14.2.1. Filtering XML Data

If you remember, the Foobar Public Library allowed books on several different subjects to be entered into its system; mytechbooks.com wants only the books about computer-related subjects. Fortunately, the library captured this information in the subject attribute of the book element for each book in its XML data. The first task is to filter out any book whose subject is not "Computers". Once the technical books have been obtained, they should be formatted into an HTML page that can be shown to customers visiting mytechbooks.com.

For this company and application, there is no static HTML, since the page showing new listings must be generated each time it is accessed. I'm going to use a servlet here for handling these responses. Although Apache Cocoon would be an excellent choice for converting the XML data from the library into an HTML response, mytechbooks.com is under tremendous time pressure to make these book listings available, and does not want to introduce such a large change into its system immediately; instead, it would prefer to use XML parsers and processors and then add Cocoon in as a second-phase addition. This means that you'll have to handle conversion from XML to HTML as well as the filtering of the data and the addition of other presentation-specific items, such as a company logo and menu bar.

While this doesn't yet filter the incoming XML data or transform that data, it does take care of the HTML interface for the user. Often it is much easier to take care of these presentation details first, and then add the transformation-specific logic afterwards.

NOTE: When developing XSL stylesheets, particularly for web applications, you should test the results with your XSLT Processor using its command-line capabilities. This can help you ensure that the stylesheet is transforming your document as you expect at each step of its development; trying to debug a large stylesheet's problems once it is complete is much more difficult. For this example, you could access the supplyBooks.pl script in a web browser, save the results to an XML file, and test that and the stylesheet as you follow the examples.

Before filtering the content, you need to add a template for outputting HTML content from a single book element's entry. As you recall, an entry looks like this:

<book subject="Computers">
 <title><![CDATA[Running Linux]]></title>
 <author><![CDATA[Matt Welsh]]></author>
 <publisher><![CDATA[O'Reilly & Associates]]></publisher>
 <numPages>729</numPages>
 <saleDetails>
  <isbn> 156592469X</isbn>
  <price>39.95</price>
 </saleDetails>
 <description><![CDATA[In the tradition of all O'Reilly books, Running 
  Linux features clear, step-by-step instructions that always seem to 
  provide just the right amount of information.]]></description>
</book>

<?xml version="1.0"?>

<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
                version="1.0"
>

 <xsl:template match="books">
   <!-- Presentation of User Interface -->
 </xsl:template>

 <xsl:template match="book">
  <table border="0" cellspacing="1" bgcolor="#000000">
   <tr>
    <td>
     <table border="0" cellpadding="3" cellspacing="0">
      <tr>
       <td width="100%" bgcolor="#00659C" nowrap="nowrap" align="center">
        <b><font color="#FFFFFF">
         <xsl:value-of select="title" />
        </font></b>
       </td>
      </tr>
      <tr>
       <td width="100%" align="center" nowrap="nowrap" bgcolor="#FFFFFF">
        <font color="#000000"><b>
         Author: <xsl:value-of select="author" /><br />
         Publisher: <xsl:value-of select="publisher" /><br />
         Pages: <xsl:value-of select="numPages" /><br />
         Price: <xsl:value-of select="saleDetails/price" /><br />
         <br />          
        </b></font>
        <xsl:element name="a">
         <xsl:attribute name="href">/servlets/BuyBookServlet?isbn=
          <xsl:value-of select="saleDetails/isbn" />
         </xsl:attribute>
         <font color="#00659C">Buy the Book!</font>
        </xsl:element>
       </td>
      </tr>
     </table>
    </td>
   </tr>
  </table>
  <br />
 </xsl:template>

</xsl:stylesheet>

This template matches the book element, and then creates a table with a heading in one row, and contents in the second row. The entire table is within another table with a black background, which results in the appearance of the table being surrounded by a beveled black border. The title is inserted into the header of the table, and the information about the book (author, publisher, pages, and price) is added to the content of the table. Finally, a link to a Java servlet, BuyBookServlet, is provided to allow easy access to purchasing the book. The value of the book's isbn element is supplied as an argument to this servlet, which enables it to load the book being purchased.

