// CSE 143, Winter 2009, Marty Stepp
//
// This program counts the occurrences of words in a large file and lets the
// user know which words appeared 1000 times or more.
// It demonstrates the use of a Map collection.

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

public class WordCountWithMap {
    public static void main(String[] args) throws FileNotFoundException {
        // timer to see how long the program takes to run
        long startTime = System.currentTimeMillis();
        
        // read file into a collection
        // word --> # of occurrences
        Map<String, Integer> wordCount = new HashMap<String, Integer>();
        
        System.out.println("Reading file...");
        Scanner input = new Scanner(new File("mobydick.txt"));
        while (input.hasNext()) {
            String word = input.next();
            if (wordCount.containsKey(word)) {
                // seen this word before; increase count by 1
                int count = wordCount.get(word);
                count++;
                wordCount.put(word, count);
            } else {
                // never seen this word before
                wordCount.put(word, 1);
            }
        }

        System.out.println(wordCount.size() + " unique words total.");
        
        System.out.println("Words that occur at least 1000 times:");
        for (String word : wordCount.keySet()) {
            int count = wordCount.get(word);
            if (count >= 1000) {
                System.out.println(word + " occurs " + count + " times.");
            }
        }
                
        long endTime = System.currentTimeMillis();
        System.out.println((endTime - startTime) + " ms elapsed.");
    }
}