Networking in Java

This chapter is from the book

Just Java 2, 5th Edition

Learn More Buy

This chapter is from the book

This chapter is from the book 

Just Java 2, 5th Edition

Learn More Buy

HTTP and Web Browsing: Retrieving HTTP Pages

Here is an example of interacting with an HTTP server to retrieve a web page from a system on the network. This shows how easy it is to post information to HTML forms. Forms are covered in more depth in the chapter on servlets, and you may want to refresh your memory on that section.

For the impatient, HTML forms allow you to type some information in your browser which is sent back to the server for processing. The information may be encoded as part of the URL, or sent separately in name/value pairs.

The Yahoo site is a wide-ranging access portal. They offer online stock quotes that you can read in your browser. I happen to know (by looking at the URL field of my browser) that a request for a stock quote for ABCD is translated to a socket connection of:

http://finance.yahoo.com/q?s=abcd

That's equivalent to opening a socket on port 80 of finance.yahoo.com and sending a "get /q?s=abcd." You can make that self same request yourself, in either of two ways. You can open a socket connection to port 80, the http port. Or you can open a URL connection which offers a simpler, higher-level interface. We'll show both of these here. Here's the stock finder done with sockets:

import java.io.*;
import java.net.*;
public class Stock {

  public static void main(String a[]) throws Exception {
    if (a.length!=1) {
      System.out.println("usage: java Stock <symbol> ");
      System.exit(0);
    }

    String yahoo = "finance.yahoo.com";
    final int httpd = 80;
    Socket sock = new Socket(yahoo, httpd);

    BufferedWriter out =
        new BufferedWriter( new OutputStreamWriter(
 							sock.getOutputStream() ) );

    String cmd = "GET /q?" +"s=" +a[0] +"\n";
    out.write(cmd);
    out.flush();

	 BufferedReader in =new BufferedReader( 
		new InputStreamReader( sock.getInputStream() ) );
	 String s=null;
    int i, j;
	 // pick out the stock price from the pile of HTML
	 // it's in bold, so get the number following "<b>"
    while ( (s=in.readLine()) != null) {
       if (s.length()<25) continue;
       if ((i=s.indexOf(a[0].toUpperCase())) < 0) continue;
       s=s.substring(i);
       if ((i=s.indexOf("<b>")) < 0) continue;
       j = s.indexOf("</b>");
       s=s.substring(i+3,j);
       System.out.println(a[0] +" is at "+s);
       break;
    }
  }
}

The Yahoo page that returns stock quotes contains thousands of characters of hrefs to ads and formatting information. It consists of many HTML lines like this:

<a href="/q?s=SUNW&d=t">SUNW</a></td><td nowrap
 align=center>12:03PM</td><td nowrap><b>9.55</b></a>

Luckily it's fairly easy to pull out the stock price. From inspecting the output, it's on a line with more than 25 chars. The line contains the stock symbol rewritten in upper case. The number we want is bracketed by <b> ... </b>, which is the HTML to print the number in bold face.

Given all that, running the program provides this output:

java Stock ibm
ibm is at 101.26

It was a lot more fun running this program in the year 2000. Here is the same program, rewritten to use the classes URL and URLConnection. Obviously, URL represents a URL, and URLConnection represents a socket connection to that URL. The code to do the same work as before, but using URLConnection is:

import java.io.*;
import java.net.*;
public class Stock2 {

  public static void main(String a[]) throws Exception {
    if (a.length!=1) {
      System.out.println("usage: java Stock <symbol> ");
      System.exit(0);
    }

    String yahoo = "http://finance.yahoo.com/q";

    URL url = new URL(yahoo);
    URLConnection conn = url.openConnection();
    conn.setDoOutput(true);
    PrintWriter pw = new PrintWriter( conn.getOutputStream());
    pw.print("s=" + a[0]);
    pw.close();

    BufferedReader in = new BufferedReader( new InputStreamReader(
                         conn.getInputStream()));
    String s=null, stock=a[0].toUpperCase();
    int i=0,j=0;
    while ( (s=in.readLine()) != null) {
	   if (s.length()<25) continue;
       if ((i=s.indexOf(a[0].toUpperCase())) < 0) continue;
       s=s.substring(i);
       if ((i=s.indexOf("<b>")) < 0) continue;
       j = s.indexOf("</b>");
       s=s.substring(i+3,j);
       System.out.println(a[0] +" is at "+s);
       break;
    }
  }
}

The main difference here is that we form a URL for the site and file (script) that we want to reference. Then we open a connection to the URL, tell it that we are going to do output to it, and write the "name=value" parameter. We finish up as before, reading what the socket writes back and extracting the characters of interest. Clearly, this program will stop working when Yahoo changes the format of the page, but it demonstrates how we can use a URL and URLConnection for a slightly higher-level interface than a socket connection. We could even go one step further and use the class HttpURLConnection which is a subclass of URLConnection. Please look at the HTML documentation for information on these classes.

