1 /* 2 * Copyright (c) 2000, 2015, Oracle and/or its affiliates. All rights reserved. 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4 * 5 * This code is free software; you can redistribute it and/or modify it 6 * under the terms of the GNU General Public License version 2 only, as 7 * published by the Free Software Foundation. Oracle designates this 8 * particular file as subject to the "Classpath" exception as provided 9 * by Oracle in the LICENSE file that accompanied this code. 10 * 11 * This code is distributed in the hope that it will be useful, but WITHOUT 12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 14 * version 2 for more details (a copy is included in the LICENSE file that 15 * accompanied this code). 16 * 17 * You should have received a copy of the GNU General Public License version 18 * 2 along with this work; if not, write to the Free Software Foundation, 19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. 20 * 21 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA 22 * or visit www.oracle.com if you need additional information or have any 23 * questions. 24 */ 25 26 package java.awt.font; 27 28 import java.io.IOException; 29 import java.io.ObjectOutputStream; 30 import java.util.Arrays; 31 import java.util.Comparator; 32 import java.util.EnumSet; 33 import java.util.Set; 34 import sun.misc.SharedSecrets; 35 36 /** 37 * The <code>NumericShaper</code> class is used to convert Latin-1 (European) 38 * digits to other Unicode decimal digits. Users of this class will 39 * primarily be people who wish to present data using 40 * national digit shapes, but find it more convenient to represent the 41 * data internally using Latin-1 (European) digits. This does not 42 * interpret the deprecated numeric shape selector character (U+206E). 43 * <p> 44 * Instances of <code>NumericShaper</code> are typically applied 45 * as attributes to text with the 46 * {@link TextAttribute#NUMERIC_SHAPING NUMERIC_SHAPING} attribute 47 * of the <code>TextAttribute</code> class. 48 * For example, this code snippet causes a <code>TextLayout</code> to 49 * shape European digits to Arabic in an Arabic context:<br> 50 * <blockquote><pre> 51 * Map map = new HashMap(); 52 * map.put(TextAttribute.NUMERIC_SHAPING, 53 * NumericShaper.getContextualShaper(NumericShaper.ARABIC)); 54 * FontRenderContext frc = ...; 55 * TextLayout layout = new TextLayout(text, map, frc); 56 * layout.draw(g2d, x, y); 57 * </pre></blockquote> 58 * <br> 59 * It is also possible to perform numeric shaping explicitly using instances 60 * of <code>NumericShaper</code>, as this code snippet demonstrates:<br> 61 * <blockquote><pre> 62 * char[] text = ...; 63 * // shape all EUROPEAN digits (except zero) to ARABIC digits 64 * NumericShaper shaper = NumericShaper.getShaper(NumericShaper.ARABIC); 65 * shaper.shape(text, start, count); 66 * 67 * // shape European digits to ARABIC digits if preceding text is Arabic, or 68 * // shape European digits to TAMIL digits if preceding text is Tamil, or 69 * // leave European digits alone if there is no preceding text, or 70 * // preceding text is neither Arabic nor Tamil 71 * NumericShaper shaper = 72 * NumericShaper.getContextualShaper(NumericShaper.ARABIC | 73 * NumericShaper.TAMIL, 74 * NumericShaper.EUROPEAN); 75 * shaper.shape(text, start, count); 76 * </pre></blockquote> 77 * 78 * <p><b>Bit mask- and enum-based Unicode ranges</b></p> 79 * 80 * <p>This class supports two different programming interfaces to 81 * represent Unicode ranges for script-specific digits: bit 82 * mask-based ones, such as {@link #ARABIC NumericShaper.ARABIC}, and 83 * enum-based ones, such as {@link NumericShaper.Range#ARABIC}. 84 * Multiple ranges can be specified by ORing bit mask-based constants, 85 * such as: 86 * <blockquote><pre> 87 * NumericShaper.ARABIC | NumericShaper.TAMIL 88 * </pre></blockquote> 89 * or creating a {@code Set} with the {@link NumericShaper.Range} 90 * constants, such as: 91 * <blockquote><pre> 92 * EnumSet.of(NumericShaper.Range.ARABIC, NumericShaper.Range.TAMIL) 93 * </pre></blockquote> 94 * The enum-based ranges are a super set of the bit mask-based ones. 95 * 96 * <p>If the two interfaces are mixed (including serialization), 97 * Unicode range values are mapped to their counterparts where such 98 * mapping is possible, such as {@code NumericShaper.Range.ARABIC} 99 * from/to {@code NumericShaper.ARABIC}. If any unmappable range 100 * values are specified, such as {@code NumericShaper.Range.BALINESE}, 101 * those ranges are ignored. 102 * 103 * <p><b>Decimal Digits Precedence</b></p> 104 * 105 * <p>A Unicode range may have more than one set of decimal digits. If 106 * multiple decimal digits sets are specified for the same Unicode 107 * range, one of the sets will take precedence as follows. 108 * 109 * <table border=1 cellspacing=3 cellpadding=0 summary="NumericShaper constants precedence."> 110 * <tr> 111 * <th class="TableHeadingColor">Unicode Range</th> 112 * <th class="TableHeadingColor"><code>NumericShaper</code> Constants</th> 113 * <th class="TableHeadingColor">Precedence</th> 114 * </tr> 115 * <tr> 116 * <td rowspan="2">Arabic</td> 117 * <td>{@link NumericShaper#ARABIC NumericShaper.ARABIC}<br> 118 * {@link NumericShaper#EASTERN_ARABIC NumericShaper.EASTERN_ARABIC}</td> 119 * <td>{@link NumericShaper#EASTERN_ARABIC NumericShaper.EASTERN_ARABIC}</td> 120 * </tr> 121 * <tr> 122 * <td>{@link NumericShaper.Range#ARABIC}<br> 123 * {@link NumericShaper.Range#EASTERN_ARABIC}</td> 124 * <td>{@link NumericShaper.Range#EASTERN_ARABIC}</td> 125 * </tr> 126 * <tr> 127 * <td>Tai Tham</td> 128 * <td>{@link NumericShaper.Range#TAI_THAM_HORA}<br> 129 * {@link NumericShaper.Range#TAI_THAM_THAM}</td> 130 * <td>{@link NumericShaper.Range#TAI_THAM_THAM}</td> 131 * </tr> 132 * </table> 133 * 134 * @since 1.4 135 */ 136 137 public final class NumericShaper implements java.io.Serializable { 138 139 // For access from java.text.Bidi 140 static { 141 if (SharedSecrets.getJavaAWTFontAccess() == null) { 142 SharedSecrets.setJavaAWTFontAccess(new JavaAWTFontAccessImpl()); 143 } 144 } 145 146 /** 147 * A {@code NumericShaper.Range} represents a Unicode range of a 148 * script having its own decimal digits. For example, the {@link 149 * NumericShaper.Range#THAI} range has the Thai digits, THAI DIGIT 150 * ZERO (U+0E50) to THAI DIGIT NINE (U+0E59). 151 * 152 * <p>The <code>Range</code> enum replaces the traditional bit 153 * mask-based values (e.g., {@link NumericShaper#ARABIC}), and 154 * supports more Unicode ranges than the bit mask-based ones. For 155 * example, the following code using the bit mask: 156 * <blockquote><pre> 157 * NumericShaper.getContextualShaper(NumericShaper.ARABIC | 158 * NumericShaper.TAMIL, 159 * NumericShaper.EUROPEAN); 160 * </pre></blockquote> 161 * can be written using this enum as: 162 * <blockquote><pre> 163 * NumericShaper.getContextualShaper(EnumSet.of( 164 * NumericShaper.Range.ARABIC, 165 * NumericShaper.Range.TAMIL), 166 * NumericShaper.Range.EUROPEAN); 167 * </pre></blockquote> 168 * 169 * @since 1.7 170 */ 171 public static enum Range { 172 // The order of EUROPEAN to MOGOLIAN must be consistent 173 // with the bitmask-based constants. 