 1347 for(int y=0; y<subResult.length; y++) 1348 temp[index++] = prefix + subResult[y]; 1349 } 1350 String[] result = new String[index]; 1351 for (int x=0; x<index; x++) 1352 result[x] = temp[x]; 1353 return result; 1354 } 1355 1356 private int getClass(int c) { 1357 return sun.text.Normalizer.getCombiningClass(c); 1358 } 1359 1360 /** 1361 * Attempts to compose input by combining the first character 1362 * with the first combining mark following it. Returns a String 1363 * that is the composition of the leading character with its first 1364 * combining mark followed by the remaining combining marks. Returns 1365 * null if the first two characters cannot be further composed. 1366 */ 1367 private String composeOneStep(String input) { 1368 int len = countChars(input, 0, 2); 1369 String firstTwoCharacters = input.substring(0, len); 1370 String result = Normalizer.normalize(firstTwoCharacters, Normalizer.Form.NFC); 1371 1372 if (result.equals(firstTwoCharacters)) 1373 return null; 1374 else { 1375 String remainder = input.substring(len); 1376 return result + remainder; 1377 } 1378 } 1379 1380 /** 1381 * Preprocess any \Q...\E sequences in `temp', meta-quoting them. 1382 * See the description of `quotemeta' in perlfunc(1). 1383 */ 1384 private void RemoveQEQuoting() { 1385 final int pLen = patternLength; 1386 int i = 0; 1387 while (i < pLen-1) { 1388 if (temp[i] != '\\') 1389 i += 1; 1390 else if (temp[i + 1] != 'Q') 1391 i += 2; 1392 else 1393 break; 1394 } 1395 if (i >= pLen - 1) // No \Q sequence found 1396 return; 1397 int j = i; 1398 i += 2; 1399 int[] newtemp = new int[j + 2*(pLen-i) + 2]; 1400 System.arraycopy(temp, 0, newtemp, 0, j); 1401 1402 boolean inQuote = true; 1403 while (i < pLen) { 1404 int c = temp[i++]; 1405 if (! ASCII.isAscii(c) || ASCII.isAlnum(c)) { 1406 newtemp[j++] = c; 1407 } else if (c != '\\') { 1408 if (inQuote) newtemp[j++] = '\\'; 1409 newtemp[j++] = c; 1410 } else if (inQuote) { 1411 if (temp[i] == 'E') { 1412 i++; 1413 inQuote = false; 1414 } else { 1415 newtemp[j++] = '\\'; 1416 newtemp[j++] = '\\'; 1417 } 1418 } else { 1419 if (temp[i] == 'Q') { 1420 i++; 1421 inQuote = true; 1422 } else { 1423 newtemp[j++] = c; 1424 if (i != pLen) 1425 newtemp[j++] = temp[i++]; 1426 } 1427 } 1428 } 1429 1430 patternLength = j; 1431 temp = Arrays.copyOf(newtemp, j + 2); // double zero termination 1432 } 1433 1434 /** 1435 * Copies regular expression to an int array and invokes the parsing 1436 * of the expression which will create the object tree. 1437 */ 1438 private void compile() { 1439 // Handle canonical equivalences 1440 if (has(CANON_EQ) && !has(LITERAL)) { 1441 normalize(); 1442 } else { 1443 normalizedPattern = pattern; 1444 } 1445 patternLength = normalizedPattern.length(); 1446 1447 // Copy pattern to int array for convenience 1448 // Use double zero to terminate pattern 1449 temp = new int[patternLength + 2]; 1450 1451 boolean hasSupplementary = false; 1452 int c, count = 0; 1453 // Convert all chars into code points 1454 for (int x = 0; x < patternLength; x += Character.charCount(c)) { 1455 c = normalizedPattern.codePointAt(x); 1456 if (isSupplementary(c)) { 1457 hasSupplementary = true; 1458 } 1459 temp[count++] = c; 1460 } 1461 1462 patternLength = count; // patternLength now in code points 1463 1464 if (! has(LITERAL)) 1465 RemoveQEQuoting(); 1466 1467 // Allocate all temporary objects here. 1468 buffer = new int[32]; 1469 groupNodes = new GroupHead[10]; 1470 1471 if (has(LITERAL)) { 1472 // Literal pattern handling 1473 matchRoot = newSlice(temp, patternLength, hasSupplementary); 1474 matchRoot.next = lastAccept; 1475 } else { 1476 // Start recursive descent parsing 1477 matchRoot = expr(lastAccept); 1478 // Check extra pattern characters 1479 if (patternLength != cursor) { 1480 if (peek() == ')') { 1481 throw error("Unmatched closing ')'"); 1482 } else { 1483 throw error("Unexpected internal error"); 1484 } 1485 } 1486 } 1487 1488 // Peephole optimization 1489 if (matchRoot instanceof Slice) { 1490 root = BnM.optimize(matchRoot); 1491 if (root == matchRoot) { 1492 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1493 } 1494 } else if (matchRoot instanceof Begin || matchRoot instanceof First) { 1495 root = matchRoot; 1496 } else { 1497 root = hasSupplementary ? new StartS(matchRoot) : new Start(matchRoot); 1498 } 1499 1500 // Release temporary storage 1501 temp = null; 1502 buffer = null; 1503 groupNodes = null; 1504 patternLength = 0; 1505 compiled = true; 1506 } 1507 1508 /** 1509 * Used to print out a subtree of the Pattern to help with debugging. 1510 */ 1511 private static void printObjectTree(Node node) { 1512 while(node != null) { 1513 if (node instanceof Prolog) { 1514 System.out.println(node); 1515 printObjectTree(((Prolog)node).loop); 1516 System.out.println("**** end contents prolog loop"); 1517 } else if (node instanceof Loop) { 1518 System.out.println(node); 1519 printObjectTree(((Loop)node).body); 1520 System.out.println("**** end contents Loop body"); 1521 } else if (node instanceof Curly) { 1522 System.out.println(node); 1523 printObjectTree(((Curly)node).atom); 1524 System.out.println("**** end contents Curly body"); 1525 } else if (node instanceof GroupCurly) { 1526 System.out.println(node); 1527 printObjectTree(((GroupCurly)node).atom); 1528 System.out.println("**** end contents GroupCurly body"); 1529 } else if (node instanceof GroupTail) { 1530 System.out.println(node); 1531 System.out.println("Tail next is "+node.next); 1532 return; 1533 } else { 1534 System.out.println(node); 1535 } 1536 node = node.next; 1537 if (node != null) 1538 System.out.println("->next:"); 1539 if (node == Pattern.accept) { 1540 System.out.println("Accept Node"); 1541 node = null; 1542 } 1543 } 1544 } 1545 1546 /** 1547 * Used to accumulate information about a subtree of the object graph 1548 * so that optimizations can be applied to the subtree. 1549 */ 1550 static final class TreeInfo { 1551 int minLength; 1552 int maxLength; 1553 boolean maxValid; 1554 boolean deterministic; 1555 1556 TreeInfo() { 1557 reset(); 1558 } 1559 void reset() { 1560 minLength = 0; 1561 maxLength = 0; 1562 maxValid = true; 1563 deterministic = true; 1564 } 1565 } 1566 1567 /* 1568 * The following private methods are mainly used to improve the 1569 * readability of the code. In order to let the Java compiler easily 1570 * inline them, we should not put many assertions or error checks in them. 1571 */ 1572 1573 /** 1574 * Indicates whether a particular flag is set or not. 1575 */ 1576 private boolean has(int f) { 1577 return (flags & f) != 0; 1578 } 1579 1580 /** 1581 * Match next character, signal error if failed. 1582 */ 1583 private void accept(int ch, String s) { 1584 int testChar = temp[cursor++]; 1585 if (has(COMMENTS)) 1586 testChar = parsePastWhitespace(testChar); 1587 if (ch != testChar) { 1588 throw error(s); 1589 } 1590 } 1591 1592 /** 1593 * Mark the end of pattern with a specific character. 1594 */ 1595 private void mark(int c) { 1596 temp[patternLength] = c; 1597 } 1598 1599 /** 1600 * Peek the next character, and do not advance the cursor. 1601 */ 1602 private int peek() { 1603 int ch = temp[cursor]; 1604 if (has(COMMENTS)) 1605 ch = peekPastWhitespace(ch); 1606 return ch; 1607 } 1608 1609 /** 1610 * Read the next character, and advance the cursor by one. 1611 */ 1612 private int read() { 1613 int ch = temp[cursor++]; 1614 if (has(COMMENTS)) 1615 ch = parsePastWhitespace(ch); 1616 return ch; 1617 } 1618 1619 /** 1620 * Read the next character, and advance the cursor by one, 1621 * ignoring the COMMENTS setting 1622 */ 1623 private int readEscaped() { 1624 int ch = temp[cursor++]; 1625 return ch; 1626 } 1627 1628 /** 1629 * Advance the cursor by one, and peek the next character. 1630 */ 1631 private int next() { 1632 int ch = temp[++cursor]; 1633 if (has(COMMENTS)) 1634 ch = peekPastWhitespace(ch); 1635 return ch; 1636 } 1637 1638 /** 1639 * Advance the cursor by one, and peek the next character, 1640 * ignoring the COMMENTS setting 1641 */ 1642 private int nextEscaped() { 1643 int ch = temp[++cursor]; 1644 return ch; 1645 } 1646 1647 /** 1648 * If in xmode peek past whitespace and comments. 1649 */ 1650 private int peekPastWhitespace(int ch) { 1651 while (ASCII.isSpace(ch) || ch == '#') { 1652 while (ASCII.isSpace(ch)) 1653 ch = temp[++cursor]; 1654 if (ch == '#') { 1655 ch = peekPastLine(); 1656 } 1657 } 1658 return ch; 1659 } 1660 1661 /** 1662 * If in xmode parse past whitespace and comments. 1663 */ 1664 private int parsePastWhitespace(int ch) { 1665 while (ASCII.isSpace(ch) || ch == '#') { 1666 while (ASCII.isSpace(ch)) 1667 ch = temp[cursor++]; 1668 if (ch == '#') 1669 ch = parsePastLine(); 1670 } 1671 return ch; 1672 } 1673 1674 /** 1675 * xmode parse past comment to end of line. 1676 */ 1677 private int parsePastLine() { 1678 int ch = temp[cursor++]; 1679 while (ch != 0 && !isLineSeparator(ch)) 1680 ch = temp[cursor++]; 1681 return ch; 1682 } 1683 1684 /** 1685 * xmode peek past comment to end of line. 1686 */ 1687 private int peekPastLine() { 1688 int ch = temp[++cursor]; 1689 while (ch != 0 && !isLineSeparator(ch)) 1690 ch = temp[++cursor]; 1691 return ch; 1692 } 1693 1694 /** 1695 * Determines if character is a line separator in the current mode 1696 */ 1697 private boolean isLineSeparator(int ch) { 1698 if (has(UNIX_LINES)) { 1699 return ch == '\n'; 1700 } else { 1701 return (ch == '\n' || 1702 ch == '\r' || 1703 (ch|1) == '\u2029' || 1704 ch == '\u0085'); 1705 } 1706 } 1707 1708 /** 1709 * Read the character after the next one, and advance the cursor by two. 1710 */ 1711 private int skip() { 1712 int i = cursor; 1713 int ch = temp[i+1]; 1714 cursor = i + 2; 1715 return ch; 1716 } 1717 1718 /** 1719 * Unread one next character, and retreat cursor by one. 1720 */ 1721 private void unread() { 1722 cursor--; 1723 } 1724 1725 /** 1726 * Internal method used for handling all syntax errors. The pattern is 1727 * displayed with a pointer to aid in locating the syntax error. 1728 */ 1729 private PatternSyntaxException error(String s) { 1730 return new PatternSyntaxException(s, normalizedPattern, cursor - 1); 1731 } 1732 1733 /** 1734 * Determines if there is any supplementary character or unpaired 1735 * surrogate in the specified range. 1736 */ 1737 private boolean findSupplementary(int start, int end) { 1738 for (int i = start; i < end; i++) { 1739 if (isSupplementary(temp[i])) 1740 return true; 1741 } 1742 return false; 1743 } 1744 1745 /** 1746 * Determines if the specified code point is a supplementary 1747 * character or unpaired surrogate. 1748 */ 1749 private static final boolean isSupplementary(int ch) { 1750 return ch >= Character.MIN_SUPPLEMENTARY_CODE_POINT || isSurrogate(ch); 1751 } 1752 1753 /** 1754 * The following methods handle the main parsing. They are sorted 1755 * according to their precedence order, the lowest one first. 1756 */ 1757 1758 /** 1759 * The expression is parsed with branch nodes added for alternations. 1760 * This may be called recursively to parse sub expressions that may 1761 * contain alternations. 1762 */ 1763 private Node expr(Node end) { 1764 Node prev = null; 1765 Node firstTail = null; 1766 Node branchConn = null; 1767 1768 for (;;) { 1769 Node node = sequence(end); 1770 Node nodeTail = root; //double return 1771 if (prev == null) { 1772 prev = node; 1773 firstTail = nodeTail; 1774 } else { 1775 // Branch 1776 if (branchConn == null) { 1777 branchConn = new BranchConn(); 1778 branchConn.next = end; 1779 } 1780 if (node == end) { 1781 // if the node returned from sequence() is "end" 1782 // we have an empty expr, set a null atom into 1783 // the branch to indicate to go "next" directly. 1784 node = null; 1785 } else { 1786 // the "tail.next" of each atom goes to branchConn 1787 nodeTail.next = branchConn; 1788 } 1789 if (prev instanceof Branch) { 1790 ((Branch)prev).add(node); 1791 } else { 1792 if (prev == end) { 1793 prev = null; 1794 } else { 1795 // replace the "end" with "branchConn" at its tail.next 1796 // when put the "prev" into the branch as the first atom. 1797 firstTail.next = branchConn; 1798 } 1799 prev = new Branch(prev, node, branchConn); 1800 } 1801 } 1802 if (peek() != '|') { 1803 return prev; 1804 } 1805 next(); 1806 } 1807 } 1808 1809 /** 1810 * Parsing of sequences between alternations. 1811 */ 1812 private Node sequence(Node end) { 1813 Node head = null; 1814 Node tail = null; 1815 Node node = null; 1816 LOOP: 1817 for (;;) { 1818 int ch = peek(); 1819 switch (ch) { 1820 case '(': 1821 // Because group handles its own closure, 1822 // we need to treat it differently 1823 node = group0(); 1824 // Check for comment or flag group 1825 if (node == null) 1826 continue; 1827 if (head == null) 1828 head = node; 1829 else 1830 tail.next = node; 1831 // Double return: Tail was returned in root 1832 tail = root; 1833 continue; 1834 case '[': 1835 node = clazz(true); 1836 break; 1837 case '\\': 1838 ch = nextEscaped(); 1839 if (ch == 'p' || ch == 'P') { 1840 boolean oneLetter = true; 1841 boolean comp = (ch == 'P'); 1842 ch = next(); // Consume { if present 1843 if (ch != '{') { 1844 unread(); 1845 } else { 1846 oneLetter = false; 1847 } 1848 node = family(oneLetter).maybeComplement(comp); 1849 } else { 1850 unread(); 1851 node = atom(); 1852 } 1853 break; 1854 case '^': 1855 next(); 1856 if (has(MULTILINE)) { 1857 if (has(UNIX_LINES)) 1858 node = new UnixCaret(); 1859 else 1860 node = new Caret(); 1861 } else { 1862 node = new Begin(); 1863 } 1864 break; 1865 case '$': 1866 next(); 1867 if (has(UNIX_LINES)) 1868 node = new UnixDollar(has(MULTILINE)); 1869 else 1870 node = new Dollar(has(MULTILINE)); 1871 break; 1872 case '.': 1873 next(); 1874 if (has(DOTALL)) { 1875 node = new All(); 1876 } else { 1877 if (has(UNIX_LINES)) 1878 node = new UnixDot(); 1879 else { 1880 node = new Dot(); 1881 } 1882 } 1883 break; 1884 case '|': 1885 case ')': 1886 break LOOP; 1887 case ']': // Now interpreting dangling ] and } as literals 1888 case '}': 1889 node = atom(); 1890 break; 1891 case '?': 1892 case '*': 1893 case '+': 1894 next(); 1895 throw error("Dangling meta character '" + ((char)ch) + "'"); 1896 case 0: 1897 if (cursor >= patternLength) { 1898 break LOOP; 1899 } 1900 // Fall through 1901 default: 1902 node = atom(); 1903 break; 1904 } 1905 1906 node = closure(node); 1907 1908 if (head == null) { 1909 head = tail = node; 1910 } else { 1911 tail.next = node; 1912 tail = node; 1913 } 1914 } 1915 if (head == null) { 1916 return end; 1917 } 1918 tail.next = end; 1919 root = tail; //double return 1920 return head; 1921 } 1922 1923 /** 1924 * Parse and add a new Single or Slice. 