dataset.hpp

Go to the documentation of this file.

#pragma once
 
#include <pineforge/pineforge.h>
 
#include <cstddef>
#include <filesystem>
#include <memory>
#include <vector>
 
namespace pineforge {
namespace hpo {
 
enum class BinaryVerification { Full, Sidecar };
 
class Dataset final {
public:
    explicit Dataset(std::vector<pf_bar_t> bars);
 
    static Dataset load_csv(const std::filesystem::path& path);
 
    static Dataset load_bin(const std::filesystem::path& path,
                            BinaryVerification verification = BinaryVerification::Full);
 
    const pf_bar_t* data() const noexcept {
        return mapped_bars_ ? mapped_bars_.get() : bars_.data();
    }
    std::size_t size() const noexcept { return mapped_bars_ ? mapped_count_ : bars_.size(); }
    bool empty() const noexcept { return size() == 0; }
    const std::vector<pf_bar_t>& bars() const;
 
    void verify_unchanged() const;
 
private:
    struct MappedFile;
 
    Dataset(std::shared_ptr<const MappedFile> mapped_file,
            std::shared_ptr<const pf_bar_t> mapped_bars, std::size_t count);
 
    std::vector<pf_bar_t> bars_;
    std::shared_ptr<const MappedFile> mapped_file_;
    std::shared_ptr<const pf_bar_t> mapped_bars_;
    std::size_t mapped_count_ = 0;
};
 
struct DatasetView final {
    explicit DatasetView(std::shared_ptr<const Dataset> dataset);
 
    DatasetView(std::shared_ptr<const Dataset> dataset, std::size_t first, std::size_t count);
 
    const std::shared_ptr<const Dataset> owner;
    const std::size_t begin;
    const std::size_t length;
 
    const pf_bar_t* data() const noexcept;
    std::size_t size() const noexcept { return length; }
    bool empty() const noexcept { return length == 0; }
};
 
}  // namespace hpo
}  // namespace pineforge