These stubs are intended to be used with a type checker like mypy
to provide static type checking of python code, as well as analysis and completion in IDEs like PyCharm and VSCode with Pylance.
Supported libraries and applications:
Note that pymel now has very excellent stubs included (more info here).
These are distributed as "stubs-only" python packages, so you can just pip install
whichever packages you need:
pip install types-usd types-houdini types-katana types-mari types-nuke types-opencolorio types-PySide2 types-substance_painter
You only need to do this if your goal is to help improve the stubs. Otherwise, just use pip
,
as explained above.
Building the stubs requires python 3.9 or greater.
You can do this using pipx
:
pipx install nox
Or by creating a virtualenv:
cd cg-stubs
python3 -m venv .venv
. .venv/bin/activate
pip install nox
nox -h
Look at the .env
file within the project that you want to build. Either uncomment the necessary
environment variables and paste in the proper values, or configure your shell environment to set
these variables before the next step, for example using a package manager like rez
.
Replace maya
with the project you want to build:
nox -s 'generate(maya)'
The easiest way to use the stubs while you're devleoping them is to create an editable install.
Simply create a .pth
file in the site-packages directory of the venv where your other deps live:
echo "/path/to/cg-stubs/maya/stubs/" > /path/to/venv/lib/python3.7/site-packages/maya.pth
The name of the .pth file does not matter.
Note that if you're using the mypy daemon, be sure to run dmypy stop
to reread freshly modified stubs.
The USD stubs currently require you to build a special fork of USD, until the necessary changes are merged.
git clone https://github.com/chadrik/USD
git checkout doc-stubs2
python3.9 -m venv .venv
. .venv/bin/activate
pip install PySide6 PyOpenGL
python3.9 build_scripts/build_usd.py --python-docs --docs .build-23.08-py39
Then update the variables in stubgen_usd.sh
and generate as normal.
These must be generated from within the UI, because I could not figure out how to run a standlone interpreter.
import mypy.stubgen;mypy.stubgen.main(['-p', '_substance_painter'])
Then generate as normal to cleanup the stubs.
The Houdini stubs currently use a completely different approach to building which will eventually be ported to nox
. So this may be unapproachable for most users for now.
To publish to pypi.org, first run the nox installation steps from the Generating section, then run
the publish
task (replacing maya
with the package to publish):
nox -s 'publish(maya)'