A worker script (
scanner.py) loops through the Twitch API's list of streams and spins until it inserts all streamers it finds matching the search criteria (default zero viewers), then it starts again. These stale streams are pruned after a fresh reload occurs on the assumption that someone will view them and then they won't have zero viewers any more so should not be served for too long; we'll pick them back up on the next load if they're still viewerless.
Environment variables needed for both scanner and app:
NOBODY_HOST: the database host
NOBODY_DATABASE: the database name
NOBODY_USER: the database user
NOBODY_PASSWORD: the database password
Environment variables to be set for the scanner only:
CLIENT_ID: Your Twitch application client ID (found at https://dev.twitch.tv/console)
CLIENT_SECRET: Your Twitch application client secret (found at https://dev.twitch.tv/console)
Meanwhile, the Sanic app in
app.py serves the index and the endpoint to get a random streamer.
Note that the database schema makes use of slightly unusual extensions, namely
pg_trgm (trigram indices) and
tsm_system_rows. You may need to install your operating system's flavor of
postgresql-contrib in order for them to work correctly.
python scanner.py). This will need to run continuously. Be sure to include your database credentials in the environment variables. This also performs the initial database schema migration; additional modifications to the schema may require dropping tables/etc. See
db_utils.pyfor the exact schema.
python app.py). Be sure to include your database credentials in the environment variables.
This is obviously not production ready; you'll need to make sure all services are running as daemons (an example
supervisord config is included in
etc) and that your Sanic app is running safely (e.g. behind gunicorn/nginx/pick your poison).
Update direct dependencies in
pip-compile to compile them down to
requirements.txt if you update them.