#include <string>
#include <vector>
#include <fstream>
#include <iostream>
#include <sstream>
#include <stdexcept>

#include "Dataset.hpp"

static void fill_from_string(std::vector<double>& v, const string & str) {
    double val;
    istringstream in(str);
    while (!in.eof()) {
        in >> val;
        v.push_back(val);
    }
}

Dataset::Dataset(string fname, double enc0, double enc1){
    ifstream ifs(fname);
    if (!ifs.is_open()) {
        throw runtime_error("Could not open data file.");
    }
    string buffer;
    while (!ifs.eof()){
        getline(ifs, buffer);
        if (buffer.length() == 0) break;
        vector<double> v;
        fill_from_string(v,buffer);
        targetc.push_back(v[v.size()-1]); // put label in
        v.resize(v.size()-1); // Lose the class label
        invec.push_back(v);
    }
    // Determine number of classes
    size_t maxc = 0;
    for (auto c : targetc){
        if (c>maxc) maxc = c;
    }
    nclasses = maxc;

    // Fill the classwise counts:
    ninc.resize(nclasses+1);
    for (auto & n : ninc) n=0;
    ninc[0]=nclasses;
    for (auto c : targetc){
        ninc[c]++;
    }

    // Prepare prototypes for MLP classifiers:
    for (size_t i = 0; i<nclasses; ++i){
        vector <double> p(nclasses);
        for (size_t j = 0; j<nclasses; ++j) {
            p[j] = (i==j)?enc1:enc0;
        }
        prototypes.push_back(p);
    }
}

size_t Dataset::getNClasses() const {return nclasses;}

size_t Dataset::getNRows() const {return invec.size();}

size_t Dataset::getTargetClass(size_t ind) const {
    return targetc[ind];
}

size_t Dataset::getNRowsInClass(size_t c) const{
    return ninc[c];
}

const vector <double> &
Dataset::prototype(size_t row){
    return prototypes[targetc[row]-1];
}


const vector <double> &
Dataset::row(size_t i) const {
    return invec[i];
}

void
Dataset::toStream(ostream & ost) const {
    for (size_t irow = 0; irow<getNRows(); ++irow){
        const vector<double> & r = row(irow);
        for (size_t i=0;i<r.size();++i){
            ost << r[i];
            ost << " ";
            //if (i<row.size()-1) ost << " ";
        }
        ost << targetc[irow];
        ost << endl;
    }
}