2 ***************************************************************************
3 * Copyright (C) 1999-2008 International Business Machines Corporation *
4 * and others. All rights reserved. *
5 ***************************************************************************
7 **********************************************************************
8 * Date Name Description
9 * 10/22/99 alan Creation.
10 * 11/11/99 rgillam Complete port from Java.
11 **********************************************************************
17 #include "unicode/utypes.h"
21 * \brief C++ API: Rule Based Break Iterator
24 #if !UCONFIG_NO_BREAK_ITERATION
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"
38 struct RBBIDataHeader
;
39 class RuleBasedBreakIteratorTables
;
41 class RBBIDataWrapper
;
43 class LanguageBreakEngine
;
44 class UnhandledEngine
;
45 struct RBBIStateTable
;
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>
57 * <p>See the ICU User Guide for information on Break Iterator Rules.</p>
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>
65 class U_COMMON_API RuleBasedBreakIterator
: public BreakIterator
{
69 * The UText through which this BreakIterator accesses the text
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.
79 CharacterIterator
*fCharIter
;
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.
86 StringCharacterIterator
*fSCharIter
;
89 * When the input text is provided by a UText, this
90 * dummy CharacterIterator over an empty string will
91 * be returned from getText()
93 UCharCharacterIterator
*fDCharIter
;
96 * The rule data for this BreakIterator instance
99 RBBIDataWrapper
*fData
;
101 /** Index of the Rule {tag} values for the most recent match.
104 int32_t fLastRuleStatusIndex
;
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.
112 UBool fLastStatusIndexValid
;
115 * Counter for the number of characters encountered with the "dictionary"
119 uint32_t fDictionaryCharCount
;
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.
128 int32_t* fCachedBreakPositions
;
131 * The number of elements in fCachedBreakPositions
134 int32_t fNumCachedBreakPositions
;
137 * if fCachedBreakPositions is not null, this indicates which item in the
138 * cache the current iteration position refers to
141 int32_t fPositionInCache
;
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.
150 UStack
*fLanguageBreakEngines
;
154 * If present, the special LanguageBreakEngine used for handling
155 * characters that are in the dictionary set, but not handled by any
156 * LangugageBreakEngine.
159 UnhandledEngine
*fUnhandledBreakEngine
;
163 * The type of the break iterator, or -1 if it has not been set.
169 //=======================================================================
171 //=======================================================================
174 * Constant to be used in the constructor
175 * RuleBasedBreakIterator(RBBIDataHeader*, EDontAdopt, UErrorCode &);
176 * which does not adopt the memory indicated by the RBBIDataHeader*
186 * Constructor from a flattened set of RBBI data in malloced memory.
187 * RulesBasedBreakIterators built from a custom set of rules
188 * are created via this constructor; the rules are compiled
189 * into memory, then the break iterator is constructed here.
191 * The break iterator adopts the memory, and will
195 RuleBasedBreakIterator(RBBIDataHeader
* data
, UErrorCode
&status
);
198 * Constructor from a flattened set of RBBI data in memory which need not
199 * be malloced (e.g. it may be a memory-mapped file, etc.).
201 * This version does not adopt the memory, and does not
205 RuleBasedBreakIterator(const RBBIDataHeader
* data
, enum EDontAdopt dontAdopt
, UErrorCode
&status
);
208 friend class RBBIRuleBuilder
;
210 friend class BreakIterator
;
216 /** Default constructor. Creates an empty shell of an iterator, with no
217 * rules or text to iterate over. Object can subsequently be assigned to.
220 RuleBasedBreakIterator();
223 * Copy constructor. Will produce a break iterator with the same behavior,
224 * and which iterates over the same text, as the one passed in.
225 * @param that The RuleBasedBreakIterator passed to be copied
228 RuleBasedBreakIterator(const RuleBasedBreakIterator
& that
);
231 * Construct a RuleBasedBreakIterator from a set of rules supplied as a string.
232 * @param rules The break rules to be used.
233 * @param parseError In the event of a syntax error in the rules, provides the location
234 * within the rules of the problem.
235 * @param status Information on any errors encountered.
238 RuleBasedBreakIterator( const UnicodeString
&rules
,
239 UParseError
&parseError
,
244 * This constructor uses the udata interface to create a BreakIterator
245 * whose internal tables live in a memory-mapped file. "image" is an
246 * ICU UDataMemory handle for the pre-compiled break iterator tables.
247 * @param image handle to the memory image for the break iterator data.
248 * Ownership of the UDataMemory handle passes to the Break Iterator,
249 * which will be responsible for closing it when it is no longer needed.
