Do you use the filename guesswork feature?

[jonaswinkler]

You can name files in certain ways and paperless will pick up metadata from that, such as tags, dates and titles.

See guesswork in the documentation.

Do you use it?

[brianwanderson]

Hi,

Here is some feedback on how I use paperless:

I really like the PAPERLESS_FILENAME_DATE_ORDER=YMD setting in the config file. I prefix all of my documents with YYYY.MM.DD so that I end up with file names like: "2020.12.01 - Document Title.pdf". That way, I can always sort my documents chronologically in a file browser. It's really convenient that paperless will automatically set the documents date from the file name on import.

Beyond grabbing the date, I don't use any of the filename guesswork features. Personally, I think anything more is overkill. I don't want to be bothered by having to remember some special file name format like "Correspondent - Title - tag,tag,tag.pdf" It's easier for me to name my files however I wish and to assign correspondents and tags in the GUI after import.

[qcasey]

Basically the same here, I'd see myself using filename guesswork on dates when bulk importing a bunch of documents. Paperless has been very good at filling in titles and dates otherwise.

I personally don't use correspondents or tags at all. The search function is really good enough.

[rknightion]

Personally no, as my scanner gives random filenames and going through and renaming would be more effort when the existing search and auto tagging capabilities are so good 👍

[jonaswinkler]

I'll probably reduce that to date detection only at some point.

[jonaswinkler]

I do have a PR (almost) ready for showing a swagger UI for the paperless-ng API. This might be helpful for developers that want to play with the API to e.g. write a small program to upload a document from their scanner (guess who might want to do that :-)).

Interesting. The Django REST framework also provides a user-browsable API, with filtering and all that, so I'm curious how that compares. Sounds redundant right now, but lets see.

Furthermore, I've implemented a user-definable meta-data model but I'm currently struggling with the frontend integration.

Curious, this is on my todo list and I've already got pretty definite ideas about how to do it. See #274 (comment), and the rest of that thread as well. Is your current progress available somewhere?

Ideas are on using django-haystack for search (to be able to scale),

I'm using whoosh directly, without haystack, since certain search features (more like this and faceting) are not available through haystack.

cloud storage (hey we might want to bring back file-encryption for this),

Propose a good plan about how to do it that also takes the database, search index and filenames into account and does not store the decryption key on the server. I've concluded that it's pretty much impossible without redesigning major parts of the application.

my target is to do pip install paperless-ng, paperless-ng start

Paperless depends on external non-python binaries. The docker setup has a pretty good time-to-awesome, download some files, pull, go.

[DocLambda]

I'm probably hijacking this discussion but well... :-)

The Django REST framework also provides a user-browsable API.

That's what I used to provide the OpenAPI specification directly derived from the django url-router. On top of it I added an Angular component to integrate swagger-ui.

I have to cleanup the Angular-mess I made and try to submit this as a PR next week (hopefully).

#274

I will comment there.

haystack

Let's discuss this later. Maybe I can try to make those features available through haystack... :-)

Propose a good plan about how to do it that also takes the database, search index and filenames into account and does not store the decryption key on the server. I've concluded that it's pretty much impossible without redesigning major parts of the application.

Well to be honest this is not necessary in my case. What I want is to put the document files in the cloud storage but keep the database, index, etc. locally. This way I only need to "protect" the files but can keep the rest plain-text.

Paperless depends on external non-python binaries. The docker setup has a pretty good time-to-awesome, download some files, pull, go.

Well true, but for one I don't like docker too much as a "solution to setup things that are too complicated otherwise" and secondly I hate to remember which commands to run when developing and starting the server, the consumer, the worker queue... Don't get me wrong, paperless-ng is awesome, but I had a hard time to start this from a git clone even when reading your documentation. Especially the starting of a (development) server instance is too hard. But I'm convinced wrapping this is not too complicated...

Thanks for all your work!

[jonaswinkler]

Well to be honest this is not necessary in my case. What I want is to put the document files in the cloud storage but keep the database, index, etc. locally. This way I only need to "protect" the files but can keep the rest plain-text.

Let's discuss this in #40 .

Don't get me wrong, paperless-ng is awesome, but I had a hard time to start this from a git clone even when reading your documentation.

I know that getting the development environment up and running isn't exactly straight forward and that the documentation is severly lacking in that regard.

[DocLambda]

Cool. Will try to work on it a bit and let you know.

[jonaswinkler]

That's what I used to provide the OpenAPI specification directly derived from the django url-router. On top of it I added an Angular component to integrate swagger-ui.

Regarding this, I don't think the average user needs to access the API directly, so integrating that into the front end is questionable.

[RosaEinhorn]

Is there a feature to add tags based on the name of a file in the /consume folder? I cannot find it. :-(

[jonaswinkler]

There is not. The previous implementation of that was not pretty good and was causing many issues with miss-named tags, correspondents, and incomplete titles.

[tanderson1992]

I know the guesswork feature is already gone, but I'll add that I used it extensively. I found it easy to bulk rename documents (years of bank statements, for example) with the proper names to be consumed into the correct correspondents, titles, dates, and tags. As I mentioned in another discussion, I thought my wife would balk, but after a few tries she was regularly scanning documents with her phone and naming them the correct way to be consumed. We had an occasional hiccup, but for the most part it worked well for us.

[dennisdek]

Hi Jonas,
this would be exactly what I'm looking for at the moment, but it seems it has been removed based on the replies above?
My use case: My scanner allows up to 10 FTP profiles. So my ideal would be to setup several profiles, all leading to the same consume folder - only difference is the filename that is generated for the scan.
E.g. a profile 'tax', another one 'school'. That way I can just select the right profile in the scanner and the document gets tagged correctly based on the file name, without any further action in the app itself.