The build context is the repository root and there was no .dockerignore, so
'COPY backend/ ./' swept backend/venv (327 MB), backend/media (the actual
library, 224 MB), backend/logs, backend/staticfiles and backend/.env
(SECRET_KEY plus the database password) into the backend image: 1.43 GB per
architecture, including secrets headed for the registry. The frontend build
stage also copied the host's node_modules over the fresh install.
- Root .dockerignore excludes .git, virtualenvs, __pycache__, db.sqlite3,
logs/staticfiles, .env files, media/, node_modules, dist and the deploy
runtime state (data/, tailscale-state/).
- The backend Dockerfile now asserts the context is clean (.env, venv,
media/library, db.sqlite3 all absent) before collectstatic, so a missing
ignore file fails the build instead of leaking.
- Rebuilt: backend 1.43 GB -> 876 MB ('COPY backend/' is now 268 kB),
frontend stays at 65 MB. Verified by booting the compose stack with the
new image: migrations applied, /health ok, no .env or venv inside, and
/app/media is the mounted (empty) volume; the scheduler runs too.
- Removed the stale local images that still contained the library and the
dev .env.
J621
Self-hosted media library and e621 archive manager, rebuilt as a React SPA + Django REST API.
Structure
backend/ Django 6 + DRF API (MariaDB + Redis via docker compose)
frontend/ Vite + React + TypeScript SPA
deploy/ Docker images, compose variants and Tailscale serve configs
extras/ shell integrations (fish_greeting with fastfetch)
Development
Backend
cd backend
source venv/bin/activate
python manage.py migrate
python manage.py scan_files # index the watched folder
python manage.py runserver
Copy .env.example to .env and set WATCHED_FOLDER before scanning.
Frontend
cd frontend
npm install
npm run dev # http://localhost:5173, proxies /api to :8000
Production
Docker: see deploy/ for the two images (SPA on static
nginx, API on gunicorn), the three compose variants (both / frontend-only /
backend-only) behind a shared nginx service and a Tailscale sidecar, and the
public-funnel or tailnet-only serve configs. deploy/push_*.sh builds and
pushes the multi-arch images to the Gitea registry.
Random image endpoint
Used by the SPA's Random page and by shell greetings (fish_greeting + fastfetch):
curl -H "Authorization: Token <token>" \
"https://j621.example.ts.net/api/random/?rating=s,q&fastfetch=1"
{
"j_id": "J-59",
"filename": "J-59.jpg",
"extension": "jpg",
"rating": "e",
"url": "https://j621.example.ts.net/api/files/J-59/raw/?sig=…",
"download_url": "https://j621.example.ts.net/api/files/J-59/raw/?sig=…&download=1",
"thumbnail_url": "https://j621.example.ts.net/api/files/J-59/thumbnail/?sig=…",
"fastfetch": true
}
rating— comma separated subset ofs,q,e(default: any).fastfetch=1, or any request whose User-Agent containsfastfetch, limits the roll topng/jpg/gifso terminals can display it. Images are the only candidates in both modes.urlis absolute, and signed for authenticated callers, so fastfetch can load it without headers. Guests get an unsigned URL and only see guest-visible items./randomand/random/are aliases of/api/random/for scripts. Behind the bundled nginx those aliases negotiate onAccept: browsers get the SPA page, requesters like curl/wget/fastfetch get the JSON./api/random/is the unambiguous path for scripts; 404 when nothing matches the filters.- A ready-made shell greeting that uses this endpoint lives in
extras/fish_greeting/. - Scoped tokens for scripts: the Account page's Shell tokens section
(also
/tokens) issuesj621r_…tokens that only authenticate/api/random/— the rest of the API rejects them. They are stored as hashes, shown once, and revocable any time (/api/auth/greeting-tokens/).
Licence
Source-available, non-commercial: personal and other non-commercial use is welcome under the Jake Labs Non-Commercial Software Licence, which requires attribution and keeps derivative works under the same licence. Commercial use is not permitted. Third-party dependencies keep their own licences (all permissive: MIT, BSD, Apache-2.0, ISC).