rbbi.h

Go to the documentation of this file.

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
***************************************************************************
*   Copyright (C) 1999-2016 International Business Machines Corporation   *
*   and others. All rights reserved.                                      *
***************************************************************************

**********************************************************************
*   Date        Name        Description
*   10/22/99    alan        Creation.
*   11/11/99    rgillam     Complete port from Java.
**********************************************************************
*/

#ifndef RBBI_H
#define RBBI_H

#include "unicode/utypes.h"

#if !UCONFIG_NO_BREAK_ITERATION

#include "unicode/brkiter.h"
#include "unicode/udata.h"
#include "unicode/parseerr.h"
#include "unicode/schriter.h"

U_NAMESPACE_BEGIN

class  LanguageBreakEngine;
struct RBBIDataHeader;
class  RBBIDataWrapper;
class  UnhandledEngine;
class  UStack;

class U_COMMON_API RuleBasedBreakIterator /*U_FINAL*/ : public BreakIterator {

private:
    UText  fText;

#ifndef U_HIDE_INTERNAL_API
public:
#endif /* U_HIDE_INTERNAL_API */

    RBBIDataWrapper    *fData;
private:

    int32_t         fPosition;

    int32_t         fRuleStatusIndex;

    class BreakCache;
    BreakCache         *fBreakCache;

    class DictionaryCache;
    DictionaryCache *fDictionaryCache;

    UStack              *fLanguageBreakEngines;

    UnhandledEngine     *fUnhandledBreakEngine;

    uint32_t            fDictionaryCharCount;

    CharacterIterator  *fCharIter;

    StringCharacterIterator fSCharIter;

    UBool           fDone;

    //=======================================================================
    // constructors
    //=======================================================================

    RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status);

    friend class RBBIRuleBuilder;
    friend class BreakIterator;

public:

    RuleBasedBreakIterator();

    RuleBasedBreakIterator(const RuleBasedBreakIterator& that);

    RuleBasedBreakIterator( const UnicodeString    &rules,
                             UParseError           &parseError,
                             UErrorCode            &status);

    RuleBasedBreakIterator(const uint8_t *compiledRules,
                           uint32_t       ruleLength,
                           UErrorCode    &status);

    RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status);

    virtual ~RuleBasedBreakIterator();

    RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that);

    virtual UBool operator==(const BreakIterator& that) const;

    inline UBool operator!=(const BreakIterator& that) const;

    virtual BreakIterator* clone() const;

    virtual int32_t hashCode(void) const;

    virtual const UnicodeString& getRules(void) const;

    //=======================================================================
    // BreakIterator overrides
    //=======================================================================

    virtual  CharacterIterator& getText(void) const;


     virtual UText *getUText(UText *fillIn, UErrorCode &status) const;

    virtual void adoptText(CharacterIterator* newText);

    virtual void setText(const UnicodeString& newText);

    virtual void  setText(UText *text, UErrorCode &status);

    virtual int32_t first(void);

    virtual int32_t last(void);

    virtual int32_t next(int32_t n);

    virtual int32_t next(void);

    virtual int32_t previous(void);

    virtual int32_t following(int32_t offset);

    virtual int32_t preceding(int32_t offset);

    virtual UBool isBoundary(int32_t offset);

    virtual int32_t current(void) const;


    virtual int32_t getRuleStatus() const;

    virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status);

    virtual UClassID getDynamicClassID(void) const;

    static UClassID U_EXPORT2 getStaticClassID(void);

    virtual BreakIterator *  createBufferClone(void *stackBuffer,
                                               int32_t &BufferSize,
                                               UErrorCode &status);


    virtual const uint8_t *getBinaryRules(uint32_t &length);

    virtual RuleBasedBreakIterator &refreshInputText(UText *input, UErrorCode &status);


private:
    //=======================================================================
    // implementation
    //=======================================================================
    void reset(void);

    void init(UErrorCode &status);

    int32_t handleSafePrevious(int32_t fromPosition);

    int32_t handleNext();


    const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c);

  public:
#ifndef U_HIDE_INTERNAL_API

     void dumpCache();

    void dumpTables();

#endif  /* U_HIDE_INTERNAL_API */
};

//------------------------------------------------------------------------------
//
//   Inline Functions Definitions ...
//
//------------------------------------------------------------------------------

inline UBool RuleBasedBreakIterator::operator!=(const BreakIterator& that) const {
    return !operator==(that);
}

U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_BREAK_ITERATION */

#endif