250 * @param status Information on any errors encountered.
252 * @see #getBinaryRules
255 RuleBasedBreakIterator(UDataMemory
* image
, UErrorCode
&status
);
261 virtual ~RuleBasedBreakIterator();
264 * Assignment operator. Sets this iterator to have the same behavior,
265 * and iterate over the same text, as the one passed in.
266 * @param that The RuleBasedBreakItertor passed in
267 * @return the newly created RuleBasedBreakIterator
270 RuleBasedBreakIterator
& operator=(const RuleBasedBreakIterator
& that
);
273 * Equality operator. Returns TRUE if both BreakIterators are of the
274 * same class, have the same behavior, and iterate over the same text.
275 * @param that The BreakIterator to be compared for equality
276 * @return TRUE if both BreakIterators are of the
277 * same class, have the same behavior, and iterate over the same text.
280 virtual UBool
operator==(const BreakIterator
& that
) const;
283 * Not-equal operator. If operator== returns TRUE, this returns FALSE,
285 * @param that The BreakIterator to be compared for inequality
286 * @return TRUE if both BreakIterators are not same.
289 UBool
operator!=(const BreakIterator
& that
) const;
292 * Returns a newly-constructed RuleBasedBreakIterator with the same
293 * behavior, and iterating over the same text, as this one.
294 * Differs from the copy constructor in that it is polymorphic, and
295 * will correctly clone (copy) a derived class.
296 * clone() is thread safe. Multiple threads may simultaeneously
297 * clone the same source break iterator.
298 * @return a newly-constructed RuleBasedBreakIterator
301 virtual BreakIterator
* clone() const;
304 * Compute a hash code for this BreakIterator
305 * @return A hash code
308 virtual int32_t hashCode(void) const;
311 * Returns the description used to create this iterator
312 * @return the description used to create this iterator
315 virtual const UnicodeString
& getRules(void) const;
317 //=======================================================================
318 // BreakIterator overrides
319 //=======================================================================
323 * Return a CharacterIterator over the text being analyzed.
324 * The returned character iterator is owned by the break iterator, and must
325 * not be deleted by the caller. Repeated calls to this function may
326 * return the same CharacterIterator.
329 * The returned character iterator must not be used concurrently with
330 * the break iterator. If concurrent operation is needed, clone the
331 * returned character iterator first and operate on the clone.
334 * When the break iterator is operating on text supplied via a UText,
335 * this function will fail. Lacking any way to signal failures, it
336 * returns an CharacterIterator containing no text.
337 * The function getUText() provides similar functionality,
338 * is reliable, and is more efficient.
341 * TODO: deprecate this function?
343 * @return An iterator over the text being analyzed.
346 virtual CharacterIterator
& getText(void) const;
350 * Get a UText for the text being analyzed.
351 * The returned UText is a shallow clone of the UText used internally
352 * by the break iterator implementation. It can safely be used to
353 * access the text without impacting any break iterator operations,
354 * but the underlying text itself must not be altered.
356 * @param fillIn A UText to be filled in. If NULL, a new UText will be
357 * allocated to hold the result.
358 * @param status receives any error codes.
359 * @return The current UText for this break iterator. If an input
360 * UText was provided, it will always be returned.
363 virtual UText
*getUText(UText
*fillIn
, UErrorCode
&status
) const;
366 * Set the iterator to analyze a new piece of text. This function resets
367 * the current iteration position to the beginning of the text.
368 * @param newText An iterator over the text to analyze. The BreakIterator
369 * takes ownership of the character iterator. The caller MUST NOT delete it!
372 virtual void adoptText(CharacterIterator
* newText
);
375 * Set the iterator to analyze a new piece of text. This function resets
376 * the current iteration position to the beginning of the text.
377 * @param newText The text to analyze.
380 virtual void setText(const UnicodeString
& newText
);
383 * Reset the break iterator to operate over the text represented by
384 * the UText. The iterator position is reset to the start.
386 * This function makes a shallow clone of the supplied UText. This means
387 * that the caller is free to immediately close or otherwise reuse the
388 * Utext that was passed as a parameter, but that the underlying text itself
389 * must not be altered while being referenced by the break iterator.
391 * @param text The UText used to change the text.
392 * @param status Receives any error codes.
395 virtual void setText(UText
*text
, UErrorCode
&status
);
398 * Sets the current iteration position to the beginning of the text.
399 * @return The offset of the beginning of the text.
402 virtual int32_t first(void);
405 * Sets the current iteration position to the end of the text.
