for ease of organization and additional testing, we need to split out a section of code in a project from a single file into a directory of files. basically the project is large, this section needs some additional isolated tests, and this is the best way - workflow wise and without gutting a ton of code - to handle our needs.
the directory now looks like this:
- __init__.py
- core.py
- api.py
- misc.py
- tests.py ( unittest.TestCase examples )
the application just deals with the methods from api.py ; however the other files interface with one another.
this is where i'm having a bit of an issue / second-guessing myself.
i'd usually do something like this to be explicit in my imports :
from .api import *
that works fine in the application, however when on the commandline to run local tests:
$ python tests.py
Python doesn't like the import.
This seems to be because I'm wanting to accomplish an anti-pattern ( at least according to Guido van Rossum - https://stackoverflow.com/a/6466139/442650 )
I haven't been able to find any better docs or recommendations on how to better set this up, aside from treating this part of the application as an external module -- which is not an option. that would create a ton of work on our build/deploy script.
The only other thing I can think of, is an application structure like the following, where i manage the package outside the application and then symlink back into it :
* /src/application_parts
* /src/application_parts/mymodule
* /src/application_parts/mymodule/tests.py
* /src/application_parts/mymodule/lib
* /src/application
* /src/application/mymodule ( symlinks to /src/application_parts/mymodule/lib )
this looks like something even worse though.
