import java.util.Vector;
import java.util.Date;
import java.net.*;
import java.io.*;

/** An <CODE>AltaVistaFacade</CODE> defines an Alta Vista search engine.  This engine 
* submits queries to the AltaVista site and parses the results into a 
* <CODE>SearchResults</CODE> Collection
*   
* @version 1.0
* @author Deric Bertrand - October 1999
*/ 
public class AltaVistaFacade extends SearchEngineFacade
{
	/** S_AltaVistaURLString  Defines the AltaVista URL */
	private static final String S_AltaVistaURLString = "http://www.altavista.com/cgi-bin/query";
	
	/** 
	* Submits a query request to the <CODE>AdditionalResults.ASP</CODE> and unmarshalls the 
	* searialized results into a <CODE>SearchResult</CODE> collection
	* @param queryText  The query string  sent to the AltaVista search engine for processing
	* @param listingIndex  Starting index of the query
	* @param theCollection  The <CODE>Vector</CODE> of <CODE>SearchResult's</CODE>
	* @return  The number of results obtained from the current search request
	*/
	public int SubmitQuery(String queryText, int listingIndex, Vector theCollection)
		throws java.net.MalformedURLException, java.io.IOException, Exception
	{
		int resultCount, startidx;
		HttpURLConnection theConnection;
		BufferedReader in;
		SearchResult searchResult;
		URL theURL;
		String	resultLine;
		StringBuffer resultString = new StringBuffer();;

		// Connect to Alta Vista and retrieve the HTML page of results 
		theURL = new URL( this.buildURLString( queryText, listingIndex));
		theConnection = (HttpURLConnection) theURL.openConnection();
		
		// added exception handling - 11/11/99 - Deric Bertrand
		if (( theConnection.getResponseCode() == -1) ||(theConnection.getResponseCode() != HttpURLConnection.HTTP_OK))
			throw new IOException("Bad response from URLconnect");
					
		in = new BufferedReader(new InputStreamReader(theConnection.getInputStream()));
		while ((resultLine = in.readLine()) != null)
			resultString.append(resultLine);

		// Find where the result set starts
		startidx	= 0;	
		resultCount = 0;
		while ((startidx = resultString.toString().indexOf("<dl><dt><b>", startidx)) != -1) 
		{			
			startidx += 11;		// Move past the <dl><dt><b> token
			searchResult = parseHTML(resultString.toString(), startidx);
			// Are we looping - 11/09/99 - Deric Bertrand	
			if ( searchResult.getRank() < listingIndex)
				throw new Exception(
					"Invalid Search Result retrieved for listingIndex"
				);
			theCollection.addElement( searchResult);
			resultCount++;
		}
		
		// Close the connection - 11/11/99 - Deric Bertrand
		in.close();	
		return resultCount;
	}
	
	
	/** 
	* Formats a query to send to AltaVista (URL and the CGI Query String)
	* @param queryText  The query string  sent to the AltaVista search engine for processing
	* @param listingIndex  Starting index of the query
	* @return  The formatted AltaVista URL and Query String
	*/
	private String buildURLString( String queryText, int listingIndex)
	{
		StringBuffer queryString = new StringBuffer();
		
		queryString.append(S_AltaVistaURLString + "?");	// AltaVista URL
		queryString.append("pg=aq");				// Advanced Query
		queryString.append("&kl=XX");				// Language Type	(CODE VALUE)
		queryString.append("&q=" + queryText);		// Query Text 
		queryString.append("&r=");				// Sort By
		queryString.append("&d0=");				// Start Date (DATE FORMAT:  DD%2FMM%2FYY)
		queryString.append("&d1=");				// End Date
			
		// Is there a specific starting Item	
		if ( listingIndex > 0)
			queryString.append("&stq=" + listingIndex + "&c9k");
		
		return(queryString.toString());
	}


	/** 
	* Parses the AltaVista Search Results (HTML page) into <CODE>SearchResult</CODE> instances 
	* @param htmlPage  The HTML page of search results returned from the AltaVista query
	* @param startidx  Starting index of parse
	* @return  A <CODE>SearchResult</CODE> instance
	* deprecated Date(string)
	*/
	private SearchResult parseHTML( String htmlPage, int startidx)
		throws MalformedURLException
	{
		SearchResult searchResult = null;
		String urlString = null;
	 	String title = null;
	 	String description = null;
	 	Date lastModified = null;
	 	int rank = 0;
	 	int estimatedSize = -1;
		int endidx = 0;
	
		// Rank 
		endidx = htmlPage.indexOf(".", startidx);
		rank = (Integer.parseInt(htmlPage.substring(startidx, endidx)));
		startidx = endidx + 1;

		// URL
		endidx = htmlPage.indexOf("<a href=", startidx);
		startidx = endidx + 9;
		endidx = htmlPage.indexOf("><b>", startidx);
		urlString = htmlPage.substring(startidx, endidx - 1);
		startidx = endidx + 4;
	
		// Title
		endidx = htmlPage.indexOf("</b></a>", startidx);
		title = htmlPage.substring(startidx, endidx);
		startidx = endidx + 8;

		// Description
		endidx = htmlPage.indexOf("<dd>", startidx);
		startidx = endidx + 4;
		endidx = htmlPage.indexOf("</dd>", startidx);
		description = htmlPage.substring(startidx, endidx);
		startidx = endidx + 5;
		
		// LastModified
		endidx = htmlPage.indexOf("Last modified on:", startidx);
		startidx = endidx + 17;
		endidx = htmlPage.indexOf(" - ", startidx);
		lastModified = new Date( htmlPage.substring(startidx, endidx));
		startidx = endidx + 3;
			
		// Find the PageSize 
		endidx = htmlPage.indexOf("K bytes", startidx);
		estimatedSize = (Integer.parseInt(htmlPage.substring(startidx, endidx)) * 1024);
			
		// Build the search result instance	
		searchResult = new SearchResult(urlString, title, rank, description, lastModified, estimatedSize);
	
		return(searchResult);
	}
}