Lint all files in a directory

[matusvalo]

Type of Changes

	Type
	🐛 Bug fix
✓	✨ New feature
	🔨 Refactoring
	📜 Docs

Description

This PR adds new parameter --discover-files which switches pylint into mode when it iterates over FS subtree and executes pylint on all .py files within this subtree regardless they are in package or not. But in the other hand this PR honours package name in the final output.

Example consider following subtree:

- test
     `+- a.py
      +- b.py
      +- c
          `+- __init__.py
           +- d.py 

The pylint output with no switch will be following:

$ pylint test
************* Module test
test/__init__.py:1:0: F0010: error while code parsing: Unable to load file test/__init__.py:
[Errno 2] No such file or directory: 'test/__init__.py' (parse-error)

The pylint output with switch will be following:

Note: Each file contains single line simulating error 1 == 1

$ pylint --discover-files test
************* Module a
test/a.py:1:0: W0104: Statement seems to have no effect (pointless-statement)
test/a.py:1:0: R0124: Redundant comparison - 1 == 1 (comparison-with-itself)
************* Module b
test/b.py:1:0: W0104: Statement seems to have no effect (pointless-statement)
test/b.py:1:0: R0124: Redundant comparison - 1 == 1 (comparison-with-itself)
************* Module c
test/c/__init__.py:1:0: W0104: Statement seems to have no effect (pointless-statement)
test/c/__init__.py:1:0: R0124: Redundant comparison - 1 == 1 (comparison-with-itself)
************* Module c.d
test/c/d.py:1:0: W0104: Statement seems to have no effect (pointless-statement)
test/c/d.py:1:0: R0124: Redundant comparison - 1 == 1 (comparison-with-itself)

------------------------------------------------------------------
Your code has been rated at 0.00/10 (previous run: 0.00/10, +0.00)

Closes #352

[Pierre-Sassoulas]

I did not look into detail but this looks promising, thank you @matusvalo. #352 is the base issue, #4217 is another approach for reference and it would need to be closed if we merge this.

[matusvalo]

This PR is now in draft mode. Pending points:

• Fix CI, implement tests
• Fine tune code - type annotations, docstrings etc.
• Finish documentation

I would like to ask just for high level review whether the approach is right one.

[matusvalo]

I did not look into detail but this looks promising, thank you @matusvalo. #352 is the base issue, #4217 is another approach for reference and it would need to be closed if we merge this.

I missed #4217 since it is not linked to the issue. I let the decision to maintainers which PR should be finished (or maybe merged together - I had not look into #4217 yet). Please let me know what is your preference.

[Pierre-Sassoulas]

The other PR got stalled for a long time because of this scary comment. (I've not looked at the issue for a long time and maybe not sufficiently in depth at the time but it seems it was overwhelming for the original implementer). Do you think this would be a problem with your approach ?

[matusvalo]

The other PR got stalled for a long time because of this scary comment. (I've not looked at the issue for a long time and maybe not sufficiently in depth at the time but it seems it was overwhelming for the original implementer). Do you think this would be a problem with your approach ?

I will check that.

[matusvalo]

@Pierre-Sassoulas I have created additional PR trying to solve the issue with problems around sys.path adjustments. Please have a look. My apologies for so many parallel opened PRs, but I am still not sure which approach is the best.

[matusvalo]

Closing in favour of #5682