174 /** 175 * The Latin (European) range with the Latin (ASCII) digits. 176 */ 177 EUROPEAN ('\u0030', '\u0000', '\u0300'), 178 /** 179 * The Arabic range with the Arabic-Indic digits. 180 */ 181 ARABIC ('\u0660', '\u0600', '\u0780'), 182 /** 183 * The Arabic range with the Eastern Arabic-Indic digits. 184 */ 185 EASTERN_ARABIC ('\u06f0', '\u0600', '\u0780'), 186 /** 187 * The Devanagari range with the Devanagari digits. 188 */ 189 DEVANAGARI ('\u0966', '\u0900', '\u0980'), 190 /** 191 * The Bengali range with the Bengali digits. 192 */ 193 BENGALI ('\u09e6', '\u0980', '\u0a00'), 194 /** 195 * The Gurmukhi range with the Gurmukhi digits. 196 */ 197 GURMUKHI ('\u0a66', '\u0a00', '\u0a80'), 198 /** 199 * The Gujarati range with the Gujarati digits. 200 */ 201 GUJARATI ('\u0ae6', '\u0b00', '\u0b80'), 202 /** 203 * The Oriya range with the Oriya digits. 204 */ 205 ORIYA ('\u0b66', '\u0b00', '\u0b80'), 206 /** 207 * The Tamil range with the Tamil digits. 208 */ 209 TAMIL ('\u0be6', '\u0b80', '\u0c00'), 210 /** 211 * The Telugu range with the Telugu digits. 212 */ 213 TELUGU ('\u0c66', '\u0c00', '\u0c80'), 214 /** 215 * The Kannada range with the Kannada digits. 216 */ 217 KANNADA ('\u0ce6', '\u0c80', '\u0d00'), 218 /** 219 * The Malayalam range with the Malayalam digits. 220 */ 221 MALAYALAM ('\u0d66', '\u0d00', '\u0d80'), 222 /** 223 * The Thai range with the Thai digits. 224 */ 225 THAI ('\u0e50', '\u0e00', '\u0e80'), 226 /** 227 * The Lao range with the Lao digits. 228 */ 229 LAO ('\u0ed0', '\u0e80', '\u0f00'), 230 /** 231 * The Tibetan range with the Tibetan digits. 232 */ 233 TIBETAN ('\u0f20', '\u0f00', '\u1000'), 234 /** 235 * The Myanmar range with the Myanmar digits. 236 */ 237 MYANMAR ('\u1040', '\u1000', '\u1080'), 238 /** 239 * The Ethiopic range with the Ethiopic digits. Ethiopic 240 * does not have a decimal digit 0 so Latin (European) 0 is 241 * used. 242 */ 243 ETHIOPIC ('\u1369', '\u1200', '\u1380') { 244 @Override 245 char getNumericBase() { return 1; } 246 }, 247 /** 248 * The Khmer range with the Khmer digits. 249 */ 250 KHMER ('\u17e0', '\u1780', '\u1800'), 251 /** 252 * The Mongolian range with the Mongolian digits. 253 */ 254 MONGOLIAN ('\u1810', '\u1800', '\u1900'), 255 // The order of EUROPEAN to MOGOLIAN must be consistent 256 // with the bitmask-based constants. 257 258 /** 259 * The N'Ko range with the N'Ko digits. 260 */ 261 NKO ('\u07c0', '\u07c0', '\u0800'), 262 /** 263 * The Myanmar range with the Myanmar Shan digits. 264 */ 265 MYANMAR_SHAN ('\u1090', '\u1000', '\u10a0'), 266 /** 267 * The Limbu range with the Limbu digits. 268 */ 269 LIMBU ('\u1946', '\u1900', '\u1950'), 270 /** 271 * The New Tai Lue range with the New Tai Lue digits. 272 */ 273 NEW_TAI_LUE ('\u19d0', '\u1980', '\u19e0'), 274 /** 275 * The Balinese range with the Balinese digits. 276 */ 277 BALINESE ('\u1b50', '\u1b00', '\u1b80'), 278 /** 279 * The Sundanese range with the Sundanese digits. 280 */ 281 SUNDANESE ('\u1bb0', '\u1b80', '\u1bc0'), 282 /** 283 * The Lepcha range with the Lepcha digits. 284 */ 285 LEPCHA ('\u1c40', '\u1c00', '\u1c50'), 286 /** 287 * The Ol Chiki range with the Ol Chiki digits. 288 */ 289 OL_CHIKI ('\u1c50', '\u1c50', '\u1c80'), 290 /** 291 * The Vai range with the Vai digits. 292 */ 293 VAI ('\ua620', '\ua500', '\ua640'), 294 /** 295 * The Saurashtra range with the Saurashtra digits. 296 */ 297 SAURASHTRA ('\ua8d0', '\ua880', '\ua8e0'), 298 /** 299 * The Kayah Li range with the Kayah Li digits. 300 */ 301 KAYAH_LI ('\ua900', '\ua900', '\ua930'), 302 /** 303 * The Cham range with the Cham digits. 304 */ 305 CHAM ('\uaa50', '\uaa00', '\uaa60'), 306 /** 307 * The Tai Tham Hora range with the Tai Tham Hora digits. 308 */ 309 TAI_THAM_HORA ('\u1a80', '\u1a20', '\u1ab0'), 310 /** 311 * The Tai Tham Tham range with the Tai Tham Tham digits. 312 */ 313 TAI_THAM_THAM ('\u1a90', '\u1a20', '\u1ab0'), 314 /** 315 * The Javanese range with the Javanese digits. 316 */ 317 JAVANESE ('\ua9d0', '\ua980', '\ua9e0'), 318 /** 319 * The Meetei Mayek range with the Meetei Mayek digits. 320 */ 321 MEETEI_MAYEK ('\uabf0', '\uabc0', '\uac00'), 322 /** 323 * The Sinhala range with the Sinhala digits. 324 * @since 1.9 325 */ 326 SINHALA ('\u0de6', '\u0d80', '\u0e00'), 327 /** 328 * The Myanmar Extended-B range with the Myanmar Tai Laing digits. 329 * @since 1.9 330 */ 331 MYANMAR_TAI_LAING ('\ua9f0', '\ua9e0', '\uaa00'); 332 333 private static int toRangeIndex(Range script) { 334 int index = script.ordinal(); 335 return index < NUM_KEYS ? index : -1; 336 } 337 338 private static Range indexToRange(int index) { 339 return index < NUM_KEYS ? Range.values()[index] : null; 340 } 341 342 private static int toRangeMask(Set<Range> ranges) { 343 int m = 0; 344 for (Range range : ranges) { 345 int index = range.ordinal(); 346 if (index < NUM_KEYS) { 347 m |= 1 << index; 348 } 349 } 350 return m; 351 } 352 353 private static Set<Range> maskToRangeSet(int mask) { 354 Set<Range> set = EnumSet.noneOf(Range.class); 355 Range[] a = Range.values(); 356 for (int i = 0; i < NUM_KEYS; i++) { 357 if ((mask & (1 << i)) != 0) { 358 set.add(a[i]); 359 } 360 } 361 return set; 362 } 363 364 // base character of range digits 365 private final int base; 366 // Unicode range 367 private final int start, // inclusive 368 end; // exclusive 369 370 private Range(int base, int start, int end) { 371 this.base = base - ('0' + getNumericBase()); 372 this.start = start; 373 this.end = end; 374 } 375 376 private int getDigitBase() { 377 return base; 378 } 379 380 char getNumericBase() { 381 return 0; 382 } 383 384 private boolean inRange(int c) { 385 return start <= c && c < end; 386 } 387 } 388 389 /** index of context for contextual shaping - values range from 0 to 18 */ 390 private int key; 391 392 /** flag indicating whether to shape contextually (high bit) and which 393 * digit ranges to shape (bits 0-18) 394 */ 395 private int mask; 396 397 /** 398 * The context {@code Range} for contextual shaping or the {@code 399 * Range} for non-contextual shaping. {@code null} for the bit 400 * mask-based API. 401 * 402 * @since 1.7 403 */ 404 private Range shapingRange; 405 406 /** 407 * {@code Set<Range>} indicating which Unicode ranges to 408 * shape. {@code null} for the bit mask-based API. 409 */ 410 private transient Set<Range> rangeSet; 411 412 /** 413 * rangeSet.toArray() value. Sorted by Range.base when the number 414 * of elements is greater then BSEARCH_THRESHOLD. 