1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
4 ***************************************************************************
5 * Copyright (C) 1999-2016 International Business Machines Corporation *
6 * and others. All rights reserved. *
7 ***************************************************************************
9 **********************************************************************
10 * Date Name Description
11 * 10/22/99 alan Creation.
12 * 11/11/99 rgillam Complete port from Java.
13 **********************************************************************
19 #include "unicode/utypes.h"
23 * \brief C++ API: Rule Based Break Iterator
26 #if !UCONFIG_NO_BREAK_ITERATION
28 #include "unicode/brkiter.h"
29 #include "unicode/udata.h"
30 #include "unicode/parseerr.h"
31 #include "unicode/schriter.h"
32 #include "unicode/uchriter.h"
40 struct RBBIDataHeader;
41 class RuleBasedBreakIteratorTables;
43 class RBBIDataWrapper;
45 class LanguageBreakEngine;
46 class UnhandledEngine;
47 struct RBBIStateTable;
54 * A subclass of BreakIterator whose behavior is specified using a list of rules.
55 * <p>Instances of this class are most commonly created by the factory methods of
56 * BreakIterator::createWordInstance(), BreakIterator::createLineInstance(), etc.,
57 * and then used via the abstract API in class BreakIterator</p>
59 * <p>See the ICU User Guide for information on Break Iterator Rules.</p>
61 * <p>This class is not intended to be subclassed.</p>
63 class U_COMMON_API RuleBasedBreakIterator /*U_FINAL*/ : public BreakIterator {
67 * The UText through which this BreakIterator accesses the text
73 * A character iterator that refers to the same text as the UText, above.
74 * Only included for compatibility with old API, which was based on CharacterIterators.
75 * Value may be adopted from outside, or one of fSCharIter or fDCharIter, below.
77 CharacterIterator *fCharIter;
80 * When the input text is provided by a UnicodeString, this will point to
81 * a characterIterator that wraps that data. Needed only for the
82 * implementation of getText(), a backwards compatibility issue.
84 StringCharacterIterator *fSCharIter;
87 * When the input text is provided by a UText, this
88 * dummy CharacterIterator over an empty string will
89 * be returned from getText()
91 UCharCharacterIterator *fDCharIter;
94 * The rule data for this BreakIterator instance
97 RBBIDataWrapper *fData;
99 /** Index of the Rule {tag} values for the most recent match.
102 int32_t fLastRuleStatusIndex;
105 * Rule tag value valid flag.
106 * Some iterator operations don't intrinsically set the correct tag value.
107 * This flag lets us lazily compute the value if we are ever asked for it.
110 UBool fLastStatusIndexValid;
113 * Counter for the number of characters encountered with the "dictionary"
117 uint32_t fDictionaryCharCount;
120 * When a range of characters is divided up using the dictionary, the break
121 * positions that are discovered are stored here, preventing us from having
122 * to use either the dictionary or the state table again until the iterator
123 * leaves this range of text. Has the most impact for line breaking.
126 int32_t* fCachedBreakPositions;
129 * The number of elements in fCachedBreakPositions
132 int32_t fNumCachedBreakPositions;
135 * if fCachedBreakPositions is not null, this indicates which item in the
136 * cache the current iteration position refers to
139 int32_t fPositionInCache;
143 * If present, UStack of LanguageBreakEngine objects that might handle
144 * dictionary characters. Searched from top to bottom to find an object to
145 * handle a given character.
148 UStack *fLanguageBreakEngines;
152 * If present, the special LanguageBreakEngine used for handling
153 * characters that are in the dictionary set, but not handled by any
154 * LangugageBreakEngine.
157 UnhandledEngine *fUnhandledBreakEngine;
161 * The type of the break iterator, or -1 if it has not been set.
166 //=======================================================================
168 //=======================================================================
171 * Constructor from a flattened set of RBBI data in malloced memory.
172 * RulesBasedBreakIterators built from a custom set of rules
173 * are created via this constructor; the rules are compiled
174 * into memory, then the break iterator is constructed here.
