Parse CDATA value with xslt 1.0

Question

I'd like to transform some xml files using xslt, but encountered a big problem.

In my input files i have CDATA tags, which contain xml subtree-s, and should be copied to the output tag. We know some informations about these CDATA tags:

They have a specific name: SUBXML1, SUBXML2, SUBXML3, and so on and so forth. Their content is fixed: some of there SUBXMLn-s contain only one element (<a>...</a>), some of them contain a list of elements (<a>...</a><a>...</a>), or multiple and/or list of elements (<a>...</a><b>...</b><b>...</b>).

I would like to transform them using IBM Message Broker (xsl transform node, we use it for like a hundred types messages, but didn't have this problem so far). We have a strict rule for these transformations: we can't use anything else, we have to achieve out goal with XLS; we can use only xslt 1.0 (broker doesn't support 2.0 or 3.0); we can't change the structure of the input/output messages (lots of systems use the defined structures so we can't touch them).

So here's the thing:

I have an input XML document:

<ns1:myOperation xmlns:ns1="http://mynamespace/">
    <SUBXML1>
        <![CDATA[<metaData><value>value1</value></metaData><metaData><value>value2</value></metaData><notMetaData><notValue>value3</notValue></notMetaData>]]>
    </SUBXML1>
</ns1:myOperation>

And i have my xsl like this:

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet version="1.0"
    xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:ns1="http://mynamespace/">
    <xsl:output method="xml" version="1.0" encoding="UTF-8"
        indent="yes" omit-xml-declaration="yes" />

    <xsl:template match="/">
        <xsl:apply-templates select="ns1:myOperation" />
    </xsl:template>

    <xsl:template match="ns1:myOperation">
        <ns1:myOperation>
            <xsl:apply-templates select="SUBXML1" />
        </ns1:myOperation>
    </xsl:template>

    <xsl:template match="SUBXML1">
        <xsl:element name="one">
            <xsl:element name="metaData">
                <xsl:value-of
                    select="substring-before(substring-after(.,'metaData&gt;'), '&lt;/metaData')" disable-output-escaping="yes" />
            </xsl:element>
        </xsl:element>
        <xsl:element name="all">
            <xsl:element name="allSUBXML1">
                <xsl:value-of
                    select="." disable-output-escaping="yes" />
            </xsl:element>
        </xsl:element>
    </xsl:template>
</xsl:stylesheet>

The output of this will be:

<ns1:myOperation xmlns:ns1="http://mynamespace/">
    <one>
        <metaData>
            <value>value1</value>
        </metaData>
    </one>
    <all>
        <allSUBXML1 />
    </all>
</ns1:myOperation>

Note: if you're testing this example in eclipse it will work (the <allSUBXML1> wont be empty), BUT xsl:value-of select should work only in elements! Thus in message broker it will generate an emtpy <allSUBXML1 /> element, but in eclipse somehow (a bug maybe?) it'll generate a "full output" so, it'll populate that field with the SUBXML1 content.

Back to my issue: the transformed xml has nothing in the <allSUBXML1 /> node, because i have no elements (only a big text) under <SUBXML1>...</SUBXML1> in my input xml.

How can i achieve my goal (to copy everything from SUBXML1 to my output's desired element)?

I can't use value-of select, because there are no elements under SUBXML1, only a big string. I can't use copy-of, because if won't convert the CDATA properly (i don't want to see "<" and ">" in my output, it should be a valid XML subtree).

As you can see i've tried to do some tricks using substring-before(substrin-after(.,...), ...) to get one inner element from SUBXML1, and it works, but that's not enough.

Is there any tricky way to copy everything from <SUBXML1> to <allSUBXML1> (so all inner elements)?

Thanks in forward, Tamas

Answer 1

Are you dead set on using XSLT? You could use the parse as bitstream clause in ESQL to do this pretty easily I would think.