Watchtower Pipeline ​

Script to fetch data for Watchtower. This can also be used as a library for custom integrations.

Development Setup ​

• Ensure you have Poetry installed
• Create a virtualenv
• poetry install

Running the script ​

• Run as a module with python -m watchtower_pipeline.{connector}, where {connector} is:
  • example
  • kitsu

Files overview ​

These are the building blocks of the pipeline:

• models.py: Dataclass representations of all Watchtower data structures
• writers.py: Utilities to write out JSON files
• ffprobe.py: Wrapper around ffprobe, needed to calculate the duration of a video file

This is how those blocks can be used:

• example.py: Generate synthetic data for demo purposes
• kitsu.py: Fetch data from Kitsu (this is the setup used at Blender Studio)

Developing a custom connector ​

The fastest way to develop a custom connector is to create {connector_name}.py in the watchtower_pipeline directory and modify example.py to suit your needs.

JSON files data source ​

Watchtower runs as static web application, and use JSON file as data source. Here is how such data should be structured:

projects/
├── <project-id>/
│   ├── assets.json
│   ├── casting.json
│   ├── edit.json
│   ├── edit.mp4
│   ├── project.json
│   ├── sequences.json
│   └── shots.json
└── context.json

Publish an update to Pypi ​

• Navigate to client-web and then yarn run build-for-py
• Navigate to pipeline and run poetry publish --build

Publish the docs ​

• Navigate to docs and ensure you have a .env file
• Run yarn run docs:publish