1 /* 2 * Copyright (C) 2007 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17 package java.util.regex; 18 19 import java.io.IOException; 20 import java.io.ObjectInputStream; 21 import java.io.Serializable; 22 23 /** 24 * Patterns are compiled regular expressions. In many cases, convenience methods such as 25 * {@link String#matches String.matches}, {@link String#replaceAll String.replaceAll} and 26 * {@link String#split String.split} will be preferable, but if you need to do a lot of work 27 * with the same regular expression, it may be more efficient to compile it once and reuse it. 28 * The {@code Pattern} class and its companion, {@link Matcher}, also offer more functionality 29 * than the small amount exposed by {@code String}. 30 * 31 * <pre> 32 * // String convenience methods: 33 * boolean sawFailures = s.matches("Failures: \\d+"); 34 * String farewell = s.replaceAll("Hello, (\\S+)", "Goodbye, $1"); 35 * String[] fields = s.split(":"); 36 * 37 * // Direct use of Pattern: 38 * Pattern p = Pattern.compile("Hello, (\\S+)"); 39 * Matcher m = p.matcher(inputString); 40 * while (m.find()) { // Find each match in turn; String can't do this. 41 * String name = m.group(1); // Access a submatch group; String can't do this. 42 * } 43 * </pre> 44 * 45 * <h3>Regular expression syntax</h3> 46 * <span class="datatable"> 47 * <style type="text/css"> 48 * .datatable td { padding-right: 20px; } 49 * </style> 50 * 51 * <p>Java supports a subset of Perl 5 regular expression syntax. An important gotcha is that Java 52 * has no regular expression literals, and uses plain old string literals instead. This means that 53 * you need an extra level of escaping. For example, the regular expression {@code \s+} has to 54 * be represented as the string {@code "\\s+"}. 55 * 56 * <h3>Escape sequences</h3> 57 * <p><table> 58 * <tr> <td> \ </td> <td>Quote the following metacharacter (so {@code \.} matches a literal {@code .}).</td> </tr> 59 * <tr> <td> \Q </td> <td>Quote all following metacharacters until {@code \E}.</td> </tr> 60 * <tr> <td> \E </td> <td>Stop quoting metacharacters (started by {@code \Q}).</td> </tr> 61 * <tr> <td> \\ </td> <td>A literal backslash.</td> </tr> 62 * <tr> <td> \u<i>hhhh</i> </td> <td>The Unicode character U+hhhh (in hex).</td> </tr> 63 * <tr> <td> \x<i>hh</i> </td> <td>The Unicode character U+00hh (in hex).</td> </tr> 64 * <tr> <td> \c<i>x</i> </td> <td>The ASCII control character ^x (so {@code \cH} would be ^H, U+0008).</td> </tr> 65 * 66 * <tr> <td> \a </td> <td>The ASCII bell character (U+0007).</td> </tr> 67 * <tr> <td> \e </td> <td>The ASCII ESC character (U+001b).</td> </tr> 68 * <tr> <td> \f </td> <td>The ASCII form feed character (U+000c).</td> </tr> 69 * <tr> <td> \n </td> <td>The ASCII newline character (U+000a).</td> </tr> 70 * <tr> <td> \r </td> <td>The ASCII carriage return character (U+000d).</td> </tr> 71 * <tr> <td> \t </td> <td>The ASCII tab character (U+0009).</td> </tr> 72 * </table> 73 * 74 * <h3>Character classes</h3> 75 * <p>It's possible to construct arbitrary character classes using set operations: 76 * <table> 77 * <tr> <td> [abc] </td> <td>Any one of {@code a}, {@code b}, or {@code c}. (Enumeration.)</td> </tr> 78 * <tr> <td> [a-c] </td> <td>Any one of {@code a}, {@code b}, or {@code c}. (Range.)</td> </tr> 79 * <tr> <td> [^abc] </td> <td>Any character <i>except</i> {@code a}, {@code b}, or {@code c}. (Negation.)</td> </tr> 80 * <tr> <td> [[a-f][0-9]] </td> <td>Any character in either range. (Union.)</td> </tr> 81 * <tr> <td> [[a-z]&&[jkl]] </td> <td>Any character in both ranges. (Intersection.)</td> </tr> 82 * </table> 83 * <p>Most of the time, the built-in character classes are more useful: 84 * <table> 85 * <tr> <td> \d </td> <td>Any digit character.</td> </tr> 86 * <tr> <td> \D </td> <td>Any non-digit character.</td> </tr> 87 * <tr> <td> \s </td> <td>Any whitespace character.</td> </tr> 88 * <tr> <td> \S </td> <td>Any non-whitespace character.