1 /* 2 *************************************************************************** 3 * Copyright (C) 1999-2007 International Business Machines Corporation * 4 * and others. All rights reserved. * 5 *************************************************************************** 6 7 ********************************************************************** 8 * Date Name Description 9 * 10/22/99 alan Creation. 10 * 11/11/99 rgillam Complete port from Java. 11 ********************************************************************** 12 */ 13 14 #ifndef RBBI_H 15 #define RBBI_H 16 17 #include "unicode/utypes.h" 18 19 /** 20 * \file 21 * \brief C++ API: Rule Based Break Iterator 22 */ 23 24 #if !UCONFIG_NO_BREAK_ITERATION 25 26 #include "unicode/brkiter.h" 27 #include "unicode/udata.h" 28 #include "unicode/parseerr.h" 29 #include "unicode/schriter.h" 30 #include "unicode/uchriter.h" 31 32 33 struct UTrie; 34 35 U_NAMESPACE_BEGIN 36 37 /** @internal */ 38 struct RBBIDataHeader; 39 class RuleBasedBreakIteratorTables; 40 class BreakIterator; 41 class RBBIDataWrapper; 42 class UStack; 43 class LanguageBreakEngine; 44 class UnhandledEngine; 45 struct RBBIStateTable; 46 47 48 49 50 /** 51 * 52 * A subclass of BreakIterator whose behavior is specified using a list of rules. 53 * <p>Instances of this class are most commonly created by the factory methods of 54 * BreakIterator::createWordInstance(), BreakIterator::createLineInstance(), etc., 55 * and then used via the abstract API in class BreakIterator</p> 56 * 57 * <p>See the ICU User Guide for information on Break Iterator Rules.</p> 58 * 59 * <p>This class is not intended to be subclassed. (Class DictionaryBasedBreakIterator 60 * is a subclass, but that relationship is effectively internal to the ICU 61 * implementation. The subclassing interface to RulesBasedBreakIterator is 62 * not part of the ICU API, and may not remain stable.</p> 63 * 64 */ 65 class U_COMMON_API RuleBasedBreakIterator : public BreakIterator { 66 67 protected: 68 /** 69 * The UText through which this BreakIterator accesses the text 70 * @internal 71 */ 72 UText *fText; 73 74 /** 75 * A character iterator that refers to the same text as the UText, above. 76 * Only included for compatibility with old API, which was based on CharacterIterators. 77 * Value may be adopted from outside, or one of fSCharIter or fDCharIter, below. 78 */ 79 CharacterIterator *fCharIter; 80 81 /** 82 * When the input text is provided by a UnicodeString, this will point to 83 * a characterIterator that wraps that data. Needed only for the 84 * implementation of getText(), a backwards compatibility issue. 85 */ 86 StringCharacterIterator *fSCharIter; 87 88 /** 89 * When the input text is provided by a UText, this 90 * dummy CharacterIterator over an empty string will 91 * be returned from getText() 92 */ 93 UCharCharacterIterator *fDCharIter; 94 95 /** 96 * The rule data for this BreakIterator instance 97 * @internal 98 */ 99 RBBIDataWrapper *fData; 100 101 /** Index of the Rule {tag} values for the most recent match. 102 * @internal 103 */ 104 int32_t fLastRuleStatusIndex; 105 106 /** 107 * Rule tag value valid flag. 108 * Some iterator operations don't intrinsically set the correct tag value. 109 * This flag lets us lazily compute the value if we are ever asked for it. 110 * @internal 111 */ 112 UBool fLastStatusIndexValid; 113 114 /** 115 * Counter for the number of characters encountered with the "dictionary" 116 * flag set. 117 * @internal 118 */ 119 uint32_t fDictionaryCharCount; 120 121 /** 122 * When a range of characters is divided up using the dictionary, the break 123 * positions that are discovered are stored here, preventing us from having 124 * to use either the dictionary or the state table again until the iterator 125 * leaves this range of text. Has the most impact for line breaking. 126 * @internal 127 */ 128 int32_t* fCachedBreakPositions; 129 130 /** 131 * The number of elements in fCachedBreakPositions 132 * @internal 133 */ 134 int32_t fNumCachedBreakPositions; 135 136 /** 137 * if fCachedBreakPositions is not null, this indicates which item in the 138 * cache the current iteration position refers to 139 * @internal 140 */ 141 int32_t fPositionInCache; 142 143 /** 144 * 145 * If present, UStack of LanguageBreakEngine objects that might handle 146 * dictionary characters. Searched from top to bottom to find an object to 147 * handle a given character. 