// AltavistaReader.java
//
import java.io.*;
import java.net.*;
import java.util.*;
import gnu.regexp.*;


/**
 * AltavistaReader parses a given stream until it is empty,
 * and returns an array of Hits.
 *
 * @author Andrew Walenstein
 */
public class AltavistaReader
{
  static final int MAX_HITS_PER_PAGE = 10;

  /**
   * Empties the input, parsing the HTML return result contained therein,
   * and returns the salient contents.
   *
   * @param inStream an open stream of Altavista "text mode" results
   * @returns AltavistaPageContents information contained in page
   * @throws IOException if something goes wrong reading the HTML
   */
  private static
  Vector parsePage ( Reader inStream ) throws IOException
  {
    RE lineStartRE   = null;
    RE contentLineRE = null;
    RE statsLineRE   = null;

    try
      {
	lineStartRE = new RE
	  ( "^<dl><dt><b>([0-9]+). *</b><a href=.([^\"]+).><b>([^<]+)</b></a><dd>$" );
	contentLineRE = new RE
	  ( "^([^<]+)<" );

	statsLineRE = new RE
	  ( "^Last modified ([0-9]*-...-..) - page size ([0-9]+)K - in (\\w+)" );
      }
    catch ( REException e )
      {
	System.out.println ( "Exc: " + e );
      }

    final BufferedReader bRead = new BufferedReader ( inStream );
    final Vector hits = new Vector ( MAX_HITS_PER_PAGE );

    int numHits = 0;
    String line;

    while ( (line = bRead.readLine()) != null )
      {
	final REMatch part1 = lineStartRE.getMatch ( line );

	if ( part1 == null ) continue;
	if ( (line = bRead.readLine()) == null ) break;

	final REMatch part2 = contentLineRE.getMatch ( line );

	if ( part2 == null ) continue;
	if ( (line = bRead.readLine()) == null ) break;

	final REMatch part3 = statsLineRE.getMatch ( line );

	if ( part3 == null ) continue;

	hits.addElement ( new Hit ( part1.toString(1),
				    part1.toString(2),
				    part1.toString(3),
				    part2.toString(1),
				    part3.toString(1),
				    part3.toString(2),
				    part3.toString(3) ) );
      }

    inStream.close();

    return hits;
  }

  /**
   * Given an Altavista query string and a maximum number of hits,
   * this function returns an array of hits returned from Altavista.
   *
   * @param query query string in Altavista format
   * @param maxHits maximum number of hits to return
   * @returns an array of Hits returned by Altavista, null if invalid
   * @throws IOException if querying error
   */
  public static
  Hit[] getQuery ( String query, int maxHits ) throws IOException
  {
    if ( maxHits <= 0 ) return null;

    final String avURL     = "http://altavista.digital.com/cgi-bin/query?";
    final String formQuery = URLEncoder.encode ( query );

    final String queryBase = avURL + "pg=q&q=" + formQuery;
    final String suffix    = "&c9k";

    final Vector[] hitCollect = new Vector[(int)(maxHits/MAX_HITS_PER_PAGE)+1];

    int numPages = 0;
    int totHits  = 0;

    while ( totHits < maxHits )
      {
	try
	  {
	    URL queryURL;

	    queryURL = new URL ( queryBase + "&stq=" + (int)(numPages*10) +
				 suffix );

	    InputStream resultStream = queryURL.openStream();

	    hitCollect[numPages] = parsePage ( new InputStreamReader ( resultStream ) );
	    if ( hitCollect[numPages].size() <= 0 ) break;

	    totHits += hitCollect[numPages].size();
	    numPages++;

	    if ( totHits < MAX_HITS_PER_PAGE ) break;
	  }
	catch ( MalformedURLException e )
	  {
	    System.err.println ( "Bad URL formed: " + e.toString() );
	  }
      }

    Hit[] collect = new Hit[totHits];
    for ( int i = 0, hitNum = 0 ; i < numPages ; ++i )
      {
	for ( Enumeration e = hitCollect[i].elements() ;
	      e.hasMoreElements() ; )
	  {
	    collect[hitNum++] = (Hit)e.nextElement();
	  }
      }

    return collect;
  }
  
  public static void main ( String[] args )
  {
    for ( int i = 0 ; i < args.length ; ++i )
      {
	try
	  {
	    Hit[] result = getQuery ( args[i], 10 );
	    for ( int q = 0 ; q < result.length ; ++q )
	      {
		Hit r = result[q];
		System.out.println ( "\nRecord #" + r.number );
		System.out.println ( "URL:    " + r.theURL );
		if ( r.title.length() < 50 )
		  {
		    System.out.println ( "Title:  " + r.title );
		  }
		else
		  {
		    System.out.println ( "Title:  " + r.title.substring(0,49));
		  }
		if ( r.body.length() < 50 )
		  {
		    System.out.println ( "Body:   " + r.body );
		  }
		else
		  {
		    System.out.println ( "Body:   " + r.body.substring(0,49) );
		  }
		System.out.println ( "Size(K):" + r.sizeK );
		System.out.println ( "Lang:   " + r.language );
	      }
	  }
	catch ( IOException e )
	  {
	    System.err.println ( "IO " + e.toString() );
	  }
      }
  }
}