Every time I have to deal with python packaging I hate it even more than before
Like, sure, serialising your API as HTML5 is totally fine and not unspeakably cursed way to do it
....yes, they invented a JSON version, but it's not supported by our version of internal registry, and updating it will take somewhere between two hours and half year
and, the most convenient way to parse such amazing markup with python stdlib, is regular expressions. Because xml parsers can't do html, and included html.parser is only half-done
LMAO, even half-done stub in stdlib uses regular expressions
after three different full featured xml parsers in module next to, this feels like an insult
https://github.com/python/cpython/blob/3.14/Lib/html/parser.py
GitHubcpython/Lib/html/parser.py at 3.14 · python/cpythonThe Python programming language. Contribute to python/cpython development by creating an account on GitHub.There's several completely incompatible formats, and zero official tooling to deal with that
except pip, but it works only for cli
well, I sorta hacked it with pip & grep via subprocess.run
slow as fuck, but at least I don't have to manually parse 5 different formats
and our repo is small anyways
Note to self: never, ever try dealing with not installed python packages
Create venv in /tmp, install without deps, and read metadata/source from site-packages
Overkill? In case of python — no.