1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | *************************************************************************** |
5 | * Copyright (C) 1999-2016 International Business Machines Corporation * |
6 | * and others. All rights reserved. * |
7 | *************************************************************************** |
8 | |
9 | ********************************************************************** |
10 | * Date Name Description |
11 | * 10/22/99 alan Creation. |
12 | * 11/11/99 rgillam Complete port from Java. |
13 | ********************************************************************** |
14 | */ |
15 | |
16 | #ifndef RBBI_H |
17 | #define RBBI_H |
18 | |
19 | #include "unicode/utypes.h" |
20 | |
21 | #if U_SHOW_CPLUSPLUS_API |
22 | |
23 | /** |
24 | * \file |
25 | * \brief C++ API: Rule Based Break Iterator |
26 | */ |
27 | |
28 | #if !UCONFIG_NO_BREAK_ITERATION |
29 | |
30 | #include "unicode/brkiter.h" |
31 | #include "unicode/udata.h" |
32 | #include "unicode/parseerr.h" |
33 | #include "unicode/schriter.h" |
34 | |
35 | struct UCPTrie; |
36 | |
37 | U_NAMESPACE_BEGIN |
38 | |
39 | /** @internal */ |
40 | class LanguageBreakEngine; |
41 | struct ; |
42 | class RBBIDataWrapper; |
43 | class UnhandledEngine; |
44 | class UStack; |
45 | |
46 | /** |
47 | * |
48 | * A subclass of BreakIterator whose behavior is specified using a list of rules. |
49 | * <p>Instances of this class are most commonly created by the factory methods of |
50 | * BreakIterator::createWordInstance(), BreakIterator::createLineInstance(), etc., |
51 | * and then used via the abstract API in class BreakIterator</p> |
52 | * |
53 | * <p>See the ICU User Guide for information on Break Iterator Rules.</p> |
54 | * |
55 | * <p>This class is not intended to be subclassed.</p> |
56 | */ |
57 | class U_COMMON_API RuleBasedBreakIterator /*final*/ : public BreakIterator { |
58 | |
59 | private: |
60 | /** |
61 | * The UText through which this BreakIterator accesses the text |
62 | * @internal (private) |
63 | */ |
64 | UText fText = UTEXT_INITIALIZER; |
65 | |
66 | #ifndef U_HIDE_INTERNAL_API |
67 | public: |
68 | #endif /* U_HIDE_INTERNAL_API */ |
69 | /** |
70 | * The rule data for this BreakIterator instance. |
71 | * Not for general use; Public only for testing purposes. |
72 | * @internal |
73 | */ |
74 | RBBIDataWrapper *fData = nullptr; |
75 | |
76 | private: |
77 | /** |
78 | * The saved error code associated with this break iterator. |
79 | * This is the value to be returned by copyErrorTo(). |
80 | */ |
81 | UErrorCode fErrorCode = U_ZERO_ERROR; |
82 | |
83 | /** |
84 | * The current position of the iterator. Pinned, 0 < fPosition <= text.length. |
85 | * Never has the value UBRK_DONE (-1). |
86 | */ |
87 | int32_t fPosition = 0; |
88 | |
89 | /** |
90 | * TODO: |
91 | */ |
92 | int32_t fRuleStatusIndex = 0; |
93 | |
94 | /** |
95 | * Cache of previously determined boundary positions. |
96 | */ |
97 | class BreakCache; |
98 | BreakCache *fBreakCache = nullptr; |
99 | |
100 | /** |
101 | * Cache of boundary positions within a region of text that has been |
102 | * sub-divided by dictionary based breaking. |
103 | */ |
104 | class DictionaryCache; |
105 | DictionaryCache *fDictionaryCache = nullptr; |
106 | |
107 | /** |
108 | * |
109 | * If present, UStack of LanguageBreakEngine objects that might handle |
110 | * dictionary characters. Searched from top to bottom to find an object to |
111 | * handle a given character. |
112 | * @internal (private) |
113 | */ |
114 | UStack *fLanguageBreakEngines = nullptr; |
115 | |
116 | /** |
117 | * |
118 | * If present, the special LanguageBreakEngine used for handling |
119 | * characters that are in the dictionary set, but not handled by any |
120 | * LanguageBreakEngine. |
121 | * @internal (private) |
122 | */ |
123 | UnhandledEngine *fUnhandledBreakEngine = nullptr; |
124 | |
125 | /** |
126 | * Counter for the number of characters encountered with the "dictionary" |
127 | * flag set. |
128 | * @internal (private) |
129 | */ |
130 | uint32_t fDictionaryCharCount = 0; |
131 | |
132 | /** |
133 | * A character iterator that refers to the same text as the UText, above. |
134 | * Only included for compatibility with old API, which was based on CharacterIterators. |
135 | * Value may be adopted from outside, or one of fSCharIter or fDCharIter, below. |
136 | */ |
137 | CharacterIterator *fCharIter = &fSCharIter; |
