XSLT Constructors

XSLT Constructors Normally, to create a result element in your output, you just type it literally:

[...]

However, there are some circumstances when you can't do that. For instance, the name of the element you want to create may change based on the XML input. For instance: If you're processing a TEI list element, you may need to produce either ul or ol in your output, depending on whether it's a bulleted or numbered list: <list rend="bulleted"> → <ul> and <list rend="numbered"> → <ol> We can handle this with the XSLT element constructor.

The element constructor [...contents of the h2 element...]

An example element constructor [...] The XPath logic is contained in {curly braces}. Be careful with nested quotes!

The attribute constructor

Just as you can create an element with a constructor, you can also create an attribute: [... value of the class attribute ...]

You might do this if you need to generate the attribute value dynamically based on the content.

Like xsl:variable, the value can be given as the value of a select attribute, or as the content.

An example attribute constructor

XSL constructors: Task 1

Write a constructor for an element called div with an attribute called "class", whose value is "chapter".

XSL constructors: Task 1 answer

A constructor for an element div with a class attribute whose value is chapter:

chapter

XSL constructors: Task 1 discussion

A constructor for an element div with a class attribute whose value is chapter:

[What can go in this location?] chapter [What can go in this location?]

XSL constructors: Task 1 discussion

A constructor for an element div with a class attribute whose value is chapter:

[Only attribute constructors can go here.] chapter [Lots of things can go here, including <xsl:apply-templates>, other attribute constructors, and other element constructors.]

Copy constructors: xsl:copy and xsl:copy-of Sometimes, you simply want to copy some of the XML in your tree to the output document. This is often the case when you're using XSLT to make small modifications to your XML files. You can use xsl:copy and xsl:copy-of to do this. Although they look similar, they have completely different effects.

xsl:copy-of xsl:copy-of is the simpler of the two copy constructors. It makes a complete copy of the node, including its attributes and descendants, in the output tree. For instance, if your input is this:

This is my first paragraph.

and your template is this:

then your output will be this:

This is my first paragraph.

xsl:copy xsl:copy copies only the node itself, and not its descendants or attributes to the output tree. For instance, if your input is this:

This is my second paragraph.

and your template is this: then your output will be this:

Only the node (p) is copied. Its xml:id attribute and text content are ignored. However, xsl:copy has content that is also processed and put into the output tree. For instance, if your input is the same and your template is this: Hi Martin! I just messed up your paragraph. :-) then your output will be this:

Hi Martin! I just messed up your paragraph. :-)

Why copy constructors are useful: the identity transform You're probably wondering why copy constructors would be useful. After all, usually we're converting our XML into something else such as XHTML, so we can't just copy TEI nodes to the output document. However, we also use XSLT to edit, amend or improve existing TEI documents, transforming TEI into TEI, with minor changes. This is done through the so-called identity transform(ation). One version of it looks like this:

Can you figure out what this is doing?

An example use of the identity transform Imagine you have an old TEI file which has a lot of div0 and div1 elements in it: Title...

Some stuff...

[...]

Some more stuff...

[...] You have, quite sensibly, decided that both div0 and div1 should be turned into plain div elements. We can do this with an identity transform.

Changing div0 and div1 to div with an identity transform

This is the identity transform we need:

XSL constructors: Task 2 When would I ever actually use this? We've already come across a perfect use-case in the Hamlet transformation:

This matches head elements for both Act headings and Scene headings. What's the problem with it? Open hamlet.xml and the example_01.xsl file. Set up a transformation scenario (elicit the steps). Do the transform, and show the problem. Discuss possible solutions. Arrive at the counting of ancestor divs. Elicit the initial fix, which would use h{count(ancestor::div)+1}. Once that's working, elicit the problem with the general case (h7 etc.). Elicit the solution to it (h{min(count(ancestor::div)+1), 6)}.