ICU 62.1  62.1
brkiter.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 ********************************************************************************
5 * Copyright (C) 1997-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 ********************************************************************************
8 *
9 * File brkiter.h
10 *
11 * Modification History:
12 *
13 * Date Name Description
14 * 02/18/97 aliu Added typedef for TextCount. Made DONE const.
15 * 05/07/97 aliu Fixed DLL declaration.
16 * 07/09/97 jfitz Renamed BreakIterator and interface synced with JDK
17 * 08/11/98 helena Sync-up JDK1.2.
18 * 01/13/2000 helena Added UErrorCode parameter to createXXXInstance methods.
19 ********************************************************************************
20 */
21 
22 #ifndef BRKITER_H
23 #define BRKITER_H
24 
25 #include "unicode/utypes.h"
26 
32 #if UCONFIG_NO_BREAK_ITERATION
33 
35 
36 /*
37  * Allow the declaration of APIs with pointers to BreakIterator
38  * even when break iteration is removed from the build.
39  */
40 class BreakIterator;
41 
43 
44 #else
45 
46 #include "unicode/uobject.h"
47 #include "unicode/unistr.h"
48 #include "unicode/chariter.h"
49 #include "unicode/locid.h"
50 #include "unicode/ubrk.h"
51 #include "unicode/strenum.h"
52 #include "unicode/utext.h"
53 #include "unicode/umisc.h"
54 
56 
103 public:
108  virtual ~BreakIterator();
109 
123  virtual UBool operator==(const BreakIterator&) const = 0;
124 
131  UBool operator!=(const BreakIterator& rhs) const { return !operator==(rhs); }
132 
138  virtual BreakIterator* clone(void) const = 0;
139 
145  virtual UClassID getDynamicClassID(void) const = 0;
146 
151  virtual CharacterIterator& getText(void) const = 0;
152 
153 
168  virtual UText *getUText(UText *fillIn, UErrorCode &status) const = 0;
169 
181  virtual void setText(const UnicodeString &text) = 0;
182 
201  virtual void setText(UText *text, UErrorCode &status) = 0;
202 
211  virtual void adoptText(CharacterIterator* it) = 0;
212 
213  enum {
219  DONE = (int32_t)-1
220  };
221 
227  virtual int32_t first(void) = 0;
228 
234  virtual int32_t last(void) = 0;
235 
242  virtual int32_t previous(void) = 0;
243 
250  virtual int32_t next(void) = 0;
251 
257  virtual int32_t current(void) const = 0;
258 
267  virtual int32_t following(int32_t offset) = 0;
268 
277  virtual int32_t preceding(int32_t offset) = 0;
278 
287  virtual UBool isBoundary(int32_t offset) = 0;
288 
298  virtual int32_t next(int32_t n) = 0;
299 
313  virtual int32_t getRuleStatus() const;
314 
343  virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status);
344 
364  static BreakIterator* U_EXPORT2
365  createWordInstance(const Locale& where, UErrorCode& status);
366 
388  static BreakIterator* U_EXPORT2
389  createLineInstance(const Locale& where, UErrorCode& status);
390 
410  static BreakIterator* U_EXPORT2
411  createCharacterInstance(const Locale& where, UErrorCode& status);
412 
431  static BreakIterator* U_EXPORT2
432  createSentenceInstance(const Locale& where, UErrorCode& status);
433 
456  static BreakIterator* U_EXPORT2
457  createTitleInstance(const Locale& where, UErrorCode& status);
458 
468  static const Locale* U_EXPORT2 getAvailableLocales(int32_t& count);
469 
479  static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale,
480  const Locale& displayLocale,
481  UnicodeString& name);
482 
491  static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale,
492  UnicodeString& name);
493 
513  virtual BreakIterator * createBufferClone(void *stackBuffer,
514  int32_t &BufferSize,
515  UErrorCode &status) = 0;
516 
517 #ifndef U_HIDE_DEPRECATED_API
518 
525  inline UBool isBufferClone(void);
526 
527 #endif /* U_HIDE_DEPRECATED_API */
528 
529 #if !UCONFIG_NO_SERVICE
530 
545  static URegistryKey U_EXPORT2 registerInstance(BreakIterator* toAdopt,
546  const Locale& locale,
547  UBreakIteratorType kind,
548  UErrorCode& status);
549 
562  static UBool U_EXPORT2 unregister(URegistryKey key, UErrorCode& status);
563 
570  static StringEnumeration* U_EXPORT2 getAvailableLocales(void);
571 #endif
572 
578  Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const;
579 
580 #ifndef U_HIDE_INTERNAL_API
581 
587  const char *getLocaleID(ULocDataLocaleType type, UErrorCode& status) const;
588 #endif /* U_HIDE_INTERNAL_API */
589 
615  virtual BreakIterator &refreshInputText(UText *input, UErrorCode &status) = 0;
616 
617  private:
618  static BreakIterator* buildInstance(const Locale& loc, const char *type, UErrorCode& status);
619  static BreakIterator* createInstance(const Locale& loc, int32_t kind, UErrorCode& status);
620  static BreakIterator* makeInstance(const Locale& loc, int32_t kind, UErrorCode& status);
621 
622  friend class ICUBreakIteratorFactory;
623  friend class ICUBreakIteratorService;
624 
625 protected:
626  // Do not enclose protected default/copy constructors with #ifndef U_HIDE_INTERNAL_API
627  // or else the compiler will create a public ones.