138 | |
139 | /** |
140 | * When the input text is provided by a UnicodeString, this will point to |
141 | * a characterIterator that wraps that data. Needed only for the |
142 | * implementation of getText(), a backwards compatibility issue. |
143 | */ |
144 | UCharCharacterIterator fSCharIter {u"" , 0}; |
145 | |
146 | /** |
147 | * True when iteration has run off the end, and iterator functions should return UBRK_DONE. |
148 | */ |
149 | bool fDone = false; |
150 | |
151 | /** |
152 | * Array of look-ahead tentative results. |
153 | */ |
154 | int32_t *fLookAheadMatches = nullptr; |
155 | |
156 | /** |
157 | * A flag to indicate if phrase based breaking is enabled. |
158 | */ |
159 | UBool fIsPhraseBreaking = false; |
160 | |
161 | //======================================================================= |
162 | // constructors |
163 | //======================================================================= |
164 | |
165 | /** |
166 | * Constructor from a flattened set of RBBI data in malloced memory. |
167 | * RulesBasedBreakIterators built from a custom set of rules |
168 | * are created via this constructor; the rules are compiled |
169 | * into memory, then the break iterator is constructed here. |
170 | * |
171 | * The break iterator adopts the memory, and will |
172 | * free it when done. |
173 | * @internal (private) |
174 | */ |
175 | (RBBIDataHeader* data, UErrorCode &status); |
176 | |
177 | /** |
178 | * This constructor uses the udata interface to create a BreakIterator |
179 | * whose internal tables live in a memory-mapped file. "image" is an |
180 | * ICU UDataMemory handle for the pre-compiled break iterator tables. |
181 | * @param image handle to the memory image for the break iterator data. |
182 | * Ownership of the UDataMemory handle passes to the Break Iterator, |
183 | * which will be responsible for closing it when it is no longer needed. |
184 | * @param status Information on any errors encountered. |
185 | * @param isPhraseBreaking true if phrase based breaking is required, otherwise false. |
186 | * @see udata_open |
187 | * @see #getBinaryRules |
188 | * @internal (private) |
189 | */ |
190 | RuleBasedBreakIterator(UDataMemory* image, UBool isPhraseBreaking, UErrorCode &status); |
191 | |
192 | /** @internal */ |
193 | friend class RBBIRuleBuilder; |
194 | /** @internal */ |
195 | friend class BreakIterator; |
196 | |
197 | /** |
198 | * Default constructor with an error code parameter. |
199 | * Aside from error handling, otherwise identical to the default constructor. |
200 | * Internally, handles common initialization for other constructors. |
201 | * @internal (private) |
202 | */ |
203 | RuleBasedBreakIterator(UErrorCode *status); |
204 | |
205 | public: |
206 | |
207 | /** Default constructor. Creates an empty shell of an iterator, with no |
208 | * rules or text to iterate over. Object can subsequently be assigned to, |
209 | * but is otherwise unusable. |
210 | * @stable ICU 2.2 |
211 | */ |
212 | RuleBasedBreakIterator(); |
213 | |
214 | /** |
215 | * Copy constructor. Will produce a break iterator with the same behavior, |
216 | * and which iterates over the same text, as the one passed in. |
217 | * @param that The RuleBasedBreakIterator passed to be copied |
218 | * @stable ICU 2.0 |
219 | */ |
220 | RuleBasedBreakIterator(const RuleBasedBreakIterator& that); |
221 | |
222 | /** |
223 | * Construct a RuleBasedBreakIterator from a set of rules supplied as a string. |
224 | * @param rules The break rules to be used. |
225 | * @param parseError In the event of a syntax error in the rules, provides the location |
226 | * within the rules of the problem. |
227 | * @param status Information on any errors encountered. |
228 | * @stable ICU 2.2 |
229 | */ |
230 | RuleBasedBreakIterator( const UnicodeString &rules, |
231 | UParseError &parseError, |
232 | UErrorCode &status); |
233 | |
234 | /** |
235 | * Construct a RuleBasedBreakIterator from a set of precompiled binary rules. |
236 | * Binary rules are obtained from RulesBasedBreakIterator::getBinaryRules(). |
237 | * Construction of a break iterator in this way is substantially faster than |
238 | * construction from source rules. |
239 | * |
240 | * Ownership of the storage containing the compiled rules remains with the |
241 | * caller of this function. The compiled rules must not be modified or |
242 | * deleted during the life of the break iterator. |
