read a csv file and and add its all data into vector in c++

Question

For example to add the following CSV data:

I am trying to add CSV file into a 2D array string vector and get the sum of each column. The following program didn't work properly,

vector<string> read_csv(string filename){

    vector<string> result;
    fstream fin;
    fin.open(filename, ios::in);

    if(!fin.is_open())
        throw std::runtime_error("Could not open file");

    std::string line, colname;
    int val;

    // Read the column names
    if(fin.good())
    {
        std::getline(fin, line);
        std::stringstream ss(line);
        while(std::getline(ss, colname, ',')){
            result.push_back(colname);
            cout << colname << endl;
        }
    }

    while(std::getline(fin, line))
    {
        std::stringstream ss(line);
        int colIdx = 0;
        while(ss >> val){

            if(ss.peek() == ',') ss.ignore();
            colIdx++;
        }
    }
    fin.close();
    return result;
}

when I tried to go through the vector, I didn't get a proper result. It showed only the column names.

for (int i = 0; i < vectorCsv.size(); ++i) 
{
        cout << vectorCsv[i] << endl;
}

I couldn't find whether the error is in read_csv() function or in the forloop. Thank you for looking at this problem.

Answer 1

In your while loop, you never pushed any values to your vector.

It looks like you have everything you need to read the csv into a vector right here. Only problem is you stopped at column names.

// Read the column names
    if(fin.good())
    {
        std::getline(fin, line);
        std::stringstream ss(line);
        while(std::getline(ss, colname, ',')){
            result.push_back(colname);
            cout << colname << endl;
        }
    }

Try changing the code I copied above to:

// Read the column names
    while(std::getline(fin, line))
    {
        std::getline(fin, line);
        std::stringstream ss(line);
        while(std::getline(ss, colname, ',')){
            result.push_back(colname);
            cout << colname << endl;
        }
    }

Answer 2

1. Don't try to create vectors of std::string s, that's probably not very efficient - each string being allocated and de-allocated separately.

2. Don't read CSV's yourself - you're reinventing the wheel. Use an existing library. Here's a question about finding one at Software Recommendations StackExchange:

   Modern C++ CSV reader (and maybe writer) library

Answer 3

I cannot believe that we are using a library for such an ultra simple thing like splitting a std::string into tokens.

C++ has, since long, a build in and dedicated functionality, specifically designed for this purpose, to tokenize strings (split strings into tokens). And because such a simple dedicated function, designed for this purpose, is available, it simply should be used. There is no need for external libraries or complicated constructs. Simply use the std::sregex_token_iterator .

This is an iterator (like many other iterators), that iterates over tokens (sub-strings) of a string. So, what we want.

We can then use the std::vector s range constructor to write something simple like this:

std::vector tokens(std::sregex_token_iterator(line.begin(), line.end(), delimiter, -1), {}));

So, we define a variable with the name "tokens" of type std::vector (with CTAD the type of the vector is automatically deduced). We use its range constructor and provide a begin and an end iterator. The begin iterator is the std::sregex_token_iterator and the end-iterator is its default-initialized counterpart.

To put such a vector into a 2D Vector, we use the outer vectors emplace_back function and do an inplace construction for the inner vector.

So you read the whole CSV-File with 2 statements

• a simple for loop
• a simple emplace back with the std::sregex_token_iterator

        // We will read all lines of the source file with a simple for loop and std::getline
        for (std::string line{}; std::getline(csvFile, line); ) {

            // We will split the one big string into tokens (sub-strings) and add it to our 2D array
            csvData.emplace_back(std::vector<std::string>(std::sregex_token_iterator(line.begin(), line.end(), delimiter, -1), {}));
        }

So, why should you use a library for such a simple task that you can do with 2 statements? I personally fail to understand that. Therefore, I find that the advise in the accepted answer is flat wrong. But, to avoid starting religious discussions: This is my very personal humble opinion and everybody can do what he wants.

Please see a complete working example, which solves your problem, with just a few lines of code . . .

#include <iostream>
#include <fstream>
#include <vector>
#include <regex>

const std::string csvFileName{ "r:\\csv.csv" };
const std::regex delimiter{ "," };

int main() {

    // Open the file and check, if it could be opened
    if (std::ifstream csvFile(csvFileName); csvFile) {

        // This is our "2D array string vector" as described in your post
        std::vector<std::vector<std::string>> csvData{};


        // Read the complete CSV FIle into a 2D vector ----------------------------------------------------
        // We will read all lines of the source file with a simple for loop and std::getline
        for (std::string line{}; std::getline(csvFile, line); ) {

            // We will split the one big string into tokens (sub-strings) and add it to our 2D array
            csvData.emplace_back(std::vector<std::string>(std::sregex_token_iterator(line.begin(), line.end(), delimiter, -1), {}));
        }
        // -------------------------------------------------------------------------------------------------


        // This is for summing up values
        double DP{}, Dta{}, Dts{};

        // Iterate in a simple for loop through all elements of the 2D vector, convert the vlaues to double and sum them up
        for (size_t i = 1U; i < csvData.size(); ++i) {

            DP += std::stod(csvData[i].at(1));
            Dta += std::stod(csvData[i].at(2));
            Dts += std::stod(csvData[i].at(3));
        }

        // Sho the result to the user
        std::cout << "\nSums:  DP: " << DP << "  Dta: " << Dta << "  Dts: " << Dts << "\n";
    }
    else { // In case that we could not open the source file
        std::cerr << "\n*** Error. Could not open file " << csvFileName << "\n\n";
    }
    return 0;
}

But as said, everybdoy can do whatever he wants.