1925 */ 1926 private Node atom() { 1927 int first = 0; 1928 int prev = -1; 1929 boolean hasSupplementary = false; 1930 int ch = peek(); 1931 for (;;) { 1932 switch (ch) { 1933 case '*': 1934 case '+': 1935 case '?': 1936 case '{': 1937 if (first > 1) { 1938 cursor = prev; // Unwind one character 1939 first--; 1940 } 1941 break; 1942 case '$': 1943 case '.': 1944 case '^': 1945 case '(': 1946 case '[': 1947 case '|': 1948 case ')': 1949 break; 1950 case '\\': 1951 ch = nextEscaped(); 1952 if (ch == 'p' || ch == 'P') { // Property 1953 if (first > 0) { // Slice is waiting; handle it first 1954 unread(); 1955 break; 1956 } else { // No slice; just return the family node 1957 boolean comp = (ch == 'P'); 1958 boolean oneLetter = true; 1959 ch = next(); // Consume { if present 1960 if (ch != '{') 1961 unread(); 1962 else 1963 oneLetter = false; 1964 return family(oneLetter).maybeComplement(comp); 1965 } 1966 } 1967 unread(); 1968 prev = cursor; 1969 ch = escape(false, first == 0); 1970 if (ch >= 0) { 1971 append(ch, first); 1972 first++; 1973 if (isSupplementary(ch)) { 1974 hasSupplementary = true; 1975 } 1976 ch = peek(); 1977 continue; 1978 } else if (first == 0) { 1979 return root; 1980 } 1981 // Unwind meta escape sequence 1982 cursor = prev; 1983 break; 1984 case 0: 1985 if (cursor >= patternLength) { 1986 break; 1987 } 1988 // Fall through 1989 default: 1990 prev = cursor; 1991 append(ch, first); 1992 first++; 1993 if (isSupplementary(ch)) { 1994 hasSupplementary = true; 1995 } 1996 ch = next(); 1997 continue; 1998 } 1999 break; 2000 } 2001 if (first == 1) { 2002 return newSingle(buffer[0]); 2003 } else { 2004 return newSlice(buffer, first, hasSupplementary); 2005 } 2006 } 2007 2008 private void append(int ch, int len) { 2009 if (len >= buffer.length) { 2010 int[] tmp = new int[len+len]; 2011 System.arraycopy(buffer, 0, tmp, 0, len); 2012 buffer = tmp; 2013 } 2014 buffer[len] = ch; 2015 } 2016 2017 /** 2018 * Parses a backref greedily, taking as many numbers as it 2019 * can. The first digit is always treated as a backref, but 2020 * multi digit numbers are only treated as a backref if at 2021 * least that many backrefs exist at this point in the regex. 2022 */ 2023 private Node ref(int refNum) { 2024 boolean done = false; 2025 while(!done) { 2026 int ch = peek(); 2027 switch(ch) { 2028 case '0': 2029 case '1': 2030 case '2': 2031 case '3': 2032 case '4': 2033 case '5': 2034 case '6': 2035 case '7': 2036 case '8': 2037 case '9': 2038 int newRefNum = (refNum * 10) + (ch - '0'); 2039 // Add another number if it doesn't make a group 2040 // that doesn't exist 2041 if (capturingGroupCount - 1 < newRefNum) { 2042 done = true; 2043 break; 2044 } 2045 refNum = newRefNum; 2046 read(); 2047 break; 2048 default: 2049 done = true; 2050 break; 2051 } 2052 } 2053 if (has(CASE_INSENSITIVE)) 2054 return new CIBackRef(refNum, has(UNICODE_CASE)); 2055 else 2056 return new BackRef(refNum); 2057 } 2058 2059 /** 2060 * Parses an escape sequence to determine the actual value that needs 2061 * to be matched. 2062 * If -1 is returned and create was true a new object was added to the tree 2063 * to handle the escape sequence. 2064 * If the returned value is greater than zero, it is the value that 2065 * matches the escape sequence. 2066 */ 2067 private int escape(boolean inclass, boolean create) { 2068 int ch = skip(); 2069 switch (ch) { 2070 case '0': 2071 return o(); 2072 case '1': 2073 case '2': 2074 case '3': 2075 case '4': 2076 case '5': 2077 case '6': 2078 case '7': 2079 case '8': 2080 case '9': 2081 if (inclass) break; 2082 if (create) { 2083 root = ref((ch - '0')); 2084 } 2085 return -1; 2086 case 'A': 2087 if (inclass) break; 2088 if (create) root = new Begin(); 2089 return -1; 2090 case 'B': 2091 if (inclass) break; 2092 if (create) root = new Bound(Bound.NONE); 2093 return -1; 2094 case 'C': 2095 break; 2096 case 'D': 2097 if (create) root = new Ctype(ASCII.DIGIT).complement(); 2098 return -1; 2099 case 'E': 2100 case 'F': 2101 break; 2102 case 'G': 2103 if (inclass) break; 2104 if (create) root = new LastMatch(); 2105 return -1; 2106 case 'H': 2107 case 'I': 2108 case 'J': 2109 case 'K': 2110 case 'L': 2111 case 'M': 2112 case 'N': 2113 case 'O': 2114 case 'P': 2115 case 'Q': 2116 case 'R': 2117 break; 2118 case 'S': 2119 if (create) root = new Ctype(ASCII.SPACE).complement(); 2120 return -1; 2121 case 'T': 2122 case 'U': 2123 case 'V': 2124 break; 2125 case 'W': 2126 if (create) root = new Ctype(ASCII.WORD).complement(); 2127 return -1; 2128 case 'X': 2129 case 'Y': 2130 break; 2131 case 'Z': 2132 if (inclass) break; 2133 if (create) { 2134 if (has(UNIX_LINES)) 2135 root = new UnixDollar(false); 2136 else 2137 root = new Dollar(false); 2138 } 2139 return -1; 2140 case 'a': 2141 return '\007'; 2142 case 'b': 2143 if (inclass) break; 2144 if (create) root = new Bound(Bound.BOTH); 2145 return -1; 2146 case 'c': 2147 return c(); 2148 case 'd': 2149 if (create) root = new Ctype(ASCII.DIGIT); 2150 return -1; 2151 case 'e': 2152 return '\033'; 2153 case 'f': 2154 return '\f'; 2155 case 'g': 2156 case 'h': 2157 case 'i': 2158 case 'j': 2159 case 'k': 2160 case 'l': 2161 case 'm': 2162 break; 2163 case 'n': 2164 return '\n'; 2165 case 'o': 2166 case 'p': 2167 case 'q': 2168 break; 2169 case 'r': 2170 return '\r'; 2171 case 's': 2172 if (create) root = new Ctype(ASCII.SPACE); 2173 return -1; 2174 case 't': 2175 return '\t'; 2176 case 'u': 2177 return u(); 2178 case 'v': 2179 return '\013'; 2180 case 'w': 2181 if (create) root = new Ctype(ASCII.WORD); 2182 return -1; 2183 case 'x': 2184 return x(); 2185 case 'y': 2186 break; 2187 case 'z': 2188 if (inclass) break; 2189 if (create) root = new End(); 2190 return -1; 2191 default: 2192 return ch; 2193 } 2194 throw error("Illegal/unsupported escape sequence"); 2195 } 2196 2197 /** 2198 * Parse a character class, and return the node that matches it. 2199 * 2200 * Consumes a ] on the way out if consume is true. Usually consume 2201 * is true except for the case of [abc&&def] where def is a separate 2202 * right hand node with "understood" brackets. 2203 */ 2204 private CharProperty clazz(boolean consume) { 2205 CharProperty prev = null; 2206 CharProperty node = null; 2207 BitClass bits = new BitClass(); 2208 boolean include = true; 2209 boolean firstInClass = true; 2210 int ch = next(); 2211 for (;;) { 2212 switch (ch) { 2213 case '^': 2214 // Negates if first char in a class, otherwise literal 2215 if (firstInClass) { 2216 if (temp[cursor-1] != '[') 2217 break; 2218 ch = next(); 2219 include = !include; 2220 continue; 2221 } else { 2222 // ^ not first in class, treat as literal 2223 break; 2224 } 2225 case '[': 2226 firstInClass = false; 2227 node = clazz(true); 2228 if (prev == null) 2229 prev = node; 2230 else 2231 prev = union(prev, node); 2232 ch = peek(); 2233 continue; 2234 case '&': 2235 firstInClass = false; 2236 ch = next(); 2237 if (ch == '&') { 2238 ch = next(); 2239 CharProperty rightNode = null; 2240 while (ch != ']' && ch != '&') { 2241 if (ch == '[') { 2242 if (rightNode == null) 2243 rightNode = clazz(true); 2244 else 2245 rightNode = union(rightNode, clazz(true)); 2246 } else { // abc&&def 2247 unread(); 2248 rightNode = clazz(false); 2249 } 2250 ch = peek(); 2251 } 2252 if (rightNode != null) 2253 node = rightNode; 2254 if (prev == null) { 2255 if (rightNode == null) 2256 throw error("Bad class syntax"); 2257 else 2258 prev = rightNode; 2259 } else { 2260 prev = intersection(prev, node); 2261 } 2262 } else { 2263 // treat as a literal & 2264 unread(); 2265 break; 2266 } 2267 continue; 2268 case 0: 2269 firstInClass = false; 2270 if (cursor >= patternLength) 2271 throw error("Unclosed character class"); 2272 break; 2273 case ']': 2274 firstInClass = false; 2275 if (prev != null) { 2276 if (consume) 2277 next(); 2278 return prev; 2279 } 2280 break; 2281 default: 2282 firstInClass = false; 2283 break; 2284 } 2285 node = range(bits); 2286 if (include) { 2287 if (prev == null) { 2288 prev = node; 2289 } else { 2290 if (prev != node) 2291 prev = union(prev, node); 2292 } 2293 } else { 2294 if (prev == null) { 2295 prev = node.complement(); 2296 } else { 2297 if (prev != node) 2298 prev = setDifference(prev, node); 2299 } 2300 } 2301 ch = peek(); 2302 } 2303 } 2304 2305 private CharProperty bitsOrSingle(BitClass bits, int ch) { 2306 /* Bits can only handle codepoints in [u+0000-u+00ff] range. 2307 Use "single" node instead of bits when dealing with unicode 2308 case folding for codepoints listed below. 2309 (1)Uppercase out of range: u+00ff, u+00b5 2310 toUpperCase(u+00ff) -> u+0178 2311 toUpperCase(u+00b5) -> u+039c 2312 (2)LatinSmallLetterLongS u+17f 2313 toUpperCase(u+017f) -> u+0053 2314 (3)LatinSmallLetterDotlessI u+131 2315 toUpperCase(u+0131) -> u+0049 2316 (4)LatinCapitalLetterIWithDotAbove u+0130 2317 toLowerCase(u+0130) -> u+0069 2318 (5)KelvinSign u+212a 2319 toLowerCase(u+212a) ==> u+006B 2320 (6)AngstromSign u+212b 2321 toLowerCase(u+212b) ==> u+00e5 2322 */ 2323 int d; 2324 if (ch < 256 && 2325 !(has(CASE_INSENSITIVE) && has(UNICODE_CASE) && 2326 (ch == 0xff || ch == 0xb5 || 2327 ch == 0x49 || ch == 0x69 || //I and i 2328 ch == 0x53 || ch == 0x73 || //S and s 2329 ch == 0x4b || ch == 0x6b || //K and k 2330 ch == 0xc5 || ch == 0xe5))) //A+ring 2331 return bits.add(ch, flags()); 2332 return newSingle(ch); 2333 } 2334 2335 /** 2336 * Parse a single character or a character range in a character class 2337 * and return its representative node. 2338 */ 2339 private CharProperty range(BitClass bits) { 2340 int ch = peek(); 2341 if (ch == '\\') { 2342 ch = nextEscaped(); 2343 if (ch == 'p' || ch == 'P') { // A property 2344 boolean comp = (ch == 'P'); 2345 boolean oneLetter = true; 2346 // Consume { if present 2347 ch = next(); 2348 if (ch != '{') 2349 unread(); 2350 else 2351 oneLetter = false; 2352 return family(oneLetter).maybeComplement(comp); 2353 } else { // ordinary escape 2354 unread(); 2355 ch = escape(true, true); 2356 if (ch == -1) 2357 return (CharProperty) root; 2358 } 2359 } else { 2360 ch = single(); 2361 } 2362 if (ch >= 0) { 2363 if (peek() == '-') { 2364 int endRange = temp[cursor+1]; 2365 if (endRange == '[') { 2366 return bitsOrSingle(bits, ch); 2367 } 2368 if (endRange != ']') { 2369 next(); 2370 int m = single(); 2371 if (m < ch) 2372 throw error("Illegal character range"); 2373 if (has(CASE_INSENSITIVE)) 2374 return caseInsensitiveRangeFor(ch, m); 2375 else 2376 return rangeFor(ch, m); 2377 } 2378 } 2379 return bitsOrSingle(bits, ch); 2380 } 2381 throw error("Unexpected character '"+((char)ch)+"'"); 2382 } 2383 2384 private int single() { 2385 int ch = peek(); 2386 switch (ch) { 2387 case '\\': 2388 return escape(true, false); 2389 default: 2390 next(); 2391 return ch; 2392 } 2393 } 2394 2395 /** 2396 * Parses a Unicode character family and returns its representative node. 2397 */ 2398 private CharProperty family(boolean singleLetter) { 2399 next(); 2400 String name; 2401 2402 if (singleLetter) { 2403 int c = temp[cursor]; 2404 if (!Character.isSupplementaryCodePoint(c)) { 2405 name = String.valueOf((char)c); 2406 } else { 2407 name = new String(temp, cursor, 1); 2408 } 2409 read(); 2410 } else { 2411 int i = cursor; 2412 mark('}'); 2413 while(read() != '}') { 2414 } 2415 mark('\000'); 2416 int j = cursor; 2417 if (j > patternLength) 2418 throw error("Unclosed character family"); 2419 if (i + 1 >= j) 2420 throw error("Empty character family"); 2421 name = new String(temp, i, j-i-1); 2422 } 2423 2424 if (name.startsWith("In")) { 2425 return unicodeBlockPropertyFor(name.substring(2)); 2426 } else { 2427 if (name.startsWith("Is")) 2428 name = name.substring(2); 2429 return charPropertyNodeFor(name); 2430 } 2431 } 2432 2433 /** 2434 * Returns a CharProperty matching all characters in a UnicodeBlock. 2435 */ 2436 private CharProperty unicodeBlockPropertyFor(String name) { 2437 final Character.UnicodeBlock block; 2438 try { 2439 block = Character.UnicodeBlock.forName(name); 2440 } catch (IllegalArgumentException iae) { 2441 throw error("Unknown character block name {" + name + "}"); 2442 } 2443 return new CharProperty() { 2444 boolean isSatisfiedBy(int ch) { 2445 return block == Character.UnicodeBlock.of(ch);}}; 2446 } 2447 2448 /** 2449 * Returns a CharProperty matching all characters in a named property. 2450 */ 2451 private CharProperty charPropertyNodeFor(String name) { 2452 CharProperty p = CharPropertyNames.charPropertyFor(name); 2453 if (p == null) 2454 throw error("Unknown character property name {" + name + "}"); 2455 return p; 2456 } 2457 2458 /** 2459 * Parses a group and returns the head node of a set of nodes that process 2460 * the group. Sometimes a double return system is used where the tail is 2461 * returned in root. 2462 */ 2463 private Node group0() { 2464 boolean capturingGroup = false; 2465 Node head = null; 2466 Node tail = null; 2467 int save = flags; 2468 root = null; 2469 int ch = next(); 2470 if (ch == '?') { 2471 ch = skip(); 2472 switch (ch) { 2473 case ':': // (?:xxx) pure group 2474 head = createGroup(true); 2475 tail = root; 2476 head.next = expr(tail); 2477 break; 2478 case '=': // (?