243 | * |
244 | * The compiled rules are not compatible across different major versions of ICU. |
245 | * The compiled rules are compatible only between machines with the same |
246 | * byte ordering (little or big endian) and the same base character set family |
247 | * (ASCII or EBCDIC). |
248 | * |
249 | * @see #getBinaryRules |
250 | * @param compiledRules A pointer to the compiled break rules to be used. |
251 | * @param ruleLength The length of the compiled break rules, in bytes. This |
252 | * corresponds to the length value produced by getBinaryRules(). |
253 | * @param status Information on any errors encountered, including invalid |
254 | * binary rules. |
255 | * @stable ICU 4.8 |
256 | */ |
257 | RuleBasedBreakIterator(const uint8_t *compiledRules, |
258 | uint32_t ruleLength, |
259 | UErrorCode &status); |
260 | |
261 | /** |
262 | * This constructor uses the udata interface to create a BreakIterator |
263 | * whose internal tables live in a memory-mapped file. "image" is an |
264 | * ICU UDataMemory handle for the pre-compiled break iterator tables. |
265 | * @param image handle to the memory image for the break iterator data. |
266 | * Ownership of the UDataMemory handle passes to the Break Iterator, |
267 | * which will be responsible for closing it when it is no longer needed. |
268 | * @param status Information on any errors encountered. |
269 | * @see udata_open |
270 | * @see #getBinaryRules |
271 | * @stable ICU 2.8 |
272 | */ |
273 | RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status); |
274 | |
275 | /** |
276 | * Destructor |
277 | * @stable ICU 2.0 |
278 | */ |
279 | virtual ~RuleBasedBreakIterator(); |
280 | |
281 | /** |
282 | * Assignment operator. Sets this iterator to have the same behavior, |
283 | * and iterate over the same text, as the one passed in. |
284 | * @param that The RuleBasedBreakItertor passed in |
285 | * @return the newly created RuleBasedBreakIterator |
286 | * @stable ICU 2.0 |
287 | */ |
288 | RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that); |
289 | |
290 | /** |
291 | * Equality operator. Returns true if both BreakIterators are of the |
292 | * same class, have the same behavior, and iterate over the same text. |
293 | * @param that The BreakIterator to be compared for equality |
294 | * @return true if both BreakIterators are of the |
295 | * same class, have the same behavior, and iterate over the same text. |
296 | * @stable ICU 2.0 |
297 | */ |
298 | virtual bool operator==(const BreakIterator& that) const override; |
299 | |
300 | /** |
301 | * Not-equal operator. If operator== returns true, this returns false, |
302 | * and vice versa. |
303 | * @param that The BreakIterator to be compared for inequality |
304 | * @return true if both BreakIterators are not same. |
305 | * @stable ICU 2.0 |
306 | */ |
307 | inline bool operator!=(const BreakIterator& that) const { |
308 | return !operator==(that); |
309 | } |
310 | |
311 | /** |
312 | * Returns a newly-constructed RuleBasedBreakIterator with the same |
313 | * behavior, and iterating over the same text, as this one. |
314 | * Differs from the copy constructor in that it is polymorphic, and |
315 | * will correctly clone (copy) a derived class. |
316 | * clone() is thread safe. Multiple threads may simultaneously |
317 | * clone the same source break iterator. |
318 | * @return a newly-constructed RuleBasedBreakIterator |
319 | * @stable ICU 2.0 |
320 | */ |
321 | virtual RuleBasedBreakIterator* clone() const override; |
322 | |
323 | /** |
324 | * Compute a hash code for this BreakIterator |
325 | * @return A hash code |
326 | * @stable ICU 2.0 |
327 | */ |
328 | virtual int32_t hashCode(void) const; |
329 | |
330 | /** |
331 | * Returns the description used to create this iterator |
332 | * @return the description used to create this iterator |
333 | * @stable ICU 2.0 |
334 | */ |
335 | virtual const UnicodeString& getRules(void) const; |
336 | |
337 | //======================================================================= |
338 | // BreakIterator overrides |
339 | //======================================================================= |
340 | |
341 | /** |
342 | * <p> |
343 | * Return a CharacterIterator over the text being analyzed. |
344 | * The returned character iterator is owned by the break iterator, and must |
345 | * not be deleted by the caller. Repeated calls to this function may |