415 */ 416 private transient Range[] rangeArray; 417 418 /** 419 * If more than BSEARCH_THRESHOLD ranges are specified, binary search is used. 420 */ 421 private static final int BSEARCH_THRESHOLD = 3; 422 423 private static final long serialVersionUID = -8022764705923730308L; 424 425 /** Identifies the Latin-1 (European) and extended range, and 426 * Latin-1 (European) decimal base. 427 */ 428 public static final int EUROPEAN = 1<<0; 429 430 /** Identifies the ARABIC range and decimal base. */ 431 public static final int ARABIC = 1<<1; 432 433 /** Identifies the ARABIC range and ARABIC_EXTENDED decimal base. */ 434 public static final int EASTERN_ARABIC = 1<<2; 435 436 /** Identifies the DEVANAGARI range and decimal base. */ 437 public static final int DEVANAGARI = 1<<3; 438 439 /** Identifies the BENGALI range and decimal base. */ 440 public static final int BENGALI = 1<<4; 441 442 /** Identifies the GURMUKHI range and decimal base. */ 443 public static final int GURMUKHI = 1<<5; 444 445 /** Identifies the GUJARATI range and decimal base. */ 446 public static final int GUJARATI = 1<<6; 447 448 /** Identifies the ORIYA range and decimal base. */ 449 public static final int ORIYA = 1<<7; 450 451 /** Identifies the TAMIL range and decimal base. */ 452 // TAMIL DIGIT ZERO was added in Unicode 4.1 453 public static final int TAMIL = 1<<8; 454 455 /** Identifies the TELUGU range and decimal base. */ 456 public static final int TELUGU = 1<<9; 457 458 /** Identifies the KANNADA range and decimal base. */ 459 public static final int KANNADA = 1<<10; 460 461 /** Identifies the MALAYALAM range and decimal base. */ 462 public static final int MALAYALAM = 1<<11; 463 464 /** Identifies the THAI range and decimal base. */ 465 public static final int THAI = 1<<12; 466 467 /** Identifies the LAO range and decimal base. */ 468 public static final int LAO = 1<<13; 469 470 /** Identifies the TIBETAN range and decimal base. */ 471 public static final int TIBETAN = 1<<14; 472 473 /** Identifies the MYANMAR range and decimal base. */ 474 public static final int MYANMAR = 1<<15; 475 476 /** Identifies the ETHIOPIC range and decimal base. */ 477 public static final int ETHIOPIC = 1<<16; 478 479 /** Identifies the KHMER range and decimal base. */ 480 public static final int KHMER = 1<<17; 481 482 /** Identifies the MONGOLIAN range and decimal base. */ 483 public static final int MONGOLIAN = 1<<18; 484 485 /** Identifies all ranges, for full contextual shaping. 486 * 487 * <p>This constant specifies all of the bit mask-based 488 * ranges. Use {@code EnumSet.allOf(NumericShaper.Range.class)} to 489 * specify all of the enum-based ranges. 490 */ 491 public static final int ALL_RANGES = 0x0007ffff; 492 493 private static final int EUROPEAN_KEY = 0; 494 private static final int ARABIC_KEY = 1; 495 private static final int EASTERN_ARABIC_KEY = 2; 496 private static final int DEVANAGARI_KEY = 3; 497 private static final int BENGALI_KEY = 4; 498 private static final int GURMUKHI_KEY = 5; 499 private static final int GUJARATI_KEY = 6; 500 private static final int ORIYA_KEY = 7; 501 private static final int TAMIL_KEY = 8; 502 private static final int TELUGU_KEY = 9; 503 private static final int KANNADA_KEY = 10; 504 private static final int MALAYALAM_KEY = 11; 505 private static final int THAI_KEY = 12; 506 private static final int LAO_KEY = 13; 507 private static final int TIBETAN_KEY = 14; 508 private static final int MYANMAR_KEY = 15; 509 private static final int ETHIOPIC_KEY = 16; 510 private static final int KHMER_KEY = 17; 511 private static final int MONGOLIAN_KEY = 18; 512 513 private static final int NUM_KEYS = MONGOLIAN_KEY + 1; // fixed 514 515 private static final int CONTEXTUAL_MASK = 1<<31; 516 517 private static final char[] bases = { 518 '\u0030' - '\u0030', // EUROPEAN 519 '\u0660' - '\u0030', // ARABIC-INDIC 520 '\u06f0' - '\u0030', // EXTENDED ARABIC-INDIC (EASTERN_ARABIC) 521 '\u0966' - '\u0030', // DEVANAGARI 522 '\u09e6' - '\u0030', // BENGALI 523 '\u0a66' - '\u0030', // GURMUKHI 524 '\u0ae6' - '\u0030', // GUJARATI 525 '\u0b66' - '\u0030', // ORIYA 526 '\u0be6' - '\u0030', // TAMIL - zero was added in Unicode 4.1 527 '\u0c66' - '\u0030', // TELUGU 528 '\u0ce6' - '\u0030', // KANNADA 529 '\u0d66' - '\u0030', // MALAYALAM 530 '\u0e50' - '\u0030', // THAI 531 '\u0ed0' - '\u0030', // LAO 532 '\u0f20' - '\u0030', // TIBETAN 533 '\u1040' - '\u0030', // MYANMAR 534 '\u1369' - '\u0031', // ETHIOPIC - no zero 535 '\u17e0' - '\u0030', // KHMER 536 '\u1810' - '\u0030', // MONGOLIAN 537 }; 538 539 // some ranges adjoin or overlap, rethink if we want to do a binary search on this 540 541 private static final char[] contexts = { 542 '\u0000', '\u0300', // 'EUROPEAN' (really latin-1 and extended) 543 '\u0600', '\u0780', // ARABIC 544 '\u0600', '\u0780', // EASTERN_ARABIC -- note overlap with arabic 545 '\u0900', '\u0980', // DEVANAGARI 546 '\u0980', '\u0a00', // BENGALI 547 '\u0a00', '\u0a80', // GURMUKHI 548 '\u0a80', '\u0b00', // GUJARATI 549 '\u0b00', '\u0b80', // ORIYA 550 '\u0b80', '\u0c00', // TAMIL 551 '\u0c00', '\u0c80', // TELUGU 552 '\u0c80', '\u0d00', // KANNADA 553 '\u0d00', '\u0d80', // MALAYALAM 554 '\u0e00', '\u0e80', // THAI 555 '\u0e80', '\u0f00', // LAO 556 '\u0f00', '\u1000', // TIBETAN 557 '\u1000', '\u1080', // MYANMAR 558 '\u1200', '\u1380', // ETHIOPIC - note missing zero 559 '\u1780', '\u1800', // KHMER 560 '\u1800', '\u1900', // MONGOLIAN 561 '\uffff', 562 }; 563 564 // assume most characters are near each other so probing the cache is infrequent, 565 // and a linear probe is ok. 566 567 private static int ctCache = 0; 568 private static int ctCacheLimit = contexts.length - 2; 569 570 // warning, synchronize access to this as it modifies state 571 private static int getContextKey(char c) { 572 if (c < contexts[ctCache]) { 573 while (ctCache > 0 && c < contexts[ctCache]) --ctCache; 574 } else if (c >= contexts[ctCache + 1]) { 575 while (ctCache < ctCacheLimit && c >= contexts[ctCache + 1]) ++ctCache; 576 } 577 578 // if we're not in a known range, then return EUROPEAN as the range key 579 return (ctCache & 0x1) == 0 ? (ctCache / 2) : EUROPEAN_KEY; 580 } 581 582 // cache for the NumericShaper.Range version 583 private transient volatile Range currentRange = Range.EUROPEAN; 584 585 private Range rangeForCodePoint(final int codepoint) { 586 if (currentRange.inRange(codepoint)) { 587 return currentRange; 588 } 589 590 final Range[] ranges = rangeArray; 591 if (ranges.length > BSEARCH_THRESHOLD) { 592 int lo = 0; 593 int hi = ranges.length - 1; 594 while (lo <= hi) { 595 int mid = (lo + hi) / 2; 596 Range range = ranges[mid]; 597 if (codepoint < range.start) { 598 hi = mid - 1; 599 } else if (codepoint >= range.end) { 600 lo = mid + 1; 601 } else { 602 currentRange = range; 603 return range; 604 } 605 } 606 } else { 607 for (int i = 0; i < ranges.length; i++) { 608 if (ranges[i].inRange(codepoint)) { 609 return ranges[i]; 610 } 611 } 612 } 613 return Range.EUROPEAN; 614 } 615 616 /* 617 * A range table of strong directional characters (types L, R, AL). 