jah 1.3

Autohinter for data science assignments

JAH (Jeff's AutoHinter)

For data science and public policy courses at Georgetown. This repo combines two formerly-separate repos: one for the Python package that gets imported in each student's notebook, and another for the documentation of the package.

1 Publishing to PyPI

There are a few steps where, if one is skipped, the project won't update on PyPI!

1. Increment the version number (can use semantic versioning etc.) in pyproject.toml
2. Run python3 -m build, which will create new tar.gz files in the dist directory corresponding to the new version number
3. Run add-dots.sh, which just handles the annoying-ness of JupyterHub's interface not allowing me to see dotfiles
4. Push to github like usual

1.1 Testing the PyPI Version

1. Create a new virtual environment with e.g. python3 -m venv .venv
2. Activate that environment with source .venv/bin/activate
3. Install jah into the virtual environment with python3 -m pip install jah
4. Start a Python REPL shell with python
5. Try importing and using jah
6. Use quit() to exit the Python REPL shell
7. Uninstall with python3 -m pip uninstall jah
8. Deactivate the virtual environment by just typing deactivate

2 Package Documentation

The documentation of the package uses Zensical. The goal of Zensical, in theory, is to streamline the conversion of codebases into documentation (for example, by automatically converting in-code comment strings into pieces of the documentation). Here we don't use thse fancy features, so it essentially just becomes a... static site publishing system like Jekyll.

All of the Zensical operations here should be run within the custom virtual environment for the docs, by first running:

source .venv/bin/activate

You should then see a (jag) prefix in the Terminal shell, indicating that you're working within this virtual environment.

2.1 Previewing Documentation

The documentation pages are drawn from the docs subfolder. To preview what it looks like, run

zensical serve

The main configuration of the site is in zensical.toml (as opposed to pyproject.toml, which contains the configuration for the Python package itself!)

2.2 Publishing Documentation

Once the documentation is ready to be pushed, you can build the Zensical site structure using

zensical build

And, if that runs without error, you can then push it to GitHub via

git add -A .
git commit -m "Your message here"
git push

The updates should then be reflected on https://jjacobs.me/jah shortly thereafter.