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 RegexRule {
public:
    RegexRule(std::string patternString,
              int annotationType,
              std::string value,
              bool caseSensitive = true)
              throw(ConcordiaException);

    virtual ~RegexRule();

    void apply(TokenizedSentence & sentence);

private:
    int _annotationType;

    std::string _value;

    boost::u32regex _pattern;
};

#endif