preprocess.py

import json

from tqdm import tqdm
import typer

def preprocess(data_path, processed_data_path):
    with open(data_path) as f:
        data = json.loads(f.read())

    with open(processed_data_path, "w") as f:
        for grant in tqdm(data["grants"]):
            if any([org["name"] == "The Wellcome Trust" for org in grant["fundingOrganization"]]):
                f.write(json.dumps(grant) + "\n")

if __name__ == "__main__":
    typer.run(preprocess)