176 * The break iterator adopts the memory, and will
180 RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status);
183 friend class RBBIRuleBuilder;
185 friend class BreakIterator;
191 /** Default constructor. Creates an empty shell of an iterator, with no
192 * rules or text to iterate over. Object can subsequently be assigned to.
195 RuleBasedBreakIterator();
198 * Copy constructor. Will produce a break iterator with the same behavior,
199 * and which iterates over the same text, as the one passed in.
200 * @param that The RuleBasedBreakIterator passed to be copied
203 RuleBasedBreakIterator(const RuleBasedBreakIterator& that);
206 * Construct a RuleBasedBreakIterator from a set of rules supplied as a string.
207 * @param rules The break rules to be used.
208 * @param parseError In the event of a syntax error in the rules, provides the location
209 * within the rules of the problem.
210 * @param status Information on any errors encountered.
213 RuleBasedBreakIterator( const UnicodeString &rules,
214 UParseError &parseError,
218 * Contruct a RuleBasedBreakIterator from a set of precompiled binary rules.
219 * Binary rules are obtained from RulesBasedBreakIterator::getBinaryRules().
220 * Construction of a break iterator in this way is substantially faster than
221 * constuction from source rules.
223 * Ownership of the storage containing the compiled rules remains with the
224 * caller of this function. The compiled rules must not be modified or
225 * deleted during the life of the break iterator.
227 * The compiled rules are not compatible across different major versions of ICU.
228 * The compiled rules are comaptible only between machines with the same
229 * byte ordering (little or big endian) and the same base character set family
232 * @see #getBinaryRules
233 * @param compiledRules A pointer to the compiled break rules to be used.
234 * @param ruleLength The length of the compiled break rules, in bytes. This
235 * corresponds to the length value produced by getBinaryRules().
236 * @param status Information on any errors encountered, including invalid
240 RuleBasedBreakIterator(const uint8_t *compiledRules,
245 * This constructor uses the udata interface to create a BreakIterator
246 * whose internal tables live in a memory-mapped file. "image" is an
247 * ICU UDataMemory handle for the pre-compiled break iterator tables.
248 * @param image handle to the memory image for the break iterator data.
249 * Ownership of the UDataMemory handle passes to the Break Iterator,
250 * which will be responsible for closing it when it is no longer needed.
251 * @param status Information on any errors encountered.
253 * @see #getBinaryRules
256 RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status);
262 virtual ~RuleBasedBreakIterator();
265 * Assignment operator. Sets this iterator to have the same behavior,
266 * and iterate over the same text, as the one passed in.
267 * @param that The RuleBasedBreakItertor passed in
268 * @return the newly created RuleBasedBreakIterator
271 RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that);
274 * Equality operator. Returns TRUE if both BreakIterators are of the
275 * same class, have the same behavior, and iterate over the same text.
276 * @param that The BreakIterator to be compared for equality
277 * @return TRUE if both BreakIterators are of the
278 * same class, have the same behavior, and iterate over the same text.
281 virtual UBool operator==(const BreakIterator& that) const;
284 * Not-equal operator. If operator== returns TRUE, this returns FALSE,
286 * @param that The BreakIterator to be compared for inequality
287 * @return TRUE if both BreakIterators are not same.
290 UBool operator!=(const BreakIterator& that) const;
293 * Returns a newly-constructed RuleBasedBreakIterator with the same
294 * behavior, and iterating over the same text, as this one.
295 * Differs from the copy constructor in that it is polymorphic, and
296 * will correctly clone (copy) a derived class.
297 * clone() is thread safe. Multiple threads may simultaeneously
298 * clone the same source break iterator.
299 * @return a newly-constructed RuleBasedBreakIterator
302 virtual BreakIterator* clone() const;
305 * Compute a hash code for this BreakIterator
306 * @return A hash code
309 virtual int32_t hashCode(void) const;
312 * Returns the description used to create this iterator
313 * @return the description used to create this iterator
316 virtual const UnicodeString& getRules(void) const;
318 //=======================================================================
319 // BreakIterator overrides
320 //=======================================================================
324 * Return a CharacterIterator over the text being analyzed.
