Templates as Mapping Rules

We've been a little formulaic in our conversion from the simplified stylesheet to the full stylesheet that we have now – we take every <xsl:for-each> and <xsl:value-of> and turn it into an <xsl:apply-templates>. The introduction of templates has really just been a way of modularizing the code.

But there's another way of thinking about templates that can make your stylesheets more elegant and more extensible, and that's to consider them as a mapping rule. Each template describes how to map a particular node in the source XML on to a result that you're after. In this section, we'll first look at using templates with mixed content, and at how using templates as mapping rules is particularly useful when processing document-oriented XML. Then, we'll go on to look at how we could apply the same kind of technique to our stylesheet and the impact of doing so.

Processing Document-Oriented XML

Let's first look at the problem of generating output from document-oriented XML. Document-oriented XML arises when you take a paragraph of text and mark up particular words and phrases within that paragraph, giving mixed content – elements and text intermingled. This is in contrast to data-oriented XML, which is concerned with storing data and usually results in element-only and text-only content.

Take another look at the XML structure that we're using to store information about TV programs:

While most of the XML structure is oriented around providing data, the description of the TV program is more document-oriented. You could imagine wanting to add a bit of document-oriented markup to the <Description> element, perhaps a link to the EastEnders biography for Jamie and highlights around the names of the characters:

The <Description> element now holds mixed content. Looking at the node tree representation of that piece of XML makes this clearer:

We want the elements that we use in the description to be transformed into HTML elements instead. In the HTML version of the page, we want to use <span> elements around the character names, and transform the <Link> elements into <a> elements, to give:

If you imagine processing the content of the <Description> element with <xsl:for-each>, you'll see that you run into problems. We haven't looked at how to yet, but it's possible to iterate over all the nodes that are children of the <Description> element, and test what kind of node they are to decide what to do with them. But even if you did that, you still need to take account of nested elements, so you'd get very long and very deep conditional processing to cover all levels of nesting.

However, with templates it's a lot easier. You can create a template for each kind of element that you know can occur in the content of the <Description> element, which describes how to map between that element and the result that you want. In this example, there are two elements, <Character> and <Link>, so you need a template for each. Within the <span> and <a> elements that these templates create, you apply templates to the content of the <Character> or <Link> element to account for possible nested elements:

Whenever you add a new type of element that you might include in the description, you can add a new template that describes how to map that on to HTML.

We can add support for lots of different elements that we want to be able to use within the <Description> element. Highlighting character names and providing links to other web sites is useful, but you might also want to add elements for emphasis, foreign words, names of directors, series, channels, films, and so on – different elements for the different types of words and phrases that can appear in descriptions of TV programs and series.

We'll add just a few of these elements in TVGuide2.xml, to create TVGuide3.xml, which looks
as follows:

Try using TVGuide6.xsl with TVGuide3.xml, which uses <xsl:apply-templates> to apply templates to the <Description> element. There aren't any templates for <Character> or <Link> elements, so the built-in templates are used instead. The result of the transformation of the <Description> element looks just the same as before, because the built-in templates automatically show any text within an element.

We want a new version of the stylesheet (TVGuide8.xsl), which generates HTML where the words and phrases in the description that we've picked out with <Character> and <Link> elements are displayed and behave slightly differently from the rest of the text. Links should be links, for example, and character names should be slightly larger than the surrounding text.

To make the marked-up text display and act differently, we need to introduce templates for these new elements: one for the <Link> element, to create a hypertext link with an HTML <a> element:

and one for the <Character> element, to create a <span> element with a class of character around the character names:

To make the character names slightly bigger, we'll add a rule to TVGuide.css, to create TVGuide2.css, which contains:

Putting the final touches on TVGuide8.xsl, we need the HTML that it creates to point to TVGuide2.css rather than TVGuide.css, so the <link> element generated in the template matching the root node needs to be altered slightly:

Having made this final change, transform TVGuide3.xml with TVGuide8.xsl. The result of the transformation should look something like the following:

The character names are slightly larger than the rest of the text, and clicking on the word "Jamie" takes you to the EastEnders site with Jamie Mitchell's biography. Feel free to add your own elements to that description, and add your own templates matching them to present the document-oriented XML.

Part 1 : XSLT Templates XSLT Part 2 : Defining Templates XSLT Part 3 : The Node Tree XSLT Part 4 : XSLT Processing Model XSLT Part 5 : The Built-in Templates XSLT Part 6 : Templates as Mapping Rules XSLT Part 7 : Context-Dependent Processing XSLT Part 8 : Resolving Conflicts Between Templates XSLT Part 9 : Processing with Push and Pull XSLT Part 10 : Using Moded Templates XSLT Part 11 : Summary Buy Beginning XSLT here © Copyright 2002 Wrox Press This chapter is written by Jeni Tennison and taken from "Beginning XSLT" published by Wrox Press Limited in June 2002; ISBN 1861005946; copyright © Wrox Press Limited 2002; all rights reserved. No part of these chapters may be reproduced, stored in a retrieval system or transmitted in any form or by any means -- electronic, electrostatic, mechanical, photocopying, recording or otherwise -- without the prior written permission of the publisher, except in the case of brief quotations embodied in critical articles or reviews.