Probably wouldn't hurt to toss in hooks to the NCX and OPF as well.
Would it make sense to be able to access all files in the ebook via some sort of iterable dictionary or class structure?
Code:
for sheet in book.css:
if sheet.name == file_name_parsed_from_link_in_current_xml:
do_stuff
With the ability to add/remove files to the ebook's structure?
Just spitballin'.