406 * @return The text's past-the-end offset.
409 virtual int32_t last(void);
412 * Advances the iterator either forward or backward the specified number of steps.
413 * Negative values move backward, and positive values move forward. This is
414 * equivalent to repeatedly calling next() or previous().
415 * @param n The number of steps to move. The sign indicates the direction
416 * (negative is backwards, and positive is forwards).
417 * @return The character offset of the boundary position n boundaries away from
421 virtual int32_t next(int32_t n
);
424 * Advances the iterator to the next boundary position.
425 * @return The position of the first boundary after this one.
428 virtual int32_t next(void);
431 * Moves the iterator backwards, to the last boundary preceding this one.
432 * @return The position of the last boundary position preceding this one.
435 virtual int32_t previous(void);
438 * Sets the iterator to refer to the first boundary position following
439 * the specified position.
440 * @param offset The position from which to begin searching for a break position.
441 * @return The position of the first break after the current position.
444 virtual int32_t following(int32_t offset
);
447 * Sets the iterator to refer to the last boundary position before the
448 * specified position.
449 * @param offset The position to begin searching for a break from.
450 * @return The position of the last boundary before the starting position.
453 virtual int32_t preceding(int32_t offset
);
456 * Returns true if the specfied position is a boundary position. As a side
457 * effect, leaves the iterator pointing to the first boundary position at
459 * @param offset the offset to check.
460 * @return True if "offset" is a boundary position.
463 virtual UBool
isBoundary(int32_t offset
);
466 * Returns the current iteration position.
467 * @return The current iteration position.
470 virtual int32_t current(void) const;
474 * Return the status tag from the break rule that determined the most recently
475 * returned break position. For break rules that do not specify a
476 * status, a default value of 0 is returned. If more than one break rule
477 * would cause a boundary to be located at some position in the text,
478 * the numerically largest of the applicable status values is returned.
480 * Of the standard types of ICU break iterators, only word break and
481 * line break provide status values. The values are defined in
482 * the header file ubrk.h. For Word breaks, the status allows distinguishing between words
483 * that contain alphabetic letters, "words" that appear to be numbers,
484 * punctuation and spaces, words containing ideographic characters, and
485 * more. For Line Break, the status distinguishes between hard (mandatory) breaks
486 * and soft (potential) break positions.
488 * <code>getRuleStatus()</code> can be called after obtaining a boundary
489 * position from <code>next()</code>, <code>previous()</code>, or
490 * any other break iterator functions that returns a boundary position.
492 * When creating custom break rules, one is free to define whatever
493 * status values may be convenient for the application.
495 * Note: this function is not thread safe. It should not have been
496 * declared const, and the const remains only for compatibility
497 * reasons. (The function is logically const, but not bit-wise const).
499 * @return the status from the break rule that determined the most recently
500 * returned break position.
505 virtual int32_t getRuleStatus() const;
508 * Get the status (tag) values from the break rule(s) that determined the most
509 * recently returned break position.
511 * The returned status value(s) are stored into an array provided by the caller.
512 * The values are stored in sorted (ascending) order.
513 * If the capacity of the output array is insufficient to hold the data,
514 * the output will be truncated to the available length, and a
515 * U_BUFFER_OVERFLOW_ERROR will be signaled.
517 * @param fillInVec an array to be filled in with the status values.
518 * @param capacity the length of the supplied vector. A length of zero causes
519 * the function to return the number of status values, in the
520 * normal way, without attemtping to store any values.
521 * @param status receives error codes.
522 * @return The number of rule status values from rules that determined
523 * the most recent boundary returned by the break iterator.
524 * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value
525 * is the total number of status values that were available,
526 * not the reduced number that were actually returned.
530 virtual int32_t getRuleStatusVec(int32_t *fillInVec
, int32_t capacity
, UErrorCode
&status
);
533 * Returns a unique class ID POLYMORPHICALLY. Pure virtual override.
534 * This method is to implement a simple version of RTTI, since not all
535 * C++ compilers support genuine RTTI. Polymorphic operator==() and
536 * clone() methods call this method.
538 * @return The class ID for this object. All objects of a
539 * given class have the same class ID. Objects of
540 * other classes have different class IDs.
543 virtual UClassID
getDynamicClassID(void) const;
546 * Returns the class ID for this class. This is useful only for
547 * comparing to a return value from getDynamicClassID(). For example:
549 * Base* polymorphic_pointer = createPolymorphicObject();
550 * if (polymorphic_pointer->getDynamicClassID() ==
551 * Derived::getStaticClassID()) ...