148 * @internal 149 */ 150 UStack *fLanguageBreakEngines; 151 152 /** 153 * 154 * If present, the special LanguageBreakEngine used for handling 155 * characters that are in the dictionary set, but not handled by any 156 * LangugageBreakEngine. 157 * @internal 158 */ 159 UnhandledEngine *fUnhandledBreakEngine; 160 161 /** 162 * 163 * The type of the break iterator, or -1 if it has not been set. 164 * @internal 165 */ 166 int32_t fBreakType; 167 168 protected: 169 //======================================================================= 170 // constructors 171 //======================================================================= 172 173 /** 174 * Constructor from a flattened set of RBBI data in malloced memory. 175 * RulesBasedBreakIterators built from a custom set of rules 176 * are created via this constructor; the rules are compiled 177 * into memory, then the break iterator is constructed here. 178 * 179 * The break iterator adopts the memory, and will 180 * free it when done. 181 * @internal 182 */ 183 RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status); 184 185 186 friend class RBBIRuleBuilder; 187 /** @internal */ 188 friend class BreakIterator; 189 190 191 192 public: 193 194 /** Default constructor. Creates an empty shell of an iterator, with no 195 * rules or text to iterate over. Object can subsequently be assigned to. 196 * @stable ICU 2.2 197 */ 198 RuleBasedBreakIterator(); 199 200 /** 201 * Copy constructor. Will produce a break iterator with the same behavior, 202 * and which iterates over the same text, as the one passed in. 203 * @param that The RuleBasedBreakIterator passed to be copied 204 * @stable ICU 2.0 205 */ 206 RuleBasedBreakIterator(const RuleBasedBreakIterator& that); 207 208 /** 209 * Construct a RuleBasedBreakIterator from a set of rules supplied as a string. 210 * @param rules The break rules to be used. 211 * @param parseError In the event of a syntax error in the rules, provides the location 212 * within the rules of the problem. 213 * @param status Information on any errors encountered. 214 * @stable ICU 2.2 215 */ 216 RuleBasedBreakIterator( const UnicodeString &rules, 217 UParseError &parseError, 218 UErrorCode &status); 219 220 221 /** 222 * This constructor uses the udata interface to create a BreakIterator 223 * whose internal tables live in a memory-mapped file. "image" is an 224 * ICU UDataMemory handle for the pre-compiled break iterator tables. 225 * @param image handle to the memory image for the break iterator data. 226 * Ownership of the UDataMemory handle passes to the Break Iterator, 227 * which will be responsible for closing it when it is no longer needed. 228 * @param status Information on any errors encountered. 229 * @see udata_open 230 * @see #getBinaryRules 231 * @stable ICU 2.8 232 */ 233 RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status); 234 235 /** 236 * Destructor 237 * @stable ICU 2.0 238 */ 239 virtual ~RuleBasedBreakIterator(); 240 241 /** 242 * Assignment operator. Sets this iterator to have the same behavior, 243 * and iterate over the same text, as the one passed in. 244 * @param that The RuleBasedBreakItertor passed in 245 * @return the newly created RuleBasedBreakIterator 246 * @stable ICU 2.0 247 */ 248 RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that); 249 250 /** 251 * Equality operator. Returns TRUE if both BreakIterators are of the 252 * same class, have the same behavior, and iterate over the same text. 253 * @param that The BreakIterator to be compared for equality 254 * @return TRUE if both BreakIterators are of the 255 * same class, have the same behavior, and iterate over the same text. 256 * @stable ICU 2.0 257 */ 258 virtual UBool operator==(const BreakIterator& that) const; 259 260 /** 261 * Not-equal operator. If operator== returns TRUE, this returns FALSE, 262 * and vice versa. 263 * @param that The BreakIterator to be compared for inequality 264 * @return TRUE if both BreakIterators are not same. 265 * @stable ICU 2.0 266 */ 267 UBool operator!