Lambdarado puts together:

• A Flask app written in Python

• A Docker image that contains the app code and dependencies

• AWS Lambda + AWS Gateway to run the app in the AWS

• Werkzeug to test app locally

It runs the relevant code depending on where it runs.

On the local computer, it runs the a debug server, serving requests to 127.0.0.1 with your app. You can start it directly (python3 main.py) or from a container (docker run ...) to test the app.

In the AWS Cloud the requests are handled with the same app, but in a different way. Lambdarado creates a handler, that is compatible with the combination of API Gateway + Lambda.

$ pip3 install lambdarado 

FROM public.ecr.aws/lambda/python:3.8

# ... here should be the code that creates the image ...

ENTRYPOINT ["python", "main.py"]

You build the image as usual, but the ENTRYPOINT is just a call to a .py file in the project root. And there is no CMD.

import lambdarado import start

def get_app():
  # this function must return WSGI app, e.g. Flask
  from my_app_module import app
  return app 
  
start(get_app)

When starting the Lambda function instance, the get_app method will run once, but the main.py module will be imported twice. Make sure that the app is only created when get_app is called, not when main.py is imported.

In other words, simply running python3 main.py without calling start should NOT do anything heavy and probably should not even declare or import the app.

Running shell command on development machine:

$ python3 main.py

This will start Werkzeug server listening to http://127.0.0.1:5000.

Command-line:

$ docker run -p 5005:5000 docker-image-name

This will start Werkzeug server listening to http://0.0.0.0:5000 (inside the docker). The server is accessible as http://127.0.0.1:5005 from the development (host) machine.

After deploying the same image as a Lambda function, it will serve the requests to the AWS Gateway with your app.

• You should connect the AWS Gateway to your Lambda function. For the function to receive all HTTP requests, you may need to redirect the /{proxy+} route to the function and make lambda:InvokeFunction policy less restrictive

Under the hood:

• The awslambdaric will receive requests from and send requests to the Lambda service
• The apig_wsgi will translate requests received by awslambdaric from the AWS Gateway. So your application doesn't have to handle calls from the gateway directly. For the application, requests will look like normal HTTP