Lofi Girl Scrobbler 🎧

Lofi Girl Scrobbler helps you scrobble (mark as listened) Lofi Girl (previously known as chilledcow) YouTube live stream music tracks. It is a rewrite of a previous weekend project of mine. Now it supports both LastFM and ListenBrainz with now playing support on both platforms as well. This time it's written in Rust and it's completely panic free!

Getting Started

Pre-Requisites

This project uses opencv library to capture/process images and tesseract-ocr to make an image to text analysis.

Modules

This project includes different modules and several features which you can choose according to your preference. The list of binaries/frontend which are compiled on releases are:

• Lofigirl Server - A REST API http broadcasting server module which does the image processing, ocr and serves it on a selected port. Requires all the dependencies to be present in the system. It keeps user sessions and sends the listening information to multiple backends.
• Lofigirl CLI Client - Includes Optional multi-os notification system.
  • Default Client - A client-only version which doesn't require opencv or tesseract-ocr dependencies. It uses the given server configuration to communicate with the server module using the REST API.
  • Standalone - The standalone version which does the images processing/OCR and sends the listening information data on a regular interval. Requires all the dependencies to be present in the system.
• Lofigirl Web Client - Compiled to wasm, runs on browser and communicates with the server module using the REST API.
• Lofigirl Flutter Client - Flutter multi platform front-end that communicates with the server module using the REST API.

Installing all dependencies

Arch Linux

On a minimal arch installation, these packages (besides rust and cargo) were required to be able compile and run the system:

pacman -S openssl pkgconf opencv vtk hdf5 qt5-base glew tesseract tesseract-data-eng clang

Windows

Using vcpkg, it should be possible to compile and use the system.

vcpkg install llvm opencv4[contrib,nonfree] tesseract

MacOS

brew install opencv tesseract leptonica

Compiling

Server side uses sqlx which does compile time query checking so a db must be present on the compilation time. To set the compile time db and run migrations;

cargo install sqlx-cli
export DATABASE_URL=sqlite:token.db
sqlx db create 
sqlx migrate run

Instead of sqlx-cli, sqlite3 also works to create the db. However, the compilation still requires the DATABASE_URL environment variable.

sqlite3 token.db < migrations/20210525000135_table.sql 
export DATABASE_URL=sqlite:token.db

Compile all;

cargo build --release

Check server, client, web-client and flutter-client for more information on specific module.

Configuration

To use the system with LastFM, you need a API key and secret. These can be obtained here. To use with ListenBrainz, you'll need to give your user token which can be found here.

On the previous project, the system would use the youtube channel information to get live streams however, since it's required YouTube API as well, in this version I decided to simply give the youtube video link directly. Since the livestreams last months, it shouldn't be a big deal. Additionally, it's now also possible to give the second livestream link as well if you want to scrobble that instead. Check usage for how to do it.

All of the configuration can be put into a toml file like in the example:

[lastfm] # client optional - server ignore
username = "username" # will be removed after first run and turned into session_key 
password = "password" # will be removed after first run and turned into session_key

[lastfm_api] #standalone client and server use. The others ignore
api_key = "api_key"
api_secret = "api_secret"

[listenbrainz] # client optional - server ignore
token = "token"

[video] # standalone client and server use, others ignore
links = [
    "https::///www.youtube.com/something",
    "https::///www.youtube.com/something", # optional
    "https::///www.youtube.com/something", # optional
]

[server] # client only mandatory, others would ignore.
link = "http://127.0.0.1:8080"

[server_settings] # server uses, others ignore
token_db = "token.db"
port = 8888

Both LastFM and ListenBrainz are optional. You can use one or both depending however you want.

LastFM username and password only used once to receive the session_key and they are not stored. Only LastFM session_key and ListenBrainz token are stored on server side which can be seen in the server table schema.

Usage

Check server, client or web-client usage in their sections.

Docker

Docker images includes all binaries (without notification support).

docker pull gokberkkocak/lofigirl

The default entry point of the Docker image is the server module.

Use -v to pass your configuration file and token db file to the container and -p for port arrangement.

docker run -d -v /path/to/config.toml:/config.toml \
              -v /path/to/token.db:/token.db -p 8888:8888 gokberkkocak/lofigirl:latest 

To use with other modules, use --entrypoint flag.

docker run -d -v /path/to/your/config.toml:/config.toml --entrypoint {lofigirl_standalone|lofigirl} gokberkkocak/lofigirl:latest 

How does it Work

Main stream as an example

• The program takes the video link and extracts the raw video stream link using ytextract/rustube.
• Using opencv, it opens the stream and captures a single frame as an image periodically.
• The image gets cropped.
• The background is removed by a mask.
• The final image is checked by tesseract-ocr.
• The info is sent to LastFM and/or ListenBrainz.

Second stream

It works the same way for the second stream as well.

• Full image
• Cropped image
• Masked
• Sending to LastFM and ListenBrainz.