disk_index_impl.h

Go to the documentation of this file.

#ifndef META_INDEX_DISK_INDEX_IMPL_H_
#define META_INDEX_DISK_INDEX_IMPL_H_

#include <mutex>

#include "index/disk_index.h"
#include "index/string_list.h"
#include "index/vocabulary_map.h"
#include "util/disk_vector.h"
#include "util/invertible_map.h"
#include "util/optional.h"

namespace meta
{
namespace index
{

class string_list_writer;

enum index_file
{
    DOC_IDS_MAPPING = 0,
    DOC_IDS_MAPPING_INDEX,
    DOC_SIZES,
    DOC_LABELS,
    DOC_UNIQUETERMS,
    LABEL_IDS_MAPPING,
    POSTINGS,
    TERM_IDS_MAPPING,
    TERM_IDS_MAPPING_INVERSE
};

class disk_index::disk_index_impl
{
  public:
    friend disk_index;

    const static std::vector<const char*> files;

    void initialize_metadata(uint64_t num_docs = 0);

    void load_doc_sizes(uint64_t num_docs = 0);

    void load_labels(uint64_t num_docs = 0);

    void load_unique_terms(uint64_t num_docs = 0);

    void load_doc_id_mapping();

    void load_term_id_mapping();

    void load_label_id_mapping();

    void load_postings();

    void save_label_id_mapping();

    string_list_writer make_doc_id_writer(uint64_t num_docs) const;

    void set_label(doc_id id, const class_label& label);

    void set_length(doc_id id, uint64_t length);

    void set_unique_terms(doc_id id, uint64_t terms);

    const io::mmap_file& postings() const;

    uint64_t total_unique_terms() const;

    label_id doc_label_id(doc_id id) const;

    std::vector<class_label> class_labels() const;

  private:
    label_id get_label_id(const class_label& lbl);

    std::string index_name_;

    util::optional<string_list> doc_id_mapping_;

    util::optional<util::disk_vector<double>> doc_sizes_;

    util::optional<util::disk_vector<label_id>> labels_;

    util::optional<util::disk_vector<uint64_t>> unique_terms_;

    util::optional<vocabulary_map> term_id_mapping_;

    util::invertible_map<class_label, label_id> label_ids_;

    util::optional<io::mmap_file> postings_;

    mutable std::mutex mutex_;
};
}
}
#endif