Servlets DevWorks

Introduction to Java Servlet technology
Skill Level: Introductory Roy W. Miller (roy@roywmiller.com) Programmer Independent
07 Dec 2004 Java developer and trainer Roy Miller revamps our existing introductory servlet material into this single easy-to-follow, hands-on tutorial. Roy introduces explains what servlets are, how they work, how you can use them to create Web applications of any degree of sophistication you can imagine, and how you can use servlets most effectively as a professional programmer.
Section 1. Before you start

About this tutorial
How do the pages you're reading in your favorite Web browser show up there? When you log into your favorite Web site, how does the Web site know that you're you? And how do Web retailers handle taking your order online? Those capabilities are possible because of code running on servers behind the scenes that interact with you in a Web session, access stored information throughout that process, and oftentimes present dynamic information in one or more Web pages. The core of those capabilities is provided in the Java language world by servlets. The goal of this tutorial is to introduce you to servlets. It describes what servlets are, how they work, how you can use them to create Web applications of any degree of sophistication you can imagine, and how you can use servlets most effectively as a professional programmer. The content of this tutorial is geared toward Java programmers who are unfamiliar, or only vaguely familiar, with servlets. It assumes a general knowledge of downloading and installing software, and a general knowledge of the Java language
Introduction to Java Servlet technology Copyright IBM Corporation 2004. All rights reserved. Trademarks Page 1 of 41
developerWorks
ibm.com/developerWorks
(creating classes, importing classes, etc.), but doesn't assume knowledge of servlets. The tutorial includes a simple example to illustrate basic servlet concepts, and a more involved example that illustrates more sophisticated use of servlets in a small contact management application. Should I take this tutorial? If you've been writing Web apps for a decade, this tutorial is not for you. If you don't know what servlets are, or are only vaguely familiar with them, keep reading. There is much more to servlets than what is included in this tutorial, but this is a good place to start your learning. You should, however, have the fundamentals of Java programming well in hand. If you're not quite there yet, try my Introduction to Java programming tutorial to get started.
Tools and code

To run the examples or sample code in this tutorial, you'll need to have at least JDK 1.4.2 or higher, along with the Eclipse IDE, installed on your machine. We'll walk through the process of installing the Tomcat plugin for Eclipse, which will allow you to develop servlet applications easily. All code examples in this tutorial have been tested with J2SE 1.4.2 on the Windows XP platform, but should work without modification using J2SE 1.4.1, or even 5.0. To install Tomcat, go to the Jakarta Web site (see Resources) and download the binary distribution of Tomcat 5.0.28 (the most current version that works with J2SE 1.4.2, as of this writing). The package comes with a Windows installer that makes installing on that platform a breeze. Follow the instructions in the readme files, and you'll be set. To install the Tomcat plugin for Eclipse, go to the Sysdeo Web site (see Resources ) and download the plugin zip file (tomcatPluginV3.zip, as of this writing). Then simply extract it to your plugins directory, and follow the instructions at the bottom of the download page to set up the plugin. To be sure your plugin is working correctly, work through the very simple HelloWorld servlet setup "tutorial" that is linked at the bottom of the Sysdeo page (see Resources for a direct link). Once you have Tomcat and the plugin installed, you're ready to begin this tutorial.
About the author

Roy Miller is an independent software development coach, programmer, and author.
Introduction to Java Servlet technology Copyright IBM Corporation 2004. All rights reserved.
Trademarks Page 2 of 41
developerWorks
He began his career at Andersen Consulting (now Accenture), and most recently spent three years using the Java platform professionally at RoleModel Software, Inc., in Holly Springs, NC. He has developed software, managed teams, and coached other programmers at clients ranging from two-person start-ups to Fortune 50 companies. For technical questions or comments about the content of this tutorial, contact Roy at roy@roywmiller.com.
Section 2. An introduction to servlets

What does a servlet do?
When you use an interactive Web site, all that you see is what's in the browser. Behind the scenes, there's a Web server taking requests from you in a session, possibly handing off to other code (maybe on other servers) to process the request and access data, and generating results to display in the browser. A servlet is the gatekeeper for that process. It lives on the Web server and handles incoming requests and outgoing responses. It has nothing to do with presentation, generally speaking, and really shouldn't. You can use a servlet to write to a stream that adds content to a Web page, but that's usually not a good idea either, because it tends to encourage mixing presentation with business logic.
Servlet alternatives
Servlets aren't the only way to serve up Web pages. One of the earliest technologies for this purpose was the common gateway interface (CGI), but that forked a separate process for each request, which wasn't very efficient. There were also proprietary server extensions like the Netscape Server API (NSAPI), but those were, well, proprietary. And in the Microsoft world, there is the active server pages (ASP) standard. Servlets provide an alternative to all of these, and they offer several advantages: They're as platform independent as the Java language They give you full access to the entire Java language API, including libraries for data access (like JDBC) They're (in most cases) inherently more efficient than CGI, because
developerWorks
servlets spawn new threads, rather than separate processes, for requests There is extensive industry support for servlets, including containers for most popular Web and application servers Servlets are a powerful addition to the professional programmer's toolbox.
But what is a servlet?

Most Java servlets that you'll encounter as a professional programmer are designed to respond to HTTP requests in the context of a Web application. As such, the HTTP-specific classes in the javax.servlet and javax.servlet.http packages are the ones you'll care about. When you create a Java servlet, you typically subclass HttpServlet. This class has methods that give you access to the request and response wrappers you can use to handle requests and create responses. The HTTP protocol isn't Java-specific, of course. It is simply a specification that defines what service requests and responses have to look like. The Java servlet classes wrap those low-level constructs in Java classes with convenience methods that make them easier to deal with in a Java language context. When a user issues a request via a URL, the Java servlet classes convert it to an HttpServletRequest and send it to the target pointed to by the URL, as defined in configuration files for the particular servlet container you're using. When the server side has done its work, the Java Runtime Environment packages the results in an HttpServletResponse and then sends a raw HTTP response back to the client that made the request. When you're interacting with a Web app, you usually make multiple requests and get multiple responses. All of them are within the context of a session, which Java language wraps in an HttpSession object. You can access this object when you process requests, and add stuff to it when you create responses. It provides some cross-request context. A container, like Tomcat, manages the runtime environment for servlets. You can configure the container to customize the way in which the J2EE server functions, and you must configure it to expose your servlets to the world. As we'll see, through various configuration files in the container, you provide a bridge from a URL (entered by a user in a browser) to the server-side components that handle the request that you want the URL to translate into. When your app runs, the container loads and initializes your servlet(s), and manages their lifecycle. When we say that servlets have a lifecycle, we simply mean that things happen in a predictable way when a servlet is invoked. In other words, certain methods on any servlet you create will always get called in the same order. Here's a typical scenario:
developerWorks
A user enters a URL in his browser. Your Web server configuration file says that this URL points to a servlet managed by a servlet container running on your server. If an instance of the servlet hasn't been created yet (there's only one instance of a servlet for an application), the container loads the class and instantiates it. The container calls init() on the servlet. The container calls service() on the servlet, and passes in a wrapped HttpServletRequest and HttpServletResponse. The servlet typically accesses elements in the request, delegates to other server-side classes to perform the requested service and to access resources like databases, then populates the response using that information If necessary, when the servlet's useful life is done, the container calls destroy() on the servlet to finalize it.
How do you "run" a servlet?

