1 // © 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 *************************************************************************** 5 * Copyright (C) 1999-2016 International Business Machines Corporation * 6 * and others. All rights reserved. * 7 *************************************************************************** 8 9 ********************************************************************** 10 * Date Name Description 11 * 10/22/99 alan Creation. 12 * 11/11/99 rgillam Complete port from Java. 13 ********************************************************************** 14 */ 15 16 #ifndef RBBI_H 17 #define RBBI_H 18 19 #include "unicode/utypes.h" 20 21 #if U_SHOW_CPLUSPLUS_API 22 23 /** 24 * \file 25 * \brief C++ API: Rule Based Break Iterator 26 */ 27 28 #if !UCONFIG_NO_BREAK_ITERATION 29 30 #include "unicode/brkiter.h" 31 #include "unicode/udata.h" 32 #include "unicode/parseerr.h" 33 #include "unicode/schriter.h" 34 35 U_NAMESPACE_BEGIN 36 37 /** @internal */ 38 class LanguageBreakEngine; 39 struct RBBIDataHeader; 40 class RBBIDataWrapper; 41 class UnhandledEngine; 42 class UStack; 43 44 /** 45 * 46 * A subclass of BreakIterator whose behavior is specified using a list of rules. 47 * <p>Instances of this class are most commonly created by the factory methods of 48 * BreakIterator::createWordInstance(), BreakIterator::createLineInstance(), etc., 49 * and then used via the abstract API in class BreakIterator</p> 50 * 51 * <p>See the ICU User Guide for information on Break Iterator Rules.</p> 52 * 53 * <p>This class is not intended to be subclassed.</p> 54 */ 55 class U_COMMON_API RuleBasedBreakIterator /*U_FINAL*/ : public BreakIterator { 56 57 private: 58 /** 59 * The UText through which this BreakIterator accesses the text 60 * @internal (private) 61 */ 62 UText fText; 63 64 #ifndef U_HIDE_INTERNAL_API 65 public: 66 #endif /* U_HIDE_INTERNAL_API */ 67 /** 68 * The rule data for this BreakIterator instance. 69 * Not for general use; Public only for testing purposes. 70 * @internal 71 */ 72 RBBIDataWrapper *fData; 73 private: 74 75 /** 76 * The current position of the iterator. Pinned, 0 < fPosition <= text.length. 77 * Never has the value UBRK_DONE (-1). 78 */ 79 int32_t fPosition; 80 81 /** 82 * TODO: 83 */ 84 int32_t fRuleStatusIndex; 85 86 /** 87 * Cache of previously determined boundary positions. 88 */ 89 class BreakCache; 90 BreakCache *fBreakCache; 91 92 /** 93 * Cache of boundary positions within a region of text that has been 94 * sub-divided by dictionary based breaking. 95 */ 96 class DictionaryCache; 97 DictionaryCache *fDictionaryCache; 98 99 /** 100 * 101 * If present, UStack of LanguageBreakEngine objects that might handle 102 * dictionary characters. Searched from top to bottom to find an object to 103 * handle a given character. 104 * @internal (private) 105 */ 106 UStack *fLanguageBreakEngines; 107 108 /** 109 * 110 * If present, the special LanguageBreakEngine used for handling 111 * characters that are in the dictionary set, but not handled by any 112 * LanguageBreakEngine. 113 * @internal (private) 114 */ 115 UnhandledEngine *fUnhandledBreakEngine; 116 117 /** 118 * Counter for the number of characters encountered with the "dictionary" 119 * flag set. 120 * @internal (private) 121 */ 122 uint32_t fDictionaryCharCount; 123 124 /** 125 * A character iterator that refers to the same text as the UText, above. 126 * Only included for compatibility with old API, which was based on CharacterIterators. 127 * Value may be adopted from outside, or one of fSCharIter or fDCharIter, below. 128 */ 129 CharacterIterator *fCharIter; 130 131 /** 132 * When the input text is provided by a UnicodeString, this will point to 133 * a characterIterator that wraps that data. Needed only for the 134 * implementation of getText(), a backwards compatibility issue. 135 */ 136 StringCharacterIterator fSCharIter; 137 138 /** 139 * True when iteration has run off the end, and iterator functions should return UBRK_DONE. 140 */ 141 UBool fDone; 142 143 //======================================================================= 144 // constructors 145 //======================================================================= 146 147 /** 148 * Constructor from a flattened set of RBBI data in malloced memory. 