The code that appears between the <% and %> delimiters is called a scriptlet. By the way, we really hate the term “scriptlet.” It seems to imply (falsely) a completeness that isn’t there. It is too parallel to the term “applet,” which is a complete Java program that runs inside a browser. A scriptlet isn’t necessarily a complete anything. It’s a snippet of code that gets dropped inside the code of the servlet generated from the JSP source.

Recall that servlets may have a doPost() and a doGet() methods, which we collapsed in our example by having them both call the doBoth() method. Same sort of thing is happening here with the JSP, and the doBoth() ends up doing all the output of the HTML. Any snippets of Java code from within the <% and %> delimiters get dropped right in place between those output calls, becoming just a part of a method.

It can be useful to keep this in mind when writing JSP. It helps you answer the questions of scope—who has access to what, where are variables getting declared and how long will they be around? (Can you answer that last question? Since any variable declared inside the <% and %> will be in the JSP equivalent of our doBoth() method, then that variable will only be around for the duration of that one call to the doBoth(), which is the result of one GET (or POST) from the browser.)

The source code snippets can be just pieces of Java, so long as it makes a complete construct when all is converted. For example, we can write:

<% if (acct != null) { // acct.getParent() %>
    <a href="BudgetPro?func=back">
    <img src="/back.gif">
    </a>
<% } else { %>
    <img src="/back.gif">
<% } %>

Notice how the if-else construct is broken up into three separate scriptlets—that is, snippets of code. Between them, in the body of the if and the else, is plain HTML. Here is what that will get translated into after the JSP conversion:

if (acct != null) { // acct.getParent()
    out.println("<a href="BudgetPro?func=back">");
    out.println("<img src="/back.gif">");
    out.println("</a>");
} else {
    out.println("<img src="/back.gif">");
}

Do you also see why we describe it as being “turned inside out”? What was delimited becomes undelimited; what was straight HTML becomes delimited strings in output statements.

As long as we’re on the topic of conversion, let’s consider comments. There are two ways to write comments in JSP, either in HTML or in Java. In HTML we would write:

<!-- HTML comment format -->

but since we can put Java inside of delimiters, we can use Java comments, too:

<% // Java comment format %>

or even:

<% /*
       * Larger comments, too.
       */
%>

If you’ve been following what we’ve been saying about translation of JSP into Java code, you may have figured out the difference. The Java comments, when compiled, will be removed, as all comments are, from the final executable.

The HTML-based comments, however, will be part of the final HTML output. This means that you’ll see the HTML comments in the HTML that reaches the browser. (Use the View Source command in your browser to see them. As HTML comments, they aren’t displayed on the page, but they are sent to the browser.) This is especially something to be aware of when writing a loop. Remember our loop for generating the table?

<% // for each subaccount:
  for (Iterator actit = acct.getAllSubs(); actit.hasNext(); ) {
    Account suba = (Account) actit.next();
%>
    <!-- Next Row -->
    <tr>
    <td><a href="BPControl?name=<%= suba.getName() %>&func=cd">
...
<% } // next acct %>

We’ve put a comment just prior to the <tr> tag. What will happen is that the comment will be part of the generated HTML, and since this is a loop, the comment, just like the <tr> and other tags, will be repeated for each iteration of the loop. Now we’re not saying this is undesirable—in fact it makes the resultant HTML more readable. But be aware that these comments will be visible to the end user. Be careful in what you say in them. The additional transmission time required for these few extra bytes is probably imperceptible, unless your comments are large and repeated many times.

The other place that code can be placed is outside the doGet() and doPost(). It is still inside the class definition for the servlet class that gets generated from the JSP, but it is not inside any method. Such code is delimited like this:

<%! code %>

The exclamation mark makes all the difference. Since it’s outside any method, such code typically includes things like variable declarations and complete method declarations. For example:

<%! public static MyType varbl;

public long
countEm()
{
    long retval = 0L;
    retval *= varbl.toLong();
    return retval;
}
%>

If you tried to do something like this inside of a scriptlet, you would get errors when the server tries to compile your JSP. Such syntax belongs at the outer lexical level. The use of the <%! ... %> syntax puts it there.

This delimiter is a shorthand for getting output from a very small bit of Java into the output stream. It’s not a complete Java statement, only an expression that evaluates into a String. Here’s an example:

<h4>As of <%= new java.util.Date() %></h4>

which will create a Java Date object (initialized, by default, with the current date/time) and then call the toString() method on that object. This yields a date/time stamp as part of an <h4> heading.

Any methods and variables defined inside the previously described delimiters are OK to use with this expression shorthand.

There are also a few predefined servlet variables.