=xxx) and (?!xxx) lookahead 2479 case '!': 2480 head = createGroup(true); 2481 tail = root; 2482 head.next = expr(tail); 2483 if (ch == '=') { 2484 head = tail = new Pos(head); 2485 } else { 2486 head = tail = new Neg(head); 2487 } 2488 break; 2489 case '>': // (?>xxx) independent group 2490 head = createGroup(true); 2491 tail = root; 2492 head.next = expr(tail); 2493 head = tail = new Ques(head, INDEPENDENT); 2494 break; 2495 case '<': // (?<xxx) look behind 2496 ch = read(); 2497 int start = cursor; 2498 head = createGroup(true); 2499 tail = root; 2500 head.next = expr(tail); 2501 tail.next = lookbehindEnd; 2502 TreeInfo info = new TreeInfo(); 2503 head.study(info); 2504 if (info.maxValid == false) { 2505 throw error("Look-behind group does not have " 2506 + "an obvious maximum length"); 2507 } 2508 boolean hasSupplementary = findSupplementary(start, patternLength); 2509 if (ch == '=') { 2510 head = tail = (hasSupplementary ? 2511 new BehindS(head, info.maxLength, 2512 info.minLength) : 2513 new Behind(head, info.maxLength, 2514 info.minLength)); 2515 } else if (ch == '!') { 2516 head = tail = (hasSupplementary ? 2517 new NotBehindS(head, info.maxLength, 2518 info.minLength) : 2519 new NotBehind(head, info.maxLength, 2520 info.minLength)); 2521 } else { 2522 throw error("Unknown look-behind group"); 2523 } 2524 break; 2525 case '$': 2526 case '@': 2527 throw error("Unknown group type"); 2528 default: // (?xxx:) inlined match flags 2529 unread(); 2530 addFlag(); 2531 ch = read(); 2532 if (ch == ')') { 2533 return null; // Inline modifier only 2534 } 2535 if (ch != ':') { 2536 throw error("Unknown inline modifier"); 2537 } 2538 head = createGroup(true); 2539 tail = root; 2540 head.next = expr(tail); 2541 break; 2542 } 2543 } else { // (xxx) a regular group 2544 capturingGroup = true; 2545 head = createGroup(false); 2546 tail = root; 2547 head.next = expr(tail); 2548 } 2549 2550 accept(')', "Unclosed group"); 2551 flags = save; 2552 2553 // Check for quantifiers 2554 Node node = closure(head); 2555 if (node == head) { // No closure 2556 root = tail; 2557 return node; // Dual return 2558 } 2559 if (head == tail) { // Zero length assertion 2560 root = node; 2561 return node; // Dual return 2562 } 2563 2564 if (node instanceof Ques) { 2565 Ques ques = (Ques) node; 2566 if (ques.type == POSSESSIVE) { 2567 root = node; 2568 return node; 2569 } 2570 tail.next = new BranchConn(); 2571 tail = tail.next; 2572 if (ques.type == GREEDY) { 2573 head = new Branch(head, null, tail); 2574 } else { // Reluctant quantifier 2575 head = new Branch(null, head, tail); 2576 } 2577 root = tail; 2578 return head; 2579 } else if (node instanceof Curly) { 2580 Curly curly = (Curly) node; 2581 if (curly.type == POSSESSIVE) { 2582 root = node; 2583 return node; 2584 } 2585 // Discover if the group is deterministic 2586 TreeInfo info = new TreeInfo(); 2587 if (head.study(info)) { // Deterministic 2588 GroupTail temp = (GroupTail) tail; 2589 head = root = new GroupCurly(head.next, curly.cmin, 2590 curly.cmax, curly.type, 2591 ((GroupTail)tail).localIndex, 2592 ((GroupTail)tail).groupIndex, 2593 capturingGroup); 2594 return head; 2595 } else { // Non-deterministic 2596 int temp = ((GroupHead) head).localIndex; 2597 Loop loop; 2598 if (curly.type == GREEDY) 2599 loop = new Loop(this.localCount, temp); 2600 else // Reluctant Curly 2601 loop = new LazyLoop(this.localCount, temp); 2602 Prolog prolog = new Prolog(loop); 2603 this.localCount += 1; 2604 loop.cmin = curly.cmin; 2605 loop.cmax = curly.cmax; 2606 loop.body = head; 2607 tail.next = loop; 2608 root = loop; 2609 return prolog; // Dual return 2610 } 2611 } 2612 throw error("Internal logic error"); 2613 } 2614 2615 /** 2616 * Create group head and tail nodes using double return. If the group is 2617 * created with anonymous true then it is a pure group and should not 2618 * affect group counting. 2619 */ 2620 private Node createGroup(boolean anonymous) { 2621 int localIndex = localCount++; 2622 int groupIndex = 0; 2623 if (!anonymous) 2624 groupIndex = capturingGroupCount++; 2625 GroupHead head = new GroupHead(localIndex); 2626 root = new GroupTail(localIndex, groupIndex); 2627 if (!anonymous && groupIndex < 10) 2628 groupNodes[groupIndex] = head; 2629 return head; 2630 } 2631 2632 /** 2633 * Parses inlined match flags and set them appropriately. 2634 */ 2635 private void addFlag() { 2636 int ch = peek(); 2637 for (;;) { 2638 switch (ch) { 2639 case 'i': 2640 flags |= CASE_INSENSITIVE; 2641 break; 2642 case 'm': 2643 flags |= MULTILINE; 2644 break; 2645 case 's': 2646 flags |= DOTALL; 2647 break; 2648 case 'd': 2649 flags |= UNIX_LINES; 2650 break; 2651 case 'u': 2652 flags |= UNICODE_CASE; 2653 break; 2654 case 'c': 2655 flags |= CANON_EQ; 2656 break; 2657 case 'x': 2658 flags |= COMMENTS; 2659 break; 2660 case '-': // subFlag then fall through 2661 ch = next(); 2662 subFlag(); 2663 default: 2664 return; 2665 } 2666 ch = next(); 2667 } 2668 } 2669 2670 /** 2671 * Parses the second part of inlined match flags and turns off 2672 * flags appropriately. 2673 */ 2674 private void subFlag() { 2675 int ch = peek(); 2676 for (;;) { 2677 switch (ch) { 2678 case 'i': 2679 flags &= ~CASE_INSENSITIVE; 2680 break; 2681 case 'm': 2682 flags &= ~MULTILINE; 2683 break; 2684 case 's': 2685 flags &= ~DOTALL; 2686 break; 2687 case 'd': 2688 flags &= ~UNIX_LINES; 2689 break; 2690 case 'u': 2691 flags &= ~UNICODE_CASE; 2692 break; 2693 case 'c': 2694 flags &= ~CANON_EQ; 2695 break; 2696 case 'x': 2697 flags &= ~COMMENTS; 2698 break; 2699 default: 2700 return; 2701 } 2702 ch = next(); 2703 } 2704 } 2705 2706 static final int MAX_REPS = 0x7FFFFFFF; 2707 2708 static final int GREEDY = 0; 2709 2710 static final int LAZY = 1; 2711 2712 static final int POSSESSIVE = 2; 2713 2714 static final int INDEPENDENT = 3; 2715 2716 /** 2717 * Processes repetition. If the next character peeked is a quantifier 2718 * then new nodes must be appended to handle the repetition. 2719 * Prev could be a single or a group, so it could be a chain of nodes. 2720 */ 2721 private Node closure(Node prev) { 2722 Node atom; 2723 int ch = peek(); 2724 switch (ch) { 2725 case '?': 2726 ch = next(); 2727 if (ch == '?') { 2728 next(); 2729 return new Ques(prev, LAZY); 2730 } else if (ch == '+') { 2731 next(); 2732 return new Ques(prev, POSSESSIVE); 2733 } 2734 return new Ques(prev, GREEDY); 2735 case '*': 2736 ch = next(); 2737 if (ch == '?') { 2738 next(); 2739 return new Curly(prev, 0, MAX_REPS, LAZY); 2740 } else if (ch == '+') { 2741 next(); 2742 return new Curly(prev, 0, MAX_REPS, POSSESSIVE); 2743 } 2744 return new Curly(prev, 0, MAX_REPS, GREEDY); 2745 case '+': 2746 ch = next(); 2747 if (ch == '?') { 2748 next(); 2749 return new Curly(prev, 1, MAX_REPS, LAZY); 2750 } else if (ch == '+') { 2751 next(); 2752 return new Curly(prev, 1, MAX_REPS, POSSESSIVE); 2753 } 2754 return new Curly(prev, 1, MAX_REPS, GREEDY); 2755 case '{': 2756 ch = temp[cursor+1]; 2757 if (ASCII.isDigit(ch)) { 2758 skip(); 2759 int cmin = 0; 2760 do { 2761 cmin = cmin * 10 + (ch - '0'); 2762 } while (ASCII.isDigit(ch = read())); 2763 int cmax = cmin; 2764 if (ch == ',') { 2765 ch = read(); 2766 cmax = MAX_REPS; 2767 if (ch != '}') { 2768 cmax = 0; 2769 while (ASCII.isDigit(ch)) { 2770 cmax = cmax * 10 + (ch - '0'); 2771 ch = read(); 2772 } 2773 } 2774 } 2775 if (ch != '}') 2776 throw error("Unclosed counted closure"); 2777 if (((cmin) | (cmax) | (cmax - cmin)) < 0) 2778 throw error("Illegal repetition range"); 2779 Curly curly; 2780 ch = peek(); 2781 if (ch == '?') { 2782 next(); 2783 curly = new Curly(prev, cmin, cmax, LAZY); 2784 } else if (ch == '+') { 2785 next(); 2786 curly = new Curly(prev, cmin, cmax, POSSESSIVE); 2787 } else { 2788 curly = new Curly(prev, cmin, cmax, GREEDY); 2789 } 2790 return curly; 2791 } else { 2792 throw error("Illegal repetition"); 2793 } 2794 default: 2795 return prev; 2796 } 2797 } 2798 2799 /** 2800 * Utility method for parsing control escape sequences. 2801 */ 2802 private int c() { 2803 if (cursor < patternLength) { 2804 return read() ^ 64; 2805 } 2806 throw error("Illegal control escape sequence"); 2807 } 2808 2809 /** 2810 * Utility method for parsing octal escape sequences. 2811 */ 2812 private int o() { 2813 int n = read(); 2814 if (((n-'0')|('7'-n)) >= 0) { 2815 int m = read(); 2816 if (((m-'0')|('7'-m)) >= 0) { 2817 int o = read(); 2818 if ((((o-'0')|('7'-o)) >= 0) && (((n-'0')|('3'-n)) >= 0)) { 2819 return (n - '0') * 64 + (m - '0') * 8 + (o - '0'); 2820 } 2821 unread(); 2822 return (n - '0') * 8 + (m - '0'); 2823 } 2824 unread(); 2825 return (n - '0'); 2826 } 2827 throw error("Illegal octal escape sequence"); 2828 } 2829 2830 /** 2831 * Utility method for parsing hexadecimal escape sequences. 2832 */ 2833 private int x() { 2834 int n = read(); 2835 if (ASCII.isHexDigit(n)) { 2836 int m = read(); 2837 if (ASCII.isHexDigit(m)) { 2838 return ASCII.toDigit(n) * 16 + ASCII.toDigit(m); 2839 } 2840 } 2841 throw error("Illegal hexadecimal escape sequence"); 2842 } 2843 2844 /** 2845 * Utility method for parsing unicode escape sequences. 2846 */ 2847 private int cursor() { 2848 return cursor; 2849 } 2850 2851 private void setcursor(int pos) { 2852 cursor = pos; 2853 } 2854 2855 private int uxxxx() { 2856 int n = 0; 2857 for (int i = 0; i < 4; i++) { 2858 int ch = read(); 2859 if (!ASCII.isHexDigit(ch)) { 2860 throw error("Illegal Unicode escape sequence"); 2861 } 2862 n = n * 16 + ASCII.toDigit(ch); 2863 } 2864 return n; 2865 } 2866 2867 private int u() { 2868 int n = uxxxx(); 2869 if (Character.isHighSurrogate((char)n)) { 2870 int cur = cursor(); 2871 if (read() == '\\' && read() == 'u') { 2872 int n2 = uxxxx(); 2873 if (Character.isLowSurrogate((char)n2)) 2874 return Character.toCodePoint((char)n, (char)n2); 2875 } 2876 setcursor(cur); 2877 } 2878 return n; 2879 } 2880 2881 // 2882 // Utility methods for code point support 2883 // 2884 2885 /** 2886 * Tests a surrogate value. 2887 */ 2888 private static final boolean isSurrogate(int c) { 2889 return c >= Character.MIN_HIGH_SURROGATE && c <= Character.MAX_LOW_SURROGATE; 2890 } 2891 2892 private static final int countChars(CharSequence seq, int index, 2893 int lengthInCodePoints) { 2894 // optimization 2895 if (lengthInCodePoints == 1 && !Character.isHighSurrogate(seq.charAt(index))) { 2896 assert (index >= 0 && index < seq.length()); 2897 return 1; 2898 } 2899 int length = seq.length(); 2900 int x = index; 2901 if (lengthInCodePoints >= 0) { 2902 assert (index >= 0 && index < length); 2903 for (int i = 0; x < length && i < lengthInCodePoints; i++) { 2904 if (Character.isHighSurrogate(seq.charAt(x++))) { 2905 if (x < length && Character.isLowSurrogate(seq.charAt(x))) { 2906 x++; 2907 } 2908 } 2909 } 2910 return x - index; 2911 } 2912 2913 assert (index >= 0 && index <= length); 2914 if (index == 0) { 2915 return 0; 2916 } 2917 int len = -lengthInCodePoints; 2918 for (int i = 0; x > 0 && i < len; i++) { 2919 if (Character.isLowSurrogate(seq.charAt(--x))) { 2920 if (x > 0 && Character.isHighSurrogate(seq.charAt(x-1))) { 2921 x--; 2922 } 2923 } 2924 } 2925 return index - x; 2926 } 2927 2928 private static final int countCodePoints(CharSequence seq) { 2929 int length = seq.length(); 2930 int n = 0; 2931 for (int i = 0; i < length; ) { 2932 n++; 2933 if (Character.isHighSurrogate(seq.charAt(i++))) { 2934 if (i < length && Character.isLowSurrogate(seq.charAt(i))) { 2935 i++; 2936 } 2937 } 2938 } 2939 return n; 2940 } 2941 2942 /** 2943 * Creates a bit vector for matching Latin-1 values. A normal BitClass 2944 * never matches values above Latin-1, and a complemented BitClass always 2945 * matches values above Latin-1. 2946 */ 2947 private static final class BitClass extends BmpCharProperty { 2948 final boolean[] bits; 2949 BitClass() { bits = new boolean[256]; } 2950 private BitClass(boolean[] bits) { this.bits = bits; } 2951 BitClass add(int c, int flags) { 2952 assert c >= 0 && c <= 255; 2953 if ((flags & CASE_INSENSITIVE) != 0) { 2954 if (ASCII.isAscii(c)) { 2955 bits[ASCII.toUpper(c)] = true; 2956 bits[ASCII.toLower(c)] = true; 2957 } else if ((flags & UNICODE_CASE) != 0) { 2958 bits[Character.toLowerCase(c)] = true; 2959 bits[Character.toUpperCase(c)] = true; 2960 } 2961 } 2962 bits[c] = true; 2963 return this; 2964 } 2965 boolean isSatisfiedBy(int ch) { 2966 return ch < 256 && bits[ch]; 2967 } 2968 } 2969 2970 /** 2971 * Returns a suitably optimized, single character matcher. 2972 */ 2973 private CharProperty newSingle(final int ch) { 2974 if (has(CASE_INSENSITIVE)) { 2975 int lower, upper; 2976 if (has(UNICODE_CASE)) { 2977 upper = Character.toUpperCase(ch); 2978 lower = Character.toLowerCase(upper); 2979 if (upper != lower) 2980 return new SingleU(lower); 2981 } else if (ASCII.isAscii(ch)) { 2982 lower = ASCII.toLower(ch); 2983 upper = ASCII.toUpper(ch); 2984 if (lower != upper) 2985 return new SingleI(lower, upper); 2986 } 2987 } 2988 if (isSupplementary(ch)) 2989 return new SingleS(ch); // Match a given Unicode character 2990 return new Single(ch); // Match a given BMP character 2991 } 2992 2993 /** 2994 * Utility method for creating a string slice matcher. 2995 */ 2996 private Node newSlice(int[] buf, int count, boolean hasSupplementary) { 2997 int[] tmp = new int[count]; 2998 if (has(CASE_INSENSITIVE)) { 2999 if (has(UNICODE_CASE)) { 3000 for (int i = 0; i < count; i++) { 3001 tmp[i] = Character.toLowerCase( 3002 Character.toUpperCase(buf[i])); 3003 } 3004 return hasSupplementary? new SliceUS(tmp) : new SliceU(tmp); 3005 } 3006 for (int i = 0; i < count; i++) { 3007 tmp[i] = ASCII.toLower(buf[i]); 3008 } 3009 return hasSupplementary? new SliceIS(tmp) : new SliceI(tmp); 3010 } 3011 for (int i = 0; i < count; i++) { 3012 tmp[i] = buf[i]; 3013 } 3014 return hasSupplementary ? new SliceS(tmp) : new Slice(tmp); 3015 } 3016 3017 /** 3018 * The following classes are the building components of the object 3019 * tree that represents a compiled regular expression. The object tree 3020 * is made of individual elements that handle constructs in the Pattern. 