149 * RulesBasedBreakIterators built from a custom set of rules 150 * are created via this constructor; the rules are compiled 151 * into memory, then the break iterator is constructed here. 152 * 153 * The break iterator adopts the memory, and will 154 * free it when done. 155 * @internal (private) 156 */ 157 RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status); 158 159 /** @internal */ 160 friend class RBBIRuleBuilder; 161 /** @internal */ 162 friend class BreakIterator; 163 164 public: 165 166 /** Default constructor. Creates an empty shell of an iterator, with no 167 * rules or text to iterate over. Object can subsequently be assigned to. 168 * @stable ICU 2.2 169 */ 170 RuleBasedBreakIterator(); 171 172 /** 173 * Copy constructor. Will produce a break iterator with the same behavior, 174 * and which iterates over the same text, as the one passed in. 175 * @param that The RuleBasedBreakIterator passed to be copied 176 * @stable ICU 2.0 177 */ 178 RuleBasedBreakIterator(const RuleBasedBreakIterator& that); 179 180 /** 181 * Construct a RuleBasedBreakIterator from a set of rules supplied as a string. 182 * @param rules The break rules to be used. 183 * @param parseError In the event of a syntax error in the rules, provides the location 184 * within the rules of the problem. 185 * @param status Information on any errors encountered. 186 * @stable ICU 2.2 187 */ 188 RuleBasedBreakIterator( const UnicodeString &rules, 189 UParseError &parseError, 190 UErrorCode &status); 191 192 /** 193 * Construct a RuleBasedBreakIterator from a set of precompiled binary rules. 194 * Binary rules are obtained from RulesBasedBreakIterator::getBinaryRules(). 195 * Construction of a break iterator in this way is substantially faster than 196 * construction from source rules. 197 * 198 * Ownership of the storage containing the compiled rules remains with the 199 * caller of this function. The compiled rules must not be modified or 200 * deleted during the life of the break iterator. 201 * 202 * The compiled rules are not compatible across different major versions of ICU. 203 * The compiled rules are compatible only between machines with the same 204 * byte ordering (little or big endian) and the same base character set family 205 * (ASCII or EBCDIC). 206 * 207 * @see #getBinaryRules 208 * @param compiledRules A pointer to the compiled break rules to be used. 209 * @param ruleLength The length of the compiled break rules, in bytes. This 210 * corresponds to the length value produced by getBinaryRules(). 211 * @param status Information on any errors encountered, including invalid 212 * binary rules. 213 * @stable ICU 4.8 214 */ 215 RuleBasedBreakIterator(const uint8_t *compiledRules, 216 uint32_t ruleLength, 217 UErrorCode &status); 218 219 /** 220 * This constructor uses the udata interface to create a BreakIterator 221 * whose internal tables live in a memory-mapped file. "image" is an 222 * ICU UDataMemory handle for the pre-compiled break iterator tables. 223 * @param image handle to the memory image for the break iterator data. 224 * Ownership of the UDataMemory handle passes to the Break Iterator, 225 * which will be responsible for closing it when it is no longer needed. 226 * @param status Information on any errors encountered. 227 * @see udata_open 228 * @see #getBinaryRules 229 * @stable ICU 2.8 230 */ 231 RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status); 232 233 /** 234 * Destructor 235 * @stable ICU 2.0 236 */ 237 virtual ~RuleBasedBreakIterator(); 238 239 /** 240 * Assignment operator. Sets this iterator to have the same behavior, 241 * and iterate over the same text, as the one passed in. 242 * @param that The RuleBasedBreakItertor passed in 243 * @return the newly created RuleBasedBreakIterator 244 * @stable ICU 2.0 245 */ 246 RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that); 247 248 /** 249 * Equality operator. Returns TRUE if both BreakIterators are of the 250 * same class, have the same behavior, and iterate over the same text. 251 * @param that The BreakIterator to be compared for equality 252 * @return TRUE if both BreakIterators are of the 253 * same class, have the same behavior, and iterate over the same text. 254 * @stable ICU 2.0 255 */ 256 virtual UBool operator==(const BreakIterator& that) const; 257 258 /** 259 * Not-equal operator. If operator== returns TRUE, this returns FALSE, 260 * and vice versa. 