346 | * return the same CharacterIterator. |
347 | * </p> |
348 | * <p> |
349 | * The returned character iterator must not be used concurrently with |
350 | * the break iterator. If concurrent operation is needed, clone the |
351 | * returned character iterator first and operate on the clone. |
352 | * </p> |
353 | * <p> |
354 | * When the break iterator is operating on text supplied via a UText, |
355 | * this function will fail, returning a CharacterIterator containing no text. |
356 | * The function getUText() provides similar functionality, |
357 | * is reliable, and is more efficient. |
358 | * </p> |
359 | * |
360 | * TODO: deprecate this function? |
361 | * |
362 | * @return An iterator over the text being analyzed. |
363 | * @stable ICU 2.0 |
364 | */ |
365 | virtual CharacterIterator& getText(void) const override; |
366 | |
367 | |
368 | /** |
369 | * Get a UText for the text being analyzed. |
370 | * The returned UText is a shallow clone of the UText used internally |
371 | * by the break iterator implementation. It can safely be used to |
372 | * access the text without impacting any break iterator operations, |
373 | * but the underlying text itself must not be altered. |
374 | * |
375 | * @param fillIn A UText to be filled in. If nullptr, a new UText will be |
376 | * allocated to hold the result. |
377 | * @param status receives any error codes. |
378 | * @return The current UText for this break iterator. If an input |
379 | * UText was provided, it will always be returned. |
380 | * @stable ICU 3.4 |
381 | */ |
382 | virtual UText *getUText(UText *fillIn, UErrorCode &status) const override; |
383 | |
384 | /** |
385 | * Set the iterator to analyze a new piece of text. This function resets |
386 | * the current iteration position to the beginning of the text. |
387 | * @param newText An iterator over the text to analyze. The BreakIterator |
388 | * takes ownership of the character iterator. The caller MUST NOT delete it! |
389 | * @stable ICU 2.0 |
390 | */ |
391 | virtual void adoptText(CharacterIterator* newText) override; |
392 | |
393 | /** |
394 | * Set the iterator to analyze a new piece of text. This function resets |
395 | * the current iteration position to the beginning of the text. |
396 | * |
397 | * The BreakIterator will retain a reference to the supplied string. |
398 | * The caller must not modify or delete the text while the BreakIterator |
399 | * retains the reference. |
400 | * |
401 | * @param newText The text to analyze. |
402 | * @stable ICU 2.0 |
403 | */ |
404 | virtual void setText(const UnicodeString& newText) override; |
405 | |
406 | /** |
407 | * Reset the break iterator to operate over the text represented by |
408 | * the UText. The iterator position is reset to the start. |
409 | * |
410 | * This function makes a shallow clone of the supplied UText. This means |
411 | * that the caller is free to immediately close or otherwise reuse the |
412 | * Utext that was passed as a parameter, but that the underlying text itself |
413 | * must not be altered while being referenced by the break iterator. |
414 | * |
415 | * @param text The UText used to change the text. |
416 | * @param status Receives any error codes. |
417 | * @stable ICU 3.4 |
418 | */ |
419 | virtual void setText(UText *text, UErrorCode &status) override; |
420 | |
421 | /** |
422 | * Sets the current iteration position to the beginning of the text, position zero. |
423 | * @return The offset of the beginning of the text, zero. |
424 | * @stable ICU 2.0 |
425 | */ |
426 | virtual int32_t first(void) override; |
427 | |
428 | /** |
429 | * Sets the current iteration position to the end of the text. |
430 | * @return The text's past-the-end offset. |
431 | * @stable ICU 2.0 |
432 | */ |
433 | virtual int32_t last(void) override; |
434 | |
435 | /** |
436 | * Advances the iterator either forward or backward the specified number of steps. |
437 | * Negative values move backward, and positive values move forward. This is |
438 | * equivalent to repeatedly calling next() or previous(). |
439 | * @param n The number of steps to move. The sign indicates the direction |
440 | * (negative is backwards, and positive is forwards). |
441 | * @return The character offset of the boundary position n boundaries away from |
442 | * the current one. |
443 | * @stable ICU 2.0 |
444 | */ |
445 | virtual int32_t next(int32_t n) override; |
446 | |
447 | /** |
