/* Use, modification, and distribution are subject to the Boost Software 
License, Version 1.0. (See accompanying file LICENSE_1.0.txt or copy at 
www.boost.org/LICENSE_1.0.txt) */

// Page Class
// This class stores information about a specific webpage.
// It contains a vector of Word objects and a vector of links.

import java.util.*;

/**
 * This class stores information about a webpage. It uses HashMap objects to
 * store the {@link Link} and {@link Word} objects that occur on the page. 
 * @author Aaron Miller, Mike Ottum
 * @version 1.7
 */

public class Page
{
    protected HashMap linkList;
    protected HashMap wordList;
    protected String pageURL;
    protected boolean index;
    protected boolean follow;
    protected long checkSum;
    protected String header;
    protected String text;


    public void addText(String newText) {
	text = text + newText;
	if (text.charAt(text.length()-1) != ' ') {
	    text = text.concat(" ");
	}
    }

    public String getText() {
	return text;
    }

    public void addHeader(String newHeader) {
	header = header + newHeader;
	if (header.charAt(header.length()-1) != ' ') {
	    header = header.concat(" ");
	}	    
    }

    public String getHeader() {
	return header;
    }
    
    /**
     * Initializes the url string associated with the page and creates the
     * HashMaps.
     * @param url The absolute url of the webpage.
     */
    public Page(String url)
    {
	pageURL = new String(url);
	linkList = new HashMap();
	wordList = new HashMap();
	index = true;
	follow = true;
	header = "";
	text = "";
    }
    
    /**
     * Sets the noIndex boolean value.  If noIndex is true, then don't index
     * this page.  If it is false, then do.
     * @param set Value to set noIndex equal to.
     */
    public void setNoIndex(boolean set)
    {
	index = !set;
    }


    /**
    * Sets the noFollow boolean value.  If noFollow is true, then do not
    * crawl the links on this page.  If it is false, then do.
    * @param set Value to set noFollow equal to.
    */
		     
    public void setNoFollow(boolean set)
    {
	follow = !set;
    }

    /**
     * Tests a Page object for equality with the current Page. The test uses
     * the pageURL member variable as the equality test. Thus, if two pages
     * have the same URL, they are equal.
     * @param p The Page object to compare to the current Page.
     * @return True if the Pages are equal. False otherwise.
     */
    public boolean equals(Page p)
    {
	return pageURL.equals(p.getURL());
    }

    /**
     * Returns the current Page objects URL.
     * @return The pageURL member variable.
     */
    public String getURL()
    {
	return pageURL;
    }
    
    /**
     * Adds a specific word and location to the hash of words on the page.
     * @param s The word string to be added.
     * @param location The location of the word on the page.
     * @param tags Tags associated with the string at the given location.
     */
    public void addWord(String s, int location, String tags)
    {
	s = s.toLowerCase();
	Word w;
	if(!wordList.containsKey(s))
	    {
		w = new Word(s, location);
		wordList.put(s, w);
	    }
	else
	    {
		// put the information into the word object that already
		// exists
		w = (Word)(wordList.get(s));
	    }
	w.addTags(tags, location);
    }
    
    /**
     * Adds a specific link and its position on the page to the link hash.
     * @param url The absolute url of the link.
     * @param pos The position of the link on the page.
     */
    public void addLink(String url, int pos)
    {
	Link l;
	if(!linkList.containsKey(url))
	    {
		l = new Link(url, getURL(), pos);
		linkList.put(url,l);
	    }
	else
	    {
		l = (Link)(linkList.get(url));
		l.addPosition(pos);
	    }
    }
    
    /**
     * Returns an {@link Iterator} to the {@link Set} of URLs contained
     * on the page.
     */
    public Iterator getLinks()
	// returns an iterator to a Set of Strings of links.
    {
	Iterator it = linkList.keySet().iterator();
	return it;
    }
    
    /**
     * Returns an {@link Iterator} to the {@link Set} of words contained
     * on the page.
     */
    public Iterator getWords()
	// returns an iterator to a Set of Strings of words.
    {
	Iterator it = wordList.keySet().iterator();
	return it;
    }
	
    public Word keyToWord(String wordkey)
	//returns the word object from the HashMap
    {
	return (Word) wordList.get( wordkey );
    }
	
    public Link keyToLink(String linkey)
	//returns the link object from the hashmap
    {
	return (Link) linkList.get( linkey );
    }

    public void setCheckSum(long check)
    {
	checkSum = check;
    }

}