"Running" a servlet is like running a Java program. Once you've configured your container to know about the servlet, and to know that certain URLs should cause the container to invoke the servlet, the container will call the lifecycle methods in the proscribed order. So, running a servlet essentially means configuring it correctly, then pointing a browser to the right URL. Of course, the code in the servlet is where the interesting business logic happens. You don't have to worry about the low-level stuff that's going on, unless something goes wrong. Unfortunately, something does go wrong frustratingly often, especially when you're setting up a servlet. The single biggest cause of headaches with servlet applications is configuration files. You can't effectively debug them. You simply have to get them right by trial and error, as you try to decipher the error messages that you may or may not see in your browser.
Section 3. A simple servlet

What our simple servlet does
developerWorks
Our first servlet will do very little, but it will expose all of the basics of writing a servlet. The output will be some simple, unformatted text in a browser window:
Hello, World!
In creating this servlet, we'll be able to confirm that Tomcat functions as it ought to, and that we can use Eclipse to create a Web project like we should be able to. We'll also walk through the process of configuring your Web app in the Tomcat servlet container, which can be a bear if you happen make a slight mistake in an XML file. Don't worry: In this tutorial, at least, it will all work. In this first example, we'll write output to the browser directly from our servlet. This will be the last time in the tutorial that we use that approach.
Setting up Eclipse
There are a few things we need to do to make sure we can create and manage Tomcat projects in Eclipse. If you've installed the plugin (by simply extracting the Sysdeo zip file to your eclipse/plugins directory), you should get some additional menu items and tools on your toolbar. These are indicated in Figure 1. Figure 1. Tomcat plugin features
The toolbar buttons let you start, stop, and restart Tomcat, which you'll have to do when you want to run your servlets. To allow us to create Tomcat projects, which have the correct layout to facilitate
developerWorks
Tomcat deployment, we have to tell Eclipse a few things. If you click Window>Preferences, you'll see the standard Eclipse preferences dialog, with a new category called Tomcat at the bottom of the list. Clicking on that will show you the main Tomcat preferences page (see Figure 2). Figure 2. Tomcat preferences
Select Version 5.x, and specify the Tomcat home location. (On my system, this location is C:\Program Files\Apache Software Foundation\Tomcat 5.0, but yours may vary.) Select Context files as the context declaration mode. Then click on the JVM Settings preferences subcategory and make sure there's a valid JRE selected in the drop-down menu at the top of the page. You can use the default JRE, or you can point to your JDK, which you can tell Eclipse about in the Java>Installed JREs preferences page. When you're done, click OK. We're now ready to create a Tomcat project.
Creating a Tomcat project

developerWorks
The Tomcat plugin makes life much easier for the Web developer using Tomcat. If you click File>New>Project, and expand the Java wizard category in the dialog (see Figure 3), you'll see a new kind of a project wizard there: a Tomcat project. Figure 3. New Tomcat project
Click Next, name the project "HelloWorld," then click Finish. If you switch to the Java perspective in Eclipse, you'll be able to see that new project. It has a structure that will facilitate deployment to Tomcat (see Figure 4). Figure 4. Tomcat project structure
developerWorks
The work, WEB-INF, and WEB-INF/src directories are particularly important, as we'll see later.
Testing Tomcat
Click the Start Tomcat toolbar button. Eclipse should update your console with informational statements as Tomcat tries to launch. If it launches without showing any stack traces, you're set. If you see any stack traces in there, life becomes slightly more difficult. Unfortunately, trial and error (with your good friend Google) is the only way to track down any errors that occur. The good news is that starting with a fresh, new project like we did should eliminate the possibility of any nasty errors. When Tomcat starts, you won't see anything (except the console content). You'll have to test it to be sure it's working. If you want a quick indication, try opening a browser and entering the following URL:
http://localhost:8080/
If all goes well, you should see either a nice Tomcat welcome page, or a directory listing for the Tomcat "ROOT context." Don't worry about the second one. We'll prove Tomcat is working when we run our first servlet.
Declaring the class

A servlet is a class, so let's create a basic one. In Eclipse, create a class called HelloWorldServlet in your HelloWorld project. It should look like this:
public class HelloWorldServlet extends HttpServlet { public void service(HttpServletRequest request,
developerWorks
HttpServletResponse response) throws ServletException, IOException { PrintWriter writer = response.getWriter(); writer.println("Hello, World!"); writer.close(); } }
Enter this code, then press Ctrl+Shift+O to organize your import statements. Eclipse should give you imports for the following classes: java.io.IOException java.io.PrintWriter javax.servlet.ServletException javax.servlet.HttpServlet javax.servlet.HttpServletRequest javax.servlet.HttpServletResponse Notice that we subclassed HttpServlet, and that we overrode the service() method. The service() method is the most basic processing method that the servlet engine will call in our servlet's lifecycle. It takes a request wrapper and a response wrapper, which we can access in our method. We have no need to do so here, though, since we're just doing something basic to get a servlet working. We could have overridden doGet(), but service() will give us what we need. In our service() method, we called getWriter() on our response wrapper to allow us to print a string literal to that stream. Then we closed the stream. That's typical in a servlet that produces output: You perform the logic you need to perform, then you write to the output stream.
Configuring the Web application

The Java programming work is done, but now we have to do the necessary work with the configuration files. In my opinion, this is the biggest headache of Web development. Fortunately, the Tomcat plugin takes away some of the pain. Right-click on the HelloWorld project and select Properties. Select the Tomcat category of properties. You should see a context for the project that looks something like this:
/HelloWorld
Now go look at the filesystem in your Tomcat home. Drill down to the
developerWorks
conf/Catalina/localhost subdirectory. There, you should see a set of XML files. Specifically, you should see a HelloWorld.xml file. Open that. This file defines a Web application context for Tomcat.
<Context path="/HelloWorld" reloadable="true" docBase="path to your project\HelloWorld" workDir="path to your project\HelloWorld\work" />
When Tomcat launches, it reads these context files to tell the servlet container where to find your classes (which include your servlets). If you look back at the INFO statements Tomcat spits out to the console when it's loading, you'll see information about your Web application context in the list. The last step for configuring your Web application in Tomcat is to create a web.xml file, which needs to live in the WEB-INF directory of your project. (Note: Do not put it in the WEB-INF/src directory -- that's for other things.) Here's what the file should look like for this simple example:
<!DOCTYPE web-app PUBLIC '-//Sun Microsystems, Inc.//DTD Web Application 2.3//EN' 'http://java.sun.com/dtd/web-app_2_3.dtd'> <web-app> <servlet> <servlet-name>hello</servlet-name> <servlet-class>HelloWorldServlet</servlet-class> </servlet> <servlet-mapping> <servlet-name>hello</servlet-name> <url-pattern>/hello</url-pattern> </servlet-mapping> </web-app>
This file defines your Web application to Tomcat. The servlet-name element names your servlet for use in this file. The servlet-class element maps that name to a particular class that defines the servlet -- HelloWorldServlet, in our example. The servlet-mapping element tells Tomcat that URLs of the form (in this case) /hello map to our servlet, which is defined by the mapped servlet class. Once we have this file in place, we can fire up Tomcat and see our servlet load.
Running the servlet

