Dependencies with Different Import Names Incorrectly Reported as Unused

[ViaDav1d]

I'm experiencing a false positive with deptry, where scikit-learn is marked as an unused dependency despite being used in my project. This appears to be due to module aliasing, where the installed package (scikit-learn) has a different import name (sklearn).

I noticed that PR #1033 aimed to fix issues related to transitive dependencies with different module names, but this issue still persists in the latest version of deptry.

Steps to Reproduce

1. Install scikit-learn:
   pip install scikit-learn

2. Create a simple Python script (example.py):
   from sklearn.model_selection import train_test_split

3. Run deptry in the project directory:
   deptry --verbose

4. Observe that scikit-learn is incorrectly flagged as an unused dependency (DEP002).
   Expected Behavior -
   eptry should recognize that scikit-learn is in use and not mark it as an unused dependency.

Actual Behavior
deptry incorrectly reports:

Unused dependencies found:
scikit-learn

Environment
deptry version: 0.23.0
Python version: 3.13.2

Additional Information
This issue might be related to how deptry detects dependencies when the installed package name differs from the imported module name.

Other libraries with similar aliasing may also be affected.

Can this be investigated further to properly handle cases where the installed package name differs from the import name? If there's a known workaround, I'd love to hear about it.

Thanks for the great work on deptry! 🚀

[robsonpiere]

Another example is the python-dateutil package

pip install python-dateutil

from dateutil.relativedelta import relativedelta

[m-aciek]

If there's a known workaround, I'd love to hear about it.

@ViaDav1d You can configure this case in a package module name map.

[mkniewallner]

Do you run deptry from a virtual environment where the dependencies are installed? Without that, deptry cannot guess the mapping from the package and the module (unless you specify it manually as @m-aciek shared):

When using pip to install deptry, make sure you install it within the virtual environment of your project. Installing deptry globally will not work, since it needs to have access to the metadata of the packages in the virtual environment.