Hektor backend

This is Hektor's Python backend. It's responsible for creating images based on a set of rules defined by the artist, Isabelle Gagné. It also provides a REST API to let the frontend app retrieve the data it needs.

Images processing can take a significant amount of time since we need to trigger a style transfer through FloydHub by running a temporary Model API based on our own fork of anishathalye/neural-style, thus we append every request to a queue using Django-Q and a Redis backend.

Requirements

These should be installed on the host machine prior to setting up the app:

• Python >= 3.7.3
• Redis >= 5.0.4

The style transfer app must be set up on the same machine: https://github.com/hektorbot/neural-style

Setup

1. Clone the project locally and cd into it:

$ git clone [email protected]:hektorbot/backend.git hektor-backend
$ cd hektor-backend

2. Create a virtual environment and activate it (optional):

$ python3 -m venv venv
$ . venv/bin/activate

3. Install required modules:

$ pip install -r requirements.txt

4. Create a .env file based on .env.example and set the variables as needed:

$ cp .env.example .env

5. Run migrations:

$ django manage.py migrate

6. A Vision API-authorized GCP service account key is required for reverse image search, make sure one is saved locally and that the GOOGLE_APPLICATION_CREDENTIALS references it properly

7. Pixelsort must be cloned locally and the PIXEL_SORT_PATH variable should be the absolute path to pixelsort.py

8. Run the app:

$ django manage.py runserver

9. To enable image processing, you'll need to start django-q's cluster in the background (make sure Redis server is running first):

$ nohup python manage.py qcluster&

The app should now be running on http://127.0.0.1:8000/