How to deal with presence or not of xml namespaces using xslt.

Question

I have some XML/TEI documents, and i'm writing an XSLT 2.0 to extract their content.

Almost all TEI documents has no namespace, but one has the default namespace (xmlns="http://www.tei-c.org/ns/1.0"). So all documents has the same aspect, with unqulified tags like <TEI> or <teiHeader>, but if I try to extract the content, all works with "non-namespaced-documents", but nothing (of course) is extracted from the namespaced-document.

So i used the attribute xpath-default-namespace="http://www.tei-c.org/ns/1.0" and now (of course) the only document working is the namespaced one.

I can't edit documents at all, so what I'm asking is if there's a way to change dynamically the xpath-default-namespace in order to make work xpaths like //teiHeader both with namespaced and non-namespaced documents

Answer 1

There isn't really a clean way to do precisely what you are asking. However, there are workarounds available. You could use a two stage process whereby you strip the namespace from the document if it's present and then pass it through the same templates for all content.

There is a good example (in XSLT 1) of doing this in the DocBook XSLT. Take a look at html/docbook.xsl and common/stripns.xsl

Basically, you would need to assign the result of stripping the namespace to a variable and then call your existing templates (for the non namespaced) content but select the variable.

Answer 2

It is ugly, but this gives you what you want:

//*[name()='teiHeader']

If you use this style for all location steps in any XPath expression, the XPath expressions will select elements only by name, regardless whether or not the elements belong to any namespace.