448 | * Advances the iterator to the next boundary position. |
449 | * @return The position of the first boundary after this one. |
450 | * @stable ICU 2.0 |
451 | */ |
452 | virtual int32_t next(void) override; |
453 | |
454 | /** |
455 | * Moves the iterator backwards, to the last boundary preceding this one. |
456 | * @return The position of the last boundary position preceding this one. |
457 | * @stable ICU 2.0 |
458 | */ |
459 | virtual int32_t previous(void) override; |
460 | |
461 | /** |
462 | * Sets the iterator to refer to the first boundary position following |
463 | * the specified position. |
464 | * @param offset The position from which to begin searching for a break position. |
465 | * @return The position of the first break after the current position. |
466 | * @stable ICU 2.0 |
467 | */ |
468 | virtual int32_t following(int32_t offset) override; |
469 | |
470 | /** |
471 | * Sets the iterator to refer to the last boundary position before the |
472 | * specified position. |
473 | * @param offset The position to begin searching for a break from. |
474 | * @return The position of the last boundary before the starting position. |
475 | * @stable ICU 2.0 |
476 | */ |
477 | virtual int32_t preceding(int32_t offset) override; |
478 | |
479 | /** |
480 | * Returns true if the specified position is a boundary position. As a side |
481 | * effect, leaves the iterator pointing to the first boundary position at |
482 | * or after "offset". |
483 | * @param offset the offset to check. |
484 | * @return True if "offset" is a boundary position. |
485 | * @stable ICU 2.0 |
486 | */ |
487 | virtual UBool isBoundary(int32_t offset) override; |
488 | |
489 | /** |
490 | * Returns the current iteration position. Note that UBRK_DONE is never |
491 | * returned from this function; if iteration has run to the end of a |
492 | * string, current() will return the length of the string while |
493 | * next() will return UBRK_DONE). |
494 | * @return The current iteration position. |
495 | * @stable ICU 2.0 |
496 | */ |
497 | virtual int32_t current(void) const override; |
498 | |
499 | |
500 | /** |
501 | * Return the status tag from the break rule that determined the boundary at |
502 | * the current iteration position. For break rules that do not specify a |
503 | * status, a default value of 0 is returned. If more than one break rule |
504 | * would cause a boundary to be located at some position in the text, |
505 | * the numerically largest of the applicable status values is returned. |
506 | * <p> |
507 | * Of the standard types of ICU break iterators, only word break and |
508 | * line break provide status values. The values are defined in |
509 | * the header file ubrk.h. For Word breaks, the status allows distinguishing between words |
510 | * that contain alphabetic letters, "words" that appear to be numbers, |
511 | * punctuation and spaces, words containing ideographic characters, and |
512 | * more. For Line Break, the status distinguishes between hard (mandatory) breaks |
513 | * and soft (potential) break positions. |
514 | * <p> |
515 | * <code>getRuleStatus()</code> can be called after obtaining a boundary |
516 | * position from <code>next()</code>, <code>previous()</code>, or |
517 | * any other break iterator functions that returns a boundary position. |
518 | * <p> |
519 | * Note that <code>getRuleStatus()</code> returns the value corresponding to |
520 | * <code>current()</code> index even after <code>next()</code> has returned DONE. |
521 | * <p> |
522 | * When creating custom break rules, one is free to define whatever |
523 | * status values may be convenient for the application. |
524 | * <p> |
525 | * @return the status from the break rule that determined the boundary |
526 | * at the current iteration position. |
527 | * |
528 | * @see UWordBreak |
529 | * @stable ICU 2.2 |
530 | */ |
531 | virtual int32_t getRuleStatus() const override; |
532 | |
533 | /** |
534 | * Get the status (tag) values from the break rule(s) that determined the boundary |
535 | * at the current iteration position. |
536 | * <p> |
537 | * The returned status value(s) are stored into an array provided by the caller. |
538 | * The values are stored in sorted (ascending) order. |
539 | * If the capacity of the output array is insufficient to hold the data, |
540 | * the output will be truncated to the available length, and a |
