[xsl] Performance Tuning

[Bhupendra Singh <skbhupendra@xxxxxxxxxxx>]

Hi,
I have an XML something like this.
<Document>
	- <ParentGroup> Node1 </ParentGroup>
	- <ChildGroup> Node1.1 </ChildGroup>
	- <Child Node1>1.1.1 </Child Node1>
	- <Child Node2>1.1.2 </Child Node2>
	- <Child Node3>1.1.3 </Child Node3>	
	- <ParentGroup> Node1 </ParentGroup>
	- <ChildGroup> Node1.2 </ChildGroup>
	- <Child Node1>1.2.1 </Child Node1>
	- <Child Node2>1.2.2 </Child Node2>
	- <Child Node3>1.2.3 </Child Node3>	
	- <ParentGroup> Node1 </ParentGroup>
	- <ChildGroup> Node1.3 </ChildGroup>
	- <Child Node1>1.3.1 </Child Node1>
	- <Child Node2>1.3.2 </Child Node2>
	- <Child Node3>1.3.3 </Child Node3>	
	- <ParentGroup> Node1 </ParentGroup>
	- <ChildGroup> Node1.3 </ChildGroup>
	- <Child Node1>1.3.x </Child Node1>
	- <Child Node2>1.3.y </Child Node2>
	- <Child Node3>1.3.z </Child Node3>	

	- <ParentGroup> Node2 </ParentGroup>
	- <ChildGroup> Node2.1 </ChildGroup>
	- <Child Node1>2.1.1 </Child Node1>
	- <Child Node2>2.1.2 </Child Node2>
	- <Child Node3>2.1.3 </Child Node3>	
	- <ParentGroup> Node2 </ParentGroup>
	- <ChildGroup> Node2.2 </ChildGroup>
	- <Child Node1>2.2.1 </Child Node1>
	- <Child Node2>2.2.2 </Child Node2>
	- <Child Node3>2.2.3 </Child Node3>	
	- <ParentGroup> Node2 </ParentGroup>
	- <ChildGroup> Node2.3 </ChildGroup>
	- <Child Node1>2.3.1 </Child Node1>
	- <Child Node2>2.3.2 </Child Node2>
	- <Child Node3>2.3.3 </Child Node3>	
...........
</Document>

Please note that all the nodes above are at the same
depth. I want to transform the above XML into depth
wise soemthing like this:

<Document>
  <ParentGroup>	
  <PRIMARY_CUST_DESC>Node1</PRIMARY_CUST_DESC>
     <ChildGroup Node1.1 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>1.1.1 </Child Node1>
	- <Child Node2>1.1.2 </Child Node2>
	- <Child Node3>1.1.3 </Child Node3>
	<ACCOUNT_SUMMARY>
     </ChildGroup>
     <ChildGroup Node1.2 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>1.2.1 </Child Node1>
	- <Child Node2>1.2.2 </Child Node2>
	- <Child Node3>1.2.3 </Child Node3>
	</ACCOUNT_SUMMARY>
     </ChildGroup>
     <ChildGroup Node1.3 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>1.3.1 </Child Node1>
	- <Child Node2>1.3.2 </Child Node2>
	- <Child Node3>1.3.3 </Child Node3>
	</ACCOUNT_SUMMARY>
     </ChildGroup>
     <ChildGroup Node1.3 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>1.3.x </Child Node1>
	- <Child Node2>1.3.y </Child Node2>
	- <Child Node3>1.3.z </Child Node3>
	</ACCOUNT_SUMMARY>
     </ChildGroup>
  </ParentGroup>	
  <ParentGroup>	
  <PRIMARY_CUST_DESC>Node2</PRIMARY_CUST_DESC>
    <ChildGroup Node2.1 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>2.1.1 </Child Node1>
	- <Child Node2>2.1.2 </Child Node2>
	- <Child Node3>2.1.3 </Child Node3>
	</ACCOUNT_SUMMARY>
    </ChildGroup>
    <ChildGroup Node2.2 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>2.2.1 </Child Node1>
	- <Child Node2>2.2.2 </Child Node2>
	- <Child Node3>2.2.3 </Child Node3>
	</ACCOUNT_SUMMARY>
    </ChildGroup>
    <ChildGroup Node2.3 >
	<ACCOUNT_SUMMARY>
	- <Child Node1>2.3.1 </Child Node1>
	- <Child Node2>2.3.2 </Child Node2>
	- <Child Node3>2.3.3 </Child Node3>
	</ACCOUNT_SUMMARY>
    </ChildGroup>
  </ParentGroup>	
...........
</Document>


The XSL that I use has  three "preceding-sibling::"
The time taken to parse a 150KB XML in Xalan is 1.9
sec and with Saxon8 its 1.6 secs.

Since the above parsing is done frequently in my
application with different input XMLs. I need to
further reduce the transformation timing (in
milliseconds may be). The only bottleneck I see are
the 3 preceding-sibling. Can we somehow remove them or
use some other logic, to reduce my transformation
time.

Here is the XSL:

<xsl:param name="priroot">ParentGroup</xsl:param>
<xsl:param name="secroot">ChildGroup</xsl:param>

<xsl:key name="PRI" match="ParentGroup" use="."/>
<xsl:key name="SEC" match="ChildGroup" use="."/>

<CUSTOMER_RELATIONS>
 <xsl:variable name="var_1"
select="Document/ParentGroup"/>
 <xsl:variable name="var_2"
select="Document/ChildGroup" />
 <xsl:variable name="var_3" select="Document/Child
Node1" />
 <xsl:variable name="var_4" select="Document/Child
Node2" />
 <xsl:variable name="var_5" select="Document/Child
Node3" />
   <xsl:for-each select="$var_1[count(. | key('PRI',
.)[1]) = 1]">
   <xsl:element name="{$priroot}">
	<xsl:variable name="priId" select="." />
	<xsl:variable name="p"
select="count(preceding-sibling::ParentGroup) + 1" />
	<PRIMARY_CUST_DESC><xsl:value-of
select="$priId"/></PRIMARY_CUST_DESC>
	<xsl:for-each select="$var_2[count(. | key('SEC',
.)[1]) = 1]">
   	   <xsl:variable name="m"
select="count(preceding-sibling::ChildGroup) + 1" />
	   <xsl:if test="$var_1[$m] = $priId">
	   <xsl:variable name="id" select="." />
	   <xsl:element name="{$secroot}">
	   <xsl:for-each select="$var_2[. = $id]">
		<xsl:variable name="n"
select="count(preceding-sibling::ChildGroup) + 1" />
		<ACCOUNT_SUMMARY> 		 	       
<UNIQUE_ID>A<xsl:value-of select="$n"/></UNIQUE_ID>
		<xsl:copy-of select="." />
		<xsl:copy-of select="$var_3[$n]" />
		<xsl:copy-of select="$var_4[$n]" />
		</ACCOUNT_SUMMARY>
	   </xsl:for-each>
	   </xsl:element>
	   </xsl:if>				
	</xsl:for-each>
	</xsl:element>
   </xsl:for-each>
</CUSTOMER_RELATIONS>

I would appreciate any suggestions to reduce the
timings.

Thanks in advance,
Bhupendra.



	
	
		
___________________________________________________________ALL-NEW Yahoo! Messenger - all new features - even more fun! http://uk.messenger.yahoo.com