Nowadays, when I read mostly papers in PDF or books I have scanned in or ebooks, I do screenshots and copy/paste to OmniOutliner, an outliner program.
It's all still a hack though. I'm still waiting for a clean way from within standard ebook reading apps. For example, an API by iBooks that would allow any 3rd party flashcard app to interact with my notes/highlights.
I used to think this use case is so obvious that it'd be a matter of months that someone would come up with a way to make it more easy to both read and write/organize content this way. As it is, now I'm increasingly asking myself whether I'm an exception or whether this is not the way non-literary books _should_ be read after all.
I could see your way of doing things being useful, though. I suspect it hasn't been partly because it's harder than it sounds (look at the design process for HTML, and all of the mistakes that were made and alternatives that were proposed) and partly because so much of the relevant IP (and thus market) is already locked up pretty tightly.