Support for PDF receipts

[bram-atmire]

Not sure if this use case is shared among others: I use Scanbot to scan my receipts as multi-page PDFs. Would be great if this tool could work on these pdfs.

Scanbot does a sort of OCR itself, but it doesn't seem to be that good, in the sense that it adds too much noise: a receipt contains so much text, and I'm only interested in the articles, price per article, to see price evolution across multiple weeks.

[mre]

Your use-case makes a lot of sense to me.
We could use pdf2image as a preprocessor before recognizing the text. I think that would be the easiest thing to try.
Alternatively you could try OCRmyPDF to see if it works with your inputs out of the box.