Question

I am trying to tokenize a database dump separated by commas. I only need to read the first word, which will tell me if this is the line I need and then tokenize the line and save each separated string in a vector.

I have had trouble keeping all of the datatypes in order. I use a method of getline:

string line;
    vector<string> tokens;

// Iterate through each line of the file
while( getline( file, line ) )
{
    // Here is where i want to tokenize. strtok however uses a character array and not a string.
}

The thing is, I only want to continue reading and tokenize a line if the first word is what I am after. Here is a sample of a line from the file:

example,1,200,200,220,10,550,550,550,0,100,0,-84,255

So, if I am after the string example, it goes ahead and tokenizes the rest of the line for my use and then stops reading from the file.

Should I be using strtok, stringstream or something else?

Thank you!

Answer 1

#include <iostream>
#include <fstream>
#include <vector>
using namespace std;

void do(ifstream& file) {
    string line;
    string prefix = "example,";

    // Get all lines from the file
    while (getline(file,line).good()) {
        // Compare the beginning for your prefix
        if (line.compare(0, prefix.size(), prefix) == 0) {
            // Homemade tokenization
            vector<string> tokens;
            int oldpos = 0;
            int pos;
            while ((pos = line.find(',', oldpos)) != string::npos) {
                tokens.push_back(line.substr(oldpos, pos-oldpos));
                oldpos = pos + 1;
            }
            tokens.push_back(line.substr(oldpos)); // don't forget the last bit
            // And here you are!
        }
    }
}