Python - PIP : Is it possible to specify architecture?

Question

I am building a Python Lambda function code to be executed on AWS (Linux).
My Dev machine is not Linux.
If I do pip install, I am getting Windows/MacOS related modules.

Is there a way to fetch Linux based dependencies always? Such as a CLI argument for target architecture.

Answer 1

I just had the exact same problem with Lambda, Python, and trying to get Pillow to work. After much searching around, I came across this article that explained the problem, and provided a great solution.

https://medium.freecodecamp.org/escaping-lambda-function-hell-using-docker-40b187ec1e48

Basically, you can run docker on your Mac, which allows you to run a Linux instance. Then using pip, install whatever library you want to use. You can then include those files in your ZIP file and upload it to Lambda. Works like a charm.

These docker images approximate the AWS Lambda environment pretty accurately. You can use them to create packages.