3021 * Each type of object knows how to match its equivalent construct with 3022 * the match() method. 3023 */ 3024 3025 /** 3026 * Base class for all node classes. Subclasses should override the match() 3027 * method as appropriate. This class is an accepting node, so its match() 3028 * always returns true. 3029 */ 3030 static class Node extends Object { 3031 Node next; 3032 Node() { 3033 next = Pattern.accept; 3034 } 3035 /** 3036 * This method implements the classic accept node. 3037 */ 3038 boolean match(Matcher matcher, int i, CharSequence seq) { 3039 matcher.last = i; 3040 matcher.groups[0] = matcher.first; 3041 matcher.groups[1] = matcher.last; 3042 return true; 3043 } 3044 /** 3045 * This method is good for all zero length assertions. 3046 */ 3047 boolean study(TreeInfo info) { 3048 if (next != null) { 3049 return next.study(info); 3050 } else { 3051 return info.deterministic; 3052 } 3053 } 3054 } 3055 3056 static class LastNode extends Node { 3057 /** 3058 * This method implements the classic accept node with 3059 * the addition of a check to see if the match occurred 3060 * using all of the input. 3061 */ 3062 boolean match(Matcher matcher, int i, CharSequence seq) { 3063 if (matcher.acceptMode == Matcher.ENDANCHOR && i != matcher.to) 3064 return false; 3065 matcher.last = i; 3066 matcher.groups[0] = matcher.first; 3067 matcher.groups[1] = matcher.last; 3068 return true; 3069 } 3070 } 3071 3072 /** 3073 * Used for REs that can start anywhere within the input string. 3074 * This basically tries to match repeatedly at each spot in the 3075 * input string, moving forward after each try. An anchored search 3076 * or a BnM will bypass this node completely. 3077 */ 3078 static class Start extends Node { 3079 int minLength; 3080 Start(Node node) { 3081 this.next = node; 3082 TreeInfo info = new TreeInfo(); 3083 next.study(info); 3084 minLength = info.minLength; 3085 } 3086 boolean match(Matcher matcher, int i, CharSequence seq) { 3087 if (i > matcher.to - minLength) { 3088 matcher.hitEnd = true; 3089 return false; 3090 } 3091 boolean ret = false; 3092 int guard = matcher.to - minLength; 3093 for (; i <= guard; i++) { 3094 if (ret = next.match(matcher, i, seq)) 3095 break; 3096 if (i == guard) 3097 matcher.hitEnd = true; 3098 } 3099 if (ret) { 3100 matcher.first = i; 3101 matcher.groups[0] = matcher.first; 3102 matcher.groups[1] = matcher.last; 3103 } 3104 return ret; 3105 } 3106 boolean study(TreeInfo info) { 3107 next.study(info); 3108 info.maxValid = false; 3109 info.deterministic = false; 3110 return false; 3111 } 3112 } 3113 3114 /* 3115 * StartS supports supplementary characters, including unpaired surrogates. 3116 */ 3117 static final class StartS extends Start { 3118 StartS(Node node) { 3119 super(node); 3120 } 3121 boolean match(Matcher matcher, int i, CharSequence seq) { 3122 if (i > matcher.to - minLength) { 3123 matcher.hitEnd = true; 3124 return false; 3125 } 3126 boolean ret = false; 3127 int guard = matcher.to - minLength; 3128 while (i <= guard) { 3129 if ((ret = next.match(matcher, i, seq)) || i == guard) 3130 break; 3131 // Optimization to move to the next character. This is 3132 // faster than countChars(seq, i, 1). 3133 if (Character.isHighSurrogate(seq.charAt(i++))) { 3134 if (i < seq.length() && Character.isLowSurrogate(seq.charAt(i))) { 3135 i++; 3136 } 3137 } 3138 if (i == guard) 3139 matcher.hitEnd = true; 3140 } 3141 if (ret) { 3142 matcher.first = i; 3143 matcher.groups[0] = matcher.first; 3144 matcher.groups[1] = matcher.last; 3145 } 3146 return ret; 3147 } 3148 } 3149 3150 /** 3151 * Node to anchor at the beginning of input. This object implements the 3152 * match for a \A sequence, and the caret anchor will use this if not in 3153 * multiline mode. 3154 */ 3155 static final class Begin extends Node { 3156 boolean match(Matcher matcher, int i, CharSequence seq) { 3157 int fromIndex = (matcher.anchoringBounds) ? 3158 matcher.from : 0; 3159 if (i == fromIndex && next.match(matcher, i, seq)) { 3160 matcher.first = i; 3161 matcher.groups[0] = i; 3162 matcher.groups[1] = matcher.last; 3163 return true; 3164 } else { 3165 return false; 3166 } 3167 } 3168 } 3169 3170 /** 3171 * Node to anchor at the end of input. This is the absolute end, so this 3172 * should not match at the last newline before the end as $ will. 3173 */ 3174 static final class End extends Node { 3175 boolean match(Matcher matcher, int i, CharSequence seq) { 3176 int endIndex = (matcher.anchoringBounds) ? 3177 matcher.to : matcher.getTextLength(); 3178 if (i == endIndex) { 3179 matcher.hitEnd = true; 3180 return next.match(matcher, i, seq); 3181 } 3182 return false; 3183 } 3184 } 3185 3186 /** 3187 * Node to anchor at the beginning of a line. This is essentially the 3188 * object to match for the multiline ^. 3189 */ 3190 static final class Caret extends Node { 3191 boolean match(Matcher matcher, int i, CharSequence seq) { 3192 int startIndex = matcher.from; 3193 int endIndex = matcher.to; 3194 if (!matcher.anchoringBounds) { 3195 startIndex = 0; 3196 endIndex = matcher.getTextLength(); 3197 } 3198 // Perl does not match ^ at end of input even after newline 3199 if (i == endIndex) { 3200 matcher.hitEnd = true; 3201 return false; 3202 } 3203 if (i > startIndex) { 3204 char ch = seq.charAt(i-1); 3205 if (ch != '\n' && ch != '\r' 3206 && (ch|1) != '\u2029' 3207 && ch != '\u0085' ) { 3208 return false; 3209 } 3210 // Should treat /r/n as one newline 3211 if (ch == '\r' && seq.charAt(i) == '\n') 3212 return false; 3213 } 3214 return next.match(matcher, i, seq); 3215 } 3216 } 3217 3218 /** 3219 * Node to anchor at the beginning of a line when in unixdot mode. 3220 */ 3221 static final class UnixCaret extends Node { 3222 boolean match(Matcher matcher, int i, CharSequence seq) { 3223 int startIndex = matcher.from; 3224 int endIndex = matcher.to; 3225 if (!matcher.anchoringBounds) { 3226 startIndex = 0; 3227 endIndex = matcher.getTextLength(); 3228 } 3229 // Perl does not match ^ at end of input even after newline 3230 if (i == endIndex) { 3231 matcher.hitEnd = true; 3232 return false; 3233 } 3234 if (i > startIndex) { 3235 char ch = seq.charAt(i-1); 3236 if (ch != '\n') { 3237 return false; 3238 } 3239 } 3240 return next.match(matcher, i, seq); 3241 } 3242 } 3243 3244 /** 3245 * Node to match the location where the last match ended. 3246 * This is used for the \G construct. 3247 */ 3248 static final class LastMatch extends Node { 3249 boolean match(Matcher matcher, int i, CharSequence seq) { 3250 if (i != matcher.oldLast) 3251 return false; 3252 return next.match(matcher, i, seq); 3253 } 3254 } 3255 3256 /** 3257 * Node to anchor at the end of a line or the end of input based on the 3258 * multiline mode. 3259 * 3260 * When not in multiline mode, the $ can only match at the very end 3261 * of the input, unless the input ends in a line terminator in which 3262 * it matches right before the last line terminator. 3263 * 3264 * Note that \r\n is considered an atomic line terminator. 3265 * 3266 * Like ^ the $ operator matches at a position, it does not match the 3267 * line terminators themselves. 3268 */ 3269 static final class Dollar extends Node { 3270 boolean multiline; 3271 Dollar(boolean mul) { 3272 multiline = mul; 3273 } 3274 boolean match(Matcher matcher, int i, CharSequence seq) { 3275 int endIndex = (matcher.anchoringBounds) ? 3276 matcher.to : matcher.getTextLength(); 3277 if (!multiline) { 3278 if (i < endIndex - 2) 3279 return false; 3280 if (i == endIndex - 2) { 3281 char ch = seq.charAt(i); 3282 if (ch != '\r') 3283 return false; 3284 ch = seq.charAt(i + 1); 3285 if (ch != '\n') 3286 return false; 3287 } 3288 } 3289 // Matches before any line terminator; also matches at the 3290 // end of input 3291 // Before line terminator: 3292 // If multiline, we match here no matter what 3293 // If not multiline, fall through so that the end 3294 // is marked as hit; this must be a /r/n or a /n 3295 // at the very end so the end was hit; more input 3296 // could make this not match here 3297 if (i < endIndex) { 3298 char ch = seq.charAt(i); 3299 if (ch == '\n') { 3300 // No match between \r\n 3301 if (i > 0 && seq.charAt(i-1) == '\r') 3302 return false; 3303 if (multiline) 3304 return next.match(matcher, i, seq); 3305 } else if (ch == '\r' || ch == '\u0085' || 3306 (ch|1) == '\u2029') { 3307 if (multiline) 3308 return next.match(matcher, i, seq); 3309 } else { // No line terminator, no match 3310 return false; 3311 } 3312 } 3313 // Matched at current end so hit end 3314 matcher.hitEnd = true; 3315 // If a $ matches because of end of input, then more input 3316 // could cause it to fail! 3317 matcher.requireEnd = true; 3318 return next.match(matcher, i, seq); 3319 } 3320 boolean study(TreeInfo info) { 3321 next.study(info); 3322 return info.deterministic; 3323 } 3324 } 3325 3326 /** 3327 * Node to anchor at the end of a line or the end of input based on the 3328 * multiline mode when in unix lines mode. 3329 */ 3330 static final class UnixDollar extends Node { 3331 boolean multiline; 3332 UnixDollar(boolean mul) { 3333 multiline = mul; 3334 } 3335 boolean match(Matcher matcher, int i, CharSequence seq) { 3336 int endIndex = (matcher.anchoringBounds) ? 3337 matcher.to : matcher.getTextLength(); 3338 if (i < endIndex) { 3339 char ch = seq.charAt(i); 3340 if (ch == '\n') { 3341 // If not multiline, then only possible to 3342 // match at very end or one before end 3343 if (multiline == false && i != endIndex - 1) 3344 return false; 3345 // If multiline return next.match without setting 3346 // matcher.hitEnd 3347 if (multiline) 3348 return next.match(matcher, i, seq); 3349 } else { 3350 return false; 3351 } 3352 } 3353 // Matching because at the end or 1 before the end; 3354 // more input could change this so set hitEnd 3355 matcher.hitEnd = true; 3356 // If a $ matches because of end of input, then more input 3357 // could cause it to fail! 3358 matcher.requireEnd = true; 3359 return next.match(matcher, i, seq); 3360 } 3361 boolean study(TreeInfo info) { 3362 next.study(info); 3363 return info.deterministic; 3364 } 3365 } 3366 3367 /** 3368 * Abstract node class to match one character satisfying some 3369 * boolean property. 3370 */ 3371 private static abstract class CharProperty extends Node { 3372 abstract boolean isSatisfiedBy(int ch); 3373 CharProperty complement() { 3374 return new CharProperty() { 3375 boolean isSatisfiedBy(int ch) { 3376 return ! CharProperty.this.isSatisfiedBy(ch);}}; 3377 } 3378 CharProperty maybeComplement(boolean complement) { 3379 return complement ? complement() : this; 3380 } 3381 boolean match(Matcher matcher, int i, CharSequence seq) { 3382 if (i < matcher.to) { 3383 int ch = Character.codePointAt(seq, i); 3384 return isSatisfiedBy(ch) 3385 && next.match(matcher, i+Character.charCount(ch), seq); 3386 } else { 3387 matcher.hitEnd = true; 3388 return false; 3389 } 3390 } 3391 boolean study(TreeInfo info) { 3392 info.minLength++; 3393 info.maxLength++; 3394 return next.study(info); 3395 } 3396 } 3397 3398 /** 3399 * Optimized version of CharProperty that works only for 3400 * properties never satisfied by Supplementary characters. 3401 */ 3402 private static abstract class BmpCharProperty extends CharProperty { 3403 boolean match(Matcher matcher, int i, CharSequence seq) { 3404 if (i < matcher.to) { 3405 return isSatisfiedBy(seq.charAt(i)) 3406 && next.match(matcher, i+1, seq); 3407 } else { 3408 matcher.hitEnd = true; 3409 return false; 3410 } 3411 } 3412 } 3413 3414 /** 3415 * Node class that matches a Supplementary Unicode character 3416 */ 3417 static final class SingleS extends CharProperty { 3418 final int c; 3419 SingleS(int c) { this.c = c; } 3420 boolean isSatisfiedBy(int ch) { 3421 return ch == c; 3422 } 3423 } 3424 3425 /** 3426 * Optimization -- matches a given BMP character 3427 */ 3428 static final class Single extends BmpCharProperty { 3429 final int c; 3430 Single(int c) { this.c = c; } 3431 boolean isSatisfiedBy(int ch) { 3432 return ch == c; 3433 } 3434 } 3435 3436 /** 3437 * Case insensitive matches a given BMP character 3438 */ 3439 static final class SingleI extends BmpCharProperty { 3440 final int lower; 3441 final int upper; 3442 SingleI(int lower, int upper) { 3443 this.lower = lower; 3444 this.upper = upper; 3445 } 3446 boolean isSatisfiedBy(int ch) { 3447 return ch == lower || ch == upper; 3448 } 3449 } 3450 3451 /** 3452 * Unicode case insensitive matches a given Unicode character 3453 */ 3454 static final class SingleU extends CharProperty { 3455 final int lower; 3456 SingleU(int lower) { 3457 this.lower = lower; 3458 } 3459 boolean isSatisfiedBy(int ch) { 3460 return lower == ch || 3461 lower == Character.toLowerCase(Character.toUpperCase(ch)); 3462 } 3463 } 3464 3465 /** 3466 * Node class that matches a Unicode category. 3467 */ 3468 static final class Category extends CharProperty { 3469 final int typeMask; 3470 Category(int typeMask) { this.typeMask = typeMask; } 3471 boolean isSatisfiedBy(int ch) { 3472 return (typeMask & (1 << Character.getType(ch))) != 0; 3473 } 3474 } 3475 3476 /** 3477 * Node class that matches a POSIX type. 3478 */ 3479 static final class Ctype extends BmpCharProperty { 3480 final int ctype; 3481 Ctype(int ctype) { this.ctype = ctype; } 3482 boolean isSatisfiedBy(int ch) { 3483 return ch < 128 && ASCII.isType(ch, ctype); 3484 } 3485 } 3486 3487 /** 3488 * Base class for all Slice nodes 3489 */ 3490 static class SliceNode extends Node { 3491 int[] buffer; 3492 SliceNode(int[] buf) { 3493 buffer = buf; 3494 } 3495 boolean study(TreeInfo info) { 3496 info.minLength += buffer.length; 3497 info.maxLength += buffer.length; 3498 return next.study(info); 3499 } 3500 } 3501 3502 /** 3503 * Node class for a case sensitive/BMP-only sequence of literal 3504 * characters. 