I’m just trying to gauge the typical usage of queries from within the API. I’d like to know if they are mainly for documents, document fragments or a mixture of both.

I’m not clear on the terminolgy.

By documents do you mean

XML documents ie XML decl+prolog+document Element
or document Elements?

By fragments to you mean

nodes which are children of the document elements
for example sub elements
or do you mean elements and text nodes together
or what?