Hello,
In the spirit of talk radio, I'm a first-time poster, long-time listener. This site has been a gold mine of XSLT resources. Thanks. This will sound like a SOAP question, but the solution is using XSLT.

I am calling legacy SOAP rpc/encoded services from a newer platform that does not support the entire SOAP rpc spec. I need to transform xml formatted with multireference accessors(href/id) to a more traditional inline xml. Our COTS software expects one child under the SOAP Body, and multireference rpc payloads contain multiple children under the SOAP Body. I am trying to transform this:

<env:Envelope xmlns:env="http://schemas.xmlsoap.org/soap/envelope/" xmlns:tns="http://target.org" env:encodingStyle="http://schemas.xmlsoap.org/soap/encoding/">
<env:Body>
<tns:legacyRpcServiceResponse>
<result href="#ID1"/>
</tns:legacyRpcServiceResponse>
<tns:LookupTableResult id="ID1" xsi:type="tns:LookupTableResult">
<returnCode xsi:type="xsd:string">0</returnCode>
<returnMessage xsi:type="xsd:string" xsi:nil="1"/>
<tableArray href="#ID2"/>
</tns:LookupTableResult>
<tns:ArrayOfTableArray id="ID2" xsi:type="enc:Array" enc:arrayType="tns:TableArray[2]">
<item href="#ID3"/>
<item href="#ID4"/>
</tns:ArrayOfTableArray>
<tns:TableArray id="ID3" xsi:type="tns:TableArray">
<lastChangeDt xsi:type="xsd:string">2009-05-12-13.13.42.145</lastChangeDt>
<tableName xsi:type="xsd:string">NameTable</tableName>
</tns:TableArray>
<tns:TableArray id="ID4" xsi:type="tns:TableArray">
<lastChangeDt xsi:type="xsd:string"/>
<tableName xsi:type="xsd:string">JobTable</tableName>
</tns:TableArray>
</env:Body>
</env:Envelope>

to this inline version:

<?xml version="1.0" encoding="UTF-8"?>
<env:Envelope xmlns:env="http://schemas.xmlsoap.org/soap/envelope/" xmlns:tns="http://target.org" env:encodingStyle="http://schemas.xmlsoap.org/soap/encoding/">
<env:Body>
<tns:legacyRpcServiceResponse>
<result>
<tns:LookupTableResult type="tns:LookupTableResult">
<returnCode type="xsd:string">0</returnCode>
<returnMessage type="xsd:string" nil="1"/>
<tableArray>
<tns:ArrayOfTableArray type="enc:Array" arrayType="tns:TableArray[2]">
<item>
<tns:TableArray type="tns:TableArray">
<lastChangeDt type="xsd:string">2009-05-12-13.13.42.145</lastChangeDt>
<tableName type="xsd:string">NameTable</tableName>
</tns:TableArray>
</item>
<item>
<tns:TableArray type="tns:TableArray">
<lastChangeDt type="xsd:string"/>
<tableName type="xsd:string">JobTable</tableName>
</tns:TableArray>
</item>
</tns:ArrayOfTableArray>
</tableArray>
</tns:LookupTableResult>
</result>
</tns:legacyRpcServiceResponse>
</env:Body>
</env:Envelope>

My XSLT below uses templates to copy all nodes. When it encounters a node containing an href attribute, it copies the nodeset with the matching id attribute. I had to use a mode to keep from copying the elements with an id attribute twice. Some of the services I call can contain 1,000's of references. The bulk of my xslt is spent selecting the elements referenced by the href. I have saved a ton of time using a key, but believe that there is still a lot of room for improvement. Are there tricks out there to:
1) select an element by id from a distinct large list of elements?
2) stop processing the document once I've finished processing the first child of the soap:Body ?

My xslt is:

<xsl:stylesheet version="1.0" xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<xsl:key name="linkableItems" match="//*[@id]" use="@id"/>
<xsl:template match="/">
<xsl:apply-templates/>
</xsl:template>
<xsl:template match="@*[name()='id']|@*[name()='href']"/>
<xsl:template match="node()" mode="normalize">
<xsl:copy>
<xsl:apply-templates select="@*|node()"/>
</xsl:copy>
</xsl:template>
<xsl:template match="node()[@href]">
<xsl:copy>
<xsl:apply-templates select="key('linkableItems',substring-after(current()/@href,'#'))" mode="normalize"/>
</xsl:copy>
</xsl:template>
<xsl:template match="node()[@id]"/>
<xsl:template match="@*|node()">
<xsl:copy>
<xsl:apply-templates select="@*|node()"/>
</xsl:copy>
</xsl:template>
</xsl:stylesheet>

Thank you.

There are several ways to select the first child:

As for selecting an element by id from a distinct large list of element will depend on what you want to select. in the input xml you posted, what to do you want to select?

As my XSLT is copying nodes, for any node containing an href attribute, I want to go fetch the element with the matching id attribute and copy it and its children into the current node. So if the original xml is

<tns:legacyRpcServiceResponse>
<result href="#ID1"/>
</tns:legacyRpcServiceResponse>
...
the output xml after resolving the href to id references would be

<tns:legacyRpcServiceResponse>
<result>
<tns:LookupTableResult type="tns:LookupTableResult">
...
</tns:LookupTableResult>
</result>
</tns:legacyRpcServiceResponse>

I defined a Key on the @id attribute to significantly improve the selection of the element whose @id attribute matches the @href attribute of the context node.
98% of my transformation is spent on this selection, so I was trying to see if there's an even faster way to do it. So if I know I have a list of 10,000 elements with distinct @id values, is there a way to instruct the processor to stop searching the full node list once a match is found? I think the XSL engine is trying to find all nodes who match the criteria, even though I know it will only result in one node.

Adding the [1] predicate at the end happens after the fact and only selects the 1st node of a node list that contains one element. At least that's how I understand it to work.
Any other thoughts?

I am a bit confused. Your first post gives a desired output:

which is what your xslt outputs.

I think his question is not "how do I do this" but "is this the fastest way of doing this".

Your performance will vary greatly depending on the XSLT processor you use and how they have implemented keys etc. Some processors may do exactly what you want when faced with the [1] predicate and stop once they have found the first key, however others may not. Different processors will also likely have better or worse routines for doing the key indexing.

The only way to know for sure would be to develop some test cases and try running them against different XSLT processors. Basically you're going to have to benchmark.

__________________
/- Sam Judson : Wrox Technical Editor -/

Think before you post: What have you tried?

I would think any XSLT processor with a half-decent optimizer would ensure that xxx[1] stops searching xxx after finding the first item in the sequence. But perhaps there are some XSLT processors around without half-decent optimizers. The only way to find out is to do measurements.

__________________
Michael Kay
http://www.saxonica.com/
Author, XSLT 2.0 and XPath 2.0 Programmer\'s Reference

Thanks for the tip on how the [1] predicate effects optimization. I did not know that.
The processor definitely makes a difference. I have a 4mb xml file with over 10k href attributes. XMLSpy processes the doc in 24 - 30 seconds, yet our Oracle SOA Suite server, using Xalan 2.7.0, processes the doc in under 7 seconds. This is fine for now since we only get a couple of these requests per day. The majority of our docs are in the 1 to 2kb range and process sub second. I'm just getting greedy and wanted to see how far I could optimize this xslt.

I appreciate all of your tips. Thanks!