1 // 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html#License 3 /* 4 ******************************************************************************* 5 * Copyright (C) 1996-2016, International Business Machines Corporation and 6 * others. All Rights Reserved. 7 ******************************************************************************* 8 */ 9 package com.ibm.icu.text; 10 11 import java.io.IOException; 12 import java.text.ParsePosition; 13 import java.util.ArrayList; 14 import java.util.Collection; 15 import java.util.Collections; 16 import java.util.Iterator; 17 import java.util.NoSuchElementException; 18 import java.util.TreeSet; 19 20 import com.ibm.icu.impl.BMPSet; 21 import com.ibm.icu.impl.Norm2AllModes; 22 import com.ibm.icu.impl.PatternProps; 23 import com.ibm.icu.impl.RuleCharacterIterator; 24 import com.ibm.icu.impl.SortedSetRelation; 25 import com.ibm.icu.impl.StringRange; 26 import com.ibm.icu.impl.UBiDiProps; 27 import com.ibm.icu.impl.UCaseProps; 28 import com.ibm.icu.impl.UCharacterProperty; 29 import com.ibm.icu.impl.UPropertyAliases; 30 import com.ibm.icu.impl.UnicodeSetStringSpan; 31 import com.ibm.icu.impl.Utility; 32 import com.ibm.icu.lang.CharSequences; 33 import com.ibm.icu.lang.UCharacter; 34 import com.ibm.icu.lang.UProperty; 35 import com.ibm.icu.lang.UScript; 36 import com.ibm.icu.util.Freezable; 37 import com.ibm.icu.util.ICUUncheckedIOException; 38 import com.ibm.icu.util.OutputInt; 39 import com.ibm.icu.util.ULocale; 40 import com.ibm.icu.util.VersionInfo; 41 42 /** 43 * A mutable set of Unicode characters and multicharacter strings. 44 * Objects of this class represent <em>character classes</em> used 45 * in regular expressions. A character specifies a subset of Unicode 46 * code points. Legal code points are U+0000 to U+10FFFF, inclusive. 47 * 48 * Note: method freeze() will not only make the set immutable, but 49 * also makes important methods much higher performance: 50 * contains(c), containsNone(...), span(...), spanBack(...) etc. 51 * After the object is frozen, any subsequent call that wants to change 52 * the object will throw UnsupportedOperationException. 53 * 54 * <p>The UnicodeSet class is not designed to be subclassed. 55 * 56 * <p><code>UnicodeSet</code> supports two APIs. The first is the 57 * <em>operand</em> API that allows the caller to modify the value of 58 * a <code>UnicodeSet</code> object. It conforms to Java 2's 59 * <code>java.util.Set</code> interface, although 60 * <code>UnicodeSet</code> does not actually implement that 61 * interface. All methods of <code>Set</code> are supported, with the 62 * modification that they take a character range or single character 63 * instead of an <code>Object</code>, and they take a 64 * <code>UnicodeSet</code> instead of a <code>Collection</code>. The 65 * operand API may be thought of in terms of boolean logic: a boolean 66 * OR is implemented by <code>add</code>, a boolean AND is implemented 67 * by <code>retain</code>, a boolean XOR is implemented by 68 * <code>complement</code> taking an argument, and a boolean NOT is 69 * implemented by <code>complement</code> with no argument. In terms 70 * of traditional set theory function names, <code>add</code> is a 71 * union, <code>retain</code> is an intersection, <code>remove</code> 72 * is an asymmetric difference, and <code>complement</code> with no 73 * argument is a set complement with respect to the superset range 74 * <code>MIN_VALUE-MAX_VALUE</code> 75 * 76 * <p>The second API is the 77 * <code>applyPattern()</code>/<code>toPattern()</code> API from the 78 * <code>java.text.Format</code>-derived classes. Unlike the 79 * methods that add characters, add categories, and control the logic 80 * of the set, the method <code>applyPattern()</code> sets all 81 * attributes of a <code>UnicodeSet</code> at once, based on a 82 * string pattern. 83 * 84 * <p><b>Pattern syntax</b></p> 85 * 86 * Patterns are accepted by the constructors and the 87 * <code>applyPattern()</code> methods and returned by the 88 * <code>toPattern()</code> method. These patterns follow a syntax 89 * similar to that employed by version 8 regular expression character 90 * classes. Here are some simple examples: 91 * 92 * <blockquote> 93 * <table> 94 * <tr style="vertical-align: top"> 95 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[]</code></td> 96 * <td style="vertical-align: top;">No characters</td> 97 * </tr><tr style="vertical-align: top"> 98 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[a]</code></td> 99 * <td style="vertical-align: top;">The character 'a'</td> 100 * </tr><tr style="vertical-align: top"> 101 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[ae]</code></td> 102 * <td style="vertical-align: top;">The characters 'a' and 'e'</td> 103 * </tr> 104 * <tr> 105 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[a-e]</code></td> 106 * <td style="vertical-align: top;">The characters 'a' through 'e' inclusive, in Unicode code 107 * point order</td> 108 * </tr> 109 * <tr> 110 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[\\u4E01]</code></td> 111 * <td style="vertical-align: top;">The character U+4E01</td> 112 * </tr> 113 * <tr> 114 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[a{ab}{ac}]</code></td> 115 * <td style="vertical-align: top;">The character 'a' and the multicharacter strings "ab" and 116 * "ac"</td> 117 * </tr> 118 * <tr> 119 * <td style="white-space: nowrap; vertical-align: top; horizontal-align: left;"><code>[\p{Lu}]</code></td> 120 * <td style="vertical-align: top;">All characters in the general category Uppercase Letter</td> 121 * </tr> 122 * </table> 123 * </blockquote> 124 * 125 * Any character may be preceded by a backslash in order to remove any special 126 * meaning. White space characters, as defined by the Unicode Pattern_White_Space property, are 127 * ignored, unless they are escaped. 128 * 129 * <p>Property patterns specify a set of characters having a certain 130 * property as defined by the Unicode standard. Both the POSIX-like 131 * "[:Lu:]" and the Perl-like syntax "\p{Lu}" are recognized. For a 132 * complete list of supported property patterns, see the User's Guide 133 * for UnicodeSet at 134 * <a href="http://www.icu-project.org/userguide/unicodeSet.html"> 135 * http://www.icu-project.org/userguide/unicodeSet.html</a>. 136 * Actual determination of property data is defined by the underlying 137 * Unicode database as implemented by UCharacter. 138 * 139 * <p>Patterns specify individual characters, ranges of characters, and 140 * Unicode property sets. When elements are concatenated, they 141 * specify their union. To complement a set, place a '^' immediately 142 * after the opening '['. Property patterns are inverted by modifying 143 * their delimiters; "[:^foo]" and "\P{foo}". In any other location, 144 * '^' has no special meaning. 145 * 146 * <p>Ranges are indicated by placing two a '-' between two 147 * characters, as in "a-z". This specifies the range of all 148 * characters from the left to the right, in Unicode order. If the 149 * left character is greater than or equal to the 150 * right character it is a syntax error. If a '-' occurs as the first 151 * character after the opening '[' or '[^', or if it occurs as the 152 * last character before the closing ']', then it is taken as a 153 * literal. Thus "[a\\-b]", "[-ab]", and "[ab-]" all indicate the same 154 * set of three characters, 'a', 'b', and '-'. 155 * 156 * <p>Sets may be intersected using the '&' operator or the asymmetric 157 * set difference may be taken using the '-' operator, for example, 158 * "[[:L:]&[\\u0000-\\u0FFF]]" indicates the set of all Unicode letters 159 * with values less than 4096. Operators ('&' and '|') have equal 160 * precedence and bind left-to-right. Thus 161 * "[[:L:]-[a-z]-[\\u0100-\\u01FF]]" is equivalent to 162 * "[[[:L:]-[a-z]]-[\\u0100-\\u01FF]]". This only really matters for 163 * difference; intersection is commutative. 164 * 165 * <table> 166 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[a]</code><td>The set containing 'a' 167 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[a-z]</code><td>The set containing 'a' 168 * through 'z' and all letters in between, in Unicode order 169 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[^a-z]</code><td>The set containing 170 * all characters but 'a' through 'z', 171 * that is, U+0000 through 'a'-1 and 'z'+1 through U+10FFFF 172 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[[<em>pat1</em>][<em>pat2</em>]]</code> 173 * <td>The union of sets specified by <em>pat1</em> and <em>pat2</em> 174 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[[<em>pat1</em>]&[<em>pat2</em>]]</code> 175 * <td>The intersection of sets specified by <em>pat1</em> and <em>pat2</em> 176 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[[<em>pat1</em>]-[<em>pat2</em>]]</code> 177 * <td>The asymmetric difference of sets specified by <em>pat1</em> and 178 * <em>pat2</em> 179 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[:Lu:] or \p{Lu}</code> 180 * <td>The set of characters having the specified 181 * Unicode property; in 182 * this case, Unicode uppercase letters 183 * <tr style="vertical-align: top;"><td style="white-space: nowrap;"><code>[:^Lu:] or \P{Lu}</code> 184 * <td>The set of characters <em>not</em> having the given 185 * Unicode property 186 * </table> 187 * 188 * <p><b>Warning</b>: you cannot add an empty string ("") to a UnicodeSet.</p> 189 * 190 * <p><b>Formal syntax</b></p> 191 * 192 * <blockquote> 193 * <table> 194 * <tr style="vertical-align: top"> 195 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>pattern := </code></td> 196 * <td style="vertical-align: top;"><code>('[' '^'? item* ']') | 197 * property</code></td> 198 * </tr> 199 * <tr style="vertical-align: top"> 200 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>item := </code></td> 201 * <td style="vertical-align: top;"><code>char | (char '-' char) | pattern-expr<br> 202 * </code></td> 203 * </tr> 204 * <tr style="vertical-align: top"> 205 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>pattern-expr := </code></td> 206 * <td style="vertical-align: top;"><code>pattern | pattern-expr pattern | 207 * pattern-expr op pattern<br> 208 * </code></td> 209 * </tr> 210 * <tr style="vertical-align: top"> 211 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>op := </code></td> 212 * <td style="vertical-align: top;"><code>'&' | '-'<br> 213 * </code></td> 214 * </tr> 215 * <tr style="vertical-align: top"> 216 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>special := </code></td> 217 * <td style="vertical-align: top;"><code>'[' | ']' | '-'<br> 218 * </code></td> 219 * </tr> 220 * <tr style="vertical-align: top"> 221 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>char := </code></td> 222 * <td style="vertical-align: top;"><em>any character that is not</em><code> special<br> 223 * | ('\\' </code><em>any character</em><code>)<br> 224 * | ('\u' hex hex hex hex)<br> 225 * </code></td> 226 * </tr> 227 * <tr style="vertical-align: top"> 228 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>hex := </code></td> 229 * <td style="vertical-align: top;"><em>any character for which 230 * </em><code>Character.digit(c, 16)</code><em> 231 * returns a non-negative result</em></td> 232 * </tr> 233 * <tr> 234 * <td style="white-space: nowrap; vertical-align: top;" align="right"><code>property := </code></td> 235 * <td style="vertical-align: top;"><em>a Unicode property set pattern</em></td> 236 * </tr> 237 * </table> 238 * <br> 239 * <table border="1"> 240 * <tr> 241 * <td>Legend: <table> 242 * <tr> 243 * <td style="white-space: nowrap; vertical-align: top;"><code>a := b</code></td> 244 * <td style="width: 20; vertical-align: top;"> </td> 245 * <td style="vertical-align: top;"><code>a</code> may be replaced by <code>b</code> </td> 246 * </tr> 247 * <tr> 248 * <td style="white-space: nowrap; vertical-align: top;"><code>a?</code></td> 249 * <td style="vertical-align: top;"></td> 250 * <td style="vertical-align: top;">zero or one instance of <code>a</code><br> 251 * </td> 252 * </tr> 253 * <tr> 254 * <td style="white-space: nowrap; vertical-align: top;"><code>a*</code></td> 255 * <td style="vertical-align: top;"></td> 256 * <td style="vertical-align: top;">one or more instances of <code>a</code><br> 257 * </td> 258 * </tr> 259 * <tr> 260 * <td style="white-space: nowrap; vertical-align: top;"><code>a | b</code></td> 261 * <td style="vertical-align: top;"></td> 262 * <td style="vertical-align: top;">either <code>a</code> or <code>b</code><br> 263 * </td> 264 * </tr> 265 * <tr> 266 * <td style="white-space: nowrap; vertical-align: top;"><code>'a'</code></td> 267 * <td style="vertical-align: top;"></td> 268 * <td style="vertical-align: top;">the literal string between the quotes </td> 269 * </tr> 270 * </table> 271 * </td> 272 * </tr> 273 * </table> 274 * </blockquote> 275 * <p>To iterate over contents of UnicodeSet, the following are available: 276 * <ul><li>{@link #ranges()} to iterate through the ranges</li> 277 * <li>{@link #strings()} to iterate through the strings</li> 278 * <li>{@link #iterator()} to iterate through the entire contents in a single loop. 279 * That method is, however, not particularly efficient, since it "boxes" each code point into a String. 280 * </ul> 281 * All of the above can be used in <b>for</b> loops. 282 * The {@link com.ibm.icu.text.UnicodeSetIterator UnicodeSetIterator} can also be used, but not in <b>for</b> loops. 283 * <p>To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}. 284 * 285 * @author Alan Liu 286 * @stable ICU 2.0 287 * @see UnicodeSetIterator 288 * @see UnicodeSetSpanner 289 */ 290 public class UnicodeSet extends UnicodeFilter implements Iterable<String>, Comparable<UnicodeSet>, Freezable<UnicodeSet> { 291 292 /** 293 * Constant for the empty set. 294 * @stable ICU 4.8 295 */ 296 public static final UnicodeSet EMPTY = new UnicodeSet().freeze(); 297 /** 298 * Constant for the set of all code points. (Since UnicodeSets can include strings, does not include everything that a UnicodeSet can.) 299 * @stable ICU 4.8 300 */ 301 public static final UnicodeSet ALL_CODE_POINTS = new UnicodeSet(0, 0x10FFFF).freeze(); 302 303 private static XSymbolTable XSYMBOL_TABLE = null; // for overriding the the function processing 304 305 private static final int LOW = 0x000000; // LOW <= all valid values. ZERO for codepoints 306 private static final int HIGH = 0x110000; // HIGH > all valid values. 10000 for code units. 307 // 110000 for codepoints 308 309 /** 310 * Minimum value that can be stored in a UnicodeSet. 311 * @stable ICU 2.0 312 */ 313 public static final int MIN_VALUE = LOW; 314 315 /** 316 * Maximum value that can be stored in a UnicodeSet. 317 * @stable ICU 2.0 318 */ 319 public static final int MAX_VALUE = HIGH - 1; 320 321 private int len; // length used; list may be longer to minimize reallocs 322 private int[] list; // MUST be terminated with HIGH 323 private int[] rangeList; // internal buffer 324 private int[] buffer; // internal buffer 325 326 // NOTE: normally the field should be of type SortedSet; but that is missing a public clone!! 327 // is not private so that UnicodeSetIterator can get access 328 TreeSet<String> strings = new TreeSet<String>(); 329 330 /** 331 * The pattern representation of this set. This may not be the 332 * most economical pattern. It is the pattern supplied to 333 * applyPattern(), with variables substituted and whitespace 334 * removed. For sets constructed without applyPattern(), or 335 * modified using the non-pattern API, this string will be null, 336 * indicating that toPattern() must generate a pattern 337 * representation from the inversion list. 338 */ 339 private String pat = null; 340 341 private static final int START_EXTRA = 16; // initial storage. Must be >= 0 342 private static final int GROW_EXTRA = START_EXTRA; // extra amount for growth. Must be >= 0 343 344 // Special property set IDs 345 private static final String ANY_ID = "ANY"; // [\u0000-\U0010FFFF] 346 private static final String ASCII_ID = "ASCII"; // [\u0000-\u007F] 347 private static final String ASSIGNED = "Assigned"; // [:^Cn:] 348 349 /** 350 * A set of all characters _except_ the second through last characters of 351 * certain ranges. These ranges are ranges of characters whose 352 * properties are all exactly alike, e.g. CJK Ideographs from 353 * U+4E00 to U+9FA5. 354 */ 355 private static UnicodeSet INCLUSIONS[] = null; 356 357 private volatile BMPSet bmpSet; // The set is frozen if bmpSet or stringSpan is not null. 358 private volatile UnicodeSetStringSpan stringSpan; 359 //---------------------------------------------------------------- 360 // Public API 361 //---------------------------------------------------------------- 362 363 /** 364 * Constructs an empty set. 365 * @stable ICU 2.0 366 */ 367 public UnicodeSet() { 368 list = new int[1 + START_EXTRA]; 369 list[len++] = HIGH; 370 } 371 372 /** 373 * Constructs a copy of an existing set. 374 * @stable ICU 2.0 375 */ 376 public UnicodeSet(UnicodeSet other) { 377 set(other); 378 } 379 380 /** 381 * Constructs a set containing the given range. If <code>end > 382 * start</code> then an empty set is created. 383 * 384 * @param start first character, inclusive, of range 385 * @param end last character, inclusive, of range 386 * @stable ICU 2.0 387 */ 388 public UnicodeSet(int start, int end) { 389 this(); 390 complement(start, end); 391 } 392 393 /** 394 * Quickly constructs a set from a set of ranges <s0, e0, s1, e1, s2, e2, ..., sn, en>. 395 * There must be an even number of integers, and they must be all greater than zero, 396 * all less than or equal to Character.MAX_CODE_POINT. 397 * In each pair (..., si, ei, ...) it must be true that si <= ei 398 * Between adjacent pairs (...ei, sj...), it must be true that ei+1 < sj 399 * @param pairs pairs of character representing ranges 400 * @stable ICU 4.4 401 */ 402 public UnicodeSet(int... pairs) { 403 if ((pairs.length & 1) != 0) { 404 throw new IllegalArgumentException("Must have even number of integers"); 405 } 406 list = new int[pairs.length + 1]; // don't allocate extra space, because it is likely that this is a fixed set. 407 len = list.length; 408 int last = -1; // used to ensure that the results are monotonically increasing. 409 int i = 0; 410 while (i < pairs.length) { 411 // start of pair 412 int start = pairs[i]; 413 if (last >= start) { 414 throw new IllegalArgumentException("Must be monotonically increasing."); 415 } 416 list[i++] = last = start; 417 // end of pair 418 int end = pairs[i] + 1; 419 if (last >= end) { 420 throw new IllegalArgumentException("Must be monotonically increasing."); 421 } 422 list[i++] = last = end; 423 } 424 list[i] = HIGH; // terminate 425 } 426 427 /** 428 * Constructs a set from the given pattern. See the class description 429 * for the syntax of the pattern language. Whitespace is ignored. 430 * @param pattern a string specifying what characters are in the set 431 * @exception java.lang.IllegalArgumentException if the pattern contains 432 * a syntax error. 433 * @stable ICU 2.0 434 */ 435 public UnicodeSet(String pattern) { 436 this(); 437 applyPattern(pattern, null, null, IGNORE_SPACE); 438 } 439 440 /** 441 * Constructs a set from the given pattern. See the class description 442 * for the syntax of the pattern language. 443 * @param pattern a string specifying what characters are in the set 444 * @param ignoreWhitespace if true, ignore Unicode Pattern_White_Space characters 445 * @exception java.lang.IllegalArgumentException if the pattern contains 446 * a syntax error. 447 * @stable ICU 2.0 448 */ 449 public UnicodeSet(String pattern, boolean ignoreWhitespace) { 450 this(); 451 applyPattern(pattern, null, null, ignoreWhitespace ? IGNORE_SPACE : 0); 452 } 453 454 /** 455 * Constructs a set from the given pattern. See the class description 456 * for the syntax of the pattern language. 457 * @param pattern a string specifying what characters are in the set 458 * @param options a bitmask indicating which options to apply. 459 * Valid options are IGNORE_SPACE and CASE. 460 * @exception java.lang.IllegalArgumentException if the pattern contains 461 * a syntax error. 462 * @stable ICU 3.8 463 */ 464 public UnicodeSet(String pattern, int options) { 465 this(); 466 applyPattern(pattern, null, null, options); 467 } 468 469 /** 470 * Constructs a set from the given pattern. See the class description 471 * for the syntax of the pattern language. 472 * @param pattern a string specifying what characters are in the set 473 * @param pos on input, the position in pattern at which to start parsing. 474 * On output, the position after the last character parsed. 