We’ve described how the JSP is converted into a servlet—the HTML statements become println() calls. This all happens inside of an HttpServlet-like class, just like our BudgetProServlet extends HttpServlet in the previous chapter. In such a class, the method called when a request arrives from a browser looks very much like our doBoth() method:

doBoth(HttpServletRequest request, HttpServletResponse response)

Tip

If you want to see the source for the servlet that gets generated when a JSP is converted, and if you’re using NetBeans, right-click on the filename (in the Explorer view) and, from this menu, choose Compile. Then do it again and you’ll notice that the second choice on the menu is View Servlet (Figure 19.1).

Figure 19.1. Viewing the converted JSP in NetBeans

If you are using Apache Tomcat as your Web server, just look in the work subdirectory in the directory where Tomcat is installed. In the appropriate subdirectory you will find both the .java and .class files for your converted JSP with the .jsp suffix converted to $jsp.java and $jsp.class respectively. For example, BPAcct.jsp becomes BPAcct$jsp.java and is compiled into BPAcct$jsp.class.

The point here is that a request object and a response object are defined by the way the servlet is generated. They are called, oddly enough, request and response. In addition to these, a session is defined and initialized, just like we did in our servlet example. (What were we thinking?)

There are a few other variables that the converted servlet has created that we can use. We’ll summarize them in Table 19.1. To read more about how to use them, look up the Javadoc page for their class definition.

Remember that these can be used not only by the <%= %> expressions, but also by the <% %> snippets of code.

The last of the special delimiters that we will discuss is the one that doesn’t directly involve Java code. The <%@ ... %> delimiter encompasses a wide variety of directives to the JSP converter. We don’t have the space or the patience to cover them all, so we’ll cover the few that you are most likely to need early on in your use of JSP. We have some good JSP references at the end of this chapter for those who want all the gory details of this feature.

<%@page import="package.name.*" %>

is the way to provide Java import statements for your JSP. We bet you can guess what that happens in the generated servlet.

Here’s another useful page directive:

<%@page contentType="text/html" %>

You’ll see this as the opening line of our JSP, to identify the output MIME type for our servlet.

JSP also has an include directive:

<%@include file="relative path" %>

This directive is, for some applications, worth the price of admission alone. That is, it is such a useful feature that even if they use nothing else, they could use JSP just for this feature. It will include the named file when converting the JSP—that is, at compile time.

It can be used for common header and footer files for a family of Web pages. (If you’re a C programmer, think #include.) By defining one header file and then using this directive in each JSP, you could give all your JSP the same look—say, a corporate logo and title at the top of page and a standard copyright statement and hyperlink to your webmaster’s e-mail address at the bottom.

Be aware that this inclusion happens at compile time and is a source-level inclusion. That is, you are inserting additional source into the JSP, so if your included file contains snippets of Java code, they will be part of the resulting program. For example, you could define a variable in the included file and reference in the including file.

Also, since this inclusion happens at compile time, if you later change the included file, the change will not become visible until the JSP files that do the including are recompiled. (On Linux, this is simply a matter of touching all the JSP, as in:

$ touch *.jsp

assuming all your JSP files are in that directory. Touching them updates their time of last modification, so the Web server thinks they’ve been modified so the next access will cause them to be reconverted and their generated servlets reloaded. You get the idea.

There is another way to do an include in JSP—one that happens not at compile time, but at runtime. The syntax is different than the directives we’ve seen so far, but more on that in minute. First, an example of this kind of include:

<jsp:include page="URL" flush="true" />

In this format, the page specified by the URL (relative to this Web application’s root) is visited and its output is included in place amongst the output of this JSP, the one doing the include.

A few quick notes:

But what about that new syntax? It’s an XML-conformant syntax, and it’s the syntax for all the newer features added to JSP. In fact, even the old JSP syntax, the statements that we’ve discussed, have an alternative new syntax (Table 19.2). Prior to JSP 2.0, that syntax was reserved for JSP that produce XML rather than HTML. (That’s a whole other can of worms that we won’t open now.) Now, as of JSP 2.0, both forms can be used, if your Web server is JSP 2.0 compliant.

You can see that the old syntax is more compact and less distracting than the large tags. We suspect that means the old syntax is likely to continue to be used for a long time yet.^[1]

This new syntax is also used for the last two parts of JSP that we will cover, useBean and tag libraries.

For those who really want to avoid doing any Java coding inside of a JSP, there is additional syntax that will provide for a lot of capability but without having to explicitly write any Java statements. Instead, you write a lot of arcane JSP directives, as we’ll show you in just a bit. Is this any better? In some ways yes, but in other ways, no, it’s just different syntax.

