LFReader

LFReader is a self-hosted Local-first Feed Reader written in Python and Preact/React.

Note

This app is still in its early stage. Breaking changes may occur from time to time.

Features

• PWA (Progressive Web App) support: provide better loading experience and can be installed to a device
• Support for various feed formats: RSS, Atom, CDF, and JSON feeds
• Local-first: Feeds and entries are stored in a sqlite3 database so that you can read local feeds even without Internet access. Futhermore, it supports archiving resources like images in the entries.
• Dark mode support: Users can choose between light and dark modes
• Flexible archiving options: User can use regex to filter the html tags and values when archiving
• Responsive UI: it supports both large-screen and small-screen devices

Screenshots

Light mode:

Dark mode:

Config options:

Installation

Using docker/podman

# create data dir
mkdir -p data
docker run -d -p 8080:80 --name lfreader -v $PWD/data:/app/data dcsunset/lfreader

Then access http://localhost:8080 in browser.

The database and archived files will be stored in the volume /app/data in container (or where you mount it in the host). The log of backend server is stored at /app/logs/backend.log inside the container. You can quickly check the log by docker exec lfreader tail /app/logs/backend.log.

The default config file can be found in docker/config.json in this repo. You can use your own config simply by mounting it to /app/config.json:

# suppose your config is at ./config.json
docker run -d -p 8080:80 --name lfreader -v $PWD/config.json:/app/config.json -v $PWD/data:/app/data dcsunset/lfreader

Using pip

To install backend server only, you can use pip to install the backend dir in this repo:

pip install -e "git+https://github.com/DCsunset/LFReader.git#egg=lfreader_server&subdirectory=backend"

Using Nix

This package is available in NUR. The output of the derivation includes both backend executable (bin/lfreader-server) and frontend static files (in share/lfreader).

Besides, a NixOS module for backend server is provided in the Nix flake output. Simply import the flake and import the module modules.lfreader and add the following NixOS configuration:

{
  imports = [ nur-dcsunset.modules.lfreader ];
  services.lfreader = {
    enable = true;
    host = "::";
    port = 3000;
    openFirewall = true;
    settings = {
      db_file = "/data/db.sqlite";
      archiver = {
        base_dir = "/data/archives";
      };
      log_level = "info";
    };
  };
}

From source code

First, clone this repo.

Then, build the frontend (bundled files are in directory frontend/dist):

cd frontend
npm i
npm run build
cd ..

Next, install dependencies and run backend:

# use Nix
nix run .#backend-prod

# or manually
cd backend
pip install -r requirements.txt
uvicorn lfreader_server.app:app --host 0.0.0.0 --port 3000

Finally, use your favorite web server to serve the frontend files (frontend/dist) and set up the reverse proxy to pass routes prefixed by /api to backend. If archiving is enabled (default), make sure the archives directory is served at /archives.

Configuration

Frontend

Available config options can be found in the sidebar UI.

One important option is to enable/disable archiving. For feeds that contain many images or videos, it might be slow and expensive to archive them. You can disable archiving globally in such case.

Backend

The following environment variables are supported when running the backend:

Variable	Description
LFREADER_CONFIG	Config file path

The config file is in JSON format. Please see refer to Config class in file backend/config.py for available options and default values.

Development

Frontend

Change directory to frontend. Create a symlink public/archivesto the backend archive directory to serve resources. Finally, run npm run dev.

Backend

With Nix, run dev server directly: nix run .#backend-dev

Or install all the dependencies (you could also use venv here):

pip install -r fastapi uvicorn
uvicorn lfreader_server.app:app --reload --port 3000

To test if the backend pyproject can build successfully:

cd backend
## optionally set up vevn
# python -m venv venv
# . ./venv/bin/activate

## pip install
pip install .

Migration

v3.0.0

Steps to migrate database from v2.0.0+ to v3.0.0+:

1. Make a backup of both the old database and archive directory
2. Change base_url, base_dir, and archive_options to match your old settings and put the old archive directory at the specified location
3. Run python scripts/migrate_v3.0.0.py <old_db> <new_db> to create the new db and update the archive directory
4. Remove unnecessary subdirectories n archive directory. The new archives directory should only contain files
5. Test the migrated db and archive directory with v3.0.0+
6. It's recommended to use the db with v3.1.0+ as it optimizes the storage

v2.0.0

Move original environment variables to config file.

v1.2.0

Steps to migrate database from below v1.2.0 to v1.2.0:

1. Make a backup of the old db first.
2. Create a new database by running LFReader server v1.2.0.
3. Add all previous feeds through API or Web UI
4. Run python scripts/migrate_v1.2.0.py <old_db> <new_db> to migrate all previous entries

License

AGPL-3.0