475 * @param symbols a symbol table mapping variables to char[] arrays 476 * and chars to UnicodeSets 477 * @exception java.lang.IllegalArgumentException if the pattern 478 * contains a syntax error. 479 * @stable ICU 2.0 480 */ 481 public UnicodeSet(String pattern, ParsePosition pos, SymbolTable symbols) { 482 this(); 483 applyPattern(pattern, pos, symbols, IGNORE_SPACE); 484 } 485 486 /** 487 * Constructs a set from the given pattern. See the class description 488 * for the syntax of the pattern language. 489 * @param pattern a string specifying what characters are in the set 490 * @param pos on input, the position in pattern at which to start parsing. 491 * On output, the position after the last character parsed. 492 * @param symbols a symbol table mapping variables to char[] arrays 493 * and chars to UnicodeSets 494 * @param options a bitmask indicating which options to apply. 495 * Valid options are IGNORE_SPACE and CASE. 496 * @exception java.lang.IllegalArgumentException if the pattern 497 * contains a syntax error. 498 * @stable ICU 3.2 499 */ 500 public UnicodeSet(String pattern, ParsePosition pos, SymbolTable symbols, int options) { 501 this(); 502 applyPattern(pattern, pos, symbols, options); 503 } 504 505 506 /** 507 * Return a new set that is equivalent to this one. 508 * @stable ICU 2.0 509 */ 510 @Override 511 public Object clone() { 512 if (isFrozen()) { 513 return this; 514 } 515 UnicodeSet result = new UnicodeSet(this); 516 result.bmpSet = this.bmpSet; 517 result.stringSpan = this.stringSpan; 518 return result; 519 } 520 521 /** 522 * Make this object represent the range <code>start - end</code>. 523 * If <code>end > start</code> then this object is set to an 524 * an empty range. 525 * 526 * @param start first character in the set, inclusive 527 * @param end last character in the set, inclusive 528 * @stable ICU 2.0 529 */ 530 public UnicodeSet set(int start, int end) { 531 checkFrozen(); 532 clear(); 533 complement(start, end); 534 return this; 535 } 536 537 /** 538 * Make this object represent the same set as <code>other</code>. 539 * @param other a <code>UnicodeSet</code> whose value will be 540 * copied to this object 541 * @stable ICU 2.0 542 */ 543 public UnicodeSet set(UnicodeSet other) { 544 checkFrozen(); 545 list = other.list.clone(); 546 len = other.len; 547 pat = other.pat; 548 strings = new TreeSet<String>(other.strings); 549 return this; 550 } 551 552 /** 553 * Modifies this set to represent the set specified by the given pattern. 554 * See the class description for the syntax of the pattern language. 555 * Whitespace is ignored. 556 * @param pattern a string specifying what characters are in the set 557 * @exception java.lang.IllegalArgumentException if the pattern 558 * contains a syntax error. 559 * @stable ICU 2.0 560 */ 561 public final UnicodeSet applyPattern(String pattern) { 562 checkFrozen(); 563 return applyPattern(pattern, null, null, IGNORE_SPACE); 564 } 565 566 /** 567 * Modifies this set to represent the set specified by the given pattern, 568 * optionally ignoring whitespace. 569 * See the class description for the syntax of the pattern language. 570 * @param pattern a string specifying what characters are in the set 571 * @param ignoreWhitespace if true then Unicode Pattern_White_Space characters are ignored 572 * @exception java.lang.IllegalArgumentException if the pattern 573 * contains a syntax error. 574 * @stable ICU 2.0 575 */ 576 public UnicodeSet applyPattern(String pattern, boolean ignoreWhitespace) { 577 checkFrozen(); 578 return applyPattern(pattern, null, null, ignoreWhitespace ? IGNORE_SPACE : 0); 579 } 580 581 /** 582 * Modifies this set to represent the set specified by the given pattern, 583 * optionally ignoring whitespace. 584 * See the class description for the syntax of the pattern language. 585 * @param pattern a string specifying what characters are in the set 586 * @param options a bitmask indicating which options to apply. 587 * Valid options are IGNORE_SPACE and CASE. 588 * @exception java.lang.IllegalArgumentException if the pattern 589 * contains a syntax error. 590 * @stable ICU 3.8 591 */ 592 public UnicodeSet applyPattern(String pattern, int options) { 593 checkFrozen(); 594 return applyPattern(pattern, null, null, options); 595 } 596 597 /** 598 * Return true if the given position, in the given pattern, appears 599 * to be the start of a UnicodeSet pattern. 600 * @stable ICU 2.0 601 */ 602 public static boolean resemblesPattern(String pattern, int pos) { 603 return ((pos+1) < pattern.length() && 604 pattern.charAt(pos) == '[') || 605 resemblesPropertyPattern(pattern, pos); 606 } 607 608 /** 609 * TODO: create Appendable version of UTF16.append(buf, c), 610 * maybe in new class Appendables? 611 * @throws IOException 612 */ 613 private static void appendCodePoint(Appendable app, int c) { 614 assert 0 <= c && c <= 0x10ffff; 615 try { 616 if (c <= 0xffff) { 617 app.append((char) c); 618 } else { 619 app.append(UTF16.getLeadSurrogate(c)).append(UTF16.getTrailSurrogate(c)); 620 } 621 } catch (IOException e) { 622 throw new ICUUncheckedIOException(e); 623 } 624 } 625 626 /** 627 * TODO: create class Appendables? 628 * @throws IOException 629 */ 630 private static void append(Appendable app, CharSequence s) { 631 try { 632 app.append(s); 633 } catch (IOException e) { 634 throw new ICUUncheckedIOException(e); 635 } 636 } 637 638 /** 639 * Append the <code>toPattern()</code> representation of a 640 * string to the given <code>Appendable</code>. 641 */ 642 private static <T extends Appendable> T _appendToPat(T buf, String s, boolean escapeUnprintable) { 643 int cp; 644 for (int i = 0; i < s.length(); i += Character.charCount(cp)) { 645 cp = s.codePointAt(i); 646 _appendToPat(buf, cp, escapeUnprintable); 647 } 648 return buf; 649 } 650 651 /** 652 * Append the <code>toPattern()</code> representation of a 653 * character to the given <code>Appendable</code>. 654 */ 655 private static <T extends Appendable> T _appendToPat(T buf, int c, boolean escapeUnprintable) { 656 try { 657 if (escapeUnprintable && Utility.isUnprintable(c)) { 658 // Use hex escape notation (<backslash>uxxxx or <backslash>Uxxxxxxxx) for anything 659 // unprintable 660 if (Utility.escapeUnprintable(buf, c)) { 661 return buf; 662 } 663 } 664 // Okay to let ':' pass through 665 switch (c) { 666 case '[': // SET_OPEN: 667 case ']': // SET_CLOSE: 668 case '-': // HYPHEN: 669 case '^': // COMPLEMENT: 670 case '&': // INTERSECTION: 671 case '\\': //BACKSLASH: 672 case '{': 673 case '}': 674 case '$': 675 case ':': 676 buf.append('\\'); 677 break; 678 default: 679 // Escape whitespace 680 if (PatternProps.isWhiteSpace(c)) { 681 buf.append('\\'); 682 } 683 break; 684 } 685 appendCodePoint(buf, c); 686 return buf; 687 } catch (IOException e) { 688 throw new ICUUncheckedIOException(e); 689 } 690 } 691 692 /** 693 * Returns a string representation of this set. If the result of 694 * calling this function is passed to a UnicodeSet constructor, it 695 * will produce another set that is equal to this one. 696 * @stable ICU 2.0 697 */ 698 @Override 699 public String toPattern(boolean escapeUnprintable) { 700 if (pat != null && !escapeUnprintable) { 701 return pat; 702 } 703 StringBuilder result = new StringBuilder(); 704 return _toPattern(result, escapeUnprintable).toString(); 705 } 706 707 /** 708 * Append a string representation of this set to result. This will be 709 * a cleaned version of the string passed to applyPattern(), if there 710 * is one. Otherwise it will be generated. 711 */ 712 private <T extends Appendable> T _toPattern(T result, 713 boolean escapeUnprintable) { 714 if (pat == null) { 715 return appendNewPattern(result, escapeUnprintable, true); 716 } 717 try { 718 if (!escapeUnprintable) { 719 result.append(pat); 720 return result; 721 } 722 boolean oddNumberOfBackslashes = false; 723 for (int i=0; i<pat.length(); ) { 724 int c = pat.codePointAt(i); 725 i += Character.charCount(c); 726 if (Utility.isUnprintable(c)) { 727 // If the unprintable character is preceded by an odd 728 // number of backslashes, then it has been escaped 729 // and we omit the last backslash. 730 Utility.escapeUnprintable(result, c); 731 oddNumberOfBackslashes = false; 732 } else if (!oddNumberOfBackslashes && c == '\\') { 733 // Temporarily withhold an odd-numbered backslash. 734 oddNumberOfBackslashes = true; 735 } else { 736 if (oddNumberOfBackslashes) { 737 result.append('\\'); 738 } 739 appendCodePoint(result, c); 740 oddNumberOfBackslashes = false; 741 } 742 } 743 if (oddNumberOfBackslashes) { 744 result.append('\\'); 745 } 746 return result; 747 } catch (IOException e) { 748 throw new ICUUncheckedIOException(e); 749 } 750 } 751 752 /** 753 * Generate and append a string representation of this set to result. 754 * This does not use this.pat, the cleaned up copy of the string 755 * passed to applyPattern(). 756 * @param result the buffer into which to generate the pattern 757 * @param escapeUnprintable escape unprintable characters if true 758 * @stable ICU 2.0 759 */ 760 public StringBuffer _generatePattern(StringBuffer result, boolean escapeUnprintable) { 761 return _generatePattern(result, escapeUnprintable, true); 762 } 763 764 /** 765 * Generate and append a string representation of this set to result. 766 * This does not use this.pat, the cleaned up copy of the string 767 * passed to applyPattern(). 768 * @param includeStrings if false, doesn't include the strings. 769 * @stable ICU 3.8 770 */ 771 public StringBuffer _generatePattern(StringBuffer result, 772 boolean escapeUnprintable, boolean includeStrings) { 773 return appendNewPattern(result, escapeUnprintable, includeStrings); 774 } 775 776 private <T extends Appendable> T appendNewPattern( 777 T result, boolean escapeUnprintable, boolean includeStrings) { 778 try { 779 result.append('['); 780 781 int count = getRangeCount(); 782 783 // If the set contains at least 2 intervals and includes both 784 // MIN_VALUE and MAX_VALUE, then the inverse representation will 785 // be more economical. 786 if (count > 1 && 787 getRangeStart(0) == MIN_VALUE && 788 getRangeEnd(count-1) == MAX_VALUE) { 789 790 // Emit the inverse 791 result.append('^'); 792 793 for (int i = 1; i < count; ++i) { 794 int start = getRangeEnd(i-1)+1; 795 int end = getRangeStart(i)-1; 796 _appendToPat(result, start, escapeUnprintable); 797 if (start != end) { 798 if ((start+1) != end) { 799 result.append('-'); 800 } 801 _appendToPat(result, end, escapeUnprintable); 802 } 803 } 804 } 805 806 // Default; emit the ranges as pairs 807 else { 808 for (int i = 0; i < count; ++i) { 809 int start = getRangeStart(i); 810 int end = getRangeEnd(i); 811 _appendToPat(result, start, escapeUnprintable); 812 if (start != end) { 813 if ((start+1) != end) { 814 result.append('-'); 815 } 816 _appendToPat(result, end, escapeUnprintable); 817 } 818 } 819 } 820 821 if (includeStrings && strings.size() > 0) { 822 for (String s : strings) { 823 result.append('{'); 824 _appendToPat(result, s, escapeUnprintable); 825 result.append('}'); 826 } 827 } 828 result.append(']'); 829 return result; 830 } catch (IOException e) { 831 throw new ICUUncheckedIOException(e); 832 } 833 } 834 835 /** 836 * Returns the number of elements in this set (its cardinality) 837 * Note than the elements of a set may include both individual 838 * codepoints and strings. 839 * 840 * @return the number of elements in this set (its cardinality). 841 * @stable ICU 2.0 842 */ 843 public int size() { 844 int n = 0; 845 int count = getRangeCount(); 846 for (int i = 0; i < count; ++i) { 847 n += getRangeEnd(i) - getRangeStart(i) + 1; 848 } 849 return n + strings.size(); 850 } 851 852 /** 853 * Returns <tt>true</tt> if this set contains no elements. 854 * 855 * @return <tt>true</tt> if this set contains no elements. 856 * @stable ICU 2.0 857 */ 858 public boolean isEmpty() { 859 return len == 1 && strings.size() == 0; 860 } 861 862 /** 863 * Implementation of UnicodeMatcher API. Returns <tt>true</tt> if 864 * this set contains any character whose low byte is the given 865 * value. This is used by <tt>RuleBasedTransliterator</tt> for 866 * indexing. 867 * @stable ICU 2.0 868 */ 869 @Override 870 public boolean matchesIndexValue(int v) { 871 /* The index value v, in the range [0,255], is contained in this set if 872 * it is contained in any pair of this set. Pairs either have the high 873 * bytes equal, or unequal. If the high bytes are equal, then we have 874 * aaxx..aayy, where aa is the high byte. Then v is contained if xx <= 875 * v <= yy. If the high bytes are unequal we have aaxx..bbyy, bb>aa. 876 * Then v is contained if xx <= v || v <= yy. (This is identical to the 877 * time zone month containment logic.) 878 */ 879 for (int i=0; i<getRangeCount(); ++i) { 880 int low = getRangeStart(i); 881 int high = getRangeEnd(i); 882 if ((low & ~0xFF) == (high & ~0xFF)) { 883 if ((low & 0xFF) <= v && v <= (high & 0xFF)) { 884 return true; 885 } 886 } else if ((low & 0xFF) <= v || v <= (high & 0xFF)) { 887 return true; 888 } 889 } 890 if (strings.size() != 0) { 891 for (String s : strings) { 892 //if (s.length() == 0) { 893 // // Empty strings match everything 894 // return true; 895 //} 896 // assert(s.length() != 0); // We enforce this elsewhere 897 int c = UTF16.charAt(s, 0); 898 if ((c & 0xFF) == v) { 899 return true; 900 } 901 } 902 } 903 return false; 904 } 905 906 /** 907 * Implementation of UnicodeMatcher.matches(). Always matches the 908 * longest possible multichar string. 909 * @stable ICU 2.0 910 */ 911 @Override 912 public int matches(Replaceable text, 913 int[] offset, 914 int limit, 915 boolean incremental) { 916 917 if (offset[0] == limit) { 918 // Strings, if any, have length != 0, so we don't worry 919 // about them here. If we ever allow zero-length strings 920 // we much check for them here. 921 if (contains(UnicodeMatcher.ETHER)) { 922 return incremental ? U_PARTIAL_MATCH : U_MATCH; 923 } else { 924 return U_MISMATCH; 925 } 926 } else { 927 if (strings.size() != 0) { // try strings first 928 929 // might separate forward and backward loops later 930 // for now they are combined 931 932 // TODO Improve efficiency of this, at least in the forward 933 // direction, if not in both. In the forward direction we 934 // can assume the strings are sorted. 935 936 boolean forward = offset[0] < limit; 937 938 // firstChar is the leftmost char to match in the 939 // forward direction or the rightmost char to match in 940 // the reverse direction. 941 char firstChar = text.charAt(offset[0]); 942 943 // If there are multiple strings that can match we 944 // return the longest match. 945 int highWaterLength = 0; 946 947 for (String trial : strings) { 948 //if (trial.length() == 0) { 949 // return U_MATCH; // null-string always matches 950 //} 951 // assert(trial.length() != 0); // We ensure this elsewhere 952 953 char c = trial.charAt(forward ? 0 : trial.length() - 1); 954 955 // Strings are sorted, so we can optimize in the 956 // forward direction. 957 if (forward && c > firstChar) break; 958 if (c != firstChar) continue; 959 960 int length = matchRest(text, offset[0], limit, trial); 961 962 if (incremental) { 963 int maxLen = forward ? limit-offset[0] : offset[0]-limit; 964 if (length == maxLen) { 965 // We have successfully matched but only up to limit. 966 return U_PARTIAL_MATCH; 967 } 968 } 969 970 if (length == trial.length()) { 971 // We have successfully matched the whole string. 972 if (length > highWaterLength) { 973 highWaterLength = length; 974 } 975 // In the forward direction we know strings 976 // are sorted so we can bail early. 977 if (forward && length < highWaterLength) { 978 break; 979 } 980 continue; 981 } 982 } 983 984 // We've checked all strings without a partial match. 985 // If we have full matches, return the longest one. 986 if (highWaterLength != 0) { 987 offset[0] += forward ? highWaterLength : -highWaterLength; 988 return U_MATCH; 989 } 990 } 991 return super.matches(text, offset, limit, incremental); 992 } 993 } 994 995 /** 996 * Returns the longest match for s in text at the given position. 997 * If limit > start then match forward from start+1 to limit 998 * matching all characters except s.charAt(0). If limit < start, 999 * go backward starting from start-1 matching all characters 1000 * except s.charAt(s.length()-1). This method assumes that the 1001 * first character, text.charAt(start), matches s, so it does not 1002 * check it. 1003 * @param text the text to match 1004 * @param start the first character to match. In the forward 1005 * direction, text.charAt(start) is matched against s.charAt(0). 1006 * In the reverse direction, it is matched against 1007 * s.charAt(s.length()-1). 1008 * @param limit the limit offset for matching, either last+1 in 1009 * the forward direction, or last-1 in the reverse direction, 1010 * where last is the index of the last character to match. 1011 * @return If part of s matches up to the limit, return |limit - 1012 * start|. If all of s matches before reaching the limit, return 1013 * s.length(). If there is a mismatch between s and text, return 1014 * 0 1015 */ 1016 private static int matchRest (Replaceable text, int start, int limit, String s) { 1017 int maxLen; 1018 int slen = s.length(); 1019 if (start < limit) { 1020 maxLen = limit - start; 1021 if (maxLen > slen) maxLen = slen; 1022 for (int i = 1; i < maxLen; ++i) { 1023 if (text.charAt(start + i) != s.charAt(i)) return 0; 1024 } 1025 } else { 1026 maxLen = start - limit; 1027 if (maxLen > slen) maxLen = slen; 1028 --slen; // <=> slen = s.length() - 1; 1029 for (int i = 1; i < maxLen; ++i) { 1030 if (text.charAt(start - i) != s.charAt(slen - i)) return 0; 1031 } 1032 } 1033 return maxLen; 1034 } 1035 1036 /** 1037 * Tests whether the text matches at the offset. If so, returns the end of the longest substring that it matches. If not, returns -1. 1038 * @internal 1039 * @deprecated This API is ICU internal only. 1040 */ 1041 @Deprecated 1042 public int matchesAt(CharSequence text, int offset) { 1043 int lastLen = -1; 1044 strings: 1045 if (strings.size() != 0) { 1046 char firstChar = text.charAt(offset); 1047 String trial = null; 1048 // find the first string starting with firstChar 1049 Iterator<String> it = strings.iterator(); 1050 while (it.hasNext()) { 1051 trial = it.next(); 1052 char firstStringChar = trial.charAt(0); 1053 if (firstStringChar < firstChar) continue; 1054 if (firstStringChar > firstChar) break strings; 1055 } 1056 1057 // now keep checking string until we get the longest one 1058 for (;;) { 1059 int tempLen = matchesAt(text, offset, trial); 1060 if (lastLen > tempLen) break strings; 1061 lastLen = tempLen; 1062 if (!it.hasNext()) break; 1063 trial = it.next(); 1064 } 1065 } 1066 1067 if (lastLen < 2) { 1068 int cp = UTF16.charAt(text, offset); 1069 if (contains(cp)) lastLen = UTF16.getCharCount(cp); 1070 } 1071 1072 return offset+lastLen; 1073 } 1074 1075 /** 1076 * Does one string contain another, starting at a specific offset? 1077 * @param text text to match 1078 * @param offsetInText offset within that text 1079 * @param substring substring to match at offset in text 1080 * @return -1 if match fails, otherwise other.length() 1081 */ 1082 // Note: This method was moved from CollectionUtilities 1083 private static int matchesAt(CharSequence text, int offsetInText, CharSequence substring) { 1084 int len = substring.length(); 1085 int textLength = text.length(); 1086 if (textLength + offsetInText > len) { 1087 return -1; 1088 } 1089 int i = 0; 1090 for (int j = offsetInText; i < len; ++i, ++j) { 1091 char pc = substring.charAt(i); 1092 char tc = text.charAt(j); 1093 if (pc != tc) return -1; 1094 } 1095 return i; 1096 } 1097 1098 /** 1099 * Implementation of UnicodeMatcher API. Union the set of all 1100 * characters that may be matched by this object into the given 1101 * set. 1102 * @param toUnionTo the set into which to union the source characters 1103 * @stable ICU 2.2 1104 */ 1105 @Override 1106 public void addMatchSetTo(UnicodeSet toUnionTo) { 1107 toUnionTo.addAll(this); 1108 } 1109 1110 /** 1111 * Returns the index of the given character within this set, where 1112 * the set is ordered by ascending code point. If the character 1113 * is not in this set, return -1. The inverse of this method is 1114 * <code>charAt()</code>. 1115 * @return an index from 0..size()-1, or -1 1116 * @stable ICU 2.0 1117 */ 1118 public int indexOf(int c) { 1119 if (c < MIN_VALUE || c > MAX_VALUE) { 1120 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(c, 6)); 1121 } 1122 int i = 0; 1123 int n = 0; 1124 for (;;) { 1125 int start = list[i++]; 1126 if (c < start) { 1127 return -1; 1128 } 1129 int limit = list[i++]; 1130 if (c < limit) { 1131 return n + c - start; 1132 } 1133 n += limit - start; 1134 } 1135 } 1136 1137 /** 1138 * Returns the character at the given index within this set, where 1139 * the set is ordered by ascending code point. If the index is 1140 * out of range, return -1. The inverse of this method is 1141 * <code>indexOf()</code>. 