Building an Instant Messaging Application Using Jabber/XMPP

An adventure with Smack and Wildfire

Jun. 23, 2006 01:15 PM

This article will describe our experiences with developing a Java-based instant messenger application using Jabber/XMPP (Extensible Messaging and Presence Protocol) - a free, open and public protocol and technology for instant messaging. According to the Jabber Software Foundation, "Under the hood, Jabber is a set of streaming XML protocols and technologies that enable any two entities on the Internet to exchange messages, presence, and other structured information in close to real-time."

Google Talk uses the standard Jabber/XMPP protocol for authenticating, presence, and messaging. But using Jabber goes beyond instant messaging to almost real-time server-to-server communication.

This article will describe the Jabber/XMPP protocol for messaging, the Jabber/XMPP client program based on JFace and Eclipse, and the Jabber/XMPP Java server. These will be illustrated through Open Source Smack and the Wildfire Server 2.4.4. It will offer examples of a login-class plug-in for custom authentication, a plug-in and server extension for custom messages, and a server-side extension for database interactions. Each example will contain the XML messages, client-side code, and server-side code.

What's Different About Jabber?
Compared to traditional IM programs, Jabber:

Is an XML-based messaging protocol that is open and extensible
It lets you build custom client IM applications. We'll talk about Smack, a project based on JFace and Eclipse that provides tools for building custom client applications
It lets you extend the functionality of the server to process custom messages by writing server-side plug-ins. We'll describe the plug-in architecture of the Wildfire Server

What Are the Basic Features of Instant Messaging?
Any instant messaging system will have these basic features: It connects to server, registers new users, logs in, gets presence information, exchanges messages, and does custom interactions (VoIP, Web conferencing, etc.). Below each of these IM features are described using XMPP:

Connect to Server - This is done through a XML stream header exchange. The client sends:

The client receives:

Register - The client sends a message to discover the server's requirements for registration:

The server responds with a message that lists the fields required for registration:

<iq type='result' id='reg1'> <query xmlns='jabber:iq:register'>
      <instructions>

Choose a username and password for use with this service.
Please also provide your email address.        </instructions>
     <username/>
     <password/>
     <email/>
   </query></iq>

The client responds by sending the requested information:

<iq type='set' id='reg2'>
   <query xmlns='jabber:iq:register'>
     <username>a3</username>
     <password>password</password>
     <email>a3@sow.com</email>
   </query>
</iq>

On successful registration, the server responds:

Declare Presence and Get Presence Information
To declare presence information, the client sends:

Note the "to" attribute isn't required in the above message because the client knows which server it's talking to.

To get presence information, client sends:

The server responds with presence data about two people in the roster:

Exchange Messages
User a1 sends message to user a2:

<message xmlns="" id="H49LH-12" to="a1@sow" type="chat" from="a2 @sow/1139385809707">
<body>Good Morning, I am testing chat in Jabber Client.</body>
</message>

The elements above form the core of XMPP (Extensible Messaging and Presence Protocol) XML syntax. It also gives you a sense of the flow of XML messages between the client and server to accomplish a task.

Jabber/XMPP Client and Server
Since Jabber/XMPP is an open protocol scores of implementations have been created for the client and server, some Open Source and others commercial. There are literally hundreds of Jabber clients available. (See References.)

In this article we'll use the Smack library (www.jivesoftware.org/smack/) to illustrate client-side functionality. Smack provides an API for implementing the GUI and managing XML data. Managing XML data involves creating XML messages, sending the messages, receiving messages from the server, and processing the incoming messages.

We'll use the Wildfire Server (www.jivesoftware.org/wildfire/) for illustrating creating server-side plug-ins. For basic IM needs, no server-side programming is required. For other needs Wildfire provides a plug-in architecture for extending the server's functionality. We'll look next at a use case to illustrate how custom messages can be created and processed at both the client and server.