3505 */ 3506 static final class Slice extends SliceNode { 3507 Slice(int[] buf) { 3508 super(buf); 3509 } 3510 boolean match(Matcher matcher, int i, CharSequence seq) { 3511 int[] buf = buffer; 3512 int len = buf.length; 3513 for (int j=0; j<len; j++) { 3514 if ((i+j) >= matcher.to) { 3515 matcher.hitEnd = true; 3516 return false; 3517 } 3518 if (buf[j] != seq.charAt(i+j)) 3519 return false; 3520 } 3521 return next.match(matcher, i+len, seq); 3522 } 3523 } 3524 3525 /** 3526 * Node class for a case_insensitive/BMP-only sequence of literal 3527 * characters. 3528 */ 3529 static class SliceI extends SliceNode { 3530 SliceI(int[] buf) { 3531 super(buf); 3532 } 3533 boolean match(Matcher matcher, int i, CharSequence seq) { 3534 int[] buf = buffer; 3535 int len = buf.length; 3536 for (int j=0; j<len; j++) { 3537 if ((i+j) >= matcher.to) { 3538 matcher.hitEnd = true; 3539 return false; 3540 } 3541 int c = seq.charAt(i+j); 3542 if (buf[j] != c && 3543 buf[j] != ASCII.toLower(c)) 3544 return false; 3545 } 3546 return next.match(matcher, i+len, seq); 3547 } 3548 } 3549 3550 /** 3551 * Node class for a unicode_case_insensitive/BMP-only sequence of 3552 * literal characters. Uses unicode case folding. 3553 */ 3554 static final class SliceU extends SliceNode { 3555 SliceU(int[] buf) { 3556 super(buf); 3557 } 3558 boolean match(Matcher matcher, int i, CharSequence seq) { 3559 int[] buf = buffer; 3560 int len = buf.length; 3561 for (int j=0; j<len; j++) { 3562 if ((i+j) >= matcher.to) { 3563 matcher.hitEnd = true; 3564 return false; 3565 } 3566 int c = seq.charAt(i+j); 3567 if (buf[j] != c && 3568 buf[j] != Character.toLowerCase(Character.toUpperCase(c))) 3569 return false; 3570 } 3571 return next.match(matcher, i+len, seq); 3572 } 3573 } 3574 3575 /** 3576 * Node class for a case sensitive sequence of literal characters 3577 * including supplementary characters. 3578 */ 3579 static final class SliceS extends SliceNode { 3580 SliceS(int[] buf) { 3581 super(buf); 3582 } 3583 boolean match(Matcher matcher, int i, CharSequence seq) { 3584 int[] buf = buffer; 3585 int x = i; 3586 for (int j = 0; j < buf.length; j++) { 3587 if (x >= matcher.to) { 3588 matcher.hitEnd = true; 3589 return false; 3590 } 3591 int c = Character.codePointAt(seq, x); 3592 if (buf[j] != c) 3593 return false; 3594 x += Character.charCount(c); 3595 if (x > matcher.to) { 3596 matcher.hitEnd = true; 3597 return false; 3598 } 3599 } 3600 return next.match(matcher, x, seq); 3601 } 3602 } 3603 3604 /** 3605 * Node class for a case insensitive sequence of literal characters 3606 * including supplementary characters. 3607 */ 3608 static class SliceIS extends SliceNode { 3609 SliceIS(int[] buf) { 3610 super(buf); 3611 } 3612 int toLower(int c) { 3613 return ASCII.toLower(c); 3614 } 3615 boolean match(Matcher matcher, int i, CharSequence seq) { 3616 int[] buf = buffer; 3617 int x = i; 3618 for (int j = 0; j < buf.length; j++) { 3619 if (x >= matcher.to) { 3620 matcher.hitEnd = true; 3621 return false; 3622 } 3623 int c = Character.codePointAt(seq, x); 3624 if (buf[j] != c && buf[j] != toLower(c)) 3625 return false; 3626 x += Character.charCount(c); 3627 if (x > matcher.to) { 3628 matcher.hitEnd = true; 3629 return false; 3630 } 3631 } 3632 return next.match(matcher, x, seq); 3633 } 3634 } 3635 3636 /** 3637 * Node class for a case insensitive sequence of literal characters. 3638 * Uses unicode case folding. 3639 */ 3640 static final class SliceUS extends SliceIS { 3641 SliceUS(int[] buf) { 3642 super(buf); 3643 } 3644 int toLower(int c) { 3645 return Character.toLowerCase(Character.toUpperCase(c)); 3646 } 3647 } 3648 3649 private static boolean inRange(int lower, int ch, int upper) { 3650 return lower <= ch && ch <= upper; 3651 } 3652 3653 /** 3654 * Returns node for matching characters within an explicit value range. 3655 */ 3656 private static CharProperty rangeFor(final int lower, 3657 final int upper) { 3658 return new CharProperty() { 3659 boolean isSatisfiedBy(int ch) { 3660 return inRange(lower, ch, upper);}}; 3661 } 3662 3663 /** 3664 * Returns node for matching characters within an explicit value 3665 * range in a case insensitive manner. 3666 */ 3667 private CharProperty caseInsensitiveRangeFor(final int lower, 3668 final int upper) { 3669 if (has(UNICODE_CASE)) 3670 return new CharProperty() { 3671 boolean isSatisfiedBy(int ch) { 3672 if (inRange(lower, ch, upper)) 3673 return true; 3674 int up = Character.toUpperCase(ch); 3675 return inRange(lower, up, upper) || 3676 inRange(lower, Character.toLowerCase(up), upper);}}; 3677 return new CharProperty() { 3678 boolean isSatisfiedBy(int ch) { 3679 return inRange(lower, ch, upper) || 3680 ASCII.isAscii(ch) && 3681 (inRange(lower, ASCII.toUpper(ch), upper) || 3682 inRange(lower, ASCII.toLower(ch), upper)); 3683 }}; 3684 } 3685 3686 /** 3687 * Implements the Unicode category ALL and the dot metacharacter when 3688 * in dotall mode. 3689 */ 3690 static final class All extends CharProperty { 3691 boolean isSatisfiedBy(int ch) { 3692 return true; 3693 } 3694 } 3695 3696 /** 3697 * Node class for the dot metacharacter when dotall is not enabled. 3698 */ 3699 static final class Dot extends CharProperty { 3700 boolean isSatisfiedBy(int ch) { 3701 return (ch != '\n' && ch != '\r' 3702 && (ch|1) != '\u2029' 3703 && ch != '\u0085'); 3704 } 3705 } 3706 3707 /** 3708 * Node class for the dot metacharacter when dotall is not enabled 3709 * but UNIX_LINES is enabled. 3710 */ 3711 static final class UnixDot extends CharProperty { 3712 boolean isSatisfiedBy(int ch) { 3713 return ch != '\n'; 3714 } 3715 } 3716 3717 /** 3718 * The 0 or 1 quantifier. This one class implements all three types. 3719 */ 3720 static final class Ques extends Node { 3721 Node atom; 3722 int type; 3723 Ques(Node node, int type) { 3724 this.atom = node; 3725 this.type = type; 3726 } 3727 boolean match(Matcher matcher, int i, CharSequence seq) { 3728 switch (type) { 3729 case GREEDY: 3730 return (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)) 3731 || next.match(matcher, i, seq); 3732 case LAZY: 3733 return next.match(matcher, i, seq) 3734 || (atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq)); 3735 case POSSESSIVE: 3736 if (atom.match(matcher, i, seq)) i = matcher.last; 3737 return next.match(matcher, i, seq); 3738 default: 3739 return atom.match(matcher, i, seq) && next.match(matcher, matcher.last, seq); 3740 } 3741 } 3742 boolean study(TreeInfo info) { 3743 if (type != INDEPENDENT) { 3744 int minL = info.minLength; 3745 atom.study(info); 3746 info.minLength = minL; 3747 info.deterministic = false; 3748 return next.study(info); 3749 } else { 3750 atom.study(info); 3751 return next.study(info); 3752 } 3753 } 3754 } 3755 3756 /** 3757 * Handles the curly-brace style repetition with a specified minimum and 3758 * maximum occurrences. The * quantifier is handled as a special case. 3759 * This class handles the three types. 3760 */ 3761 static final class Curly extends Node { 3762 Node atom; 3763 int type; 3764 int cmin; 3765 int cmax; 3766 3767 Curly(Node node, int cmin, int cmax, int type) { 3768 this.atom = node; 3769 this.type = type; 3770 this.cmin = cmin; 3771 this.cmax = cmax; 3772 } 3773 boolean match(Matcher matcher, int i, CharSequence seq) { 3774 int j; 3775 for (j = 0; j < cmin; j++) { 3776 if (atom.match(matcher, i, seq)) { 3777 i = matcher.last; 3778 continue; 3779 } 3780 return false; 3781 } 3782 if (type == GREEDY) 3783 return match0(matcher, i, j, seq); 3784 else if (type == LAZY) 3785 return match1(matcher, i, j, seq); 3786 else 3787 return match2(matcher, i, j, seq); 3788 } 3789 // Greedy match. 3790 // i is the index to start matching at 3791 // j is the number of atoms that have matched 3792 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 3793 if (j >= cmax) { 3794 // We have matched the maximum... continue with the rest of 3795 // the regular expression 3796 return next.match(matcher, i, seq); 3797 } 3798 int backLimit = j; 3799 while (atom.match(matcher, i, seq)) { 3800 // k is the length of this match 3801 int k = matcher.last - i; 3802 if (k == 0) // Zero length match 3803 break; 3804 // Move up index and number matched 3805 i = matcher.last; 3806 j++; 3807 // We are greedy so match as many as we can 3808 while (j < cmax) { 3809 if (!atom.match(matcher, i, seq)) 3810 break; 3811 if (i + k != matcher.last) { 3812 if (match0(matcher, matcher.last, j+1, seq)) 3813 return true; 3814 break; 3815 } 3816 i += k; 3817 j++; 3818 } 3819 // Handle backing off if match fails 3820 while (j >= backLimit) { 3821 if (next.match(matcher, i, seq)) 3822 return true; 3823 i -= k; 3824 j--; 3825 } 3826 return false; 3827 } 3828 return next.match(matcher, i, seq); 3829 } 3830 // Reluctant match. At this point, the minimum has been satisfied. 3831 // i is the index to start matching at 3832 // j is the number of atoms that have matched 3833 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 3834 for (;;) { 3835 // Try finishing match without consuming any more 3836 if (next.match(matcher, i, seq)) 3837 return true; 3838 // At the maximum, no match found 3839 if (j >= cmax) 3840 return false; 3841 // Okay, must try one more atom 3842 if (!atom.match(matcher, i, seq)) 3843 return false; 3844 // If we haven't moved forward then must break out 3845 if (i == matcher.last) 3846 return false; 3847 // Move up index and number matched 3848 i = matcher.last; 3849 j++; 3850 } 3851 } 3852 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 3853 for (; j < cmax; j++) { 3854 if (!atom.match(matcher, i, seq)) 3855 break; 3856 if (i == matcher.last) 3857 break; 3858 i = matcher.last; 3859 } 3860 return next.match(matcher, i, seq); 3861 } 3862 boolean study(TreeInfo info) { 3863 // Save original info 3864 int minL = info.minLength; 3865 int maxL = info.maxLength; 3866 boolean maxV = info.maxValid; 3867 boolean detm = info.deterministic; 3868 info.reset(); 3869 3870 atom.study(info); 3871 3872 int temp = info.minLength * cmin + minL; 3873 if (temp < minL) { 3874 temp = 0xFFFFFFF; // arbitrary large number 3875 } 3876 info.minLength = temp; 3877 3878 if (maxV & info.maxValid) { 3879 temp = info.maxLength * cmax + maxL; 3880 info.maxLength = temp; 3881 if (temp < maxL) { 3882 info.maxValid = false; 3883 } 3884 } else { 3885 info.maxValid = false; 3886 } 3887 3888 if (info.deterministic && cmin == cmax) 3889 info.deterministic = detm; 3890 else 3891 info.deterministic = false; 3892 3893 return next.study(info); 3894 } 3895 } 3896 3897 /** 3898 * Handles the curly-brace style repetition with a specified minimum and 3899 * maximum occurrences in deterministic cases. This is an iterative 3900 * optimization over the Prolog and Loop system which would handle this 3901 * in a recursive way. The * quantifier is handled as a special case. 3902 * If capture is true then this class saves group settings and ensures 3903 * that groups are unset when backing off of a group match. 3904 */ 3905 static final class GroupCurly extends Node { 3906 Node atom; 3907 int type; 3908 int cmin; 3909 int cmax; 3910 int localIndex; 3911 int groupIndex; 3912 boolean capture; 3913 3914 GroupCurly(Node node, int cmin, int cmax, int type, int local, 3915 int group, boolean capture) { 3916 this.atom = node; 3917 this.type = type; 3918 this.cmin = cmin; 3919 this.cmax = cmax; 3920 this.localIndex = local; 3921 this.groupIndex = group; 3922 this.capture = capture; 3923 } 3924 boolean match(Matcher matcher, int i, CharSequence seq) { 3925 int[] groups = matcher.groups; 3926 int[] locals = matcher.locals; 3927 int save0 = locals[localIndex]; 3928 int save1 = 0; 3929 int save2 = 0; 3930 3931 if (capture) { 3932 save1 = groups[groupIndex]; 3933 save2 = groups[groupIndex+1]; 3934 } 3935 3936 // Notify GroupTail there is no need to setup group info 3937 // because it will be set here 3938 locals[localIndex] = -1; 3939 3940 boolean ret = true; 3941 for (int j = 0; j < cmin; j++) { 3942 if (atom.match(matcher, i, seq)) { 3943 if (capture) { 3944 groups[groupIndex] = i; 3945 groups[groupIndex+1] = matcher.last; 3946 } 3947 i = matcher.last; 3948 } else { 3949 ret = false; 3950 break; 3951 } 3952 } 3953 if (ret) { 3954 if (type == GREEDY) { 3955 ret = match0(matcher, i, cmin, seq); 3956 } else if (type == LAZY) { 3957 ret = match1(matcher, i, cmin, seq); 3958 } else { 3959 ret = match2(matcher, i, cmin, seq); 3960 } 3961 } 3962 if (!ret) { 3963 locals[localIndex] = save0; 3964 if (capture) { 3965 groups[groupIndex] = save1; 3966 groups[groupIndex+1] = save2; 3967 } 3968 } 3969 return ret; 3970 } 3971 // Aggressive group match 3972 boolean match0(Matcher matcher, int i, int j, CharSequence seq) { 3973 int[] groups = matcher.groups; 3974 int save0 = 0; 3975 int save1 = 0; 3976 if (capture) { 3977 save0 = groups[groupIndex]; 3978 save1 = groups[groupIndex+1]; 3979 } 3980 for (;;) { 3981 if (j >= cmax) 3982 break; 3983 if (!atom.match(matcher, i, seq)) 3984 break; 3985 int k = matcher.last - i; 3986 if (k <= 0) { 3987 if (capture) { 3988 groups[groupIndex] = i; 3989 groups[groupIndex+1] = i + k; 3990 } 3991 i = i + k; 3992 break; 3993 } 3994 for (;;) { 3995 if (capture) { 3996 groups[groupIndex] = i; 3997 groups[groupIndex+1] = i + k; 3998 } 3999 i = i + k; 4000 if (++j >= cmax) 4001 break; 4002 if (!atom.match(matcher, i, seq)) 4003 break; 4004 if (i + k != matcher.last) { 4005 if (match0(matcher, i, j, seq)) 4006 return true; 4007 break; 4008 } 4009 } 4010 while (j > cmin) { 4011 if (next.match(matcher, i, seq)) { 4012 if (capture) { 4013 groups[groupIndex+1] = i; 4014 groups[groupIndex] = i - k; 4015 } 4016 i = i - k; 4017 return true; 4018 } 4019 // backing off 4020 if (capture) { 4021 groups[groupIndex+1] = i; 4022 groups[groupIndex] = i - k; 4023 } 4024 i = i - k; 4025 j--; 4026 } 4027 break; 4028 } 4029 if (capture) { 4030 