261 * @param that The BreakIterator to be compared for inequality 262 * @return TRUE if both BreakIterators are not same. 263 * @stable ICU 2.0 264 */ 265 inline UBool operator!=(const BreakIterator& that) const; 266 267 /** 268 * Returns a newly-constructed RuleBasedBreakIterator with the same 269 * behavior, and iterating over the same text, as this one. 270 * Differs from the copy constructor in that it is polymorphic, and 271 * will correctly clone (copy) a derived class. 272 * clone() is thread safe. Multiple threads may simultaneously 273 * clone the same source break iterator. 274 * @return a newly-constructed RuleBasedBreakIterator 275 * @stable ICU 2.0 276 */ 277 virtual RuleBasedBreakIterator* clone() const; 278 279 /** 280 * Compute a hash code for this BreakIterator 281 * @return A hash code 282 * @stable ICU 2.0 283 */ 284 virtual int32_t hashCode(void) const; 285 286 /** 287 * Returns the description used to create this iterator 288 * @return the description used to create this iterator 289 * @stable ICU 2.0 290 */ 291 virtual const UnicodeString& getRules(void) const; 292 293 //======================================================================= 294 // BreakIterator overrides 295 //======================================================================= 296 297 /** 298 * <p> 299 * Return a CharacterIterator over the text being analyzed. 300 * The returned character iterator is owned by the break iterator, and must 301 * not be deleted by the caller. Repeated calls to this function may 302 * return the same CharacterIterator. 303 * </p> 304 * <p> 305 * The returned character iterator must not be used concurrently with 306 * the break iterator. If concurrent operation is needed, clone the 307 * returned character iterator first and operate on the clone. 308 * </p> 309 * <p> 310 * When the break iterator is operating on text supplied via a UText, 311 * this function will fail. Lacking any way to signal failures, it 312 * returns an CharacterIterator containing no text. 313 * The function getUText() provides similar functionality, 314 * is reliable, and is more efficient. 315 * </p> 316 * 317 * TODO: deprecate this function? 318 * 319 * @return An iterator over the text being analyzed. 320 * @stable ICU 2.0 321 */ 322 virtual CharacterIterator& getText(void) const; 323 324 325 /** 326 * Get a UText for the text being analyzed. 327 * The returned UText is a shallow clone of the UText used internally 328 * by the break iterator implementation. It can safely be used to 329 * access the text without impacting any break iterator operations, 330 * but the underlying text itself must not be altered. 331 * 332 * @param fillIn A UText to be filled in. If NULL, a new UText will be 333 * allocated to hold the result. 334 * @param status receives any error codes. 335 * @return The current UText for this break iterator. If an input 336 * UText was provided, it will always be returned. 337 * @stable ICU 3.4 338 */ 339 virtual UText *getUText(UText *fillIn, UErrorCode &status) const; 340 341 /** 342 * Set the iterator to analyze a new piece of text. This function resets 343 * the current iteration position to the beginning of the text. 344 * @param newText An iterator over the text to analyze. The BreakIterator 345 * takes ownership of the character iterator. The caller MUST NOT delete it! 346 * @stable ICU 2.0 347 */ 348 virtual void adoptText(CharacterIterator* newText); 349 350 /** 351 * Set the iterator to analyze a new piece of text. This function resets 352 * the current iteration position to the beginning of the text. 353 * 354 * The BreakIterator will retain a reference to the supplied string. 355 * The caller must not modify or delete the text while the BreakIterator 356 * retains the reference. 357 * 358 * @param newText The text to analyze. 359 * @stable ICU 2.0 360 */ 361 virtual void setText(const UnicodeString& newText); 362 363 /** 364 * Reset the break iterator to operate over the text represented by 365 * the UText. The iterator position is reset to the start. 366 * 367 * This function makes a shallow clone of the supplied UText. This means 368 * that the caller is free to immediately close or otherwise reuse the 369 * Utext that was passed as a parameter, but that the underlying text itself 370 * must not be altered while being referenced by the break iterator. 