As I mentioned earlier, "running the servlet" simply involves starting Tomcat and pointing a Web browser to the URL that should invoke it. Start Tomcat using the appropriate toolbar button (you'll need to stop it and start again if it's already running). Once Tomcat finishes launching, start a browser and enter the following URL:
developerWorks
http://localhost:8080/HelloWorld/hello
You should see a nice message in your browser window.
Section 4. Action servlets

Introduction
In the early days of Web development, many professional programmers had to figure out how to use servlets well as they went along. One of the most common results was an explosion of servlets on the server. There was one for each type of request. That quickly became painful, so programmers started to include conditional logic in their servlets to make them more adaptable, handling multiple types of requests. Over time, that also produced some ugly code. There's a better way, called an action servlet, which implements a concept called Model 2. As far as I know, this idea was first written about by David M. Geary (see Resources for more from him) but has been used well in popular servlet libraries, such as the Jakarta Struts project. In an action servlet, you don't have conditional logic that directs the servlet's behavior. Instead, you have actions (programmer-defined classes) that the servlet delegates to in order to handle different types of requests. Most of the time, that's a much better object-oriented (OO) approach than having multiple servlets, or multiple if conditions in a single servlet.
What our sample action servlet does

Our sample action servlet will be the gatekeeper for a simplistic browser-based application that will allow us to create, store, view, and delete contact list entries. Those entries will be nicely formatted. Eventually, users will have to log into the application in order to use it, but we'll add that feature later, in Users and data .
Setting up the project

Create a new Tomcat project in Eclipse, just like you did for HelloWorld. Note that the name of your project is the default context value for the servlet, so you'll use it
developerWorks
when we enter the URL that accesses the servlet. If you configured Tomcat to use context files, it should automatically create one for this project. Eclipse also should have created a project that has the correct structure, with the following important directories: WEB-INF WEB-INF/src work The first directory (WEB-INF) stores important configuration files, specifically the web.xml file that we'll discuss later. It also contains our compiled code, in the classes directory. The second directory (WEB-INF/src) stores the source code for our Java classes. The third directory (work) contains the compiled code for our JavaServer Pages (JSP) files, which Tomcat creates automatically for us whenever we hit a JSP page for the first time after code has changed (we'll talk about JSP technology more in the next section). The root of the project contains all of our JSP source files, as well as our database file. Note that you can see all of this structure in the Resource perspective in Eclipse, but you'll see only the WEB-INF/src and work directories in the Java Browsing perspective. All of these files are contained in the contacts.jar file included with this tutorial (see Resources for a link). To import them, simply create a new Tomcat project, then import contacts.jar (use the Import>Zip file option). That will bring all of the files in at the right locations, except the source code. The source code will end up in the src directory at the root of the project. Move the contents of that folder to WEB-INF/src and you should be all set.
Presentation
This is a tutorial about servlets, after all, which have almost nothing to do with presentation. Still, without seeing some results on the screen somewhere, we'd really be telling only part of the story. You certainly can write servlets that aren't involved with presentation at all, but most Web apps present information in a browser, which means that you have to choose a presentation mechanism to use. JavaServer Pages technology is one typical alternative and is used widely. With JSP technology, you can create dynamic Web pages. They support static HTML (or other markup, such as XML) and dynamic code elements that, as the name implies, can create content dynamically. Under the covers, JSP pages are compiled into servlets (that is, into Java code) by a container like Tomcat. You almost never will have to care about that, however. Just know that the following flow
developerWorks
occurs: A user types a URL in a browser that the J2EE servlet container points to a servlet The servlet does its job and puts information in the session, or in a bean, and forwards to the JSP page The JSP code translates information in the bean and/or the session, and sends the response to the browser You can create simple JSP pages easily and run them in Tomcat with only minor configuration changes in our Web app and without downloading additional code libraries, so we'll use them here (see Resources for much more detailed information about JSP technology). Our Contacts application will have one primary JSP page or listing existing contacts and adding new ones. Later, we'll add pages for login and logout. It's important to remember that JSP technology is just one presentation alternative. There are others. One that is gaining great popularity is the Jakarta Velocity templating package (see Resources). JSP technology does have a major drawback, which is that complicated, feature-rich apps tend to require ridiculously complex JSP pages, along with some extra server work to create custom tags if you want to keep your logic and presentation separate. Another drawback is that JSP technology offers a frequently irresistible temptation to mix business logic and presentation, which makes for brittle systems that bring on maintenance nightmares. In my opinion, JSP technology is frequently the wrong choice, and Velocity (or some other templating approach) is frequently the right one. But for our simple example, it will serve the purpose of illustrating the concepts we need to cover. In such a simple case, mixing a little logic and a little presentation is acceptable. Professionally, however, it's unwise most of the time, even though many programmers do it.
The web.xml file

In order for us to use the JSP page we're about to create, we have to tell Tomcat how to handle it. To do that, we have to create a web.xml file in our WEB-INF directory. It should look like this:
<!DOCTYPE web-app PUBLIC '-//Sun Microsystems, Inc.//DTD Web Application 2.3//EN' 'http://java.sun.com/dtd/web-app_2_3.dtd'> <web-app> <servlet> <servlet-name>contacts</servlet-name> <servlet-class> com.roywmiller.contacts.model2.ContactsServlet</servlet-class>
developerWorks
</servlet> <servlet-mapping> <servlet-name>contacts</servlet-name> <url-pattern>/index.htm</url-pattern> </servlet-mapping> <servlet-mapping> <servlet-name>contacts</servlet-name> <url-pattern>*.perform</url-pattern> </servlet-mapping> <servlet> <servlet-name>jspAssign</servlet-name> <servlet-class> org.apache.jasper.servlet.JspServlet</servlet-class> <init-param> <param-name>logVerbosityLevel</param-name> <param-value>WARNING</param-value> </init-param> <init-param> <param-name>fork</param-name> <param-value>false</param-value> </init-param> <load-on-startup>3</load-on-startup> </servlet> <servlet-mapping> <servlet-name>jspAssign</servlet-name> <url-pattern>/*.jsp</url-pattern> </servlet-mapping> </web-app>
We created a basic web.xml file for our HelloWorldServlet, but it was the pretty minimal. As your application becomes more complex, your web.xml file has to become more savvy. Let's analyze this file quickly. The <servlet> tag specifies a name alias for our servlet that we'll use elsewhere in the file. It also tells Tomcat which class to instantiate so as to create the servlet in memory. In my Eclipse workspace, I created a com.roywmiller.contacts.model2 package to hold the servlet class. You can call your package whatever you want, but the path to your servlet has to match what's in your <servlet-class> element here. The second servlet we define is one that comes with Tomcat when you download it, and you don't need to change it. It's simply the JSP-handling servlet. The <servlet-mapping> tells Tomcat which servlet to execute when certain URLs come to the server. We have three mappings here. The first maps the default page that the Web server looks for (<index.htm>) to our servlet. The second tells Tomcat to map any URL ending in .perform to our servlet. URLs of that form will tell our servlet which action to implement (we'll discuss how this works in more detail later). The third mapping tells Tomcat to use the JSP servlet to handle JSP pages.
The user view of the JSP page
developerWorks
In our simple example, we're not going to spend much time talking about JSP technology. That's to ensure that we keep things simple, and don't get bogged down in the details of presentation generally and JSP technology in particular. (Again, see Resources for more information.) We're also going to put everything on a single page, at least initially, even though that's somewhat unrealistic. That should minimize the number of pages we have to create just to illustrate the important concepts of using servlets. Our initial page will present the list of contacts, which will come from an object that contains the list. It will also contain a form for adding a new contact. The page will look like Figure 5. Figure 5. ContactList page
While not a work of art, the page displays all of our contacts in nicely formatted rows at the top. Each one has a Delete link that the user can click on to delete that particular contact. The form contains fields for name and address values, and radio buttons for the type of contact (family or acquaintance, in our simple example). This
developerWorks
simple page will allow us to explore how to use a simple action framework in our servlet application. It also will let us explore how to use the request and response our servlet receives from the browser during a user session. Now we're ready to create the page.
Coding the JSP page

