/*
 * WebWorm Solution.   CSE 341.
 * 
 * Jason Hartline
 */

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


/*
 * this class reads in web pages and will, when printed, display
 * the links in the web page.
 */
class WebWorm {

   /*
    * this holds the links that we have found thus far.
    */
   Vector links;
   
   /* 
    * this is the URL that we are going to look for links on
    */
   String url;
   
   /* 
    * sequences of characters denotes a link.
    */
   static final String linkPrefix = "<a";
   static final String linkSuffix = ">";
   
   /*
    * This contstructor makes a WebWorm specifing the URL to open.
    * It does not actually try to open the URL or verify that it is
    * valid.
    */
   public WebWorm(String urlIn)
   {
      url = urlIn;
      links = null;
   }
   
   /*
    * this function reads in the URL
    * and processes the URL filling up
    * 'links' with all the links found.
    */
   public boolean process()
   {
      
      /* 
       * prepare to process
       */
      clearError();
      links = null;
      
      
      /* 
       * try to open our connection to the web server
       * with the URL.
       */
      DataInputStream in = null;
      
      try 
      {
         URL address = new URL(url);
         in = new DataInputStream(address.openStream());
      }
      catch (IOException e)
      {
         setError("Could not open URL connection.");
         return false;
      }
      
      
      
      /*
       * initialize our list of hyperlinks.
       */
      links = new Vector(); 
      String line = null;
      
      while (true) // break when line = null.
      {
         /*
          * try to read in a line.
          */
         try
         {
            line = in.readLine();
         }
         catch (IOException e)
         {
            links = null;
            setError("Error reading from URL connection.");

            /*
             * close the file.
             */
            try
            {
               in.close();
            }
            catch (IOException e2)
            {
               // do nothing.  We are already dealing with an error.
            }
            
            return false;
         }

         /*
          * we are done if line == null.
          */
         if (line == null)
            break;
      
         /*
          * process the line.
          */
         processLine(line);      
      } 

          
      /*
       * close the URL connection.
       */
      try
      {
         in.close();
      }
      catch (IOException e)
      {
         links = null;
         setError("Could not close URL connection.");
         return false;
      }
      
      /*
       * success!!!
       */
      return true;
   }
   
   /*
    * This function is a helper function
    * that takes a line and adds any URLs 
    * found in that line, to 'links'.
    */
   private void processLine(String s)
   {
      /*
       * use a lowercase version of the string to be
       * case insensitive.
       */
      String lower = s.toLowerCase();
      int index=0;
      
      /*
       * for each link...
       */ 
      for (index=lower.indexOf(linkPrefix,0); 
           index >= 0; 
           index = lower.indexOf(linkPrefix,index+1))
      {
         /*
          * find the end of the link.
          */
         int endIndex = lower.indexOf(linkSuffix, index);
         
         if (endIndex < 0)
            endIndex = s.length();
         
         /*
          * add the link to 'links'
          */
         links.addElement(s.substring(index+2,endIndex));
         
         /*
          * prepare to find the next link.
          * on this line.
          */
         index = endIndex;
      }
   }
   
   /*
    * printing out a WebWorm will either.
    *    1. if the URL has not been processed, print out 
    *       that the WebWorm will process the link.
    *    2. if there was an error processing the URL, 
    *       print out the error message.
    *    3. if the URL was successfully processed, print
    *       out all of the links.
    */
   public String toString()
   {
      /*
       * oops! there was an error.
       */
      if (hasError())
         return("WebWorm error: " + getErrorMessage());
         
      /*
       * URL has not been processed yet.
       */
      if (links == null)
         return("WebWorm to load: " + url);
      
      /*
       * print out header.
       */
      StringBuffer out = new StringBuffer("Links found in " + url);

      /*
       * print out links.
       */
      for(Enumeration e = links.elements(); e.hasMoreElements(); )
         out.append("\n"+e.nextElement());
      
      return(out.toString());
   }
 
   /*
    * main program.
    * 
    * - create a WebWorm.
    * - have it process the URL.
    * - print it out.
    */
   public static void main(String [] args)
   {
      if (args.length != 1)
      {
         System.err.println("USAGE: WebWorm url");
         System.exit(-1);
      }
      
      WebWorm worm = new WebWorm(args[0]);
      
      worm.process();
      
      System.out.println(worm);
   }


   /*
    * dealing with errors.
    * 
    * setError, clearError, hasError, and getError
    */
   
   private String errorMessage = null;
   
   private void setError(String error)
   {
      errorMessage = error;
   }
   
   private void clearError()
   {
      errorMessage = null;
   }
   
   public boolean hasError()
   {
      return (errorMessage != null);
   }
   
   public String getErrorMessage()
   {
      return errorMessage;
   }
   
}