csv.hpp

/****************************************************************************
 *  hipipe library
 *  Copyright (c) 2017, Cognexa Solutions s.r.o.
 *  Copyright (c) 2018, Iterait a.s.
 *  Author(s) Filip Matzner
 *
 *  This file is distributed under the MIT License.
 *  See the accompanying file LICENSE.txt for the complete license agreement.
 ****************************************************************************/
 
#pragma once
 
#include <hipipe/core/dataframe.hpp>
 
#include <range/v3/iterator/default_sentinel.hpp>
 
#include <experimental/filesystem>
#include <fstream>
#include <iostream>
#include <vector>
 
namespace hipipe {
 
 
class csv_istream_range : public ranges::view_facade<csv_istream_range> {
private:
    friend ranges::range_access;
    using single_pass = std::true_type;
    enum class RowPosition{Normal, Last, End};
 
    std::istream* in_;
    char separator_;
    char quote_;
    char escape_;
 
    std::vector<std::string> row_;
    RowPosition row_position_ = RowPosition::Normal;
 
    class cursor {
    private:
        csv_istream_range* rng_;
 
    public:
        cursor() = default;
        explicit cursor(csv_istream_range& rng) noexcept
          : rng_{&rng}
        {}
 
        void next()
        {
            rng_->next();
        }
 
        std::vector<std::string>& read() const noexcept
        {
            return rng_->row_;
        }
 
        std::vector<std::string>&& move() const noexcept
        {
            return std::move(rng_->row_);
        }
 
        bool equal(ranges::default_sentinel_t) const noexcept
        {
            return rng_->row_position_ == RowPosition::End;
        }
    };
 
    // parse csv field and return whether the next separator is found
    std::tuple<std::string, bool> parse_field();
 
    // parse csv row
    void next();
 
    cursor begin_cursor() { return cursor{*this}; }
 
public:
    csv_istream_range() = default;
 
    explicit csv_istream_range(std::istream& in,
                               char separator = ',',
                               char quote = '"',
                               char escape = '\\');
};
 
dataframe read_csv(
  std::istream& in,
  int drop = 0,
  bool has_header = true,
  char separator = ',',
  char quote = '"',
  char escape = '\\');
 
 
dataframe read_csv(
  const std::experimental::filesystem::path& file,
  int drop = 0,
  bool header = true,
  char separator = ',',
  char quote = '"',
  char escape = '\\');
 
 
std::ostream& write_csv_row(
  std::ostream& out,
  const std::vector<std::string>& row,
  char separator = ',',
  char quote = '"',
  char escape = '\\');
 
 
std::ostream& write_csv(
  std::ostream& out,
  const dataframe& df,
  char separator = ',',
  char quote = '"',
  char escape = '\\');
 
 
void write_csv(
  const std::experimental::filesystem::path& file,
  const dataframe& df,
  char separator = ',',
  char quote = '"',
  char escape = '\\');
 
}  // namespace hipipe