=(const BreakIterator& that) const; 268 269 /** 270 * Returns a newly-constructed RuleBasedBreakIterator with the same 271 * behavior, and iterating over the same text, as this one. 272 * Differs from the copy constructor in that it is polymorphic, and 273 * will correctly clone (copy) a derived class. 274 * clone() is thread safe. Multiple threads may simultaeneously 275 * clone the same source break iterator. 276 * @return a newly-constructed RuleBasedBreakIterator 277 * @stable ICU 2.0 278 */ 279 virtual BreakIterator* clone() const; 280 281 /** 282 * Compute a hash code for this BreakIterator 283 * @return A hash code 284 * @stable ICU 2.0 285 */ 286 virtual int32_t hashCode(void) const; 287 288 /** 289 * Returns the description used to create this iterator 290 * @return the description used to create this iterator 291 * @stable ICU 2.0 292 */ 293 virtual const UnicodeString& getRules(void) const; 294 295 //======================================================================= 296 // BreakIterator overrides 297 //======================================================================= 298 299 /** 300 * <p> 301 * Return a CharacterIterator over the text being analyzed. 302 * The returned character iterator is owned by the break iterator, and must 303 * not be deleted by the caller. Repeated calls to this function may 304 * return the same CharacterIterator. 305 * </p> 306 * <p> 307 * The returned character iterator must not be used concurrently with 308 * the break iterator. If concurrent operation is needed, clone the 309 * returned character iterator first and operate on the clone. 310 * </p> 311 * <p> 312 * When the break iterator is operating on text supplied via a UText, 313 * this function will fail. Lacking any way to signal failures, it 314 * returns an CharacterIterator containing no text. 315 * The function getUText() provides similar functionality, 316 * is reliable, and is more efficient. 317 * </p> 318 * 319 * TODO: deprecate this function? 320 * 321 * @return An iterator over the text being analyzed. 322 * @stable ICU 2.0 323 */ 324 virtual CharacterIterator& getText(void) const; 325 326 327 /** 328 * Get a UText for the text being analyzed. 329 * The returned UText is a shallow clone of the UText used internally 330 * by the break iterator implementation. It can safely be used to 331 * access the text without impacting any break iterator operations, 332 * but the underlying text itself must not be altered. 333 * 334 * @param fillIn A UText to be filled in. If NULL, a new UText will be 335 * allocated to hold the result. 336 * @param status receives any error codes. 337 * @return The current UText for this break iterator. If an input 338 * UText was provided, it will always be returned. 339 * @stable ICU 3.4 340 */ 341 virtual UText *getUText(UText *fillIn, UErrorCode &status) const; 342 343 /** 344 * Set the iterator to analyze a new piece of text. This function resets 345 * the current iteration position to the beginning of the text. 346 * @param newText An iterator over the text to analyze. The BreakIterator 347 * takes ownership of the character iterator. The caller MUST NOT delete it! 348 * @stable ICU 2.0 349 */ 350 virtual void adoptText(CharacterIterator* newText); 351 352 /** 353 * Set the iterator to analyze a new piece of text. This function resets 354 * the current iteration position to the beginning of the text. 355 * @param newText The text to analyze. 356 * @stable ICU 2.0 357 */ 358 virtual void setText(const UnicodeString& newText); 359 360 /** 361 * Reset the break iterator to operate over the text represented by 362 * the UText. The iterator position is reset to the start. 363 * 364 * This function makes a shallow clone of the supplied UText. This means 365 * that the caller is free to immediately close or otherwise reuse the 366 * Utext that was passed as a parameter, but that the underlying text itself 367 * must not be altered while being referenced by the break iterator. 368 * 369 * @param text The UText used to change the text. 370 * @param status Receives any error codes. 