Here's the code for our JSP page:
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <%@ page import="java.util.*" %> <%@ page import="com.roywmiller.contacts.model.*" %> <html> <head> <title>Contacts List 1.0</title> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <style type="text/css"> body, table, hr { color: black; background: silver; font-family: Verdana, sans-serif; font-size: x-small; } </style> </head> <body> <jsp:useBean id="contacts" scope="session" class="com.roywmiller.contacts.model.ContactList"/> <h2>Contact List 1.0</h2> <hr size="2"/> <table frame="below" width="100%"> <tr> <th align="left"></th> <th align="left">Name</th> <th align="left">Street</th> <th align="left">City</th> <th align="left">State</th> <th align="left">Zip</th> <th align="left">Type</th> </tr> <% List list = contacts.getContacts(); for (Iterator i = list.iterator(); i.hasNext();) { Contact contact = (Contact)i.next(); %> <tr> <td width="100">< a href="removeContactAction.perform?id=<%= contact.getId()%>" >Delete</a></td> <td width="200"><% =contact.getFirstname()%> <%=contact.getLastname()%></td> <td width="150"><%=contact.getStreet()%></td> <td width="100"><%=contact.getCity()%></td> <td width="100"><%=contact.getState()%></td> <td width="100"><%=contact.getZip()%></td> <td width="100"><%=contact.getType()%></td>
developerWorks
</tr> <% } %> </table> <br/> <br/> <br/> <fieldset> <legend><b>Add Contact</b></legend> <form method="post" action="addContactAction.perform"> <table> <tr> <td>First Name:<td> <td><input type="text" size="30" name="firstname"></td> </tr> <tr> <td>Last Name:<td> <td><input type="text" size="30" name="lastname"></td> </tr> <tr> <td>Street:<td> <td><input type="text" size="30" name="street"></td> </tr> <tr> <td>City:<td> <td><input type="text" size="30" name="city"></td> </tr> <tr> <td>State:<td> <td><input type="text" size="30" name="state"></td> </tr> <tr> <td>Zip:<td> <td><input type="text" size="30" name="zip"></td> </tr> <tr> <td>Type:<td> <td><input type="radio" size="30" name="type" value="family"> Family <input type="radio" size="30" name="type" value="acquaintance" checked> Acquaintance</td> </tr> </table> <br/> <input type="submit" name="addContact" value=" Add "> </form> </fieldset> </body> </html>
At this point, most of what you see there is probably Greek. We won't dissect all of it, but over the next few sections we'll hit the high points for understanding how our servlet will interact with this page.
Anatomy of a simple JSP page

developerWorks
In a JSP page, HTML is HTML. Java code embedded in the page appears like so:
<% Java code %>
In order to embed Java code in the page, you have to tell the JSP page where the classes are, just like you do in a Java class. You do that with statements like this:
<%@ page import="java.util.*" %>
Our page displays a list of contacts, which come from a ContactList instance that the JSP page knows about because of this line:
<jsp:useBean id="contacts" scope="session" class="com.roywmiller.contacts.model.ContactList"/>
This line tells the JSP page to use a bean, called contacts elsewhere in the page. It's an instance of com.roywmiller.contacts.model.ContactList, and it has session scope. Notice that we have a Java for loop in the body of the page:
List list = contacts.getContacts(); for (Iterator i = list.iterator(); i.hasNext();) { Contact contact = (Contact)i.next(); %> <tr> <td width="100"> <a href="removeContactAction.perform?id=<%= contact.getId()%>" >Delete</a></td> <td width="200"><%=contact.getFirstname()%> <%= contact.getLastname()%></td> <td width="150"><%=contact.getStreet()%></td> <td width="100"><%=contact.getCity()%></td> <td width="100"><%=contact.getState()%></td> <td width="100"><%=contact.getZip()%></td> <td width="100"><%=contact.getType()%></td> </tr> <% }
This illustrates how JSP technology lets you mix HTML and Java statements. Here we loop through the contact list of our contact object. Each time through the loop, we add a <tr> element to our HTML table. Within each row of the table, one per contact, we call getters on the Contact instance to populate our table cells. For the first cell, we need to create a Delete link for each row. We set the href attribute to the following string:
removeContactAction.perform?id=<%= contact.getId()%>
developerWorks
When a user clicks that link, that string will be appended to the end of the URL that gets sent to the server, after an initial slash (/) is added. The question mark is a delimiter for request parameters, which follow in name=value pairs. In this case, we send along the ID of each contact. This same kind of thing happens elsewhere in the page, such as in the form to add a new contact. Notice the <form> tag:
<form method="post" action="addContactAction.perform">
When a user clicks the Add button (the submit button at the bottom of the form), addContactAction.perform gets appended to the URL. That's all there is to it! Some of this syntactical magic is part of the reason many professional programmers either begrudgingly use JSP technology, or create various helper classes (such as custom JSP tags) that help make pages easier to create, read, and maintain. But now that we have the page, we can get to writing some code.
Creating the servlet

Our servlet is similar to HelloWorldServlet, with the addition of our action handling capability:
import java.io.IOException; import import import import javax.servlet.ServletException; javax.servlet.http.HttpServlet; javax.servlet.http.HttpServletRequest; javax.servlet.http.HttpServletResponse;
import com.roywmiller.contacts.actions.Action; public class ContactsServlet extends HttpServlet { protected ActionFactory factory = new ActionFactory(); public ContactsServlet() { super(); } protected String getActionName(HttpServletRequest request) { String path = request.getServletPath(); return path.substring(1, path.lastIndexOf(".")); } public void service(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { Action action = factory.create(getActionName(request));
developerWorks
String url = action.perform(request, response); if (url != null) getServletContext().getRequestDispatcher(url).forward(request, response); } }
We extend HttpServlet and override the service() method, just like before. In that method, we: Derive the action name from the URL that caused the servlet to be invoked Instantiate the correct action based on the name Tell the action to perform itself Forward the response to the URL to which our action points us We derive the action name from the URL that caused the servlet to be invoked, which we get from request.servletPath(). Remember that all of the URLs that cause us to invoke an action will have the form *.perform. We parse that to get the string to the left of the dot, which is the action name, then pass that action name to our ActionFactory to instantiate the right action. Now you see why we told our Web app how to handle URLs of that form, and why we used those "magic" strings in our JSP page. It was so we could decode them here and use actions to our advantage. What's the alternative? Lots and lots of if statements, and lots of extra code. With actions, as we'll see, each action we need to perform is neatly encapsulated. This is fine, but we need some additional classes to get the job done. That's where our action framework comes in.
A simple action framework