groups[groupIndex] = save0; 4031 groups[groupIndex+1] = save1; 4032 } 4033 return next.match(matcher, i, seq); 4034 } 4035 // Reluctant matching 4036 boolean match1(Matcher matcher, int i, int j, CharSequence seq) { 4037 for (;;) { 4038 if (next.match(matcher, i, seq)) 4039 return true; 4040 if (j >= cmax) 4041 return false; 4042 if (!atom.match(matcher, i, seq)) 4043 return false; 4044 if (i == matcher.last) 4045 return false; 4046 if (capture) { 4047 matcher.groups[groupIndex] = i; 4048 matcher.groups[groupIndex+1] = matcher.last; 4049 } 4050 i = matcher.last; 4051 j++; 4052 } 4053 } 4054 // Possessive matching 4055 boolean match2(Matcher matcher, int i, int j, CharSequence seq) { 4056 for (; j < cmax; j++) { 4057 if (!atom.match(matcher, i, seq)) { 4058 break; 4059 } 4060 if (capture) { 4061 matcher.groups[groupIndex] = i; 4062 matcher.groups[groupIndex+1] = matcher.last; 4063 } 4064 if (i == matcher.last) { 4065 break; 4066 } 4067 i = matcher.last; 4068 } 4069 return next.match(matcher, i, seq); 4070 } 4071 boolean study(TreeInfo info) { 4072 // Save original info 4073 int minL = info.minLength; 4074 int maxL = info.maxLength; 4075 boolean maxV = info.maxValid; 4076 boolean detm = info.deterministic; 4077 info.reset(); 4078 4079 atom.study(info); 4080 4081 int temp = info.minLength * cmin + minL; 4082 if (temp < minL) { 4083 temp = 0xFFFFFFF; // Arbitrary large number 4084 } 4085 info.minLength = temp; 4086 4087 if (maxV & info.maxValid) { 4088 temp = info.maxLength * cmax + maxL; 4089 info.maxLength = temp; 4090 if (temp < maxL) { 4091 info.maxValid = false; 4092 } 4093 } else { 4094 info.maxValid = false; 4095 } 4096 4097 if (info.deterministic && cmin == cmax) { 4098 info.deterministic = detm; 4099 } else { 4100 info.deterministic = false; 4101 } 4102 4103 return next.study(info); 4104 } 4105 } 4106 4107 /** 4108 * A Guard node at the end of each atom node in a Branch. It 4109 * serves the purpose of chaining the "match" operation to 4110 * "next" but not the "study", so we can collect the TreeInfo 4111 * of each atom node without including the TreeInfo of the 4112 * "next". 4113 */ 4114 static final class BranchConn extends Node { 4115 BranchConn() {}; 4116 boolean match(Matcher matcher, int i, CharSequence seq) { 4117 return next.match(matcher, i, seq); 4118 } 4119 boolean study(TreeInfo info) { 4120 return info.deterministic; 4121 } 4122 } 4123 4124 /** 4125 * Handles the branching of alternations. Note this is also used for 4126 * the ? quantifier to branch between the case where it matches once 4127 * and where it does not occur. 4128 */ 4129 static final class Branch extends Node { 4130 Node[] atoms = new Node[2]; 4131 int size = 2; 4132 Node conn; 4133 Branch(Node first, Node second, Node branchConn) { 4134 conn = branchConn; 4135 atoms[0] = first; 4136 atoms[1] = second; 4137 } 4138 4139 void add(Node node) { 4140 if (size >= atoms.length) { 4141 Node[] tmp = new Node[atoms.length*2]; 4142 System.arraycopy(atoms, 0, tmp, 0, atoms.length); 4143 atoms = tmp; 4144 } 4145 atoms[size++] = node; 4146 } 4147 4148 boolean match(Matcher matcher, int i, CharSequence seq) { 4149 for (int n = 0; n < size; n++) { 4150 if (atoms[n] == null) { 4151 if (conn.next.match(matcher, i, seq)) 4152 return true; 4153 } else if (atoms[n].match(matcher, i, seq)) { 4154 return true; 4155 } 4156 } 4157 return false; 4158 } 4159 4160 boolean study(TreeInfo info) { 4161 int minL = info.minLength; 4162 int maxL = info.maxLength; 4163 boolean maxV = info.maxValid; 4164 4165 int minL2 = Integer.MAX_VALUE; //arbitrary large enough num 4166 int maxL2 = -1; 4167 for (int n = 0; n < size; n++) { 4168 info.reset(); 4169 if (atoms[n] != null) 4170 atoms[n].study(info); 4171 minL2 = Math.min(minL2, info.minLength); 4172 maxL2 = Math.max(maxL2, info.maxLength); 4173 maxV = (maxV & info.maxValid); 4174 } 4175 4176 minL += minL2; 4177 maxL += maxL2; 4178 4179 info.reset(); 4180 conn.next.study(info); 4181 4182 info.minLength += minL; 4183 info.maxLength += maxL; 4184 info.maxValid &= maxV; 4185 info.deterministic = false; 4186 return false; 4187 } 4188 } 4189 4190 /** 4191 * The GroupHead saves the location where the group begins in the locals 4192 * and restores them when the match is done. 4193 * 4194 * The matchRef is used when a reference to this group is accessed later 4195 * in the expression. The locals will have a negative value in them to 4196 * indicate that we do not want to unset the group if the reference 4197 * doesn't match. 4198 */ 4199 static final class GroupHead extends Node { 4200 int localIndex; 4201 GroupHead(int localCount) { 4202 localIndex = localCount; 4203 } 4204 boolean match(Matcher matcher, int i, CharSequence seq) { 4205 int save = matcher.locals[localIndex]; 4206 matcher.locals[localIndex] = i; 4207 boolean ret = next.match(matcher, i, seq); 4208 matcher.locals[localIndex] = save; 4209 return ret; 4210 } 4211 boolean matchRef(Matcher matcher, int i, CharSequence seq) { 4212 int save = matcher.locals[localIndex]; 4213 matcher.locals[localIndex] = ~i; // HACK 4214 boolean ret = next.match(matcher, i, seq); 4215 matcher.locals[localIndex] = save; 4216 return ret; 4217 } 4218 } 4219 4220 /** 4221 * Recursive reference to a group in the regular expression. It calls 4222 * matchRef because if the reference fails to match we would not unset 4223 * the group. 4224 */ 4225 static final class GroupRef extends Node { 4226 GroupHead head; 4227 GroupRef(GroupHead head) { 4228 this.head = head; 4229 } 4230 boolean match(Matcher matcher, int i, CharSequence seq) { 4231 return head.matchRef(matcher, i, seq) 4232 && next.match(matcher, matcher.last, seq); 4233 } 4234 boolean study(TreeInfo info) { 4235 info.maxValid = false; 4236 info.deterministic = false; 4237 return next.study(info); 4238 } 4239 } 4240 4241 /** 4242 * The GroupTail handles the setting of group beginning and ending 4243 * locations when groups are successfully matched. It must also be able to 4244 * unset groups that have to be backed off of. 4245 * 4246 * The GroupTail node is also used when a previous group is referenced, 4247 * and in that case no group information needs to be set. 4248 */ 4249 static final class GroupTail extends Node { 4250 int localIndex; 4251 int groupIndex; 4252 GroupTail(int localCount, int groupCount) { 4253 localIndex = localCount; 4254 groupIndex = groupCount + groupCount; 4255 } 4256 boolean match(Matcher matcher, int i, CharSequence seq) { 4257 int tmp = matcher.locals[localIndex]; 4258 if (tmp >= 0) { // This is the normal group case. 4259 // Save the group so we can unset it if it 4260 // backs off of a match. 4261 int groupStart = matcher.groups[groupIndex]; 4262 int groupEnd = matcher.groups[groupIndex+1]; 4263 4264 matcher.groups[groupIndex] = tmp; 4265 matcher.groups[groupIndex+1] = i; 4266 if (next.match(matcher, i, seq)) { 4267 return true; 4268 } 4269 matcher.groups[groupIndex] = groupStart; 4270 matcher.groups[groupIndex+1] = groupEnd; 4271 return false; 4272 } else { 4273 // This is a group reference case. We don't need to save any 4274 // group info because it isn't really a group. 4275 matcher.last = i; 4276 return true; 4277 } 4278 } 4279 } 4280 4281 /** 4282 * This sets up a loop to handle a recursive quantifier structure. 4283 */ 4284 static final class Prolog extends Node { 4285 Loop loop; 4286 Prolog(Loop loop) { 4287 this.loop = loop; 4288 } 4289 boolean match(Matcher matcher, int i, CharSequence seq) { 4290 return loop.matchInit(matcher, i, seq); 4291 } 4292 boolean study(TreeInfo info) { 4293 return loop.study(info); 4294 } 4295 } 4296 4297 /** 4298 * Handles the repetition count for a greedy Curly. The matchInit 4299 * is called from the Prolog to save the index of where the group 4300 * beginning is stored. A zero length group check occurs in the 4301 * normal match but is skipped in the matchInit. 4302 */ 4303 static class Loop extends Node { 4304 Node body; 4305 int countIndex; // local count index in matcher locals 4306 int beginIndex; // group beginning index 4307 int cmin, cmax; 4308 Loop(int countIndex, int beginIndex) { 4309 this.countIndex = countIndex; 4310 this.beginIndex = beginIndex; 4311 } 4312 boolean match(Matcher matcher, int i, CharSequence seq) { 4313 // Avoid infinite loop in zero-length case. 4314 if (i > matcher.locals[beginIndex]) { 4315 int count = matcher.locals[countIndex]; 4316 4317 // This block is for before we reach the minimum 4318 // iterations required for the loop to match 4319 if (count < cmin) { 4320 matcher.locals[countIndex] = count + 1; 4321 boolean b = body.match(matcher, i, seq); 4322 // If match failed we must backtrack, so 4323 // the loop count should NOT be incremented 4324 if (!b) 4325 matcher.locals[countIndex] = count; 4326 // Return success or failure since we are under 4327 // minimum 4328 return b; 4329 } 4330 // This block is for after we have the minimum 4331 // iterations required for the loop to match 4332 if (count < cmax) { 4333 matcher.locals[countIndex] = count + 1; 4334 boolean b = body.match(matcher, i, seq); 4335 // If match failed we must backtrack, so 4336 // the loop count should NOT be incremented 4337 if (!b) 4338 matcher.locals[countIndex] = count; 4339 else 4340 return true; 4341 } 4342 } 4343 return next.match(matcher, i, seq); 4344 } 4345 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4346 int save = matcher.locals[countIndex]; 4347 boolean ret = false; 4348 if (0 < cmin) { 4349 matcher.locals[countIndex] = 1; 4350 ret = body.match(matcher, i, seq); 4351 } else if (0 < cmax) { 4352 matcher.locals[countIndex] = 1; 4353 ret = body.match(matcher, i, seq); 4354 if (ret == false) 4355 ret = next.match(matcher, i, seq); 4356 } else { 4357 ret = next.match(matcher, i, seq); 4358 } 4359 matcher.locals[countIndex] = save; 4360 return ret; 4361 } 4362 boolean study(TreeInfo info) { 4363 info.maxValid = false; 4364 info.deterministic = false; 4365 return false; 4366 } 4367 } 4368 4369 /** 4370 * Handles the repetition count for a reluctant Curly. The matchInit 4371 * is called from the Prolog to save the index of where the group 4372 * beginning is stored. A zero length group check occurs in the 4373 * normal match but is skipped in the matchInit. 4374 */ 4375 static final class LazyLoop extends Loop { 4376 LazyLoop(int countIndex, int beginIndex) { 4377 super(countIndex, beginIndex); 4378 } 4379 boolean match(Matcher matcher, int i, CharSequence seq) { 4380 // Check for zero length group 4381 if (i > matcher.locals[beginIndex]) { 4382 int count = matcher.locals[countIndex]; 4383 if (count < cmin) { 4384 matcher.locals[countIndex] = count + 1; 4385 boolean result = body.match(matcher, i, seq); 4386 // If match failed we must backtrack, so 4387 // the loop count should NOT be incremented 4388 if (!result) 4389 matcher.locals[countIndex] = count; 4390 return result; 4391 } 4392 if (next.match(matcher, i, seq)) 4393 return true; 4394 if (count < cmax) { 4395 matcher.locals[countIndex] = count + 1; 4396 boolean result = body.match(matcher, i, seq); 4397 // If match failed we must backtrack, so 4398 // the loop count should NOT be incremented 4399 if (!result) 4400 matcher.locals[countIndex] = count; 4401 return result; 4402 } 4403 return false; 4404 } 4405 return next.match(matcher, i, seq); 4406 } 4407 boolean matchInit(Matcher matcher, int i, CharSequence seq) { 4408 int save = matcher.locals[countIndex]; 4409 boolean ret = false; 4410 if (0 < cmin) { 4411 matcher.locals[countIndex] = 1; 4412 ret = body.match(matcher, i, seq); 4413 } else if (next.match(matcher, i, seq)) { 4414 ret = true; 4415 } else if (0 < cmax) { 4416 matcher.locals[countIndex] = 1; 4417 ret = body.match(matcher, i, seq); 4418 } 4419 matcher.locals[countIndex] = save; 4420 return ret; 4421 } 4422 boolean study(TreeInfo info) { 4423 info.maxValid = false; 4424 info.deterministic = false; 4425 return false; 4426 } 4427 } 4428 4429 /** 4430 * Refers to a group in the regular expression. Attempts to match 4431 * whatever the group referred to last matched. 4432 */ 4433 static class BackRef extends Node { 4434 int groupIndex; 4435 BackRef(int groupCount) { 4436 super(); 4437 groupIndex = groupCount + groupCount; 4438 } 4439 boolean match(Matcher matcher, int i, CharSequence seq) { 4440 int j = matcher.groups[groupIndex]; 4441 int k = matcher.groups[groupIndex+1]; 4442 4443 int groupSize = k - j; 4444 4445 // If the referenced group didn't match, neither can this 4446 if (j < 0) 4447 return false; 4448 4449 // If there isn't enough input left no match 4450 if (i + groupSize > matcher.to) { 4451 matcher.hitEnd = true; 4452 return false; 4453 } 4454 4455 // Check each new char to make sure it matches what the group 4456 // referenced matched last time around 4457 for (int index=0; index<groupSize; index++) 4458 if (seq.charAt(i+index) != seq.charAt(j+index)) 4459 return false; 4460 4461 return next.match(matcher, i+groupSize, seq); 4462 } 4463 boolean study(TreeInfo info) { 4464 info.maxValid = false; 4465 return next.study(info); 4466 } 4467 } 4468 4469 static class CIBackRef extends Node { 4470 int groupIndex; 4471 boolean doUnicodeCase; 4472 CIBackRef(int groupCount, boolean doUnicodeCase) { 4473 super(); 4474 groupIndex = groupCount + groupCount; 4475 this.doUnicodeCase = doUnicodeCase; 4476 } 4477 boolean match(Matcher matcher, int i, CharSequence seq) { 4478 int j = matcher.groups[groupIndex]; 4479 int k = matcher.groups[groupIndex+1]; 4480 4481 int groupSize = k - j; 4482 4483 // If the referenced group didn't match, neither can this 4484 if (j < 0) 4485 return false; 4486 4487 // If there isn't enough input left no match 4488 if (i + groupSize > matcher.to) { 4489 matcher.hitEnd = true; 4490 return false; 4491 } 4492 4493 // Check each new char to make sure it matches what the group 4494 // referenced matched last time around 4495 int x = i; 4496 for (int index=0; index<groupSize; index++) { 4497 int c1 = Character.codePointAt(seq, x); 4498 int c2 = Character.codePointAt(seq, j); 4499 if (c1 != c2) { 4500 if (doUnicodeCase) { 4501 int cc1 = Character.toUpperCase(c1); 4502 int cc2 = Character.toUpperCase(c2); 4503 if (cc1 != cc2 && 4504 Character.toLowerCase(cc1) != 4505 Character.toLowerCase(cc2)) 4506 return false; 4507 } else { 4508 if (ASCII.toLower(c1) != ASCII.toLower(c2)) 4509 return false; 4510 } 4511 } 4512 x += Character.charCount(c1); 4513 j += Character.charCount(c2); 4514 } 4515 4516 return next.match(matcher, i+groupSize, seq); 4517 } 4518 boolean study(TreeInfo info) { 4519 info.maxValid = false; 4520 return next.study(info); 4521 } 4522 } 4523 4524 /** 4525 * Searches until the next instance of its atom. This is useful for 4526 * finding the atom efficiently without passing an instance of it 4527 * (greedy problem) and without a lot of wasted search time (reluctant 4528 * problem). 