Skip to content

Latest commit

 

History

History
105 lines (70 loc) · 3.06 KB

README.rst

File metadata and controls

105 lines (70 loc) · 3.06 KB
Raw

github-hooks-server

Custom handling of GitHub pull requests.

It manages the labels for a PR to keep track of the review state for a PR:

  • needs-review
  • needs-changes
  • needs-merge

The GitHub Token needs to have triage permissions to the managed repo, and the account must be a member of the organization to be able to see teams.

The code is public but the package is private. It is not / and should not be published on PyPi.

Also tests are designed to run from Chevah VPN with access to staging Trac instance.

Virtual environment

To create a working virtual environment:

virtualenv -p 3.8 venv  # Last Python version supported by serverless-azure-functions
. venv/bin/activate
pip install poetry
poetry install
nodeenv venv/node -n 17.1.0
. venv/node/bin/activate
npm install -g
npm install -g serverless

# Install the Serverless plugin `serverless-azure-functions`.
# https://www.serverless.com/framework/docs/guides/plugins#installing-plugins
npm install serverless-azure-functions

To activate the virtual environment:

. venv/bin/activate
. venv/node/bin/activate

Running offline

To run offline for testing purposes, once you have a virtual environment:

. venv/bin/activate
. venv/node/bin/activate
serverless offline

Now you can send a POST request to the hook function, or a GET request to the ping function:

http://localhost:7071/api/ping

To expose the offline server running on port 7071 to the Web, you can use PageKite:

pagekite 7071 yourname.pagekite.me

This lets you test the GitHub hooks while easily iterating.

Deployment

Make sure config.ini has an appropriate github-token value.

We can not deploy using Serverless on Azure Functions anymore. The upload of the package succeeds, but updating the Function App does not.

Therefore, we have to deploy using Azure CLI and Azure Functions Core Tools. Serverless is still used, because it generates the proper package.

Install Azure CLI and Azure Functions Core Tools.

Also install pip_tools (via pip) for pip-compile, to generate hashes for requirements.txt, as a workaround to a poetry issue.

Then, in the virtual env:

poetry export -f requirements.txt --output requirements.txt
pip-compile --generate-hashes -o requirements.txt requirements.txt
# Poetry won't pin setuptools, but Azure wants it to prevent tampering.
echo 'setuptools==59.6.0 --hash=sha256:4ce92f1e1f8f01233ee9952c04f6b81d1e02939d6e1b488428154974a4d0783e' >> requirements.txt
serverless package
cd .serverless/
unzip githubhooks.zip
az login
func azure functionapp publish sls-weur-dev-githubhooks --python

Courtesy of this comment.