553 * @return The class ID for all objects of this class.
556 static UClassID U_EXPORT2
getStaticClassID(void);
559 * Create a clone (copy) of this break iterator in memory provided
560 * by the caller. The idea is to increase performance by avoiding
561 * a storage allocation. Use of this functoin is NOT RECOMMENDED.
562 * Performance gains are minimal, and correct buffer management is
563 * tricky. Use clone() instead.
565 * @param stackBuffer The pointer to the memory into which the cloned object
566 * should be placed. If NULL, allocate heap memory
567 * for the cloned object.
568 * @param BufferSize The size of the buffer. If zero, return the required
569 * buffer size, but do not clone the object. If the
570 * size was too small (but not zero), allocate heap
571 * storage for the cloned object.
573 * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be
574 * returned if the the provided buffer was too small, and
575 * the clone was therefore put on the heap.
577 * @return Pointer to the clone object. This may differ from the stackBuffer
578 * address if the byte alignment of the stack buffer was not suitable
579 * or if the stackBuffer was too small to hold the clone.
582 virtual BreakIterator
* createBufferClone(void *stackBuffer
,
588 * Return the binary form of compiled break rules,
589 * which can then be used to create a new break iterator at some
590 * time in the future. Creating a break iterator from pre-compiled rules
591 * is much faster than building one from the source form of the
594 * The binary data can only be used with the same version of ICU
595 * and on the same platform type (processor endian-ness)
597 * @param length Returns the length of the binary data. (Out paramter.)
599 * @return A pointer to the binary (compiled) rule data. The storage
600 * belongs to the RulesBasedBreakIterator object, not the
601 * caller, and must not be modified or deleted.
604 virtual const uint8_t *getBinaryRules(uint32_t &length
);
608 //=======================================================================
610 //=======================================================================
612 * Dumps caches and performs other actions associated with a complete change
613 * in text or iteration position.
616 virtual void reset(void);
620 * Return true if the category lookup for this char
621 * indicates that it is in the set of dictionary lookup chars.
622 * This function is intended for use by dictionary based break iterators.
623 * @return true if the category lookup for this char
624 * indicates that it is in the set of dictionary lookup chars.
627 virtual UBool
isDictionaryChar(UChar32
);
630 * Get the type of the break iterator.
633 virtual int32_t getBreakType() const;
637 * Set the type of the break iterator.
640 virtual void setBreakType(int32_t type
);
643 * Common initialization function, used by constructors and bufferClone.
644 * (Also used by DictionaryBasedBreakIterator::createBufferClone().)
652 * This method backs the iterator back up to a "safe position" in the text.
653 * This is a position that we know, without any context, must be a break position.
654 * The various calling methods then iterate forward from this safe position to
655 * the appropriate position to return. (For more information, see the description
656 * of buildBackwardsStateTable() in RuleBasedBreakIterator.Builder.)
657 * @param statetable state table used of moving backwards
660 int32_t handlePrevious(const RBBIStateTable
*statetable
);
663 * This method is the actual implementation of the next() method. All iteration
664 * vectors through here. This method initializes the state machine to state 1
665 * and advances through the text character by character until we reach the end
666 * of the text or the state machine transitions to state 0. We update our return
667 * value every time the state machine passes through a possible end state.
668 * @param statetable state table used of moving forwards
671 int32_t handleNext(const RBBIStateTable
*statetable
);
676 * This is the function that actually implements dictionary-based
677 * breaking. Covering at least the range from startPos to endPos,
678 * it checks for dictionary characters, and if it finds them determines
679 * the appropriate object to deal with them. It may cache found breaks in
680 * fCachedBreakPositions as it goes. It may well also look at text outside
681 * the range startPos to endPos.
682 * If going forward, endPos is the normal Unicode break result, and
683 * if goind in reverse, startPos is the normal Unicode break result
684 * @param startPos The start position of a range of text
685 * @param endPos The end position of a range of text
686 * @param reverse The call is for the reverse direction
689 int32_t checkDictionary(int32_t startPos
, int32_t endPos
, UBool reverse
);
694 * This function returns the appropriate LanguageBreakEngine for a
696 * @param c A character in the dictionary set
699 const LanguageBreakEngine
*getLanguageBreakEngine(UChar32 c
);
704 void makeRuleStatusValid();
708 //------------------------------------------------------------------------------
710 // Inline Functions Definitions ...
712 //------------------------------------------------------------------------------
714 inline UBool
RuleBasedBreakIterator::operator!=(const BreakIterator
& that
) const {
715 return !operator==(that
);
720 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */