There is no XML data in a CDATA section. CDATA means "character data" - it's only purpose is to tell the parser that anything inside it, even if it looks like XML, isn't actually XML but plain text.
If someone makes the design mistake of putting XML in a CDATA section (and it's often done), then the only thing you can do is extract the character data and supply it as input to an XML parser, which will interpret it as XML.
In Saxon the saxon:parse() extension function can be used to do this.
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference