ARG Monitoring

A semi-properly-not-really tool that checks the integrity of the ARG.

This repository runs scripts everyday at 00:00 GMT+00:00 to do the following:

• Check the integrity of the YouTube videos
• Check the integrity of metadata (e.g. number of video uploads, number of soundcloud tracks, etc)

It makes use of GitHub Actions cache to store the "known" states. The output of this repository is a RSS feed, which is pushed to this very same repository on the feed branch. Here is a handy link; add it to your RSS reader.

(If you want the "live" JSON, use this link)

Twitch Intro Sequence

As part of monitoring, a intro sequence detection has been implemented. This requires priming; to do so, navigate to feed-generator/src/sources/twitch-source.

It involves the following steps:

1. Create a bunch of reference squares using frameviewer.py (this is a little slow). (Parameters: input video, and output directory, either neuro or evil).
2. Analyze the SSIM mean and confidence interval between many videos using analyze.py. (Parameters: source directory, either neuro or evil, detection square, either detectors/neuro_detector.png or detectors/evil_detector.png. Modify the script's YOUTUBE_VIDEOS list to check it against multiple videos.
3. Rename the output (result.npz) to either neuro.npz or evil.npz. This is the thresholds file.
4. Run scrutinize.py to check the threshold against a particular video.

If the above steps are successful, the neuro/, evil/ directories, and neruo.npz, evil.npz files can be committed to be used as threshold values for the upcoming stream.

There is also a Twitch Stream trigger, which relies on:

• A valid Twitch Client ID and Secret
• A valid GitHub Personal Access Token
• A valid callback URL (either ngrok or equivalent)

Run the file hook_listener.py to create the first skeleton for a config file, which will be secrets.ini.

Audio Monitoring

Audio monitoring is achieved with @owobred's pleep program.

Contributing

The repository is split into two parts: feed-generator and web.

Feed Generator

The Feed Generator is written in Python.

cd feed-generator/
pip install -e . // or poetry install

If you hate red squiggly lines, run this as well:

mypy --install-types

Web

The Web interface is written both in JavaScript and Rust (because I wanted to).

Install pre-requisites:

cd web/
npm install -g wasm-pack
cd www/
npm install
cd -

Build the project:

wasm-pack build

To serve the development HTTP server, try:

cd www/
npm start