1 // © 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 * Legacy version of RuleBasedBreakIterator from ICU 57,
11 * only for use by Apple RuleBasedTokenizer
12 **********************************************************************
18 #include "unicode/utypes.h"
22 * \brief C++ API: Rule Based Break Iterator
25 #if !UCONFIG_NO_BREAK_ITERATION
27 #include "unicode/brkiter.h"
28 #include "unicode/udata.h"
29 #include "unicode/parseerr.h"
30 #include "unicode/schriter.h"
31 #include "unicode/uchriter.h"
32 #include "rbbidata57.h"
40 struct RBBIDataHeader57
;
41 class RuleBasedBreakIteratorTables
;
43 class RBBIDataWrapper57
;
45 class LanguageBreakEngine
;
46 class UnhandledEngine
;
47 struct RBBIStateTable
;
51 * A subclass of BreakIterator whose behavior is specified using a list of rules.
53 * <p>This class was not intended to be subclassed. The subclassing interface to
54 * RulesBasedBreakIterator is not part of the ICU API, and may not remain stable.</p>
56 * <p>However, the ICU 57 version <em>was<em> subclassed for the Apple RuleBasedTokenizer.
57 * So this legacy version of RuleBasedBreakIterator is preserved for that usage,
58 * since versions of RuleBasedBreakIterator in ICU 58 and later broke that
59 * subclassing (in ICU 58 it just by changing the access to certain internal fields,
60 * but in ICU 60 RuleBasedBreakIterator was completely rewritten).</p>
62 class U_COMMON_API RuleBasedBreakIterator57
/*U_FINAL*/ : public BreakIterator
{
66 * The UText through which this BreakIterator accesses the text
72 * A character iterator that refers to the same text as the UText, above.
73 * Only included for compatibility with old API, which was based on CharacterIterators.
74 * Value may be adopted from outside, or one of fSCharIter or fDCharIter, below.
76 CharacterIterator
*fCharIter
;
79 * When the input text is provided by a UnicodeString, this will point to
80 * a characterIterator that wraps that data. Needed only for the
81 * implementation of getText(), a backwards compatibility issue.
83 StringCharacterIterator
*fSCharIter
;
86 * When the input text is provided by a UText, this
87 * dummy CharacterIterator over an empty string will
88 * be returned from getText()
90 UCharCharacterIterator
*fDCharIter
;
93 * The rule data for this BreakIterator instance
96 RBBIDataWrapper57
*fData
;
98 /** Index of the Rule {tag} values for the most recent match.
101 int32_t fLastRuleStatusIndex
;
104 * Rule tag value valid flag.
105 * Some iterator operations don't intrinsically set the correct tag value.
106 * This flag lets us lazily compute the value if we are ever asked for it.
109 UBool fLastStatusIndexValid
;
112 * Counter for the number of characters encountered with the "dictionary"
116 uint32_t fDictionaryCharCount
;
119 * When a range of characters is divided up using the dictionary, the break
120 * positions that are discovered are stored here, preventing us from having
121 * to use either the dictionary or the state table again until the iterator
122 * leaves this range of text. Has the most impact for line breaking.
125 int32_t* fCachedBreakPositions
;
128 * The number of elements in fCachedBreakPositions
131 int32_t fNumCachedBreakPositions
;
134 * if fCachedBreakPositions is not null, this indicates which item in the
135 * cache the current iteration position refers to
138 int32_t fPositionInCache
;
142 * If present, UStack of LanguageBreakEngine objects that might handle
143 * dictionary characters. Searched from top to bottom to find an object to
144 * handle a given character.
147 UStack
*fLanguageBreakEngines
;
151 * If present, the special LanguageBreakEngine used for handling
152 * characters that are in the dictionary set, but not handled by any
153 * LangugageBreakEngine.
156 UnhandledEngine
*fUnhandledBreakEngine
;
160 * The type of the break iterator, or -1 if it has not been set.
165 //=======================================================================
167 //=======================================================================
170 * Constant to be used in the constructor
171 * RuleBasedBreakIterator57(RBBIDataHeader57*, EDontAdopt, UErrorCode &);
172 * which does not adopt the memory indicated by the RBBIDataHeader57*
175 * @internal, used by rbtok.cpp, urbtok57.cpp
182 * Constructor from a flattened set of RBBI data in malloced memory.
183 * RulesBasedBreakIterators built from a custom set of rules
184 * are created via this constructor; the rules are compiled
185 * into memory, then the break iterator is constructed here.
