michael@0: /**
michael@0:  *******************************************************************************
michael@0:  * Copyright (C) 2006,2012-2013, International Business Machines Corporation   *
michael@0:  * and others. All Rights Reserved.                                            *
michael@0:  *******************************************************************************
michael@0:  */
michael@0: 
michael@0: #ifndef DICTBE_H
michael@0: #define DICTBE_H
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uniset.h"
michael@0: #include "unicode/utext.h"
michael@0: 
michael@0: #include "brkeng.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: class DictionaryMatcher;
michael@0: 
michael@0: /*******************************************************************
michael@0:  * DictionaryBreakEngine
michael@0:  */
michael@0: 
michael@0: /**
michael@0:  * <p>DictionaryBreakEngine is a kind of LanguageBreakEngine that uses a
michael@0:  * dictionary to determine language-specific breaks.</p>
michael@0:  *
michael@0:  * <p>After it is constructed a DictionaryBreakEngine may be shared between
michael@0:  * threads without synchronization.</p>
michael@0:  */
michael@0: class DictionaryBreakEngine : public LanguageBreakEngine {
michael@0:  private:
michael@0:     /**
michael@0:      * The set of characters handled by this engine
michael@0:      * @internal
michael@0:      */
michael@0: 
michael@0:   UnicodeSet    fSet;
michael@0: 
michael@0:     /**
michael@0:      * The set of break types handled by this engine
michael@0:      * @internal
michael@0:      */
michael@0: 
michael@0:   uint32_t      fTypes;
michael@0: 
michael@0:   /**
michael@0:    * <p>Default constructor.</p>
michael@0:    *
michael@0:    */
michael@0:   DictionaryBreakEngine();
michael@0: 
michael@0:  public:
michael@0: 
michael@0:   /**
michael@0:    * <p>Constructor setting the break types handled.</p>
michael@0:    *
michael@0:    * @param breakTypes A bitmap of types handled by the engine.
michael@0:    */
michael@0:   DictionaryBreakEngine( uint32_t breakTypes );
michael@0: 
michael@0:   /**
michael@0:    * <p>Virtual destructor.</p>
michael@0:    */
michael@0:   virtual ~DictionaryBreakEngine();
michael@0: 
michael@0:   /**
michael@0:    * <p>Indicate whether this engine handles a particular character for
michael@0:    * a particular kind of break.</p>
michael@0:    *
michael@0:    * @param c A character which begins a run that the engine might handle
michael@0:    * @param breakType The type of text break which the caller wants to determine
michael@0:    * @return TRUE if this engine handles the particular character and break
michael@0:    * type.
michael@0:    */
michael@0:   virtual UBool handles( UChar32 c, int32_t breakType ) const;
michael@0: 
michael@0:   /**
michael@0:    * <p>Find any breaks within a run in the supplied text.</p>
michael@0:    *
michael@0:    * @param text A UText representing the text. The iterator is left at
michael@0:    * the end of the run of characters which the engine is capable of handling 
michael@0:    * that starts from the first (or last) character in the range.
michael@0:    * @param startPos The start of the run within the supplied text.
michael@0:    * @param endPos The end of the run within the supplied text.
michael@0:    * @param reverse Whether the caller is looking for breaks in a reverse
michael@0:    * direction.
michael@0:    * @param breakType The type of break desired, or -1.
michael@0:    * @param foundBreaks An allocated C array of the breaks found, if any
michael@0:    * @return The number of breaks found.
michael@0:    */
michael@0:   virtual int32_t findBreaks( UText *text,
michael@0:                               int32_t startPos,
michael@0:                               int32_t endPos,
michael@0:                               UBool reverse,
michael@0:                               int32_t breakType,
michael@0:                               UStack &foundBreaks ) const;
michael@0: 
michael@0:  protected:
michael@0: 
michael@0:  /**
michael@0:   * <p>Set the character set handled by this engine.</p>
michael@0:   *
michael@0:   * @param set A UnicodeSet of the set of characters handled by the engine
michael@0:   */
michael@0:   virtual void setCharacters( const UnicodeSet &set );
michael@0: 
michael@0:  /**
michael@0:   * <p>Set the break types handled by this engine.</p>
michael@0:   *
michael@0:   * @param breakTypes A bitmap of types handled by the engine.