1142 * @param index an index from 0..size()-1 1143 * @return the character at the given index, or -1. 1144 * @stable ICU 2.0 1145 */ 1146 public int charAt(int index) { 1147 if (index >= 0) { 1148 // len2 is the largest even integer <= len, that is, it is len 1149 // for even values and len-1 for odd values. With odd values 1150 // the last entry is UNICODESET_HIGH. 1151 int len2 = len & ~1; 1152 for (int i=0; i < len2;) { 1153 int start = list[i++]; 1154 int count = list[i++] - start; 1155 if (index < count) { 1156 return start + index; 1157 } 1158 index -= count; 1159 } 1160 } 1161 return -1; 1162 } 1163 1164 /** 1165 * Adds the specified range to this set if it is not already 1166 * present. If this set already contains the specified range, 1167 * the call leaves this set unchanged. If <code>end > start</code> 1168 * then an empty range is added, leaving the set unchanged. 1169 * 1170 * @param start first character, inclusive, of range to be added 1171 * to this set. 1172 * @param end last character, inclusive, of range to be added 1173 * to this set. 1174 * @stable ICU 2.0 1175 */ 1176 public UnicodeSet add(int start, int end) { 1177 checkFrozen(); 1178 return add_unchecked(start, end); 1179 } 1180 1181 /** 1182 * Adds all characters in range (uses preferred naming convention). 1183 * @param start The index of where to start on adding all characters. 1184 * @param end The index of where to end on adding all characters. 1185 * @return a reference to this object 1186 * @stable ICU 4.4 1187 */ 1188 public UnicodeSet addAll(int start, int end) { 1189 checkFrozen(); 1190 return add_unchecked(start, end); 1191 } 1192 1193 // for internal use, after checkFrozen has been called 1194 private UnicodeSet add_unchecked(int start, int end) { 1195 if (start < MIN_VALUE || start > MAX_VALUE) { 1196 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6)); 1197 } 1198 if (end < MIN_VALUE || end > MAX_VALUE) { 1199 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6)); 1200 } 1201 if (start < end) { 1202 add(range(start, end), 2, 0); 1203 } else if (start == end) { 1204 add(start); 1205 } 1206 return this; 1207 } 1208 1209 // /** 1210 // * Format out the inversion list as a string, for debugging. Uncomment when 1211 // * needed. 1212 // */ 1213 // public final String dump() { 1214 // StringBuffer buf = new StringBuffer("["); 1215 // for (int i=0; i<len; ++i) { 1216 // if (i != 0) buf.append(", "); 1217 // int c = list[i]; 1218 // //if (c <= 0x7F && c != '\n' && c != '\r' && c != '\t' && c != ' ') { 1219 // // buf.append((char) c); 1220 // //} else { 1221 // buf.append("U+").append(Utility.hex(c, (c<0x10000)?4:6)); 1222 // //} 1223 // } 1224 // buf.append("]"); 1225 // return buf.toString(); 1226 // } 1227 1228 /** 1229 * Adds the specified character to this set if it is not already 1230 * present. If this set already contains the specified character, 1231 * the call leaves this set unchanged. 1232 * @stable ICU 2.0 1233 */ 1234 public final UnicodeSet add(int c) { 1235 checkFrozen(); 1236 return add_unchecked(c); 1237 } 1238 1239 // for internal use only, after checkFrozen has been called 1240 private final UnicodeSet add_unchecked(int c) { 1241 if (c < MIN_VALUE || c > MAX_VALUE) { 1242 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(c, 6)); 1243 } 1244 1245 // find smallest i such that c < list[i] 1246 // if odd, then it is IN the set 1247 // if even, then it is OUT of the set 1248 int i = findCodePoint(c); 1249 1250 // already in set? 1251 if ((i & 1) != 0) return this; 1252 1253 // HIGH is 0x110000 1254 // assert(list[len-1] == HIGH); 1255 1256 // empty = [HIGH] 1257 // [start_0, limit_0, start_1, limit_1, HIGH] 1258 1259 // [..., start_k-1, limit_k-1, start_k, limit_k, ..., HIGH] 1260 // ^ 1261 // list[i] 1262 1263 // i == 0 means c is before the first range 1264 // TODO: Is the "list[i]-1" a typo? Even if you pass MAX_VALUE into 1265 // add_unchecked, the maximum value that "c" will be compared to 1266 // is "MAX_VALUE-1" meaning that "if (c == MAX_VALUE)" will 1267 // never be reached according to this logic. 1268 if (c == list[i]-1) { 1269 // c is before start of next range 1270 list[i] = c; 1271 // if we touched the HIGH mark, then add a new one 1272 if (c == MAX_VALUE) { 1273 ensureCapacity(len+1); 1274 list[len++] = HIGH; 1275 } 1276 if (i > 0 && c == list[i-1]) { 1277 // collapse adjacent ranges 1278 1279 // [..., start_k-1, c, c, limit_k, ..., HIGH] 1280 // ^ 1281 // list[i] 1282 System.arraycopy(list, i+1, list, i-1, len-i-1); 1283 len -= 2; 1284 } 1285 } 1286 1287 else if (i > 0 && c == list[i-1]) { 1288 // c is after end of prior range 1289 list[i-1]++; 1290 // no need to chcek for collapse here 1291 } 1292 1293 else { 1294 // At this point we know the new char is not adjacent to 1295 // any existing ranges, and it is not 10FFFF. 1296 1297 1298 // [..., start_k-1, limit_k-1, start_k, limit_k, ..., HIGH] 1299 // ^ 1300 // list[i] 1301 1302 // [..., start_k-1, limit_k-1, c, c+1, start_k, limit_k, ..., HIGH] 1303 // ^ 1304 // list[i] 1305 1306 // Don't use ensureCapacity() to save on copying. 1307 // NOTE: This has no measurable impact on performance, 1308 // but it might help in some usage patterns. 1309 if (len+2 > list.length) { 1310 int[] temp = new int[len + 2 + GROW_EXTRA]; 1311 if (i != 0) System.arraycopy(list, 0, temp, 0, i); 1312 System.arraycopy(list, i, temp, i+2, len-i); 1313 list = temp; 1314 } else { 1315 System.arraycopy(list, i, list, i+2, len-i); 1316 } 1317 1318 list[i] = c; 1319 list[i+1] = c+1; 1320 len += 2; 1321 } 1322 1323 pat = null; 1324 return this; 1325 } 1326 1327 /** 1328 * Adds the specified multicharacter to this set if it is not already 1329 * present. If this set already contains the multicharacter, 1330 * the call leaves this set unchanged. 1331 * Thus "ch" => {"ch"} 1332 * <br><b>Warning: you cannot add an empty string ("") to a UnicodeSet.</b> 1333 * @param s the source string 1334 * @return this object, for chaining 1335 * @stable ICU 2.0 1336 */ 1337 public final UnicodeSet add(CharSequence s) { 1338 checkFrozen(); 1339 int cp = getSingleCP(s); 1340 if (cp < 0) { 1341 strings.add(s.toString()); 1342 pat = null; 1343 } else { 1344 add_unchecked(cp, cp); 1345 } 1346 return this; 1347 } 1348 1349 /** 1350 * Utility for getting code point from single code point CharSequence. 1351 * See the public UTF16.getSingleCodePoint() 1352 * @return a code point IF the string consists of a single one. 1353 * otherwise returns -1. 1354 * @param s to test 1355 */ 1356 private static int getSingleCP(CharSequence s) { 1357 if (s.length() < 1) { 1358 throw new IllegalArgumentException("Can't use zero-length strings in UnicodeSet"); 1359 } 1360 if (s.length() > 2) return -1; 1361 if (s.length() == 1) return s.charAt(0); 1362 1363 // at this point, len = 2 1364 int cp = UTF16.charAt(s, 0); 1365 if (cp > 0xFFFF) { // is surrogate pair 1366 return cp; 1367 } 1368 return -1; 1369 } 1370 1371 /** 1372 * Adds each of the characters in this string to the set. Thus "ch" => {"c", "h"} 1373 * If this set already any particular character, it has no effect on that character. 1374 * @param s the source string 1375 * @return this object, for chaining 1376 * @stable ICU 2.0 1377 */ 1378 public final UnicodeSet addAll(CharSequence s) { 1379 checkFrozen(); 1380 int cp; 1381 for (int i = 0; i < s.length(); i += UTF16.getCharCount(cp)) { 1382 cp = UTF16.charAt(s, i); 1383 add_unchecked(cp, cp); 1384 } 1385 return this; 1386 } 1387 1388 /** 1389 * Retains EACH of the characters in this string. Note: "ch" == {"c", "h"} 1390 * If this set already any particular character, it has no effect on that character. 1391 * @param s the source string 1392 * @return this object, for chaining 1393 * @stable ICU 2.0 1394 */ 1395 public final UnicodeSet retainAll(CharSequence s) { 1396 return retainAll(fromAll(s)); 1397 } 1398 1399 /** 1400 * Complement EACH of the characters in this string. Note: "ch" == {"c", "h"} 1401 * If this set already any particular character, it has no effect on that character. 1402 * @param s the source string 1403 * @return this object, for chaining 1404 * @stable ICU 2.0 1405 */ 1406 public final UnicodeSet complementAll(CharSequence s) { 1407 return complementAll(fromAll(s)); 1408 } 1409 1410 /** 1411 * Remove EACH of the characters in this string. Note: "ch" == {"c", "h"} 1412 * If this set already any particular character, it has no effect on that character. 1413 * @param s the source string 1414 * @return this object, for chaining 1415 * @stable ICU 2.0 1416 */ 1417 public final UnicodeSet removeAll(CharSequence s) { 1418 return removeAll(fromAll(s)); 1419 } 1420 1421 /** 1422 * Remove all strings from this UnicodeSet 1423 * @return this object, for chaining 1424 * @stable ICU 4.2 1425 */ 1426 public final UnicodeSet removeAllStrings() { 1427 checkFrozen(); 1428 if (strings.size() != 0) { 1429 strings.clear(); 1430 pat = null; 1431 } 1432 return this; 1433 } 1434 1435 /** 1436 * Makes a set from a multicharacter string. Thus "ch" => {"ch"} 1437 * <br><b>Warning: you cannot add an empty string ("") to a UnicodeSet.</b> 1438 * @param s the source string 1439 * @return a newly created set containing the given string 1440 * @stable ICU 2.0 1441 */ 1442 public static UnicodeSet from(CharSequence s) { 1443 return new UnicodeSet().add(s); 1444 } 1445 1446 1447 /** 1448 * Makes a set from each of the characters in the string. Thus "ch" => {"c", "h"} 1449 * @param s the source string 1450 * @return a newly created set containing the given characters 1451 * @stable ICU 2.0 1452 */ 1453 public static UnicodeSet fromAll(CharSequence s) { 1454 return new UnicodeSet().addAll(s); 1455 } 1456 1457 1458 /** 1459 * Retain only the elements in this set that are contained in the 1460 * specified range. If <code>end > start</code> then an empty range is 1461 * retained, leaving the set empty. 1462 * 1463 * @param start first character, inclusive, of range to be retained 1464 * to this set. 1465 * @param end last character, inclusive, of range to be retained 1466 * to this set. 1467 * @stable ICU 2.0 1468 */ 1469 public UnicodeSet retain(int start, int end) { 1470 checkFrozen(); 1471 if (start < MIN_VALUE || start > MAX_VALUE) { 1472 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6)); 1473 } 1474 if (end < MIN_VALUE || end > MAX_VALUE) { 1475 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6)); 1476 } 1477 if (start <= end) { 1478 retain(range(start, end), 2, 0); 1479 } else { 1480 clear(); 1481 } 1482 return this; 1483 } 1484 1485 /** 1486 * Retain the specified character from this set if it is present. 1487 * Upon return this set will be empty if it did not contain c, or 1488 * will only contain c if it did contain c. 1489 * @param c the character to be retained 1490 * @return this object, for chaining 1491 * @stable ICU 2.0 1492 */ 1493 public final UnicodeSet retain(int c) { 1494 return retain(c, c); 1495 } 1496 1497 /** 1498 * Retain the specified string in this set if it is present. 1499 * Upon return this set will be empty if it did not contain s, or 1500 * will only contain s if it did contain s. 1501 * @param cs the string to be retained 1502 * @return this object, for chaining 1503 * @stable ICU 2.0 1504 */ 1505 public final UnicodeSet retain(CharSequence cs) { 1506 1507 int cp = getSingleCP(cs); 1508 if (cp < 0) { 1509 String s = cs.toString(); 1510 boolean isIn = strings.contains(s); 1511 if (isIn && size() == 1) { 1512 return this; 1513 } 1514 clear(); 1515 strings.add(s); 1516 pat = null; 1517 } else { 1518 retain(cp, cp); 1519 } 1520 return this; 1521 } 1522 1523 /** 1524 * Removes the specified range from this set if it is present. 1525 * The set will not contain the specified range once the call 1526 * returns. If <code>end > start</code> then an empty range is 1527 * removed, leaving the set unchanged. 1528 * 1529 * @param start first character, inclusive, of range to be removed 1530 * from this set. 1531 * @param end last character, inclusive, of range to be removed 1532 * from this set. 1533 * @stable ICU 2.0 1534 */ 1535 public UnicodeSet remove(int start, int end) { 1536 checkFrozen(); 1537 if (start < MIN_VALUE || start > MAX_VALUE) { 1538 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6)); 1539 } 1540 if (end < MIN_VALUE || end > MAX_VALUE) { 1541 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6)); 1542 } 1543 if (start <= end) { 1544 retain(range(start, end), 2, 2); 1545 } 1546 return this; 1547 } 1548 1549 /** 1550 * Removes the specified character from this set if it is present. 1551 * The set will not contain the specified character once the call 1552 * returns. 1553 * @param c the character to be removed 1554 * @return this object, for chaining 1555 * @stable ICU 2.0 1556 */ 1557 public final UnicodeSet remove(int c) { 1558 return remove(c, c); 1559 } 1560 1561 /** 1562 * Removes the specified string from this set if it is present. 1563 * The set will not contain the specified string once the call 1564 * returns. 1565 * @param s the string to be removed 1566 * @return this object, for chaining 1567 * @stable ICU 2.0 1568 */ 1569 public final UnicodeSet remove(CharSequence s) { 1570 int cp = getSingleCP(s); 1571 if (cp < 0) { 1572 strings.remove(s.toString()); 1573 pat = null; 1574 } else { 1575 remove(cp, cp); 1576 } 1577 return this; 1578 } 1579 1580 /** 1581 * Complements the specified range in this set. Any character in 1582 * the range will be removed if it is in this set, or will be 1583 * added if it is not in this set. If <code>end > start</code> 1584 * then an empty range is complemented, leaving the set unchanged. 1585 * 1586 * @param start first character, inclusive, of range to be removed 1587 * from this set. 1588 * @param end last character, inclusive, of range to be removed 1589 * from this set. 1590 * @stable ICU 2.0 1591 */ 1592 public UnicodeSet complement(int start, int end) { 1593 checkFrozen(); 1594 if (start < MIN_VALUE || start > MAX_VALUE) { 1595 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6)); 1596 } 1597 if (end < MIN_VALUE || end > MAX_VALUE) { 1598 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6)); 1599 } 1600 if (start <= end) { 1601 xor(range(start, end), 2, 0); 1602 } 1603 pat = null; 1604 return this; 1605 } 1606 1607 /** 1608 * Complements the specified character in this set. The character 1609 * will be removed if it is in this set, or will be added if it is 1610 * not in this set. 1611 * @stable ICU 2.0 1612 */ 1613 public final UnicodeSet complement(int c) { 1614 return complement(c, c); 1615 } 1616 1617 /** 1618 * This is equivalent to 1619 * <code>complement(MIN_VALUE, MAX_VALUE)</code>. 1620 * @stable ICU 2.0 1621 */ 1622 public UnicodeSet complement() { 1623 checkFrozen(); 1624 if (list[0] == LOW) { 1625 System.arraycopy(list, 1, list, 0, len-1); 1626 --len; 1627 } else { 1628 ensureCapacity(len+1); 1629 System.arraycopy(list, 0, list, 1, len); 1630 list[0] = LOW; 1631 ++len; 1632 } 1633 pat = null; 1634 return this; 1635 } 1636 1637 /** 1638 * Complement the specified string in this set. 1639 * The set will not contain the specified string once the call 1640 * returns. 1641 * <br><b>Warning: you cannot add an empty string ("") to a UnicodeSet.</b> 1642 * @param s the string to complement 1643 * @return this object, for chaining 1644 * @stable ICU 2.0 1645 */ 1646 public final UnicodeSet complement(CharSequence s) { 1647 checkFrozen(); 1648 int cp = getSingleCP(s); 1649 if (cp < 0) { 1650 String s2 = s.toString(); 1651 if (strings.contains(s2)) { 1652 strings.remove(s2); 1653 } else { 1654 strings.add(s2); 1655 } 1656 pat = null; 1657 } else { 1658 complement(cp, cp); 1659 } 1660 return this; 1661 } 1662 1663 /** 1664 * Returns true if this set contains the given character. 1665 * @param c character to be checked for containment 1666 * @return true if the test condition is met 1667 * @stable ICU 2.0 1668 */ 1669 @Override 1670 public boolean contains(int c) { 1671 if (c < MIN_VALUE || c > MAX_VALUE) { 1672 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(c, 6)); 1673 } 1674 if (bmpSet != null) { 1675 return bmpSet.contains(c); 1676 } 1677 if (stringSpan != null) { 1678 return stringSpan.contains(c); 1679 } 1680 1681 /* 1682 // Set i to the index of the start item greater than ch 1683 // We know we will terminate without length test! 1684 int i = -1; 1685 while (true) { 1686 if (c < list[++i]) break; 1687 } 1688 */ 1689 1690 int i = findCodePoint(c); 1691 1692 return ((i & 1) != 0); // return true if odd 1693 } 1694 1695 /** 1696 * Returns the smallest value i such that c < list[i]. Caller 1697 * must ensure that c is a legal value or this method will enter 1698 * an infinite loop. This method performs a binary search. 1699 * @param c a character in the range MIN_VALUE..MAX_VALUE 1700 * inclusive 1701 * @return the smallest integer i in the range 0..len-1, 1702 * inclusive, such that c < list[i] 1703 */ 1704 private final int findCodePoint(int c) { 1705 /* Examples: 1706 findCodePoint(c) 1707 set list[] c=0 1 3 4 7 8 1708 === ============== =========== 1709 [] [110000] 0 0 0 0 0 0 1710 [\u0000-\u0003] [0, 4, 110000] 1 1 1 2 2 2 1711 [\u0004-\u0007] [4, 8, 110000] 0 0 0 1 1 2 1712 [:all:] [0, 110000] 1 1 1 1 1 1 1713 */ 1714 1715 // Return the smallest i such that c < list[i]. Assume 1716 // list[len - 1] == HIGH and that c is legal (0..HIGH-1). 1717 if (c < list[0]) return 0; 1718 // High runner test. c is often after the last range, so an 1719 // initial check for this condition pays off. 1720 if (len >= 2 && c >= list[len-2]) return len-1; 1721 int lo = 0; 1722 int hi = len - 1; 1723 // invariant: c >= list[lo] 1724 // invariant: c < list[hi] 1725 for (;;) { 1726 int i = (lo + hi) >>> 1; 1727 if (i == lo) return hi; 1728 if (c < list[i]) { 1729 hi = i; 1730 } else { 1731 lo = i; 1732 } 1733 } 1734 } 1735 1736 // //---------------------------------------------------------------- 1737 // // Unrolled binary search 1738 // //---------------------------------------------------------------- 1739 // 1740 // private int validLen = -1; // validated value of len 1741 // private int topOfLow; 1742 // private int topOfHigh; 1743 // private int power; 1744 // private int deltaStart; 1745 // 1746 // private void validate() { 1747 // if (len <= 1) { 1748 // throw new IllegalArgumentException("list.len==" + len + "; must be >1"); 1749 // } 1750 // 1751 // // find greatest power of 2 less than or equal to len 1752 // for (power = exp2.length-1; power > 0 && exp2[power] > len; power--) {} 1753 // 1754 // // assert(exp2[power] <= len); 1755 // 1756 // // determine the starting points 1757 // topOfLow = exp2[power] - 1; 1758 // topOfHigh = len - 1; 1759 // deltaStart = exp2[power-1]; 1760 // validLen = len; 1761 // } 1762 // 1763 // private static final int exp2[] = { 1764 // 0x1, 0x2, 0x4, 0x8, 1765 // 0x10, 0x20, 0x40, 0x80, 1766 // 0x100, 0x200, 0x400, 0x800, 1767 // 0x1000, 0x2000, 0x4000, 0x8000, 1768 // 0x10000, 0x20000, 0x40000, 0x80000, 1769 // 0x100000, 0x200000, 0x400000, 0x800000, 1770 // 0x1000000, 0x2000000, 0x4000000, 0x8000000, 1771 // 0x10000000, 0x20000000 // , 0x40000000 // no unsigned int in Java 1772 // }; 1773 // 1774 // /** 1775 // * Unrolled lowest index GT. 1776 // */ 1777 // private final int leastIndexGT(int searchValue) { 1778 // 1779 // if (len != validLen) { 1780 // if (len == 1) return 0; 1781 // validate(); 1782 // } 1783 // int temp; 1784 // 1785 // // set up initial range to search. Each subrange is a power of two in length 1786 // int high = searchValue < list[topOfLow] ? topOfLow : topOfHigh; 1787 // 1788 // // Completely unrolled binary search, folhighing "Programming Pearls" 1789 // // Each case deliberately falls through to the next 1790 // // Logically, list[-1] < all_search_values && list[count] > all_search_values 1791 // // although the values -1 and count are never actually touched. 