MS Office OCR works great for this, it did the Rogue.c jpegs in no time, but there are missing routines. I wonder if an AI could fill the gap.
A listing made on a dot matrix printer that contain PETSCII graphic characters might make it struggle. It'd be interesting to see if the LLM could infer the intended character.
I ment fixing the missing code blocks, of would suppose that an LLM could fix the mangled code without OCR training...