Python uses a list of “Trove classifiers ” to add some structured meta data to projects. It can be a bit tedius to add these manually. Combining the ease of uvx with a Python package, we can simplify this.

add

While I have brainstormed about a script to automate some of this, a post from adamghill informed me there is a trove-classifiers project that has all the classifiers already included. If I could combine this with something like peco for interactive filtering, this could work!

After a brief search, I found the pzp project which was the missing part I need. While recent Python versions have tomllib built in, I decided to go with tomlkit so that I could write it back. With these two pieces, uvx add-classifiers add was created.

With this command, you can easily get a list of classifiers as an interactive filtered list that you can easily add to your pyproject.toml

suggest

From the same thread Ash_Crow made a futher great suggestion.

a useful one could be checking that the classifiers in [project] are consistent with the versions actually supported by what is declared in [dependencies] (for Django) or [requires-python] (for Python)

From this uvx add-classifiers suggest was born.

From the packaging library, we can use the Requirement class and SpecifierSet class to more easily and more reliabilly parse anything in project.python-requires and project.dependencies.

With these two classes, it was not much additional work to process a set of suggested_addition and suggested_removal that can automatally be applied to our pyproject.toml

summary

With these existing packages, the filter + add method ends up being fairly simple.

https://codeberg.org/kfdm/add-classifiers/src/tag/v0.2.0/src/add_classifiers/__init__.py#L48-L68

The suggest method has to do a bit more but it is also not significantly more complex.

https://codeberg.org/kfdm/add-classifiers/src/tag/v0.2.0/src/add_classifiers/__init__.py#L71-L126

I will probably continue to poke around and make some small improvements, but the intial version already seems like it will work well and works about as good as my initial brainstorm hoped.

https://codeberg.org/kfdm/add-classifiers/issues https://pypi.org/project/add-classifiers/