segmenter.h

Go to the documentation of this file.

#ifndef META_UTF_SEGMENTER_H_
#define META_UTF_SEGMENTER_H_

#include <string>
#include <vector>
#include "util/pimpl.h"

namespace meta
{
namespace utf
{

class segmenter
{
  public:
    class segment
    {
      public:
        segment(int32_t begin, int32_t end);

      private:
        friend segmenter;
        // using int32_t here because of ICU, which accepts only int32_t as
        // its indexes
        int32_t begin_;
        int32_t end_;
    };

    segmenter();

    segmenter(const segmenter&);

    ~segmenter();

    void set_content(const std::string& str);

    std::vector<segment> sentences() const;

    std::vector<segment> words() const;

    std::vector<segment> words(const segment& seg) const;

    std::string content(const segment& seg) const;

  private:
    class impl;
    util::pimpl<impl> impl_;
};
}
}
#endif