Use a heuristic to keep out PDFs that don't seem to have fields

[nonprofittechy]

1. If it has no fields [after processing by formfyxer], it's not a form
2. If it doesn't have at least one page with more than 2 fields detected, it's not a form

[BryceStevenWilley]

Counterpoint: I'm not sure we can trust jurisdictions to actually use Acroform PDF with fields. I'm pretty sure I've seen a few jurisdictions have forms without PDF fields.

If you are thinking post field detection, this could work, but it would be a bit less reliable, there are usually enough false positives.

(EDIT: apologies, didn't see the original conversation until now)

[nonprofittechy]

Yes, this would be post field detection. Take a look at https://suffolklitlab.org/form-explorer/list/PA/ for inspiration for this issue. There are definitely lots of false positives, but not usually more than 2 per page.

[nonprofittechy]

(EDIT: apologies, didn't see the original conversation until now)

No worries, good reminder to add full context to the GitHub issue. No guarantee we'd remember it in a week or two when we can come back to this task.