/*
 * Kyle Pierce
 * CSE 143
 *
 * Client program making use of Lists, Sets, and Maps to glean
 * certain word-related information from "moby dick" (or "the whale").
 */
import java.io.*;
import java.util.*;

public class WordCount {
   public static void main(String[] args) throws FileNotFoundException {

      /*
       * WARNING: This produces a lot of output.  You probably want
       * to comment out certain parts to run them in isolation so you
       * can actually see the results more clearly.
       */

      File f = new File("moby.txt");
      Scanner input = new Scanner(f);

      long start = System.currentTimeMillis();
      int count1 = countUniquesWithList(input);
      long stop = System.currentTimeMillis();

      System.out.println("time with a list = " + ((stop - start) / 1000.0) + "s");
      System.out.println("number of words  = " + count1);
      System.out.println();

      input = new Scanner(f);  // effectively reset the Scanner

      start = System.currentTimeMillis();
      int count2 = countUniquesWithSet(input);
      stop = System.currentTimeMillis();

      System.out.println("time with a set  = " + ((stop - start) / 1000.0) + "s");
      System.out.println("number of words  = " + count2);
      System.out.println();

      input = new Scanner(f);  // effectively reset the Scanner

      printCounts(input);
   }

   // post: returns the number of unique words in the given Scanner
   public static int countUniquesWithList(Scanner input) {
      List<String> words = new ArrayList<>();
      while (input.hasNext()) {
         String word = input.next();

         if (!words.contains(word)) {
            words.add(word);
         }
      }

      return words.size();
   }

   // post: returns the number of unique even-length words in the given
   //       Scanner and prints them all out
   public static int countUniquesWithSet(Scanner input) {
      Set<String> words = new HashSet<>();
      while (input.hasNext()) {
         String word = input.next();
         words.add(word);
      }

      // remove all the words with odd length
      // cannot use a for-each loop here, must be iterator
      Iterator<String> itr = words.iterator();
      while (itr.hasNext()) {
         String word = itr.next();
         if (word.length() % 2 == 1) {
            // careful not to say words.remove(word) or we
            // would still get an exception
            itr.remove();
         }
      }

      // print out all of the words, one per line
      for (String word : words) {
         System.out.println(word);
      }

      return words.size();
   }

   // post: prints each word in the given Scanner followed by the number
   //       of occurrences of that word
   public static void printCounts(Scanner input) {
      Map<String, Integer> words = new TreeMap<>();
      while (input.hasNext()) {
         String word = input.next();

         // either add the word into the map for the first time
         // or update the existing count if already present
         if (!words.containsKey(word)) {
            words.put(word, 1);
         } else {
            int oldCount = words.get(word);
            words.put(word, oldCount + 1);
         }
      }

      // print out all the words and their # of occurrences
      for (String word : words.keySet()) {
         System.out.println(word + ": " + words.get(word) + " occurrences");
      }
   }
}