Our simple action framework has four main components: An ActionFactory. This factory translates action names in requests to action classes our servlet can use to do work. An Action interface. This interface defines the very simple public interface of all actions. An abstract class called ContactsAction. This class implements one method common to all actions, and forces subclasses to implement the other (perform()). Three subclasses of ContactsAction. These subclasses enable our servlet to bootstrap itself, add new contacts, and remove contacts.
developerWorks
In the service() method of our servlet, the process begins with ActionFactory.
The ActionFactory
Here's our ActionFactory:
import import import import import import java.util.HashMap; java.util.Map; com.roywmiller.contacts.actions.Action; com.roywmiller.contacts.actions.AddContactAction; com.roywmiller.contacts.actions.BootstrapAction; com.roywmiller.contacts.actions.RemoveContactAction;
public class ActionFactory { protected Map map = defaultMap(); public ActionFactory() { super(); } public Action create(String actionName) { Class klass = (Class) map.get(actionName); if (klass == null) throw new RuntimeException(getClass() + " was unable to find an action named '" + actionName + "'."); Action actionInstance = null; try { actionInstance = (Action) klass.newInstance(); } catch (Exception e) { e.printStackTrace(); } return actionInstance; } protected Map defaultMap() { Map map = new HashMap(); map.put("index", BootstrapAction.class); map.put("addContactAction", AddContactAction.class); map.put("removeContactAction", RemoveContactAction.class); return map; } }
An ActionFactory is quite simple. It has a Map of action classes and their names. We use the names in our pages to tell our servlet which actions to perform. In our case, we have three actions: BootstrapAction AddContactAction RemoveContactAction Recall that the actions to add and remove contacts were sent as URLs to our servlet by the Add form and the Delete link, respectively. The BootstrapAction simply
developerWorks
fits the calling of /index.htm into our action framework. When we tell the factory to create an Action, it instantiates the class and gives us back the instance. Adding a new action to the factory is as simple as creating the class for the action, then adding a new entry in the factory's action Map.
Actions
The Action interface looks like this:
import javax.servlet.http.HttpServletRequest; import javax.servlet.http.HttpServletResponse; public interface Action { public String perform(HttpServletRequest request, HttpServletResponse response); public void writeToResponseStream(HttpServletResponse response, String output); }
The method we'll be using extensively now is perform(). The other method, writeToReponseStream(), allows an action to write directly to the output stream of the response, for passthrough to the JSP page. Whatever gets written (text, HTML, etc.) shows up on the page. We won't be using that method for the moment, but it's available on ContactsAction for you to see how it's done. Remember that we used the code in the body of this method in our HelloWorldServlet, so it shouldn't be unfamiliar to you.
Starting up with BootstrapAction

The simplest subclass of ContactsAction we have is BootstrapAction, and it's a good pattern for the others:
import javax.servlet.http.HttpServletRequest; import javax.servlet.http.HttpServletResponse; public class BootstrapAction extends ContactsAction { public String perform(HttpServletRequest request, HttpServletResponse response) { return "/" + "contactList.jsp"; } }
We simply implement perform() to do what we want. In this case, all we do is return a URL that points to contactList.jsp (the initial slash is important, so don't forget it). Now look back at the service() method on ContactsServlet:
public void service(HttpServletRequest request, HttpServletResponse response)
developerWorks
throws ServletException, IOException { Action action = factory.create(getActionName(request)); String url = action.perform(request, response); if (url != null) getServletContext().getRequestDispatcher(url).forward(request, response); }
Our action either returns a URL string or writes to the output stream for printing on the JSP page. If our action returns a URL string, which our BootstrapAction does, we get the ServletContext, ask it for a RequestDispatcher on our URL, and finally forward the request and response to the JSP servlet so it can construct the page. Our action servlet regains control after that, and can do any remaining work, as long as it doesn't write to the JSP page's PrintStream, which is now closed. We could call sendRedirect() on our response if we wanted to, rather than using a RequestDispatcher:
response.sendRedirect("http://...");
But there's a price to pay for doing that. When we use a dispatcher, we're delegating the request and response to the JSP servlet, which also forwards the existing HttpSession. That preserves the contents of the session. Forwarding to another URL wouldn't. At the moment, when we're displaying the page initially, there's nothing in the session that we care about, so the effect would be the same. But very soon it will be important to preserve the session contents. To tell our action framework about this new available action, we added this line to the factory's action Map:
map.put("index", BootstrapAction.class);
Adding contacts
An app that shows a page that doesn't let you do anything isn't very useful. We need to be able to add contacts. To do that, we must: Create a class called AddContactAction Implement perform() to add a new Contact instance to the ContactList we maintain in our Web session Tell the factory about the new action
developerWorks
Telling the factory about the action is as simple as adding another entry in the factory's map, as we did with BootstrapAction. The AddContactAction class, with its perform() method implemented, looks like this:
import javax.servlet.http.HttpServletRequest; import javax.servlet.http.HttpServletResponse; import javax.servlet.http.HttpSession; import com.roywmiller.contacts.model.Contact; import com.roywmiller.contacts.model.ContactList; public class AddContactAction extends ContactsAction { public String perform(HttpServletRequest request, HttpServletResponse response) { Contact newContact = createContact(request); HttpSession session = request.getSession(); ContactList contacts = (ContactList) session.getAttribute("contacts"); contacts.addContact(newContact); session.setAttribute("contacts", contacts); return "/contactList.jsp"; } protected Contact createContact(HttpServletRequest request) { Contact contact = new Contact(); contact.setFirstname(request.getParameter(RequestParameters.FIRSTNAME)); contact.setLastname(request.getParameter(RequestParameters.LASTNAME)); contact.setStreet(request.getParameter(RequestParameters.STREET)); contact.setCity(request.getParameter(RequestParameters.CITY)); contact.setState(request.getParameter(RequestParameters.STATE)); contact.setZip(request.getParameter(RequestParameters.ZIP)); contact.setType(request.getParameter(RequestParameters.TYPE)); return contact; } }
All we do here is call createContact() to create a new Contact and set its instance variables to the corresponding values containing the request parameters. Then we add the new Contact to the ContactList in the HttpSession. Finally, we tell our servlet to return to /contactList.jsp. Recall that every time we create a Contact, the constructor assigns it a unique ID. Look back at the JSP code for a moment. You'll see two important things in it with regard to what we're doing in this action. First, notice that we guaranteed we'd always have a ContactList instance in our session by adding this line:
<jsp:useBean id="contacts" scope="session" class="com.roywmiller.contacts.model.ContactList"/>
When the JSP page is first compiled and displayed (as a result of forwarding to the JSP page after BootstrapAction performs), it will instantiate a ContactList.
developerWorks
This object has nothing in it, which is why our list of contacts shows up empty when we start the app. In AddContactAction, we're modifying that object to add new contact information, then reinserting it in the session. When the page displays after that, it will read the ContactList's list of Contact instances and display them. Second, notice that our form to add contacts looks like this:
<form method="post" action="addContactAction.perform"> table with labels and text input fields <input type="submit" name="addContact" value=" </form> Add ">
The action for the form causes addContactAction.perform to be sent in the request to our servlet. It then extracts the addContactAction portion as the action name, looks it up in the factory, and creates an instance of AddContactsAction.
Removing contacts
Adding contacts is nice, but being able to remove them is just as important. Being able to edit them would be nice, but this tutorial is only so long. Besides, adding edit capability would be as simple as adding another action. So for now, we'll just add the ability to remove contacts and then move on to more interesting things. As before, all we have to do is add a new action class, implement its perform() method, and tell the factory about it. We also have to be sure that our JSP code tells our servlet to invoke the action at the right time. Look at the Delete link for each row of the contacts table in our JSP page:
<a href="removeContactAction.perform?id=<%= contact.getId()%>" >Delete</a>
That link tells our servlet to ask the factory for the right action class for the name removeContactAction. It also passes a parameter called id in the request, with a value set to the current contact's ID. Our class looks like this:
import javax.servlet.http.HttpServletRequest; import javax.servlet.http.HttpServletResponse; import javax.servlet.http.HttpSession; import com.roywmiller.contacts.model.ContactList; public class RemoveContactAction extends ContactsAction {
developerWorks
public String perform(HttpServletRequest request, HttpServletResponse response) { int contactId = Integer.parseInt(request.getParameter("id")); HttpSession session = request.getSession(); ContactList contacts = (ContactList) session.getAttribute("contacts"); contacts.removeContact(contactId); session.setAttribute("contacts", contacts); return "/contactList.jsp"; } }
All we do here is grab the id parameter and the ContactList from the session, tell the list to remove the Contact with that id, then replace the list on the session. Last, but not least, we tell the servlet to forward back to contactList.jsp.
Running the app

