#include <iostream>
#include <iomanip>
#include <gmpxx.h>
#include <string.h>
#include <locale>

#include <time.h>
#include <cmath>
#include <omp.h>

using namespace std;

const long long DEFAULT_LIMIT = 1'000'000;

class comma_numpunct : public std::numpunct<char>
{
  protected:
    virtual char do_thousands_sep() const
    {
        return ',';
    }

    virtual std::string do_grouping() const
    {
        return "\03";
    }
};

int main(int argc, char * argv[]) {

    long double sum = 0;
    long double term;
    long long limit = DEFAULT_LIMIT;

    timespec beginTS, endTS;

    clock_t start, stop;

    int sign = 1;
    int threads = 0;
    long long i;

    i = 1;
    while (i < argc) {
       if ( !strcmp("-l",argv[i]) and argv[i+1] != nullptr) {
          i++;
  	      limit= atoll(argv[i]);
	      i++;
       } else if ( !strcmp("-p",argv[i]) and argv[i+1] != nullptr) {
          i++;
	      threads = atoi(argv[i]);
	      i++;
       } else {
          cout << "Error: " << argv[i] << endl;
	      i++;
       }

    }

    if (limit < 0) {
       limit = DEFAULT_LIMIT;
    }

    if (threads > 0) {
        omp_set_num_threads(threads);
    }

    cout << endl;
    locale comma_locale(locale(), new comma_numpunct());
    cout.imbue(comma_locale);
    cout << "The limit is " <<  limit << endl;

    cout << setprecision(20);

    start = clock();
    clock_gettime(CLOCK_REALTIME, &beginTS);

    #pragma omp parallel
    #pragma omp master
    {
       cout << "Using " << omp_get_num_threads() << " threads " << endl;
    }


    #pragma omp parallel 
    {
        long long v;

        #pragma omp for reduction(+:sum)

        for(i=0;i<= limit;i++) {
            //cout << omp_get_thread_num() << " has " << i  << endl;
    
            if (i%2==0) {
	           sign = 1;
        	} else {
        	   sign = -1;
        	}

        	v = 2*i+1;

            term = sign *  4.0/(long double)(v);
            sum += term;
        }
    }

    stop = clock();
    clock_gettime(CLOCK_REALTIME, &endTS);

    cout << endl;

    cout << "Pi = " << sum << endl;
    cout << "Error:  " << fabsl(M_PIl - sum) << endl;
    cout << endl;

    cout << setprecision(5);
    cout << "Process time: " << (stop-start)/double(CLOCKS_PER_SEC) 
         << " seconds" << endl;
    time_t sec = endTS.tv_sec - beginTS.tv_sec;
    long nsec = endTS.tv_nsec - beginTS.tv_nsec;
    if (nsec < 0) {
       nsec+= 1'000'000'000;
    }
 
    nsec /=10000000;

    cout << "Wall clock time: " << sec << "." << nsec << " seconds" << endl;
    cout << endl;

    return 0;
}