Python API for checking correctness & knowledge tracing for Koconut (now Codeitz). Benji Xie (bxie@uw.edu)
The Koconut API runs on Python 3 (tested on 3.7.3, but 3.6 should work). All Python library dependencies (namely Flask, NumPy, Pandas) are listed in the requirements.txt file. Ensure you have Python3 running and have downloaded all dependencies (via pip). If you don't want to make global changes to your Python configuration, consider creating a virtual environment (venv) for the Koconut API.
- Clone the repo.
- In the parent directory of the repo, type the following command:
export FLASK_APP=run_python.py. This command sets the environment variable and tells Flask to run the run_python.py file on execution.- optional: set the FLASK_ENV variable to development to have the server reload on change
export FLASK_ENV=development
- optional: set the FLASK_ENV variable to development to have the server reload on change
- Run flask with the following command:
flask run. You should see a response specifying that Flask is serving the app. - To verify the app is running, go to 127.0.0.1:5000 and see a very exciting "Hello python service!" message. You should also see a 200 response in your console where Flask is running.
To do something more interesting, let's execute a POST request (e.g w/ Postman)! to the /bkt endpoint (127.0.0.1:5000/bkt) with the JSON object in example_bkt_data.json in the body of the request. After doing that, you should see another 200 response on your console and a request body which includes a "pkNew" attribute being set ot 0.3383...
Congratulations! You've run the koconut API locally. 👏🏽
Heroku is set up to automatically deploy any new code pushed to the master branch to codeitz.herokuapp.com.
The Procfile is already setup to use gunicorn, so this code should just be able to be deployed to Heroku. After a successful deployment, verify that / and /bkt endpoints work (just as was done in running API locally).
- Flask dev server: for local deployment on Flask