/********************************************************************
*          World Wide Web Page Download Using Lynx in Java          *
*===================================================================*
*                       Name:   Wen-Chen Hu                         *
*                       Dated:  October 1, 1999                     *
*********************************************************************

*********************************************************************
*                The Purpose of this Project                        *
*********************************************************************

   Demonstrate how to use Java language to download Web pages.

*********************************************************************
*                   How to Use This Program robot                   *
*********************************************************************

   > javac lynxj.java
   > java lynxj url number
     1. lynxj : this class name
     2. url   : a seed URL
     3. number: number of Web pages downloaded

   For example,
     > java lynxj http://www.eng.auburn.edu/~wenchen/ 50

*********************************************************************
********************************************************************/

import java.io.*;
import java.lang.*;


/*******************************************************************
*                          Main Program                            *
*******************************************************************/

public class lynxj  {
   public static void main (String args[])  {
   
      // args[0]: a seed URL
      // args[1]: number of Web pages downloaded

      String        url = args[0];
      downLoadHTML  t   = new downLoadHTML();
 
      for (int i = 0;  i < Integer.valueOf(args[1]).intValue( ); i++)  {
	 t.loading (url, "work");      // work is a temporary workarea

         // start text processing on the work,
         // enter the extracted information into database, and 
         // assign the url with a new URL .
      }
   }
}
 

class downLoadHTML  { 
   void loading (String url, String work)  {

      //Open the URL for reading
      try  {
         Runtime run = Runtime.getRuntime( );
         try  {
            OutputStream  to_file = new FileOutputStream(work);
            String    command = "lynx -dump " + url;
	    DataInputStream  theHTML = new DataInputStream (run.exec(command).getInputStream());
            byte[]  buffer = new byte[8096];
            int     bytes_read;
            while ((bytes_read = theHTML.read(buffer)) != -1)
               to_file.write (buffer, 0, bytes_read);
            to_file.close( );
         } // end try
         catch (Exception e)  {
            System.err.println(e);
         }
      } // end try
      catch (Exception e)  {
         System.err.println(e);
      }
   }
}