Dynamic imports don't work if the python path is a level up from their location

[ingrinder]

Given a directory structure like this:

project_folder/
└── my_module/
    ├── some_submodules/
    │   ├── module_a.py
    │   └── module_b.py
    └── main.py

If I use absolute imports in my code (e.g. import my_module.some_submodules.module_a), then my $PYTHONPATH must be project_folder, so I execute stickytape in this folder and --add-python-path .

However, if instead of importing the submodules directly in main.py, and instead import using importlib.import_module...

from importlib import import_module
import_module('my_module.some_submodules.module_a.py')

...when I run the stickytape script, I get a traceback ending in ModuleNotFoundError: No module named 'my_module'.

Obviously this is a simplified example but it demonstrates the behaviour I'm getting in some software where modules are dynamically imported with importlib in this way. Perhaps it could be possible to add these submodules to stickytape somehow, like how you can specify hidden modules in PyInstaller?

[mwilliamson]

master now has an option --add-python-module in the CLI to explicitly include Python modules. So, in your example, you'd include --add-python-module my_module.some_submodules.module_a. Does that work for you?

[ingrinder]

Yep, that works perfectly. Thanks!