325 * The returned character iterator is owned by the break iterator, and must
326 * not be deleted by the caller. Repeated calls to this function may
327 * return the same CharacterIterator.
330 * The returned character iterator must not be used concurrently with
331 * the break iterator. If concurrent operation is needed, clone the
332 * returned character iterator first and operate on the clone.
335 * When the break iterator is operating on text supplied via a UText,
336 * this function will fail. Lacking any way to signal failures, it
337 * returns an CharacterIterator containing no text.
338 * The function getUText() provides similar functionality,
339 * is reliable, and is more efficient.
342 * TODO: deprecate this function?
344 * @return An iterator over the text being analyzed.
347 virtual CharacterIterator& getText(void) const;
351 * Get a UText for the text being analyzed.
352 * The returned UText is a shallow clone of the UText used internally
353 * by the break iterator implementation. It can safely be used to
354 * access the text without impacting any break iterator operations,
355 * but the underlying text itself must not be altered.
357 * @param fillIn A UText to be filled in. If NULL, a new UText will be
358 * allocated to hold the result.
359 * @param status receives any error codes.
360 * @return The current UText for this break iterator. If an input
361 * UText was provided, it will always be returned.
364 virtual UText *getUText(UText *fillIn, UErrorCode &status) const;
367 * Set the iterator to analyze a new piece of text. This function resets
368 * the current iteration position to the beginning of the text.
369 * @param newText An iterator over the text to analyze. The BreakIterator
370 * takes ownership of the character iterator. The caller MUST NOT delete it!
373 virtual void adoptText(CharacterIterator* newText);
376 * Set the iterator to analyze a new piece of text. This function resets
377 * the current iteration position to the beginning of the text.
379 * The BreakIterator will retain a reference to the supplied string.
380 * The caller must not modify or delete the text while the BreakIterator
381 * retains the reference.
383 * @param newText The text to analyze.
386 virtual void setText(const UnicodeString& newText);
389 * Reset the break iterator to operate over the text represented by
390 * the UText. The iterator position is reset to the start.
392 * This function makes a shallow clone of the supplied UText. This means
393 * that the caller is free to immediately close or otherwise reuse the
394 * Utext that was passed as a parameter, but that the underlying text itself
395 * must not be altered while being referenced by the break iterator.
397 * @param text The UText used to change the text.
398 * @param status Receives any error codes.
401 virtual void setText(UText *text, UErrorCode &status);
404 * Sets the current iteration position to the beginning of the text, position zero.
405 * @return The offset of the beginning of the text, zero.
408 virtual int32_t first(void);
411 * Sets the current iteration position to the end of the text.
412 * @return The text's past-the-end offset.
415 virtual int32_t last(void);
418 * Advances the iterator either forward or backward the specified number of steps.
419 * Negative values move backward, and positive values move forward. This is
420 * equivalent to repeatedly calling next() or previous().
421 * @param n The number of steps to move. The sign indicates the direction
422 * (negative is backwards, and positive is forwards).
423 * @return The character offset of the boundary position n boundaries away from
427 virtual int32_t next(int32_t n);
430 * Advances the iterator to the next boundary position.
431 * @return The position of the first boundary after this one.
434 virtual int32_t next(void);
437 * Moves the iterator backwards, to the last boundary preceding this one.
438 * @return The position of the last boundary position preceding this one.
441 virtual int32_t previous(void);
444 * Sets the iterator to refer to the first boundary position following
445 * the specified position.
446 * @param offset The position from which to begin searching for a break position.
447 * @return The position of the first break after the current position.
450 virtual int32_t following(int32_t offset);
453 * Sets the iterator to refer to the last boundary position before the
454 * specified position.
455 * @param offset The position to begin searching for a break from.
456 * @return The position of the last boundary before the starting position.
459 virtual int32_t preceding(int32_t offset);
462 * Returns true if the specfied position is a boundary position. As a side
463 * effect, leaves the iterator pointing to the first boundary position at
465 * @param offset the offset to check.