If curiosity hasn't gotten the better of you already, you should now run the app to see how it works. Launch a browser and enter the following URL:
http://localhost:8080/contacts/
If Tomcat works its magic correctly, you should be looking at contactList.jsp, with no contacts in the list. Type in some values for the text fields on the add form and click the Add button. You should see a new contact in the list, complete with a Delete link to the left of the contact's name. Unless you changed it, its type should be set to Acquaintance (the default radio button selection for type). To keep things simple, we didn't do any validation of the form, so you can enter multiple contacts with exactly the same values for all fields. Each contact has a unique ID, so each one will show up separately, and you can delete them individually. That's it -- we have a working Web application! But we can't save our list of contacts, so we have to reenter them every time we launch the app. Worse, every user of our app has the same list of contacts. We can fix that by adding support for unique users, and by storing data in a file (the simplest database that could possibly work). We'll do both in the next section.
Section 5. Users and data
developerWorks
Enhancing the app

In this section, we're going to refactor our code and our existing JSP page somewhat to handle persistent contact data for unique users. In a nutshell, we'll do the following: Create a ContactsUser object Give each ContactsUser a username, password, and contact list Change our JSP page's <jsp:useBean/> tag to use a ContactsUser Add login.jsp as the first page of the app Change contactList.jsp to include a friendly welcome message for the logged-in user Add a Logout link to contactList.jsp to invoke LogoutAction Add a goodbye.jsp to display a personalized goodbye message Add a LoginAction and a LogoutAction Add a UsersDatabase to handle storing and retrieving Contacts from usersDatabase.txt Initialize the ContactsDatabase by overriding init() on our servlet Override destroy() on our servlet to tell UsersDatabase to close usersDatabase.txt Not so bad, really. The only real new concepts are using files (just more standard Java language work) and directing to new pages. All of the action handling mechanisms are the same. This will illustrate the power of our action framework, which took precious little time to create. And it isn't nearly as complicated as something like Jakarta's Struts framework (see Resources), which just might be overkill for what you're doing in your application.
The ContactsUser
Our ContactsUser object looks like this, minus the import statements and accessors (you can find the complete source code in contacts.jar):
public class ContactsUser { protected String username = ""; protected String password = ""; protected List contactList = new ArrayList();
developerWorks
public ContactsUser() { } public ContactsUser(String username, String password, List contactList) { this.username = username; this.password = password; this.contactList.addAll(contactList); } public boolean hasContacts() { return !contactList.isEmpty(); } public void addContact(Contact aContact) { contactList.add(aContact); } public void removeContact(Contact aContact) { contactList.remove(aContact); } public void removeContact(int id) { Contact toRemove = findContact(id); contactList.remove(toRemove); } protected Contact findContact(int id) { Contact found = null; Iterator iterator = contactList.iterator(); while (iterator.hasNext()) { Contact current = (Contact) iterator.next(); if (current.getId() == id) found = current; } return found; } accessors... }
This class holds information about users of the application. For the most part, that's all it does. It holds the user's username and password, and maintains a list of contacts for the user. It allows various actions in our action framework to add Contacts to the user, and remove them. The no-argument constructor is there for unit testing purposes. The other constructor, the one that takes three arguments, is the one the app uses. You might be asking yourself, "Why doesn't this class have a ContactList instance variable?" After all, we went to the trouble of creating it in the first place. Why don't we use it? The simple answer is that we really don't need that class anymore. It wrapped an ArrayList and gave us some helper methods. Those helper methods really make more sense on ContactUser. If we used a ContactList, we'd be calling methods on it from ContactUser that have the same names, and that we want to accomplish the same thing. For example, if ContactUser had a ContactList, and we named that instance variable contactList, here's what addContact() would look like:
developerWorks
public void addContact(Contact aContact) { contactList.addContact(aContact); }
Delegating to that other object here is a little silly. So we deleted the ContactList class. That's what refactoring is all about. We simplified our code, and reduced the number of classes in our system, but still got the same job done. Having ContactList was an intermediate step in creating our system. It let us get up and running, and helped us create our action framework. Then its useful life ended and we got rid of it. Just because you wrote some code doesn't mean you have to keep it around forever.
Changing contactList.jsp
Changing the JSP page to use our new ContactUser is straightforward. There are three changes we need to make. The first is to change the <jsp:useBean> tag to look like this:
<jsp:useBean id="user" scope="session" class="com.roywmiller.contacts.model.ContactsUser"/>
Now our page will instantiate a ContactsUser instead of a ContactList. The second change is to update the table row-building logic in the page to use our new user variable:
<% List list = user.getContacts(); for (Iterator i = list.iterator(); i.hasNext();) { Contact contact = (Contact)i.next(); %>
The third change is to add a link for a user to log out:

<a href="logoutAction.perform">Logout</a>
We'll put this link next to the "Contacts 1.0" header. When a user clicks the link, our servlet will perform the LogoutAction.
Adding the login/logout pages