1792 // 1793 // // The bounds at each point are low & high, 1794 // // where low == high - delta*2 1795 // // so high - delta is the midpoint 1796 // 1797 // // The invariant AFTER each line is that list[low] < searchValue <= list[high] 1798 // 1799 // switch (power) { 1800 // //case 31: if (searchValue < list[temp = high-0x40000000]) high = temp; // no unsigned int in Java 1801 // case 30: if (searchValue < list[temp = high-0x20000000]) high = temp; 1802 // case 29: if (searchValue < list[temp = high-0x10000000]) high = temp; 1803 // 1804 // case 28: if (searchValue < list[temp = high- 0x8000000]) high = temp; 1805 // case 27: if (searchValue < list[temp = high- 0x4000000]) high = temp; 1806 // case 26: if (searchValue < list[temp = high- 0x2000000]) high = temp; 1807 // case 25: if (searchValue < list[temp = high- 0x1000000]) high = temp; 1808 // 1809 // case 24: if (searchValue < list[temp = high- 0x800000]) high = temp; 1810 // case 23: if (searchValue < list[temp = high- 0x400000]) high = temp; 1811 // case 22: if (searchValue < list[temp = high- 0x200000]) high = temp; 1812 // case 21: if (searchValue < list[temp = high- 0x100000]) high = temp; 1813 // 1814 // case 20: if (searchValue < list[temp = high- 0x80000]) high = temp; 1815 // case 19: if (searchValue < list[temp = high- 0x40000]) high = temp; 1816 // case 18: if (searchValue < list[temp = high- 0x20000]) high = temp; 1817 // case 17: if (searchValue < list[temp = high- 0x10000]) high = temp; 1818 // 1819 // case 16: if (searchValue < list[temp = high- 0x8000]) high = temp; 1820 // case 15: if (searchValue < list[temp = high- 0x4000]) high = temp; 1821 // case 14: if (searchValue < list[temp = high- 0x2000]) high = temp; 1822 // case 13: if (searchValue < list[temp = high- 0x1000]) high = temp; 1823 // 1824 // case 12: if (searchValue < list[temp = high- 0x800]) high = temp; 1825 // case 11: if (searchValue < list[temp = high- 0x400]) high = temp; 1826 // case 10: if (searchValue < list[temp = high- 0x200]) high = temp; 1827 // case 9: if (searchValue < list[temp = high- 0x100]) high = temp; 1828 // 1829 // case 8: if (searchValue < list[temp = high- 0x80]) high = temp; 1830 // case 7: if (searchValue < list[temp = high- 0x40]) high = temp; 1831 // case 6: if (searchValue < list[temp = high- 0x20]) high = temp; 1832 // case 5: if (searchValue < list[temp = high- 0x10]) high = temp; 1833 // 1834 // case 4: if (searchValue < list[temp = high- 0x8]) high = temp; 1835 // case 3: if (searchValue < list[temp = high- 0x4]) high = temp; 1836 // case 2: if (searchValue < list[temp = high- 0x2]) high = temp; 1837 // case 1: if (searchValue < list[temp = high- 0x1]) high = temp; 1838 // } 1839 // 1840 // return high; 1841 // } 1842 // 1843 // // For debugging only 1844 // public int len() { 1845 // return len; 1846 // } 1847 // 1848 // //---------------------------------------------------------------- 1849 // //---------------------------------------------------------------- 1850 1851 /** 1852 * Returns true if this set contains every character 1853 * of the given range. 1854 * @param start first character, inclusive, of the range 1855 * @param end last character, inclusive, of the range 1856 * @return true if the test condition is met 1857 * @stable ICU 2.0 1858 */ 1859 public boolean contains(int start, int end) { 1860 if (start < MIN_VALUE || start > MAX_VALUE) { 1861 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6)); 1862 } 1863 if (end < MIN_VALUE || end > MAX_VALUE) { 1864 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6)); 1865 } 1866 //int i = -1; 1867 //while (true) { 1868 // if (start < list[++i]) break; 1869 //} 1870 int i = findCodePoint(start); 1871 return ((i & 1) != 0 && end < list[i]); 1872 } 1873 1874 /** 1875 * Returns <tt>true</tt> if this set contains the given 1876 * multicharacter string. 1877 * @param s string to be checked for containment 1878 * @return <tt>true</tt> if this set contains the specified string 1879 * @stable ICU 2.0 1880 */ 1881 public final boolean contains(CharSequence s) { 1882 1883 int cp = getSingleCP(s); 1884 if (cp < 0) { 1885 return strings.contains(s.toString()); 1886 } else { 1887 return contains(cp); 1888 } 1889 } 1890 1891 /** 1892 * Returns true if this set contains all the characters and strings 1893 * of the given set. 1894 * @param b set to be checked for containment 1895 * @return true if the test condition is met 1896 * @stable ICU 2.0 1897 */ 1898 public boolean containsAll(UnicodeSet b) { 1899 // The specified set is a subset if all of its pairs are contained in 1900 // this set. This implementation accesses the lists directly for speed. 1901 // TODO: this could be faster if size() were cached. But that would affect building speed 1902 // so it needs investigation. 1903 int[] listB = b.list; 1904 boolean needA = true; 1905 boolean needB = true; 1906 int aPtr = 0; 1907 int bPtr = 0; 1908 int aLen = len - 1; 1909 int bLen = b.len - 1; 1910 int startA = 0, startB = 0, limitA = 0, limitB = 0; 1911 while (true) { 1912 // double iterations are such a pain... 1913 if (needA) { 1914 if (aPtr >= aLen) { 1915 // ran out of A. If B is also exhausted, then break; 1916 if (needB && bPtr >= bLen) { 1917 break; 1918 } 1919 return false; 1920 } 1921 startA = list[aPtr++]; 1922 limitA = list[aPtr++]; 1923 } 1924 if (needB) { 1925 if (bPtr >= bLen) { 1926 // ran out of B. Since we got this far, we have an A and we are ok so far 1927 break; 1928 } 1929 startB = listB[bPtr++]; 1930 limitB = listB[bPtr++]; 1931 } 1932 // if B doesn't overlap and is greater than A, get new A 1933 if (startB >= limitA) { 1934 needA = true; 1935 needB = false; 1936 continue; 1937 } 1938 // if B is wholy contained in A, then get a new B 1939 if (startB >= startA && limitB <= limitA) { 1940 needA = false; 1941 needB = true; 1942 continue; 1943 } 1944 // all other combinations mean we fail 1945 return false; 1946 } 1947 1948 if (!strings.containsAll(b.strings)) return false; 1949 return true; 1950 } 1951 1952 // /** 1953 // * Returns true if this set contains all the characters and strings 1954 // * of the given set. 1955 // * @param c set to be checked for containment 1956 // * @return true if the test condition is met 1957 // * @stable ICU 2.0 1958 // */ 1959 // public boolean containsAllOld(UnicodeSet c) { 1960 // // The specified set is a subset if all of its pairs are contained in 1961 // // this set. It's possible to code this more efficiently in terms of 1962 // // direct manipulation of the inversion lists if the need arises. 1963 // int n = c.getRangeCount(); 1964 // for (int i=0; i<n; ++i) { 1965 // if (!contains(c.getRangeStart(i), c.getRangeEnd(i))) { 1966 // return false; 1967 // } 1968 // } 1969 // if (!strings.containsAll(c.strings)) return false; 1970 // return true; 1971 // } 1972 1973 /** 1974 * Returns true if there is a partition of the string such that this set contains each of the partitioned strings. 1975 * For example, for the Unicode set [a{bc}{cd}]<br> 1976 * containsAll is true for each of: "a", "bc", ""cdbca"<br> 1977 * containsAll is false for each of: "acb", "bcda", "bcx"<br> 1978 * @param s string containing characters to be checked for containment 1979 * @return true if the test condition is met 1980 * @stable ICU 2.0 1981 */ 1982 public boolean containsAll(String s) { 1983 int cp; 1984 for (int i = 0; i < s.length(); i += UTF16.getCharCount(cp)) { 1985 cp = UTF16.charAt(s, i); 1986 if (!contains(cp)) { 1987 if (strings.size() == 0) { 1988 return false; 1989 } 1990 return containsAll(s, 0); 1991 } 1992 } 1993 return true; 1994 } 1995 1996 /** 1997 * Recursive routine called if we fail to find a match in containsAll, and there are strings 1998 * @param s source string 1999 * @param i point to match to the end on 2000 * @return true if ok 2001 */ 2002 private boolean containsAll(String s, int i) { 2003 if (i >= s.length()) { 2004 return true; 2005 } 2006 int cp= UTF16.charAt(s, i); 2007 if (contains(cp) && containsAll(s, i+UTF16.getCharCount(cp))) { 2008 return true; 2009 } 2010 for (String setStr : strings) { 2011 if (s.startsWith(setStr, i) && containsAll(s, i+setStr.length())) { 2012 return true; 2013 } 2014 } 2015 return false; 2016 2017 } 2018 2019 /** 2020 * Get the Regex equivalent for this UnicodeSet 2021 * @return regex pattern equivalent to this UnicodeSet 2022 * @internal 2023 * @deprecated This API is ICU internal only. 2024 */ 2025 @Deprecated 2026 public String getRegexEquivalent() { 2027 if (strings.size() == 0) { 2028 return toString(); 2029 } 2030 StringBuilder result = new StringBuilder("(?:"); 2031 appendNewPattern(result, true, false); 2032 for (String s : strings) { 2033 result.append('|'); 2034 _appendToPat(result, s, true); 2035 } 2036 return result.append(")").toString(); 2037 } 2038 2039 /** 2040 * Returns true if this set contains none of the characters 2041 * of the given range. 2042 * @param start first character, inclusive, of the range 2043 * @param end last character, inclusive, of the range 2044 * @return true if the test condition is met 2045 * @stable ICU 2.0 2046 */ 2047 public boolean containsNone(int start, int end) { 2048 if (start < MIN_VALUE || start > MAX_VALUE) { 2049 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6)); 2050 } 2051 if (end < MIN_VALUE || end > MAX_VALUE) { 2052 throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6)); 2053 } 2054 int i = -1; 2055 while (true) { 2056 if (start < list[++i]) break; 2057 } 2058 return ((i & 1) == 0 && end < list[i]); 2059 } 2060 2061 /** 2062 * Returns true if none of the characters or strings in this UnicodeSet appears in the string. 2063 * For example, for the Unicode set [a{bc}{cd}]<br> 2064 * containsNone is true for: "xy", "cb"<br> 2065 * containsNone is false for: "a", "bc", "bcd"<br> 2066 * @param b set to be checked for containment 2067 * @return true if the test condition is met 2068 * @stable ICU 2.0 2069 */ 2070 public boolean containsNone(UnicodeSet b) { 2071 // The specified set is a subset if some of its pairs overlap with some of this set's pairs. 2072 // This implementation accesses the lists directly for speed. 2073 int[] listB = b.list; 2074 boolean needA = true; 2075 boolean needB = true; 2076 int aPtr = 0; 2077 int bPtr = 0; 2078 int aLen = len - 1; 2079 int bLen = b.len - 1; 2080 int startA = 0, startB = 0, limitA = 0, limitB = 0; 2081 while (true) { 2082 // double iterations are such a pain... 2083 if (needA) { 2084 if (aPtr >= aLen) { 2085 // ran out of A: break so we test strings 2086 break; 2087 } 2088 startA = list[aPtr++]; 2089 limitA = list[aPtr++]; 2090 } 2091 if (needB) { 2092 if (bPtr >= bLen) { 2093 // ran out of B: break so we test strings 2094 break; 2095 } 2096 startB = listB[bPtr++]; 2097 limitB = listB[bPtr++]; 2098 } 2099 // if B is higher than any part of A, get new A 2100 if (startB >= limitA) { 2101 needA = true; 2102 needB = false; 2103 continue; 2104 } 2105 // if A is higher than any part of B, get new B 2106 if (startA >= limitB) { 2107 needA = false; 2108 needB = true; 2109 continue; 2110 } 2111 // all other combinations mean we fail 2112 return false; 2113 } 2114 2115 if (!SortedSetRelation.hasRelation(strings, SortedSetRelation.DISJOINT, b.strings)) return false; 2116 return true; 2117 } 2118 2119 // /** 2120 // * Returns true if none of the characters or strings in this UnicodeSet appears in the string. 2121 // * For example, for the Unicode set [a{bc}{cd}]<br> 2122 // * containsNone is true for: "xy", "cb"<br> 2123 // * containsNone is false for: "a", "bc", "bcd"<br> 2124 // * @param c set to be checked for containment 2125 // * @return true if the test condition is met 2126 // * @stable ICU 2.0 2127 // */ 2128 // public boolean containsNoneOld(UnicodeSet c) { 2129 // // The specified set is a subset if all of its pairs are contained in 2130 // // this set. It's possible to code this more efficiently in terms of 2131 // // direct manipulation of the inversion lists if the need arises. 2132 // int n = c.getRangeCount(); 2133 // for (int i=0; i<n; ++i) { 2134 // if (!containsNone(c.getRangeStart(i), c.getRangeEnd(i))) { 2135 // return false; 2136 // } 2137 // } 2138 // if (!SortedSetRelation.hasRelation(strings, SortedSetRelation.DISJOINT, c.strings)) return false; 2139 // return true; 2140 // } 2141 2142 /** 2143 * Returns true if this set contains none of the characters 2144 * of the given string. 2145 * @param s string containing characters to be checked for containment 2146 * @return true if the test condition is met 2147 * @stable ICU 2.0 2148 */ 2149 public boolean containsNone(CharSequence s) { 2150 return span(s, SpanCondition.NOT_CONTAINED) == s.length(); 2151 } 2152 2153 /** 2154 * Returns true if this set contains one or more of the characters 2155 * in the given range. 2156 * @param start first character, inclusive, of the range 2157 * @param end last character, inclusive, of the range 2158 * @return true if the condition is met 2159 * @stable ICU 2.0 2160 */ 2161 public final boolean containsSome(int start, int end) { 2162 return !containsNone(start, end); 2163 } 2164 2165 /** 2166 * Returns true if this set contains one or more of the characters 2167 * and strings of the given set. 2168 * @param s set to be checked for containment 2169 * @return true if the condition is met 2170 * @stable ICU 2.0 2171 */ 2172 public final boolean containsSome(UnicodeSet s) { 2173 return !containsNone(s); 2174 } 2175 2176 /** 2177 * Returns true if this set contains one or more of the characters 2178 * of the given string. 2179 * @param s string containing characters to be checked for containment 2180 * @return true if the condition is met 2181 * @stable ICU 2.0 2182 */ 2183 public final boolean containsSome(CharSequence s) { 2184 return !containsNone(s); 2185 } 2186 2187 2188 /** 2189 * Adds all of the elements in the specified set to this set if 2190 * they're not already present. This operation effectively 2191 * modifies this set so that its value is the <i>union</i> of the two 2192 * sets. The behavior of this operation is unspecified if the specified 2193 * collection is modified while the operation is in progress. 2194 * 2195 * @param c set whose elements are to be added to this set. 2196 * @stable ICU 2.0 2197 */ 2198 public UnicodeSet addAll(UnicodeSet c) { 2199 checkFrozen(); 2200 add(c.list, c.len, 0); 2201 strings.addAll(c.strings); 2202 return this; 2203 } 2204 2205 /** 2206 * Retains only the elements in this set that are contained in the 2207 * specified set. In other words, removes from this set all of 2208 * its elements that are not contained in the specified set. This 2209 * operation effectively modifies this set so that its value is 2210 * the <i>intersection</i> of the two sets. 2211 * 2212 * @param c set that defines which elements this set will retain. 2213 * @stable ICU 2.0 2214 */ 2215 public UnicodeSet retainAll(UnicodeSet c) { 2216 checkFrozen(); 2217 retain(c.list, c.len, 0); 2218 strings.retainAll(c.strings); 2219 return this; 2220 } 2221 2222 /** 2223 * Removes from this set all of its elements that are contained in the 2224 * specified set. This operation effectively modifies this 2225 * set so that its value is the <i>asymmetric set difference</i> of 2226 * the two sets. 2227 * 2228 * @param c set that defines which elements will be removed from 2229 * this set. 2230 * @stable ICU 2.0 2231 */ 2232 public UnicodeSet removeAll(UnicodeSet c) { 2233 checkFrozen(); 2234 retain(c.list, c.len, 2); 2235 strings.removeAll(c.strings); 2236 return this; 2237 } 2238 2239 /** 2240 * Complements in this set all elements contained in the specified 2241 * set. Any character in the other set will be removed if it is 2242 * in this set, or will be added if it is not in this set. 2243 * 2244 * @param c set that defines which elements will be complemented from 2245 * this set. 2246 * @stable ICU 2.0 2247 */ 2248 public UnicodeSet complementAll(UnicodeSet c) { 2249 checkFrozen(); 2250 xor(c.list, c.len, 0); 2251 SortedSetRelation.doOperation(strings, SortedSetRelation.COMPLEMENTALL, c.strings); 2252 return this; 2253 } 2254 2255 /** 2256 * Removes all of the elements from this set. This set will be 2257 * empty after this call returns. 2258 * @stable ICU 2.0 2259 */ 2260 public UnicodeSet clear() { 2261 checkFrozen(); 2262 list[0] = HIGH; 2263 len = 1; 2264 pat = null; 2265 strings.clear(); 2266 return this; 2267 } 2268 2269 /** 2270 * Iteration method that returns the number of ranges contained in 2271 * this set. 2272 * @see #getRangeStart 2273 * @see #getRangeEnd 2274 * @stable ICU 2.0 2275 */ 2276 public int getRangeCount() { 2277 return len/2; 2278 } 2279 2280 /** 2281 * Iteration method that returns the first character in the 2282 * specified range of this set. 2283 * @exception ArrayIndexOutOfBoundsException if index is outside 2284 * the range <code>0..getRangeCount()-1</code> 2285 * @see #getRangeCount 2286 * @see #getRangeEnd 2287 * @stable ICU 2.0 2288 */ 2289 public int getRangeStart(int index) { 2290 return list[index*2]; 2291 } 2292 2293 /** 2294 * Iteration method that returns the last character in the 2295 * specified range of this set. 2296 * @exception ArrayIndexOutOfBoundsException if index is outside 2297 * the range <code>0..getRangeCount()-1</code> 2298 * @see #getRangeStart 2299 * @see #getRangeEnd 2300 * @stable ICU 2.0 2301 */ 2302 public int getRangeEnd(int index) { 2303 return (list[index*2 + 1] - 1); 2304 } 2305 2306 /** 2307 * Reallocate this objects internal structures to take up the least 2308 * possible space, without changing this object's value. 2309 * @stable ICU 2.0 2310 */ 2311 public UnicodeSet compact() { 2312 checkFrozen(); 2313 if (len != list.length) { 2314 int[] temp = new int[len]; 2315 System.arraycopy(list, 0, temp, 0, len); 2316 list = temp; 2317 } 2318 rangeList = null; 2319 buffer = null; 2320 return this; 2321 } 2322 2323 /** 2324 * Compares the specified object with this set for equality. Returns 2325 * <tt>true</tt> if the specified object is also a set, the two sets 2326 * have the same size, and every member of the specified set is 2327 * contained in this set (or equivalently, every member of this set is 2328 * contained in the specified set). 2329 * 2330 * @param o Object to be compared for equality with this set. 2331 * @return <tt>true</tt> if the specified Object is equal to this set. 2332 * @stable ICU 2.0 2333 */ 2334 @Override 2335 public boolean equals(Object o) { 2336 if (o == null) { 2337 return false; 2338 } 2339 if (this == o) { 2340 return true; 2341 } 2342 try { 2343 UnicodeSet that = (UnicodeSet) o; 2344 if (len != that.len) return false; 2345 for (int i = 0; i < len; ++i) { 2346 if (list[i] != that.list[i]) return false; 2347 } 2348 if (!strings.equals(that.strings)) return false; 2349 } catch (Exception e) { 2350 return false; 2351 } 2352 return true; 2353 } 2354 2355 /** 2356 * Returns the hash code value for this set. 2357 * 2358 * @return the hash code value for this set. 2359 * @see java.lang.Object#hashCode() 2360 * @stable ICU 2.0 2361 */ 2362 @Override 2363 public int hashCode() { 2364 int result = len; 2365 for (int i = 0; i < len; ++i) { 2366 result *= 1000003; 2367 result += list[i]; 2368 } 2369 return result; 2370 } 2371 2372 /** 2373 * Return a programmer-readable string representation of this object. 2374 * @stable ICU 2.0 2375 */ 2376 @Override 2377 public String toString() { 2378 return toPattern(true); 2379 } 2380 2381 //---------------------------------------------------------------- 2382 // Implementation: Pattern parsing 2383 //---------------------------------------------------------------- 2384 2385 /** 2386 * Parses the given pattern, starting at the given position. The character 2387 * at pattern.charAt(pos.getIndex()) must be '[', or the parse fails. 2388 * Parsing continues until the corresponding closing ']'. If a syntax error 2389 * is encountered between the opening and closing brace, the parse fails. 2390 * Upon return from a successful parse, the ParsePosition is updated to 2391 * point to the character following the closing ']', and an inversion 2392 * list for the parsed pattern is returned. This method 2393 * calls itself recursively to parse embedded subpatterns. 2394 * 2395 * @param pattern the string containing the pattern to be parsed. The 2396 * portion of the string from pos.getIndex(), which must be a '[', to the 2397 * corresponding closing ']', is parsed. 2398 * @param pos upon entry, the position at which to being parsing. The 2399 * character at pattern.charAt(pos.getIndex()) must be a '['. Upon return 2400 * from a successful parse, pos.getIndex() is either the character after the 2401 * closing ']' of the parsed pattern, or pattern.length() if the closing ']' 2402 * is the last character of the pattern string. 2403 * @return an inversion list for the parsed substring 2404 * of <code>pattern</code> 2405 * @exception java.lang.IllegalArgumentException if the parse fails. 2406 * @internal 2407 * @deprecated This API is ICU internal only. 2408 */ 2409 @Deprecated 2410 public UnicodeSet applyPattern(String pattern, 2411 ParsePosition pos, 2412 SymbolTable symbols, 2413 int options) { 2414 2415 // Need to build the pattern in a temporary string because 2416 // _applyPattern calls add() etc., which set pat to empty. 