618 * Even (left) indexes are starts of ranges of non-strong-directional (or undefined) 619 * characters, odd (right) indexes are starts of ranges of strong directional 620 * characters. 621 */ 622 private static int[] strongTable = { 623 0x0000, 0x0041, 624 0x005b, 0x0061, 625 0x007b, 0x00aa, 626 0x00ab, 0x00b5, 627 0x00b6, 0x00ba, 628 0x00bb, 0x00c0, 629 0x00d7, 0x00d8, 630 0x00f7, 0x00f8, 631 0x02b9, 0x02bb, 632 0x02c2, 0x02d0, 633 0x02d2, 0x02e0, 634 0x02e5, 0x02ee, 635 0x02ef, 0x0370, 636 0x0374, 0x0376, 637 0x0378, 0x037a, 638 0x037e, 0x037f, 639 0x0380, 0x0386, 640 0x0387, 0x0388, 641 0x038b, 0x038c, 642 0x038d, 0x038e, 643 0x03a2, 0x03a3, 644 0x03f6, 0x03f7, 645 0x0483, 0x048a, 646 0x0530, 0x0531, 647 0x0557, 0x0559, 648 0x0560, 0x0561, 649 0x0588, 0x0589, 650 0x058a, 0x0590, 651 0x0591, 0x05be, 652 0x05bf, 0x05c0, 653 0x05c1, 0x05c3, 654 0x05c4, 0x05c6, 655 0x05c7, 0x05c8, 656 0x0600, 0x0608, 657 0x0609, 0x060b, 658 0x060c, 0x060d, 659 0x060e, 0x061b, 660 0x064b, 0x066d, 661 0x0670, 0x0671, 662 0x06d6, 0x06e5, 663 0x06e7, 0x06ee, 664 0x06f0, 0x06fa, 665 0x0711, 0x0712, 666 0x0730, 0x074b, 667 0x07a6, 0x07b1, 668 0x07eb, 0x07f4, 669 0x07f6, 0x07fa, 670 0x0816, 0x081a, 671 0x081b, 0x0824, 672 0x0825, 0x0828, 673 0x0829, 0x082e, 674 0x0859, 0x085c, 675 0x08e4, 0x0903, 676 0x093a, 0x093b, 677 0x093c, 0x093d, 678 0x0941, 0x0949, 679 0x094d, 0x094e, 680 0x0951, 0x0958, 681 0x0962, 0x0964, 682 0x0981, 0x0982, 683 0x0984, 0x0985, 684 0x098d, 0x098f, 685 0x0991, 0x0993, 686 0x09a9, 0x09aa, 687 0x09b1, 0x09b2, 688 0x09b3, 0x09b6, 689 0x09ba, 0x09bd, 690 0x09c1, 0x09c7, 691 0x09c9, 0x09cb, 692 0x09cd, 0x09ce, 693 0x09cf, 0x09d7, 694 0x09d8, 0x09dc, 695 0x09de, 0x09df, 696 0x09e2, 0x09e6, 697 0x09f2, 0x09f4, 698 0x09fb, 0x0a03, 699 0x0a04, 0x0a05, 700 0x0a0b, 0x0a0f, 701 0x0a11, 0x0a13, 702 0x0a29, 0x0a2a, 703 0x0a31, 0x0a32, 704 0x0a34, 0x0a35, 705 0x0a37, 0x0a38, 706 0x0a3a, 0x0a3e, 707 0x0a41, 0x0a59, 708 0x0a5d, 0x0a5e, 709 0x0a5f, 0x0a66, 710 0x0a70, 0x0a72, 711 0x0a75, 0x0a83, 712 0x0a84, 0x0a85, 713 0x0a8e, 0x0a8f, 714 0x0a92, 0x0a93, 715 0x0aa9, 0x0aaa, 716 0x0ab1, 0x0ab2, 717 0x0ab4, 0x0ab5, 718 0x0aba, 0x0abd, 719 0x0ac1, 0x0ac9, 720 0x0aca, 0x0acb, 721 0x0acd, 0x0ad0, 722 0x0ad1, 0x0ae0, 723 0x0ae2, 0x0ae6, 724 0x0af1, 0x0b02, 725 0x0b04, 0x0b05, 726 0x0b0d, 0x0b0f, 727 0x0b11, 0x0b13, 728 0x0b29, 0x0b2a, 729 0x0b31, 0x0b32, 730 0x0b34, 0x0b35, 731 0x0b3a, 0x0b3d, 732 0x0b3f, 0x0b40, 733 0x0b41, 0x0b47, 734 0x0b49, 0x0b4b, 735 0x0b4d, 0x0b57, 736 0x0b58, 0x0b5c, 737 0x0b5e, 0x0b5f, 738 0x0b62, 0x0b66, 739 0x0b78, 0x0b83, 740 0x0b84, 0x0b85, 741 0x0b8b, 0x0b8e, 742 0x0b91, 0x0b92, 743 0x0b96, 0x0b99, 744 0x0b9b, 0x0b9c, 745 0x0b9d, 0x0b9e, 746 0x0ba0, 0x0ba3, 747 0x0ba5, 0x0ba8, 748 0x0bab, 0x0bae, 749 0x0bba, 0x0bbe, 750 0x0bc0, 0x0bc1, 751 0x0bc3, 0x0bc6, 752 0x0bc9, 0x0bca, 753 0x0bcd, 0x0bd0, 754 0x0bd1, 0x0bd7, 755 0x0bd8, 0x0be6, 756 0x0bf3, 0x0c01, 757 0x0c04, 0x0c05, 758 0x0c0d, 0x0c0e, 759 0x0c11, 0x0c12, 760 0x0c29, 0x0c2a, 761 0x0c3a, 0x0c3d, 762 0x0c3e, 0x0c41, 763 0x0c45, 0x0c58, 764 0x0c5a, 0x0c60, 765 0x0c62, 0x0c66, 766 0x0c70, 0x0c7f, 767 0x0c80, 0x0c82, 768 0x0c84, 0x0c85, 769 0x0c8d, 0x0c8e, 770 0x0c91, 0x0c92, 771 0x0ca9, 0x0caa, 772 0x0cb4, 0x0cb5, 773 0x0cba, 0x0cbd, 774 0x0cc5, 0x0cc6, 775 0x0cc9, 0x0cca, 776 0x0ccc, 0x0cd5, 777 0x0cd7, 0x0cde, 778 0x0cdf, 0x0ce0, 779 0x0ce2, 0x0ce6, 780 0x0cf0, 0x0cf1, 781 0x0cf3, 0x0d02, 782 0x0d04, 0x0d05, 783 0x0d0d, 0x0d0e, 784 0x0d11, 0x0d12, 785 0x0d3b, 0x0d3d, 786 0x0d41, 0x0d46, 787 0x0d49, 0x0d4a, 788 0x0d4d, 0x0d4e, 789 0x0d4f, 0x0d57, 790 0x0d58, 0x0d60, 791 0x0d62, 0x0d66, 792 0x0d76, 0x0d79, 793 0x0d80, 0x0d82, 794 0x0d84, 0x0d85, 795 0x0d97, 0x0d9a, 796 0x0db2, 0x0db3, 797 0x0dbc, 0x0dbd, 798 0x0dbe, 0x0dc0, 799 0x0dc7, 0x0dcf, 800 0x0dd2, 0x0dd8, 801 0x0de0, 0x0de6, 802 0x0df0, 0x0df2, 803 0x0df5, 0x0e01, 804 0x0e31, 0x0e32, 805 0x0e34, 0x0e40, 806 0x0e47, 0x0e4f, 807 0x0e5c, 0x0e81, 808 0x0e83, 0x0e84, 809 0x0e85, 0x0e87, 810 0x0e89, 0x0e8a, 811 0x0e8b, 0x0e8d, 812 0x0e8e, 0x0e94, 813 0x0e98, 0x0e99, 814 0x0ea0, 0x0ea1, 815 0x0ea4, 0x0ea5, 816 0x0ea6, 0x0ea7, 817 0x0ea8, 0x0eaa, 818 0x0eac, 0x0ead, 819 0x0eb1, 0x0eb2, 820 0x0eb4, 0x0ebd, 821 0x0ebe, 0x0ec0, 822 0x0ec5, 0x0ec6, 823 0x0ec7, 0x0ed0, 824 0x0eda, 0x0edc, 825 0x0ee0, 0x0f00, 826 0x0f18, 0x0f1a, 827 0x0f35, 0x0f36, 828 0x0f37, 0x0f38, 829 0x0f39, 0x0f3e, 830 0x0f48, 0x0f49, 831 0x0f6d, 0x0f7f, 832 0x0f80, 0x0f85, 833 0x0f86, 0x0f88, 834 0x0f8d, 0x0fbe, 835 0x0fc6, 0x0fc7, 836 0x0fcd, 0x0fce, 837 0x0fdb, 0x1000, 838 0x102d, 0x1031, 839 0x1032, 0x1038, 840 0x1039, 0x103b, 841 0x103d, 0x103f, 842 0x1058, 0x105a, 843 0x105e, 0x1061, 844 0x1071, 0x1075, 845 0x1082, 0x1083, 846 0x1085, 0x1087, 847 0x108d, 0x108e, 848 0x109d, 0x109e, 849 0x10c6, 0x10c7, 850 0x10c8, 0x10cd, 851 0x10ce, 0x10d0, 852 0x1249, 0x124a, 853 0x124e, 0x1250, 854 0x1257, 0x1258, 855 0x1259, 0x125a, 856 0x125e, 0x1260, 857 0x1289, 0x128a, 858 0x128e, 0x1290, 859 0x12b1, 0x12b2, 860 0x12b6, 0x12b8, 861 0x12bf, 0x12c0, 862 0x12c1, 0x12c2, 863 0x12c6, 0x12c8, 864 0x12d7, 0x12d8, 865 0x1311, 0x1312, 866 0x1316, 0x1318, 867 0x135b, 0x1360, 868 0x137d, 0x1380, 869 0x1390, 0x13a0, 870 0x13f5, 0x1401, 871 0x1680, 0x1681, 872 0x169b, 0x16a0, 873 0x16f9, 0x1700, 874 0x170d, 0x170e, 875 0x1712, 0x1720, 876 0x1732, 0x1735, 877 0x1737, 0x1740, 878 0x1752, 0x1760, 879 0x176d, 0x176e, 880 0x1771, 0x1780, 881 0x17b4, 0x17b6, 882 0x17b7, 0x17be, 883 0x17c6, 0x17c7, 884 0x17c9, 0x17d4, 885 0x17db, 0x17dc, 886 0x17dd, 0x17e0, 887 0x17ea, 0x1810, 888 0x181a, 0x1820, 889 0x1878, 0x1880, 890 0x18a9, 0x18aa, 891 0x18ab, 0x18b0, 892 0x18f6, 0x1900, 893 0x191f, 0x1923, 894 0x1927, 0x1929, 895 0x192c, 0x1930, 896 0x1932, 0x1933, 897 0x1939, 0x1946, 898 0x196e, 0x1970, 899 0x1975, 0x1980, 900 0x19ac, 0x19b0, 901 0x19ca, 0x19d0, 902 0x19db, 0x1a00, 903 0x1a17, 0x1a19, 904 0x1a1b, 0x1a1e, 905 0x1a56, 0x1a57, 906 0x1a58, 0x1a61, 907 0x1a62, 0x1a63, 908 0x1a65, 0x1a6d, 909 0x1a73, 0x1a80, 910 0x1a8a, 0x1a90, 911 0x1a9a, 0x1aa0, 912 0x1aae, 0x1b04, 913 0x1b34, 0x1b35, 914 0x1b36, 0x1b3b, 915 0x1b3c, 0x1b3d, 916 0x1b42, 0x1b43, 917 0x1b4c, 0x1b50, 918 0x1b6b, 0x1b74, 919 0x1b7d, 0x1b82, 920 0x1ba2, 0x1ba6, 921 0x1ba8, 0x1baa, 922 0x1bab, 0x1bae, 923 0x1be6, 0x1be7, 924 0x1be8, 0x1bea, 925 0x1bed, 0x1bee, 926 0x1bef, 0x1bf2, 927 0x1bf4, 0x1bfc, 928 0x1c2c, 0x1c34, 929 0x1c36, 0x1c3b, 930 0x1c4a, 0x1c4d, 931 0x1c80, 0x1cc0, 932 0x1cc8, 0x1cd3, 933 0x1cd4, 0x1ce1, 934 0x1ce2, 0x1ce9, 935 0x1ced, 0x1cee, 936 0x1cf4, 0x1cf5, 937 0x1cf7, 0x1d00, 938 0x1dc0, 0x1e00, 939 0x1f16, 0x1f18, 940 0x1f1e, 0x1f20, 941 0x1f46, 0x1f48, 942 0x1f4e, 0x1f50, 943 0x1f58, 0x1f59, 944 0x1f5a, 0x1f5b, 945 0x1f5c, 0x1f5d, 946 0x1f5e, 0x1f5f, 947 0x1f7e, 0x1f80, 948 0x1fb5, 0x1fb6, 949 0x1fbd, 