[jdom-interest] How would you do this kind of Document processing?

[Aleksi Kallio]

I need to create a selective "view" to a bunch XML-documents (for generating HTML with XSLT from those documents). And it needs to be fast.

I can remove and add contents, but the problem is that the originals are broken after the operation. I can do a reverse operation to recreate the originals, but it turns out to be awfully hard, because I'm not just collecting Documents under a one root, but making a more sophisticated selection. Of course I could copy, but it isn't fast.

Can you come up with a trick that makes it possible to compose a temporal Document out of multiple documents, without affecting the originals (or automatically restoring them)?