371 * 372 * @param text The UText used to change the text. 373 * @param status Receives any error codes. 374 * @stable ICU 3.4 375 */ 376 virtual void setText(UText *text, UErrorCode &status); 377 378 /** 379 * Sets the current iteration position to the beginning of the text, position zero. 380 * @return The offset of the beginning of the text, zero. 381 * @stable ICU 2.0 382 */ 383 virtual int32_t first(void); 384 385 /** 386 * Sets the current iteration position to the end of the text. 387 * @return The text's past-the-end offset. 388 * @stable ICU 2.0 389 */ 390 virtual int32_t last(void); 391 392 /** 393 * Advances the iterator either forward or backward the specified number of steps. 394 * Negative values move backward, and positive values move forward. This is 395 * equivalent to repeatedly calling next() or previous(). 396 * @param n The number of steps to move. The sign indicates the direction 397 * (negative is backwards, and positive is forwards). 398 * @return The character offset of the boundary position n boundaries away from 399 * the current one. 400 * @stable ICU 2.0 401 */ 402 virtual int32_t next(int32_t n); 403 404 /** 405 * Advances the iterator to the next boundary position. 406 * @return The position of the first boundary after this one. 407 * @stable ICU 2.0 408 */ 409 virtual int32_t next(void); 410 411 /** 412 * Moves the iterator backwards, to the last boundary preceding this one. 413 * @return The position of the last boundary position preceding this one. 414 * @stable ICU 2.0 415 */ 416 virtual int32_t previous(void); 417 418 /** 419 * Sets the iterator to refer to the first boundary position following 420 * the specified position. 421 * @param offset The position from which to begin searching for a break position. 422 * @return The position of the first break after the current position. 423 * @stable ICU 2.0 424 */ 425 virtual int32_t following(int32_t offset); 426 427 /** 428 * Sets the iterator to refer to the last boundary position before the 429 * specified position. 430 * @param offset The position to begin searching for a break from. 431 * @return The position of the last boundary before the starting position. 432 * @stable ICU 2.0 433 */ 434 virtual int32_t preceding(int32_t offset); 435 436 /** 437 * Returns true if the specified position is a boundary position. As a side 438 * effect, leaves the iterator pointing to the first boundary position at 439 * or after "offset". 440 * @param offset the offset to check. 441 * @return True if "offset" is a boundary position. 442 * @stable ICU 2.0 443 */ 444 virtual UBool isBoundary(int32_t offset); 445 446 /** 447 * Returns the current iteration position. Note that UBRK_DONE is never 448 * returned from this function; if iteration has run to the end of a 449 * string, current() will return the length of the string while 450 * next() will return UBRK_DONE). 451 * @return The current iteration position. 452 * @stable ICU 2.0 453 */ 454 virtual int32_t current(void) const; 455 456 457 /** 458 * Return the status tag from the break rule that determined the boundary at 459 * the current iteration position. For break rules that do not specify a 460 * status, a default value of 0 is returned. If more than one break rule 461 * would cause a boundary to be located at some position in the text, 462 * the numerically largest of the applicable status values is returned. 463 * <p> 464 * Of the standard types of ICU break iterators, only word break and 465 * line break provide status values. The values are defined in 466 * the header file ubrk.h. For Word breaks, the status allows distinguishing between words 467 * that contain alphabetic letters, "words" that appear to be numbers, 468 * punctuation and spaces, words containing ideographic characters, and 469 * more. For Line Break, the status distinguishes between hard (mandatory) breaks 470 * and soft (potential) break positions. 471 * <p> 472 * <code>getRuleStatus()</code> can be called after obtaining a boundary 473 * position from <code>next()</code>, <code>previous()</code>, or 474 * any other break iterator functions that returns a boundary position. 