document.h

Go to the documentation of this file.

#ifndef META_DOCUMENT_H_
#define META_DOCUMENT_H_

#include <string>
#include <unordered_map>

#include "meta.h"
#include "util/optional.h"

namespace meta
{
namespace corpus
{

class document
{
  public:
    document(const std::string& path = "[NONE]", doc_id d_id = doc_id{0},
             const class_label& label = class_label{"[NONE]"});

    void increment(const std::string& term, double amount);

    std::string path() const;

    const class_label& label() const;

    std::string name() const;

    void name(const std::string& n);

    uint64_t length() const;

    double count(const std::string& term) const;

    const std::unordered_map<std::string, double>& counts() const;

    void content(const std::string& content,
                 const std::string& encoding = "utf-8");

    void encoding(const std::string& encoding);

    const std::string& content() const;

    const std::string& encoding() const;

    doc_id id() const;

    bool contains_content() const;

    void label(class_label label);

  private:
    std::string path_;

    doc_id d_id_;

    class_label label_;

    std::string name_;

    size_t length_;

    std::unordered_map<std::string, double> counts_;

    util::optional<std::string> content_;

    std::string encoding_;
};
}
}

#endif