0x1fbe, 950 0x1fbf, 0x1fc2, 951 0x1fc5, 0x1fc6, 952 0x1fcd, 0x1fd0, 953 0x1fd4, 0x1fd6, 954 0x1fdc, 0x1fe0, 955 0x1fed, 0x1ff2, 956 0x1ff5, 0x1ff6, 957 0x1ffd, 0x200e, 958 0x2010, 0x2071, 959 0x2072, 0x207f, 960 0x2080, 0x2090, 961 0x209d, 0x2102, 962 0x2103, 0x2107, 963 0x2108, 0x210a, 964 0x2114, 0x2115, 965 0x2116, 0x2119, 966 0x211e, 0x2124, 967 0x2125, 0x2126, 968 0x2127, 0x2128, 969 0x2129, 0x212a, 970 0x212e, 0x212f, 971 0x213a, 0x213c, 972 0x2140, 0x2145, 973 0x214a, 0x214e, 974 0x2150, 0x2160, 975 0x2189, 0x2336, 976 0x237b, 0x2395, 977 0x2396, 0x249c, 978 0x24ea, 0x26ac, 979 0x26ad, 0x2800, 980 0x2900, 0x2c00, 981 0x2c2f, 0x2c30, 982 0x2c5f, 0x2c60, 983 0x2ce5, 0x2ceb, 984 0x2cef, 0x2cf2, 985 0x2cf4, 0x2d00, 986 0x2d26, 0x2d27, 987 0x2d28, 0x2d2d, 988 0x2d2e, 0x2d30, 989 0x2d68, 0x2d6f, 990 0x2d71, 0x2d80, 991 0x2d97, 0x2da0, 992 0x2da7, 0x2da8, 993 0x2daf, 0x2db0, 994 0x2db7, 0x2db8, 995 0x2dbf, 0x2dc0, 996 0x2dc7, 0x2dc8, 997 0x2dcf, 0x2dd0, 998 0x2dd7, 0x2dd8, 999 0x2ddf, 0x3005, 1000 0x3008, 0x3021, 1001 0x302a, 0x302e, 1002 0x3030, 0x3031, 1003 0x3036, 0x3038, 1004 0x303d, 0x3041, 1005 0x3097, 0x309d, 1006 0x30a0, 0x30a1, 1007 0x30fb, 0x30fc, 1008 0x3100, 0x3105, 1009 0x312e, 0x3131, 1010 0x318f, 0x3190, 1011 0x31bb, 0x31f0, 1012 0x321d, 0x3220, 1013 0x3250, 0x3260, 1014 0x327c, 0x327f, 1015 0x32b1, 0x32c0, 1016 0x32cc, 0x32d0, 1017 0x32ff, 0x3300, 1018 0x3377, 0x337b, 1019 0x33de, 0x33e0, 1020 0x33ff, 0x3400, 1021 0x4db6, 0x4e00, 1022 0x9fcd, 0xa000, 1023 0xa48d, 0xa4d0, 1024 0xa60d, 0xa610, 1025 0xa62c, 0xa640, 1026 0xa66f, 0xa680, 1027 0xa69e, 0xa6a0, 1028 0xa6f0, 0xa6f2, 1029 0xa6f8, 0xa722, 1030 0xa788, 0xa789, 1031 0xa78f, 0xa790, 1032 0xa7ae, 0xa7b0, 1033 0xa7b2, 0xa7f7, 1034 0xa802, 0xa803, 1035 0xa806, 0xa807, 1036 0xa80b, 0xa80c, 1037 0xa825, 0xa827, 1038 0xa828, 0xa830, 1039 0xa838, 0xa840, 1040 0xa874, 0xa880, 1041 0xa8c4, 0xa8ce, 1042 0xa8da, 0xa8f2, 1043 0xa8fc, 0xa900, 1044 0xa926, 0xa92e, 1045 0xa947, 0xa952, 1046 0xa954, 0xa95f, 1047 0xa97d, 0xa983, 1048 0xa9b3, 0xa9b4, 1049 0xa9b6, 0xa9ba, 1050 0xa9bc, 0xa9bd, 1051 0xa9ce, 0xa9cf, 1052 0xa9da, 0xa9de, 1053 0xa9e5, 0xa9e6, 1054 0xa9ff, 0xaa00, 1055 0xaa29, 0xaa2f, 1056 0xaa31, 0xaa33, 1057 0xaa35, 0xaa40, 1058 0xaa43, 0xaa44, 1059 0xaa4c, 0xaa4d, 1060 0xaa4e, 0xaa50, 1061 0xaa5a, 0xaa5c, 1062 0xaa7c, 0xaa7d, 1063 0xaab0, 0xaab1, 1064 0xaab2, 0xaab5, 1065 0xaab7, 0xaab9, 1066 0xaabe, 0xaac0, 1067 0xaac1, 0xaac2, 1068 0xaac3, 0xaadb, 1069 0xaaec, 0xaaee, 1070 0xaaf6, 0xab01, 1071 0xab07, 0xab09, 1072 0xab0f, 0xab11, 1073 0xab17, 0xab20, 1074 0xab27, 0xab28, 1075 0xab2f, 0xab30, 1076 0xab60, 0xab64, 1077 0xab66, 0xabc0, 1078 0xabe5, 0xabe6, 1079 0xabe8, 0xabe9, 1080 0xabed, 0xabf0, 1081 0xabfa, 0xac00, 1082 0xd7a4, 0xd7b0, 1083 0xd7c7, 0xd7cb, 1084 0xd7fc, 0xe000, 1085 0xfa6e, 0xfa70, 1086 0xfada, 0xfb00, 1087 0xfb07, 0xfb13, 1088 0xfb18, 0xfb1d, 1089 0xfb1e, 0xfb1f, 1090 0xfb29, 0xfb2a, 1091 0xfd3e, 0xfd40, 1092 0xfdd0, 0xfdf0, 1093 0xfdfd, 0xfdfe, 1094 0xfe00, 0xfe70, 1095 0xfeff, 0xff21, 1096 0xff3b, 0xff41, 1097 0xff5b, 0xff66, 1098 0xffbf, 0xffc2, 1099 0xffc8, 0xffca, 1100 0xffd0, 0xffd2, 1101 0xffd8, 0xffda, 1102 0xffdd, 0x10000, 1103 0x1000c, 0x1000d, 1104 0x10027, 0x10028, 1105 0x1003b, 0x1003c, 1106 0x1003e, 0x1003f, 1107 0x1004e, 0x10050, 1108 0x1005e, 0x10080, 1109 0x100fb, 0x10100, 1110 0x10101, 0x10102, 1111 0x10103, 0x10107, 1112 0x10134, 0x10137, 1113 0x10140, 0x101d0, 1114 0x101fd, 0x10280, 1115 0x1029d, 0x102a0, 1116 0x102d1, 0x10300, 1117 0x10324, 0x10330, 1118 0x1034b, 0x10350, 1119 0x10376, 0x10380, 1120 0x1039e, 0x1039f, 1121 0x103c4, 0x103c8, 1122 0x103d6, 0x10400, 1123 0x1049e, 0x104a0, 1124 0x104aa, 0x10500, 1125 0x10528, 0x10530, 1126 0x10564, 0x1056f, 1127 0x10570, 0x10600, 1128 0x10737, 0x10740, 1129 0x10756, 0x10760, 1130 0x10768, 0x10800, 1131 0x1091f, 0x10920, 1132 0x10a01, 0x10a04, 1133 0x10a05, 0x10a07, 1134 0x10a0c, 0x10a10, 1135 0x10a38, 0x10a3b, 1136 0x10a3f, 0x10a40, 1137 0x10ae5, 0x10ae7, 1138 0x10b39, 0x10b40, 1139 0x10e60, 0x10e7f, 1140 0x11001, 0x11002, 1141 0x11038, 0x11047, 1142 0x1104e, 0x11066, 1143 0x11070, 0x11082, 1144 0x110b3, 0x110b7, 1145 0x110b9, 0x110bb, 1146 0x110c2, 0x110d0, 1147 0x110e9, 0x110f0, 1148 0x110fa, 0x11103, 1149 0x11127, 0x1112c, 1150 0x1112d, 0x11136, 1151 0x11144, 0x11150, 1152 0x11173, 0x11174, 1153 0x11177, 0x11182, 1154 0x111b6, 0x111bf, 1155 0x111c9, 0x111cd, 1156 0x111ce, 0x111d0, 1157 0x111db, 0x111e1, 1158 0x111f5, 0x11200, 1159 0x11212, 0x11213, 1160 0x1122f, 0x11232, 1161 0x11234, 0x11235, 1162 0x11236, 0x11238, 1163 0x1123e, 0x112b0, 1164 0x112df, 0x112e0, 1165 0x112e3, 0x112f0, 1166 0x112fa, 0x11302, 1167 0x11304, 0x11305, 1168 0x1130d, 0x1130f, 1169 0x11311, 0x11313, 1170 0x11329, 0x1132a, 1171 0x11331, 0x11332, 1172 0x11334, 0x11335, 1173 0x1133a, 0x1133d, 1174 0x11340, 0x11341, 1175 0x11345, 0x11347, 1176 0x11349, 0x1134b, 1177 0x1134e, 0x11357, 1178 0x11358, 0x1135d, 1179 0x11364, 0x11480, 1180 0x114b3, 0x114b9, 1181 0x114ba, 0x114bb, 1182 0x114bf, 0x114c1, 1183 0x114c2, 0x114c4, 1184 0x114c8, 0x114d0, 1185 0x114da, 0x11580, 1186 0x115b2, 0x115b8, 1187 0x115bc, 0x115be, 1188 0x115bf, 0x115c1, 1189 0x115ca, 0x11600, 1190 0x11633, 0x1163b, 1191 0x1163d, 0x1163e, 1192 0x1163f, 0x11641, 1193 0x11645, 0x11650, 1194 0x1165a, 0x11680, 1195 0x116ab, 0x116ac, 1196 0x116ad, 0x116ae, 1197 0x116b0, 0x116b6, 1198 0x116b7, 0x116c0, 1199 0x116ca, 0x118a0, 1200 0x118f3, 0x118ff, 1201 0x11900, 0x11ac0, 1202 0x11af9, 0x12000, 1203 0x12399, 0x12400, 1204 0x1246f, 0x12470, 1205 0x12475, 0x13000, 1206 0x1342f, 0x16800, 1207 0x16a39, 0x16a40, 1208 0x16a5f, 0x16a60, 1209 0x16a6a, 0x16a6e, 1210 0x16a70, 0x16ad0, 1211 0x16aee, 0x16af5, 1212 0x16af6, 0x16b00, 1213 0x16b30, 0x16b37, 1214 0x16b46, 0x16b50, 1215 0x16b5a, 0x16b5b, 1216 0x16b62, 0x16b63, 1217 0x16b78, 0x16b7d, 1218 0x16b90, 0x16f00, 1219 0x16f45, 0x16f50, 1220 0x16f7f, 0x16f93, 1221 0x16fa0, 0x1b000, 1222 0x1b002, 0x1bc00, 1223 0x1bc6b, 0x1bc70, 1224 0x1bc7d, 0x1bc80, 1225 0x1bc89, 0x1bc90, 1226 0x1bc9a, 0x1bc9c, 1227 0x1bc9d, 0x1bc9f, 1228 0x1bca0, 0x1d000, 1229 0x1d0f6, 0x1d100, 1230 0x1d127, 0x1d129, 1231 0x1d167, 0x1d16a, 1232 0x1d173, 0x1d183, 1233 0x1d185, 0x1d18c, 1234 0x1d1aa, 0x1d1ae, 1235 0x1d1de, 0x1d360, 1236 0x1d372, 0x1d400, 1237 0x1d455, 0x1d456, 1238 0x1d49d, 0x1d49e, 1239 0x1d4a0, 0x1d4a2, 1240 0x1d4a3, 0x1d4a5, 1241 0x1d4a7, 0x1d4a9, 1242 0x1d4ad, 0x1d4ae, 1243 0x1d4ba, 0x1d4bb, 1244 0x1d4bc, 0x1d4bd, 1245 0x1d4c4, 0x1d4c5, 1246 0x1d506, 0x1d507, 1247 0x1d50b, 0x1d50d, 1248 0x1d515, 0x1d516, 1249 0x1d51d, 0x1d51e, 1250 0x1d53a, 0x1d53b, 1251 0x1d53f, 0x1d540, 1252 0x1d545, 0x1d546, 1253 0x1d547, 0x1d54a, 1254 0x1d551, 0x1d552, 1255 0x1d6a6, 0x1d6a8, 1256 0x1d6db, 0x1d6dc, 1257 0x1d715, 0x1d716, 1258 0x1d74f, 0x1d750, 1259 0x1d789, 0x1d78a, 1260 0x1d7c3, 0x1d7c4, 1261 0x1d7cc, 0x1e800, 1262 0x1e8d0, 0x1e8d7, 1263 0x1eef0, 0x1eef2, 