It's at least partially a copyright claim, isn't it? So the method -- OCR or scraping -- doesn't matter, I think.