package com.testpbc4;

import java.io.File;
import java.io.FileWriter;
import java.io.PrintWriter;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.Random;
import java.util.Scanner;

import org.apache.commons.lang3.StringUtils;

import weka.classifiers.trees.PBC4cip;
import weka.core.Instances;
import weka.core.converters.ConverterUtils.DataSource;

public class App {
    static int numFolds = 5;
    static String path = "data/";
    static String datasetsFile = path + "datasets.txt";
    static String outputFile = "results_pbc4.txt";
    static boolean debug = false;

    public static void printAndExit(String s) {
        System.out.println(s);
        System.exit(0);
    }

    private static Instances getInstances(String fileName) {
        DataSource source = null;
        try {
            source = new DataSource(String.format(path + "%s.arff", fileName));
            return source.getDataSet();
        } catch (Exception e) {
            printAndExit(
                    String.format("*** Error trying to read " + path + "%s.arff file... (%s)", fileName,
                            e.getMessage()));
        }
        return null;
    }

    private static void validation(Instances instancesIn, int seed, ArrayList<Double> accuracy,
            ArrayList<Double> timeSpent, ArrayList<Double> complexity) throws Exception {
        PBC4cip pb4 = new PBC4cip();
        Random rand = new Random(seed);
        pb4.setSeed(seed);
        Instances instances;
        Instances training = null;
        Instances test = null;

        instances = instancesIn;
        instances.randomize(rand);
        instances.stratify(numFolds);
        for (int i = 0; i < numFolds; i++) {
            long startTime = System.currentTimeMillis();
            training = instances.trainCV(numFolds, i);
            test = instances.testCV(numFolds, i);
            pb4.buildClassifier(training);
            // let's classify
            int aciertos = 0;
            for (int j = 0; j < test.numInstances(); j++) {
                double v = pb4.classifyInstance(test.instance(j));
                double trueValue = test.instance(j).value(test.classIndex());
                if (trueValue == v)
                    aciertos++;
            }
            double acc = aciertos / (double) test.numInstances();
            double tspent = (System.currentTimeMillis() - startTime) / 1000.0;
            if (debug) {
                System.err.println(String.format("%f, %f", acc, tspent));
            }
            accuracy.add(acc);
            timeSpent.add(tspent);
            complexity.add(0.0);
        }
    }

    private static double[] meanAndDeviation(ArrayList<Double> input) {
        double sum = 0.0;
        int i = 0;
        double[] results = new double[4]; // 0 -> mean, 1 -> std
        Iterator it = input.iterator();
        // Compute mean
        while (it.hasNext()) {
            i++;
            sum += (double) it.next();
        }
        results[0] = i != 0 ? sum / (double) i : 0;
        it = input.iterator();
        sum = 0.0;
        while (it.hasNext()) {
            sum += Math.pow((double) it.next() - results[0], 2);
        }
        results[1] = i != 0 ? Math.sqrt(sum / (double) i) : 0;
        return results;
    }

    private static void initializeOutput() {
        try {
            File f = new File(outputFile);
            f.delete();
        } catch (Exception e) {
            printAndExit(
                    String.format("*** Error trying to delete the output file %s... (%s)", outputFile, e.getMessage()));
        }
    }

    private static void store(String dataset, double[] accuracy, double[] timeSpent, double[] complexity)
            throws Exception {
        // Append output
        FileWriter file = new FileWriter(outputFile, true);
        PrintWriter linePrint = new PrintWriter(file);
        linePrint.printf("%s; %f; %f; %f; %f; %f; %f%n", dataset, accuracy[0], accuracy[1], timeSpent[0], timeSpent[1],
                complexity[0], complexity[1]);
        linePrint.close();
    }

    public static void main(String[] args) throws Exception {
        Instances instances = null;
        Scanner sc;
        File file = null;
        Random gen;
        double[] accuracyStat = new double[2];
        double[] timeStat = new double[2];
        double[] complexityStat = new double[2];
        int[] seeds = { 57, 31, 1714, 17, 23, 79, 83, 97, 7, 1 };
        String dataset;
        ArrayList<Double> timeSpent;
        ArrayList<Double> accuracy;
        ArrayList<Double> complexity;

        try {
            file = new File(datasetsFile);
        } catch (Exception e) {
            printAndExit(String.format("*** Error trying to read datasets file... (%s)",
                    e.getMessage()));
        }
        System.out.println(StringUtils.center(String.format("%d fold cross validation stratified", numFolds), 64));
        initializeOutput();
        sc = new Scanner(file);
        System.out.println(String.format(" #  %-30s %-29s", "Dataset", "Seeds"));
        System.out.println("--- " + "-".repeat(30) + " " + "-".repeat(29));
        int number = 0;
        while (sc.hasNextLine()) {
            dataset = sc.nextLine();
            System.out.print(String.format("%3d %-30s ", number++, dataset));
            if (debug && dataset.equals("balloons")) {
                printAndExit("* Check error output. Debug End.");
            }
            timeSpent = new ArrayList<>();
            accuracy = new ArrayList<>();
            complexity = new ArrayList<>();
            for (int seed : seeds) {
                System.out.print(String.format("%d ", seed));
                // Establece la semilla
                gen = new Random(seed);
                // Obtiene los datos
                instances = getInstances(dataset);
                instances.randomize(gen);
                instances.setClassIndex(instances.numAttributes() - 1);
                try {
                    validation(instances, seed, accuracy, timeSpent, complexity);
                } catch (Exception e) {
                    printAndExit(String.format("*** Error training dataset %s... (%s)", dataset,
                            e.getMessage()));
                }

            }
            System.out.println("");
            accuracyStat = meanAndDeviation(accuracy);
            timeStat = meanAndDeviation(timeSpent);
            complexityStat = meanAndDeviation(complexity);
            try {
                store(dataset, accuracyStat, timeStat, complexityStat);
            } catch (Exception e) {
                printAndExit(String.format("*** Error storing results of dataset %s... (%s)",
                        dataset, e.getMessage()));
            }
        }
        sc.close();
        printAndExit("* End.");
    }
}