1264 0x1f000, 0x1f110, 1265 0x1f12f, 0x1f130, 1266 0x1f16a, 0x1f170, 1267 0x1f19b, 0x1f1e6, 1268 0x1f203, 0x1f210, 1269 0x1f23b, 0x1f240, 1270 0x1f249, 0x1f250, 1271 0x1f252, 0x20000, 1272 0x2a6d7, 0x2a700, 1273 0x2b735, 0x2b740, 1274 0x2b81e, 0x2f800, 1275 0x2fa1e, 0xf0000, 1276 0xffffe, 0x100000, 1277 0x10fffe, 0x10ffff // sentinel 1278 }; 1279 1280 1281 // use a binary search with a cache 1282 1283 private transient volatile int stCache = 0; 1284 1285 private boolean isStrongDirectional(char c) { 1286 int cachedIndex = stCache; 1287 if (c < strongTable[cachedIndex]) { 1288 cachedIndex = search(c, strongTable, 0, cachedIndex); 1289 } else if (c >= strongTable[cachedIndex + 1]) { 1290 cachedIndex = search(c, strongTable, cachedIndex + 1, 1291 strongTable.length - cachedIndex - 1); 1292 } 1293 boolean val = (cachedIndex & 0x1) == 1; 1294 stCache = cachedIndex; 1295 return val; 1296 } 1297 1298 private static int getKeyFromMask(int mask) { 1299 int key = 0; 1300 while (key < NUM_KEYS && ((mask & (1<<key)) == 0)) { 1301 ++key; 1302 } 1303 if (key == NUM_KEYS || ((mask & ~(1<<key)) != 0)) { 1304 throw new IllegalArgumentException("invalid shaper: " + Integer.toHexString(mask)); 1305 } 1306 return key; 1307 } 1308 1309 /** 1310 * Returns a shaper for the provided unicode range. All 1311 * Latin-1 (EUROPEAN) digits are converted 1312 * to the corresponding decimal unicode digits. 1313 * @param singleRange the specified Unicode range 1314 * @return a non-contextual numeric shaper 1315 * @throws IllegalArgumentException if the range is not a single range 1316 */ 1317 public static NumericShaper getShaper(int singleRange) { 1318 int key = getKeyFromMask(singleRange); 1319 return new NumericShaper(key, singleRange); 1320 } 1321 1322 /** 1323 * Returns a shaper for the provided Unicode 1324 * range. All Latin-1 (EUROPEAN) digits are converted to the 1325 * corresponding decimal digits of the specified Unicode range. 1326 * 1327 * @param singleRange the Unicode range given by a {@link 1328 * NumericShaper.Range} constant. 1329 * @return a non-contextual {@code NumericShaper}. 1330 * @throws NullPointerException if {@code singleRange} is {@code null} 1331 * @since 1.7 1332 */ 1333 public static NumericShaper getShaper(Range singleRange) { 1334 return new NumericShaper(singleRange, EnumSet.of(singleRange)); 1335 } 1336 1337 /** 1338 * Returns a contextual shaper for the provided unicode range(s). 1339 * Latin-1 (EUROPEAN) digits are converted to the decimal digits 1340 * corresponding to the range of the preceding text, if the 1341 * range is one of the provided ranges. Multiple ranges are 1342 * represented by or-ing the values together, such as, 1343 * <code>NumericShaper.ARABIC | NumericShaper.THAI</code>. The 1344 * shaper assumes EUROPEAN as the starting context, that is, if 1345 * EUROPEAN digits are encountered before any strong directional 1346 * text in the string, the context is presumed to be EUROPEAN, and 1347 * so the digits will not shape. 1348 * @param ranges the specified Unicode ranges 1349 * @return a shaper for the specified ranges 1350 */ 1351 public static NumericShaper getContextualShaper(int ranges) { 1352 ranges |= CONTEXTUAL_MASK; 1353 return new NumericShaper(EUROPEAN_KEY, ranges); 1354 } 1355 1356 /** 1357 * Returns a contextual shaper for the provided Unicode 1358 * range(s). The Latin-1 (EUROPEAN) digits are converted to the 1359 * decimal digits corresponding to the range of the preceding 1360 * text, if the range is one of the provided ranges. 1361 * 1362 * <p>The shaper assumes EUROPEAN as the starting context, that 1363 * is, if EUROPEAN digits are encountered before any strong 1364 * directional text in the string, the context is presumed to be 1365 * EUROPEAN, and so the digits will not shape. 1366 * 1367 * @param ranges the specified Unicode ranges 1368 * @return a contextual shaper for the specified ranges 1369 * @throws NullPointerException if {@code ranges} is {@code null}. 1370 * @since 1.7 1371 */ 1372 public static NumericShaper getContextualShaper(Set<Range> ranges) { 1373 NumericShaper shaper = new NumericShaper(Range.EUROPEAN, ranges); 1374 shaper.mask = CONTEXTUAL_MASK; 1375 return shaper; 1376 } 1377 1378 /** 1379 * Returns a contextual shaper for the provided unicode range(s). 1380 * Latin-1 (EUROPEAN) digits will be converted to the decimal digits 1381 * corresponding to the range of the preceding text, if the 1382 * range is one of the provided ranges. Multiple ranges are 1383 * represented by or-ing the values together, for example, 1384 * <code>NumericShaper.ARABIC | NumericShaper.THAI</code>. The 1385 * shaper uses defaultContext as the starting context. 1386 * @param ranges the specified Unicode ranges 1387 * @param defaultContext the starting context, such as 1388 * <code>NumericShaper.EUROPEAN</code> 1389 * @return a shaper for the specified Unicode ranges. 1390 * @throws IllegalArgumentException if the specified 1391 * <code>defaultContext</code> is not a single valid range. 1392 */ 1393 public static NumericShaper getContextualShaper(int ranges, int defaultContext) { 1394 int key = getKeyFromMask(defaultContext); 1395 ranges |= CONTEXTUAL_MASK; 1396 return new NumericShaper(key, ranges); 1397 } 1398 1399 /** 1400 * Returns a contextual shaper for the provided Unicode range(s). 1401 * The Latin-1 (EUROPEAN) digits will be converted to the decimal 1402 * digits corresponding to the range of the preceding text, if the 1403 * range is one of the provided ranges. The shaper uses {@code 1404 * defaultContext} as the starting context. 1405 * 1406 * @param ranges the specified Unicode ranges 1407 * @param defaultContext the starting context, such as 1408 * {@code NumericShaper.Range.EUROPEAN} 1409 * @return a contextual shaper for the specified Unicode ranges. 1410 * @throws NullPointerException 1411 * if {@code ranges} or {@code defaultContext} is {@code null} 1412 * @since 1.7 1413 */ 1414 public static NumericShaper getContextualShaper(Set<Range> ranges, 1415 Range defaultContext) { 1416 if (defaultContext == null) { 1417 throw new NullPointerException(); 1418 } 1419 NumericShaper shaper = new NumericShaper(defaultContext, ranges); 1420 shaper.mask = CONTEXTUAL_MASK; 1421 return shaper; 1422 } 1423 1424 /** 1425 * Private constructor. 1426 */ 1427 private NumericShaper(int key, int mask) { 1428 this.key = key; 1429 this.mask = mask; 1430 } 1431 1432 private NumericShaper(Range defaultContext, Set<Range> ranges) { 1433 shapingRange = defaultContext; 1434 rangeSet = EnumSet.copyOf(ranges); // throws NPE if ranges is null. 1435 1436 // Give precedence to EASTERN_ARABIC if both ARABIC and 1437 // EASTERN_ARABIC are specified. 1438 if (rangeSet.contains(Range.EASTERN_ARABIC) 1439 && rangeSet.contains(Range.ARABIC)) { 1440 rangeSet.remove(Range.ARABIC); 1441 } 1442 1443 // As well as the above case, give precedence to TAI_THAM_THAM if both 1444 // TAI_THAM_HORA and TAI_THAM_THAM are specified. 