corpus_utils.hpp

Go to the documentation of this file.

#ifndef QNLP_CORPUS_LOADER
#define QNLP_CORPUS_LOADER

#include <unordered_map>
#include "db_helper.hpp"

#ifndef QNLP_DB_FILE
#define QNLP_DB_FILE "qnlp_tagged_corpus.sqlite"
#endif

namespace QNLP{
    class CorpusUtils{
    public:

        CorpusUtils();
        CorpusUtils(const std::string filename);
        ~CorpusUtils();

        typedef std::unordered_map<std::string, std::unordered_map<std::string, unsigned int> > NTB;
        typedef std::unordered_map<std::string, std::unordered_map<unsigned int, std::string> > BTN;

        NTB& getNameToBin();
        BTN& getBinToName();

        void loadData(const std::string data_type, const std::string table="corpus");
        DBHelper& getDBHelper();

        void printData(const std::string type, const std::string table);

        const std::string get_database_filename();

        void clearData();

    private:
        //Maps for strings to binary pattern and reverse maps. Outer map keys are grammatical
        //types (nouns, verbs, etc.). Returned maps then hold unique entries. 
        NTB name_to_bin;
        BTN bin_to_name;

        //Database helper boject
        DBHelper db_helper;
        std::string database_filename;
    };
}

#endif