475 * <p> 476 * Note that <code>getRuleStatus()</code> returns the value corresponding to 477 * <code>current()</code> index even after <code>next()</code> has returned DONE. 478 * <p> 479 * When creating custom break rules, one is free to define whatever 480 * status values may be convenient for the application. 481 * <p> 482 * @return the status from the break rule that determined the boundary 483 * at the current iteration position. 484 * 485 * @see UWordBreak 486 * @stable ICU 2.2 487 */ 488 virtual int32_t getRuleStatus() const; 489 490 /** 491 * Get the status (tag) values from the break rule(s) that determined the boundary 492 * at the current iteration position. 493 * <p> 494 * The returned status value(s) are stored into an array provided by the caller. 495 * The values are stored in sorted (ascending) order. 496 * If the capacity of the output array is insufficient to hold the data, 497 * the output will be truncated to the available length, and a 498 * U_BUFFER_OVERFLOW_ERROR will be signaled. 499 * 500 * @param fillInVec an array to be filled in with the status values. 501 * @param capacity the length of the supplied vector. A length of zero causes 502 * the function to return the number of status values, in the 503 * normal way, without attempting to store any values. 504 * @param status receives error codes. 505 * @return The number of rule status values from the rules that determined 506 * the boundary at the current iteration position. 507 * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value 508 * is the total number of status values that were available, 509 * not the reduced number that were actually returned. 510 * @see getRuleStatus 511 * @stable ICU 3.0 512 */ 513 virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status); 514 515 /** 516 * Returns a unique class ID POLYMORPHICALLY. Pure virtual override. 517 * This method is to implement a simple version of RTTI, since not all 518 * C++ compilers support genuine RTTI. Polymorphic operator==() and 519 * clone() methods call this method. 520 * 521 * @return The class ID for this object. All objects of a 522 * given class have the same class ID. Objects of 523 * other classes have different class IDs. 524 * @stable ICU 2.0 525 */ 526 virtual UClassID getDynamicClassID(void) const; 527 528 /** 529 * Returns the class ID for this class. This is useful only for 530 * comparing to a return value from getDynamicClassID(). For example: 531 * 532 * Base* polymorphic_pointer = createPolymorphicObject(); 533 * if (polymorphic_pointer->getDynamicClassID() == 534 * Derived::getStaticClassID()) ... 535 * 536 * @return The class ID for all objects of this class. 537 * @stable ICU 2.0 538 */ 539 static UClassID U_EXPORT2 getStaticClassID(void); 540 541 #ifndef U_FORCE_HIDE_DEPRECATED_API 542 /** 543 * Deprecated functionality. Use clone() instead. 544 * 545 * Create a clone (copy) of this break iterator in memory provided 546 * by the caller. The idea is to increase performance by avoiding 547 * a storage allocation. Use of this function is NOT RECOMMENDED. 548 * Performance gains are minimal, and correct buffer management is 549 * tricky. Use clone() instead. 550 * 551 * @param stackBuffer The pointer to the memory into which the cloned object 552 * should be placed. If NULL, allocate heap memory 553 * for the cloned object. 554 * @param BufferSize The size of the buffer. If zero, return the required 555 * buffer size, but do not clone the object. If the 556 * size was too small (but not zero), allocate heap 557 * storage for the cloned object. 558 * 559 * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be 560 * returned if the provided buffer was too small, and 561 * the clone was therefore put on the heap. 562 * 563 * @return Pointer to the clone object. This may differ from the stackBuffer 564 * address if the byte alignment of the stack buffer was not suitable 565 * or if the stackBuffer was too small to hold the clone. 566 * @deprecated ICU 52. Use clone() instead. 567 */ 568 virtual RuleBasedBreakIterator *createBufferClone(void *stackBuffer, 569 int32_t &BufferSize, 570 UErrorCode &status); 571 #endif // U_FORCE_HIDE_DEPRECATED_API 572 573 /** 574 * Return the binary form of compiled break rules, 575 * which can then be used to create a new break iterator at some 576 * time in the future. Creating a break iterator from pre-compiled rules 577 * is much faster than building one from the source form of the 578 * break rules. 