Actually I'm already manipulating my database heavily, same like i was doing with MS Access DB used by Alfa eBook Manager, and as I did in my
Thumbnail browser for calibre database and seems that it doing the job without any conflicts.
Anyway I agree that it will be better to use calibre itself, but I don't know Python and have no time to learn it, so to overcome issues for example I'm deleting books from DB by connecting to calibre through command line instead of manipulating DB and files directly, i think the same can be done with metadata because there is also a command line for that, actually the really big issue is to grab metadata and TOC from the providers or from PDF files directly (not speaking of DJVU).