Remove nodes on missing child reference

I have (thanks to SO!!) already some XSLT to "filter and remove / xsl:copy" using clearly defined conditions, but now I'm facing a kind of "Wildcards / search everything on everywhere", and I want to ask on an abstract example for the basic way to start.

Fake Example: I have hundreds of factories, producing thousands of products. The unmodified XML looks like this:

<data ID="1">
    <type>factory.tires</type>
    <Country>a</Country>
</data>

<data ID="2">
    <type>factory.glass</type>
    <Country>a</Country>
</data>

<data ID="3">
    <type>factory.steel</type>
    <Country>b</Country>
</data>

<data ID="4">
    <type>factory.tires</type>
    <Country>c</Country>
</data>

<!---- xxxxxxxxxxxxxx -->

<data ID="11">
    <type>product.screws</type>
    <factory REF="3" />
</data>

<data ID="12">
    <type>product.screws</type>
    <factory REF="3" />
</data>

<data ID="13">
    <type>product.cups</type>
    <factory REF="2" />
</data>

<data ID="20">
    <type>product.wheel</type>
    <factory REF="1" />
</data>

<data ID="21">
    <type>product.wheel</type>
    <factory REF="4" />
</data>

Now we closed factories 1 and 2, the data of the factories is already removed:

<data ID="3">
    <type>factory.steel</type>
    <Country>b</Country>
</data>

<data ID="4">
    <type>factory.tires</type>
    <Country>c</Country>
</data>

<!---- xxxxxxxxxxxxxx -->

<data ID="11">
    <type>product.screws</type>
    <factory REF="3" />
</data>

<data ID="12">
    <type>product.screws</type>
    <factory REF="3" />
</data>

<data ID="13">
    <type>product.cups</type>
    <factory REF="2" />
</data>

<data ID="20">
    <type>product.wheel</type>
    <factory REF="1" />
</data>

<data ID="21">
    <type>product.wheel</type>
    <factory REF="4" />
</data>

Challenge: All products which where made in the closed factories have to be removed. But I don't know which factories are closed - dozens ... So the idea is "

search every node with child "factory"
search if the ID of the REF still exists somewhere (the ID is unique)
if you don't find the ID (the factory is closed and removed), then remove the node (the product)

This should be the result - IDs 13 and 20 are removed:

<data ID="3">
    <type>factory.steel</type>
    <Country>b</Country>
</data>

<data ID="4">
    <type>factory.tires</type>
    <Country>c</Country>
</data>

<!---- xxxxxxxxxxxxxx -->

<data ID="11">
    <type>product.screws</type>
    <factory REF="3" />
</data>

<data ID="12">
    <type>product.screws</type>
    <factory REF="3" />
</data>


<data ID="21">
    <type>product.wheel</type>
    <factory REF="4" />
</data>

Any help will be appreciated very much! Thanks in advance.

Solution

If I understand your question correctly, you want to do:

XSLT 1.0

<xsl:stylesheet version="1.0" 
xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<xsl:output method="xml" version="1.0" encoding="UTF-8" indent="yes"/>
<xsl:strip-space elements="*"/>

<xsl:key name="data-by-ID" match="data" use="@ID" />

<!-- identity transform -->
<xsl:template match="@*|node()">
    <xsl:copy>
        <xsl:apply-templates select="@*|node()"/>
    </xsl:copy>
</xsl:template>

<!-- remove products that do not have a related factory -->
<xsl:template match="data[factory][not(key('data-by-ID', factory/@REF))]"/>

</xsl:stylesheet>

This is assuming that every data element that represents a product has a factory child.

If every data element that represents a factory has a Country child, you could make this more efficient by defining the key as:

<xsl:key name="data-by-ID" match="data[Country]" use="@ID" />