CSVReader.hpp

//
// Created by jstark on 23.12.21.
//

#ifndef OPENFPM_IO_CSVREADER_HPP
#define OPENFPM_IO_CSVREADER_HPP

#include <iostream>
#include <sstream>
#include <string>

#include "Vector/map_vector.hpp"
#include "util/PathsAndFiles.hpp"
#include "VCluster/VCluster.hpp"

template <typename T>
T string_to_type(const std::string & string_to_convert)
{
    T x_new_type;
    std::istringstream istream_string_to_convert(string_to_convert);
    istream_string_to_convert >> x_new_type;
    return x_new_type;
}


template <typename T>
void read_csv_to_vector(const std::string & input_file, openfpm::vector<T> & output_vector, size_t & m, size_t & n)
{
    size_t total_size = 0;
    m = 0;
    n = 0;
    auto & v_cl = create_vcluster();
    // File is accessed and read only by one process
    if (v_cl.rank() == 0)
    {
        if(!check_if_file_exists(input_file))
        {
            std::cout << "Cannot find < " <<  input_file << " >. Please check path specification. Aborting..."
                    << std::endl;
            abort();
        }
        std::ifstream file(input_file); // Create file pointer and open file
        std::string line, element;
        while (getline(file, line)) // Read entire row and store as one string in line
        {
            std::stringstream stream_line(line); // Needed to break line into elements later one
            while (getline(stream_line, element, ',')) // Read every column of line and store content as string in element
            {
                output_vector.add(string_to_type<T>(element)); // Convert element from string to type T and append to
                // output_vector
                total_size += 1; // Increase n by one for each element read
            }
            m += 1; // Increase m by one for each row read
        }
        if(m >= 1) n = total_size / m; // If at least one row present, divide number of elements read by number of rows to get
        // the number of columns
    }
    // Distribute size of the lin. vector to all other processes s.t. their output_vector can be resized accordingly.
    MPI_Bcast(&total_size,1,MPI_UNSIGNED,0,MPI_COMM_WORLD);
    MPI_Bcast(&m,1,MPI_UNSIGNED,0,MPI_COMM_WORLD);
    MPI_Bcast(&n,1,MPI_UNSIGNED,0,MPI_COMM_WORLD);
    
    if (v_cl.rank() != 0)
    {
        output_vector.resize(total_size);
    }
    // After the output_vector has been resized to correct size, it can receive the content from process 0.
    v_cl.Bcast(output_vector, 0);
    v_cl.execute();
}

#endif //OPENFPM_IO_CSVREADER_HPP