Put the data source in a separate repo (and in a more language independent format)

[zachleat]

Idea from https://fediverse.zachleat.com/@dominik@nona.social/109382540011847100

it might be a good idea to separate the implementation from the database. This way other implementations in other languages can use the same canonical list.

[nhoizey]

@zachleat should the implementation part move to tweetback and we could keep this repo for the mapping file?

[zachleat]

Ah, I hope not! I use this as a dependency on a few standalone web sites too (11ty.dev).

I think it can be as simple as using a JSON file instead, though I would acknowledge that in my experience the strictness of JSON creates more friction with pull requests that have syntax errors.

Maybe generating the JSON file with a GitHub action or as part of the release process?