This is a collection of useful NLP tools as Python wheels built for a Windows isolated environment.
Wheels are separated into multiple zipped directories. Go to Releases and download whichever wheel directories fit your use case.
If you chose data_wheels.zip then extract the zip file and run
python -m pip install --no-index --find-links=data_wheels -r requirements.txt, where there exists
a requirements.txt that specifies the required packages from the wheel directory (see
wheel_requirements/
to determine which packages are provided by each zip file).
Note that you can provide multiple paths to --find-links, which allows you to combine wheel
directories if needed.
New libraries can be added to the appropriate *_requirements.txt file in
wheel_requirements/,
or an entirely new wheels zip file can be generated by creating a new *_requirements.txt file
if needed.
SpaCy models are usually downloaded through python -m spacy download <model>, but in an isolated environment
this option isn't available.
SpaCy uses pip to download models, so we can instead add a GitHub link to a build requirements file where we want a spaCy model available like:
https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.7.1/en_core_web_sm-3.7.1.tar.gz#egg=en_core_web_sm
If you want the en_core_web_sm model available in your isolated environment, you can instead add the line
en_core_web_sm==3.7.1 to your environment's requirements file after following the steps in Installation.
SpaCy will be able to load and import the model like normal.