466 * @return True if "offset" is a boundary position.
469 virtual UBool isBoundary(int32_t offset);
472 * Returns the current iteration position.
473 * @return The current iteration position.
476 virtual int32_t current(void) const;
480 * Return the status tag from the break rule that determined the most recently
481 * returned break position. For break rules that do not specify a
482 * status, a default value of 0 is returned. If more than one break rule
483 * would cause a boundary to be located at some position in the text,
484 * the numerically largest of the applicable status values is returned.
486 * Of the standard types of ICU break iterators, only word break and
487 * line break provide status values. The values are defined in
488 * the header file ubrk.h. For Word breaks, the status allows distinguishing between words
489 * that contain alphabetic letters, "words" that appear to be numbers,
490 * punctuation and spaces, words containing ideographic characters, and
491 * more. For Line Break, the status distinguishes between hard (mandatory) breaks
492 * and soft (potential) break positions.
494 * <code>getRuleStatus()</code> can be called after obtaining a boundary
495 * position from <code>next()</code>, <code>previous()</code>, or
496 * any other break iterator functions that returns a boundary position.
498 * When creating custom break rules, one is free to define whatever
499 * status values may be convenient for the application.
501 * Note: this function is not thread safe. It should not have been
502 * declared const, and the const remains only for compatibility
503 * reasons. (The function is logically const, but not bit-wise const).
505 * @return the status from the break rule that determined the most recently
506 * returned break position.
511 virtual int32_t getRuleStatus() const;
514 * Get the status (tag) values from the break rule(s) that determined the most
515 * recently returned break position.
517 * The returned status value(s) are stored into an array provided by the caller.
518 * The values are stored in sorted (ascending) order.
519 * If the capacity of the output array is insufficient to hold the data,
520 * the output will be truncated to the available length, and a
521 * U_BUFFER_OVERFLOW_ERROR will be signaled.
523 * @param fillInVec an array to be filled in with the status values.
524 * @param capacity the length of the supplied vector. A length of zero causes
525 * the function to return the number of status values, in the
526 * normal way, without attemtping to store any values.
527 * @param status receives error codes.
528 * @return The number of rule status values from rules that determined
529 * the most recent boundary returned by the break iterator.
530 * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value
531 * is the total number of status values that were available,
532 * not the reduced number that were actually returned.
536 virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status);
539 * Returns a unique class ID POLYMORPHICALLY. Pure virtual override.
540 * This method is to implement a simple version of RTTI, since not all
541 * C++ compilers support genuine RTTI. Polymorphic operator==() and
542 * clone() methods call this method.
544 * @return The class ID for this object. All objects of a
545 * given class have the same class ID. Objects of
546 * other classes have different class IDs.
549 virtual UClassID getDynamicClassID(void) const;
552 * Returns the class ID for this class. This is useful only for
553 * comparing to a return value from getDynamicClassID(). For example:
555 * Base* polymorphic_pointer = createPolymorphicObject();
556 * if (polymorphic_pointer->getDynamicClassID() ==
557 * Derived::getStaticClassID()) ...
559 * @return The class ID for all objects of this class.
562 static UClassID U_EXPORT2 getStaticClassID(void);
565 * Deprecated functionality. Use clone() instead.
567 * Create a clone (copy) of this break iterator in memory provided
568 * by the caller. The idea is to increase performance by avoiding
569 * a storage allocation. Use of this functoin is NOT RECOMMENDED.
570 * Performance gains are minimal, and correct buffer management is
571 * tricky. Use clone() instead.
573 * @param stackBuffer The pointer to the memory into which the cloned object
574 * should be placed. If NULL, allocate heap memory
575 * for the cloned object.
576 * @param BufferSize The size of the buffer. If zero, return the required
577 * buffer size, but do not clone the object. If the
578 * size was too small (but not zero), allocate heap
579 * storage for the cloned object.
581 * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be
582 * returned if the the provided buffer was too small, and
583 * the clone was therefore put on the heap.