</td> </tr> 89 * <tr> <td> \w </td> <td>Any word character.</td> </tr> 90 * <tr> <td> \W </td> <td>Any non-word character.</td> </tr> 91 * <tr> <td> \p{<i>NAME</i>} </td> <td> Any character in the class with the given <i>NAME</i>. </td> </tr> 92 * <tr> <td> \P{<i>NAME</i>} </td> <td> Any character <i>not</i> in the named class. </td> </tr> 93 * </table> 94 * <p>There are a variety of named classes: 95 * <ul> 96 * <li><a href="../../lang/Character.html#unicode_categories">Unicode category names</a>, 97 * prefixed by {@code Is}. For example {@code \p{IsLu}} for all uppercase letters. 98 * <li>POSIX class names. These are 'Alnum', 'Alpha', 'ASCII', 'Blank', 'Cntrl', 'Digit', 99 * 'Graph', 'Lower', 'Print', 'Punct', 'Upper', 'XDigit'. 100 * <li>Unicode block names, as used by {@link java.lang.Character.UnicodeBlock#forName} prefixed 101 * by {@code In}. For example {@code \p{InHebrew}} for all characters in the Hebrew block. 102 * <li>Character method names. These are all non-deprecated methods from {@link java.lang.Character} 103 * whose name starts with {@code is}, but with the {@code is} replaced by {@code java}. 104 * For example, {@code \p{javaLowerCase}}. 105 * </ul> 106 * 107 * <h3>Quantifiers</h3> 108 * <p>Quantifiers match some number of instances of the preceding regular expression. 109 * <table> 110 * <tr> <td> * </td> <td>Zero or more.</td> </tr> 111 * <tr> <td> ? </td> <td>Zero or one.</td> </tr> 112 * <tr> <td> + </td> <td>One or more.</td> </tr> 113 * <tr> <td> {<i>n</i>} </td> <td>Exactly <i>n</i>.</td> </tr> 114 * <tr> <td> {<i>n,</i>} </td> <td>At least <i>n</i>.</td> </tr> 115 * <tr> <td> {<i>n</i>,<i>m</i>} </td> <td>At least <i>n</i> but not more than <i>m</i>.</td> </tr> 116 * </table> 117 * <p>Quantifiers are "greedy" by default, meaning that they will match the longest possible input 118 * sequence. There are also non-greedy quantifiers that match the shortest possible input sequence. 119 * They're same as the greedy ones but with a trailing {@code ?}: 120 * <table> 121 * <tr> <td> *? </td> <td>Zero or more (non-greedy).</td> </tr> 122 * <tr> <td> ?? </td> <td>Zero or one (non-greedy).</td> </tr> 123 * <tr> <td> +? </td> <td>One or more (non-greedy).</td> </tr> 124 * <tr> <td> {<i>n</i>}? </td> <td>Exactly <i>n</i> (non-greedy).</td> </tr> 125 * <tr> <td> {<i>n,</i>}? </td> <td>At least <i>n</i> (non-greedy).</td> </tr> 126 * <tr> <td> {<i>n</i>,<i>m</i>}? </td> <td>At least <i>n</i> but not more than <i>m</i> (non-greedy).</td> </tr> 127 * </table> 128 * <p>Quantifiers allow backtracking by default. There are also possessive quantifiers to prevent 129 * backtracking. They're same as the greedy ones but with a trailing {@code +}: 130 * <table> 131 * <tr> <td> *+ </td> <td>Zero or more (possessive).</td> </tr> 132 * <tr> <td> ?+ </td> <td>Zero or one (possessive).</td> </tr> 133 * <tr> <td> ++ </td> <td>One or more (possessive).</td> </tr> 134 * <tr> <td> {<i>n</i>}+ </td> <td>Exactly <i>n</i> (possessive).</td> </tr> 135 * <tr> <td> {<i>n,</i>}+ </td> <td>At least <i>n</i> (possessive).</td> </tr> 136 * <tr> <td> {<i>n</i>,<i>m</i>}+ </td> <td>At least <i>n</i> but not more than <i>m</i> (possessive).</td> </tr> 137 * </table> 138 * 139 * <h3>Zero-width assertions</h3> 140 * <p><table> 141 * <tr> <td> ^ </td> <td>At beginning of line.</td> </tr> 142 * <tr> <td> $ </td> <td>At end of line.</td> </tr> 143 * <tr> <td> \A </td> <td>At beginning of input.</td> </tr> 144 * <tr> <td> \b </td> <td>At word boundary.</td> </tr> 145 * <tr> <td> \B </td> <td>At non-word boundary.</td> </tr> 146 * <tr> <td> \G </td> <td>At end of previous match.</td> </tr> 147 * <tr> <td> \z </td> <td>At end of input.</td> </tr> 148 * <tr> <td> \Z </td> <td>At end of input, or before newline at end.</td> </tr> 149 * </table> 150 * 151 * <h3>Look-around assertions</h3> 152 * <p>Look-around assertions assert that the subpattern does (positive) or doesn't (negative) match 153 * after (look-ahead) or before (look-behind) the current position, without including the matched 154 * text in the containing match. The maximum length of possible matches for look-behind patterns 155 * must not be unbounded. 