User:Dr pda/extractRefCount.stx

<?xml version="1.0"?> <stx:transform version="1.0"

              xmlns:stx="http://stx.sourceforge.net/2002/ns"
              xmlns:m="http://www.mediawiki.org/xml/export-0.3/"
              pass-through="none"
              output-method="xml"
              exclude-result-prefixes="#all"

>

<stx:variable name="namespace-prefixes"/>

<stx:template match="m:namespace">

 <stx:if test="@key!=0">
   <stx:assign name="namespace-prefixes" select="($namespace-prefixes, .)"/>
 </stx:if>

</stx:template>

<stx:template match="/m:mediawiki">

 <mediawiki>
   <stx:process-children />
 </mediawiki>    

</stx:template>

<stx:template match="m:siteinfo">

 <stx:process-children />

</stx:template>

<stx:template match="m:namespaces">

 <stx:process-children />

</stx:template>

<stx:variable name="page-title"/> <stx:variable name="page-text"/> <stx:variable name="page-id"/>

<stx:template match="m:title">

 <stx:assign name="page-title" select="string(.)"/>    

</stx:template>

<stx:template match="m:text">

 <stx:assign name="page-text" select="string(.)"/>

</stx:template>

<stx:variable name="first-revision" select="true()"/>

<stx:template match="m:revision">

 <stx:if test="$first-revision">
   <stx:assign name="first-revision" select="false()"/>  
   <stx:process-children/>
 </stx:if>  

</stx:template>

<stx:buffer name="parsed"/> <stx:variable name="pd-count" select="0"/> <stx:variable name="pnd-count" select="0"/> <stx:variable name="found-something"/>

<stx:template match="m:id">

 <stx:if test="$first-revision">
   <stx:assign name="page-id" select="normalize-space(.)"/>
 </stx:if>  

</stx:template>

<stx:template match="m:page">

 <stx:assign name="first-revision" select="true()"/>
 <stx:process-children />
 
 <stx:variable name="prefix" select="substring-before($page-title,':')"/>
 <stx:variable name="skip" select="false()"/>
 <stx:if test="$prefix">
   <stx:value-of select="$prefix"/>    
   <stx:for-each-item name="p" select="$namespace-prefixes">
     <stx:if test="string($p) = string($prefix)">
       <stx:assign name="skip" select="true()"/> 
     </stx:if>
   </stx:for-each-item>
 </stx:if>  
 <stx:if test="not($skip)">
   <stx:assign name="found-something" select="false()"/>
   <stx:assign name="pnd-count" select="0"/>
   <stx:assign name="pd-count" select="$pd-count+1"/>
   <stx:result-buffer name="parsed" clear="yes">  
     <stx:variable name="text" select="$page-text"/>
     <stx:if test="not(starts-with($text,'#REDIRECT') or starts-with($text,'#Redirect') or starts-with($text,'#redirect'))">
       <stx:while test="string-length($text) > 0">
         <stx:variable name="before" select="substring-before($text,'<')"/>
         <stx:assign name="text" select="substring-after($text,'<')"/>
         <stx:call-procedure name="template">
           <stx:with-param name="content" select="substring-before($text,'>')"/>
         </stx:call-procedure>
         <stx:assign name="text" select="substring-after($text,'>')"/>
       </stx:while>
       <template name="PND">
         <param>
           <stx:value-of select="normalize-space($pnd-count)"/>
         </param>
       </template>
     </stx:if>  
   </stx:result-buffer>  
     <stx:message>
       <stx:value-of select="$pd-count"/>
       <stx:text>/</stx:text>
       <stx:value-of select="$pnd-count"/>
     </stx:message>
     <stx:text>
</stx:text> 
 <page>
   <title><stx:value-of select="$page-title"/></title>
   <id><stx:value-of select="$page-id"/></id>
   <stx:text>
</stx:text> 
   <revision>
     <parsed>  
       <stx:process-buffer name="parsed" group="copy"/>
     </parsed>
     <stx:text>
</stx:text> 
   </revision>
 </page>
     <stx:text>
</stx:text> 
 </stx:if>  

</stx:template>

<stx:group name="copy">

 <stx:template match="*">
   <stx:element name="{name(.)}">
     <stx:process-attributes/>
     <stx:process-children/>
   </stx:element>
 </stx:template>
 <stx:template match="@*">
   <stx:attribute name="{name(.)}" select="."/>
 </stx:template>
 <stx:template match="text()">
   <stx:value-of select="."/>
 </stx:template>

</stx:group>

<stx:procedure name="template">

 <stx:param name="content" required="yes"/>
 <stx:if test="starts-with($content,'ref')">
   <stx:if test="not(starts-with($content,'references'))">
     <stx:assign name="pnd-count" select="$pnd-count+1"/>
     <stx:assign name="found-something" select="true()"/>
   </stx:if>
 </stx:if>

</stx:procedure>

</stx:transform>