1445 if (rangeSet.contains(Range.TAI_THAM_THAM) 1446 && rangeSet.contains(Range.TAI_THAM_HORA)) { 1447 rangeSet.remove(Range.TAI_THAM_HORA); 1448 } 1449 1450 rangeArray = rangeSet.toArray(new Range[rangeSet.size()]); 1451 if (rangeArray.length > BSEARCH_THRESHOLD) { 1452 // sort rangeArray for binary search 1453 Arrays.sort(rangeArray, 1454 new Comparator<Range>() { 1455 public int compare(Range s1, Range s2) { 1456 return s1.base > s2.base ? 1 : s1.base == s2.base ? 0 : -1; 1457 } 1458 }); 1459 } 1460 } 1461 1462 /** 1463 * Converts the digits in the text that occur between start and 1464 * start + count. 1465 * @param text an array of characters to convert 1466 * @param start the index into <code>text</code> to start 1467 * converting 1468 * @param count the number of characters in <code>text</code> 1469 * to convert 1470 * @throws IndexOutOfBoundsException if start or start + count is 1471 * out of bounds 1472 * @throws NullPointerException if text is null 1473 */ 1474 public void shape(char[] text, int start, int count) { 1475 checkParams(text, start, count); 1476 if (isContextual()) { 1477 if (rangeSet == null) { 1478 shapeContextually(text, start, count, key); 1479 } else { 1480 shapeContextually(text, start, count, shapingRange); 1481 } 1482 } else { 1483 shapeNonContextually(text, start, count); 1484 } 1485 } 1486 1487 /** 1488 * Converts the digits in the text that occur between start and 1489 * start + count, using the provided context. 1490 * Context is ignored if the shaper is not a contextual shaper. 1491 * @param text an array of characters 1492 * @param start the index into <code>text</code> to start 1493 * converting 1494 * @param count the number of characters in <code>text</code> 1495 * to convert 1496 * @param context the context to which to convert the 1497 * characters, such as <code>NumericShaper.EUROPEAN</code> 1498 * @throws IndexOutOfBoundsException if start or start + count is 1499 * out of bounds 1500 * @throws NullPointerException if text is null 1501 * @throws IllegalArgumentException if this is a contextual shaper 1502 * and the specified <code>context</code> is not a single valid 1503 * range. 1504 */ 1505 public void shape(char[] text, int start, int count, int context) { 1506 checkParams(text, start, count); 1507 if (isContextual()) { 1508 int ctxKey = getKeyFromMask(context); 1509 if (rangeSet == null) { 1510 shapeContextually(text, start, count, ctxKey); 1511 } else { 1512 shapeContextually(text, start, count, Range.values()[ctxKey]); 1513 } 1514 } else { 1515 shapeNonContextually(text, start, count); 1516 } 1517 } 1518 1519 /** 1520 * Converts the digits in the text that occur between {@code 1521 * start} and {@code start + count}, using the provided {@code 1522 * context}. {@code Context} is ignored if the shaper is not a 1523 * contextual shaper. 1524 * 1525 * @param text a {@code char} array 1526 * @param start the index into {@code text} to start converting 1527 * @param count the number of {@code char}s in {@code text} 1528 * to convert 1529 * @param context the context to which to convert the characters, 1530 * such as {@code NumericShaper.Range.EUROPEAN} 1531 * @throws IndexOutOfBoundsException 1532 * if {@code start} or {@code start + count} is out of bounds 1533 * @throws NullPointerException 1534 * if {@code text} or {@code context} is null 1535 * @since 1.7 1536 */ 1537 public void shape(char[] text, int start, int count, Range context) { 1538 checkParams(text, start, count); 1539 if (context == null) { 1540 throw new NullPointerException("context is null"); 1541 } 1542 1543 if (isContextual()) { 1544 if (rangeSet != null) { 1545 shapeContextually(text, start, count, context); 1546 } else { 1547 int key = Range.toRangeIndex(context); 1548 if (key >= 0) { 1549 shapeContextually(text, start, count, key); 1550 } else { 1551 shapeContextually(text, start, count, shapingRange); 1552 } 1553 } 1554 } else { 1555 shapeNonContextually(text, start, count); 1556 } 1557 } 1558 1559 private void checkParams(char[] text, int start, int count) { 1560 if (text == null) { 1561 throw new NullPointerException("text is null"); 1562 } 1563 if ((start < 0) 1564 || (start > text.length) 1565 || ((start + count) < 0) 1566 || ((start + count) > text.length)) { 1567 throw new IndexOutOfBoundsException( 1568 "bad start or count for text of length " + text.length); 1569 } 1570 } 1571 1572 /** 1573 * Returns a <code>boolean</code> indicating whether or not 1574 * this shaper shapes contextually. 1575 * @return <code>true</code> if this shaper is contextual; 1576 * <code>false</code> otherwise. 1577 */ 1578 public boolean isContextual() { 1579 return (mask & CONTEXTUAL_MASK) != 0; 1580 } 1581 1582 /** 1583 * Returns an <code>int</code> that ORs together the values for 1584 * all the ranges that will be shaped. 1585 * <p> 1586 * For example, to check if a shaper shapes to Arabic, you would use the 1587 * following: 1588 * <blockquote> 1589 * {@code if ((shaper.getRanges() & shaper.ARABIC) != 0) { ... } 1590 * </blockquote> 1591 * 1592 * <p>Note that this method supports only the bit mask-based 1593 * ranges. Call {@link #getRangeSet()} for the enum-based ranges. 1594 * 1595 * @return the values for all the ranges to be shaped. 1596 */ 1597 public int getRanges() { 1598 return mask & ~CONTEXTUAL_MASK; 1599 } 1600 1601 /** 1602 * Returns a {@code Set} representing all the Unicode ranges in 1603 * this {@code NumericShaper} that will be shaped. 1604 * 1605 * @return all the Unicode ranges to be shaped. 1606 * @since 1.7 1607 */ 1608 public Set<Range> getRangeSet() { 1609 if (rangeSet != null) { 1610 return EnumSet.copyOf(rangeSet); 1611 } 1612 return Range.maskToRangeSet(mask); 1613 } 1614 1615 /** 1616 * Perform non-contextual shaping. 1617 */ 1618 private void shapeNonContextually(char[] text, int start, int count) { 1619 int base; 1620 char minDigit = '0'; 1621 if (shapingRange != null) { 1622 base = shapingRange.getDigitBase(); 1623 minDigit += shapingRange.getNumericBase(); 1624 } else { 1625 base = bases[key]; 1626 if (key == ETHIOPIC_KEY) { 1627 minDigit++; // Ethiopic doesn't use decimal zero 1628 } 1629 } 1630 for (int i = start, e = start + count; i < e; ++i) { 1631 char c = text[i]; 1632 if (c >= minDigit && c <= '\u0039') { 1633 text[i] = (char)(c + base); 1634 } 1635 } 1636 } 1637 1638 /** 1639 * Perform contextual shaping. 1640 * Synchronized to protect caches used in getContextKey. 