michael@0:   */
michael@0: //  virtual void setBreakTypes( uint32_t breakTypes );
michael@0: 
michael@0:  /**
michael@0:   * <p>Divide up a range of known dictionary characters handled by this break engine.</p>
michael@0:   *
michael@0:   * @param text A UText representing the text
michael@0:   * @param rangeStart The start of the range of dictionary characters
michael@0:   * @param rangeEnd The end of the range of dictionary characters
michael@0:   * @param foundBreaks Output of C array of int32_t break positions, or 0
michael@0:   * @return The number of breaks found
michael@0:   */
michael@0:   virtual int32_t divideUpDictionaryRange( UText *text,
michael@0:                                            int32_t rangeStart,
michael@0:                                            int32_t rangeEnd,
michael@0:                                            UStack &foundBreaks ) const = 0;
michael@0: 
michael@0: };
michael@0: 
michael@0: /*******************************************************************
michael@0:  * ThaiBreakEngine
michael@0:  */
michael@0: 
michael@0: /**
michael@0:  * <p>ThaiBreakEngine is a kind of DictionaryBreakEngine that uses a
michael@0:  * dictionary and heuristics to determine Thai-specific breaks.</p>
michael@0:  *
michael@0:  * <p>After it is constructed a ThaiBreakEngine may be shared between
michael@0:  * threads without synchronization.</p>
michael@0:  */
michael@0: class ThaiBreakEngine : public DictionaryBreakEngine {
michael@0:  private:
michael@0:     /**
michael@0:      * The set of characters handled by this engine
michael@0:      * @internal
michael@0:      */
michael@0: 
michael@0:   UnicodeSet                fThaiWordSet;
michael@0:   UnicodeSet                fEndWordSet;
michael@0:   UnicodeSet                fBeginWordSet;
michael@0:   UnicodeSet                fSuffixSet;
michael@0:   UnicodeSet                fMarkSet;
michael@0:   DictionaryMatcher  *fDictionary;
michael@0: 
michael@0:  public:
michael@0: 
michael@0:   /**
michael@0:    * <p>Default constructor.</p>
michael@0:    *
michael@0:    * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the
michael@0:    * engine is deleted.
michael@0:    */
michael@0:   ThaiBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status);
michael@0: 
michael@0:   /**
michael@0:    * <p>Virtual destructor.</p>
michael@0:    */
michael@0:   virtual ~ThaiBreakEngine();
michael@0: 
michael@0:  protected:
michael@0:  /**
michael@0:   * <p>Divide up a range of known dictionary characters handled by this break engine.</p>
michael@0:   *
michael@0:   * @param text A UText representing the text
michael@0:   * @param rangeStart The start of the range of dictionary characters
michael@0:   * @param rangeEnd The end of the range of dictionary characters
michael@0:   * @param foundBreaks Output of C array of int32_t break positions, or 0
michael@0:   * @return The number of breaks found
michael@0:   */
michael@0:   virtual int32_t divideUpDictionaryRange( UText *text,
michael@0:                                            int32_t rangeStart,
michael@0:                                            int32_t rangeEnd,
michael@0:                                            UStack &foundBreaks ) const;
michael@0: 
michael@0: };
michael@0: 
michael@0: /*******************************************************************
michael@0:  * LaoBreakEngine
michael@0:  */
michael@0: 
michael@0: /**
michael@0:  * <p>LaoBreakEngine is a kind of DictionaryBreakEngine that uses a
michael@0:  * dictionary and heuristics to determine Lao-specific breaks.</p>
michael@0:  *
michael@0:  * <p>After it is constructed a LaoBreakEngine may be shared between
michael@0:  * threads without synchronization.</p>
michael@0:  */
michael@0: class LaoBreakEngine : public DictionaryBreakEngine {
michael@0:  private:
michael@0:     /**
michael@0:      * The set of characters handled by this engine
michael@0:      * @internal
michael@0:      */
michael@0: 
michael@0:   UnicodeSet                fLaoWordSet;
michael@0:   UnicodeSet                fEndWordSet;
michael@0:   UnicodeSet                fBeginWordSet;
michael@0:   UnicodeSet                fMarkSet;
michael@0:   DictionaryMatcher  *fDictionary;
michael@0: 
michael@0:  public:
michael@0: 
michael@0:   /**
michael@0:    * <p>Default constructor.</p>
michael@0:    *
michael@0:    * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the
michael@0:    * engine is deleted.
michael@0:    */
michael@0:   LaoBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status);
michael@0: 
michael@0:   /**
michael@0:    * <p>Virtual destructor.</p>
michael@0:    */
michael@0:   virtual ~LaoBreakEngine();
michael@0: 
michael@0:  protected:
michael@0:  /**
michael@0:   * <p>Divide up a range of known dictionary characters handled by this break engine.</p>
michael@0:   *
michael@0:   * @param text A UText representing the text
michael@0:   * @param rangeStart The start of the range of dictionary characters
michael@0:   * @param rangeEnd The end of the range of dictionary characters
michael@0:   * @param foundBreaks Output of C array of int32_t break positions, or 0
michael@0:   * @return The number of breaks found
michael@0:   */
michael@0:   virtual int32_t divideUpDictionaryRange( UText *text,
michael@0:                                            int32_t rangeStart,
michael@0:                                            int32_t rangeEnd,
michael@0:                                            UStack &foundBreaks ) const;
michael@0: 
michael@0: };
michael@0: 
michael@0: /******************************************************************* 
michael@0:  * KhmerBreakEngine 
michael@0:  */ 
michael@0:  
michael@0: /** 
michael@0:  * <p>KhmerBreakEngine is a kind of DictionaryBreakEngine that uses a 
michael@0:  * DictionaryMatcher and heuristics to determine Khmer-specific breaks.</p> 
michael@0:  * 
michael@0:  * <p>After it is constructed a KhmerBreakEngine may be shared between 
michael@0:  * threads without synchronization.</p> 
michael@0:  */ 
michael@0: class KhmerBreakEngine : public DictionaryBreakEngine { 
michael@0:  private: 
michael@0:     /** 
michael@0:      * The set of characters handled by this engine 
michael@0:      * @internal 
michael@0:      */ 
michael@0:  
michael@0:   UnicodeSet                fKhmerWordSet; 
michael@0:   UnicodeSet                fEndWordSet; 
michael@0:   UnicodeSet                fBeginWordSet; 
michael@0:   UnicodeSet                fMarkSet; 
michael@0:   DictionaryMatcher  *fDictionary; 
michael@0:  
michael@0:  public: 
michael@0:  
michael@0:   /** 
michael@0:    * <p>Default constructor.</p> 
michael@0:    * 
michael@0:    * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the 
michael@0:    * engine is deleted. 
