This repository holds all of the source documents as well as output for the Computable AI blog.
Computable AI is powered by Pelican, a powerful static site generator. There are many static site generators, but Pelican was chosen in support of blogging via Jupyter.
Clone this repo with --recursive to get the ipynb plugin necessary to convert notebooks into posts. If you already cloned without --recursive, do git submodule update --init --recursive.
We use Pipenv for managing Python dependencies. TL;DR:
pip install pipenv
pipenv install
The dev branch (default) contains all of the source files. The master branch is published by GitHub Pages, and only contains the output.
Start with pipenv shell to enter the virtualenv shell for this repo.
Run make devserver to start an auto-refreshing local server hosting the site at http://localhost:8000, then run jupyter notebook and use Jupyter to write site content in content/, occasionally checking the Pelican-built version of the article to make sure things look right.
Look at an existing post's source to see how metadata works (each source notebook has some YAML in its first cell).
Here's some example metadata. Only the first four keys are required. The default category is "Miscellany". The default image is the first image found in your post. The default status is published.
- title: Title of this post
- summary: The summary/subtitle of this post
- author: Joe Blogger
- date: 2019-03-03
- category: Example
- image: /images/someimage.png
- status: draft
Images can be right-justified, left-justified, or centered by appending, e.g., #right to a source url in img tag src.
When I'm done, make publish publishes the site (by generating a production build and using ghp-import to extract the output/ into master).