[Python] Python custom src-layout: How to add another subpackage layer?

I have my Python package structured in src-layout. I run multiple github repositories with each the same super-package-name, structured as follows:

package-subpackage1
├── .git/
├── src/
│ └── package/
│ └── subpackage1/
│ ├── __init__.py
│ └── code_subpackage1.py
└── pyproject.toml


package-subpackage2/
├── .git/
├── src/
│ └── package/
│ └── subpackage2/
│ ├── __init__.py
│ └── code_subpackage2.py
└── pyproject.toml


By setting the .toml in each repository to

[tool.setuptools.packages.find]
where = ["src"]
include = ["package.*"]


I can import the modules as

import package.subpackage1

import package.subpackage2


Now I would like to add a subsubpackage-Layer to subpackage1, like

package-subpackage1/
├── .git/
├── src/
│ └── package/
│ └── subpackage1/
│ ├── __init__.py
│ ├── code_subpackage1.py
│ └── subsubpackage1/
│ ├── __init__.py
│ └── code_subsubpackage1.py
└── pyproject.toml


which I would like to be able to call via

import package.subpackage1
subpackage1.subsubpackage1.SomeClass


When I try to access the subsubpackage1 via

import package.subpackage1 as pack
pack.subsubpackage1


I get

AttributeError: module 'package.subpackage1' has no attribute 'subsubpackage1'


which is kind of expected, since I am breaking the vanilla src/-Layout and the automatic discovery seems to fail. However, if I manually import subsubpackage1 afterwards via

import package.subpackage1.subsubpackage1


all my classes in subsubpackage1.py become "magically" available via

pack.subsubpackage1.SomeClass


I don’t seem to understand how it is possible that a manual import of the subsubpackage1 makes it available in the previously imported subpackage1. Any idea why this is the case?

And if I want to design my pyproject.toml correctly, do I have to explicitly set tools.packages.packages and reassemble my files into this custom structure as suggested in here ? Or is there a more standardized way to include another subpackage layer into the src-Layout?

(Don’t be too hard on me pls, this if my first question ever on StackOverFlow )

Continue reading...