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 U_SHOW_CPLUSPLUS_API

#if !UCONFIG_NO_BREAK_ITERATION

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

struct UCPTrie;

U_NAMESPACE_BEGIN

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


#ifndef U_HIDE_INTERNAL_API

class ExternalBreakEngine : public UObject {
  public:
    virtual ~ExternalBreakEngine() {}

    virtual bool isFor(UChar32 c, const char* locale) const = 0;

    virtual bool handles(UChar32 c) const = 0;

     virtual int32_t fillBreaks(UText* text,  int32_t start, int32_t end,
                               int32_t* foundBreaks, int32_t foundBreaksCapacity,
                               UErrorCode& status) const = 0;
};
#endif  /* U_HIDE_INTERNAL_API */


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

private:
    UText  fText = UTEXT_INITIALIZER;

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

    RBBIDataWrapper    *fData = nullptr;

private:
    UErrorCode      fErrorCode = U_ZERO_ERROR;

    int32_t         fPosition = 0;

    int32_t         fRuleStatusIndex = 0;

    class BreakCache;
    BreakCache         *fBreakCache = nullptr;

    class DictionaryCache;
    DictionaryCache *fDictionaryCache = nullptr;

    UStack              *fLanguageBreakEngines = nullptr;

    UnhandledEngine     *fUnhandledBreakEngine = nullptr;

    uint32_t            fDictionaryCharCount = 0;

    CharacterIterator  *fCharIter = &fSCharIter;

    UCharCharacterIterator fSCharIter {u"", 0};

    bool           fDone = false;

    int32_t *fLookAheadMatches = nullptr;

    UBool fIsPhraseBreaking = false;

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

    RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status);

    RuleBasedBreakIterator(UDataMemory* image, UBool isPhraseBreaking, UErrorCode &status);

    friend class RBBIRuleBuilder;
    friend class BreakIterator;

    RuleBasedBreakIterator(UErrorCode *status);

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 bool operator==(const BreakIterator& that) const override;

    inline bool operator!=(const BreakIterator& that) const {
        return !operator==(that);
    }

    virtual RuleBasedBreakIterator* clone() const override;

    virtual int32_t hashCode() const;

    virtual const UnicodeString& getRules() const;

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

    virtual CharacterIterator& getText() const override;

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

    virtual void adoptText(CharacterIterator* newText) override;

    virtual void setText(const UnicodeString& newText) override;

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

    virtual int32_t first() override;

    virtual int32_t last() override;

    virtual int32_t next(int32_t n) override;

    virtual int32_t next() override;

    virtual int32_t previous() override;

    virtual int32_t following(int32_t offset) override;

    virtual int32_t preceding(int32_t offset) override;

    virtual UBool isBoundary(int32_t offset) override;

    virtual int32_t current() const override;

    virtual int32_t getRuleStatus() const override;

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

    virtual UClassID getDynamicClassID() const override;

    static UClassID U_EXPORT2 getStaticClassID();

#ifndef U_FORCE_HIDE_DEPRECATED_API

    virtual RuleBasedBreakIterator *createBufferClone(void *stackBuffer,
                                                      int32_t &BufferSize,
                                                      UErrorCode &status) override;
#endif  // U_FORCE_HIDE_DEPRECATED_API

    virtual const uint8_t *getBinaryRules(uint32_t &length);

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


private:
    //=======================================================================
    // implementation
    //=======================================================================
    int32_t handleSafePrevious(int32_t fromPosition);

    int32_t handleNext();

    /*
     * Templatized version of handleNext() and handleSafePrevious().
     *
     * There will be exactly four instantiations, two each for 8 and 16 bit tables,
     * two each for 8 and 16 bit trie.
     * Having separate instantiations for the table types keeps conditional tests of
     * the table type out of the inner loops, at the expense of replicated code.
     *
     * The template parameter for the Trie access function is a value, not a type.
     * Doing it this way, the compiler will inline the Trie function in the
     * expanded functions. (Both the 8 and 16 bit access functions have the same type
     * signature)
     */

    typedef uint16_t (*PTrieFunc)(const UCPTrie *, UChar32);

    template<typename RowType, PTrieFunc trieFunc>
    int32_t handleSafePrevious(int32_t fromPosition);

    template<typename RowType, PTrieFunc trieFunc>
    int32_t handleNext();


    const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c, const char* locale);

  public:
#ifndef U_HIDE_INTERNAL_API

     void dumpCache();

    void dumpTables();
#endif  /* U_HIDE_INTERNAL_API */

#ifndef U_HIDE_INTERNAL_API

    static void U_EXPORT2 registerExternalBreakEngine(
                  ExternalBreakEngine* toAdopt, UErrorCode& status);
#endif  /* U_HIDE_INTERNAL_API */

};


U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_BREAK_ITERATION */

#endif /* U_SHOW_CPLUSPLUS_API */

#endif