.travis.yml example for Python Lambda microservice testing and deployment

.travis-lambda-python.yml

language: python
services: docker
python: 3.6

env:
  global:

    - AWS_ACCESS_KEY_ID=...
    - AWS_DEFAULT_REGION=ap-southeast-2
    
    - LAMBDA_NAME=...
    - LAMBDA_DESCRIPTION=...
    - LAMBDA_TIMEOUT=...
    - LAMBDA_ROLE=arn:aws:iam::873114526714:role/genericLambdaRole
    - LAMBDA_ALIAS=prod

    - LAMBDA_RUNTIME=python3.6
    - LAMBDA_MODULE=lambda_function
    - LAMBDA_HANDLER=lambda_handler
    - PYPI_USERNAME=tdmalone

    # AWS_SECRET_ACCESS_KEY
    - secure: ...
    
    # PYPI_PASSWORD
    - secure: ...

install:
  - pip install --requirement requirements.txt
  - curl --output ~/.pylint.rc https://raw.githubusercontent.com/tdmalone/dotfiles/master/.pylintrc

script:
  - make lint
  - make test

# Copy the Lambda function as a workaround for a PyPi deployment.
before_deploy:
  - mkdir "${LAMBDA_NAME}"
  - cp lambda_function.py "${LAMBDA_NAME}"/

deploy:

  # Deploy to $LATEST on dev branch. (i.e. don't publish a new version).

  - on:
      branch: dev
    publish: false

    provider: lambda
    function_name: $LAMBDA_NAME
    region: $AWS_DEFAULT_REGION
    role: $LAMBDA_ROLE
    description: $LAMBDA_DESCRIPTION
    runtime: $LAMBDA_RUNTIME
    timeout: $LAMBDA_TIMEOUT
    module_name: $LAMBDA_MODULE
    handler_name: $LAMBDA_HANDLER
    access_key_id: $AWS_ACCESS_KEY_ID
    secret_access_key: $AWS_SECRET_ACCESS_KEY
    skip_cleanup: true

  # Deploy and publish a new version on master branch.

  - on:
      branch: master
    publish: true

    provider: lambda
    function_name: $LAMBDA_NAME
    region: $AWS_DEFAULT_REGION
    role: $LAMBDA_ROLE
    description: $LAMBDA_DESCRIPTION
    runtime: $LAMBDA_RUNTIME
    timeout: $LAMBDA_TIMEOUT
    handler_name: handler
    access_key_id: $AWS_ACCESS_KEY_ID
    secret_access_key: $AWS_SECRET_ACCESS_KEY
    skip_cleanup: true

  # Deploy to pypi on master with a new tag.

  - on:
      branch: master
      tags: true

    provider: pypi
    username: $PYPI_USERNAME
    password: $PYPI_PASSWORD
    skip_cleanup: true

after_deploy:

  # Set a Lambda alias to the most recently deployed version.
  - if [ "master" = "${TRAVIS_BRANCH}" ]; then
      pip install awscli --upgrade --user;
      export MOST_RECENT=$(aws lambda list-versions-by-function --function "${LAMBDA_NAME}" --max-items 10000 | node -e "let stdin=''; process.stdin.on('data',(chunk)=>{stdin+=chunk}).on('end',()=>{console.log(JSON.parse(stdin).Versions.pop().Version)})");
      aws lambda update-alias --function-name "${LAMBDA_NAME}" --name "${LAMBDA_ALIAS}" --function-version "${MOST_RECENT}";
    fi;

notifications:
  email: false
  webhooks:
    urls: https://api.tm.id.au/v2/travis/jobStatus
  slack:
    on_start: always
    rooms:
      - secure: ...