A URL can pose a security risk, since you can pass along information even by reading a URL. Requesting http://www.cia.gov/cgi-bin/cgi.exe/secretinfo passes "secretinfo" along to the CGI script, for example. Since requesting a URL can send out information just as a Socket can, requesting a URL has the same security model as access to Sockets. Namely, an applet can only open a socket connection back to the server from which the applet came. Security is also the reason that an applet may not open a socket connection to any other system except its server. Otherwise, it could look at information on its subnet behind the firewall, and send it back to crackers everywhere.

If you are behind a firewall (and who isn't these days?) you will need to tell Java the details of your proxy server and port in order to access hosts outside the firewall. You do this by defining properties, perhaps when starting the code:

java -DproxySet=true -DproxyHost=SOMEHOST -DproxyPort=SOMENUM code.java

Without this, you'll get an UnknownHostException. The proxy settings are needed for both java.net.URLConnection and for java.net.Sockets. At work, your systems administrator will know the values. At home, you won't be using a proxy server unless you set it up yourself.

How to Find the IP Address Given to a Machine Name

The class java.net.InetAddress represents IP addresses and about one dozen common operations on them. The class should have been called IP or IPAddress, but was not (presumably because such a name does not match the coding conventions for classnames). Common operations on IP addresses are things like: turning an IP address into the characters that represent the corresponding domain name, turning a host name into an IP address, determining if a given address belongs to the system you are currently executing on, and so on.

InetAddress has two subclasses:

Your programs will not use these classes directly very much, as you can create sockets using domain and host names. Further, in most of the places where a hostname is expected (such as in a URL), a String that contains an IP address will work equally well. However, if native code passes you an IP address, these classes give you a way to work on it.

The InetAddress class does not have any public constructors. Applications should use the methods getLocalHost(), getByName(), or getAllByName() to create a new InetAddress instance. The program that follows show examples of each of these.

This code will be able to find the IP address of all computers it knows about. That may mean all systems that have an entry in the local hosts table, or (if it is served by a name server) the domain of the name server, which could be as extensive as a large subnet or the entire organization.

import java.io.*;
import java.net.*;
public class addr {

  public static void main(String a[]) throws Exception {

    InetAddress me = InetAddress.getByName("localhost");
    PrintStream o = System.out;   
    o.println("localhost by name =" + me );

    InetAddress me2 = InetAddress.getLocalHost();
    o.println("localhost by getLocalHost =" + me2 );

    InetAddress[] many = InetAddress.getAllByName("microsoft.com");
    for (int i=0; i<many.length; i++) 
        o.println( many[i] );
  }
}

Run it with:

java addr

localhost by name =localhost/127.0.0.1
localhost by getLocalHost =zap/10.0.10.175
Microsoft: microsoft.com/207.46.230.218
Microsoft: microsoft.com/207.46.230.219
Microsoft: microsoft.com/207.46.197.100
Microsoft: microsoft.com/207.46.197.101
Microsoft: microsoft.com/207.46.197.102

The getAllByName() method reports all the IP addresses associated with a domain name. You can see from the output above that Microsoft.com, like most big sites, is served by multiple IP addresses, on two different subnets (probably for fault tolerance). Each of those five IP addresses probably represents load balancer hardware fanning out to dozens of server nodes.

Some Notes on Protocol and Content Handlers

Some of the Java documentation makes a big production about support for extending the MIME types known to browsers. If it is asked to browse some data whose type it doesn't recognize, it can simply download the code for the appropriate handler based on the name of the datetype and use that to grok the data. This is exactly what happens with plug-ins. If you stumble across a RealAudio file, the browser prompts you to download the plug-in that can play it. Java can make this completely automatic. Or so the theory runs. It hasn't yet been used much in practice.

The theory of the handlers is this. There are two kinds of handler that you can write: protocol handlers and content handlers.

A protocol handler talks to other programs. Both ends follow the same protocol in order to communicate structured data between themselves ("After you," "No, I insist—after you.") If you wrote an Oracle database protocol handler, it would deal with SQL queries to pull data out of an Oracle database.

A content handler knows how to decode the contents of a file. It handles data (think of it as the contents of something pointed to by a URL). It gets the bytes and assembles them into an object. If you wrote an MPEG content handler, it would then be able to play MPEG movies in your browser, once you had brought the data over there. Bringing MPEG data to your browser could be done using the FTP protocol, or you might wish to write your own high performance protocol handler.

Content handlers and protocol handlers may be particularly convenient for web browsers, and they may also be useful in stand-alone applications. There is not a lot of practical experience with these handlers in Java yet, so it is hard to offer definitive advice about their use. Some people predict they are going to be very important for decoding file formats of arbitrary kinds, while other people are ready to be convinced by an existence proof.