156 * <p><table> 157 * <tr> <td> (?=<i>a</i>) </td> <td>Zero-width positive look-ahead.</td> </tr> 158 * <tr> <td> (?!<i>a</i>) </td> <td>Zero-width negative look-ahead.</td> </tr> 159 * <tr> <td> (?<=<i>a</i>) </td> <td>Zero-width positive look-behind.</td> </tr> 160 * <tr> <td> (?<!<i>a</i>) </td> <td>Zero-width negative look-behind.</td> </tr> 161 * </table> 162 * 163 * <h3>Groups</h3> 164 * 165 * <p><table> 166 * <tr> <td> (<i>a</i>) </td> <td>A capturing group.</td> </tr> 167 * <tr> <td> (?:<i>a</i>) </td> <td>A non-capturing group.</td> </tr> 168 * <tr> <td> (?><i>a</i>) </td> <td>An independent non-capturing group. (The first match of the subgroup is the only match tried.)</td> </tr> 169 * <tr> <td> \<i>n</i> </td> <td>The text already matched by capturing group <i>n</i>.</td> </tr> 170 * </table> 171 * <p>See {@link Matcher#group} for details of how capturing groups are numbered and accessed. 172 * 173 * <h3>Operators</h3> 174 * <p><table> 175 * <tr> <td> <i>ab</i> </td> <td>Expression <i>a</i> followed by expression <i>b</i>.</td> </tr> 176 * <tr> <td> <i>a</i>|<i>b</i> </td> <td>Either expression <i>a</i> or expression <i>b</i>.</td> </tr> 177 * </table> 178 * 179 * <a name="flags"><h3>Flags</h3></a> 180 * <p><table> 181 * <tr> <td> (?dimsux-dimsux:<i>a</i>) </td> <td>Evaluates the expression <i>a</i> with the given flags enabled/disabled.</td> </tr> 182 * <tr> <td> (?dimsux-dimsux) </td> <td>Evaluates the rest of the pattern with the given flags enabled/disabled.</td> </tr> 183 * </table> 184 * 185 * <p>The flags are: 186 * <table> 187 * <tr><td>{@code i}</td> <td>{@link #CASE_INSENSITIVE}</td> <td>case insensitive matching</td></tr> 188 * <tr><td>{@code d}</td> <td>{@link #UNIX_LINES}</td> <td>only accept {@code '\n'} as a line terminator</td></tr> 189 * <tr><td>{@code m}</td> <td>{@link #MULTILINE}</td> <td>allow {@code ^} and {@code $} to match beginning/end of any line</td></tr> 190 * <tr><td>{@code s}</td> <td>{@link #DOTALL}</td> <td>allow {@code .} to match {@code '\n'} ("s" for "single line")</td></tr> 191 * <tr><td>{@code u}</td> <td>{@link #UNICODE_CASE}</td> <td>enable Unicode case folding</td></tr> 192 * <tr><td>{@code x}</td> <td>{@link #COMMENTS}</td> <td>allow whitespace and comments</td></tr> 193 * </table> 194 * <p>Either set of flags may be empty. For example, {@code (?i-m)} would turn on case-insensitivity 195 * and turn off multiline mode, {@code (?i)} would just turn on case-insensitivity, 196 * and {@code (?-m)} would just turn off multiline mode. 197 * <p>Note that on Android, {@code UNICODE_CASE} is always on: case-insensitive matching will 198 * always be Unicode-aware. 199 * <p>There are two other flags not settable via this mechanism: {@link #CANON_EQ} and 200 * {@link #LITERAL}. Attempts to use {@link #CANON_EQ} on Android will throw an exception. 201 * </span> 202 * 203 * <h3>Implementation notes</h3> 204 * 205 * <p>The regular expression implementation used in Android is provided by 206 * <a href="http://www.icu-project.org">ICU</a>. The notation for the regular 207 * expressions is mostly a superset of those used in other Java language 208 * implementations. This means that existing applications will normally work as 209 * expected, but in rare cases Android may accept a regular expression that is 210 * not accepted by other implementations. 211 * 212 * <p>In some cases, Android will recognize that a regular expression is a simple 213 * special case that can be handled more efficiently. This is true of both the convenience methods 214 * in {@code String} and the methods in {@code Pattern}. 215 * 216 * @see Matcher 217 */ 218 public final class Pattern implements Serializable { 219 220 private static final long serialVersionUID = 5073258162644648461L; 221 222 /** 223 * This constant specifies that a pattern matches Unix line endings ('\n') 224 * only against the '.', '^', and '$' meta characters. Corresponds to {@code (?d)}. 