371 * @stable ICU 3.4 372 */ 373 virtual void setText(UText *text, UErrorCode &status); 374 375 /** 376 * Sets the current iteration position to the beginning of the text. 377 * @return The offset of the beginning of the text. 378 * @stable ICU 2.0 379 */ 380 virtual int32_t first(void); 381 382 /** 383 * Sets the current iteration position to the end of the text. 384 * @return The text's past-the-end offset. 385 * @stable ICU 2.0 386 */ 387 virtual int32_t last(void); 388 389 /** 390 * Advances the iterator either forward or backward the specified number of steps. 391 * Negative values move backward, and positive values move forward. This is 392 * equivalent to repeatedly calling next() or previous(). 393 * @param n The number of steps to move. The sign indicates the direction 394 * (negative is backwards, and positive is forwards). 395 * @return The character offset of the boundary position n boundaries away from 396 * the current one. 397 * @stable ICU 2.0 398 */ 399 virtual int32_t next(int32_t n); 400 401 /** 402 * Advances the iterator to the next boundary position. 403 * @return The position of the first boundary after this one. 404 * @stable ICU 2.0 405 */ 406 virtual int32_t next(void); 407 408 /** 409 * Moves the iterator backwards, to the last boundary preceding this one. 410 * @return The position of the last boundary position preceding this one. 411 * @stable ICU 2.0 412 */ 413 virtual int32_t previous(void); 414 415 /** 416 * Sets the iterator to refer to the first boundary position following 417 * the specified position. 418 * @param offset The position from which to begin searching for a break position. 419 * @return The position of the first break after the current position. 420 * @stable ICU 2.0 421 */ 422 virtual int32_t following(int32_t offset); 423 424 /** 425 * Sets the iterator to refer to the last boundary position before the 426 * specified position. 427 * @param offset The position to begin searching for a break from. 428 * @return The position of the last boundary before the starting position. 429 * @stable ICU 2.0 430 */ 431 virtual int32_t preceding(int32_t offset); 432 433 /** 434 * Returns true if the specfied position is a boundary position. As a side 435 * effect, leaves the iterator pointing to the first boundary position at 436 * or after "offset". 437 * @param offset the offset to check. 438 * @return True if "offset" is a boundary position. 439 * @stable ICU 2.0 440 */ 441 virtual UBool isBoundary(int32_t offset); 442 443 /** 444 * Returns the current iteration position. 445 * @return The current iteration position. 446 * @stable ICU 2.0 447 */ 448 virtual int32_t current(void) const; 449 450 451 /** 452 * Return the status tag from the break rule that determined the most recently 453 * returned break position. For break rules that do not specify a 454 * status, a default value of 0 is returned. If more than one break rule 455 * would cause a boundary to be located at some position in the text, 456 * the numerically largest of the applicable status values is returned. 457 * <p> 458 * Of the standard types of ICU break iterators, only word break and 459 * line break provide status values. The values are defined in 460 * the header file ubrk.h. For Word breaks, the status allows distinguishing between words 461 * that contain alphabetic letters, "words" that appear to be numbers, 462 * punctuation and spaces, words containing ideographic characters, and 463 * more. For Line Break, the status distinguishes between hard (mandatory) breaks 464 * and soft (potential) break positions. 465 * <p> 466 * <code>getRuleStatus()</code> can be called after obtaining a boundary 467 * position from <code>next()</code>, <code>previous()</code>, or 468 * any other break iterator functions that returns a boundary position. 469 * <p> 470 * When creating custom break rules, one is free to define whatever 471 * status values may be convenient for the application. 472 * <p> 473 * Note: this function is not thread safe. It should not have been 474 * declared const, and the const remains only for compatibility 475 * reasons. (The function is logically const, but not bit-wise const). 476 * <p> 477 * @return the status from the break rule that determined the most recently 478 * returned break position. 