jonaswinkler / paperless-ng

A supercharged version of paperless: scan, index and archive all your physical documents
https://paperless-ng.readthedocs.io/en/latest/
GNU General Public License v3.0
5.37k stars 357 forks source link

[Feature Request] Auto date suggestion #450

Open igno2k opened 3 years ago

igno2k commented 3 years ago

paperless-ng often gets the right date from the document itself or file name. Sometimes it chooses the wrong date by using the first one found in the document. It would be nice if paperless-ng would show additional dates it has found in the document and we could use it by just clicking on it.

docspell (docspell.org) has exactly this nifty feature and it is really useful.

shamoon commented 3 years ago

Duplicate of #187 / #188 (I agree 😀)

Philmo67 commented 3 years ago

And maybe then store somewhere that "for this document type coming from this correspondent, I have to use this date" ?

igno2k commented 3 years ago

And maybe then store somewhere that "for this document type coming from this correspondent, I have to use this date" ?

That would be great! Like the automatic tagging paperless-ng should learn which date is the right one for one document-correspondent-type

ignoreigor commented 3 years ago

@igno2k Don't know if @jonaswinkler wants to extend this feature so far. If this is going to work, one'll suggest to extract more metadata, such as invoice-totals or invoice-numbers automatically by learning. This is exactly what e.g. insiders smartfix is doing. Nice, but a sort of other product category.

@Philmo67: this is what I thought of, when I suggested dependent rules, processed in an order in issue #447. Once you identified a correspondent (maybe by UStID or something elso), you could use a regex, who greps the correct date (or another field and put it into a fixed or custom field), and put some more tags on this document. Also some work to implement, some work for you to define (easy) rules, but I'm not sure, if this is not going to work more properly in the nearest future.

But let @jonaswinkler decide. It's his task to lead paperless-ng in it's future.