Search for efficiency...(building nodes dynamically)

Hi all,

I have an XML file of 6588 Airport elements as follows:

<apt code="AAF" city="Apalachicola" country="United States"
state="FL">Municipal Apt.</apt>
....

In an XSLT program, I am processing yet another XML document that has
the following format:

<availabilitytable>
	<avail>
		<origin>XYZ</origin>
		<destination>XYZ</destination>
	</avail>

	<avail>
		<origin>XYZ</origin>
		<destination>XYZ</destination>
	</avail>
</availabilitytable>

These are an approx of 15-20 'avail' elements.

The problem: I have written the XSL sheet in such a way that for every
'origin' and 'destination' element, the
<apt> nodeset is traversed to get the airport name of a given code.

Therefore, if there are duplicate origins/destinations (which is very
probable in my case) there will be unnecessary multiple traversals.

So I thought of building a nodeset of all origins/destinations,
removing duplicates, and then looking up the <apt> file.

The Question(s): Since I will need the result node later on in the XSL
file, where do I store it ? And how do I build a result node like this
? If this is not feasible, what is the best method to avoid multiple
traversals of such a large tree ?

Any help is appreciated,

Regards,

Ahsan