187 * The break iterator adopts the memory, and will
189 * @internal, used by rbtok.cpp
191 RuleBasedBreakIterator57(RBBIDataHeader57
* data
, UErrorCode
&status
);
194 * Constructor from a flattened set of RBBI data in memory which need not
195 * be malloced (e.g. it may be a memory-mapped file, etc.).
197 * This version does not adopt the memory, and does not
199 * @internal, used by rbtok.cpp
201 RuleBasedBreakIterator57(const RBBIDataHeader57
* data
, enum EDontAdopt dontAdopt
, UErrorCode
&status
);
204 friend class RBBIRuleBuilder57
;
206 friend class BreakIterator
;
210 /** Default constructor. Creates an empty shell of an iterator, with no
211 * rules or text to iterate over. Object can subsequently be assigned to.
214 RuleBasedBreakIterator57();
217 * Copy constructor. Will produce a break iterator with the same behavior,
218 * and which iterates over the same text, as the one passed in.
219 * @param that The RuleBasedBreakIterator57 passed to be copied
222 RuleBasedBreakIterator57(const RuleBasedBreakIterator57
& that
);
225 * Construct a RuleBasedBreakIterator57 from a set of rules supplied as a string.
226 * @param rules The break rules to be used.
227 * @param parseError In the event of a syntax error in the rules, provides the location
228 * within the rules of the problem.
229 * @param status Information on any errors encountered.
230 * @internal, used by rbtok.cpp
232 RuleBasedBreakIterator57( const UnicodeString
&rules
,
233 UParseError
&parseError
,
237 // not used by rbtok.cpp
239 * Contruct a RuleBasedBreakIterator57 from a set of precompiled binary rules.
240 * Binary rules are obtained from RulesBasedBreakIterator::getBinaryRules().
241 * Construction of a break iterator in this way is substantially faster than
242 * constuction from source rules.
244 * Ownership of the storage containing the compiled rules remains with the
245 * caller of this function. The compiled rules must not be modified or
246 * deleted during the life of the break iterator.
248 * The compiled rules are not compatible across different major versions of ICU.
249 * The compiled rules are comaptible only between machines with the same
250 * byte ordering (little or big endian) and the same base character set family
253 * @see #getBinaryRules
254 * @param compiledRules A pointer to the compiled break rules to be used.
255 * @param ruleLength The length of the compiled break rules, in bytes. This
256 * corresponds to the length value produced by getBinaryRules().
257 * @param status Information on any errors encountered, including invalid
261 RuleBasedBreakIterator57(const uint8_t *compiledRules
,
266 * This constructor uses the udata interface to create a BreakIterator
267 * whose internal tables live in a memory-mapped file. "image" is an
268 * ICU UDataMemory handle for the pre-compiled break iterator tables.
269 * @param image handle to the memory image for the break iterator data.
270 * Ownership of the UDataMemory handle passes to the Break Iterator,
271 * which will be responsible for closing it when it is no longer needed.
272 * @param status Information on any errors encountered.
274 * @see #getBinaryRules
277 RuleBasedBreakIterator57(UDataMemory
* image
, UErrorCode
&status
);
284 virtual ~RuleBasedBreakIterator57();
287 * Assignment operator. Sets this iterator to have the same behavior,
288 * and iterate over the same text, as the one passed in.
289 * @param that The RuleBasedBreakItertor passed in
290 * @return the newly created RuleBasedBreakIterator57
293 RuleBasedBreakIterator57
& operator=(const RuleBasedBreakIterator57
& that
);
296 * Equality operator. Returns TRUE if both BreakIterators are of the
297 * same class, have the same behavior, and iterate over the same text.
298 * @param that The BreakIterator to be compared for equality
299 * @return TRUE if both BreakIterators are of the
300 * same class, have the same behavior, and iterate over the same text.
303 virtual UBool
operator==(const BreakIterator
& that
) const;
306 * Not-equal operator. If operator== returns TRUE, this returns FALSE,
308 * @param that The BreakIterator to be compared for inequality
309 * @return TRUE if both BreakIterators are not same.
312 UBool
operator!=(const BreakIterator
& that
) const;
315 * Returns a newly-constructed RuleBasedBreakIterator57 with the same
316 * behavior, and iterating over the same text, as this one.
317 * Differs from the copy constructor in that it is polymorphic, and
318 * will correctly clone (copy) a derived class.
319 * clone() is thread safe. Multiple threads may simultaeneously
320 * clone the same source break iterator.
