concordia-library/concordia/token_annotation.hpp

#ifndef TOKEN_ANNOTATION_HDR
#define TOKEN_ANNOTATION_HDR

#include "concordia/common/config.hpp"
#include "concordia/interval.hpp"

#include <string>

/*!
  Class representing annotation of char sequence as a token.
  It is a type of interval that is also storing information
  about the annoation type and value.

*/

class TokenAnnotation : public Interval {
public:
    /*! Constructor.
      \param start start index of the annotation (char-level, 0-based)
      \param end end index of the annotation (char-level, 0-based)
      \param annotationType annotation type
      \param value annotation value
    */
    TokenAnnotation(const SUFFIX_MARKER_TYPE start,
                    const SUFFIX_MARKER_TYPE end,
                    const int annotationType,
                    const std::string & value);

    /*! Destructor.
    */
    virtual ~TokenAnnotation();

    /*! Getter for annotation type.
      \returns annotation type
    */
    int getType() const {
        return _annotationType;
    }

    /*! Getter for annotation value.
      \returns annotation value
    */
    std::string getValue() const {
        return _value;
    }

    /*! Named entity annotation type
    */
    static int NE;

    /*! Word annotation type
    */
    static int WORD;

    /*! Html tag annotation type
    */
    static int HTML_TAG;

    /*! Stop word annotation type
    */
    static int STOP_WORD;

protected:
    int _annotationType;

    std::string _value;
};

#endif