479 * 480 * @see UWordBreak 481 * @stable ICU 2.2 482 */ 483 virtual int32_t getRuleStatus() const; 484 485 /** 486 * Get the status (tag) values from the break rule(s) that determined the most 487 * recently returned break position. 488 * <p> 489 * The returned status value(s) are stored into an array provided by the caller. 490 * The values are stored in sorted (ascending) order. 491 * If the capacity of the output array is insufficient to hold the data, 492 * the output will be truncated to the available length, and a 493 * U_BUFFER_OVERFLOW_ERROR will be signaled. 494 * 495 * @param fillInVec an array to be filled in with the status values. 496 * @param capacity the length of the supplied vector. A length of zero causes 497 * the function to return the number of status values, in the 498 * normal way, without attemtping to store any values. 499 * @param status receives error codes. 500 * @return The number of rule status values from rules that determined 501 * the most recent boundary returned by the break iterator. 502 * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value 503 * is the total number of status values that were available, 504 * not the reduced number that were actually returned. 505 * @see getRuleStatus 506 * @stable ICU 3.0 507 */ 508 virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status); 509 510 /** 511 * Returns a unique class ID POLYMORPHICALLY. Pure virtual override. 512 * This method is to implement a simple version of RTTI, since not all 513 * C++ compilers support genuine RTTI. Polymorphic operator==() and 514 * clone() methods call this method. 515 * 516 * @return The class ID for this object. All objects of a 517 * given class have the same class ID. Objects of 518 * other classes have different class IDs. 519 * @stable ICU 2.0 520 */ 521 virtual UClassID getDynamicClassID(void) const; 522 523 /** 524 * Returns the class ID for this class. This is useful only for 525 * comparing to a return value from getDynamicClassID(). For example: 526 * 527 * Base* polymorphic_pointer = createPolymorphicObject(); 528 * if (polymorphic_pointer->getDynamicClassID() == 529 * Derived::getStaticClassID()) ... 530 * 531 * @return The class ID for all objects of this class. 532 * @stable ICU 2.0 533 */ 534 static UClassID U_EXPORT2 getStaticClassID(void); 535 536 /* 537 * Create a clone (copy) of this break iterator in memory provided 538 * by the caller. The idea is to increase performance by avoiding 539 * a storage allocation. Use of this functoin is NOT RECOMMENDED. 540 * Performance gains are minimal, and correct buffer management is 541 * tricky. Use clone() instead. 542 * 543 * @param stackBuffer The pointer to the memory into which the cloned object 544 * should be placed. If NULL, allocate heap memory 545 * for the cloned object. 546 * @param BufferSize The size of the buffer. If zero, return the required 547 * buffer size, but do not clone the object. If the 548 * size was too small (but not zero), allocate heap 549 * storage for the cloned object. 550 * 551 * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be 552 * returned if the the provided buffer was too small, and 553 * the clone was therefore put on the heap. 554 * 555 * @return Pointer to the clone object. This may differ from the stackBuffer 556 * address if the byte alignment of the stack buffer was not suitable 557 * or if the stackBuffer was too small to hold the clone. 558 * @stable ICU 2.0 559 */ 560 virtual BreakIterator * createBufferClone(void *stackBuffer, 561 int32_t &BufferSize, 562 UErrorCode &status); 563 564 565 /** 566 * Return the binary form of compiled break rules, 567 * which can then be used to create a new break iterator at some 568 * time in the future. Creating a break iterator from pre-compiled rules 569 * is much faster than building one from the source form of the 570 * break rules. 571 * 572 * The binary data can only be used with the same version of ICU 573 * and on the same platform type (processor endian-ness) 574 * 575 * @param length Returns the length of the binary data. (Out paramter.) 576 * 577 * @return A pointer to the binary (compiled) rule data. The storage 578 * belongs to the RulesBasedBreakIterator object, not the 579 * caller, and must not be modified or deleted. 