/home/palakars/workspace-inprogress/code-in-progress/aitools3/aitools3-aq-invertedindex-cpp/src/util/Vocabulary.hpp

Go to the documentation of this file.

/* Copyright (C) 2010 webis.de
 * All rights reserved.
 */
#ifndef AITOOLS_INVERTEDINDEX_VOCABULARY_HPP
#define AITOOLS_INVERTEDINDEX_VOCABULARY_HPP

#include "Externalizable.hpp"
#include <tr1/unordered_map>
#include <stdexcept>
#include <ostream>
#include <vector>
#include <string>
#include <map> // sorted!

namespace aitools {
namespace invertedindex {

class Vocabulary : public Externalizable {

public: // nested types

        typedef std::map<std::string, uint64_t> tree_map_t;

        typedef std::tr1::unordered_map<std::string, uint64_t> hash_map_t;

        typedef hash_map_t::const_iterator const_iterator;

        typedef hash_map_t::iterator iterator;

public: // class data

        static const char delim   = '\t';
        static const char newline = '\n';

public: // c'tor / d'tor

        Vocabulary();

        Vocabulary(const bfs::path& path) throw (std::invalid_argument);

        Vocabulary(const Vocabulary& vocabulary);

        ~Vocabulary();

public: // methods

        void accumulate(const std::string& word, uint64_t frequency = 1);

        const_iterator begin() const;

        iterator begin();

        void clear();

        bool contains(const std::string& word) const;

        const_iterator end() const;

        iterator end();

        const_iterator find(const std::string& word) const;

        iterator find(const std::string& word);

        uint64_t frequency(const std::string& word) const;

        void get_words(std::vector<std::string>& words) const;

        bool insert(const std::string& word, uint64_t frequency = 0);

        bool is_empty() const;

        void load(const bfs::path& file) throw (std::invalid_argument);

        void save(const bfs::path& file) throw (std::invalid_argument);

        size_t size() const;

private: // member

        hash_map_t hash_map_;

};

} // namespace invertedindex
} // namespace aitools

namespace std
{
        ostream&
        operator<<(ostream& os, const aitools::invertedindex::Vocabulary& vocab);

} // namespace std

#endif // AITOOLS_INVERTEDINDEX_VOCABULARY_HPP

---