2417 boolean parsePositionWasNull = pos == null; 2418 if (parsePositionWasNull) { 2419 pos = new ParsePosition(0); 2420 } 2421 2422 StringBuilder rebuiltPat = new StringBuilder(); 2423 RuleCharacterIterator chars = 2424 new RuleCharacterIterator(pattern, symbols, pos); 2425 applyPattern(chars, symbols, rebuiltPat, options); 2426 if (chars.inVariable()) { 2427 syntaxError(chars, "Extra chars in variable value"); 2428 } 2429 pat = rebuiltPat.toString(); 2430 if (parsePositionWasNull) { 2431 int i = pos.getIndex(); 2432 2433 // Skip over trailing whitespace 2434 if ((options & IGNORE_SPACE) != 0) { 2435 i = PatternProps.skipWhiteSpace(pattern, i); 2436 } 2437 2438 if (i != pattern.length()) { 2439 throw new IllegalArgumentException("Parse of \"" + pattern + 2440 "\" failed at " + i); 2441 } 2442 } 2443 return this; 2444 } 2445 2446 // Add constants to make the applyPattern() code easier to follow. 2447 2448 private static final int LAST0_START = 0, 2449 LAST1_RANGE = 1, 2450 LAST2_SET = 2; 2451 2452 private static final int MODE0_NONE = 0, 2453 MODE1_INBRACKET = 1, 2454 MODE2_OUTBRACKET = 2; 2455 2456 private static final int SETMODE0_NONE = 0, 2457 SETMODE1_UNICODESET = 1, 2458 SETMODE2_PROPERTYPAT = 2, 2459 SETMODE3_PREPARSED = 3; 2460 2461 /** 2462 * Parse the pattern from the given RuleCharacterIterator. The 2463 * iterator is advanced over the parsed pattern. 2464 * @param chars iterator over the pattern characters. Upon return 2465 * it will be advanced to the first character after the parsed 2466 * pattern, or the end of the iteration if all characters are 2467 * parsed. 2468 * @param symbols symbol table to use to parse and dereference 2469 * variables, or null if none. 2470 * @param rebuiltPat the pattern that was parsed, rebuilt or 2471 * copied from the input pattern, as appropriate. 2472 * @param options a bit mask of zero or more of the following: 2473 * IGNORE_SPACE, CASE. 2474 */ 2475 private void applyPattern(RuleCharacterIterator chars, SymbolTable symbols, 2476 Appendable rebuiltPat, int options) { 2477 2478 // Syntax characters: [ ] ^ - & { } 2479 2480 // Recognized special forms for chars, sets: c-c s-s s&s 2481 2482 int opts = RuleCharacterIterator.PARSE_VARIABLES | 2483 RuleCharacterIterator.PARSE_ESCAPES; 2484 if ((options & IGNORE_SPACE) != 0) { 2485 opts |= RuleCharacterIterator.SKIP_WHITESPACE; 2486 } 2487 2488 StringBuilder patBuf = new StringBuilder(), buf = null; 2489 boolean usePat = false; 2490 UnicodeSet scratch = null; 2491 Object backup = null; 2492 2493 // mode: 0=before [, 1=between [...], 2=after ] 2494 // lastItem: 0=none, 1=char, 2=set 2495 int lastItem = LAST0_START, lastChar = 0, mode = MODE0_NONE; 2496 char op = 0; 2497 2498 boolean invert = false; 2499 2500 clear(); 2501 String lastString = null; 2502 2503 while (mode != MODE2_OUTBRACKET && !chars.atEnd()) { 2504 //Eclipse stated the following is "dead code" 2505 /* 2506 if (false) { 2507 // Debugging assertion 2508 if (!((lastItem == 0 && op == 0) || 2509 (lastItem == 1 && (op == 0 || op == '-')) || 2510 (lastItem == 2 && (op == 0 || op == '-' || op == '&')))) { 2511 throw new IllegalArgumentException(); 2512 } 2513 }*/ 2514 2515 int c = 0; 2516 boolean literal = false; 2517 UnicodeSet nested = null; 2518 2519 // -------- Check for property pattern 2520 2521 // setMode: 0=none, 1=unicodeset, 2=propertypat, 3=preparsed 2522 int setMode = SETMODE0_NONE; 2523 if (resemblesPropertyPattern(chars, opts)) { 2524 setMode = SETMODE2_PROPERTYPAT; 2525 } 2526 2527 // -------- Parse '[' of opening delimiter OR nested set. 2528 // If there is a nested set, use `setMode' to define how 2529 // the set should be parsed. If the '[' is part of the 2530 // opening delimiter for this pattern, parse special 2531 // strings "[", "[^", "[-", and "[^-". Check for stand-in 2532 // characters representing a nested set in the symbol 2533 // table. 2534 2535 else { 2536 // Prepare to backup if necessary 2537 backup = chars.getPos(backup); 2538 c = chars.next(opts); 2539 literal = chars.isEscaped(); 2540 2541 if (c == '[' && !literal) { 2542 if (mode == MODE1_INBRACKET) { 2543 chars.setPos(backup); // backup 2544 setMode = SETMODE1_UNICODESET; 2545 } else { 2546 // Handle opening '[' delimiter 2547 mode = MODE1_INBRACKET; 2548 patBuf.append('['); 2549 backup = chars.getPos(backup); // prepare to backup 2550 c = chars.next(opts); 2551 literal = chars.isEscaped(); 2552 if (c == '^' && !literal) { 2553 invert = true; 2554 patBuf.append('^'); 2555 backup = chars.getPos(backup); // prepare to backup 2556 c = chars.next(opts); 2557 literal = chars.isEscaped(); 2558 } 2559 // Fall through to handle special leading '-'; 2560 // otherwise restart loop for nested [], \p{}, etc. 2561 if (c == '-') { 2562 literal = true; 2563 // Fall through to handle literal '-' below 2564 } else { 2565 chars.setPos(backup); // backup 2566 continue; 2567 } 2568 } 2569 } else if (symbols != null) { 2570 UnicodeMatcher m = symbols.lookupMatcher(c); // may be null 2571 if (m != null) { 2572 try { 2573 nested = (UnicodeSet) m; 2574 setMode = SETMODE3_PREPARSED; 2575 } catch (ClassCastException e) { 2576 syntaxError(chars, "Syntax error"); 2577 } 2578 } 2579 } 2580 } 2581 2582 // -------- Handle a nested set. This either is inline in 2583 // the pattern or represented by a stand-in that has 2584 // previously been parsed and was looked up in the symbol 2585 // table. 2586 2587 if (setMode != SETMODE0_NONE) { 2588 if (lastItem == LAST1_RANGE) { 2589 if (op != 0) { 2590 syntaxError(chars, "Char expected after operator"); 2591 } 2592 add_unchecked(lastChar, lastChar); 2593 _appendToPat(patBuf, lastChar, false); 2594 lastItem = LAST0_START; 2595 op = 0; 2596 } 2597 2598 if (op == '-' || op == '&') { 2599 patBuf.append(op); 2600 } 2601 2602 if (nested == null) { 2603 if (scratch == null) scratch = new UnicodeSet(); 2604 nested = scratch; 2605 } 2606 switch (setMode) { 2607 case SETMODE1_UNICODESET: 2608 nested.applyPattern(chars, symbols, patBuf, options); 2609 break; 2610 case SETMODE2_PROPERTYPAT: 2611 chars.skipIgnored(opts); 2612 nested.applyPropertyPattern(chars, patBuf, symbols); 2613 break; 2614 case SETMODE3_PREPARSED: // `nested' already parsed 2615 nested._toPattern(patBuf, false); 2616 break; 2617 } 2618 2619 usePat = true; 2620 2621 if (mode == MODE0_NONE) { 2622 // Entire pattern is a category; leave parse loop 2623 set(nested); 2624 mode = MODE2_OUTBRACKET; 2625 break; 2626 } 2627 2628 switch (op) { 2629 case '-': 2630 removeAll(nested); 2631 break; 2632 case '&': 2633 retainAll(nested); 2634 break; 2635 case 0: 2636 addAll(nested); 2637 break; 2638 } 2639 2640 op = 0; 2641 lastItem = LAST2_SET; 2642 2643 continue; 2644 } 2645 2646 if (mode == MODE0_NONE) { 2647 syntaxError(chars, "Missing '['"); 2648 } 2649 2650 // -------- Parse special (syntax) characters. If the 2651 // current character is not special, or if it is escaped, 2652 // then fall through and handle it below. 2653 2654 if (!literal) { 2655 switch (c) { 2656 case ']': 2657 if (lastItem == LAST1_RANGE) { 2658 add_unchecked(lastChar, lastChar); 2659 _appendToPat(patBuf, lastChar, false); 2660 } 2661 // Treat final trailing '-' as a literal 2662 if (op == '-') { 2663 add_unchecked(op, op); 2664 patBuf.append(op); 2665 } else if (op == '&') { 2666 syntaxError(chars, "Trailing '&'"); 2667 } 2668 patBuf.append(']'); 2669 mode = MODE2_OUTBRACKET; 2670 continue; 2671 case '-': 2672 if (op == 0) { 2673 if (lastItem != LAST0_START) { 2674 op = (char) c; 2675 continue; 2676 } else if (lastString != null) { 2677 op = (char) c; 2678 continue; 2679 } else { 2680 // Treat final trailing '-' as a literal 2681 add_unchecked(c, c); 2682 c = chars.next(opts); 2683 literal = chars.isEscaped(); 2684 if (c == ']' && !literal) { 2685 patBuf.append("-]"); 2686 mode = MODE2_OUTBRACKET; 2687 continue; 2688 } 2689 } 2690 } 2691 syntaxError(chars, "'-' not after char, string, or set"); 2692 break; 2693 case '&': 2694 if (lastItem == LAST2_SET && op == 0) { 2695 op = (char) c; 2696 continue; 2697 } 2698 syntaxError(chars, "'&' not after set"); 2699 break; 2700 case '^': 2701 syntaxError(chars, "'^' not after '['"); 2702 break; 2703 case '{': 2704 if (op != 0 && op != '-') { 2705 syntaxError(chars, "Missing operand after operator"); 2706 } 2707 if (lastItem == LAST1_RANGE) { 2708 add_unchecked(lastChar, lastChar); 2709 _appendToPat(patBuf, lastChar, false); 2710 } 2711 lastItem = LAST0_START; 2712 if (buf == null) { 2713 buf = new StringBuilder(); 2714 } else { 2715 buf.setLength(0); 2716 } 2717 boolean ok = false; 2718 while (!chars.atEnd()) { 2719 c = chars.next(opts); 2720 literal = chars.isEscaped(); 2721 if (c == '}' && !literal) { 2722 ok = true; 2723 break; 2724 } 2725 appendCodePoint(buf, c); 2726 } 2727 if (buf.length() < 1 || !ok) { 2728 syntaxError(chars, "Invalid multicharacter string"); 2729 } 2730 // We have new string. Add it to set and continue; 2731 // we don't need to drop through to the further 2732 // processing 2733 String curString = buf.toString(); 2734 if (op == '-') { 2735 int lastSingle = CharSequences.getSingleCodePoint(lastString == null ? "" : lastString); 2736 int curSingle = CharSequences.getSingleCodePoint(curString); 2737 if (lastSingle != Integer.MAX_VALUE && curSingle != Integer.MAX_VALUE) { 2738 add(lastSingle,curSingle); 2739 } else { 2740 try { 2741 StringRange.expand(lastString, curString, true, strings); 2742 } catch (Exception e) { 2743 syntaxError(chars, e.getMessage()); 2744 } 2745 } 2746 lastString = null; 2747 op = 0; 2748 } else { 2749 add(curString); 2750 lastString = curString; 2751 } 2752 patBuf.append('{'); 2753 _appendToPat(patBuf, curString, false); 2754 patBuf.append('}'); 2755 continue; 2756 case SymbolTable.SYMBOL_REF: 2757 // symbols nosymbols 2758 // [a-$] error error (ambiguous) 2759 // [a$] anchor anchor 2760 // [a-$x] var "x"* literal '$' 2761 // [a-$.] error literal '$' 2762 // *We won't get here in the case of var "x" 2763 backup = chars.getPos(backup); 2764 c = chars.next(opts); 2765 literal = chars.isEscaped(); 2766 boolean anchor = (c == ']' && !literal); 2767 if (symbols == null && !anchor) { 2768 c = SymbolTable.SYMBOL_REF; 2769 chars.setPos(backup); 2770 break; // literal '$' 2771 } 2772 if (anchor && op == 0) { 2773 if (lastItem == LAST1_RANGE) { 2774 add_unchecked(lastChar, lastChar); 2775 _appendToPat(patBuf, lastChar, false); 2776 } 2777 add_unchecked(UnicodeMatcher.ETHER); 2778 usePat = true; 2779 patBuf.append(SymbolTable.SYMBOL_REF).append(']'); 2780 mode = MODE2_OUTBRACKET; 2781 continue; 2782 } 2783 syntaxError(chars, "Unquoted '$'"); 2784 break; 2785 default: 2786 break; 2787 } 2788 } 2789 2790 // -------- Parse literal characters. This includes both 2791 // escaped chars ("\u4E01") and non-syntax characters 2792 // ("a"). 2793 2794 switch (lastItem) { 2795 case LAST0_START: 2796 if (op == '-' && lastString != null) { 2797 syntaxError(chars, "Invalid range"); 2798 } 2799 lastItem = LAST1_RANGE; 2800 lastChar = c; 2801 lastString = null; 2802 break; 2803 case LAST1_RANGE: 2804 if (op == '-') { 2805 if (lastString != null) { 2806 syntaxError(chars, "Invalid range"); 2807 } 2808 if (lastChar >= c) { 2809 // Don't allow redundant (a-a) or empty (b-a) ranges; 2810 // these are most likely typos. 2811 syntaxError(chars, "Invalid range"); 2812 } 2813 add_unchecked(lastChar, c); 2814 _appendToPat(patBuf, lastChar, false); 2815 patBuf.append(op); 2816 _appendToPat(patBuf, c, false); 2817 lastItem = LAST0_START; 2818 op = 0; 2819 } else { 2820 add_unchecked(lastChar, lastChar); 2821 _appendToPat(patBuf, lastChar, false); 2822 lastChar = c; 2823 } 2824 break; 2825 case LAST2_SET: 2826 if (op != 0) { 2827 syntaxError(chars, "Set expected after operator"); 2828 } 2829 lastChar = c; 2830 lastItem = LAST1_RANGE; 2831 break; 2832 } 2833 } 2834 2835 if (mode != MODE2_OUTBRACKET) { 2836 syntaxError(chars, "Missing ']'"); 2837 } 2838 2839 chars.skipIgnored(opts); 2840 2841 /** 2842 * Handle global flags (invert, case insensitivity). If this 2843 * pattern should be compiled case-insensitive, then we need 2844 * to close over case BEFORE COMPLEMENTING. This makes 2845 * patterns like /[^abc]/i work. 2846 */ 2847 if ((options & CASE) != 0) { 2848 closeOver(CASE); 2849 } 2850 if (invert) { 2851 complement(); 2852 } 2853 2854 // Use the rebuilt pattern (pat) only if necessary. Prefer the 2855 // generated pattern. 2856 if (usePat) { 2857 append(rebuiltPat, patBuf.toString()); 2858 } else { 2859 appendNewPattern(rebuiltPat, false, true); 2860 } 2861 } 2862 2863 private static void syntaxError(RuleCharacterIterator chars, String msg) { 2864 throw new IllegalArgumentException("Error: " + msg + " at \"" + 2865 Utility.escape(chars.toString()) + 2866 '"'); 2867 } 2868 2869 /** 2870 * Add the contents of the UnicodeSet (as strings) into a collection. 2871 * @param target collection to add into 2872 * @stable ICU 4.4 2873 */ 2874 public <T extends Collection<String>> T addAllTo(T target) { 2875 return addAllTo(this, target); 2876 } 2877 2878 2879 /** 2880 * Add the contents of the UnicodeSet (as strings) into a collection. 2881 * @param target collection to add into 2882 * @stable ICU 4.4 2883 */ 2884 public String[] addAllTo(String[] target) { 2885 return addAllTo(this, target); 2886 } 2887 2888 /** 2889 * Add the contents of the UnicodeSet (as strings) into an array. 2890 * @stable ICU 4.4 2891 */ 2892 public static String[] toArray(UnicodeSet set) { 2893 return addAllTo(set, new String[set.size()]); 2894 } 2895 2896 /** 2897 * Add the contents of the collection (as strings) into this UnicodeSet. 2898 * The collection must not contain null. 2899 * @param source the collection to add 2900 * @return a reference to this object 2901 * @stable ICU 4.4 2902 */ 2903 public UnicodeSet add(Iterable<?> source) { 2904 return addAll(source); 2905 } 2906 2907 /** 2908 * Add a collection (as strings) into this UnicodeSet. 2909 * Uses standard naming convention. 2910 * @param source collection to add into 2911 * @return a reference to this object 2912 * @stable ICU 4.4 2913 */ 2914 public UnicodeSet addAll(Iterable<?> source) { 2915 checkFrozen(); 2916 for (Object o : source) { 2917 add(o.toString()); 2918 } 2919 return this; 2920 } 2921 2922 //---------------------------------------------------------------- 2923 // Implementation: Utility methods 2924 //---------------------------------------------------------------- 2925 2926 private void ensureCapacity(int newLen) { 2927 if (newLen <= list.length) return; 2928 int[] temp = new int[newLen + GROW_EXTRA]; 2929 System.arraycopy(list, 0, temp, 0, len); 2930 list = temp; 2931 } 2932 2933 private void ensureBufferCapacity(int newLen) { 2934 if (buffer != null && newLen <= buffer.length) return; 2935 buffer = new int[newLen + GROW_EXTRA]; 2936 } 2937 2938 /** 2939 * Assumes start <= end. 2940 */ 2941 private int[] range(int start, int end) { 2942 if (rangeList == null) { 2943 rangeList = new int[] { start, end+1, HIGH }; 2944 } else { 2945 rangeList[0] = start; 2946 rangeList[1] = end+1; 2947 } 2948 return rangeList; 2949 } 2950 2951 //---------------------------------------------------------------- 2952 // Implementation: Fundamental operations 2953 //---------------------------------------------------------------- 2954 2955 // polarity = 0, 3 is normal: x xor y 2956 // polarity = 1, 2: x xor ~y == x === y 2957 2958 private UnicodeSet xor(int[] other, int otherLen, int polarity) { 2959 ensureBufferCapacity(len + otherLen); 2960 int i = 0, j = 0, k = 0; 2961 int a = list[i++]; 2962 int b; 2963 // TODO: Based on the call hierarchy, polarity of 1 or 2 is never used 2964 // so the following if statement will not be called. 2965 ///CLOVER:OFF 2966 if (polarity == 1 || polarity == 2) { 2967 b = LOW; 2968 if (other[j] == LOW) { // skip base if already LOW 2969 ++j; 2970 b = other[j]; 2971 } 2972 ///CLOVER:ON 2973 } else { 2974 b = other[j++]; 2975 } 2976 // simplest of all the routines 2977 // sort the values, discarding identicals! 2978 while (true) { 2979 if (a < b) { 2980 buffer[k++] = a; 2981 a = list[i++]; 2982 } else if (b < a) { 2983 buffer[k++] = b; 2984 b = other[j++]; 2985 } else if (a != HIGH) { // at this point, a == b 2986 // discard both values! 2987 a = list[i++]; 2988 b = other[j++]; 2989 } else { // DONE! 2990 buffer[k++] = HIGH; 2991 len = k; 2992 break; 2993 } 2994 } 2995 // swap list and buffer 2996 int[] temp = list; 2997 list = buffer; 2998 buffer = temp; 2999 pat = null; 3000 return this; 3001 } 3002 3003 // polarity = 0 is normal: x union y 3004 // polarity = 2: x union ~y 3005 // polarity = 1: ~x union y 3006 // polarity = 3: ~x union ~y 3007 3008 private UnicodeSet add(int[] other, int otherLen, int polarity) { 3009 ensureBufferCapacity(len + otherLen); 3010 int i = 0, j = 0, k = 0; 3011 int a = list[i++]; 3012 int b = other[j++]; 3013 // change from xor is that we have to check overlapping pairs 3014 // polarity bit 1 means a is second, bit 2 means b is. 3015 main: 3016 while (true) { 3017 switch (polarity) { 3018 case 0: // both first; take lower if unequal 3019 if (a < b) { // take a 3020 // Back up over overlapping ranges in buffer[] 3021 if (k > 0 && a <= buffer[k-1]) { 3022 // Pick latter end value in buffer[] vs. list[] 3023 a = max(list[i], buffer[--k]); 3024 } else { 3025 // No overlap 3026 buffer[k++] = a; 3027 a = list[i]; 3028 } 3029 i++; // Common if/else code factored out 3030 polarity ^= 1; 3031 } else if (b < a) { // take b 3032 if (k > 0 && b <= buffer[k-1]) { 3033 b = max(other[j], buffer[--k]); 3034 } else { 3035 buffer[k++] = b; 3036 b = other[j]; 3037 } 3038 j++; 3039 polarity ^= 2; 3040 } else { // a == b, take a, drop b 3041 if (a == HIGH) break main; 3042 // This is symmetrical; it doesn't matter if 3043 // we backtrack with a or b. - liu 3044 if (k > 0 && a <= buffer[k-1]) { 3045 a = max(list[i], buffer[--k]); 3046 } else { 3047 // No overlap 3048 buffer[k++] = a; 3049 a = list[i]; 3050 } 3051 i++; 3052 polarity ^= 1; 3053 b = other[j++]; polarity ^= 2; 3054 } 3055 break; 3056 case 3: // both second; take higher if unequal, and drop other 3057 if (b <= a) { // take a 3058 if (a == HIGH) break main; 3059 buffer[k++] = a; 3060 } else { // take b 3061 if (b == HIGH) break main; 3062 buffer[k++] = b; 3063 } 3064 a = list[i++]; polarity ^= 1; // factored common code 3065 b = other[j++]; polarity ^= 2; 3066 break; 3067 case 1: // a second, b first; if b < a, overlap 3068 if (a < b) { // no overlap, take a 3069 buffer[k++] = a; a = list[i++]; polarity ^= 1; 3070 } else if (b < a) { // OVERLAP, drop b 3071 b = other[j++]; polarity ^= 2; 3072 } else { // a == b, drop both! 3073 if (a == HIGH) break main; 3074 a = list[i++]; polarity ^= 1; 3075 b = other[j++]; polarity ^= 2; 3076 } 3077 break; 3078 case 2: // a first, b second; if a < b, overlap 3079 if (b < a) { // no overlap, take b 3080 buffer[k++] = b; b = other[j++]; polarity ^= 2; 3081 } else if (a < b) { // OVERLAP, drop a 3082 a = list[i++]; polarity ^= 1; 3083 } else { // a == b, drop both! 3084 if (a == HIGH) break main; 3085 a = list[i++]; polarity ^= 1; 3086 b = other[j++]; polarity ^= 2; 3087 } 3088 break; 3089 } 3090 } 3091 buffer[k++] = HIGH; // terminate 3092 len = k; 3093 // swap list and buffer 3094 int[] temp = list; 3095 list = buffer; 3096 buffer = temp; 3097 pat = null; 3098 return this; 3099 } 3100 3101 // polarity = 0 is normal: x intersect y 3102 // polarity = 2: x intersect ~y == set-minus 3103 // polarity = 1: ~x intersect y 3104 // polarity = 3: ~x intersect ~y 3105 3106 private UnicodeSet retain(int[] other, int otherLen, int polarity) { 3107 ensureBufferCapacity(len + otherLen); 3108 int i = 0, j = 0, k = 0; 3109 int a = list[i++]; 3110 int b = other[j++]; 3111 // change from xor is that we have to check overlapping pairs 3112 // polarity bit 1 means a is second, bit 2 means b is. 3113 main: 3114 while (true) { 3115 switch (polarity) { 3116 case 0: // both first; drop the smaller 3117 if (a < b) { // drop a 3118 a = list[i++]; polarity ^= 1; 3119 } else if (b < a) { // drop b 3120 b = other[j++]; polarity ^= 2; 3121 } else { // a == b, take one, drop other 3122 if (a == HIGH) break main; 3123 buffer[k++] = a; a = list[i++]; polarity ^= 1; 3124 b = other[j++]; polarity ^= 2; 3125 } 3126 break; 3127 case 3: // both second; take lower if unequal 3128 if (a < b) { // take a 3129 buffer[k++] = a; a = list[i++]; polarity ^= 1; 3130 } else if (b < a) { // take b 3131 buffer[k++] = b; b = other[j++]; polarity ^= 2; 3132 } else { // a == b, take one, drop other 3133 if (a == HIGH) break main; 3134 buffer[k++] = a; a = list[i++]; polarity ^= 1; 3135 b = other[j++]; polarity ^= 2; 3136 } 3137 break; 3138 case 1: // a second, b first; 3139 if (a < b) { // NO OVERLAP, drop a 3140 a = list[i++]; polarity ^= 1; 3141 } else if (b < a) { // OVERLAP, take b 3142 buffer[k++] = b; b = other[j++]; polarity ^= 2; 3143 } else { // a == b, drop both! 3144 if (a == HIGH) break main; 3145 a = list[i++]; polarity ^= 1; 3146 b = other[j++]; polarity ^= 2; 3147 } 3148 break; 3149 case 2: // a first, b second; if a < b, overlap 3150 if (b < a) { // no overlap, drop b 3151 b = other[j++]; polarity ^= 2; 3152 } else if (a < b) { // OVERLAP, take a 3153 buffer[k++] = a; a = list[i++]; polarity ^= 1; 3154 } else { // a == b, drop both! 3155 if (a == HIGH) break main; 3156 a = list[i++]; polarity ^= 1; 3157 b = other[j++]; polarity ^= 2; 3158 } 3159 break; 3160 } 3161 } 3162 buffer[k++] = HIGH; // terminate 3163 len = k; 3164 // swap list and buffer 3165 int[] temp = list; 3166 list = buffer; 3167 buffer = temp; 3168 pat = null; 3169 return this; 3170 } 3171 3172 private static final int max(int a, int b) { 3173 return (a > b) ? a : b; 3174 } 3175 3176 //---------------------------------------------------------------- 3177 // Generic filter-based scanning code 3178 //---------------------------------------------------------------- 3179 3180 private static interface Filter { 3181 boolean contains(int codePoint); 3182 } 3183 3184 private static class NumericValueFilter implements Filter { 3185 double value; 3186 NumericValueFilter(double value) { this.value = value; } 3187 @Override 3188 public boolean contains(int ch) { 3189 return UCharacter.getUnicodeNumericValue(ch) == value; 3190 } 3191 } 3192 3193 private static class GeneralCategoryMaskFilter implements Filter { 3194 int mask; 3195 GeneralCategoryMaskFilter(int mask) { this.mask = mask; } 3196 @Override 3197 public boolean contains(int ch) { 3198 return ((1 << UCharacter.getType(ch)) & mask) != 0; 3199 } 3200 } 3201 3202 private static class IntPropertyFilter implements Filter { 3203 int prop; 3204 int value; 3205 IntPropertyFilter(int prop, int value) { 3206 this.prop = prop; 3207 this.value = value; 3208 } 3209 @Override 3210 public boolean contains(int ch) { 3211 return UCharacter.getIntPropertyValue(ch, prop) == value; 3212 } 3213 } 3214 3215 private static class ScriptExtensionsFilter implements Filter { 3216 int script; 3217 ScriptExtensionsFilter(int script) { this.script = script; } 3218 @Override 3219 public boolean contains(int c) { 3220 return UScript.hasScript(c, script); 3221 } 3222 } 3223 3224 // VersionInfo for unassigned characters 3225 private static final VersionInfo NO_VERSION = VersionInfo.getInstance(0, 0, 0, 0); 3226 3227 private static class VersionFilter implements Filter { 3228 VersionInfo version; 3229 VersionFilter(VersionInfo version) { this.version = version; } 3230 @Override 3231 public boolean contains(int ch) { 3232 VersionInfo v = UCharacter.getAge(ch); 3233 // Reference comparison ok; VersionInfo caches and reuses 3234 // unique objects. 