2

I need a small, lightweight DMS for personal use, which is available on Linux.

Requirements:

  • runs on Linux
  • manages PDF
  • stores documents in the file system¹ in an easily recognizable structure²
  • offers a GUI (WebUI is fine, native Linux GUI as well)
  • not too many dependencies (e.g. no KDE libs as I don't use KDE)
  • tags, manual descriptions
  • find documents by tags, name, date, description

Strongly preferred:

  • available via standard repositories (Debian/Ubungu/Mint) or at least easy to install
  • OCR capabilities (for full-text search), or at least offering full text search for documents having text (for PDF, e.g. integrated via pdfsandwich; for LibreOffice, extracted from the document)

Nice to have:

  • maintains other formats (images, ODF (LibreOffice/OpenOffice) documents)
  • integrated document viewer capable to deal with supported formats
  • scan folders for (new) documents (so I can drop them in the right place and the DMS indexes them). A special "drop folder" is OK, but even better if I can place the documents where they belong. The folder(s) don't need to be monitored; manually triggering the index process is totally fine.
  • custom fields
  • grouping of connected documents (think of "book series" having multiple volumes). This could e.g. also be achieved by attaching/merging multiple documents onto/into a single record.

Using a database backend for metadata is fine (it it does, preferably SQLite or MySQL).

Not needed:

  • multi-user, approvals etc. (as pointed out above: personal use only)
  • Windows compatibility
  • "reference systems". I just need to manage my own documents. References between those are fine – but I'm not collecting external references.
  • encrypted storage (if that's needed, I can encrypt the file system)

Definitely unwanted:

  • cloud-based (unless self-hosted) – so no Evernote, please.

I know we have quite a few DMS questions, but none of them completely meets my requirements. Most of them are asking for collaborative features, but I only want a light-weight variant for personal use. Coming close (and how they miss):

I've also seen paperless which basically fits the above, but seems pretty complex to install. Zotero is already mentioned in the footnotes; but in addition to that it's rather for reference management.


¹ should the software die, I still can find and access the documents
² e.g. Zotero fails this one, using cryptic directory names

2
  • Have you also looked at nemakiware.com - I came across it in the past. Not used it yet but planned to when required. You can search for it on stackexchange as it is under a few answers.
    – Z Z
    Jan 7, 2019 at 11:01
  • Thanks, but that sounds too big as well: I don't need multi-account, team collaboration etc – and I don't like having to register for the download. I meanwhile settled with a directory tree of PDFSandwhiched PDFs and pdfgrep for search. Not elegant, but does the job :)
    – Izzy
    Jan 7, 2019 at 21:13

2 Answers 2

1

Due to the lack of alternatives, I've settled for a very minimalist approach meanwhile. All ingredients are available from the standard repositories on Ubuntu/Mint, though for more up-to-date versions, at least for one part I instead used the Debian package offered at the project's website:

PDFSandwich is a wrapper around some tools like unpaper (to optimize pages for better OCR results), Tesseract (for OCR'ing the PDFs) and some helpers like Ghostscript and ImageMagick. You feed it a PDF, and it spits out another one which combines the original one with its OCR results:

pdfsandwich -lang eng sample.pdf

will result in sample_ocr.pdf, which is a sandwich of the two.

Now, I organize the documents in an appropriate directory structure. And when I still don't know what PDF has the contents I'm after, I switch to the root directory and use

pdfgrep -r -i foobar

Which spits out a list of each line from all the documents containing "foobar", prefixed by the path/to/document.pdf.

No elegant GUI, but it does its job. Let's see how it meets my original criteria:

  • runs on Linux: Yes.
  • manages PDF: Indeed.
  • stores documents in the file system¹ in an easily recognizable structure²: My job, but: Yes.
  • offers a GUI (WebUI is fine, native Linux GUI as well): No, just a CLI
  • not too many dependencies (e.g. no KDE libs as I don't use KDE): Definitely.
  • tags, manual descriptions: can be realised in the file names. Again, not elegant (e.g. no "global tag rename"), but possible.
  • find documents by tags, name, date, description: Not exactly – but sufficient for my use. "By content" is done via pdfgrep, "by name" can be done via | grep <name>, "by date" via other CLI tools (like awkto isolate the file name, then ls -l or stat to get the date and filtering that way – a bit heavy, though :)

The "nice-to-haves" are only vaguely met (sure I can put all other file types into the structure – but then search would be difficult as pdfgrep only deals with PDF, etc. There's no "global index" to search). So I'm still open for alternatives – but for now, this works for me.

0

you can try using paperless.node I've created:

  • runs on Linux, Windows or Mac
  • manages PDF, images, emails
  • stores documents in the file system their original form in a flat directory
  • offers a WebUI
  • tags, manual descriptions
  • find documents by tags, name, description
  • easy to install
  • integrated document viewer
  • has a drop folder

screenshot

3
  • Great. Please what are main differences with something like github.com/paperless-ngx/paperless-ngx apart from fact that paperless.node retains original directory? Thanks.
    – Z Z
    Aug 10, 2022 at 8:16
  • @ZZ - to be perfectly honest - I was not aware of the paperless-ng project, so I do not have a full comparison. My solution is more lightweight, and has less track record (part of being a project that one person worked on it for a couple of months, vs a project multiple people worked on for multiple years...). I hope it has all the needed requirements, but I would appreciate feedback :)
    – Uri Agassi
    Aug 10, 2022 at 14:08
  • Thanks, that looks good! But I stopped reading in the middle of the first install step. I might be opinionated there, but I don't want to use NodeJS. But it indeed seems more comfortable – especially as it includes the DMS part. Will certainly be what a lot of people like!
    – Izzy
    Aug 10, 2022 at 18:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.