Setting Up Rogue Scholar

I’m not sure when I first came across his stuff, but I’ve really admired Andrew Heiss’ public scholarship, including his blog, which is full of handy stuff. At some point I noticed he had post-level DOIs minted. I knew you could mint repository-level DOIs with GitHub and Zenodo, but maintaining each post as a repository and adding them as submodules to a blog is… a lot.

Turns out, Rogue Scholar is a science blogging platform that will archive your posts and mint a DOI for them! As I learned from reading (and in communicating with the Rogue Scholar Team), the requirements are

• The blog needs to be licensed CC-BY

• Every post needs an author

• Every post needs a date

• The full post needs to be published in the RSS feed

Additionally, it’s possible to pre-assign your post a DOI using the commonmeta tool. So, this post is in part a description about how I spruced up my quarto infrastructure to enforce some of these requirements, and is also partially a test to make sure I did it right!

Enforcing metadata

I sometimes have a bad habit of creating a new quarto file for a post, and forgetting to add a date, which would be no good for the RSS feed. And if I had to go back and forth between the Rogue Scholar archive and my blog files to error-free copy-paste DOIs, this was never going to work. My approach was to put together a quarto pre-render script, which you can look at in full here.

The upshot is that I use pathlib to walk through and find all of my post-level directories, which I have (maybe annoyingly) organized like so:

blog/
└── posts/
    └── YYYY/
        └── MM/
            └── YYYY-MM-DD_slug/
                └── index.qmd

I can take the YYYY-MM-DD portion of the post directory to get the date, and I can generate a DOI using commonmeta-py like so

from commonmeta import encode_doi
encode_doi(
  "10.59350" #the Rogue Scholar prefix
)

'https://doi.org/10.59350/fdgm0-evb66'

Then, I drop this info into a post-level _metadata.yml file. If you look at the actual function I wrote to update _metadata.yml, I have a lot of conditionals because

1. If _metadata.yml already has a doi field, I don’t want to overwrite it!
2. And if _metadata.yml already has an accurate date, I don’t want waste the time on writing to it either.

A nice thing about how I set this up, though, is if I create a post directory and don’t wind up publishing the post until a few days later, the date metadata ought to get automatically updated if I just change the post’s directory name.

With my pre-render script written, I just added it to the pre-render list in my quarto configuration file.

Adjusting my GitHub action

For the most part, my GitHub publication action isn’t running code chunks because I’m freezing execution locally. But the pre-render script runs every time, and needs to point to the right python environment that has commonmeta-py and pyyaml installed. I’m using renv to manage all of my dependencies, so getting this to work involved adding the following step before Quarto publish:

- name: Set Quarto Python
  run: |
    echo "QUARTO_PYTHON=renv/python/virtualenvs/renv-python-3.12/bin/python" >> $GITHUB_ENV

Wrapping up

Hopefully I configured this all correctly!