Awesome

Cloudbuilder-unittest

This container allows us to run unittests using coverage on our Python 3.7 code with support for Google Cloud Platform environments.

Prerequisites

1. A Python (API) project/repository.
2. Tests that:
   • use the unittest Python module
   • are autodiscoverable through their file naming (e.g. api_server/openapi_server/test/test_blah.py)
   • are runnable from the application root. For VWT this is the api_server/ directory most of the time. Note that the application root is different from the project/repository root.
3. A requirements.txt file that includes all the used dependencies

Usage

One argument is required to run this container: the root directory of a Python application that contains the requirements.txt file. For most of our projects this would be the api_server subdirectory of a Python project.

Typically this container would run as a custom build step in a Cloud Build pipeline after preparing a deployment directory, but before actually deploying it to Google App Engine.

It will copy the provided directory into a temporary directory to run the test in and install dependencies, so it will not pollute the original directory.

# Run python unittests
- name: 'eu.gcr.io/vwt-d-gew1-dat-cloudbuilders/cloudbuilder-unittest'
  args: ['/workspace/my_api_dir/api_server']
  env:
    - BRANCH_NAME=$BRANCH_NAME
    - PROJECT_ID=$PROJECT_ID

Additionally, adding a .coveragerc file to the Python application root is needed to tailor the coverage output. In omitting this .coveragerc file, we'll also get coverage information about venv and test directories which pollute the stats.

[run]
omit = */venv/*, */test/*

Or locally on a dev laptop with /workspace as a mounted volume for some local directory:

docker run -v /Users/user01/git/myapp:/workspace -ti cloudbuilder-unittest /workspace/api_server

Testing

This project uses the Google Container Structure Test with a number of tests included in test_config.yaml. This is part of a custom build step in Cloud Builder.