XSLT 2.0 processing invalid node mixing text and cdata - xslt

Question

I need to parse the following node:
<media:keywords>keyword1,keyword2<![CDATA[keyword3]]></media:keywords>
into a valid string, preferably "keyword1,keyword2,keyword3" but I would settle for removing the cdata completely.
Trying to access the node gives me the text "keyword1,keyword2keyword3" and I can't tell where the CDATA begins.
original xml (simplified version of mRSS feed)
<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:media="http://search.yahoo.com/mrss/">
<channel>
<item>
<media:keywords>keyword1,keyword2<![CDATA[keyword3]]></media:keywords>
</item>
</channel>
</rss>
xsl (simplified):
<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:fn="http://www.w3.org/2005/xpath-functions" xmlns:media="http://search.yahoo.com/mrss/" exclude-result-prefixes="xs xsi fn">
<xsl:output method="xml" encoding="UTF-8" omit-xml-declaration="yes"/>
<xsl:template match="/">
<test>
<xsl:variable name="items" select="/rss/channel/item"/>
<xsl:for-each select="$items">
<xsl:variable name="mediakw" select="media:keywords"/>
<xsl:element name="mediaKeyWords">
<xsl:value-of select="$mediakw"/>
</xsl:element>
</xsl:for-each>
</test>
</xsl:template>
</xsl:stylesheet>
and the output:
<test xmlns:media="http://search.yahoo.com/mrss/"><mediaKeyWords>keyword1,keyword2keyword3</mediaKeyWords></test>
Thanks a lot!

Answer 1

XML and XSLT cannot help you here.
XSLT uses the INFOSET model in which there isn't anything as a "CDATA node" and there is just a single text() node:
"keyword1,keyword2keyword3"
The XML document needs to be corrected and a comma be inserted between the substrings "keyword2" and "keyword3"
One solution would be to process the CDATA DOM node using DOM, and only then initiate the XSLT transformation.

Answer 2

By the time the XSLT processor sees the text, the CDATA is gone. You cannot see the incoming CDATA, and have very little control over how output CDATA is generated (all or nothing for a given tag).

Answer 3

Can't be done in standard XSLT.
The input XML you're receiving,
<media:keywords>keyword1,keyword2<![CDATA[keyword3]]></media:keywords>
is indistinguishable (to XSLT) from
<media:keywords>keyword1,keyword2keyword3</media:keywords>
because the CDATA markup is just a way of escaping the data inside it. There is really no special markup to escape in this case, so the CDATA happens to be a no-op. But XSLT has no way of knowing what data was originally expressed using CDATA, what was expressed using character entities, etc.
The solution would be to tell whoever is providing this XML that they need to put a delimiter between keyword2 and keyword3.