321 * @return a newly-constructed RuleBasedBreakIterator57
324 virtual BreakIterator
* clone() const;
327 * Compute a hash code for this BreakIterator
328 * @return A hash code
331 virtual int32_t hashCode(void) const;
334 // not used by rbtok.cpp
336 * Returns the description used to create this iterator
337 * @return the description used to create this iterator
340 virtual const UnicodeString
& getRules(void) const;
342 //=======================================================================
343 // BreakIterator overrides
344 //=======================================================================
348 * Return a CharacterIterator over the text being analyzed.
349 * The returned character iterator is owned by the break iterator, and must
350 * not be deleted by the caller. Repeated calls to this function may
351 * return the same CharacterIterator.
354 * The returned character iterator must not be used concurrently with
355 * the break iterator. If concurrent operation is needed, clone the
356 * returned character iterator first and operate on the clone.
359 * When the break iterator is operating on text supplied via a UText,
360 * this function will fail. Lacking any way to signal failures, it
361 * returns an CharacterIterator containing no text.
362 * The function getUText() provides similar functionality,
363 * is reliable, and is more efficient.
366 * TODO: deprecate this function?
368 * @return An iterator over the text being analyzed.
369 * @internal, inherited from BreakIterator
371 virtual CharacterIterator
& getText(void) const;
375 * Get a UText for the text being analyzed.
376 * The returned UText is a shallow clone of the UText used internally
377 * by the break iterator implementation. It can safely be used to
378 * access the text without impacting any break iterator operations,
379 * but the underlying text itself must not be altered.
381 * @param fillIn A UText to be filled in. If NULL, a new UText will be
382 * allocated to hold the result.
383 * @param status receives any error codes.
384 * @return The current UText for this break iterator. If an input
385 * UText was provided, it will always be returned.
386 * @internal, inherited from BreakIterator
388 virtual UText
*getUText(UText
*fillIn
, UErrorCode
&status
) const;
391 * Set the iterator to analyze a new piece of text. This function resets
392 * the current iteration position to the beginning of the text.
393 * @param newText An iterator over the text to analyze. The BreakIterator
394 * takes ownership of the character iterator. The caller MUST NOT delete it!
395 * @internal, inherited from BreakIterator
397 virtual void adoptText(CharacterIterator
* newText
);
400 * Set the iterator to analyze a new piece of text. This function resets
401 * the current iteration position to the beginning of the text.
403 * The BreakIterator will retain a reference to the supplied string.
404 * The caller must not modify or delete the text while the BreakIterator
405 * retains the reference.
407 * @param newText The text to analyze.
408 * @internal, inherited from BreakIterator
410 virtual void setText(const UnicodeString
& newText
);
413 * Reset the break iterator to operate over the text represented by
414 * the UText. The iterator position is reset to the start.
416 * This function makes a shallow clone of the supplied UText. This means
417 * that the caller is free to immediately close or otherwise reuse the
418 * Utext that was passed as a parameter, but that the underlying text itself
419 * must not be altered while being referenced by the break iterator.
421 * @param text The UText used to change the text.
422 * @param status Receives any error codes.
423 * @internal, inherited from BreakIterator
425 virtual void setText(UText
*text
, UErrorCode
&status
);
428 * Sets the current iteration position to the beginning of the text, position zero.
429 * @return The offset of the beginning of the text, zero.
430 * @internal, inherited from BreakIterator
432 virtual int32_t first(void);
435 * Sets the current iteration position to the end of the text.
436 * @return The text's past-the-end offset.
437 * @internal, inherited from BreakIterator
439 virtual int32_t last(void);
442 * Advances the iterator either forward or backward the specified number of steps.
443 * Negative values move backward, and positive values move forward. This is
444 * equivalent to repeatedly calling next() or previous().
445 * @param n The number of steps to move. The sign indicates the direction
446 * (negative is backwards, and positive is forwards).
447 * @return The character offset of the boundary position n boundaries away from
449 * @internal, inherited from BreakIterator
451 virtual int32_t next(int32_t n
);
454 * Advances the iterator to the next boundary position.
455 * @return The position of the first boundary after this one.
456 * @internal, inherited from BreakIterator
458 virtual int32_t next(void);
461 * Moves the iterator backwards, to the last boundary preceding this one.
462 * @return The position of the last boundary position preceding this one.
463 * @internal, inherited from BreakIterator
465 virtual int32_t previous(void);
468 * Sets the iterator to refer to the first boundary position following
469 * the specified position.
470 * @param offset The position from which to begin searching for a break position.
471 * @return The position of the first break after the current position.