What we’ll be able to do with this additional syntax is:

The syntax looks like this:

<jsp:useBean id="myvar" class="net.multitool.servlet.AccountBean" />

which will create a variable called myvar as an instance of the AccountBean class found in the net.multitool.servlet package. Think of this as:

<%! import net.multitool.servlet.AccountBean; %>

<% AccountBean myval = new AccountBean(); %>

So can AccountBean be any class? Well, sort of. It can be any class that you want, as long as it is a bean. It doesn’t have to end in “Bean”, but it does have to be a class which has:

Otherwise, its a normal class. These restrictions mean that you can call the class a “JavaBean” or just “bean,” and there is additional JSP syntax to manipulate the class. Specifically:

<jsp:getProperty name="myvar" property="total" />

will do, in effect, the following:

<%= myvar.getTotal() %>

or

<% out.print(myvar.getTotal()); %>

Similarly, we can set a value in the JSP with this syntax:

<jsp:setProperty name="myvar" property="total" value="1234" />

which will do, in effect, the following:

<% myvar.setTotal("1234"); %>

So this would hardly seem worth it, but there are other syntax constructs that make this much more powerful. Remember that we’re working with Web-based stuff, with a JSP that will be invoked via a URL. That URL may have parameters on it, and we can map those parameters onto a bean’s properties—that is, connect the parameters to setters for a given bean. We replace the value attribute with a parameter attribute, for example:

<jsp:setProperty name="myvar" property="total" parameter="newtot" />

which works the same as:

<% myvar.setTotal ( request.getParameter("newtot") ); %>

We can take that one step further and map all the parameters that arrive in the URL to setters in one step:

<jsp:setProperty name="myvar" parameter="*" />

So if you design your JSP and your HTML well, you can get a lot done automatically for you. One other thing going on behind the scenes that we’ve glossed over is the type of the argument to the setter. The parameters all come in as Strings. However, if your setter’s type is a Java primitive, it will automatically convert to that type for you, instead of just passing you Strings.

One final twist on using beans is the duration of the bean and its values. If you don’t specify otherwise (and we have yet to show you syntax to do otherwise) your bean will be around for the duration of the request, at which time it will be available to be garbage-collected. Any values in the bean will not be there on the next visit to that URL (i.e., the next call to that servlet).

Here is the syntax to make that bean last longer:

<jsp:useBean id="myvar" class="net.multitool.servlet.AccountBean"
             scope="session" />

which will make it stay for the duration of the session. You may remember (or you can flip back and look up) how we created and used session variables in the servlet. The same mechanism is at work here, but behind the scenes. You only use the specific syntax in the useBean tag, and it does the rest (getting and storing) for you.

Well, we’re almost done with JSP, but the one topic that we have yet to cover is huge. It’s the trap door, or the way back out, through which JSP can get to lots of other code without the JSP author having to write it. Tag libraries are specially packaged libraries of Java code that can be invoked from within the JSP. Just like the useBean, they can do a lot of work behind the scenes and just return the results.

There are lots of available libraries, which is one reason for this topic to be so huge. We could spend chapters just describing all the various database access routines, HTML generating routines, and so on available to you. Perhaps the leading tag library is the JSP Standard Tag Library (JSTL).

Here are two of the most common directives used with tag libraries. First is the directive that declares a library to be used:

<%@ taglib prefix="my" uri="http://java.sun.com/jstl/core" %>

You then use the prefix as part of the tag name on subsequent tags that refer to this library. For example, if we had an out directive in our library, we could use my as the prefix, separated by a colon: <my:out ...>.

The second directive we will show is a for loop. The for loop mechanism provided by this library is in some ways simpler than using Java scriptlets. It comes in many forms, including one for explicit numeric values:

<my:forEach var="i" begin="0" end="10" step="2">

This example will loop six times with i taking on the values 0, then 2, then 4, and so on. Another variation of the forEach loop can also make it easy to set up the looping values:

<my:forEach var="stripe" items="red,white,blue">

In this example it will parse the items string into three values: red, white, and blue, assigning each, in turn, to the variable stripe. In fact the items attribute can also store an array, or collection, or iterator from the Java code that you may have declared (or that is implicit from the underlying servlet). The forEach will iterate over those values without you having to code the explicit next() calls or index your way through an array.

The bottom of the loop is delimited by the closing tag:

</my:forEach>

For more information on these and other tags, check out

Beyond the standard library of tags, there are other third-party collections of tags; you can also create your own libraries, called custom tag libraries. While a useful and powerful thing to do if you have a large JSP-based application, such details would expand this book well beyond its scope. If you’re interested in this topic, please follow up with some of the excellent references at the end of this chapter.