4529 */ 4530 static final class First extends Node { 4531 Node atom; 4532 First(Node node) { 4533 this.atom = BnM.optimize(node); 4534 } 4535 boolean match(Matcher matcher, int i, CharSequence seq) { 4536 if (atom instanceof BnM) { 4537 return atom.match(matcher, i, seq) 4538 && next.match(matcher, matcher.last, seq); 4539 } 4540 for (;;) { 4541 if (i > matcher.to) { 4542 matcher.hitEnd = true; 4543 return false; 4544 } 4545 if (atom.match(matcher, i, seq)) { 4546 return next.match(matcher, matcher.last, seq); 4547 } 4548 i += countChars(seq, i, 1); 4549 matcher.first++; 4550 } 4551 } 4552 boolean study(TreeInfo info) { 4553 atom.study(info); 4554 info.maxValid = false; 4555 info.deterministic = false; 4556 return next.study(info); 4557 } 4558 } 4559 4560 static final class Conditional extends Node { 4561 Node cond, yes, not; 4562 Conditional(Node cond, Node yes, Node not) { 4563 this.cond = cond; 4564 this.yes = yes; 4565 this.not = not; 4566 } 4567 boolean match(Matcher matcher, int i, CharSequence seq) { 4568 if (cond.match(matcher, i, seq)) { 4569 return yes.match(matcher, i, seq); 4570 } else { 4571 return not.match(matcher, i, seq); 4572 } 4573 } 4574 boolean study(TreeInfo info) { 4575 int minL = info.minLength; 4576 int maxL = info.maxLength; 4577 boolean maxV = info.maxValid; 4578 info.reset(); 4579 yes.study(info); 4580 4581 int minL2 = info.minLength; 4582 int maxL2 = info.maxLength; 4583 boolean maxV2 = info.maxValid; 4584 info.reset(); 4585 not.study(info); 4586 4587 info.minLength = minL + Math.min(minL2, info.minLength); 4588 info.maxLength = maxL + Math.max(maxL2, info.maxLength); 4589 info.maxValid = (maxV & maxV2 & info.maxValid); 4590 info.deterministic = false; 4591 return next.study(info); 4592 } 4593 } 4594 4595 /** 4596 * Zero width positive lookahead. 4597 */ 4598 static final class Pos extends Node { 4599 Node cond; 4600 Pos(Node cond) { 4601 this.cond = cond; 4602 } 4603 boolean match(Matcher matcher, int i, CharSequence seq) { 4604 int savedTo = matcher.to; 4605 boolean conditionMatched = false; 4606 4607 // Relax transparent region boundaries for lookahead 4608 if (matcher.transparentBounds) 4609 matcher.to = matcher.getTextLength(); 4610 try { 4611 conditionMatched = cond.match(matcher, i, seq); 4612 } finally { 4613 // Reinstate region boundaries 4614 matcher.to = savedTo; 4615 } 4616 return conditionMatched && next.match(matcher, i, seq); 4617 } 4618 } 4619 4620 /** 4621 * Zero width negative lookahead. 4622 */ 4623 static final class Neg extends Node { 4624 Node cond; 4625 Neg(Node cond) { 4626 this.cond = cond; 4627 } 4628 boolean match(Matcher matcher, int i, CharSequence seq) { 4629 int savedTo = matcher.to; 4630 boolean conditionMatched = false; 4631 4632 // Relax transparent region boundaries for lookahead 4633 if (matcher.transparentBounds) 4634 matcher.to = matcher.getTextLength(); 4635 try { 4636 if (i < matcher.to) { 4637 conditionMatched = !cond.match(matcher, i, seq); 4638 } else { 4639 // If a negative lookahead succeeds then more input 4640 // could cause it to fail! 4641 matcher.requireEnd = true; 4642 conditionMatched = !cond.match(matcher, i, seq); 4643 } 4644 } finally { 4645 // Reinstate region boundaries 4646 matcher.to = savedTo; 4647 } 4648 return conditionMatched && next.match(matcher, i, seq); 4649 } 4650 } 4651 4652 /** 4653 * For use with lookbehinds; matches the position where the lookbehind 4654 * was encountered. 4655 */ 4656 static Node lookbehindEnd = new Node() { 4657 boolean match(Matcher matcher, int i, CharSequence seq) { 4658 return i == matcher.lookbehindTo; 4659 } 4660 }; 4661 4662 /** 4663 * Zero width positive lookbehind. 4664 */ 4665 static class Behind extends Node { 4666 Node cond; 4667 int rmax, rmin; 4668 Behind(Node cond, int rmax, int rmin) { 4669 this.cond = cond; 4670 this.rmax = rmax; 4671 this.rmin = rmin; 4672 } 4673 4674 boolean match(Matcher matcher, int i, CharSequence seq) { 4675 int savedFrom = matcher.from; 4676 boolean conditionMatched = false; 4677 int startIndex = (!matcher.transparentBounds) ? 4678 matcher.from : 0; 4679 int from = Math.max(i - rmax, startIndex); 4680 // Set end boundary 4681 int savedLBT = matcher.lookbehindTo; 4682 matcher.lookbehindTo = i; 4683 // Relax transparent region boundaries for lookbehind 4684 if (matcher.transparentBounds) 4685 matcher.from = 0; 4686 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 4687 conditionMatched = cond.match(matcher, j, seq); 4688 } 4689 matcher.from = savedFrom; 4690 matcher.lookbehindTo = savedLBT; 4691 return conditionMatched && next.match(matcher, i, seq); 4692 } 4693 } 4694 4695 /** 4696 * Zero width positive lookbehind, including supplementary 4697 * characters or unpaired surrogates. 4698 */ 4699 static final class BehindS extends Behind { 4700 BehindS(Node cond, int rmax, int rmin) { 4701 super(cond, rmax, rmin); 4702 } 4703 boolean match(Matcher matcher, int i, CharSequence seq) { 4704 int rmaxChars = countChars(seq, i, -rmax); 4705 int rminChars = countChars(seq, i, -rmin); 4706 int savedFrom = matcher.from; 4707 int startIndex = (!matcher.transparentBounds) ? 4708 matcher.from : 0; 4709 boolean conditionMatched = false; 4710 int from = Math.max(i - rmaxChars, startIndex); 4711 // Set end boundary 4712 int savedLBT = matcher.lookbehindTo; 4713 matcher.lookbehindTo = i; 4714 // Relax transparent region boundaries for lookbehind 4715 if (matcher.transparentBounds) 4716 matcher.from = 0; 4717 4718 for (int j = i - rminChars; 4719 !conditionMatched && j >= from; 4720 j -= j>from ? countChars(seq, j, -1) : 1) { 4721 conditionMatched = cond.match(matcher, j, seq); 4722 } 4723 matcher.from = savedFrom; 4724 matcher.lookbehindTo = savedLBT; 4725 return conditionMatched && next.match(matcher, i, seq); 4726 } 4727 } 4728 4729 /** 4730 * Zero width negative lookbehind. 4731 */ 4732 static class NotBehind extends Node { 4733 Node cond; 4734 int rmax, rmin; 4735 NotBehind(Node cond, int rmax, int rmin) { 4736 this.cond = cond; 4737 this.rmax = rmax; 4738 this.rmin = rmin; 4739 } 4740 4741 boolean match(Matcher matcher, int i, CharSequence seq) { 4742 int savedLBT = matcher.lookbehindTo; 4743 int savedFrom = matcher.from; 4744 boolean conditionMatched = false; 4745 int startIndex = (!matcher.transparentBounds) ? 4746 matcher.from : 0; 4747 int from = Math.max(i - rmax, startIndex); 4748 matcher.lookbehindTo = i; 4749 // Relax transparent region boundaries for lookbehind 4750 if (matcher.transparentBounds) 4751 matcher.from = 0; 4752 for (int j = i - rmin; !conditionMatched && j >= from; j--) { 4753 conditionMatched = cond.match(matcher, j, seq); 4754 } 4755 // Reinstate region boundaries 4756 matcher.from = savedFrom; 4757 matcher.lookbehindTo = savedLBT; 4758 return !conditionMatched && next.match(matcher, i, seq); 4759 } 4760 } 4761 4762 /** 4763 * Zero width negative lookbehind, including supplementary 4764 * characters or unpaired surrogates. 4765 */ 4766 static final class NotBehindS extends NotBehind { 4767 NotBehindS(Node cond, int rmax, int rmin) { 4768 super(cond, rmax, rmin); 4769 } 4770 boolean match(Matcher matcher, int i, CharSequence seq) { 4771 int rmaxChars = countChars(seq, i, -rmax); 4772 int rminChars = countChars(seq, i, -rmin); 4773 int savedFrom = matcher.from; 4774 int savedLBT = matcher.lookbehindTo; 4775 boolean conditionMatched = false; 4776 int startIndex = (!matcher.transparentBounds) ? 4777 matcher.from : 0; 4778 int from = Math.max(i - rmaxChars, startIndex); 4779 matcher.lookbehindTo = i; 4780 // Relax transparent region boundaries for lookbehind 4781 if (matcher.transparentBounds) 4782 matcher.from = 0; 4783 for (int j = i - rminChars; 4784 !conditionMatched && j >= from; 4785 j -= j>from ? countChars(seq, j, -1) : 1) { 4786 conditionMatched = cond.match(matcher, j, seq); 4787 } 4788 //Reinstate region boundaries 4789 matcher.from = savedFrom; 4790 matcher.lookbehindTo = savedLBT; 4791 return !conditionMatched && next.match(matcher, i, seq); 4792 } 4793 } 4794 4795 /** 4796 * Returns the set union of two CharProperty nodes. 4797 */ 4798 private static CharProperty union(final CharProperty lhs, 4799 final CharProperty rhs) { 4800 return new CharProperty() { 4801 boolean isSatisfiedBy(int ch) { 4802 return lhs.isSatisfiedBy(ch) || rhs.isSatisfiedBy(ch);}}; 4803 } 4804 4805 /** 4806 * Returns the set intersection of two CharProperty nodes. 4807 */ 4808 private static CharProperty intersection(final CharProperty lhs, 4809 final CharProperty rhs) { 4810 return new CharProperty() { 4811 boolean isSatisfiedBy(int ch) { 4812 return lhs.isSatisfiedBy(ch) && rhs.isSatisfiedBy(ch);}}; 4813 } 4814 4815 /** 4816 * Returns the set difference of two CharProperty nodes. 4817 */ 4818 private static CharProperty setDifference(final CharProperty lhs, 4819 final CharProperty rhs) { 4820 return new CharProperty() { 4821 boolean isSatisfiedBy(int ch) { 4822 return ! rhs.isSatisfiedBy(ch) && lhs.isSatisfiedBy(ch);}}; 4823 } 4824 4825 /** 4826 * Handles word boundaries. Includes a field to allow this one class to 4827 * deal with the different types of word boundaries we can match. The word 4828 * characters include underscores, letters, and digits. Non spacing marks 4829 * can are also part of a word if they have a base character, otherwise 4830 * they are ignored for purposes of finding word boundaries. 4831 */ 4832 static final class Bound extends Node { 4833 static int LEFT = 0x1; 4834 static int RIGHT= 0x2; 4835 static int BOTH = 0x3; 4836 static int NONE = 0x4; 4837 int type; 4838 Bound(int n) { 4839 type = n; 4840 } 4841 int check(Matcher matcher, int i, CharSequence seq) { 4842 int ch; 4843 boolean left = false; 4844 int startIndex = matcher.from; 4845 int endIndex = matcher.to; 4846 if (matcher.transparentBounds) { 4847 startIndex = 0; 4848 endIndex = matcher.getTextLength(); 4849 } 4850 if (i > startIndex) { 4851 ch = Character.codePointBefore(seq, i); 4852 left = (ch == '_' || Character.isLetterOrDigit(ch) || 4853 ((Character.getType(ch) == Character.NON_SPACING_MARK) 4854 && hasBaseCharacter(matcher, i-1, seq))); 4855 } 4856 boolean right = false; 4857 if (i < endIndex) { 4858 ch = Character.codePointAt(seq, i); 4859 right = (ch == '_' || Character.isLetterOrDigit(ch) || 4860 ((Character.getType(ch) == Character.NON_SPACING_MARK) 4861 && hasBaseCharacter(matcher, i, seq))); 4862 } else { 4863 // Tried to access char past the end 4864 matcher.hitEnd = true; 4865 // The addition of another char could wreck a boundary 4866 matcher.requireEnd = true; 4867 } 4868 return ((left ^ right) ? (right ? LEFT : RIGHT) : NONE); 4869 } 4870 boolean match(Matcher matcher, int i, CharSequence seq) { 4871 return (check(matcher, i, seq) & type) > 0 4872 && next.match(matcher, i, seq); 4873 } 4874 } 4875 4876 /** 4877 * Non spacing marks only count as word characters in bounds calculations 4878 * if they have a base character. 4879 */ 4880 private static boolean hasBaseCharacter(Matcher matcher, int i, 4881 CharSequence seq) 4882 { 4883 int start = (!matcher.transparentBounds) ? 4884 matcher.from : 0; 4885 for (int x=i; x >= start; x--) { 4886 int ch = Character.codePointAt(seq, x); 4887 if (Character.isLetterOrDigit(ch)) 4888 return true; 4889 if (Character.getType(ch) == Character.NON_SPACING_MARK) 4890 continue; 4891 return false; 4892 } 4893 return false; 4894 } 4895 4896 /** 4897 * Attempts to match a slice in the input using the Boyer-Moore string 4898 * matching algorithm. The algorithm is based on the idea that the 4899 * pattern can be shifted farther ahead in the search text if it is 4900 * matched right to left. 4901 * <p> 4902 * The pattern is compared to the input one character at a time, from 4903 * the rightmost character in the pattern to the left. If the characters 4904 * all match the pattern has been found. If a character does not match, 4905 * the pattern is shifted right a distance that is the maximum of two 4906 * functions, the bad character shift and the good suffix shift. This 4907 * shift moves the attempted match position through the input more 4908 * quickly than a naive one position at a time check. 4909 * <p> 4910 * The bad character shift is based on the character from the text that 4911 * did not match. If the character does not appear in the pattern, the 4912 * pattern can be shifted completely beyond the bad character. If the 4913 * character does occur in the pattern, the pattern can be shifted to 4914 * line the pattern up with the next occurrence of that character. 4915 * <p> 4916 * The good suffix shift is based on the idea that some subset on the right 4917 * side of the pattern has matched. When a bad character is found, the 4918 * pattern can be shifted right by the pattern length if the subset does 4919 * not occur again in pattern, or by the amount of distance to the 4920 * next occurrence of the subset in the pattern. 