WARNING: In your XSL stylesheet, you should ensure that the line indicating the use of BuyBookServlet and the line with the xsl:value-of element selecting the book's ISBN number is actually one single line. If not, spaces and a carriage return could be inserted into the resultant URL, causing incorrect information to be passed to the servlet. The example stylesheet has this information broken into two lines because of the space constraints of the printed page.

The last addition you need to make to your stylesheet is to ensure that the new template is applied, and that only books with the subject "Computers" are passed to the new template. You can reference the value of the subject attribute with the @ symbol in your stylesheet, and filter the requests with the select attribute on the xsl:apply-templates element:

  </td>
  <td align="left">

   <!-- Handle creation of content for each new *computer* book -->
  <xsl:apply-templates select="book[@subject='Computers']" />         

  </td>
 </tr>
</table>

This references the value of the attribute and compares it to a literal, enclosed within single quotes because the entire XPath expression is enclosed within double quotes. Because you are accessing an attribute of a nested element, you'll need to reference the element by name, and surround the expression on the element's attribute with brackets. This will ensure that only books with a subject of "Computers" have templates applied, and are therefore included in the HTML output. Once the stylesheet is complete, it can be saved as computerBooks.xsl and referenced programmatically by a Java servlet, which I'll show you how to write next.

14.2.2. XSLT from a Servlet

There is currently no Java API that specifies how XSLT transformations can occur programmatically; however, each processor vendor should have classes that allow a transformation to be invoked from your Java code. I continue to look at using the Apache Xalan processor here; you should consult your processor's vendor for the method or methods to invoke in your own programs.

package com.techbooks;

import java.io.FileInputStream;
import java.io.InputStream;
import java.io.IOException;
import java.io.PrintWriter;
import java.net.URL;
import javax.servlet.*;
import javax.servlet.http.*;

// Import Xalan XSLT Processor components
import org.apache.xalan.xslt.XSLTInputSource;
import org.apache.xalan.xslt.XSLTProcessor;
import org.apache.xalan.xslt.XSLTProcessorFactory;
import org.apache.xalan.xslt.XSLTResultTarget;

public class ListBooksServlet extends HttpServlet {

    /** Host to connect to for books list */
    private static final String hostname = "newInstance.com";
    /** Port number to connect to for books list */
    private static final int portNumber = 80;
    /** File to request (URI path) for books list */
    private static final String file = "/cgi/supplyBooks.pl";
    
    /** Stylesheet to apply to XML */
    private static final String stylesheet = 
        "/home/bmclaugh/javaxml/techbooks/XSL/computerBooks.xsl";

    public void service(HttpServletRequest req, HttpServletResponse res) 
        throws ServletException, IOException {            
            
        res.setContentType("text/html");            
        
        // Connect and get XML listing of books
        URL getBooksURL = new URL("http", hostname, portNumber, file);
        InputStream in = getBooksURL.openStream();

        // Transform XML for InputStream into HTML output
        try {            
            XSLTProcessor processor = XSLTProcessorFactory.getProcessor();
            
            // Transform XML with XSL stylesheet
            processor.process(new XSLTInputSource(in),
                              new XSLTInputSource(
                                  new FileInputStream(stylesheet)),
                              new XSLTResultTarget(
                                  res.getOutputStream()));
            
        } catch (Exception e) {
            PrintWriter out = res.getWriter();            
            out.println("Error: " + e.getMessage());
            out.close();
        }
    }    
}

NOTE: I could have also used JAXP 1.1's TrAX API for performing this translation. However, JAXP 1.1 is still pretty new as of this writing, and I see few people adopting it (yet). Additionally, most servlet engines, especially Tomcat, still ship with JAXP 1.0, and many are relying on that default behavior rather than supplying a newer version of JAXP.

Figure 14-4. HTML output from mytechbooks.com new book listing