ParquetStreamer.hpp

/*
 * (c) Copyright 2021 CORSIKA Project, corsika-project@lists.kit.edu
 *
 * This software is distributed under the terms of the GNU General Public
 * Licence version 3 (GPL Version 3). See file LICENSE for a full version of
 * the license.
 */

#pragma once

#include <string>

// clang-format-off, NOTE: the order of these includes is *important*
// you will get unhelpful compiler errors about unknown
// operator definitions if these are reordered
#include <parquet/stream_writer.h>
#include <parquet/arrow/schema.h>
#include <arrow/io/file.h>
// clang-format-on

namespace corsika {

  class ParquetStreamer {

  public:
    ParquetStreamer();

    void initStreamer(std::string const& filepath);

    template <typename... TArgs>
    void addField(TArgs&&... args);

    void enableCompression(int const level = 3);

    void buildStreamer();

    void closeStreamer();

    std::shared_ptr<parquet::StreamWriter> getWriter();

    bool isInit() const { return isInit_; }

  private:
    bool isInit_ = false;                        
    parquet::WriterProperties::Builder builder_; 
    parquet::schema::NodeVector fields_;         
    std::shared_ptr<parquet::schema::GroupNode> schema_;   
    std::shared_ptr<arrow::io::FileOutputStream> outfile_; 
    std::shared_ptr<parquet::StreamWriter> writer_; 

  }; // class ParquetStreamer
} // namespace corsika

#include <corsika/detail/output/ParquetStreamer.inl>