How to structure python modules/packages according to dependecy inversion

Question

If I am working on a project, say it has this file structure:

car/
 body/
 __init__.py
 doors.py
 bonnet.py
 engine/
 cyclinderhead/
 __init__.py
 pistons.py
 __init__.py
 crankshaft.py
 engconstants.py
 utilities.py 

Now invariabley the pistons.py file will need some sort of constant (engconstants.py) or utility function (utilities.py). The pistons.py is a very nested deep file. Is it okay that it depends on files higher up in the file structure? I got the impression that the less nested files should only depend on the more nested files, and depencies going the other way are a code smell? I often get myself into circular import predicaments and am trying to implement a set of rules/guidelines to prevent that.

• WMRamadan: Are we to assume that each python file has only one class? For example pistons.py has class Pistons() and so on, or can you give an examples of how the classes are structured?

Answer 1

No it's all right when subpackages depend on higher modules or packages depend on subpackages. The circular import problem not related to that. It can appear even if files have the same nesting level.

Try to make each module the most independent to not encounter circular import. But sometimes independence can lead to violating DRY and Single Source of Truth.

If, after all, you encounter circular import, the fastest and sometimes the best solution is merging dependencies into one module.

• blnk.off: Dependencies Inversion principle is applicable for class design. Subpackages are designed for structuring related modules.