Pylint doesn't check file if it's named exactly like the directory where the file is

[goldiee36]

Bug description

If I have a file in a directory, and the filename is exactly the same as the directory name, I cannot make pylint to check that file if I am not inside of that directory.

Steps to reproduce:

mkdir something
echo "import yaml" > something/something
pylint something/something 

Nothing happens, but if I rename the file, it works:

mv something/something something/anything
pylint something/anything 

Now pylint works as expected.

If I step inside the directory (cd something) and run pylint for the file, it also works.

If I create an __init__.py file in the directory, pylint starts working from outside of the directory.

I cannot figure it out how to use pylint for these kind of files.

Command used

pylint something/something

Pylint output

pylint outputs nothing

Expected behavior

pylint does the check and outputs the result

Pylint version

pylint 2.15.6

OS / Environment

CentOS

[DanielNoord]

Does it also break when the file has a .py extension?

[dmnc98]

No, if the file has the .py extension it works. Also, it seems like it is case sensitive. While pylint something/Something and pylint something/something.py both work, pylint something/something does not (this also includes usage of the absolute path e.g. pylint /home/user/some_path/something/something).

[goldiee36]

Exactly as Paudo wrote.

[dmnc98]

Our rough folder structure:

❯ tree .
.
└── Hooks
    ├── SOMETHING_1
    │   └── SOMETHING_1 # Python Script
    └── SOMETHING_2
        └── SOMETHING_1 # Python Script

Some additional information:
So far I found out that if pylint is executed from within Hooks we get the unexpected behavior mentioned above where there is no output at all. If pylint is executed from one directory above (the folder in which Hooks is stored) the expected output is generated.
When I tried to debug it I noticed that in file pylint/pylint/lint/expand_modules.py and function expand_modules the execution of path = sys.path.copy() results in a list that includes one more entry that points to the Hooks folder via absolute path.
This leads to the file being detected as namespace with no init and not being a directory.
This finally leads to the execution of get_module_files(os.path.dirname('<REST_OF_PATH>/Hooks/Purge_Namespace/Purge_Namespace'), ['CSV'], list_all=True) which returns no files, since it is not detected as module.

[DanielNoord]

It's very likely that the namespace detection is broken by the fact that these files don't have a .py extension. That feels like a pattern that we don't directly support.
Why are the python scripts not saved as a python file?

[goldiee36]

Why are the python scripts not saved as a python file?

The python files are part of a generic automation system where you can create a job/task, and the job automatically pulls and executes the file which is named exactly like how the job was named. The system does not require to implement the entry executable in python, and as the jobs have descriptive names like onboard_cnf, and not onboard_cnf.py or onboard_cnf.sh hence the files usually omit the extension.