excel_columns_segregation.cpp

#include <iostream>
#include <cstdio>
#include <cstdlib>
#include <vector>
#include <algorithm>
#include <fstream>
#include <sstream>
#include <string>
#include <vector>

using namespace std;
//char input[960];


int main(int argc, char const *argv[])
{
	ifstream if1;
	ofstream of1;
	//char chars[] = "[],";
	
	//int sizeOfArray = sizeof(chars) / sizeof(chars[0]);;
    if1.open("super_cleaned.csv");
    of1.open("output_cleaned_data.csv");
    string str1, line;
    vector<int> csvRow;
    int prev_loc = 1908;
    /*while (getline(if1, str1))
    {
    	str1.erase(0,13);
    	for (unsigned int i = 0; i < sizeOfArray; ++i)
		   {
		      str1.erase (remove(str1.begin(), str1.end(), chars[i]), str1.end());
		   }
        of1<<str1<<endl;
    }*/
    csvRow.resize(42);   
    for (int i = 0; i < 42; i++)
        {
            csvRow[i] = 0;
        } 
    /*for (int i = 0; i < 42; ++i)
                        {
                            cout<<csvRow[i]<<":";
                        }
    cout<<endl;*/
    while(getline(if1,line))
    {
        stringstream   linestream(line);
        string value;
        int counter = 0;
        int loc;
        
        while(getline(linestream,value,','))
        {
           // cout<<"counter"<<counter<<endl;
            if (counter == 0)
            {
                stringstream convert(value);
                convert>>loc;
                
                if (loc == prev_loc)
                {
                    /* code */
                }
                else
                {
                    if (prev_loc > 1908)
                    {
                        for (int i = 0; i < 42; ++i)
                        {
                            if (i == 41)
                            {
                                of1<<csvRow[i];
                            }
                            else{of1<<csvRow[i]<<",";}
                        }
                        of1<<endl; 
                    }
                    
                    //csvRow.resize(42); 
                    for (int i = 0; i < 42; ++i)
                    {
                        csvRow[i] = 0;
                    }
                    csvRow[0] = loc;   
                }
                
                //cout << loc << ",";

            }
            else if (counter == 1)
            {
                if (value == "ACT"){csvRow[1] += 1;}
                else if (value == "AMG"){csvRow[2] += 1;}
                else if (value == "CTL"){csvRow[3] += 1;}
                else if (value == "CLD"){csvRow[4] += 1;}
                else if (value == "DIS"){csvRow[5] += 1;}
                else if (value == "DOR"){csvRow[6] += 1;}
                else if (value == "LIQ"){csvRow[7] += 1;}
                else if (value == "SOF"){csvRow[8] += 1;}
                else if (value == "UNL"){csvRow[9] += 1;}
                else if (value == "UPS"){csvRow[10] += 1;}
                else if (value == ""){}
                else{cout<<"Counter 1 fault!!!"<<value<<endl;}
                
            }
            else if (counter == 2)
            {
                if (value == "OPC"){csvRow[11] += 1;}
                else if (value == "PRV"){csvRow[12] += 1;}
                else if (value == "PUB"){csvRow[13] += 1;}
                else if (value == ""){}
                else{cout<<"Counter 2 fault!!!"<<value<<endl;}
            }
            else if (counter == 3)
            {
                if (value == "CLG"){csvRow[14] += 1;}
                else if (value == "CLS"){csvRow[15] += 1;}
                else if (value == "UCM"){csvRow[16] += 1;}
                else if (value == ""){}
                else{cout<<"Counter 3 fault!!!"<<value<<endl;}
            }
            else if (counter == 4)
            {
                if (value == "AAA"){csvRow[17] += 1;}
                else if (value == "BUS"){csvRow[18] += 1;}
                else if (value == "CPS"){csvRow[19] += 1;}
                else if (value == "CON"){csvRow[20] += 1;}
                else if (value == "EGW"){csvRow[21] += 1;}
                else if (value == "FIN"){csvRow[22] += 1;}
                else if (value == "INS"){csvRow[23] += 1;}
                else if (value == "MFS"){csvRow[24] += 1;}
                else if (value == "MLP"){csvRow[25] += 1;}
                else if (value == "MME"){csvRow[26] += 1;}
                else if (value == "MMC"){csvRow[27] += 1;}
                else if (value == "MOS"){csvRow[28] += 1;}
                else if (value == "MPP"){csvRow[29] += 1;}
                else if (value == "MTT"){csvRow[30] += 1;}
                else if (value == "MWP"){csvRow[31] += 1;}
                else if (value == "MQY"){csvRow[32] += 1;}
                else if (value == "RER"){csvRow[33] += 1;}
                else if (value == "TRD"){csvRow[34] += 1;}
                else if (value == "TSC"){csvRow[35] += 1;}
                else if (value == ""){}
                else{cout<<"Counter 4 fault!!!"<<value<<endl;}
            }
            else if (counter == 5)
            {
                //cout<<"Value"<<value<<endl;
                if (value[0] == 'C'){csvRow[36] += 1;} // I had to use the 1st character, I don't know why?
                else if (value[0] == 'G'){csvRow[37] += 1;}
                else if (value[0] == 'I'){csvRow[38] += 1;}
                else if (value[0] == 'O'){csvRow[39] += 1;}
                else if (value[0] == 'S'){csvRow[40] += 1;}
                else if (value[0] == 'U'){csvRow[41] += 1;}
                else if (value == " "){}
                else{cout<<"Counter 5 fault!!!"<<value<<endl;}
            }
            else{cout << "This is the End of the World!!!";}
            counter++;

        }
        //std::cout << "Line Finished" << std::endl;
        prev_loc = loc;

    }
    for (int i = 0; i < 42; ++i)
    {
        if (i == 41)
        {
            of1<<csvRow[i];
        }
        else{of1<<csvRow[i]<<",";}
    }
    cout<<endl; 
   // cout<<"here"<<endl;
    if1.close();
    of1.close();
    	

	return 0;
}

Made for the DNA.