585 * @return Pointer to the clone object. This may differ from the stackBuffer
586 * address if the byte alignment of the stack buffer was not suitable
587 * or if the stackBuffer was too small to hold the clone.
588 * @deprecated ICU 52. Use clone() instead.
590 virtual BreakIterator * createBufferClone(void *stackBuffer,
596 * Return the binary form of compiled break rules,
597 * which can then be used to create a new break iterator at some
598 * time in the future. Creating a break iterator from pre-compiled rules
599 * is much faster than building one from the source form of the
602 * The binary data can only be used with the same version of ICU
603 * and on the same platform type (processor endian-ness)
605 * @param length Returns the length of the binary data. (Out paramter.)
607 * @return A pointer to the binary (compiled) rule data. The storage
608 * belongs to the RulesBasedBreakIterator object, not the
609 * caller, and must not be modified or deleted.
612 virtual const uint8_t *getBinaryRules(uint32_t &length);
615 * Set the subject text string upon which the break iterator is operating
616 * without changing any other aspect of the matching state.
617 * The new and previous text strings must have the same content.
619 * This function is intended for use in environments where ICU is operating on
620 * strings that may move around in memory. It provides a mechanism for notifying
621 * ICU that the string has been relocated, and providing a new UText to access the
622 * string in its new position.
624 * Note that the break iterator implementation never copies the underlying text
625 * of a string being processed, but always operates directly on the original text
626 * provided by the user. Refreshing simply drops the references to the old text
627 * and replaces them with references to the new.
629 * Caution: this function is normally used only by very specialized,
630 * system-level code. One example use case is with garbage collection that moves
631 * the text in memory.
633 * @param input The new (moved) text string.
634 * @param status Receives errors detected by this function.
639 virtual RuleBasedBreakIterator &refreshInputText(UText *input, UErrorCode &status);
643 //=======================================================================
645 //=======================================================================
647 * Dumps caches and performs other actions associated with a complete change
648 * in text or iteration position.
654 * Set the type of the break iterator.
657 void setBreakType(int32_t type);
660 * Common initialization function, used by constructors and bufferClone.
666 * This method backs the iterator back up to a "safe position" in the text.
667 * This is a position that we know, without any context, must be a break position.
668 * The various calling methods then iterate forward from this safe position to
669 * the appropriate position to return. (For more information, see the description
670 * of buildBackwardsStateTable() in RuleBasedBreakIterator.Builder.)
671 * @param statetable state table used of moving backwards
674 int32_t handlePrevious(const RBBIStateTable *statetable);
677 * This method is the actual implementation of the next() method. All iteration
678 * vectors through here. This method initializes the state machine to state 1
679 * and advances through the text character by character until we reach the end
680 * of the text or the state machine transitions to state 0. We update our return
681 * value every time the state machine passes through a possible end state.
682 * @param statetable state table used of moving forwards
685 int32_t handleNext(const RBBIStateTable *statetable);
689 * This is the function that actually implements dictionary-based
690 * breaking. Covering at least the range from startPos to endPos,
691 * it checks for dictionary characters, and if it finds them determines
692 * the appropriate object to deal with them. It may cache found breaks in
693 * fCachedBreakPositions as it goes. It may well also look at text outside
694 * the range startPos to endPos.
695 * If going forward, endPos is the normal Unicode break result, and
696 * if goind in reverse, startPos is the normal Unicode break result
697 * @param startPos The start position of a range of text
698 * @param endPos The end position of a range of text
699 * @param reverse The call is for the reverse direction
702 int32_t checkDictionary(int32_t startPos, int32_t endPos, UBool reverse);
706 * This function returns the appropriate LanguageBreakEngine for a
708 * @param c A character in the dictionary set
711 const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c);
716 void makeRuleStatusValid();
720 //------------------------------------------------------------------------------
722 // Inline Functions Definitions ...
724 //------------------------------------------------------------------------------
726 inline UBool RuleBasedBreakIterator::operator!=(const BreakIterator& that) const {
727 return !operator==(that);
732 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */