Can't be done in XSLT: CDATA is not considered to be information-bearing in
the XPath data model. In other words,

<![CDATA[xyz]]>

is exactly the same content as

xyz

just as a="3" and a='3' are considered simply as two different ways of
writing the same thing.

If you have used CDATA in your source document to convey information that
needs to be retained, consider a textual transformation (e.g. in Perl) to
replace the CDATA delimiters by element start and end tags.

Michael Kay
http://www.saxonica.com/ 



> -----Original Message-----
> From: dalloro@xxxxxxxxx [mailto:dalloro@xxxxxxxxx] 
> Sent: 01 April 2005 10:03
> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
> Subject: [xsl] (newby) extraction of CDATA nodes to different 
> XML files
> 
> Hi,
> 
> I need to extract the contents of three particular nodes in an XML
> file to three separate XMLs.
> The elements have the strange definition shown below:
> <![CDATA[   .....snip.....   ]]>
> 
> Whenever I encounter that type of construct I would like to extract
> the data that it contains to a separate XML.
> 
> Can you give some hints to solve this task?
> 
> Thank you very much.
> Best regards.
> 
> Livio