98 implement usethis tool import linter

[nathanjmcdougall]

No description provided.

[codspeed-hq]

CodSpeed Performance Report

Merging #501 will not alter performance

_{Comparing 98-implement-usethis-tool-import-linter (3cf868b) with main (868463e)}

Summary

✅ 1 untouched benchmarks

[codecov]

Codecov Report

Attention: Patch coverage is 98.78049% with 2 lines in your changes missing coverage. Please review.

Project coverage is 97.25%. Comparing base (868463e) to head (3cf868b).

✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
src/usethis/_integrations/file/ini/io_.py	87.50%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #501      +/-   ##
==========================================
- Coverage   97.27%   97.25%   -0.03%     
==========================================
  Files          83       85       +2     
  Lines        3966     4079     +113     
==========================================
+ Hits         3858     3967     +109     
- Misses        108      112       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[nathanjmcdougall]

I think #500 is a pre-requisite to merging because otherwise the warning message here will be repeated multiple times unnecessarily:

usethis-python/src/usethis/_tool.py

Lines 958 to 964 in 6919a86

	if not root_packages:
	# Couldn't find any packages, we're assuming the package name is the name
	# of the project. Warn the user accordingly.
	name = get_project_name()
	warn_print("Could not find any importable packages.")
	warn_print(f"Assuming the package name is {name}.")
	root_packages = [name]

[nathanjmcdougall]

Based on the changes from #500 it has highlighted another issue - we might have an arbitrary number of config entries - we want to lay claim to a "family" of ConfigEntries all with the prefix importlinter:contracts: in the section name.

This is getting quite complicated, but I think it's the right choice.

The getter for an individual contract might need to use a cached function which gives all contracts values, and returns NoConfigValue if the contract doesn't exist.

So we need a way to register an abstract family of contract entries.

[nathanjmcdougall]

The CI isn't running anymore, rather inexplicitly. It might be because this PR has too many commits against it, so I'll try opening a new PR I think.