Thank you for a comment, Dale!
Unfortunately, these files don't seem to have any metadata. I've checked it with few viewers. For example, DjView shows this:
I'm starting to believe that only some type of manual process is the only thing that will work. I'm looking at various OCR software packages to see if any of them can scan only first N pages but then I'll have to extract ISBN manually anyway… or maybe let Calibre plugin extract it somehow. I'm still thinking about the best way to go about this.