michael@0:    */ 
michael@0:   KhmerBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status); 
michael@0:  
michael@0:   /** 
michael@0:    * <p>Virtual destructor.</p> 
michael@0:    */ 
michael@0:   virtual ~KhmerBreakEngine(); 
michael@0:  
michael@0:  protected: 
michael@0:  /** 
michael@0:   * <p>Divide up a range of known dictionary characters.</p> 
michael@0:   * 
michael@0:   * @param text A UText representing the text 
michael@0:   * @param rangeStart The start of the range of dictionary characters 
michael@0:   * @param rangeEnd The end of the range of dictionary characters 
michael@0:   * @param foundBreaks Output of C array of int32_t break positions, or 0 
michael@0:   * @return The number of breaks found 
michael@0:   */ 
michael@0:   virtual int32_t divideUpDictionaryRange( UText *text, 
michael@0:                                            int32_t rangeStart, 
michael@0:                                            int32_t rangeEnd, 
michael@0:                                            UStack &foundBreaks ) const; 
michael@0:  
michael@0: }; 
michael@0:  
michael@0: #if !UCONFIG_NO_NORMALIZATION
michael@0: 
michael@0: /*******************************************************************
michael@0:  * CjkBreakEngine
michael@0:  */
michael@0: 
michael@0: //indicates language/script that the CjkBreakEngine will handle
michael@0: enum LanguageType {
michael@0:     kKorean,
michael@0:     kChineseJapanese
michael@0: };
michael@0: 
michael@0: /**
michael@0:  * <p>CjkBreakEngine is a kind of DictionaryBreakEngine that uses a
michael@0:  * dictionary with costs associated with each word and
michael@0:  * Viterbi decoding to determine CJK-specific breaks.</p>
michael@0:  */
michael@0: class CjkBreakEngine : public DictionaryBreakEngine {
michael@0:  protected:
michael@0:     /**
michael@0:      * The set of characters handled by this engine
michael@0:      * @internal
michael@0:      */
michael@0:   UnicodeSet                fHangulWordSet;
michael@0:   UnicodeSet                fHanWordSet;
michael@0:   UnicodeSet                fKatakanaWordSet;
michael@0:   UnicodeSet                fHiraganaWordSet;
michael@0: 
michael@0:   DictionaryMatcher  *fDictionary;
michael@0: 
michael@0:  public:
michael@0: 
michael@0:     /**
michael@0:      * <p>Default constructor.</p>
michael@0:      *
michael@0:      * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the
michael@0:      * engine is deleted. The DictionaryMatcher must contain costs for each word
michael@0:      * in order for the dictionary to work properly.
michael@0:      */
michael@0:   CjkBreakEngine(DictionaryMatcher *adoptDictionary, LanguageType type, UErrorCode &status);
michael@0: 
michael@0:     /**
michael@0:      * <p>Virtual destructor.</p>
michael@0:      */
michael@0:   virtual ~CjkBreakEngine();
michael@0: 
michael@0:  protected:
michael@0:     /**
michael@0:      * <p>Divide up a range of known dictionary characters handled by this break engine.</p>
michael@0:      *
michael@0:      * @param text A UText representing the text
michael@0:      * @param rangeStart The start of the range of dictionary characters
michael@0:      * @param rangeEnd The end of the range of dictionary characters
michael@0:      * @param foundBreaks Output of C array of int32_t break positions, or 0
michael@0:      * @return The number of breaks found
michael@0:      */
michael@0:   virtual int32_t divideUpDictionaryRange( UText *text,
michael@0:           int32_t rangeStart,
michael@0:           int32_t rangeEnd,
michael@0:           UStack &foundBreaks ) const;
michael@0: 
michael@0: };
michael@0: 
michael@0: #endif
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0:     /* DICTBE_H */
michael@0: #endif