225 */ 226 public static final int UNIX_LINES = 0x01; 227 228 /** 229 * This constant specifies that a {@code Pattern} is matched 230 * case-insensitively. That is, the patterns "a+" and "A+" would both match 231 * the string "aAaAaA". See {@link #UNICODE_CASE}. Corresponds to {@code (?i)}. 232 */ 233 public static final int CASE_INSENSITIVE = 0x02; 234 235 /** 236 * This constant specifies that a {@code Pattern} may contain whitespace or 237 * comments. Otherwise comments and whitespace are taken as literal 238 * characters. Corresponds to {@code (?x)}. 239 */ 240 public static final int COMMENTS = 0x04; 241 242 /** 243 * This constant specifies that the meta characters '^' and '$' match only 244 * the beginning and end of an input line, respectively. Normally, they 245 * match the beginning and the end of the complete input. Corresponds to {@code (?m)}. 246 */ 247 public static final int MULTILINE = 0x08; 248 249 /** 250 * This constant specifies that the whole {@code Pattern} is to be taken 251 * literally, that is, all meta characters lose their meanings. 252 */ 253 public static final int LITERAL = 0x10; 254 255 /** 256 * This constant specifies that the '.' meta character matches arbitrary 257 * characters, including line endings, which is normally not the case. 258 * Corresponds to {@code (?s)}. 259 */ 260 public static final int DOTALL = 0x20; 261 262 /** 263 * This constant specifies that a {@code Pattern} that uses case-insensitive matching 264 * will use Unicode case folding. On Android, {@code UNICODE_CASE} is always on: 265 * case-insensitive matching will always be Unicode-aware. If your code is intended to 266 * be portable and uses case-insensitive matching on non-ASCII characters, you should 267 * use this flag. Corresponds to {@code (?u)}. 268 */ 269 public static final int UNICODE_CASE = 0x40; 270 271 /** 272 * This constant specifies that a character in a {@code Pattern} and a 273 * character in the input string only match if they are canonically 274 * equivalent. It is (currently) not supported in Android. 275 */ 276 public static final int CANON_EQ = 0x80; 277 278 private final String pattern; 279 private final int flags; 280 281 transient int address; 282 283 /** 284 * Returns a {@link Matcher} for this pattern applied to the given {@code input}. 285 * The {@code Matcher} can be used to match the {@code Pattern} against the 286 * whole input, find occurrences of the {@code Pattern} in the input, or 287 * replace parts of the input. 288 */ 289 public Matcher matcher(CharSequence input) { 290 return new Matcher(this, input); 291 } 292 293 /** 294 * Splits the given {@code input} at occurrences of this pattern. 295 * 296 * <p>If this pattern does not occur in the input, the result is an 297 * array containing the input (converted from a {@code CharSequence} to 298 * a {@code String}). 299 * 300 * <p>Otherwise, the {@code limit} parameter controls the contents of the 301 * returned array as described below. 302 * 303 * @param limit 304 * Determines the maximum number of entries in the resulting 305 * array, and the treatment of trailing empty strings. 306 * <ul> 307 * <li>For n > 0, the resulting array contains at most n 308 * entries. If this is fewer than the number of matches, the 309 * final entry will contain all remaining input. 310 * <li>For n < 0, the length of the resulting array is 311 * exactly the number of occurrences of the {@code Pattern} 312 * plus one for the text after the final separator. 313 * All entries are included. 314 * <li>For n == 0, the result is as for n < 0, except 315 * trailing empty strings will not be returned. (Note that 316 * the case where the input is itself an empty string is 317 * special, as described above, and the limit parameter does 318 * not apply there.) 