/*
 * CommonWordFinder.java
 * Created on Feb 24, 2005
 */
package word;

import java.util.LinkedList;
import java.util.List;
import java.util.ListIterator;

/**
 * Finds the most common words in a given text.
 */
public final class CommonWordFinder {
    
    private final WordBagFactory bagFactory;

    /**
     * Construct a common word finder with the specified delimeters.
     */
    public CommonWordFinder(WordBagFactory bagFactory) {
        this.bagFactory = bagFactory;
    }
    
    public List/*<StringCount>*/ findCommonWords(String text, int num) {
        CommonList commonList = new CommonList(num);
        WordBag bag = bagFactory.makeWordBag();
        char[] textChars = text.toCharArray();
        StringBuilder sb = new StringBuilder();
        
        for(int i = 0; i < textChars.length; i++) {
            if (Character.isLetter(textChars[i])) {
                sb.append(textChars[i]);
            } else if (sb.length() > 0) {
                commonList.add(bag.add(sb.toString()));
                sb.setLength(0);
            }
        }
        
        return commonList.common;
    }
    
    private static class CommonList {
        
        private int threshold = 0;
        private final int num;
        private final LinkedList/*<StringCount>*/ common;
        
        private CommonList(int num) {
            this.num = num;
            this.common = new LinkedList();
        }
        
        private void add(WordCount strCount) {
            if (strCount.count() > threshold) {
                ListIterator/*<StringCount>*/ i = common.listIterator();
                
                while(i.hasNext()) {
                    WordCount sc = (WordCount)i.next();
                    int diff = strCount.count() - sc.count();
                    
                    // if less than, add it right before
                    if (diff < 0) {
                        i.previous();
                        break;
                    }
                    // otherwise, if encounter the same string, remove it
                    else if (diff == 0 &&
                             strCount.word().equalsIgnoreCase(sc.word())) {
                    	i.remove();
                    }
                }
                
                // insert it and possibly set the new threshold
                i.add(strCount);
                if (common.size() == num + 1) {
                    common.removeFirst();
                    threshold = ((WordCount)common.getFirst()).count();
                }
            }
        }
        
    }

}