4921 * 4922 * Boyer-Moore search methods adapted from code by Amy Yu. 4923 */ 4924 static class BnM extends Node { 4925 int[] buffer; 4926 int[] lastOcc; 4927 int[] optoSft; 4928 4929 /** 4930 * Pre calculates arrays needed to generate the bad character 4931 * shift and the good suffix shift. Only the last seven bits 4932 * are used to see if chars match; This keeps the tables small 4933 * and covers the heavily used ASCII range, but occasionally 4934 * results in an aliased match for the bad character shift. 4935 */ 4936 static Node optimize(Node node) { 4937 if (!(node instanceof Slice)) { 4938 return node; 4939 } 4940 4941 int[] src = ((Slice) node).buffer; 4942 int patternLength = src.length; 4943 // The BM algorithm requires a bit of overhead; 4944 // If the pattern is short don't use it, since 4945 // a shift larger than the pattern length cannot 4946 // be used anyway. 4947 if (patternLength < 4) { 4948 return node; 4949 } 4950 int i, j, k; 4951 int[] lastOcc = new int[128]; 4952 int[] optoSft = new int[patternLength]; 4953 // Precalculate part of the bad character shift 4954 // It is a table for where in the pattern each 4955 // lower 7-bit value occurs 4956 for (i = 0; i < patternLength; i++) { 4957 lastOcc[src[i]&0x7F] = i + 1; 4958 } 4959 // Precalculate the good suffix shift 4960 // i is the shift amount being considered 4961 NEXT: for (i = patternLength; i > 0; i--) { 4962 // j is the beginning index of suffix being considered 4963 for (j = patternLength - 1; j >= i; j--) { 4964 // Testing for good suffix 4965 if (src[j] == src[j-i]) { 4966 // src[j..len] is a good suffix 4967 optoSft[j-1] = i; 4968 } else { 4969 // No match. The array has already been 4970 // filled up with correct values before. 4971 continue NEXT; 4972 } 4973 } 4974 // This fills up the remaining of optoSft 4975 // any suffix can not have larger shift amount 4976 // then its sub-suffix. Why??? 4977 while (j > 0) { 4978 optoSft[--j] = i; 4979 } 4980 } 4981 // Set the guard value because of unicode compression 4982 optoSft[patternLength-1] = 1; 4983 if (node instanceof SliceS) 4984 return new BnMS(src, lastOcc, optoSft, node.next); 4985 return new BnM(src, lastOcc, optoSft, node.next); 4986 } 4987 BnM(int[] src, int[] lastOcc, int[] optoSft, Node next) { 4988 this.buffer = src; 4989 this.lastOcc = lastOcc; 4990 this.optoSft = optoSft; 4991 this.next = next; 4992 } 4993 boolean match(Matcher matcher, int i, CharSequence seq) { 4994 int[] src = buffer; 4995 int patternLength = src.length; 4996 int last = matcher.to - patternLength; 4997 4998 // Loop over all possible match positions in text 4999 NEXT: while (i <= last) { 5000 // Loop over pattern from right to left 5001 for (int j = patternLength - 1; j >= 0; j--) { 5002 int ch = seq.charAt(i+j); 5003 if (ch != src[j]) { 5004 // Shift search to the right by the maximum of the 5005 // bad character shift and the good suffix shift 5006 i += Math.max(j + 1 - lastOcc[ch&0x7F], optoSft[j]); 5007 continue NEXT; 5008 } 5009 } 5010 // Entire pattern matched starting at i 5011 matcher.first = i; 5012 boolean ret = next.match(matcher, i + patternLength, seq); 5013 if (ret) { 5014 matcher.first = i; 5015 matcher.groups[0] = matcher.first; 5016 matcher.groups[1] = matcher.last; 5017 return true; 5018 } 5019 i++; 5020 } 5021 // BnM is only used as the leading node in the unanchored case, 5022 // and it replaced its Start() which always searches to the end 5023 // if it doesn't find what it's looking for, so hitEnd is true. 5024 matcher.hitEnd = true; 5025 return false; 5026 } 5027 boolean study(TreeInfo info) { 5028 info.minLength += buffer.length; 5029 info.maxValid = false; 5030 return next.study(info); 5031 } 5032 } 5033 5034 /** 5035 * Supplementary support version of BnM(). Unpaired surrogates are 5036 * also handled by this class. 5037 */ 5038 static final class BnMS extends BnM { 5039 int lengthInChars; 5040 5041 BnMS(int[] src, int[] lastOcc, int[] optoSft, Node next) { 5042 super(src, lastOcc, optoSft, next); 5043 for (int x = 0; x < buffer.length; x++) { 5044 lengthInChars += Character.charCount(buffer[x]); 5045 } 5046 } 5047 boolean match(Matcher matcher, int i, CharSequence seq) { 5048 int[] src = buffer; 5049 int patternLength = src.length; 5050 int last = matcher.to - lengthInChars; 5051 5052 // Loop over all possible match positions in text 5053 NEXT: while (i <= last) { 5054 // Loop over pattern from right to left 5055 int ch; 5056 for (int j = countChars(seq, i, patternLength), x = patternLength - 1; 5057 j > 0; j -= Character.charCount(ch), x--) { 5058 ch = Character.codePointBefore(seq, i+j); 5059 if (ch != src[x]) { 5060 // Shift search to the right by the maximum of the 5061 // bad character shift and the good suffix shift 5062 int n = Math.max(x + 1 - lastOcc[ch&0x7F], optoSft[x]); 5063 i += countChars(seq, i, n); 5064 continue NEXT; 5065 } 5066 } 5067 // Entire pattern matched starting at i 5068 matcher.first = i; 5069 boolean ret = next.match(matcher, i + lengthInChars, seq); 5070 if (ret) { 5071 matcher.first = i; 5072 matcher.groups[0] = matcher.first; 5073 matcher.groups[1] = matcher.last; 5074 return true; 5075 } 5076 i += countChars(seq, i, 1); 5077 } 5078 matcher.hitEnd = true; 5079 return false; 5080 } 5081 } 5082 5083 /////////////////////////////////////////////////////////////////////////////// 5084 /////////////////////////////////////////////////////////////////////////////// 5085 5086 /** 5087 * This must be the very first initializer. 5088 */ 5089 static Node accept = new Node(); 5090 5091 static Node lastAccept = new LastNode(); 5092 5093 private static class CharPropertyNames { 5094 5095 static CharProperty charPropertyFor(String name) { 5096 CharPropertyFactory m = map.get(name); 5097 return m == null ? null : m.make(); 5098 } 5099 5100 private static abstract class CharPropertyFactory { 5101 abstract CharProperty make(); 5102 } 5103 5104 private static void defCategory(String name, 5105 final int typeMask) { 5106 map.put(name, new CharPropertyFactory() { 5107 CharProperty make() { return new Category(typeMask);}}); 5108 } 5109 5110 private static void defRange(String name, 5111 final int lower, final int upper) { 5112 map.put(name, new CharPropertyFactory() { 5113 CharProperty make() { return rangeFor(lower, upper);}}); 5114 } 5115 5116 private static void defCtype(String name, 5117 final int ctype) { 5118 map.put(name, new CharPropertyFactory() { 5119 CharProperty make() { return new Ctype(ctype);}}); 5120 } 5121 5122 private static abstract class CloneableProperty 5123 extends CharProperty implements Cloneable 5124 { 5125 public CloneableProperty clone() { 5126 try { 5127 return (CloneableProperty) super.clone(); 5128 } catch (CloneNotSupportedException e) { 5129 throw new AssertionError(e); 5130 } 5131 } 5132 } 5133 5134 private static void defClone(String name, 5135 final CloneableProperty p) { 5136 map.put(name, new CharPropertyFactory() { 5137 CharProperty make() { return p.clone();}}); 5138 } 5139 5140 private static final HashMap<String, CharPropertyFactory> map 5141 = new HashMap<String, CharPropertyFactory>(); 5142 5143 static { 5144 // Unicode character property aliases, defined in 5145 // http://www.unicode.org/Public/UNIDATA/PropertyValueAliases.txt 5146 defCategory("Cn", 1<<Character.UNASSIGNED); 5147 defCategory("Lu", 1<<Character.UPPERCASE_LETTER); 5148 defCategory("Ll", 1<<Character.LOWERCASE_LETTER); 5149 defCategory("Lt", 1<<Character.TITLECASE_LETTER); 5150 defCategory("Lm", 1<<Character.MODIFIER_LETTER); 5151 defCategory("Lo", 1<<Character.OTHER_LETTER); 5152 defCategory("Mn", 1<<Character.NON_SPACING_MARK); 5153 defCategory("Me", 1<<Character.ENCLOSING_MARK); 5154 defCategory("Mc", 1<<Character.COMBINING_SPACING_MARK); 5155 defCategory("Nd", 1<<Character.DECIMAL_DIGIT_NUMBER); 5156 defCategory("Nl", 1<<Character.LETTER_NUMBER); 5157 defCategory("No", 1<<Character.OTHER_NUMBER); 5158 defCategory("Zs", 1<<Character.SPACE_SEPARATOR); 5159 defCategory("Zl", 1<<Character.LINE_SEPARATOR); 5160 defCategory("Zp", 1<<Character.PARAGRAPH_SEPARATOR); 5161 defCategory("Cc", 1<<Character.CONTROL); 5162 defCategory("Cf", 1<<Character.FORMAT); 5163 defCategory("Co", 1<<Character.PRIVATE_USE); 5164 defCategory("Cs", 1<<Character.SURROGATE); 5165 defCategory("Pd", 1<<Character.DASH_PUNCTUATION); 5166 defCategory("Ps", 1<<Character.START_PUNCTUATION); 5167 defCategory("Pe", 1<<Character.END_PUNCTUATION); 5168 defCategory("Pc", 1<<Character.CONNECTOR_PUNCTUATION); 5169 defCategory("Po", 1<<Character.OTHER_PUNCTUATION); 5170 defCategory("Sm", 1<<Character.MATH_SYMBOL); 5171 defCategory("Sc", 1<<Character.CURRENCY_SYMBOL); 5172 defCategory("Sk", 1<<Character.MODIFIER_SYMBOL); 5173 defCategory("So", 1<<Character.OTHER_SYMBOL); 5174 defCategory("Pi", 1<<Character.INITIAL_QUOTE_PUNCTUATION); 5175 defCategory("Pf", 1<<Character.FINAL_QUOTE_PUNCTUATION); 5176 defCategory("L", ((1<<Character.UPPERCASE_LETTER) | 5177 (1<<Character.LOWERCASE_LETTER) | 5178 (1<<Character.TITLECASE_LETTER) | 5179 (1<<Character.MODIFIER_LETTER) | 5180 (1<<Character.OTHER_LETTER))); 5181 defCategory("M", ((1<<Character.NON_SPACING_MARK) | 5182 (1<<Character.ENCLOSING_MARK) | 5183 (1<<Character.COMBINING_SPACING_MARK))); 5184 defCategory("N", ((1<<Character.DECIMAL_DIGIT_NUMBER) | 5185 (1<<Character.LETTER_NUMBER) | 5186 (1<<Character.OTHER_NUMBER))); 5187 defCategory("Z", ((1<<Character.SPACE_SEPARATOR) | 5188 (1<<Character.LINE_SEPARATOR) | 5189 (1<<Character.PARAGRAPH_SEPARATOR))); 5190 defCategory("C", ((1<<Character.CONTROL) | 5191 (1<<Character.FORMAT) | 5192 (1<<Character.PRIVATE_USE) | 5193 (1<<Character.SURROGATE))); // Other 5194 defCategory("P", ((1<<Character.DASH_PUNCTUATION) | 5195 (1<<Character.START_PUNCTUATION) | 5196 (1<<Character.END_PUNCTUATION) | 5197 (1<<Character.CONNECTOR_PUNCTUATION) | 5198 (1<<Character.OTHER_PUNCTUATION) | 5199 (1<<Character.INITIAL_QUOTE_PUNCTUATION) | 5200 (1<<Character.FINAL_QUOTE_PUNCTUATION))); 5201 defCategory("S", ((1<<Character.MATH_SYMBOL) | 5202 (1<<Character.CURRENCY_SYMBOL) | 5203 (1<<Character.MODIFIER_SYMBOL) | 5204 (1<<Character.OTHER_SYMBOL))); 5205 defCategory("LC", ((1<<Character.UPPERCASE_LETTER) | 5206 (1<<Character.LOWERCASE_LETTER) | 5207 (1<<Character.TITLECASE_LETTER))); 5208 defCategory("LD", ((1<<Character.UPPERCASE_LETTER) | 5209 (1<<Character.LOWERCASE_LETTER) | 5210 (1<<Character.TITLECASE_LETTER) | 5211 (1<<Character.MODIFIER_LETTER) | 5212 (1<<Character.OTHER_LETTER) | 5213 (1<<Character.DECIMAL_DIGIT_NUMBER))); 5214 defRange("L1", 0x00, 0xFF); // Latin-1 5215 map.put("all", new CharPropertyFactory() { 5216 CharProperty make() { return new All(); }}); 5217 5218 // Posix regular expression character classes, defined in 5219 // http://www.unix.org/onlinepubs/009695399/basedefs/xbd_chap09.html 5220 defRange("ASCII", 0x00, 0x7F); // ASCII 5221 defCtype("Alnum", ASCII.ALNUM); // Alphanumeric characters 5222 defCtype("Alpha", ASCII.ALPHA); // Alphabetic characters 5223 defCtype("Blank", ASCII.BLANK); // Space and tab characters 5224 defCtype("Cntrl", ASCII.CNTRL); // Control characters 5225 defRange("Digit", '0', '9'); // Numeric characters 5226 defCtype("Graph", ASCII.GRAPH); // printable and visible 5227 defRange("Lower", 'a', 'z'); // Lower-case alphabetic 5228 defRange("Print", 0x20, 0x7E); // Printable characters 5229 defCtype("Punct", ASCII.PUNCT); // Punctuation characters 5230 defCtype("Space", ASCII.SPACE); // Space characters 5231 defRange("Upper", 'A', 'Z'); // Upper-case alphabetic 5232 defCtype("XDigit",ASCII.XDIGIT); // hexadecimal digits 5233 5234 // Java character properties, defined by methods in Character.java 5235 defClone("javaLowerCase", new CloneableProperty() { 5236 boolean isSatisfiedBy(int ch) { 5237 return Character.isLowerCase(ch);}}); 5238 defClone("javaUpperCase", new CloneableProperty() { 5239 boolean isSatisfiedBy(int ch) { 5240 return Character.isUpperCase(ch);}}); 5241 defClone("javaTitleCase", new CloneableProperty() { 5242 boolean isSatisfiedBy(int ch) { 5243 return Character.isTitleCase(ch);}}); 5244 defClone("javaDigit", new CloneableProperty() { 5245 boolean isSatisfiedBy(int ch) { 5246 return Character.isDigit(ch);}}); 5247 defClone("javaDefined", new CloneableProperty() { 5248 boolean isSatisfiedBy(int ch) { 5249 return Character.isDefined(ch);}}); 5250 defClone("javaLetter", new CloneableProperty() { 5251 boolean isSatisfiedBy(int ch) { 5252 return Character.isLetter(ch);}}); 5253 defClone("javaLetterOrDigit", new CloneableProperty() { 5254 boolean isSatisfiedBy(int ch) { 5255 return Character.isLetterOrDigit(ch);}}); 5256 defClone("javaJavaIdentifierStart", new CloneableProperty() { 5257 boolean isSatisfiedBy(int ch) { 5258 return Character.isJavaIdentifierStart(ch);}}); 5259 defClone("javaJavaIdentifierPart", new CloneableProperty() { 5260 boolean isSatisfiedBy(int ch) { 5261 return Character.isJavaIdentifierPart(ch);}}); 5262 defClone("javaUnicodeIdentifierStart", new CloneableProperty() { 5263 boolean isSatisfiedBy(int ch) { 5264 return Character.isUnicodeIdentifierStart(ch);}}); 5265 defClone("javaUnicodeIdentifierPart", new CloneableProperty() { 5266 boolean isSatisfiedBy(int ch) { 5267 return Character.isUnicodeIdentifierPart(ch);}}); 5268 defClone("javaIdentifierIgnorable", new CloneableProperty() { 5269 boolean isSatisfiedBy(int ch) { 5270 return Character.isIdentifierIgnorable(ch);}}); 5271 defClone("javaSpaceChar", new CloneableProperty() { 5272 boolean isSatisfiedBy(int ch) { 5273 return Character.isSpaceChar(ch);}}); 5274 defClone("javaWhitespace", new CloneableProperty() { 5275 boolean isSatisfiedBy(int ch) { 5276 return Character.isWhitespace(ch);}}); 5277 defClone("javaISOControl", new CloneableProperty() { 5278 boolean isSatisfiedBy(int ch) { 5279 return Character.isISOControl(ch);}}); 5280 defClone("javaMirrored", new CloneableProperty() { 5281 boolean isSatisfiedBy(int ch) { 5282 return Character.isMirrored(ch);}}); 5283 } 5284 } 5285 }