472 * @internal, inherited from BreakIterator
474 virtual int32_t following(int32_t offset
);
477 * Sets the iterator to refer to the last boundary position before the
478 * specified position.
479 * @param offset The position to begin searching for a break from.
480 * @return The position of the last boundary before the starting position.
481 * @internal, inherited from BreakIterator
483 virtual int32_t preceding(int32_t offset
);
486 * Returns true if the specfied position is a boundary position. As a side
487 * effect, leaves the iterator pointing to the first boundary position at
489 * @param offset the offset to check.
490 * @return True if "offset" is a boundary position.
491 * @internal, inherited from BreakIterator
493 virtual UBool
isBoundary(int32_t offset
);
496 * Returns the current iteration position.
497 * @return The current iteration position.
498 * @internal, inherited from BreakIterator
500 virtual int32_t current(void) const;
504 * Return the status tag from the break rule that determined the most recently
505 * returned break position. For break rules that do not specify a
506 * status, a default value of 0 is returned. If more than one break rule
507 * would cause a boundary to be located at some position in the text,
508 * the numerically largest of the applicable status values is returned.
510 * Of the standard types of ICU break iterators, only word break and
511 * line break provide status values. The values are defined in
512 * the header file ubrk.h. For Word breaks, the status allows distinguishing between words
513 * that contain alphabetic letters, "words" that appear to be numbers,
514 * punctuation and spaces, words containing ideographic characters, and
515 * more. For Line Break, the status distinguishes between hard (mandatory) breaks
516 * and soft (potential) break positions.
518 * <code>getRuleStatus()</code> can be called after obtaining a boundary
519 * position from <code>next()</code>, <code>previous()</code>, or
520 * any other break iterator functions that returns a boundary position.
522 * When creating custom break rules, one is free to define whatever
523 * status values may be convenient for the application.
525 * Note: this function is not thread safe. It should not have been
526 * declared const, and the const remains only for compatibility
527 * reasons. (The function is logically const, but not bit-wise const).
529 * @return the status from the break rule that determined the most recently
530 * returned break position.
533 * @internal, inherited from BreakIterator
535 virtual int32_t getRuleStatus() const;
538 * Get the status (tag) values from the break rule(s) that determined the most
539 * recently returned break position.
541 * The returned status value(s) are stored into an array provided by the caller.
542 * The values are stored in sorted (ascending) order.
543 * If the capacity of the output array is insufficient to hold the data,
544 * the output will be truncated to the available length, and a
545 * U_BUFFER_OVERFLOW_ERROR will be signaled.
547 * @param fillInVec an array to be filled in with the status values.
548 * @param capacity the length of the supplied vector. A length of zero causes
549 * the function to return the number of status values, in the
550 * normal way, without attemtping to store any values.
551 * @param status receives error codes.
552 * @return The number of rule status values from rules that determined
553 * the most recent boundary returned by the break iterator.
554 * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value
555 * is the total number of status values that were available,
556 * not the reduced number that were actually returned.
558 * @internal, inherited from BreakIterator
560 virtual int32_t getRuleStatusVec(int32_t *fillInVec
, int32_t capacity
, UErrorCode
&status
);
563 * Returns a unique class ID POLYMORPHICALLY. Pure virtual override.
564 * This method is to implement a simple version of RTTI, since not all
565 * C++ compilers support genuine RTTI. Polymorphic operator==() and
566 * clone() methods call this method.
568 * @return The class ID for this object. All objects of a
569 * given class have the same class ID. Objects of
570 * other classes have different class IDs.
573 virtual UClassID
getDynamicClassID(void) const;
576 * Returns the class ID for this class. This is useful only for
577 * comparing to a return value from getDynamicClassID(). For example:
579 * Base* polymorphic_pointer = createPolymorphicObject();
580 * if (polymorphic_pointer->getDynamicClassID() ==
581 * Derived::getStaticClassID()) ...
583 * @return The class ID for all objects of this class.
586 static UClassID U_EXPORT2
getStaticClassID(void);
589 * Deprecated functionality. Use clone() instead.
591 * Create a clone (copy) of this break iterator in memory provided
592 * by the caller. The idea is to increase performance by avoiding
593 * a storage allocation. Use of this functoin is NOT RECOMMENDED.
594 * Performance gains are minimal, and correct buffer management is
595 * tricky. Use clone() instead.
597 * @param stackBuffer The pointer to the memory into which the cloned object
598 * should be placed. If NULL, allocate heap memory
599 * for the cloned object.
