Paperless-ngx 3.4.2-3

Paperless-ngx turns paper into a searchable archive: documents dropped into a folder (from a scanner, over SMB, or uploaded in its web interface) are read with OCR, made into searchable PDFs, tagged, dated and filed - then found again by searching their text. The app runs it natively on the NAS (no Docker). It needs the Valkey app (its task queue) - install that first.

• Its web interface is at https://paperless.<domain> through the Nginx app, if installed (added automatically), and at http://<nas-ip>:8040/ on the local network unless limited to the Nginx app on the management page. Its own phone apps and other clients use the same address.
• Logging in: user name admin and a generated password, which the management page shows, copies or replaces. Add more users (and give them permissions) in Paperless itself.
• Adding documents: drop them in its consume folder - Public/Paperless by default (created on first start; any folder in the shares can be chosen on the management page). Paperless imports each file, reads it (OCR, in English by default - German, or both, can be chosen), and removes it from the folder (so the folder chosen is made writable by the NAS's share group, if it isn't already). Scanners with "scan to network folder" can save straight into it. Uploading in the web interface works too.
• Your documents are kept in its data folder, next to the app, e.g. /mnt/HD/HD_a2/Nas_Prog/_paperless-ngx, and survive reinstalls and removal: the originals and their searchable copies (media/), its database (SQLite) and search index (data/), the page's settings, the secret key and admin password (root only), and paperless.log. Back it up (e.g. with the Kopia app); Paperless's own exporter also works over SSH: .../paperless-ngx/pl_ctl.sh manage document_exporter /mnt/HD/HD_a2/Public/paperless-export.
• It runs as its own paperless user (created by the app, ID 96) in the NAS's share group, as four processes: its web server (Granian), the consumer (watching the consume folder), a task worker (OCR and the rest, one document at a time) and a scheduler (Celery). When the NAS starts, it waits for Valkey, then updates its database if a new version needs it.
• Memory and speed: it uses about 600MB-1GB, more while reading a document. OCR on the NAS's processor takes from a few seconds to a minute a page.

What's left out

• Office documents and e-mails (Word, Excel, .eml): Paperless reads them through two separate services (Apache Tika and Gotenberg), which aren't included. PDFs, images (scans, photos) and plain text all work. To add them, run Tika and Gotenberg (e.g. in the Docker app) and point Paperless at them with its PAPERLESS_TIKA_* settings.
• Local AI models: Paperless's AI features (suggestions, chat) work with OpenAI-compatible servers or Ollama, set in its own settings - not with models run by Paperless itself (its "HuggingFace" embedding option needs PyTorch, over 1GB, which isn't included).
• Other OCR languages than English and German (Tesseract's language data isn't included).

How it's built

static/paperless-ngx prepares everything, from Paperless's own release (its ready-made web interface, and the exact package versions it pins, with their checksums):

• Paperless in its own Python 3.14 (a portable build), with the document tools its OCR uses: Tesseract (English, German), Ghostscript, pngquant, qpdf, poppler and libmagic from Debian, and unpaper (page cleaning) built against a small ffmpeg.
• They need a newer glibc than the NAS's, so the app brings its own (Debian Trixie's, in binaries/runtime): every program in it is patched to use it - the NAS's own programs are unaffected (as for the Calibre-Web app). So the app must be installed on the first volume (HD_a2).
• Paperless doesn't update itself: updates come with new versions of the app, and the NAS shows an alert when a new release is out, checked weekly.

Management page

Paperless's page in the NAS dashboard (Apps → Paperless-ngx) shows whether it's running (each of its processes, or what it's waiting for while starting), its version, its addresses and the admin login (with the password). It sets the consume folder, the documents' language and whether it's only reachable through the Nginx app, gives the admin a new password, restarts Paperless and shows its latest log lines. Only NAS admins can use the page (see Management pages).