I built this project because I think writing in a paper journal is theraputic and should not go anywhere in the computer age, but I deeply fear something happening to my precious journals. The intent is to provide a clean and simple way to digitize journal entries, complete with OCR processing, and embedding to allow for semantic search across multiple entries.
Scan journal pages, then search them. A new install uses Postgres, files on disk, Tesseract, and Ollama.
cp .env.example .env
docker compose upOpen http://localhost:8000. The first account is the admin. Signups close after that. An admin adds users with POST /api/v1/auth/users.
Migrations run on startup. Other backends and OIDC are in .env.example. AUTH_BACKEND=dev_header is for development. The Vite app is in web/.
DBOS runs in this process and uses the same database. Conductor is optional: docker-compose.dbos.yml.
{
inputs.echo.url = "github:csille00/echo";
outputs = { nixpkgs, echo, ... }: {
nixosConfigurations.myhost = nixpkgs.lib.nixosSystem {
modules = [
echo.nixosModules.default
./echo.nix
];
};
};
}Copy nix/example.nix to echo.nix. The service listens on 127.0.0.1:8000, with Postgres (pgvector) and Ollama (nomic-embed-text). Point your reverse proxy at that address. Behind HTTPS, set PUBLIC_URL and SESSION_COOKIE_SECURE. Secrets go in environmentFile.
Set database.createLocally = false and DATABASE_URL in the environment file to use an existing database. Set ollama.enable = false to skip local embeddings.
nix-shell --run "uv pip install -e '.[dev]'"
cp .env.example .env
docker compose up db ollama
python -m echopython scripts/smoke_pipeline.py path/to/page.jpg