Compared to our other pages, the ones to support logging in and logging out are
developerWorks
simple. The only differences exist within the <body> tag. Here's login.jsp:
<body> <h2>Contact List 1.0</h2> <hr size="2"/> <fieldset> <legend><b>Please Login</b></legend> <form method="post" action="loginAction.perform"> <table> <tr> <td>Username:<td> <td><input type="text" size="30" name="username"></td> </tr> <tr> <td>Password:<td> <td><input type="text" size="30" name="password"></td> </tr> </table> <br/> <input type="submit" name="login" value=" Login "> </form> </fieldset> </body>
This page has a form with two text fields and a submit button. When a user clicks Login, our servlet will perform the LoginAction. Here's goodbye.jsp:
<body> <jsp:useBean id="user" scope="session" class="com.roywmiller.contacts.model.ContactsUser"/> <h2>Contact List 1.0</h2> <hr size="2"/> Goodbye <%= user.getUsername() %>! </body>
This page calls getUsername() on a ContactsUser bean to display a personalized goodbye message. When a user tries to log in with a username and password that aren't in our database, our application gives up and routes the user to an error page, which looks like this:
<body> <h2>Contact List 1.0</h2> <hr size="2"/> <fieldset> <legend><b>Error</b></legend> There was an error: <%= session.getAttribute("errorMessage") %> </fieldset> </body>
developerWorks
This is the simplest page we have. It uses the default session variable available in all JSP pages to display an error message.
Adding LoginAction
The LoginAction class looks like this:
public class LoginAction implements Action { public String perform(HttpServletRequest request, HttpServletResponse response) { String username = request.getParameter(USERNAME); String password = request.getParameter(PASSWORD); ContactsUser user = UserDatabase.getSingleton().get(username, password); if (user != null) { ContactsUser contactsUser = (ContactsUser) user; request.getSession().setAttribute("user", contactsUser); return "/contactList.jsp"; } else request.getSession().setAttribute("errorMessage", "Invalid username/password."); return "/error.jsp"; } }
This action grabs the username and password parameters from the request, then checks to see if our database contains a user with that username/password combination. If it does, we put the user in the session and go directly to contactList.jsp. If that user doesn't exist in our database, we set an error message on the session and go to error.jsp. Adding actions should be easy for us now. We add an entry to our action factory that looks like this:
map.put("loginAction", LoginAction.class);
With the pages in place, and our factory aware of our new action, we're done. You should be able to run the app and see the login page. When you enter a username and password, regardless of what they are, you should see the error page. In just a little while, you'll be able to log in with a valid username and password, and see contactList.jsp with an empty contact list.
Adding LogoutAction
The LogoutAction class looks like this:
developerWorks
public class LogoutAction implements Action { public String perform(HttpServletRequest request, HttpServletResponse response) { UserDatabase.getSingleton().shutDown(); return "/goodbye.jsp"; } }
Here we tell our database to shutDown(). That method on UserDatabase looks like this:
public void shutDown() { writeUsers(); } protected void writeUsers() { StringBuffer buffer = new StringBuffer(); Collection allUsers = users.values(); Iterator iterator = allUsers.iterator(); while (iterator.hasNext()) { ContactsUser each = (ContactsUser) iterator.next(); UserRecord record = new UserRecord(each); buffer.append(record.getFullRecord()); } writeText(buffer.toString()); } protected synchronized void writeText(String text) { Writer writer = null; try { writer = new FileWriter(usersFile.getAbsolutePath()); writer.write(text); } catch (Exception e) { throw new RuntimeException("Unable to append to file.", e); } finally { closeWriter(writer); } }
shutDown() calls writeUsers(), which iterates over all the users we have in memory (from when we read in the file when the servlet initialized itself), creates a UserRecord for each, then passes the complete string to writeText(). That method dumps the string to the file, overwriting the existing contents. The UserRecord class is a nice helper that encapsulates all of the somewhat messy tokenizing work for each user record in the file. You can examine the code for yourself (see contacts.jar for the complete source code listing). Once the database is shut down, we tell the servlet to forward to goodbye.jsp to display our personalized goodbye.
The userDatabase.txt file

Most Web applications access data from a "database" of some kind. Many use an
developerWorks
industrial-strength RDBMS, but a text file can be a database also. It's the simplest database that could possibly work. If you wrap it well, and hide the access details behind an interface that makes accessing the data reasonably painless for other classes in your app, the form of the underlying data store really doesn't matter. In our application, we'll use a text file. That file will contain a single line per user, in the following form:
username password comma-delimited contact1 info|comma-delimited contactN info|...
The usernames in our file will be plain text, but the passwords will be Base64 encoded for (arguably too-simple) security. The entries for contacts will be comma-delimited within each. The contacts themselves will be delimited by pipe characters (|). There's no magic to this formatting. It just does what we need it to do, which is to allow us to parse the file easily. For convenience, we'll put this file at the root of our project so that the path to the file is straightforward. To keep things simple, our application doesn't support user maintenance features, which means that there's no way to add or remove a user from within the app. That means that you have to add users to userDatabase.txt manually. For example, to add a user named testuser with the password password, add the following line to the file:
testuser cGFzc3dvcmQ=
The password in each entry is encoded with Base64 encoding. You can use the EncoderDecoder class in contacts.jar to figure out the encoded version of your password. Its main() method lets you enter a plaintext string, then run the class to print the encoded password to the console.
The UserDatabase
Our UserDatabase wraps interaction with our text file. The class listing looks large, but it's not complicated (much of the perceived complexity is the extra Java coding stuff you need to deal with reading and writing files). We'll discuss a few high points in this section (see contacts.jarfor the complete code listing). The class implements the Singleton pattern. The class maintains a single instance that all users share by calling getSingleton(). The class maintains a Map of ContactsUsers, with the concatenated username
developerWorks
and password acting as the key for each entry. Anything could serve as the key, but this was convenient. In our servlet init() method, we'll tell the UserDatabase where the database file is located (based on the ServletContext), then we'll tell it to initialize itself by calling initialize(). That method looks like this:
public void initialize() { usersFile = new File(databaseFilePathname); String allUsers = retrieveText(); StringTokenizer tokenizer = new StringTokenizer(allUsers, "\n"); while (tokenizer.hasMoreTokens()) { String userEntry = tokenizer.nextToken(); UserRecord record = new UserRecord(userEntry); put(new ContactsUser(record.getName(), record.getPassword(), record.getContactList())); } }
This method reads in the complete file by calling retrieveText, tokenizes the big string, creates a UserRecord for each user, then calls put() to place the new ContactsUser in the map. The real work of this method goes on in the calls to retrieveText() and put():
protected synchronized String retrieveText() { BufferedReader bufferedReader = null; try { bufferedReader = new BufferedReader(new FileReader(usersFile.getAbsolutePath())); char charBuff[] = new char[(int) new File(usersFile.getAbsolutePath()).length()]; bufferedReader.read(charBuff); return new String(charBuff); } catch (Exception e) { throw new RuntimeException("Unable to read in the file.", e); } finally { closeReader(bufferedReader); } } protected void closeReader(BufferedReader bufferedReader) { try { if (bufferedReader != null) bufferedReader.close(); } catch (Exception ex) {} } public void put(ContactsUser user) { String userKey = user.getUsername() + user.getPassword(); users.put(userKey, user); }
The retrieveText() method does the file reading work. It creates a BufferedReader to read the entire file contents into a character buffer, which is
developerWorks
then turned into a String. In its finally clause, it calls closeReader() to do just that. The writeText() method writes output to the file, overwriting its existing contents. The method hides the same kind of file interaction details. The put() method creates the key for the user (the username plus the password) and inserts it in the map of users.
Section 6. Using servlets effectively

Introduction
In this tutorial, we've only scratched the surface of what you can do with servlets. Web apps can be as sophisticated as you can imagine. The underlying mechanisms, though, are essentially the same for all of them, and if you're writing code in the Java language, servlets are at the core. Creating more sophisticated apps is a matter of using more sophisticated tools and libraries. However, that's where many programmers make mistakes that lead them to create terrible Web applications. This section contains some suggestions on how to avoid that. Most Java programmers with Web development experience would agree with some of these. Some are more controversial. In any case, they should help get you started with servlets well.
Use a single servlet

