import java.util.*;
import java.io.*;

public class Vocabulary4 {
    public static void main(String[] q) throws FileNotFoundException {
        Scanner console = new Scanner(System.in);
        giveIntro();

        System.out.print("file #1 name? ");
        Scanner input1 = new Scanner(new File(console.nextLine())); // moby.txt
        System.out.print("file #2 name? ");
        Scanner input2 = new Scanner(new File(console.nextLine())); // hamlet.txt
        System.out.println();

        double time1 = System.currentTimeMillis();
        TreeSet<String> set1 = getWords(input1);
        TreeSet<String> set2 = getWords(input2);
        double time2 = System.currentTimeMillis();
        System.out.println("File read time=" + (time2 - time1) / 1000 + " seconds");

        time1 = System.currentTimeMillis();
        TreeSet<String> overlap = getOverlap(set1, set2);
        reportResults(set1, set2, overlap);
        time2 = System.currentTimeMillis();
        System.out.println("Process time=" + (time2 - time1) + " milliseconds");

        console.close();
    }

    // TreeSet automatically sorts the list and does not store duplicate objects
    public static TreeSet<String> getWords(Scanner input) {
        // ignore all but alphabetic characters and apostrophes
        input.useDelimiter("[^a-zA-Z']");
        // read all words and sort
        TreeSet<String> words = new TreeSet<String>();
        while (input.hasNext()) {
            String next = input.next().toLowerCase();
            words.add(next);
        }
        return words;
    }

    // Define a new TreeSet as set1 and retain only the elements it shares with set2
    public static TreeSet<String> getOverlap(TreeSet<String> set1, TreeSet<String> set2) {
        TreeSet<String> overlap = new TreeSet<String>(set1);
        overlap.retainAll(set2);
        return overlap;
    }

    // Explanation of the program via the Console
    public static void giveIntro() {
        System.out.println("This program compares the vocabulary of two");
        System.out.println("text files, reporting the number of words");
        System.out.println("in common and the percent of overlap.\n");
    }

    // Reports the percentage of overlap between the files
    public static void reportResults(TreeSet<String> set1, TreeSet<String> set2, TreeSet<String> overlap) {
        System.out.println("file #1 words = " + set1.size());
        System.out.println("file #2 words = " + set2.size());
        System.out.println("common words  = " + overlap.size());

        double percent1 = 100.0 * overlap.size() / set1.size();
        double percent2 = 100.0 * overlap.size() / set2.size();
        System.out.println("% of file 1 in overlap = " + percent1);
        System.out.println("% of file 2 in overlap = " + percent2);
    }
} // End Vocabulary4