Hello, I'm coming up against the edges of my journeyman usage of XSLT, and
concurrent with my huddling down with some books I'm hoping to get pointers
on the following.
I have 2 XML documents with the same schema, and I want to determine the set
difference (not symmetric set difference) between them. True, this has had
much written about it, but all the solutions I've seen simply show a
collection of "differenced" nodes, in a format that bears little resemblence
to the source document. Thus my twist is, I'd like to generate a set
difference, and retain enough context so I end up with a document [tree]
similar to the source--just pruned of unchanged nodes.
The "enough context" bears elaboration, but for now lets say I'd be happy if
the end result preserves tree branches that lead from the changed node, all
the way to the root. Each branch would imply at minimum (and show) a
changed leaf and its attributes, all the way through the root and its
attributes. In future I hope to be able to dynamically affect the meaning
of "enough context", but that for later.
Now some sample data (adapted from Sal Mangano's "XSLT Cookbook"):
doc1.xml
======
<doc>
<chapter n="1">
<section n="1">
<p n="1">Once upon a time...</p>
</section>
</chapter>
<chapter n="2">
<note n="1">I am still waiting for my $100000 advance.</note>
<section n="1">
<p n="1">... and they lived happily ever after.</p>
</section>
</chapter>
</doc>
doc2.xml
======
<doc>
<chapter n="1">
<section n="1.A">
<p n="1">Once upon a time...</p>
</section>
</chapter>
<chapter n="2.A">
<note n="1">I am still waiting for my $100000 advance.</note>
<section n="1">
<p n="1">... and they lived happily ever after.</p>
</section>
</chapter>
</doc>
And I'd like doc3.xml = doc2.xml \ doc1.xml to yield:
======
<doc>
<chapter n="1">
<section n="1.A">
</section>
</chapter>
<chapter n="2.A">
</chapter>
</doc>
Here I'm using "value semantics" for node equality--meaning 2 nodes are
considered equal, even if from different documents, if they have the same
namespace, local-name, and number and values of attributes. Let's assume
the same namespace(s).
Of course, one could make a case that the criteria for node equality should
be extended to mandate that descendent nodes of each node also compare equal
respectively. I suppose this would be a more orthogonal view, since then
doc3.xml above really shows all "changed" nodes: <doc> is present not just
as context; it's present because <doc> in doc2.xml is changed from <doc> in
doc1.xml by virtue of having changed descendent(s).
Thus far I've been able to determine the changed leaf nodes (this is pretty
much straight from "XSLT Cookbook", Chapter 7), and to write a template to
"print the branch" for each changed node. However I haven't been able to
"superimpose" or "union" these separate branches to get 1 output doc. I'm
possibly going about it all wrong, and hence the need to hunker down with
books.
But if somebody could suggest how I may tackle this, please do.
And if seeing any templates I've written would help I'll be glad to post
them. This is my first post to this (or any) XSLT list, and I don't want to
lenghthen an already long note.
Thanx
--A
_________________________________________________________________
Dont just search. Find. Check out the new MSN Search!
http://search.msn.click-url.com/go/onm00200636ave/direct/01/