concordia-library/concordia/regex_rule.hpp

#ifndef REGEX_ANNOTATION_HDR
#define REGEX_ANNOTATION_HDR

#include <string>
#include "concordia/common/config.hpp"
#include "concordia/tokenized_sentence.hpp"
#include "concordia/concordia_exception.hpp"
#include <boost/shared_ptr.hpp>
#include <boost/regex.hpp>
#include <boost/regex/icu.hpp>
#include <unicode/unistr.h>


typedef boost::error_info<struct my_tag, std::string> my_tag_error_info;

/*!
  Class for representing a regular expression annotation rule.
  Holds regex pattern string for matching and default value to assign
  to the annotations. Rule also has a type, given to all annotations
  produced by it.

*/
class RegexRule {
public:
    /*!
      Constructor.
        \param patternString regex pattern to match
        \param annoationType type of annotation
        \param value value to be assigned to the annotation
        \param caseSensitive case sensitivity of the pattern
    */
    RegexRule(std::string patternString,
              int annotationType,
              std::string value,
              bool caseSensitive = true)
              throw(ConcordiaException);

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

    /*! Applies regex annotation on tokenized sentence.
      \param sentence the input sentence
    */
    void apply(TokenizedSentence & sentence);

private:
    int _annotationType;

    std::string _value;

    boost::u32regex _pattern;
};

#endif