Processing MediaWiki XML with STX/Add namespaces

From Meta, a Wikimedia project coordination wiki

This is a filter that only adds an attribute with the namespace id of each page. if there is defined

<namespace key="42">Foo</namespace>

then

<title>Foo:Bar</title>

will be transformed to

<title namespace="42">Foo:Bar</title>

Tested with joost alpha build 2005-05-21

[edit] STX script

<stx:transform version="1.0"
               xmlns:stx="http://stx.sourceforge.net/2002/ns"
               xmlns:m="http://www.mediawiki.org/xml/export-0.3/"               
               output-method="xml"
>

<!-- default rule: copy all nodes, attributes and text -->
<stx:template match="node()|@*" priority="-1">
  <stx:copy>
    <stx:process-attributes />
    <stx:process-children />
  </stx:copy>  
</stx:template>

<!-- store namespace nodes in a buffer -->
<stx:buffer name="namespaces"/>
<stx:template match="m:namespace">
  <stx:result-buffer name="namespaces" clear="no">
    <stx:copy>
      <stx:process-attributes/>
      <stx:value-of select="."/>
    </stx:copy>  
  </stx:result-buffer>
  <stx:copy>
    <stx:process-attributes/>
    <stx:value-of select="."/>
  </stx:copy>      
</stx:template>


<!-- test $prefix against all namespaces -->
<stx:variable name="page-namespace"/>
<stx:variable name="prefix" />

<stx:group name="match-namespace-prefix">
  <stx:template match="m:namespace">
    <stx:if test="$page-namespace = 0"> <!-- avoids tests if already found -->
      <stx:if test="$prefix = string(.)">
        <stx:assign name="page-namespace" select="@key"/>
      </stx:if>
    </stx:if>
  </stx:template>
  <stx:template match="text()" priority="1">_</stx:template>
</stx:group>

<!-- add a 'namespace'-attribute to title elements -->
<stx:template match="m:title">
  <stx:assign name="page-namespace" select="0"/>
  <stx:assign name="prefix" select="substring-before(.,':')"/>  
  <stx:if test="$prefix">  
    <stx:process-buffer name="namespaces" group="match-namespace-prefix"/>
  </stx:if>    
  <stx:copy>  
    <stx:attribute name="namespace" select="$page-namespace"/>
    <stx:process-attributes />
    <stx:process-children />
  </stx:copy>  
</stx:template>

</stx:transform>