Extracting dates from a PDF

[mindphil]

Hello, working on extracting dates from PDF's, using PyPDF2 and re (for adding exceptions). What are some clever ways to increase the accuracy of strings extracted? Running into issues getting false positives and some dates being extracted that aren't even in the PDF to begin with.

[stefan6419846]

As a first step, consider switching to pypdf, as PyPDF2 is EOL. Further recommendations are hard to give in general, but it might help to know the approximated location on the page, combined with a text visitor function to only retrieve the relevant values.

[mindphil]

Switching the pypdf was the right call. Made some major headway by calling previous functions that identified the document type. If it is x, the right date is usually right in the header, so we read only the first page and extract the first 200 characters. If it's y, it's generally on the last page.

Already had a comprehensive (as I could determine) date string patterns, but only having to apply that to a smaller selection was huge for accuracy. I don't think it is really helpful but wrote a function to ignore obviously incorrect dates.