3235 return !Utility.sameObjects(v, NO_VERSION) && 3236 v.compareTo(version) <= 0; 3237 } 3238 } 3239 3240 private static synchronized UnicodeSet getInclusions(int src) { 3241 if (INCLUSIONS == null) { 3242 INCLUSIONS = new UnicodeSet[UCharacterProperty.SRC_COUNT]; 3243 } 3244 if(INCLUSIONS[src] == null) { 3245 UnicodeSet incl = new UnicodeSet(); 3246 switch(src) { 3247 case UCharacterProperty.SRC_CHAR: 3248 UCharacterProperty.INSTANCE.addPropertyStarts(incl); 3249 break; 3250 case UCharacterProperty.SRC_PROPSVEC: 3251 UCharacterProperty.INSTANCE.upropsvec_addPropertyStarts(incl); 3252 break; 3253 case UCharacterProperty.SRC_CHAR_AND_PROPSVEC: 3254 UCharacterProperty.INSTANCE.addPropertyStarts(incl); 3255 UCharacterProperty.INSTANCE.upropsvec_addPropertyStarts(incl); 3256 break; 3257 case UCharacterProperty.SRC_CASE_AND_NORM: 3258 Norm2AllModes.getNFCInstance().impl.addPropertyStarts(incl); 3259 UCaseProps.INSTANCE.addPropertyStarts(incl); 3260 break; 3261 case UCharacterProperty.SRC_NFC: 3262 Norm2AllModes.getNFCInstance().impl.addPropertyStarts(incl); 3263 break; 3264 case UCharacterProperty.SRC_NFKC: 3265 Norm2AllModes.getNFKCInstance().impl.addPropertyStarts(incl); 3266 break; 3267 case UCharacterProperty.SRC_NFKC_CF: 3268 Norm2AllModes.getNFKC_CFInstance().impl.addPropertyStarts(incl); 3269 break; 3270 case UCharacterProperty.SRC_NFC_CANON_ITER: 3271 Norm2AllModes.getNFCInstance().impl.addCanonIterPropertyStarts(incl); 3272 break; 3273 case UCharacterProperty.SRC_CASE: 3274 UCaseProps.INSTANCE.addPropertyStarts(incl); 3275 break; 3276 case UCharacterProperty.SRC_BIDI: 3277 UBiDiProps.INSTANCE.addPropertyStarts(incl); 3278 break; 3279 default: 3280 throw new IllegalStateException("UnicodeSet.getInclusions(unknown src "+src+")"); 3281 } 3282 INCLUSIONS[src] = incl; 3283 } 3284 return INCLUSIONS[src]; 3285 } 3286 3287 /** 3288 * Generic filter-based scanning code for UCD property UnicodeSets. 3289 */ 3290 private UnicodeSet applyFilter(Filter filter, int src) { 3291 // Logically, walk through all Unicode characters, noting the start 3292 // and end of each range for which filter.contain(c) is 3293 // true. Add each range to a set. 3294 // 3295 // To improve performance, use an inclusions set which 3296 // encodes information about character ranges that are known 3297 // to have identical properties. 3298 // getInclusions(src) contains exactly the first characters of 3299 // same-value ranges for the given properties "source". 3300 3301 clear(); 3302 3303 int startHasProperty = -1; 3304 UnicodeSet inclusions = getInclusions(src); 3305 int limitRange = inclusions.getRangeCount(); 3306 3307 for (int j=0; j<limitRange; ++j) { 3308 // get current range 3309 int start = inclusions.getRangeStart(j); 3310 int end = inclusions.getRangeEnd(j); 3311 3312 // for all the code points in the range, process 3313 for (int ch = start; ch <= end; ++ch) { 3314 // only add to the unicodeset on inflection points -- 3315 // where the hasProperty value changes to false 3316 if (filter.contains(ch)) { 3317 if (startHasProperty < 0) { 3318 startHasProperty = ch; 3319 } 3320 } else if (startHasProperty >= 0) { 3321 add_unchecked(startHasProperty, ch-1); 3322 startHasProperty = -1; 3323 } 3324 } 3325 } 3326 if (startHasProperty >= 0) { 3327 add_unchecked(startHasProperty, 0x10FFFF); 3328 } 3329 3330 return this; 3331 } 3332 3333 3334 /** 3335 * Remove leading and trailing Pattern_White_Space and compress 3336 * internal Pattern_White_Space to a single space character. 3337 */ 3338 private static String mungeCharName(String source) { 3339 source = PatternProps.trimWhiteSpace(source); 3340 StringBuilder buf = null; 3341 for (int i=0; i<source.length(); ++i) { 3342 char ch = source.charAt(i); 3343 if (PatternProps.isWhiteSpace(ch)) { 3344 if (buf == null) { 3345 buf = new StringBuilder().append(source, 0, i); 3346 } else if (buf.charAt(buf.length() - 1) == ' ') { 3347 continue; 3348 } 3349 ch = ' '; // convert to ' ' 3350 } 3351 if (buf != null) { 3352 buf.append(ch); 3353 } 3354 } 3355 return buf == null ? source : buf.toString(); 3356 } 3357 3358 //---------------------------------------------------------------- 3359 // Property set API 3360 //---------------------------------------------------------------- 3361 3362 /** 3363 * Modifies this set to contain those code points which have the 3364 * given value for the given binary or enumerated property, as 3365 * returned by UCharacter.getIntPropertyValue. Prior contents of 3366 * this set are lost. 3367 * 3368 * @param prop a property in the range 3369 * UProperty.BIN_START..UProperty.BIN_LIMIT-1 or 3370 * UProperty.INT_START..UProperty.INT_LIMIT-1 or. 3371 * UProperty.MASK_START..UProperty.MASK_LIMIT-1. 3372 * 3373 * @param value a value in the range 3374 * UCharacter.getIntPropertyMinValue(prop).. 3375 * UCharacter.getIntPropertyMaxValue(prop), with one exception. 3376 * If prop is UProperty.GENERAL_CATEGORY_MASK, then value should not be 3377 * a UCharacter.getType() result, but rather a mask value produced 3378 * by logically ORing (1 << UCharacter.getType()) values together. 3379 * This allows grouped categories such as [:L:] to be represented. 3380 * 3381 * @return a reference to this set 3382 * 3383 * @stable ICU 2.4 3384 */ 3385 public UnicodeSet applyIntPropertyValue(int prop, int value) { 3386 checkFrozen(); 3387 if (prop == UProperty.GENERAL_CATEGORY_MASK) { 3388 applyFilter(new GeneralCategoryMaskFilter(value), UCharacterProperty.SRC_CHAR); 3389 } else if (prop == UProperty.SCRIPT_EXTENSIONS) { 3390 applyFilter(new ScriptExtensionsFilter(value), UCharacterProperty.SRC_PROPSVEC); 3391 } else { 3392 applyFilter(new IntPropertyFilter(prop, value), UCharacterProperty.INSTANCE.getSource(prop)); 3393 } 3394 return this; 3395 } 3396 3397 3398 3399 /** 3400 * Modifies this set to contain those code points which have the 3401 * given value for the given property. Prior contents of this 3402 * set are lost. 3403 * 3404 * @param propertyAlias a property alias, either short or long. 3405 * The name is matched loosely. See PropertyAliases.txt for names 3406 * and a description of loose matching. If the value string is 3407 * empty, then this string is interpreted as either a 3408 * General_Category value alias, a Script value alias, a binary 3409 * property alias, or a special ID. Special IDs are matched 3410 * loosely and correspond to the following sets: 3411 * 3412 * "ANY" = [\\u0000-\\U0010FFFF], 3413 * "ASCII" = [\\u0000-\\u007F]. 3414 * 3415 * @param valueAlias a value alias, either short or long. The 3416 * name is matched loosely. See PropertyValueAliases.txt for 3417 * names and a description of loose matching. In addition to 3418 * aliases listed, numeric values and canonical combining classes 3419 * may be expressed numerically, e.g., ("nv", "0.5") or ("ccc", 3420 * "220"). The value string may also be empty. 3421 * 3422 * @return a reference to this set 3423 * 3424 * @stable ICU 2.4 3425 */ 3426 public UnicodeSet applyPropertyAlias(String propertyAlias, String valueAlias) { 3427 return applyPropertyAlias(propertyAlias, valueAlias, null); 3428 } 3429 3430 /** 3431 * Modifies this set to contain those code points which have the 3432 * given value for the given property. Prior contents of this 3433 * set are lost. 3434 * @param propertyAlias A string of the property alias. 3435 * @param valueAlias A string of the value alias. 3436 * @param symbols if not null, then symbols are first called to see if a property 3437 * is available. If true, then everything else is skipped. 3438 * @return this set 3439 * @stable ICU 3.2 3440 */ 3441 public UnicodeSet applyPropertyAlias(String propertyAlias, 3442 String valueAlias, SymbolTable symbols) { 3443 checkFrozen(); 3444 int p; 3445 int v; 3446 boolean invert = false; 3447 3448 if (symbols != null 3449 && (symbols instanceof XSymbolTable) 3450 && ((XSymbolTable)symbols).applyPropertyAlias(propertyAlias, valueAlias, this)) { 3451 return this; 3452 } 3453 3454 if (XSYMBOL_TABLE != null) { 3455 if (XSYMBOL_TABLE.applyPropertyAlias(propertyAlias, valueAlias, this)) { 3456 return this; 3457 } 3458 } 3459 3460 if (valueAlias.length() > 0) { 3461 p = UCharacter.getPropertyEnum(propertyAlias); 3462 3463 // Treat gc as gcm 3464 if (p == UProperty.GENERAL_CATEGORY) { 3465 p = UProperty.GENERAL_CATEGORY_MASK; 3466 } 3467 3468 if ((p >= UProperty.BINARY_START && p < UProperty.BINARY_LIMIT) || 3469 (p >= UProperty.INT_START && p < UProperty.INT_LIMIT) || 3470 (p >= UProperty.MASK_START && p < UProperty.MASK_LIMIT)) { 3471 try { 3472 v = UCharacter.getPropertyValueEnum(p, valueAlias); 3473 } catch (IllegalArgumentException e) { 3474 // Handle numeric CCC 3475 if (p == UProperty.CANONICAL_COMBINING_CLASS || 3476 p == UProperty.LEAD_CANONICAL_COMBINING_CLASS || 3477 p == UProperty.TRAIL_CANONICAL_COMBINING_CLASS) { 3478 v = Integer.parseInt(PatternProps.trimWhiteSpace(valueAlias)); 3479 // Anything between 0 and 255 is valid even if unused. 3480 if (v < 0 || v > 255) throw e; 3481 } else { 3482 throw e; 3483 } 3484 } 3485 } 3486 3487 else { 3488 switch (p) { 3489 case UProperty.NUMERIC_VALUE: 3490 { 3491 double value = Double.parseDouble(PatternProps.trimWhiteSpace(valueAlias)); 3492 applyFilter(new NumericValueFilter(value), UCharacterProperty.SRC_CHAR); 3493 return this; 3494 } 3495 case UProperty.NAME: 3496 { 3497 // Must munge name, since 3498 // UCharacter.charFromName() does not do 3499 // 'loose' matching. 3500 String buf = mungeCharName(valueAlias); 3501 int ch = UCharacter.getCharFromExtendedName(buf); 3502 if (ch == -1) { 3503 throw new IllegalArgumentException("Invalid character name"); 3504 } 3505 clear(); 3506 add_unchecked(ch); 3507 return this; 3508 } 3509 case UProperty.UNICODE_1_NAME: 3510 // ICU 49 deprecates the Unicode_1_Name property APIs. 3511 throw new IllegalArgumentException("Unicode_1_Name (na1) not supported"); 3512 case UProperty.AGE: 3513 { 3514 // Must munge name, since 3515 // VersionInfo.getInstance() does not do 3516 // 'loose' matching. 3517 VersionInfo version = VersionInfo.getInstance(mungeCharName(valueAlias)); 3518 applyFilter(new VersionFilter(version), UCharacterProperty.SRC_PROPSVEC); 3519 return this; 3520 } 3521 case UProperty.SCRIPT_EXTENSIONS: 3522 v = UCharacter.getPropertyValueEnum(UProperty.SCRIPT, valueAlias); 3523 // fall through to calling applyIntPropertyValue() 3524 break; 3525 default: 3526 // p is a non-binary, non-enumerated property that we 3527 // don't support (yet). 3528 throw new IllegalArgumentException("Unsupported property"); 3529 } 3530 } 3531 } 3532 3533 else { 3534 // valueAlias is empty. Interpret as General Category, Script, 3535 // Binary property, or ANY or ASCII. Upon success, p and v will 3536 // be set. 3537 UPropertyAliases pnames = UPropertyAliases.INSTANCE; 3538 p = UProperty.GENERAL_CATEGORY_MASK; 3539 v = pnames.getPropertyValueEnum(p, propertyAlias); 3540 if (v == UProperty.UNDEFINED) { 3541 p = UProperty.SCRIPT; 3542 v = pnames.getPropertyValueEnum(p, propertyAlias); 3543 if (v == UProperty.UNDEFINED) { 3544 p = pnames.getPropertyEnum(propertyAlias); 3545 if (p == UProperty.UNDEFINED) { 3546 p = -1; 3547 } 3548 if (p >= UProperty.BINARY_START && p < UProperty.BINARY_LIMIT) { 3549 v = 1; 3550 } else if (p == -1) { 3551 if (0 == UPropertyAliases.compare(ANY_ID, propertyAlias)) { 3552 set(MIN_VALUE, MAX_VALUE); 3553 return this; 3554 } else if (0 == UPropertyAliases.compare(ASCII_ID, propertyAlias)) { 3555 set(0, 0x7F); 3556 return this; 3557 } else if (0 == UPropertyAliases.compare(ASSIGNED, propertyAlias)) { 3558 // [:Assigned:]=[:^Cn:] 3559 p = UProperty.GENERAL_CATEGORY_MASK; 3560 v = (1<<UCharacter.UNASSIGNED); 3561 invert = true; 3562 } else { 3563 // Property name was never matched. 3564 throw new IllegalArgumentException("Invalid property alias: " + propertyAlias + "=" + valueAlias); 3565 } 3566 } else { 3567 // Valid propery name, but it isn't binary, so the value 3568 // must be supplied. 3569 throw new IllegalArgumentException("Missing property value"); 3570 } 3571 } 3572 } 3573 } 3574 3575 applyIntPropertyValue(p, v); 3576 if(invert) { 3577 complement(); 3578 } 3579 3580 return this; 3581 } 3582 3583 //---------------------------------------------------------------- 3584 // Property set patterns 3585 //---------------------------------------------------------------- 3586 3587 /** 3588 * Return true if the given position, in the given pattern, appears 3589 * to be the start of a property set pattern. 3590 */ 3591 private static boolean resemblesPropertyPattern(String pattern, int pos) { 3592 // Patterns are at least 5 characters long 3593 if ((pos+5) > pattern.length()) { 3594 return false; 3595 } 3596 3597 // Look for an opening [:, [:^, \p, or \P 3598 return pattern.regionMatches(pos, "[:", 0, 2) || 3599 pattern.regionMatches(true, pos, "\\p", 0, 2) || 3600 pattern.regionMatches(pos, "\\N", 0, 2); 3601 } 3602 3603 /** 3604 * Return true if the given iterator appears to point at a 3605 * property pattern. Regardless of the result, return with the 3606 * iterator unchanged. 3607 * @param chars iterator over the pattern characters. Upon return 3608 * it will be unchanged. 3609 * @param iterOpts RuleCharacterIterator options 3610 */ 3611 private static boolean resemblesPropertyPattern(RuleCharacterIterator chars, 3612 int iterOpts) { 3613 boolean result = false; 3614 iterOpts &= ~RuleCharacterIterator.PARSE_ESCAPES; 3615 Object pos = chars.getPos(null); 3616 int c = chars.next(iterOpts); 3617 if (c == '[' || c == '\\') { 3618 int d = chars.next(iterOpts & ~RuleCharacterIterator.SKIP_WHITESPACE); 3619 result = (c == '[') ? (d == ':') : 3620 (d == 'N' || d == 'p' || d == 'P'); 3621 } 3622 chars.setPos(pos); 3623 return result; 3624 } 3625 3626 /** 3627 * Parse the given property pattern at the given parse position. 3628 * @param symbols TODO 3629 */ 3630 private UnicodeSet applyPropertyPattern(String pattern, ParsePosition ppos, SymbolTable symbols) { 3631 int pos = ppos.getIndex(); 3632 3633 // On entry, ppos should point to one of the following locations: 3634 3635 // Minimum length is 5 characters, e.g. \p{L} 3636 if ((pos+5) > pattern.length()) { 3637 return null; 3638 } 3639 3640 boolean posix = false; // true for [:pat:], false for \p{pat} \P{pat} \N{pat} 3641 boolean isName = false; // true for \N{pat}, o/w false 3642 boolean invert = false; 3643 3644 // Look for an opening [:, [:^, \p, or \P 3645 if (pattern.regionMatches(pos, "[:", 0, 2)) { 3646 posix = true; 3647 pos = PatternProps.skipWhiteSpace(pattern, (pos+2)); 3648 if (pos < pattern.length() && pattern.charAt(pos) == '^') { 3649 ++pos; 3650 invert = true; 3651 } 3652 } else if (pattern.regionMatches(true, pos, "\\p", 0, 2) || 3653 pattern.regionMatches(pos, "\\N", 0, 2)) { 3654 char c = pattern.charAt(pos+1); 3655 invert = (c == 'P'); 3656 isName = (c == 'N'); 3657 pos = PatternProps.skipWhiteSpace(pattern, (pos+2)); 3658 if (pos == pattern.length() || pattern.charAt(pos++) != '{') { 3659 // Syntax error; "\p" or "\P" not followed by "{" 3660 return null; 3661 } 3662 } else { 3663 // Open delimiter not seen 3664 return null; 3665 } 3666 3667 // Look for the matching close delimiter, either :] or } 3668 int close = pattern.indexOf(posix ? ":]" : "}", pos); 3669 if (close < 0) { 3670 // Syntax error; close delimiter missing 3671 return null; 3672 } 3673 3674 // Look for an '=' sign. If this is present, we will parse a 3675 // medium \p{gc=Cf} or long \p{GeneralCategory=Format} 3676 // pattern. 3677 int equals = pattern.indexOf('=', pos); 3678 String propName, valueName; 3679 if (equals >= 0 && equals < close && !isName) { 3680 // Equals seen; parse medium/long pattern 3681 propName = pattern.substring(pos, equals); 3682 valueName = pattern.substring(equals+1, close); 3683 } 3684 3685 else { 3686 // Handle case where no '=' is seen, and \N{} 3687 propName = pattern.substring(pos, close); 3688 valueName = ""; 3689 3690 // Handle \N{name} 3691 if (isName) { 3692 // This is a little inefficient since it means we have to 3693 // parse "na" back to UProperty.NAME even though we already 3694 // know it's UProperty.NAME. If we refactor the API to 3695 // support args of (int, String) then we can remove 3696 // "na" and make this a little more efficient. 3697 valueName = propName; 3698 propName = "na"; 3699 } 3700 } 3701 3702 applyPropertyAlias(propName, valueName, symbols); 3703 3704 if (invert) { 3705 complement(); 3706 } 3707 3708 // Move to the limit position after the close delimiter 3709 ppos.setIndex(close + (posix ? 2 : 1)); 3710 3711 return this; 3712 } 3713 3714 /** 3715 * Parse a property pattern. 3716 * @param chars iterator over the pattern characters. Upon return 3717 * it will be advanced to the first character after the parsed 3718 * pattern, or the end of the iteration if all characters are 3719 * parsed. 3720 * @param rebuiltPat the pattern that was parsed, rebuilt or 3721 * copied from the input pattern, as appropriate. 3722 * @param symbols TODO 3723 */ 3724 private void applyPropertyPattern(RuleCharacterIterator chars, 3725 Appendable rebuiltPat, SymbolTable symbols) { 3726 String patStr = chars.lookahead(); 3727 ParsePosition pos = new ParsePosition(0); 3728 applyPropertyPattern(patStr, pos, symbols); 3729 if (pos.getIndex() == 0) { 3730 syntaxError(chars, "Invalid property pattern"); 3731 } 3732 chars.jumpahead(pos.getIndex()); 3733 append(rebuiltPat, patStr.substring(0, pos.getIndex())); 3734 } 3735 3736 //---------------------------------------------------------------- 3737 // Case folding API 3738 //---------------------------------------------------------------- 3739 3740 /** 3741 * Bitmask for constructor and applyPattern() indicating that 3742 * white space should be ignored. If set, ignore Unicode Pattern_White_Space characters, 3743 * unless they are quoted or escaped. This may be ORed together 3744 * with other selectors. 