629  BreakIterator();
631  BreakIterator (const BreakIterator &other);
632 #ifndef U_HIDE_INTERNAL_API
633 
634  BreakIterator (const Locale& valid, const Locale &actual);
636  BreakIterator &operator = (const BreakIterator &other);
637 #endif /* U_HIDE_INTERNAL_API */
638 
639 private:
640 
642  char actualLocale[ULOC_FULLNAME_CAPACITY];
643  char validLocale[ULOC_FULLNAME_CAPACITY];
644 };
645 
646 #ifndef U_HIDE_DEPRECATED_API
647 
649 {
650  return FALSE;
651 }
652 
653 #endif /* U_HIDE_DEPRECATED_API */
654 
656 
657 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
658 
659 #endif // BRKITER_H
660 //eof
C++ API: Unicode String.
U_EXPORT UBool operator==(const StringPiece &x, const StringPiece &y)
Global operator == for StringPiece.
UBreakIteratorType
The possible types of text boundaries.
Definition: ubrk.h:99
void * UClassID
UClassID is used to identify classes without using the compiler's RTTI.
Definition: uobject.h:93
C API:misc definitions.
#define ULOC_FULLNAME_CAPACITY
Useful constant for the maximum size of the whole locale ID (including the terminating NULL and all k...
Definition: uloc.h:264
Base class for 'pure' C++ implementations of uenum api.
Definition: strenum.h:57
C API: Abstract Unicode Text API.
#define U_NAMESPACE_BEGIN
This is used to begin a declaration of a public ICU C++ API.
Definition: uversion.h:137
Abstract class that defines an API for iteration on text objects.
Definition: chariter.h:358
The BreakIterator class implements methods for finding the location of boundaries in text...
Definition: brkiter.h:102
virtual UClassID getDynamicClassID() const
ICU4C "poor man's RTTI", returns a UClassID for the actual ICU class.
UBool isBufferClone(void)
Determine whether the BreakIterator was created in user memory by createBufferClone(), and thus should not be deleted.
Definition: brkiter.h:648
C++ API: Common ICU base class UObject.
const void * URegistryKey
Opaque type returned by registerInstance, registerFactory and unregister for service registration...
Definition: umisc.h:57
#define U_NAMESPACE_END
This is used to end a declaration of a public ICU C++ API.
Definition: uversion.h:138
UErrorCode
Error code to replace exception handling, so that the code is compatible with all C++ compilers...
Definition: utypes.h:396
ULocDataLocaleType
Constants for *_getLocale() Allow user to select whether she wants information on requested...
Definition: uloc.h:338
C API: BreakIterator.
C++ API: String Enumeration.
C++ API: Locale ID object.
UText struct.
Definition: utext.h:1345
Basic definitions for ICU, for both C and C++ APIs.
#define FALSE
The FALSE value of a UBool.
Definition: umachine.h:244
#define U_COMMON_API
Set to export library symbols from inside the common library, and to import them from outside...
Definition: utypes.h:359
UnicodeString is a string class that stores Unicode characters directly and provides similar function...
Definition: unistr.h:286
C++ API: Character Iterator.
UBool operator!=(const BreakIterator &rhs) const
Returns the complement of the result of operator==.
Definition: brkiter.h:131
UObject is the common ICU "boilerplate" class.
Definition: uobject.h:223
int8_t UBool
The ICU boolean type.
Definition: umachine.h:236
A Locale object represents a specific geographical, political, or cultural region.
Definition: locid.h:188