First of all, clone the record api repo
git clone https://github.com/data-apis/python-record-api.git
And create a new conda env for installing all the notebook's dependencies and the record api tooling.
conda create -n record-api python=3.8
conda activate record-api
cd python-record-api
pip install record_api
We have an example of a notebook under /notebooks/data
, so we need to transform it
to a python file so the record api tooling work.
cd notebooks
python notebook_to_python.py data
A new folder named scripts
under the notebooks
directory will be created with
all the files converted into python files.
Then, run the file run_record_api.sh
to generate the inferred API calls in each notebook.
The output will be under infer_apis
directory as a json per notebook file.