3745 * @stable ICU 3.8 3746 */ 3747 public static final int IGNORE_SPACE = 1; 3748 3749 /** 3750 * Bitmask for constructor, applyPattern(), and closeOver() 3751 * indicating letter case. This may be ORed together with other 3752 * selectors. 3753 * 3754 * Enable case insensitive matching. E.g., "[ab]" with this flag 3755 * will match 'a', 'A', 'b', and 'B'. "[^ab]" with this flag will 3756 * match all except 'a', 'A', 'b', and 'B'. This performs a full 3757 * closure over case mappings, e.g. U+017F for s. 3758 * 3759 * The resulting set is a superset of the input for the code points but 3760 * not for the strings. 3761 * It performs a case mapping closure of the code points and adds 3762 * full case folding strings for the code points, and reduces strings of 3763 * the original set to their full case folding equivalents. 3764 * 3765 * This is designed for case-insensitive matches, for example 3766 * in regular expressions. The full code point case closure allows checking of 3767 * an input character directly against the closure set. 3768 * Strings are matched by comparing the case-folded form from the closure 3769 * set with an incremental case folding of the string in question. 3770 * 3771 * The closure set will also contain single code points if the original 3772 * set contained case-equivalent strings (like U+00DF for "ss" or "Ss" etc.). 3773 * This is not necessary (that is, redundant) for the above matching method 3774 * but results in the same closure sets regardless of whether the original 3775 * set contained the code point or a string. 3776 * @stable ICU 3.8 3777 */ 3778 public static final int CASE = 2; 3779 3780 /** 3781 * Alias for UnicodeSet.CASE, for ease of porting from C++ where ICU4C 3782 * also has both USET_CASE and USET_CASE_INSENSITIVE (see uset.h). 3783 * @see #CASE 3784 * @stable ICU 3.4 3785 */ 3786 public static final int CASE_INSENSITIVE = 2; 3787 3788 /** 3789 * Bitmask for constructor, applyPattern(), and closeOver() 3790 * indicating letter case. This may be ORed together with other 3791 * selectors. 3792 * 3793 * Enable case insensitive matching. E.g., "[ab]" with this flag 3794 * will match 'a', 'A', 'b', and 'B'. "[^ab]" with this flag will 3795 * match all except 'a', 'A', 'b', and 'B'. This adds the lower-, 3796 * title-, and uppercase mappings as well as the case folding 3797 * of each existing element in the set. 3798 * @stable ICU 3.4 3799 */ 3800 public static final int ADD_CASE_MAPPINGS = 4; 3801 3802 // add the result of a full case mapping to the set 3803 // use str as a temporary string to avoid constructing one 3804 private static final void addCaseMapping(UnicodeSet set, int result, StringBuilder full) { 3805 if(result >= 0) { 3806 if(result > UCaseProps.MAX_STRING_LENGTH) { 3807 // add a single-code point case mapping 3808 set.add(result); 3809 } else { 3810 // add a string case mapping from full with length result 3811 set.add(full.toString()); 3812 full.setLength(0); 3813 } 3814 } 3815 // result < 0: the code point mapped to itself, no need to add it 3816 // see UCaseProps 3817 } 3818 3819 /** 3820 * Close this set over the given attribute. For the attribute 3821 * CASE, the result is to modify this set so that: 3822 * 3823 * 1. For each character or string 'a' in this set, all strings 3824 * 'b' such that foldCase(a) == foldCase(b) are added to this set. 3825 * (For most 'a' that are single characters, 'b' will have 3826 * b.length() == 1.) 3827 * 3828 * 2. For each string 'e' in the resulting set, if e != 3829 * foldCase(e), 'e' will be removed. 3830 * 3831 * Example: [aq\u00DF{Bc}{bC}{Fi}] => [aAqQ\u00DF\uFB01{ss}{bc}{fi}] 3832 * 3833 * (Here foldCase(x) refers to the operation 3834 * UCharacter.foldCase(x, true), and a == b actually denotes 3835 * a.equals(b), not pointer comparison.) 3836 * 3837 * @param attribute bitmask for attributes to close over. 3838 * Currently only the CASE bit is supported. Any undefined bits 3839 * are ignored. 3840 * @return a reference to this set. 3841 * @stable ICU 3.8 3842 */ 3843 public UnicodeSet closeOver(int attribute) { 3844 checkFrozen(); 3845 if ((attribute & (CASE | ADD_CASE_MAPPINGS)) != 0) { 3846 UCaseProps csp = UCaseProps.INSTANCE; 3847 UnicodeSet foldSet = new UnicodeSet(this); 3848 ULocale root = ULocale.ROOT; 3849 3850 // start with input set to guarantee inclusion 3851 // CASE: remove strings because the strings will actually be reduced (folded); 3852 // therefore, start with no strings and add only those needed 3853 if((attribute & CASE) != 0) { 3854 foldSet.strings.clear(); 3855 } 3856 3857 int n = getRangeCount(); 3858 int result; 3859 StringBuilder full = new StringBuilder(); 3860 3861 for (int i=0; i<n; ++i) { 3862 int start = getRangeStart(i); 3863 int end = getRangeEnd(i); 3864 3865 if((attribute & CASE) != 0) { 3866 // full case closure 3867 for (int cp=start; cp<=end; ++cp) { 3868 csp.addCaseClosure(cp, foldSet); 3869 } 3870 } else { 3871 // add case mappings 3872 // (does not add long s for regular s, or Kelvin for k, for example) 3873 for (int cp=start; cp<=end; ++cp) { 3874 result = csp.toFullLower(cp, null, full, UCaseProps.LOC_ROOT); 3875 addCaseMapping(foldSet, result, full); 3876 3877 result = csp.toFullTitle(cp, null, full, UCaseProps.LOC_ROOT); 3878 addCaseMapping(foldSet, result, full); 3879 3880 result = csp.toFullUpper(cp, null, full, UCaseProps.LOC_ROOT); 3881 addCaseMapping(foldSet, result, full); 3882 3883 result = csp.toFullFolding(cp, full, 0); 3884 addCaseMapping(foldSet, result, full); 3885 } 3886 } 3887 } 3888 if (!strings.isEmpty()) { 3889 if ((attribute & CASE) != 0) { 3890 for (String s : strings) { 3891 String str = UCharacter.foldCase(s, 0); 3892 if(!csp.addStringCaseClosure(str, foldSet)) { 3893 foldSet.add(str); // does not map to code points: add the folded string itself 3894 } 3895 } 3896 } else { 3897 BreakIterator bi = BreakIterator.getWordInstance(root); 3898 for (String str : strings) { 3899 // TODO: call lower-level functions 3900 foldSet.add(UCharacter.toLowerCase(root, str)); 3901 foldSet.add(UCharacter.toTitleCase(root, str, bi)); 3902 foldSet.add(UCharacter.toUpperCase(root, str)); 3903 foldSet.add(UCharacter.foldCase(str, 0)); 3904 } 3905 } 3906 } 3907 set(foldSet); 3908 } 3909 return this; 3910 } 3911 3912 /** 3913 * Internal class for customizing UnicodeSet parsing of properties. 3914 * TODO: extend to allow customizing of codepoint ranges 3915 * @draft ICU3.8 (retain) 3916 * @provisional This API might change or be removed in a future release. 3917 * @author medavis 3918 */ 3919 abstract public static class XSymbolTable implements SymbolTable { 3920 /** 3921 * Default constructor 3922 * @draft ICU3.8 (retain) 3923 * @provisional This API might change or be removed in a future release. 3924 */ 3925 public XSymbolTable(){} 3926 /** 3927 * Supplies default implementation for SymbolTable (no action). 3928 * @draft ICU3.8 (retain) 3929 * @provisional This API might change or be removed in a future release. 3930 */ 3931 @Override 3932 public UnicodeMatcher lookupMatcher(int i) { 3933 return null; 3934 } 3935 3936 /** 3937 * Override the interpretation of the sequence [:propertyName=propertyValue:] (and its negated and Perl-style 3938 * variant). The propertyName and propertyValue may be existing Unicode aliases, or may not be. 3939 * <p> 3940 * This routine will be called whenever the parsing of a UnicodeSet pattern finds such a 3941 * propertyName+propertyValue combination. 3942 * 3943 * @param propertyName 3944 * the name of the property 3945 * @param propertyValue 3946 * the name of the property value 3947 * @param result UnicodeSet value to change 3948 * a set to which the characters having the propertyName+propertyValue are to be added. 3949 * @return returns true if the propertyName+propertyValue combination is to be overridden, and the characters 3950 * with that property have been added to the UnicodeSet, and returns false if the 3951 * propertyName+propertyValue combination is not recognized (in which case result is unaltered). 3952 * @draft ICU3.8 (retain) 3953 * @provisional This API might change or be removed in a future release. 3954 */ 3955 public boolean applyPropertyAlias(String propertyName, String propertyValue, UnicodeSet result) { 3956 return false; 3957 } 3958 /** 3959 * Supplies default implementation for SymbolTable (no action). 3960 * @draft ICU3.8 (retain) 3961 * @provisional This API might change or be removed in a future release. 3962 */ 3963 @Override 3964 public char[] lookup(String s) { 3965 return null; 3966 } 3967 /** 3968 * Supplies default implementation for SymbolTable (no action). 3969 * @draft ICU3.8 (retain) 3970 * @provisional This API might change or be removed in a future release. 3971 */ 3972 @Override 3973 public String parseReference(String text, ParsePosition pos, int limit) { 3974 return null; 3975 } 3976 } 3977 3978 /** 3979 * Is this frozen, according to the Freezable interface? 3980 * 3981 * @return value 3982 * @stable ICU 3.8 3983 */ 3984 @Override 3985 public boolean isFrozen() { 3986 return (bmpSet != null || stringSpan != null); 3987 } 3988 3989 /** 3990 * Freeze this class, according to the Freezable interface. 3991 * 3992 * @return this 3993 * @stable ICU 4.4 3994 */ 3995 @Override 3996 public UnicodeSet freeze() { 3997 if (!isFrozen()) { 3998 // Do most of what compact() does before freezing because 3999 // compact() will not work when the set is frozen. 4000 // Small modification: Don't shrink if the savings would be tiny (<=GROW_EXTRA). 4001 4002 // Delete buffer first to defragment memory less. 4003 buffer = null; 4004 if (list.length > (len + GROW_EXTRA)) { 4005 // Make the capacity equal to len or 1. 4006 // We don't want to realloc of 0 size. 4007 int capacity = (len == 0) ? 1 : len; 4008 int[] oldList = list; 4009 list = new int[capacity]; 4010 for (int i = capacity; i-- > 0;) { 4011 list[i] = oldList[i]; 4012 } 4013 } 4014 4015 // Optimize contains() and span() and similar functions. 4016 if (!strings.isEmpty()) { 4017 stringSpan = new UnicodeSetStringSpan(this, new ArrayList<String>(strings), UnicodeSetStringSpan.ALL); 4018 } 4019 if (stringSpan == null || !stringSpan.needsStringSpanUTF16()) { 4020 // Optimize for code point spans. 4021 // There are no strings, or 4022 // all strings are irrelevant for span() etc. because 4023 // all of each string's code points are contained in this set. 4024 // However, fully contained strings are relevant for spanAndCount(), 4025 // so we create both objects. 4026 bmpSet = new BMPSet(list, len); 4027 } 4028 } 4029 return this; 4030 } 4031 4032 /** 4033 * Span a string using this UnicodeSet. 4034 * <p>To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}. 4035 * @param s The string to be spanned 4036 * @param spanCondition The span condition 4037 * @return the length of the span 4038 * @stable ICU 4.4 4039 */ 4040 public int span(CharSequence s, SpanCondition spanCondition) { 4041 return span(s, 0, spanCondition); 4042 } 4043 4044 /** 4045 * Span a string using this UnicodeSet. 4046 * If the start index is less than 0, span will start from 0. 4047 * If the start index is greater than the string length, span returns the string length. 4048 * <p>To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}. 4049 * @param s The string to be spanned 4050 * @param start The start index that the span begins 4051 * @param spanCondition The span condition 4052 * @return the string index which ends the span (i.e. exclusive) 4053 * @stable ICU 4.4 4054 */ 4055 public int span(CharSequence s, int start, SpanCondition spanCondition) { 4056 int end = s.length(); 4057 if (start < 0) { 4058 start = 0; 4059 } else if (start >= end) { 4060 return end; 4061 } 4062 if (bmpSet != null) { 4063 // Frozen set without strings, or no string is relevant for span(). 4064 return bmpSet.span(s, start, spanCondition, null); 4065 } 4066 if (stringSpan != null) { 4067 return stringSpan.span(s, start, spanCondition); 4068 } else if (!strings.isEmpty()) { 4069 int which = spanCondition == SpanCondition.NOT_CONTAINED ? UnicodeSetStringSpan.FWD_UTF16_NOT_CONTAINED 4070 : UnicodeSetStringSpan.FWD_UTF16_CONTAINED; 4071 UnicodeSetStringSpan strSpan = new UnicodeSetStringSpan(this, new ArrayList<String>(strings), which); 4072 if (strSpan.needsStringSpanUTF16()) { 4073 return strSpan.span(s, start, spanCondition); 4074 } 4075 } 4076 4077 return spanCodePointsAndCount(s, start, spanCondition, null); 4078 } 4079 4080 /** 4081 * Same as span() but also counts the smallest number of set elements on any path across the span. 4082 * <p>To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}. 4083 * @param outCount An output-only object (must not be null) for returning the count. 4084 * @return the limit (exclusive end) of the span 4085 * @internal 4086 * @deprecated This API is ICU internal only. 4087 */ 4088 @Deprecated 4089 public int spanAndCount(CharSequence s, int start, SpanCondition spanCondition, OutputInt outCount) { 4090 if (outCount == null) { 4091 throw new IllegalArgumentException("outCount must not be null"); 4092 } 4093 int end = s.length(); 4094 if (start < 0) { 4095 start = 0; 4096 } else if (start >= end) { 4097 return end; 4098 } 4099 if (stringSpan != null) { 4100 // We might also have bmpSet != null, 4101 // but fully-contained strings are relevant for counting elements. 4102 return stringSpan.spanAndCount(s, start, spanCondition, outCount); 4103 } else if (bmpSet != null) { 4104 return bmpSet.span(s, start, spanCondition, outCount); 4105 } else if (!strings.isEmpty()) { 4106 int which = spanCondition == SpanCondition.NOT_CONTAINED ? UnicodeSetStringSpan.FWD_UTF16_NOT_CONTAINED 4107 : UnicodeSetStringSpan.FWD_UTF16_CONTAINED; 4108 which |= UnicodeSetStringSpan.WITH_COUNT; 4109 UnicodeSetStringSpan strSpan = new UnicodeSetStringSpan(this, new ArrayList<String>(strings), which); 4110 return strSpan.spanAndCount(s, start, spanCondition, outCount); 4111 } 4112 4113 return spanCodePointsAndCount(s, start, spanCondition, outCount); 4114 } 4115 4116 private int spanCodePointsAndCount(CharSequence s, int start, 4117 SpanCondition spanCondition, OutputInt outCount) { 4118 // Pin to 0/1 values. 4119 boolean spanContained = (spanCondition != SpanCondition.NOT_CONTAINED); 4120 4121 int c; 4122 int next = start; 4123 int length = s.length(); 4124 int count = 0; 4125 do { 4126 c = Character.codePointAt(s, next); 4127 if (spanContained != contains(c)) { 4128 break; 4129 } 4130 ++count; 4131 next += Character.charCount(c); 4132 } while (next < length); 4133 if (outCount != null) { outCount.value = count; } 4134 return next; 4135 } 4136 4137 /** 4138 * Span a string backwards (from the end) using this UnicodeSet. 4139 * <p>To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}. 4140 * @param s The string to be spanned 4141 * @param spanCondition The span condition 4142 * @return The string index which starts the span (i.e. inclusive). 4143 * @stable ICU 4.4 4144 */ 4145 public int spanBack(CharSequence s, SpanCondition spanCondition) { 4146 return spanBack(s, s.length(), spanCondition); 4147 } 4148 4149 /** 4150 * Span a string backwards (from the fromIndex) using this UnicodeSet. 4151 * If the fromIndex is less than 0, spanBack will return 0. 4152 * If fromIndex is greater than the string length, spanBack will start from the string length. 4153 * <p>To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}. 4154 * @param s The string to be spanned 4155 * @param fromIndex The index of the char (exclusive) that the string should be spanned backwards 4156 * @param spanCondition The span condition 4157 * @return The string index which starts the span (i.e. inclusive). 4158 * @stable ICU 4.4 4159 */ 4160 public int spanBack(CharSequence s, int fromIndex, SpanCondition spanCondition) { 4161 if (fromIndex <= 0) { 4162 return 0; 4163 } 4164 if (fromIndex > s.length()) { 4165 fromIndex = s.length(); 4166 } 4167 if (bmpSet != null) { 4168 // Frozen set without strings, or no string is relevant for spanBack(). 4169 return bmpSet.spanBack(s, fromIndex, spanCondition); 4170 } 4171 if (stringSpan != null) { 4172 return stringSpan.spanBack(s, fromIndex, spanCondition); 4173 } else if (!strings.isEmpty()) { 4174 int which = (spanCondition == SpanCondition.NOT_CONTAINED) 4175 ? UnicodeSetStringSpan.BACK_UTF16_NOT_CONTAINED 4176 : UnicodeSetStringSpan.BACK_UTF16_CONTAINED; 4177 UnicodeSetStringSpan strSpan = new UnicodeSetStringSpan(this, new ArrayList<String>(strings), which); 4178 if (strSpan.needsStringSpanUTF16()) { 4179 return strSpan.spanBack(s, fromIndex, spanCondition); 4180 } 4181 } 4182 4183 // Pin to 0/1 values. 4184 boolean spanContained = (spanCondition != SpanCondition.NOT_CONTAINED); 4185 4186 int c; 4187 int prev = fromIndex; 4188 do { 4189 c = Character.codePointBefore(s, prev); 4190 if (spanContained != contains(c)) { 4191 break; 4192 } 4193 prev -= Character.charCount(c); 4194 } while (prev > 0); 4195 return prev; 4196 } 4197 4198 /** 4199 * Clone a thawed version of this class, according to the Freezable interface. 4200 * @return the clone, not frozen 4201 * @stable ICU 4.4 4202 */ 4203 @Override 4204 public UnicodeSet cloneAsThawed() { 4205 UnicodeSet result = new UnicodeSet(this); 4206 assert !result.isFrozen(); 4207 return result; 4208 } 4209 4210 // internal function 4211 private void checkFrozen() { 4212 if (isFrozen()) { 4213 throw new UnsupportedOperationException("Attempt to modify frozen object"); 4214 } 4215 } 4216 4217 // ************************ 4218 // Additional methods for integration with Generics and Collections 4219 // ************************ 4220 4221 /** 4222 * A struct-like class used for iteration through ranges, for faster iteration than by String. 4223 * Read about the restrictions on usage in {@link UnicodeSet#ranges()}. 4224 * 4225 * @stable ICU 54 4226 */ 4227 public static class EntryRange { 4228 /** 4229 * The starting code point of the range. 4230 * 4231 * @stable ICU 54 4232 */ 4233 public int codepoint; 4234 /** 4235 * The ending code point of the range 4236 * 4237 * @stable ICU 54 4238 */ 4239 public int codepointEnd; 4240 4241 EntryRange() { 4242 } 4243 4244 /** 4245 * {@inheritDoc} 4246 * 4247 * @stable ICU 54 4248 */ 4249 @Override 4250 public String toString() { 4251 StringBuilder b = new StringBuilder(); 4252 return ( 4253 codepoint == codepointEnd ? _appendToPat(b, codepoint, false) 4254 : _appendToPat(_appendToPat(b, codepoint, false).append('-'), codepointEnd, false)) 4255 .toString(); 4256 } 4257 } 4258 4259 /** 4260 * Provide for faster iteration than by String. Returns an Iterable/Iterator over ranges of code points. 4261 * The UnicodeSet must not be altered during the iteration. 4262 * The EntryRange instance is the same each time; the contents are just reset. 4263 * 4264 * <p><b>Warning: </b>To iterate over the full contents, you have to also iterate over the strings. 4265 * 4266 * <p><b>Warning: </b>For speed, UnicodeSet iteration does not check for concurrent modification. 4267 * Do not alter the UnicodeSet while iterating. 4268 * 4269 * <pre> 4270 * // Sample code 4271 * for (EntryRange range : us1.ranges()) { 4272 * // do something with code points between range.codepoint and range.codepointEnd; 4273 * } 4274 * for (String s : us1.strings()) { 4275 * // do something with each string; 4276 * } 4277 * </pre> 4278 * 4279 * @stable ICU 54 4280 */ 4281 public Iterable<EntryRange> ranges() { 4282 return new EntryRangeIterable(); 4283 } 4284 4285 private class EntryRangeIterable implements Iterable<EntryRange> { 4286 @Override 4287 public Iterator<EntryRange> iterator() { 4288 return new EntryRangeIterator(); 4289 } 4290 } 4291 4292 private class EntryRangeIterator implements Iterator<EntryRange> { 4293 int pos; 4294 EntryRange result = new EntryRange(); 4295 4296 @Override 4297 public boolean hasNext() { 4298 return pos < len-1; 4299 } 4300 @Override 4301 public EntryRange next() { 4302 if (pos < len-1) { 4303 result.codepoint = list[pos++]; 4304 result.codepointEnd = list[pos++]-1; 4305 } else { 4306 throw new NoSuchElementException(); 4307 } 4308 return result; 4309 } 4310 @Override 4311 public void remove() { 4312 throw new UnsupportedOperationException(); 4313 } 4314 } 4315 4316 4317 /** 4318 * Returns a string iterator. Uses the same order of iteration as {@link UnicodeSetIterator}. 4319 * <p><b>Warning: </b>For speed, UnicodeSet iteration does not check for concurrent modification. 4320 * Do not alter the UnicodeSet while iterating. 4321 * @see java.util.Set#iterator() 4322 * @stable ICU 4.4 4323 */ 4324 @Override 4325 public Iterator<String> iterator() { 4326 return new UnicodeSetIterator2(this); 4327 } 4328 4329 // Cover for string iteration. 