pip show ignores environment markers

[wimglenn]

Description

Perhaps the best explanation is to offer an example:

$ pip show opencv-python | grep Requires
Requires: numpy, numpy, numpy, numpy, numpy, numpy

Wow, it really wants numpy! :)

The reason is that environment markers are there:
https://github.com/opencv/opencv-python/blob/43cd7169cb92d96898d59753796678495b95c833/setup.py#L26-L36

Expected behavior

I'm not sure whether they should be deduplicated by pip:

Requires: numpy

Evaluated and merged by pip (in my case for Python-3.11 on macOS):

Requires: numpy >=1.23.5

Or just rendered without evaluation. It's pretty crazy for this distribution (opencv-python==4.8.0.74) but I guess that would be something like:

Requires: numpy (>=1.13.3) ; python_version < "3.7", numpy (>=1.21.0) ; python_version <= "3.9" and platform_system == "Darwin" and platform_machine == "arm64", numpy (>=1.21.2) ; python_version >= "3.10", numpy (>=1.21.4) ; python_version >= "3.10" and platform_system == "Darwin", numpy (>=1.23.5) ; python_version >= "3.11", numpy (>=1.19.3) ; python_version >= "3.6" and platform_system == "Linux" and platform_machine == "aarch64", numpy (>=1.17.0) ; python_version >= "3.7", numpy (>=1.17.3) ; python_version >= "3.8", numpy (>=1.19.3) ; python_version >= "3.9"

The latter is probably my preference, even though this particular package makes it look pathological. The current behavior of stripping the markers entirely in pip show's output seems worse than any of these options.

pip version

23.2

Python version

3.11

OS

macOS

How to Reproduce

pip install opencv-python
pip show opencv-python

Output

No response

Code of Conduct

• I agree to follow the PSF Code of Conduct.

[HanslettTheDev]

I have a similar issue when adding environment markers

[uranusjr]

Pull requests are always welcomed.

[wimglenn]

@uranusjr Well, which of the outputs mentioned in "Expected behavior" would you like to see in a PR?

[uranusjr]

pip show is never strictly accurate and is bounded by the current platform so I’d say numpy is enough.

[chrysle]

@wimglenn Do you want to tackle the issue? Otherwise I'd give it a try.

[wimglenn]

@chrysle go ahead