579 * 580 * The binary data can only be used with the same version of ICU 581 * and on the same platform type (processor endian-ness) 582 * 583 * @param length Returns the length of the binary data. (Out parameter.) 584 * 585 * @return A pointer to the binary (compiled) rule data. The storage 586 * belongs to the RulesBasedBreakIterator object, not the 587 * caller, and must not be modified or deleted. 588 * @stable ICU 4.8 589 */ 590 virtual const uint8_t *getBinaryRules(uint32_t &length); 591 592 /** 593 * Set the subject text string upon which the break iterator is operating 594 * without changing any other aspect of the matching state. 595 * The new and previous text strings must have the same content. 596 * 597 * This function is intended for use in environments where ICU is operating on 598 * strings that may move around in memory. It provides a mechanism for notifying 599 * ICU that the string has been relocated, and providing a new UText to access the 600 * string in its new position. 601 * 602 * Note that the break iterator implementation never copies the underlying text 603 * of a string being processed, but always operates directly on the original text 604 * provided by the user. Refreshing simply drops the references to the old text 605 * and replaces them with references to the new. 606 * 607 * Caution: this function is normally used only by very specialized, 608 * system-level code. One example use case is with garbage collection that moves 609 * the text in memory. 610 * 611 * @param input The new (moved) text string. 612 * @param status Receives errors detected by this function. 613 * @return *this 614 * 615 * @stable ICU 49 616 */ 617 virtual RuleBasedBreakIterator &refreshInputText(UText *input, UErrorCode &status); 618 619 620 private: 621 //======================================================================= 622 // implementation 623 //======================================================================= 624 /** 625 * Dumps caches and performs other actions associated with a complete change 626 * in text or iteration position. 627 * @internal (private) 628 */ 629 void reset(void); 630 631 /** 632 * Common initialization function, used by constructors and bufferClone. 633 * @internal (private) 634 */ 635 void init(UErrorCode &status); 636 637 /** 638 * Iterate backwards from an arbitrary position in the input text using the 639 * synthesized Safe Reverse rules. 640 * This locates a "Safe Position" from which the forward break rules 641 * will operate correctly. A Safe Position is not necessarily a boundary itself. 642 * 643 * @param fromPosition the position in the input text to begin the iteration. 644 * @internal (private) 645 */ 646 int32_t handleSafePrevious(int32_t fromPosition); 647 648 /** 649 * Find a rule-based boundary by running the state machine. 650 * Input 651 * fPosition, the position in the text to begin from. 652 * Output 653 * fPosition: the boundary following the starting position. 654 * fDictionaryCharCount the number of dictionary characters encountered. 655 * If > 0, the segment will be further subdivided 656 * fRuleStatusIndex Info from the state table indicating which rules caused the boundary. 657 * 658 * @internal (private) 659 */ 660 int32_t handleNext(); 661 662 663 /** 664 * This function returns the appropriate LanguageBreakEngine for a 665 * given character c. 666 * @param c A character in the dictionary set 667 * @internal (private) 668 */ 669 const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c); 670 671 public: 672 #ifndef U_HIDE_INTERNAL_API 673 /** 674 * Debugging function only. 675 * @internal 676 */ 677 void dumpCache(); 678 679 /** 680 * Debugging function only. 681 * @internal 682 */ 683 void dumpTables(); 684 685 #endif /* U_HIDE_INTERNAL_API */ 686 }; 687 688 //------------------------------------------------------------------------------ 689 // 690 // Inline Functions Definitions ... 691 // 692 //------------------------------------------------------------------------------ 693 694 inline UBool RuleBasedBreakIterator::operator!=(const BreakIterator& that) const { 695 return !operator==(that); 696 } 697 698 U_NAMESPACE_END 699 700 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */ 701 702 #endif /* U_SHOW_CPLUSPLUS_API */ 703 704 #endif 705