4330 private static class UnicodeSetIterator2 implements Iterator<String> { 4331 // Invariants: 4332 // sourceList != null then sourceList[item] is a valid character 4333 // sourceList == null then delegates to stringIterator 4334 private int[] sourceList; 4335 private int len; 4336 private int item; 4337 private int current; 4338 private int limit; 4339 private TreeSet<String> sourceStrings; 4340 private Iterator<String> stringIterator; 4341 private char[] buffer; 4342 4343 UnicodeSetIterator2(UnicodeSet source) { 4344 // set according to invariants 4345 len = source.len - 1; 4346 if (len > 0) { 4347 sourceStrings = source.strings; 4348 sourceList = source.list; 4349 current = sourceList[item++]; 4350 limit = sourceList[item++]; 4351 } else { 4352 stringIterator = source.strings.iterator(); 4353 sourceList = null; 4354 } 4355 } 4356 4357 /* (non-Javadoc) 4358 * @see java.util.Iterator#hasNext() 4359 */ 4360 @Override 4361 public boolean hasNext() { 4362 return sourceList != null || stringIterator.hasNext(); 4363 } 4364 4365 /* (non-Javadoc) 4366 * @see java.util.Iterator#next() 4367 */ 4368 @Override 4369 public String next() { 4370 if (sourceList == null) { 4371 return stringIterator.next(); 4372 } 4373 int codepoint = current++; 4374 // we have the codepoint we need, but we may need to adjust the state 4375 if (current >= limit) { 4376 if (item >= len) { 4377 stringIterator = sourceStrings.iterator(); 4378 sourceList = null; 4379 } else { 4380 current = sourceList[item++]; 4381 limit = sourceList[item++]; 4382 } 4383 } 4384 // Now return. Single code point is easy 4385 if (codepoint <= 0xFFFF) { 4386 return String.valueOf((char)codepoint); 4387 } 4388 // But Java lacks a valueOfCodePoint, so we handle ourselves for speed 4389 // allocate a buffer the first time, to make conversion faster. 4390 if (buffer == null) { 4391 buffer = new char[2]; 4392 } 4393 // compute ourselves, to save tests and calls 4394 int offset = codepoint - Character.MIN_SUPPLEMENTARY_CODE_POINT; 4395 buffer[0] = (char)((offset >>> 10) + Character.MIN_HIGH_SURROGATE); 4396 buffer[1] = (char)((offset & 0x3ff) + Character.MIN_LOW_SURROGATE); 4397 return String.valueOf(buffer); 4398 } 4399 4400 /* (non-Javadoc) 4401 * @see java.util.Iterator#remove() 4402 */ 4403 @Override 4404 public void remove() { 4405 throw new UnsupportedOperationException(); 4406 } 4407 } 4408 4409 /** 4410 * @see #containsAll(com.ibm.icu.text.UnicodeSet) 4411 * @stable ICU 4.4 4412 */ 4413 public <T extends CharSequence> boolean containsAll(Iterable<T> collection) { 4414 for (T o : collection) { 4415 if (!contains(o)) { 4416 return false; 4417 } 4418 } 4419 return true; 4420 } 4421 4422 /** 4423 * @see #containsNone(com.ibm.icu.text.UnicodeSet) 4424 * @stable ICU 4.4 4425 */ 4426 public <T extends CharSequence> boolean containsNone(Iterable<T> collection) { 4427 for (T o : collection) { 4428 if (contains(o)) { 4429 return false; 4430 } 4431 } 4432 return true; 4433 } 4434 4435 /** 4436 * @see #containsAll(com.ibm.icu.text.UnicodeSet) 4437 * @stable ICU 4.4 4438 */ 4439 public final <T extends CharSequence> boolean containsSome(Iterable<T> collection) { 4440 return !containsNone(collection); 4441 } 4442 4443 /** 4444 * @see #addAll(com.ibm.icu.text.UnicodeSet) 4445 * @stable ICU 4.4 4446 */ 4447 @SuppressWarnings("unchecked") // See ticket #11395, this is safe. 4448 public <T extends CharSequence> UnicodeSet addAll(T... collection) { 4449 checkFrozen(); 4450 for (T str : collection) { 4451 add(str); 4452 } 4453 return this; 4454 } 4455 4456 4457 /** 4458 * @see #removeAll(com.ibm.icu.text.UnicodeSet) 4459 * @stable ICU 4.4 4460 */ 4461 public <T extends CharSequence> UnicodeSet removeAll(Iterable<T> collection) { 4462 checkFrozen(); 4463 for (T o : collection) { 4464 remove(o); 4465 } 4466 return this; 4467 } 4468 4469 /** 4470 * @see #retainAll(com.ibm.icu.text.UnicodeSet) 4471 * @stable ICU 4.4 4472 */ 4473 public <T extends CharSequence> UnicodeSet retainAll(Iterable<T> collection) { 4474 checkFrozen(); 4475 // TODO optimize 4476 UnicodeSet toRetain = new UnicodeSet(); 4477 toRetain.addAll(collection); 4478 retainAll(toRetain); 4479 return this; 4480 } 4481 4482 /** 4483 * Comparison style enums used by {@link UnicodeSet#compareTo(UnicodeSet, ComparisonStyle)}. 4484 * @stable ICU 4.4 4485 */ 4486 public enum ComparisonStyle { 4487 /** 4488 * @stable ICU 4.4 4489 */ 4490 SHORTER_FIRST, 4491 /** 4492 * @stable ICU 4.4 4493 */ 4494 LEXICOGRAPHIC, 4495 /** 4496 * @stable ICU 4.4 4497 */ 4498 LONGER_FIRST 4499 } 4500 4501 /** 4502 * Compares UnicodeSets, where shorter come first, and otherwise lexigraphically 4503 * (according to the comparison of the first characters that differ). 4504 * @see java.lang.Comparable#compareTo(java.lang.Object) 4505 * @stable ICU 4.4 4506 */ 4507 @Override 4508 public int compareTo(UnicodeSet o) { 4509 return compareTo(o, ComparisonStyle.SHORTER_FIRST); 4510 } 4511 /** 4512 * Compares UnicodeSets, in three different ways. 4513 * @see java.lang.Comparable#compareTo(java.lang.Object) 4514 * @stable ICU 4.4 4515 */ 4516 public int compareTo(UnicodeSet o, ComparisonStyle style) { 4517 if (style != ComparisonStyle.LEXICOGRAPHIC) { 4518 int diff = size() - o.size(); 4519 if (diff != 0) { 4520 return (diff < 0) == (style == ComparisonStyle.SHORTER_FIRST) ? -1 : 1; 4521 } 4522 } 4523 int result; 4524 for (int i = 0; ; ++i) { 4525 if (0 != (result = list[i] - o.list[i])) { 4526 // if either list ran out, compare to the last string 4527 if (list[i] == HIGH) { 4528 if (strings.isEmpty()) return 1; 4529 String item = strings.first(); 4530 return compare(item, o.list[i]); 4531 } 4532 if (o.list[i] == HIGH) { 4533 if (o.strings.isEmpty()) return -1; 4534 String item = o.strings.first(); 4535 int compareResult = compare(item, list[i]); 4536 return compareResult > 0 ? -1 : compareResult < 0 ? 1 : 0; // Reverse the order. 4537 } 4538 // otherwise return the result if even index, or the reversal if not 4539 return (i & 1) == 0 ? result : -result; 4540 } 4541 if (list[i] == HIGH) { 4542 break; 4543 } 4544 } 4545 return compare(strings, o.strings); 4546 } 4547 4548 /** 4549 * @stable ICU 4.4 4550 */ 4551 public int compareTo(Iterable<String> other) { 4552 return compare(this, other); 4553 } 4554 4555 /** 4556 * Utility to compare a string to a code point. 4557 * Same results as turning the code point into a string (with the [ugly] new StringBuilder().appendCodePoint(codepoint).toString()) 4558 * and comparing, but much faster (no object creation). 4559 * Actually, there is one difference; a null compares as less. 4560 * Note that this (=String) order is UTF-16 order -- *not* code point order. 4561 * @stable ICU 4.4 4562 */ 4563 4564 public static int compare(CharSequence string, int codePoint) { 4565 return CharSequences.compare(string, codePoint); 4566 } 4567 4568 /** 4569 * Utility to compare a string to a code point. 4570 * Same results as turning the code point into a string and comparing, but much faster (no object creation). 4571 * Actually, there is one difference; a null compares as less. 4572 * Note that this (=String) order is UTF-16 order -- *not* code point order. 4573 * @stable ICU 4.4 4574 */ 4575 public static int compare(int codePoint, CharSequence string) { 4576 return -CharSequences.compare(string, codePoint); 4577 } 4578 4579 4580 /** 4581 * Utility to compare two iterables. Warning: the ordering in iterables is important. For Collections that are ordered, 4582 * like Lists, that is expected. However, Sets in Java violate Leibniz's law when it comes to iteration. 4583 * That means that sets can't be compared directly with this method, unless they are TreeSets without 4584 * (or with the same) comparator. Unfortunately, it is impossible to reliably detect in Java whether subclass of 4585 * Collection satisfies the right criteria, so it is left to the user to avoid those circumstances. 4586 * @stable ICU 4.4 4587 */ 4588 public static <T extends Comparable<T>> int compare(Iterable<T> collection1, Iterable<T> collection2) { 4589 return compare(collection1.iterator(), collection2.iterator()); 4590 } 4591 4592 /** 4593 * Utility to compare two iterators. Warning: the ordering in iterables is important. For Collections that are ordered, 4594 * like Lists, that is expected. However, Sets in Java violate Leibniz's law when it comes to iteration. 4595 * That means that sets can't be compared directly with this method, unless they are TreeSets without 4596 * (or with the same) comparator. Unfortunately, it is impossible to reliably detect in Java whether subclass of 4597 * Collection satisfies the right criteria, so it is left to the user to avoid those circumstances. 4598 * @internal 4599 * @deprecated This API is ICU internal only. 4600 */ 4601 @Deprecated 4602 public static <T extends Comparable<T>> int compare(Iterator<T> first, Iterator<T> other) { 4603 while (true) { 4604 if (!first.hasNext()) { 4605 return other.hasNext() ? -1 : 0; 4606 } else if (!other.hasNext()) { 4607 return 1; 4608 } 4609 T item1 = first.next(); 4610 T item2 = other.next(); 4611 int result = item1.compareTo(item2); 4612 if (result != 0) { 4613 return result; 4614 } 4615 } 4616 } 4617 4618 4619 /** 4620 * Utility to compare two collections, optionally by size, and then lexicographically. 4621 * @stable ICU 4.4 4622 */ 4623 public static <T extends Comparable<T>> int compare(Collection<T> collection1, Collection<T> collection2, ComparisonStyle style) { 4624 if (style != ComparisonStyle.LEXICOGRAPHIC) { 4625 int diff = collection1.size() - collection2.size(); 4626 if (diff != 0) { 4627 return (diff < 0) == (style == ComparisonStyle.SHORTER_FIRST) ? -1 : 1; 4628 } 4629 } 4630 return compare(collection1, collection2); 4631 } 4632 4633 /** 4634 * Utility for adding the contents of an iterable to a collection. 4635 * @stable ICU 4.4 4636 */ 4637 public static <T, U extends Collection<T>> U addAllTo(Iterable<T> source, U target) { 4638 for (T item : source) { 4639 target.add(item); 4640 } 4641 return target; 4642 } 4643 4644 /** 4645 * Utility for adding the contents of an iterable to a collection. 4646 * @stable ICU 4.4 4647 */ 4648 public static <T> T[] addAllTo(Iterable<T> source, T[] target) { 4649 int i = 0; 4650 for (T item : source) { 4651 target[i++] = item; 4652 } 4653 return target; 4654 } 4655 4656 /** 4657 * For iterating through the strings in the set. Example: 4658 * <pre> 4659 * for (String key : myUnicodeSet.strings()) { 4660 * doSomethingWith(key); 4661 * } 4662 * </pre> 4663 * @stable ICU 4.4 4664 */ 4665 public Collection<String> strings() { 4666 return Collections.unmodifiableSortedSet(strings); 4667 } 4668 4669 /** 4670 * Return the value of the first code point, if the string is exactly one code point. Otherwise return Integer.MAX_VALUE. 4671 * @internal 4672 * @deprecated This API is ICU internal only. 4673 */ 4674 @Deprecated 4675 public static int getSingleCodePoint(CharSequence s) { 4676 return CharSequences.getSingleCodePoint(s); 4677 } 4678 4679 /** 4680 * Simplify the ranges in a Unicode set by merging any ranges that are only separated by characters in the dontCare set. 4681 * For example, the ranges: \\u2E80-\\u2E99\\u2E9B-\\u2EF3\\u2F00-\\u2FD5\\u2FF0-\\u2FFB\\u3000-\\u303E change to \\u2E80-\\u303E 4682 * if the dontCare set includes unassigned characters (for a particular version of Unicode). 4683 * @param dontCare Set with the don't-care characters for spanning 4684 * @return the input set, modified 4685 * @internal 4686 * @deprecated This API is ICU internal only. 4687 */ 4688 @Deprecated 4689 public UnicodeSet addBridges(UnicodeSet dontCare) { 4690 UnicodeSet notInInput = new UnicodeSet(this).complement(); 4691 for (UnicodeSetIterator it = new UnicodeSetIterator(notInInput); it.nextRange();) { 4692 if (it.codepoint != 0 && it.codepoint != UnicodeSetIterator.IS_STRING && it.codepointEnd != 0x10FFFF && dontCare.contains(it.codepoint,it.codepointEnd)) { 4693 add(it.codepoint,it.codepointEnd); 4694 } 4695 } 4696 return this; 4697 } 4698 4699 /** 4700 * Find the first index at or after fromIndex where the UnicodeSet matches at that index. 4701 * If findNot is true, then reverse the sense of the match: find the first place where the UnicodeSet doesn't match. 4702 * If there is no match, length is returned. 4703 * @internal 4704 * @deprecated This API is ICU internal only. Use span instead. 4705 */ 4706 @Deprecated 4707 public int findIn(CharSequence value, int fromIndex, boolean findNot) { 4708 //TODO add strings, optimize, using ICU4C algorithms 4709 int cp; 4710 for (; fromIndex < value.length(); fromIndex += UTF16.getCharCount(cp)) { 4711 cp = UTF16.charAt(value, fromIndex); 4712 if (contains(cp) != findNot) { 4713 break; 4714 } 4715 } 4716 return fromIndex; 4717 } 4718 4719 /** 4720 * Find the last index before fromIndex where the UnicodeSet matches at that index. 4721 * If findNot is true, then reverse the sense of the match: find the last place where the UnicodeSet doesn't match. 4722 * If there is no match, -1 is returned. 4723 * BEFORE index is not in the UnicodeSet. 4724 * @internal 4725 * @deprecated This API is ICU internal only. Use spanBack instead. 4726 */ 4727 @Deprecated 4728 public int findLastIn(CharSequence value, int fromIndex, boolean findNot) { 4729 //TODO add strings, optimize, using ICU4C algorithms 4730 int cp; 4731 fromIndex -= 1; 4732 for (; fromIndex >= 0; fromIndex -= UTF16.getCharCount(cp)) { 4733 cp = UTF16.charAt(value, fromIndex); 4734 if (contains(cp) != findNot) { 4735 break; 4736 } 4737 } 4738 return fromIndex < 0 ? -1 : fromIndex; 4739 } 4740 4741 /** 4742 * Strips code points from source. If matches is true, script all that match <i>this</i>. If matches is false, then strip all that <i>don't</i> match. 4743 * @param source The source of the CharSequence to strip from. 4744 * @param matches A boolean to either strip all that matches or don't match with the current UnicodeSet object. 4745 * @return The string after it has been stripped. 4746 * @internal 4747 * @deprecated This API is ICU internal only. Use replaceFrom. 4748 */ 4749 @Deprecated 4750 public String stripFrom(CharSequence source, boolean matches) { 4751 StringBuilder result = new StringBuilder(); 4752 for (int pos = 0; pos < source.length();) { 4753 int inside = findIn(source, pos, !matches); 4754 result.append(source.subSequence(pos, inside)); 4755 pos = findIn(source, inside, matches); // get next start 4756 } 4757 return result.toString(); 4758 } 4759 4760 /** 4761 * Argument values for whether span() and similar functions continue while the current character is contained vs. 4762 * not contained in the set. 4763 * <p> 4764 * The functionality is straightforward for sets with only single code points, without strings (which is the common 4765 * case): 4766 * <ul> 4767 * <li>CONTAINED and SIMPLE work the same. 4768 * <li>CONTAINED and SIMPLE are inverses of NOT_CONTAINED. 4769 * <li>span() and spanBack() partition any string the 4770 * same way when alternating between span(NOT_CONTAINED) and span(either "contained" condition). 4771 * <li>Using a 4772 * complemented (inverted) set and the opposite span conditions yields the same results. 4773 * </ul> 4774 * When a set contains multi-code point strings, then these statements may not be true, depending on the strings in 4775 * the set (for example, whether they overlap with each other) and the string that is processed. For a set with 4776 * strings: 4777 * <ul> 4778 * <li>The complement of the set contains the opposite set of code points, but the same set of strings. 4779 * Therefore, complementing both the set and the span conditions may yield different results. 4780 * <li>When starting spans 4781 * at different positions in a string (span(s, ...) vs. span(s+1, ...)) the ends of the spans may be different 4782 * because a set string may start before the later position. 4783 * <li>span(SIMPLE) may be shorter than 4784 * span(CONTAINED) because it will not recursively try all possible paths. For example, with a set which 4785 * contains the three strings "xy", "xya" and "ax", span("xyax", CONTAINED) will return 4 but span("xyax", 4786 * SIMPLE) will return 3. span(SIMPLE) will never be longer than span(CONTAINED). 4787 * <li>With either "contained" condition, span() and spanBack() may partition a string in different ways. For example, 4788 * with a set which contains the two strings "ab" and "ba", and when processing the string "aba", span() will yield 4789 * contained/not-contained boundaries of { 0, 2, 3 } while spanBack() will yield boundaries of { 0, 1, 3 }. 4790 * </ul> 4791 * Note: If it is important to get the same boundaries whether iterating forward or backward through a string, then 4792 * either only span() should be used and the boundaries cached for backward operation, or an ICU BreakIterator could 4793 * be used. 4794 * <p> 4795 * Note: Unpaired surrogates are treated like surrogate code points. Similarly, set strings match only on code point 4796 * boundaries, never in the middle of a surrogate pair. 4797 * 4798 * @stable ICU 4.4 4799 */ 4800 public enum SpanCondition { 4801 /** 4802 * Continues a span() while there is no set element at the current position. 4803 * Increments by one code point at a time. 4804 * Stops before the first set element (character or string). 4805 * (For code points only, this is like while contains(current)==false). 4806 * <p> 4807 * When span() returns, the substring between where it started and the position it returned consists only of 4808 * characters that are not in the set, and none of its strings overlap with the span. 4809 * 4810 * @stable ICU 4.4 4811 */ 4812 NOT_CONTAINED, 4813 4814 /** 4815 * Spans the longest substring that is a concatenation of set elements (characters or strings). 4816 * (For characters only, this is like while contains(current)==true). 4817 * <p> 4818 * When span() returns, the substring between where it started and the position it returned consists only of set 4819 * elements (characters or strings) that are in the set. 4820 * <p> 4821 * If a set contains strings, then the span will be the longest substring for which there 4822 * exists at least one non-overlapping concatenation of set elements (characters or strings). 4823 * This is equivalent to a POSIX regular expression for <code>(OR of each set element)*</code>. 4824 * (Java/ICU/Perl regex stops at the first match of an OR.) 4825 * 4826 * @stable ICU 4.4 4827 */ 4828 CONTAINED, 4829 4830 /** 4831 * Continues a span() while there is a set element at the current position. 4832 * Increments by the longest matching element at each position. 4833 * (For characters only, this is like while contains(current)==true). 4834 * <p> 4835 * When span() returns, the substring between where it started and the position it returned consists only of set 4836 * elements (characters or strings) that are in the set. 4837 * <p> 4838 * If a set only contains single characters, then this is the same as CONTAINED. 4839 * <p> 4840 * If a set contains strings, then the span will be the longest substring with a match at each position with the 4841 * longest single set element (character or string). 4842 * <p> 4843 * Use this span condition together with other longest-match algorithms, such as ICU converters 4844 * (ucnv_getUnicodeSet()). 4845 * 4846 * @stable ICU 4.4 4847 */ 4848 SIMPLE, 4849 4850 /** 4851 * One more than the last span condition. 4852 * 4853 * @stable ICU 4.4 4854 */ 4855 CONDITION_COUNT 4856 } 4857 4858 /** 4859 * Get the default symbol table. Null means ordinary processing. For internal use only. 4860 * @return the symbol table 4861 * @internal 4862 * @deprecated This API is ICU internal only. 4863 */ 4864 @Deprecated 4865 public static XSymbolTable getDefaultXSymbolTable() { 4866 return XSYMBOL_TABLE; 4867 } 4868 4869 /** 4870 * Set the default symbol table. Null means ordinary processing. For internal use only. Will affect all subsequent parsing 4871 * of UnicodeSets. 4872 * <p> 4873 * WARNING: If this function is used with a UnicodeProperty, and the 4874 * Unassigned characters (gc=Cn) are different than in ICU other than in ICU, you MUST call 4875 * {@code UnicodeProperty.ResetCacheProperties} afterwards. If you then call {@code UnicodeSet.setDefaultXSymbolTable} 4876 * with null to clear the value, you MUST also call {@code UnicodeProperty.ResetCacheProperties}. 4877 * 4878 * @param xSymbolTable the new default symbol table. 4879 * @internal 4880 * @deprecated This API is ICU internal only. 4881 */ 4882 @Deprecated 4883 public static void setDefaultXSymbolTable(XSymbolTable xSymbolTable) { 4884 INCLUSIONS = null; // If the properties override inclusions, these have to be regenerated. 4885 XSYMBOL_TABLE = xSymbolTable; 4886 } 4887 } 4888 //eof 4889