541 | * U_BUFFER_OVERFLOW_ERROR will be signaled. |
542 | * |
543 | * @param fillInVec an array to be filled in with the status values. |
544 | * @param capacity the length of the supplied vector. A length of zero causes |
545 | * the function to return the number of status values, in the |
546 | * normal way, without attempting to store any values. |
547 | * @param status receives error codes. |
548 | * @return The number of rule status values from the rules that determined |
549 | * the boundary at the current iteration position. |
550 | * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value |
551 | * is the total number of status values that were available, |
552 | * not the reduced number that were actually returned. |
553 | * @see getRuleStatus |
554 | * @stable ICU 3.0 |
555 | */ |
556 | virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status) override; |
557 | |
558 | /** |
559 | * Returns a unique class ID POLYMORPHICALLY. Pure virtual override. |
560 | * This method is to implement a simple version of RTTI, since not all |
561 | * C++ compilers support genuine RTTI. Polymorphic operator==() and |
562 | * clone() methods call this method. |
563 | * |
564 | * @return The class ID for this object. All objects of a |
565 | * given class have the same class ID. Objects of |
566 | * other classes have different class IDs. |
567 | * @stable ICU 2.0 |
568 | */ |
569 | virtual UClassID getDynamicClassID(void) const override; |
570 | |
571 | /** |
572 | * Returns the class ID for this class. This is useful only for |
573 | * comparing to a return value from getDynamicClassID(). For example: |
574 | * |
575 | * Base* polymorphic_pointer = createPolymorphicObject(); |
576 | * if (polymorphic_pointer->getDynamicClassID() == |
577 | * Derived::getStaticClassID()) ... |
578 | * |
579 | * @return The class ID for all objects of this class. |
580 | * @stable ICU 2.0 |
581 | */ |
582 | static UClassID U_EXPORT2 getStaticClassID(void); |
583 | |
584 | #ifndef U_FORCE_HIDE_DEPRECATED_API |
585 | /** |
586 | * Deprecated functionality. Use clone() instead. |
587 | * |
588 | * Create a clone (copy) of this break iterator in memory provided |
589 | * by the caller. The idea is to increase performance by avoiding |
590 | * a storage allocation. Use of this function is NOT RECOMMENDED. |
591 | * Performance gains are minimal, and correct buffer management is |
592 | * tricky. Use clone() instead. |
593 | * |
594 | * @param stackBuffer The pointer to the memory into which the cloned object |
595 | * should be placed. If nullptr, allocate heap memory |
596 | * for the cloned object. |
597 | * @param BufferSize The size of the buffer. If zero, return the required |
598 | * buffer size, but do not clone the object. If the |
599 | * size was too small (but not zero), allocate heap |
600 | * storage for the cloned object. |
601 | * |
602 | * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be |
603 | * returned if the provided buffer was too small, and |
604 | * the clone was therefore put on the heap. |
605 | * |
606 | * @return Pointer to the clone object. This may differ from the stackBuffer |
607 | * address if the byte alignment of the stack buffer was not suitable |
608 | * or if the stackBuffer was too small to hold the clone. |
609 | * @deprecated ICU 52. Use clone() instead. |
610 | */ |
611 | virtual RuleBasedBreakIterator *createBufferClone(void *stackBuffer, |
612 | int32_t &BufferSize, |
613 | UErrorCode &status) override; |
614 | #endif // U_FORCE_HIDE_DEPRECATED_API |
615 | |
616 | /** |
617 | * Return the binary form of compiled break rules, |
618 | * which can then be used to create a new break iterator at some |
619 | * time in the future. Creating a break iterator from pre-compiled rules |
620 | * is much faster than building one from the source form of the |
621 | * break rules. |
622 | * |
623 | * The binary data can only be used with the same version of ICU |
624 | * and on the same platform type (processor endian-ness) |
625 | * |
626 | * @param length Returns the length of the binary data. (Out parameter.) |
627 | * |
628 | * @return A pointer to the binary (compiled) rule data. The storage |
629 | * belongs to the RulesBasedBreakIterator object, not the |
630 | * caller, and must not be modified or deleted. |
631 | * @stable ICU 4.8 |
632 | */ |
633 | virtual const uint8_t *getBinaryRules(uint32_t &length); |
634 | |
635 | /** |