600 * @param BufferSize The size of the buffer. If zero, return the required
601 * buffer size, but do not clone the object. If the
602 * size was too small (but not zero), allocate heap
603 * storage for the cloned object.
605 * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be
606 * returned if the the provided buffer was too small, and
607 * the clone was therefore put on the heap.
609 * @return Pointer to the clone object. This may differ from the stackBuffer
610 * address if the byte alignment of the stack buffer was not suitable
611 * or if the stackBuffer was too small to hold the clone.
612 * @deprecated ICU 52. Use clone() instead.
614 virtual BreakIterator
* createBufferClone(void *stackBuffer
,
620 * Return the binary form of compiled break rules,
621 * which can then be used to create a new break iterator at some
622 * time in the future. Creating a break iterator from pre-compiled rules
623 * is much faster than building one from the source form of the
626 * The binary data can only be used with the same version of ICU
627 * and on the same platform type (processor endian-ness)
629 * @param length Returns the length of the binary data. (Out paramter.)
631 * @return A pointer to the binary (compiled) rule data. The storage
632 * belongs to the RulesBasedBreakIterator object, not the
633 * caller, and must not be modified or deleted.
634 * @internal, used by urbtok57.cpp
636 virtual const uint8_t *getBinaryRules(uint32_t &length
);
639 * Set the subject text string upon which the break iterator is operating
640 * without changing any other aspect of the matching state.
641 * The new and previous text strings must have the same content.
643 * This function is intended for use in environments where ICU is operating on
644 * strings that may move around in memory. It provides a mechanism for notifying
645 * ICU that the string has been relocated, and providing a new UText to access the
646 * string in its new position.
648 * Note that the break iterator implementation never copies the underlying text
649 * of a string being processed, but always operates directly on the original text
650 * provided by the user. Refreshing simply drops the references to the old text
651 * and replaces them with references to the new.
653 * Caution: this function is normally used only by very specialized,
654 * system-level code. One example use case is with garbage collection that moves
655 * the text in memory.
657 * @param input The new (moved) text string.
658 * @param status Receives errors detected by this function.
661 * @internal, inherited from BreakIterator
663 virtual RuleBasedBreakIterator57
&refreshInputText(UText
*input
, UErrorCode
&status
);
667 //=======================================================================
669 //=======================================================================
671 * Dumps caches and performs other actions associated with a complete change
672 * in text or iteration position.
675 virtual void reset(void);
678 * Set the type of the break iterator.
681 virtual void setBreakType(int32_t type
);
684 * Common initialization function, used by constructors and bufferClone.
692 * This method backs the iterator back up to a "safe position" in the text.
693 * This is a position that we know, without any context, must be a break position.
694 * The various calling methods then iterate forward from this safe position to
695 * the appropriate position to return. (For more information, see the description
696 * of buildBackwardsStateTable() in RuleBasedBreakIterator57.Builder.)
697 * @param statetable state table used of moving backwards
700 int32_t handlePrevious(const RBBIStateTable
*statetable
);
703 * This method is the actual implementation of the next() method. All iteration
704 * vectors through here. This method initializes the state machine to state 1
705 * and advances through the text character by character until we reach the end
706 * of the text or the state machine transitions to state 0. We update our return
707 * value every time the state machine passes through a possible end state.
708 * @param statetable state table used of moving forwards
711 int32_t handleNext(const RBBIStateTable
*statetable
);
716 * This is the function that actually implements dictionary-based
717 * breaking. Covering at least the range from startPos to endPos,
718 * it checks for dictionary characters, and if it finds them determines
719 * the appropriate object to deal with them. It may cache found breaks in
720 * fCachedBreakPositions as it goes. It may well also look at text outside
721 * the range startPos to endPos.
722 * If going forward, endPos is the normal Unicode break result, and
723 * if goind in reverse, startPos is the normal Unicode break result
724 * @param startPos The start position of a range of text
725 * @param endPos The end position of a range of text
726 * @param reverse The call is for the reverse direction
729 int32_t checkDictionary(int32_t startPos
, int32_t endPos
, UBool reverse
);
734 * This function returns the appropriate LanguageBreakEngine for a
736 * @param c A character in the dictionary set
739 const LanguageBreakEngine
*getLanguageBreakEngine(UChar32 c
);
744 void makeRuleStatusValid();
748 //------------------------------------------------------------------------------
750 // Inline Functions Definitions ...
752 //------------------------------------------------------------------------------
754 inline UBool
RuleBasedBreakIterator57::operator!=(const BreakIterator
& that
) const {
755 return !operator==(that
);
760 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */