Template Rules - Page 5

An XSLT stylesheet has a collection of template rules. Each template rule has a pattern that identifies the source tree nodes to which the pattern applies and a template that is added to the result tree when the XSLT processor applies that template rule to a matched node.

In a stylesheet document, the template rules that comprise a stylesheet are represented as xsl:template elements; the stylesheet above has three. The value of each xsl:template element’s match attribute is the pattern that gets matched against source tree nodes. The element’s content—that is, everything between its start- and end- tag— is the template that gets added to the result tree for each source tree node that corresponds to the match pattern. An xsl:template element essentially tells the XSLT processor, "as you go through the source tree, when you find a node of that tree whose name matches the value of my match attribute, add my contents to the result tree."

Figure 1.5 The two parts of a template rule

For example, the first template rule in the preceding stylesheet tells the XSLT processor what to do when it sees a year element node as the child of another node in the source tree. (The "year" attribute value is actually an abbreviation of "child::year.") The template rule has one element as its template to add to the result tree: a vintage element. This element contains an xsl:apply-templates element that tells the processor to apply any relevant templates to the children of the matched element node (in this case, year). The ultimate result of this template is the contents of the input year element surrounded by vintage tags—in effect, renaming the source tree’s year element to a vintage element for the result tree. Figure 1.5 shows where the pattern and template are in one example of a template rule.

The specialized elements in a template from the XSLT namespace are sometimes called "instructions," because they are instructions to the XSLT processor to add something to the result tree. What does this make the elements in the template that don’t use the "xsl" namespace prefix, such as the vintage element? The stylesheet is a legal, well-formed XML document, and the vintage element is an element in that stylesheet. Because this element is not from the XSLT namespace, the XSLT processor will pass it along just as it is to the result tree. In XSLT, this is known as a "literal result element."

Like all template rules, the second xsl:template rule in the stylesheet on page 9 tells the XSLT processor "if you find a source tree node whose name matches the value of my match attribute, add my contents to the result tree." The string "price" is the pattern to match, but what are the template’s contents? There are no contents; it’s an empty element. So, when the XSLT processor sees a price element in the source tree, the processor will add nothing to the result tree—in effect, deleting the price element.

Because the stylesheet is an XML document, the template rule would have the same effect if it were written as a single-tag empty element, like this:

<xsl:template match="price"/>

XSLT has other ways to delete elements when copying a source tree to a result tree, but a template rule with no template is the simplest.

Unlike the first two template rules, the third one is not aimed at one specific element type. It has a more complex match pattern that uses some XPath abbreviations to make it a bit cryptic but powerful. The pattern matches any element, attribute, or text node, and the xsl:copy and xsl:apply-templates elements copy any element, attribute, or text node children of the selected nodes to the result tree. Actually, the pattern doesn’t match any element—an XSLT processor uses the most specific template it can find to process each node of the source tree, so it will process any year and price elements using the stylesheet’s templates designed to match those specific tree nodes. Because the processor will look for the most specific template it can find, it doesn’t matter whether the applicable template is at the beginning of the stylesheet or at the end—the order of the templates in a stylesheet means nothing to an XSLT processor.

Tip: If more than one xsl:template template rule is tied for being most appropriate for a particular source tree node, the XSLT processor may output an error message or it may just apply the last one to the node and continue.

The values of all of the xsl:template elements’ match attributes are considered "patterns." Patterns are like XPath expressions that limit you to using the child and attribute axes, which still gives you a lot of power. (see chapter 2, "XPath," on page 23, for more on axes and the abbreviations used in XPath expressions and patterns.) The "year" and "price" strings are match patterns just as much as "*|@*|text()" is, even though they don’t take advantage of any abbreviations or function calls.

That’s the whole stylesheet. It copies a source tree to a result tree, deleting the price elements and renaming year elements to vintage elements. For example, the stylesheet turns this wine element

<wine grape="chardonnay">
<product>Carneros</product>
<year>1997</year>
<price>10.99</price>
</wine>

into this:

<?xml version="1.0" encoding="utf-8"?>
<wine grape="chardonnay">
<product>Carneros</product>
<vintage>1997</vintage>
</wine>

Although the price element was deleted, the carriage returns before and after it were not, which is why the output has a blank line where the price element had been in the input. This won’t make a difference to any XML parser.

This is not an oversimplified example. Developers often use XSLT to copy a doct with a few small changes such as the renaming of elements or the deletion of information that shouldn’t be available at the document’s final destination.

A Simple XSLT Stylesheet - Page 4
XSLT Quickly
Running an XSLT processor - Page 6