Use Case 1: Custom Authentication and Custom Queries
Consider a case in which an IM client and a Web portal are part of a suite of collaboration applications. The Web portal has database-driven MD5-based authentication and the requirement is that IM client use the portal's authentication service. This will allow users to log in to IM and then launch applications on the portal from the IM client without logging into the portal.

Part A: Custom Authentication
The Wildfire Server uses the DefaultAuthProvider class to authenticate users and DefaultUserProvider to get user information. These use the database tables Wildfire provides.

If you want to use a custom authentication method you'll have to change the user and authentication provider in wildfire.xml.

<provider>
<user> <className>org.indent.wildfire.user.SOWUserProvider</className> </user>
<auth><className>org.indent.wildfire.auth.SOWAuthProvider</className> </auth>
</provider>

The simple extension defined in Listing 1 uses custom tables and custom encryption methods. The SOWUserProvider class is for user info and SOWAuthProvider is for authentication.

The two static variables shown define the database authentication strings. And if custom encryption is used then the encryptPassword() method provides the encryption logic.

Part B: Retrieving a Token from the Database
In this use case, after authenticating, the client is going to ask the database for an encrypted token that uniquely identifies the client. This token is later used to launch applications in the Web portal without explicitly logging into the portal. The steps are:
a) The client generates a custom message to request a token
b) The server gets the message and processes the request by getting a token from the database. The server generates a reply message that contains the token and sends it to the requesting client
c) The client gets the message and extracts the token

This process is illustrated in Figure 1.

Client to Server
The client generates a <iq> packet that looks like:

jabber:iq:token is a namespace for our custom query. This message is generated at the client using the following code:

       XMPPConnection connection = session.getConnection();
       ClientID packet = new ClientID();
       connection.sendPacket(packet);

The Class ClientID is defined in the section "Client-side Processing of Reply Packets" below.

The Server-Side Plug-In
The first step is to create a plug-in that will process the custom message with xmlns="jabber:iq:token." The plug-in is called ClientTokenPlugin. It first creates an IQTokenHandler; the handler is then added to the iq router that routes all the incoming iq messages:

public class ClientTokenPlugin implements Plugin {
    public ClientTokenPlugin() {
       IQHandler iqRDHandler = new IQTokenHandler();
       IQRouter iqRouter = XMPPServer.getInstance().getIQRouter();
       iqRouter.addHandler(iqRDHandler);
    }
    public void initializePlugin(PluginManager manager, File pluginDirectory) { }
    public void destroyPlugin() { }
}

For more on how to create a plug-in, see www.jivesoftware.org/builds/wildfire/docs/ latest/documentation/plugin-dev-guide.html.

IQTokenHandler has two key methods:

a) getInfo() which returns the type of queries in the iq messages that the token handler is going to process
b) handleIQ() which processes the incoming message/packet and creates a reply message/packet

public class IQTokenHandler extends IQHandler {
   public IQTokenHandler() { super("Client Token Handler"); }
   public IQHandlerInfo getInfo() {
     return new IQHandlerInfo("query","jabber:iq:token"); }

   public IQ handleIQ(IQ packet) {
   IQ replyPacket = IQ.createResultIQ(packet);
   Element m = replyPacket.setChildElement("query", "jabber:iq:token");
// 2 lines below are specific to our needs; change them to get data according to your needs
   ClientSession session = sessionManager.getSession(packet.getFrom());
   String token = (String)UserTokenList.get(JID.unescapeNode(session.getAddress().getNode().toLowerCase()));
   m.addElement("tokenNum").addText(token);
   return replyPacket;
   }
}

The reply message will look like:

<iq to="clientName">
   <query xmlns="jabber:iq:token">
     <tokenNum>abcdefgh</tokenNum>
   </query></iq>

About Pramod Jain
Pramod Jain is president of Innovative Decision Technologies, Inc. (INDENT, www.indent.org), in Jacksonville, FL. Their clients include Recruitmax, NASA, and NIH. Pramod has a PhD from the University of California, Berkeley.

About Mahaveer Jain
Mahaveer Jain is a lead programmer at INDENT. His expertise is in developing collaboration applications with Java technologies.