Awesome

CLA bot

CLA bot is a web server which handles GitHub webhook events to check whether PR authors have signed the CLA. It uses Google Sheets as a data source.

Configuration

GitHub Webhook setup

In your repository, go to Settings -> Webhooks, then choose Add webhook. In the Payload URL field type the URL address of this bot with /webhook endpoint. Example: https://example.com/webhook.

As a content type choose application/json. You can either choose the Send me everything option or manually select individuals events to send. You need to select at least Pull requests and Issue comments.

NOTE: If using this across an organization you can also set the CLA Bot up at the org level which will then mean any events from your repositories will be sent back to the bot.

Google Sheets setup

The sheets must contain columns with each row containing a GitHub login of the users that have signed the CLA. You can configure which columns the bot should look at.

Service health checks

When running the service exposes a /health endpoint which can be pinged to verify liveliness of the service.

How to generate the GitHub token?

Setup a GitHub user which is specifically for CLA signing activites
Navigate to Profile -> Settings -> Developer Settings for this user
Generate a new "classic" token

For the scope if only using for public repos select public_repo. If using for all repositories in an org select the repo:* scope.

IMPORTANT: You will then need to ensure that the user you have allocated for CLA activites is given push access to any repositories you want monitored - without this level of access the bot will not work.

How to generate the OAuth token?

To grant the bot access to Google Sheets we are leveraging a GCP Service Account authentication flow. To do this step you will need access to a GCP Project.

In your GCP Project navigate to the APIs dashboard: https://console.cloud.google.com/apis/dashboard
Navigate through Credentials -> + Create Credentials -> Service Account
Give it a useful name like “clabot”
Skip granting permissions -> we do not need permission in the GCP Project for this!
Select the newly created Service Account and then select Keys
Select Add Key -> JSON

Download the key locally to a file and reference it in your config at oathCredPath.

IMPORTANT: Within the GCP Project you will need to ensure that both the sheets ad drive APIs are enabled so that the service-account can access them.

Granting access to your spreadsheets?

To grant access to the spreadsheets you will need the email address of the service-account and to then grant viewer access on the spreadsheet to this email address.

Setting up the config file

Once the setup is done, we can fill our configuration file. An example configuration is shown here:

port = 8080
host = localhost

github {
  # a token from the bot account, get at https://github.com/settings/tokens
  token = GITHUB_TOKEN
}

oathCredPath = PATH_TO_CRED_JSON

cla {
  internalCLA {
    # id of the spreadsheet (the one from the spreadsheet URL)
    spreadsheetId = GOOGLE_SPREADSHEET_ID
    sheetName     = GOOGLE_SPREADSHEET_NAME
    # columns containing GitHub logins
    columns        = [ A ]
  }
  individualCLA {
    # id of the spreadsheet (the one from the spreadsheet URL)
    spreadsheetId = GOOGLE_SPREADSHEET_ID
    sheetName     = GOOGLE_SPREADSHEET_NAME
    # columns containing GitHub logins
    columns        = [ A ]
  }
  corporateCLA {
    # id of the spreadsheet (the one from the spreadsheet URL)
    spreadsheetId = GOOGLE_SPREADSHEET_ID
    sheetName     = GOOGLE_SPREADSHEET_NAME
    # columns containing GitHub logins
    columns        = [ A ]
  }
  # list of GitHub logins which do not require signing the CLA
  peopleToIgnore = [ "scala-steward" ]
}

Running

sbt assembly
java -Dconfig.file=application.conf \
  -jar clabot/target/scala-2.13/cla-bot-0.3.0.jar

We also produce a Docker image which can be found here: https://hub.docker.com/r/snowplow/cla-bot/tags

docker run \
  -p 8080:8080 \
  -v /application.conf:/application.conf \
  -v /oauth_creds.json:/oauth_creds.json \
  --network=host \
  snowplow/cla-bot:latest \
  -Dconfig.file=application.conf

How the bot algorithm works

Pull Request is opened

If user submitting the PR is a collaborator (this includes members of the organization), the bot ignores the PR.
If the user is a collaborator (and to combat failures in listing collaborators) add them to the internalCLA sheet.
If the user is not a collaborator
- and is in the peopleToIgnore list, or
- and has signed the CLA, the bot adds a cla:yes label
If the user is not a collaborator, is not in the peopleToIgnore list and has not signed the CLA, the bot adds a cla:no label and posts a comment reminding the user to sign the CLA.

The bot then listens to incoming comments in the PR. If the author of the comment is also the author of the PR ("pinging"), then the bot checks the CLA again. If the CLA is now signed, the bot posts a comment with a thank you message. Otherwise it ignores the comment.

Copyright and license

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this software except in compliance with the License.

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.