UTPython Lab

The Colaboratory notebook files for UTPython, the Introduction to Python Programming course at the University of Tokyo. https://sites.google.com/view/ut-python/

Text Book

https://utokyo-ipp.github.io/toc.html

How to set up development environment

Install the Python 3 virtualenv:

virtualenv -p python3 venv
. venv/bin/activate
pip install jupyter prog_edu_assistant_tools

Install Bazel (http://bazel.build) and build the notebooks:

bazel build ...

Build the student notebooks

./build-student.sh

The Jupyter version of the student notebooks can be found in tmp/student, and Colab version of the student notebooks is in tmp/colab_student.

Build the deployment image

docker/build-docker-image.sh

Run the service locally for testing

docker/start-local-server.sh

Deploy the autograder to Google Cloud Run

• Install the Google Cloud SDK, including Go support.

• Authorize and configure the GCP project by running

  gcloud init
  

• Create a Cloud Storage bucket for storing logs

• Enable Google Cloud Run

• Create a OAuth2 client ID and download it in JSON format.

• Create a RSA private key for JWT signing. There are multiple ways of doing that, for example, using the keytool.go from http://github.com/google/prog-edu-assistant:

  cd ../prog-edu-assistant
  go run go/cmd/keytool/keytool.go \
    --action createkey \
    --output_private_key jwt.priv \
    --output_public_key jwt.pub
  

  Another option is to use openssl command:

  openssl genrsa -out jwt.priv 2048
  

  Copy the private key file to Google Cloud storage (any bucket is okay):

  gsutil cp jwt.priv gs://<bucket-name>/
  

• Copy the file docker/cloud-run.env.template to docker/cloud-run.env and fill in the details. You may need to skip the SERVER_URL if you do not know the deployment URL yet.

• Run the deployment command:

  ./docker/deploy-cloud-run.sh
  

• If it was the first time to deploy, find the deployment URL in the Google Cloud console, and update SERVER_URL in docker/cloud-run.env. You also need to edit the Colab submission snippet (preamble.py), update the server URL and rebuild the student Colab notebooks.

How to run the integration tests

This repository includes a script for quickly sending a few requests to a locally running instance of autograder backend to verify that it can handle canonical and non-canonical submissions correctly. The contents of the submission is automatically constructed from two parts:

• The student notebook provides the template of the ipynb file (because each submission uploads the full .ipynb file that student is workin on).

• The master notebook provides submission code snippets: canonical solutions are those marked with %%solution magic, and non-canonical submissions are those marked with %%submission. The canonical solutions are expected to result in passed tests. For non-canonical submission snippets, the script expect to find a python snippet with assertions in a cell immediately following the %%submission snippet.

Run the integration tests:

# Start the local autograder backend (Docker based):
docker/start-local-server.sh

Wait until the backend is ready. It will build the Docker image and report the URL it is listening at. The test assumes the URL is `http://localhost:8000'.

# Build the student notebooks and extract them to ./tmp/student:
./build-student.sh
# Run the integration tests:
./test-all.sh

Note, that the above command prints the outcomes of submissions, which occasionally contains error messages. This is by design, as the integration test sends not only correct submissions, but incorrect submissions too. To understand whether the integration test itself passed or failed, you need to look for the final OK message printed by the script and check the status code of the script.