319 * </ul> 320 */ 321 public String[] split(CharSequence input, int limit) { 322 return Splitter.split(this, pattern, input.toString(), limit); 323 } 324 325 /** 326 * Equivalent to {@code split(input, 0)}. 327 */ 328 public String[] split(CharSequence input) { 329 return split(input, 0); 330 } 331 332 /** 333 * Returns the regular expression supplied to {@code compile}. 334 */ 335 public String pattern() { 336 return pattern; 337 } 338 339 @Override 340 public String toString() { 341 return pattern; 342 } 343 344 /** 345 * Returns the flags supplied to {@code compile}. 346 */ 347 public int flags() { 348 return flags; 349 } 350 351 /** 352 * Returns a compiled form of the given {@code regularExpression}, as modified by the 353 * given {@code flags}. See the <a href="#flags">flags overview</a> for more on flags. 354 * 355 * @throws PatternSyntaxException if the regular expression is syntactically incorrect. 356 * 357 * @see #CANON_EQ 358 * @see #CASE_INSENSITIVE 359 * @see #COMMENTS 360 * @see #DOTALL 361 * @see #LITERAL 362 * @see #MULTILINE 363 * @see #UNICODE_CASE 364 * @see #UNIX_LINES 365 */ 366 public static Pattern compile(String regularExpression, int flags) throws PatternSyntaxException { 367 return new Pattern(regularExpression, flags); 368 } 369 370 /** 371 * Equivalent to {@code Pattern.compile(pattern, 0)}. 372 */ 373 public static Pattern compile(String pattern) { 374 return new Pattern(pattern, 0); 375 } 376 377 private Pattern(String pattern, int flags) throws PatternSyntaxException { 378 if ((flags & CANON_EQ) != 0) { 379 throw new UnsupportedOperationException("CANON_EQ flag not supported"); 380 } 381 this.pattern = pattern; 382 this.flags = flags; 383 compile(); 384 } 385 386 private void compile() throws PatternSyntaxException { 387 if (pattern == null) { 388 throw new NullPointerException("pattern == null"); 389 } 390 391 String icuPattern = pattern; 392 if ((flags & LITERAL) != 0) { 393 icuPattern = quote(pattern); 394 } 395 396 // These are the flags natively supported by ICU. 397 // They even have the same value in native code. 398 int icuFlags = flags & (CASE_INSENSITIVE | COMMENTS | MULTILINE | DOTALL | UNIX_LINES); 399 400 address = compileImpl(icuPattern, icuFlags); 401 } 402 403 /** 404 * Tests whether the given {@code regularExpression} matches the given {@code input}. 405 * Equivalent to {@code Pattern.compile(regularExpression).matcher(input).matches()}. 406 * If the same regular expression is to be used for multiple operations, it may be more 407 * efficient to reuse a compiled {@code Pattern}. 408 * 409 * @see Pattern#compile(java.lang.String, int) 410 * @see Matcher#matches() 411 */ 412 public static boolean matches(String regularExpression, CharSequence input) { 413 return new Matcher(new Pattern(regularExpression, 0), input).matches(); 414 } 415 416 /** 417 * Quotes the given {@code string} using "\Q" and "\E", so that all 418 * meta-characters lose their special meaning. This method correctly 419 * escapes embedded instances of "\Q" or "\E". If the entire result 420 * is to be passed verbatim to {@link #compile}, it's usually clearer 421 * to use the {@link #LITERAL} flag instead. 422 */ 423 public static String quote(String string) { 424 StringBuilder sb = new StringBuilder(); 425 sb.append("\\Q"); 426 int apos = 0; 427 int k; 428 while ((k = string.indexOf("\\E", apos)) >= 0) { 429 sb.append(string.substring(apos, k + 2)).append("\\\\E\\Q"); 430 apos = k + 2; 431 } 432 return sb.append(string.substring(apos)).append("\\E").toString(); 433 } 434 435 @Override protected void finalize() throws Throwable { 436 try { 437 closeImpl(address); 438 } finally { 439 super.finalize(); 440 } 441 } 442 443 private void readObject(ObjectInputStream s) throws IOException, ClassNotFoundException { 444 s.defaultReadObject(); 445 compile(); 446 } 447 448 private static native void closeImpl(int addr); 449 private static native int compileImpl(String regex, int flags); 450 } 451