580 * @internal 581 */ 582 virtual const uint8_t *getBinaryRules(uint32_t &length); 583 584 585 protected: 586 //======================================================================= 587 // implementation 588 //======================================================================= 589 /** 590 * Dumps caches and performs other actions associated with a complete change 591 * in text or iteration position. 592 * @internal 593 */ 594 virtual void reset(void); 595 596 #if 0 597 /** 598 * Return true if the category lookup for this char 599 * indicates that it is in the set of dictionary lookup chars. 600 * This function is intended for use by dictionary based break iterators. 601 * @return true if the category lookup for this char 602 * indicates that it is in the set of dictionary lookup chars. 603 * @internal 604 */ 605 virtual UBool isDictionaryChar(UChar32); 606 607 /** 608 * Get the type of the break iterator. 609 * @internal 610 */ 611 virtual int32_t getBreakType() const; 612 #endif 613 614 /** 615 * Set the type of the break iterator. 616 * @internal 617 */ 618 virtual void setBreakType(int32_t type); 619 620 /** 621 * Common initialization function, used by constructors and bufferClone. 622 * (Also used by DictionaryBasedBreakIterator::createBufferClone().) 623 * @internal 624 */ 625 void init(); 626 627 private: 628 629 /** 630 * This method backs the iterator back up to a "safe position" in the text. 631 * This is a position that we know, without any context, must be a break position. 632 * The various calling methods then iterate forward from this safe position to 633 * the appropriate position to return. (For more information, see the description 634 * of buildBackwardsStateTable() in RuleBasedBreakIterator.Builder.) 635 * @param statetable state table used of moving backwards 636 * @internal 637 */ 638 int32_t handlePrevious(const RBBIStateTable *statetable); 639 640 /** 641 * This method is the actual implementation of the next() method. All iteration 642 * vectors through here. This method initializes the state machine to state 1 643 * and advances through the text character by character until we reach the end 644 * of the text or the state machine transitions to state 0. We update our return 645 * value every time the state machine passes through a possible end state. 646 * @param statetable state table used of moving forwards 647 * @internal 648 */ 649 int32_t handleNext(const RBBIStateTable *statetable); 650 651 protected: 652 653 /** 654 * This is the function that actually implements dictionary-based 655 * breaking. Covering at least the range from startPos to endPos, 656 * it checks for dictionary characters, and if it finds them determines 657 * the appropriate object to deal with them. It may cache found breaks in 658 * fCachedBreakPositions as it goes. It may well also look at text outside 659 * the range startPos to endPos. 660 * If going forward, endPos is the normal Unicode break result, and 661 * if goind in reverse, startPos is the normal Unicode break result 662 * @param startPos The start position of a range of text 663 * @param endPos The end position of a range of text 664 * @param reverse The call is for the reverse direction 665 * @internal 666 */ 667 int32_t checkDictionary(int32_t startPos, int32_t endPos, UBool reverse); 668 669 private: 670 671 /** 672 * This function returns the appropriate LanguageBreakEngine for a 673 * given character c. 674 * @param c A character in the dictionary set 675 * @internal 676 */ 677 const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c); 678 679 /** 680 * @internal 681 */ 682 void makeRuleStatusValid(); 683 684 }; 685 686 //------------------------------------------------------------------------------ 687 // 688 // Inline Functions Definitions ... 689 // 690 //------------------------------------------------------------------------------ 691 692 inline UBool RuleBasedBreakIterator::operator!=(const BreakIterator& that) const { 693 return !operator==(that); 694 } 695 696 U_NAMESPACE_END 697 698 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */ 699 700 #endif 701