In a question I answered I posted the following code:

#include <iostream>
#include <fstream>
#include <algorithm>
#include <sstream>

int main()
    fstream iFile("names.txt", ios::in);

    // This does not work so replacing with code that does
    // iFile >> file;

    // This is not the best way.
    // But the closest to the intention of the original post.

    // Get size of file.
    std::streampos          length = ifile.tellg();

    // Copy file into string.
    std::string       file(length, '\0');

    // Original code continues.

    std::istringstream ss(file);
    std::string token;

    std::vector<std::string> names;

    while(std::getline(ss, token, ',')) {

    for (unsigned int i = 0; i < names.size(); i++) {
         auto it = std::remove_if(names[i].begin(), names[i].end(), [&] (char c) { return c == '"'; });
         names[i] = std::string(names[i].begin(), it);

    for (unsigned int i = 0; i < names.size(); i++) {
        std::cout << "names["<<i<<"]: " << names[i] << std::endl;

Would this be inefficient for larger files? Would it be better if I read the file into one string, did all the manipulation on that, and then put it into a vector?

  • \$\begingroup\$ Is the code all inside main()? You can't have free-floating code like that in C++. \$\endgroup\$ Commented Dec 22, 2013 at 10:00
  • \$\begingroup\$ This code obviously does not work without main. So unless people have objects I will add main() around all the code so we are at least commenting on the code in the same way. \$\endgroup\$ Commented Dec 23, 2013 at 21:42
  • \$\begingroup\$ It still obviously does not work because iFile >> file; that is not doing very much. It reads a single space separated word. So I am going to change that to read the whole file into a string. \$\endgroup\$ Commented Dec 23, 2013 at 21:43
  • \$\begingroup\$ Now it works (as best it can). \$\endgroup\$ Commented Dec 23, 2013 at 21:51

2 Answers 2


This code looks a little wild. I'd tame it like this:

#include <algorithm> // for std::remove
#include <fstream>   // for std::ifstream
#include <string>    // for std::string and std::getline
#include <utility>   // for std::move
#include <vector>    // for std::vector

std::ifstream infile("thefile.txt");    // we don't really care where you got the name from

std::vector<std::string> names;

for (std::string line; std::getline(infile, line, ','); )
    line.erase(std::remove(line.begin(), line.end(), '"'), line.end());

// done

A few points to note:

  • Keep it simple. Don't overthink.

  • Use ready-made algorithms, don't reinvent wheels (std::remove removes values).

  • Include what you use (std::string, std::getline).

  • Use algorithms, don't hand-roll your own loops. (erase/remove)

  • Understand what algorithms can do for you (don't hand-write your own "erase"). Get familiar with the standard library.

  • Don't leak scopes. The line string is only needed within the loop, and no further.

  • Efficiency tip: You can move from the line string that you no longer need and save yourself a copy. Also, we process everything in one step; no need to loop repeatedly.

Furthermore, I suggest factoring this logic into a file:

std::vector<std::string> tokenize_file(std::string const & filename, char delimiter)
    std::vector<std::string> result;

    // ...

    return result;


std::vector<std::string> names = tokenize_file("thefile.txt", ',');

Instead of constructing an fstream, reading it all into a string (which is misleadingly named file, by the way), and creating an istringstream from that, why not just create an ifstream and call getline on it directly?

You appear to be trying to parse a CSV file where the fields may be double quoted. If that is your intention, then simply deleting all " characters is the wrong behaviour. See RFC 4180 for commonly accepted quoting rules for CSV documents. Consider using a CSV parsing library instead of rolling your own CSV parser.

Your code would be clearer if you unquoted each token before adding it to names.

  • 1
    \$\begingroup\$ Totally agree on using CSV parsing library. CSV is one of those badly (not though through enough) defined formats that has a billion edge cases that you would never think about yourself. The simple case is trivial but doing it correctly for all cases is exceptionally hard. \$\endgroup\$ Commented Dec 23, 2013 at 21:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.