ulpartofspeechtagger.h

#ifndef ULPARTOFSPEECHTAGGER_H
#define ULPARTOFSPEECHTAGGER_H

#include "ulworker.h"
#include "uldissector.h"

class ULTaggedWord
{
public:
    friend class ULPartOfSpeechTagger;

    ULTaggedWord();
    ULTaggedWord(const ULTaggedWord& other);
    virtual ~ULTaggedWord();

    void operator=(const ULTaggedWord& other);
    bool operator==(const ULTaggedWord& other) const;
    bool operator<(const ULTaggedWord& other) const;

    const ULString& getSurfaceForm() const;
    const ULList<ULDerivation>& getTagList() const;

private:
    ULString surfaceForm;
    ULList<ULDerivation> tags;
    bool ordered;
};

class ULPartOfSpeechTagger : public ULWorker
{
public:
    ULPartOfSpeechTagger();
    ULPartOfSpeechTagger(const ULPartOfSpeechTagger& other);
    virtual ~ULPartOfSpeechTagger();

    ULPartOfSpeechTagger& operator=(const ULPartOfSpeechTagger& other);
    void clear();

    // Accessors.
    const ULLanguage& getLanguage() const;
    void setDissector(ULDissector *newDissector);
    ULDissector *getDissector();

    // ULWorker interfaces.
    virtual bool isServiceAvailable(const ULServiceDescriptor& service);
    virtual void getAvailableServices(ULList<ULServiceDescriptor>& services);

    // The tagger's core services.
    ULError getTags(const ULString& sentence, ULList<ULTaggedWord>& taggedWords);
    ULError getTags(const ULString& sentence, ULList<ULDerivation>& words);

private:
    ULDissector *dissector;

    void tokenize(const ULString& s, ULList<ULString>& tokenList);
    bool resolveAmbiguityWithRule(const ULTaggingRule& rule,
                                  ULList<ULTaggedWord>& taggedWords,
                                  uluint32& nAmbiguities);
    bool dissectionListHasMatchingCategory(const ULList<ULDerivation>& wordList,
                                           const ULPartOfSpeechCategory& category);
};

#endif