636 | * Set the subject text string upon which the break iterator is operating |
637 | * without changing any other aspect of the matching state. |
638 | * The new and previous text strings must have the same content. |
639 | * |
640 | * This function is intended for use in environments where ICU is operating on |
641 | * strings that may move around in memory. It provides a mechanism for notifying |
642 | * ICU that the string has been relocated, and providing a new UText to access the |
643 | * string in its new position. |
644 | * |
645 | * Note that the break iterator implementation never copies the underlying text |
646 | * of a string being processed, but always operates directly on the original text |
647 | * provided by the user. Refreshing simply drops the references to the old text |
648 | * and replaces them with references to the new. |
649 | * |
650 | * Caution: this function is normally used only by very specialized, |
651 | * system-level code. One example use case is with garbage collection that moves |
652 | * the text in memory. |
653 | * |
654 | * @param input The new (moved) text string. |
655 | * @param status Receives errors detected by this function. |
656 | * @return *this |
657 | * |
658 | * @stable ICU 49 |
659 | */ |
660 | virtual RuleBasedBreakIterator &refreshInputText(UText *input, UErrorCode &status) override; |
661 | |
662 | |
663 | private: |
664 | //======================================================================= |
665 | // implementation |
666 | //======================================================================= |
667 | /** |
668 | * Iterate backwards from an arbitrary position in the input text using the |
669 | * synthesized Safe Reverse rules. |
670 | * This locates a "Safe Position" from which the forward break rules |
671 | * will operate correctly. A Safe Position is not necessarily a boundary itself. |
672 | * |
673 | * @param fromPosition the position in the input text to begin the iteration. |
674 | * @internal (private) |
675 | */ |
676 | int32_t handleSafePrevious(int32_t fromPosition); |
677 | |
678 | /** |
679 | * Find a rule-based boundary by running the state machine. |
680 | * Input |
681 | * fPosition, the position in the text to begin from. |
682 | * Output |
683 | * fPosition: the boundary following the starting position. |
684 | * fDictionaryCharCount the number of dictionary characters encountered. |
685 | * If > 0, the segment will be further subdivided |
686 | * fRuleStatusIndex Info from the state table indicating which rules caused the boundary. |
687 | * |
688 | * @internal (private) |
689 | */ |
690 | int32_t handleNext(); |
691 | |
692 | /* |
693 | * Templatized version of handleNext() and handleSafePrevious(). |
694 | * |
695 | * There will be exactly four instantiations, two each for 8 and 16 bit tables, |
696 | * two each for 8 and 16 bit trie. |
697 | * Having separate instantiations for the table types keeps conditional tests of |
698 | * the table type out of the inner loops, at the expense of replicated code. |
699 | * |
700 | * The template parameter for the Trie access function is a value, not a type. |
701 | * Doing it this way, the compiler will inline the Trie function in the |
702 | * expanded functions. (Both the 8 and 16 bit access functions have the same type |
703 | * signature) |
704 | */ |
705 | |
706 | typedef uint16_t (*PTrieFunc)(const UCPTrie *, UChar32); |
707 | |
708 | template<typename RowType, PTrieFunc trieFunc> |
709 | int32_t handleSafePrevious(int32_t fromPosition); |
710 | |
711 | template<typename RowType, PTrieFunc trieFunc> |
712 | int32_t handleNext(); |
713 | |
714 | |
715 | /** |
716 | * This function returns the appropriate LanguageBreakEngine for a |
717 | * given character c. |
718 | * @param c A character in the dictionary set |
719 | * @internal (private) |
720 | */ |
721 | const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c); |
722 | |
723 | public: |
724 | #ifndef U_HIDE_INTERNAL_API |
725 | /** |
726 | * Debugging function only. |
727 | * @internal |
728 | */ |
729 | void dumpCache(); |
730 | |
731 | /** |
732 | * Debugging function only. |
733 | * @internal |
734 | */ |
735 | void dumpTables(); |
736 | #endif /* U_HIDE_INTERNAL_API */ |
737 | }; |
738 | |
739 | U_NAMESPACE_END |
740 | |
741 | #endif /* #if !UCONFIG_NO_BREAK_ITERATION */ |
742 | |
743 | #endif /* U_SHOW_CPLUSPLUS_API */ |
744 | |
745 | #endif |
746 | |