If you can't use just one servlet, use as few as possible. In fact, I suggest that you use one until you absolutely can't anymore. You don't need a hundred servlets. You certainly don't need one for each type of request.
Don't hang around the servlet

Spend as little time in your servlet as possible. Servlets aren't a place for business logic. That's bad OOP design. Think of a servlet as one of two things: An extra layer behind your UI that helps get "events" to the server
developerWorks
An extra layer in front of the server that allows you to use a browser as your UI Either way, it's a place where you quickly dispatch matters to other pieces of your application, then get out.
Use actions
An action framework, even a simple one like the one we used in this tutorial, is a powerful approach. It allows you to follow my previous bit of advice: to spend as little time in your servlet as possible. It's also good OOP design. Each action class does one thing, or a very cohesive set of related things. Some people think this fragments the code, making it more difficult to understand. I think that this objection stems from two things: People aren't used to looking at object-oriented code, which certainly can feel more fragmented when you first start looking at it People favor more procedural code, even in an object Think about the alternative. Without actions (assuming you don't simply delegate to other objects that you don't call "actions"), you'll have to have lots of if statements. If you have two or three, your code will be reasonably easy to read. If you have ten, that's silly. A method that requires you to scroll through several screens is bad news. It usually means that the method does too much. At minimum, you should extract the action-like things you do in service() (or doGet(), etc.) into other methods that are named well so you know what they do. Use actions. When you need to add features, add actions.
Use service(), unless you can't

There are many folks in the servlet world who claim that you shouldn't override service(). Instead, they say, you should override doGet() and/or doPost(), possibly having them call each other, making your code look something like this:
public void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { statements } public void doPost(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { doGet(request, response); }
developerWorks
I don't know where these ideas come from. Bruce Eckle suggests that it's a holdover from the CGI days, when folks grew used to paying attention to whether a GET or POST was coming in. (See Resources for a link to a more detailed version of his theory.) I've never heard a good reason not to use service(). In any case, if you use it, and then determine later that it would be better to use one of the doX() flavors, refactor the code! Until then, use service(), because it's easier.
Don't mix presentation and business logic

Generating complex HTML strings to spit out to the JSP page's print stream is fine for a simple app, but creating a more feature-rich app that way is more painful than it needs to be. It's wiser to keep presentation in the place where it belongs: in the page. JSP technology allows you to do that, but as I said earlier, it requires a lot more work to keep business logic and presentation separate. Templating engines like Velocity are often a better choice. Whatever approach you choose, minimize the mixing of business logic and presentation.
Handle exceptions well

We didn't talk about this much in this tutorial, but handling exceptions becomes important when you're creating a Web app. There is nothing more frustrating to the user than having something unexpected happen on the server side, and then seeing a cryptic stack trace in your browser. Some of those traces can be disgustingly obtuse, and obscure. Tracking them down can be frustrating. Many of the packaged Web app development libraries, like Struts (see Resources), come with built-in frameworks for handling dynamic display of messages, including errors. Use those features.
Don't use every feature

Do you need to use every single feature of the Web app development framework or library you're using? Probably not, and using every one can make your code much more complex than it needs to be. In fact, I recommend that you not start with a framework at all, except the simplest one that could possibly work. Sometimes that feature or framework you "know" you'll be using or need to use ends up being extreme overkill for the problem at hand. Use a framework when it's wise to use one. Don't assume that you need it. Wait for your system to tell you that you do. Some programmers consider that "bad design."
developerWorks
Not so. Assuming you'll need a particular framework, or even a particular feature of that framework, can be overdesign. You should design for what you need; most often, that changes as the system grows. It's frustrating to pick a framework before development begins, then hit a brick wall because that framework doesn't support or allow something you want to do.
Section 7. Summary
Conclusion
In this tutorial, you've learned about Java servlets, and how to use them professionally. Certainly the examples in this tutorial are simple, but they illustrate most of the servlet concepts you would use to create a Web application. There are more features (configuration and otherwise) available to you, but the core of almost every Web app written with the Java language is one or more servlets acting as gatekeepers for business logic on one or more servers behind the scenes. More importantly, you've learned some techniques for using servlets well. Web applications often grow into messy piles of code. By using some simple techniques driven by fundamental OOP principles, you can avoid the mess, and create apps that are a joy to enhance and maintain.
developerWorks
Resources
Learn David Geary's Advanced JavaServer Pages (Pearson Higher Education, 2001) is a must-read. If your Web application is too complex for simple JSP components, you may wish to investigate the Velocity Template Engine or the Struts Web Application Framework. Both projects are hosted by the Apache Foundation. The Java Technology home page is the "official" Java language resource. The Java Servlet Technology section of the site offers a wealth of information on programming with servlets. The Java tutorial from Sun is an excellent resource. There is a "trail" for servlets. The developerWorks New to Java technology page is a clearinghouse for developerWorks resources for beginning Java developers, including links to tutorials and certification resources. Bruce Eckle speculates on historical reasons why many Web programmers are reluctant to override service(). You'll find articles about every aspect of Java programming in the developerWorks Java technology zone. Also see the Java technology zone tutorials page for a complete listing of free Java-focused tutorials from developerWorks. Get products and technologies Download the contacts.jar that accompanies this tutorial. Download Tomcat from the Apache Jakarta Project. You'll also need the Tomcat plug-in for Eclipse from Sysdeo. Once you've installed it, check out the HelloWorld servlet setup "tutorial" from Sysdeo. Discuss Participate in developerWorks blogs and get involved in the developerWorks community.
About the author

Roy W. Miller Roy Miller is an independent software development coach,
developerWorks
programmer, and author. He began his career at Andersen Consulting (now Accenture), and most recently spent three years using Java professionally at RoleModel Software, Inc. in Holly Springs, NC. He has developed software, managed teams, and coached other programmers at clients ranging from two-person start-ups to Fortune 50 companies.

Servlets DevWorks

Uploaded by

Copyright:

Available Formats

You might also like

Servlets DevWorks

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Servlets DevWorks

Uploaded by

Copyright:

Available Formats

Introduction to Java Servlet technology

Skill Level: Introductory Roy W. Miller (roy@roywmiller.com) Programmer Independent

Section 1. Before you start

Tools and code

About the author

Section 2. An introduction to servlets

But what is a servlet?

How do you "run" a servlet?

Section 3. A simple servlet

Creating a Tomcat project

Declaring the class

Configuring the Web application

Running the servlet

You should see a nice message in your browser window.

Section 4. Action servlets

What our sample action servlet does

Setting up the project

The web.xml file

The user view of the JSP page

Coding the JSP page

Anatomy of a simple JSP page

Creating the servlet

String url = action.perform(request, response); if (url != null) getServletContext().getRequestDispatcher(url).forward(request, response); } }

A simple action framework

Starting up with BootstrapAction

Running the app

Section 5. Users and data

Enhancing the app

public void addContact(Contact aContact) { contactList.addContact(aContact); }

The third change is to add a link for a user to log out:

Adding the login/logout pages

The userDatabase.txt file

Section 6. Using servlets effectively

Use a single servlet

Don't hang around the servlet

Use service(), unless you can't

Don't mix presentation and business logic

Handle exceptions well

Don't use every feature

About the author

You might also like