inverted_index.h

Go to the documentation of this file.

#ifndef META_INVERTED_INDEX_H_
#define META_INVERTED_INDEX_H_

#include <queue>
#include <stdexcept>

#include "index/disk_index.h"
#include "index/make_index.h"

namespace meta
{

namespace corpus
{
class corpus;
class document;
}

namespace index
{

template <class>
class chunk_handler;

template <class, class>
class postings_data;
}
}

namespace meta
{
namespace index
{

class inverted_index : public disk_index
{
  public:
    class inverted_index_exception : public std::runtime_error
    {
      public:
        using std::runtime_error::runtime_error;
    };

    using primary_key_type = term_id;
    using secondary_key_type = doc_id;
    using postings_data_type = postings_data<term_id, doc_id>;
    using index_pdata_type = postings_data<std::string, doc_id>;
    using exception = inverted_index_exception;

    template <class Index, class... Args>
    friend std::shared_ptr<Index> make_index(const std::string&, Args&&...);

    template <class Index, template <class, class> class Cache, class... Args>
    friend std::shared_ptr<cached_index<Index, Cache>>
        make_index(const std::string& config_file, Args&&... args);

  protected:
    inverted_index(const cpptoml::table& config);

  public:
    inverted_index(inverted_index&&);

    inverted_index& operator=(inverted_index&&);

    inverted_index(const inverted_index&) = delete;

    inverted_index& operator=(const inverted_index&) = delete;

    virtual ~inverted_index();

    void tokenize(corpus::document& doc);

    virtual std::shared_ptr<postings_data_type>
        search_primary(term_id t_id) const;

    uint64_t doc_freq(term_id t_id) const;

    uint64_t term_freq(term_id t_id, doc_id d_id) const;

    uint64_t total_corpus_terms();

    uint64_t total_num_occurences(term_id t_id) const;

    double avg_doc_length();

  private:
    void create_index(const std::string& config_file);

    void load_index();

    bool valid() const;

    class impl;
    util::pimpl<impl> inv_impl_;
};
}
}

#endif