1641 */ 1642 private synchronized void shapeContextually(char[] text, int start, int count, int ctxKey) { 1643 1644 // if we don't support this context, then don't shape 1645 if ((mask & (1<<ctxKey)) == 0) { 1646 ctxKey = EUROPEAN_KEY; 1647 } 1648 int lastkey = ctxKey; 1649 1650 int base = bases[ctxKey]; 1651 char minDigit = ctxKey == ETHIOPIC_KEY ? '1' : '0'; // Ethiopic doesn't use decimal zero 1652 1653 synchronized (NumericShaper.class) { 1654 for (int i = start, e = start + count; i < e; ++i) { 1655 char c = text[i]; 1656 if (c >= minDigit && c <= '\u0039') { 1657 text[i] = (char)(c + base); 1658 } 1659 1660 if (isStrongDirectional(c)) { 1661 int newkey = getContextKey(c); 1662 if (newkey != lastkey) { 1663 lastkey = newkey; 1664 1665 ctxKey = newkey; 1666 if (((mask & EASTERN_ARABIC) != 0) && 1667 (ctxKey == ARABIC_KEY || 1668 ctxKey == EASTERN_ARABIC_KEY)) { 1669 ctxKey = EASTERN_ARABIC_KEY; 1670 } else if (((mask & ARABIC) != 0) && 1671 (ctxKey == ARABIC_KEY || 1672 ctxKey == EASTERN_ARABIC_KEY)) { 1673 ctxKey = ARABIC_KEY; 1674 } else if ((mask & (1<<ctxKey)) == 0) { 1675 ctxKey = EUROPEAN_KEY; 1676 } 1677 1678 base = bases[ctxKey]; 1679 1680 minDigit = ctxKey == ETHIOPIC_KEY ? '1' : '0'; // Ethiopic doesn't use decimal zero 1681 } 1682 } 1683 } 1684 } 1685 } 1686 1687 private void shapeContextually(char[] text, int start, int count, Range ctxKey) { 1688 // if we don't support the specified context, then don't shape. 1689 if (ctxKey == null || !rangeSet.contains(ctxKey)) { 1690 ctxKey = Range.EUROPEAN; 1691 } 1692 1693 Range lastKey = ctxKey; 1694 int base = ctxKey.getDigitBase(); 1695 char minDigit = (char)('0' + ctxKey.getNumericBase()); 1696 final int end = start + count; 1697 for (int i = start; i < end; ++i) { 1698 char c = text[i]; 1699 if (c >= minDigit && c <= '9') { 1700 text[i] = (char)(c + base); 1701 continue; 1702 } 1703 if (isStrongDirectional(c)) { 1704 ctxKey = rangeForCodePoint(c); 1705 if (ctxKey != lastKey) { 1706 lastKey = ctxKey; 1707 base = ctxKey.getDigitBase(); 1708 minDigit = (char)('0' + ctxKey.getNumericBase()); 1709 } 1710 } 1711 } 1712 } 1713 1714 /** 1715 * Returns a hash code for this shaper. 1716 * @return this shaper's hash code. 1717 * @see java.lang.Object#hashCode 1718 */ 1719 public int hashCode() { 1720 int hash = mask; 1721 if (rangeSet != null) { 1722 // Use the CONTEXTUAL_MASK bit only for the enum-based 1723 // NumericShaper. A deserialized NumericShaper might have 1724 // bit masks. 1725 hash &= CONTEXTUAL_MASK; 1726 hash ^= rangeSet.hashCode(); 1727 } 1728 return hash; 1729 } 1730 1731 /** 1732 * Returns {@code true} if the specified object is an instance of 1733 * <code>NumericShaper</code> and shapes identically to this one, 1734 * regardless of the range representations, the bit mask or the 1735 * enum. For example, the following code produces {@code "true"}. 1736 * <blockquote><pre> 1737 * NumericShaper ns1 = NumericShaper.getShaper(NumericShaper.ARABIC); 1738 * NumericShaper ns2 = NumericShaper.getShaper(NumericShaper.Range.ARABIC); 1739 * System.out.println(ns1.equals(ns2)); 1740 * </pre></blockquote> 1741 * 1742 * @param o the specified object to compare to this 1743 * <code>NumericShaper</code> 1744 * @return <code>true</code> if <code>o</code> is an instance 1745 * of <code>NumericShaper</code> and shapes in the same way; 1746 * <code>false</code> otherwise. 1747 * @see java.lang.Object#equals(java.lang.Object) 1748 */ 1749 public boolean equals(Object o) { 1750 if (o != null) { 1751 try { 1752 NumericShaper rhs = (NumericShaper)o; 1753 if (rangeSet != null) { 1754 if (rhs.rangeSet != null) { 1755 return isContextual() == rhs.isContextual() 1756 && rangeSet.equals(rhs.rangeSet) 1757 && shapingRange == rhs.shapingRange; 1758 } 1759 return isContextual() == rhs.isContextual() 1760 && rangeSet.equals(Range.maskToRangeSet(rhs.mask)) 1761 && shapingRange == Range.indexToRange(rhs.key); 1762 } else if (rhs.rangeSet != null) { 1763 Set<Range> rset = Range.maskToRangeSet(mask); 1764 Range srange = Range.indexToRange(key); 1765 return isContextual() == rhs.isContextual() 1766 && rset.equals(rhs.rangeSet) 1767 && srange == rhs.shapingRange; 1768 } 1769 return rhs.mask == mask && rhs.key == key; 1770 } 1771 catch (ClassCastException e) { 1772 } 1773 } 1774 return false; 1775 } 1776 1777 /** 1778 * Returns a <code>String</code> that describes this shaper. This method 1779 * is used for debugging purposes only. 1780 * @return a <code>String</code> describing this shaper. 1781 */ 1782 public String toString() { 1783 StringBuilder buf = new StringBuilder(super.toString()); 1784 1785 buf.append("[contextual:").append(isContextual()); 1786 1787 String[] keyNames = null; 1788 if (isContextual()) { 1789 buf.append(", context:"); 1790 buf.append(shapingRange == null ? Range.values()[key] : shapingRange); 1791 } 1792 1793 if (rangeSet == null) { 1794 buf.append(", range(s): "); 1795 boolean first = true; 1796 for (int i = 0; i < NUM_KEYS; ++i) { 1797 if ((mask & (1 << i)) != 0) { 1798 if (first) { 1799 first = false; 1800 } else { 1801 buf.append(", "); 1802 } 1803 buf.append(Range.values()[i]); 1804 } 1805 } 1806 } else { 1807 buf.append(", range set: ").append(rangeSet); 1808 } 1809 buf.append(']'); 1810 1811 return buf.toString(); 1812 } 1813 1814 /** 1815 * Returns the index of the high bit in value (assuming le, actually 1816 * power of 2 >= value). value must be positive. 1817 */ 1818 private static int getHighBit(int value) { 1819 if (value <= 0) { 1820 return -32; 1821 } 1822 1823 int bit = 0; 1824 1825 if (value >= 1 << 16) { 1826 value >>= 16; 1827 bit += 16; 1828 } 1829 1830 if (value >= 1 << 8) { 1831 value >>= 8; 1832 bit += 8; 1833 } 1834 1835 if (value >= 1 << 4) { 1836 value >>= 4; 1837 bit += 4; 1838 } 1839 1840 if (value >= 1 << 2) { 1841 value >>= 2; 1842 bit += 2; 1843 } 1844 1845 if (value >= 1 << 1) { 1846 bit += 1; 1847 } 1848 1849 return bit; 1850 } 1851 1852 /** 1853 * fast binary search over subrange of array. 1854 */ 1855 private static int search(int value, int[] array, int start, int length) 1856 { 1857 int power = 1 << getHighBit(length); 1858 int extra = length - power; 1859 int probe = power; 1860 int index = start; 1861 1862 if (value >= array[index + extra]) { 1863 index += extra; 1864 } 1865 1866 while (probe > 1) { 1867 probe >>= 1; 1868 1869 if (value >= array[index + probe]) { 1870 index += probe; 1871 } 1872 } 1873 1874 return index; 1875 } 1876 1877 /** 1878 * Converts the {@code NumericShaper.Range} enum-based parameters, 1879 * if any, to the bit mask-based counterparts and writes this 1880 * object to the {@code stream}. Any enum constants that have no 1881 * bit mask-based counterparts are ignored in the conversion. 1882 * 1883 * @param stream the output stream to write to 1884 * @throws IOException if an I/O error occurs while writing to {@code stream} 1885 * @since 1.7 1886 */ 1887 private void writeObject(ObjectOutputStream stream) throws IOException { 1888 if (shapingRange != null) { 1889 int index = Range.toRangeIndex(shapingRange); 1890 if (index >= 0) { 1891 key = index; 1892 } 1893 } 1894 if (rangeSet != null) { 1895 mask |= Range.toRangeMask(rangeSet); 1896 } 1897 stream.defaultWriteObject(); 1898 } 1899 }