Virtual Tornado Siren

Streams a video of a tornado siren on the first Wednesday of every month at 11:55AM.

Setup

Pre-requisites:

Twitch account
Mastodon account (optional)
GCP account (only if deploying to GCP)

Setup Twitch and Mastodon

Setup Twitch streaming
- Find your Twitch stream key (STREAM KEY) and appropriate RTMP server (RTMP_URL)
- Retrieve your Twitch account's URL. This will be STREAM_URL.
Setup Mastodon reminder (optional)
- Create a bot account on Mastodon instance of your choice. The url of the instance will be MSTDN_URL.
- Go to settings and create an application. This will generate an access token which will be MSTDN_ACCESS_TOKEN.

Option 1: Deploy to Ubuntu

Ensure ffmpeg is installed on your machine:

sudo apt install ffmpeg

Clone the repo to your machine at /opt.

Install dependencies:

pip3 install -r requirements.txt

Find an mp4 video of a tornado siren of your choosing and save it to your machine (ex. data/siren.mp4). The path of the mp4 will be INPUT_FILE.

Create .env from sample.env and fill with your values.

Schedule a cronjob:

sudo crontab -e

55 11 * * * cd /opt/virtual-siren && python3 main.py

Option 2: Deploy to GCloud

Create a service user with Cloud Function Invoker privileges.

Find an mp4 video of a tornado siren of your choosing and host it as publically viewable on a static storage solution (Google Cloud Storage is one option). The path of the mp4 will be INPUT_FILE.

Create a .env.yaml with and populate it with proper credentials.

Deploy and schedule the function with:

# Deploy function
gcloud functions deploy stream_siren \
  --runtime python37 \
  --trigger-http \
  --region <REGION> \
  --env-vars-file .env.yaml \
  --timeout 240

# Create scheduled job
gcloud scheduler jobs create http stream-siren-job \
  --schedule "55 11 * * *" \
  --http-method GET \
  --time-zone "America/New_York" \
  --uri <TRIGGER URI> \
  --oidc-service-account-email <SERVICE ACCOUNT EMAIL>

Testing

python -m unittest discover --verbose tests

jasmaa / virtual-siren