English / 日本語

cdk2-python-library-layer turns your private Python package into a Lambda layer. This library provides a CDK Construct that you can incorporate into your CDK script.

NOTE: The branch for CDK v1 (main) is no longer maintained.

This library turns a private Python package that pip cannot resolve into an AWS Lambda layer.

Please run the following command,

npm install https://github.com/kikuomax/cdk-python-library-layer.git#v0.2.1-v2

Just import PythonLibraryLayer and new it. PythonLibraryLayer implements ILayerVersion.

Here is an example that makes a Lambda layer from a package defined in a lambda/libexample folder.

import * as path from 'path';
import { aws_lambda as lambda } from 'aws-cdk-lib';
import { Construct } from 'constructs';

import { PythonLibraryLayer } from 'cdk2-python-library-layer';

class YourCdkConstruct extends Construct {
    constructor(scope: Construct, id: string) {
        super(scope, id);
        this.layer = new PythonLibraryLayer(this, 'libexample', {
            description: 'Example Lambda layer',
            runtime: lambda.Runtime.PYTHON_3_8,
            entry: path.resolve('lambda', 'libexample'),
            compatibleArchitectures: [
                lambda.Architecture.ARM_64,
                lambda.Architecture.X86_64,
            ],
        });
    }
}

So far, a package must be configured for setuptools and have a structure similar to the following (src/ layout),

your_package/
  pyproject.toml
  setup.cfg
  src/
    your_package/

There is a working example in the example folder.

I had a project that had a lot of Python Lambda functions that shared some code among them. Not to duplicate the shared code, I packaged them as a Python package and planned to reuse it as a Lambda layer. Since the package was specific to the project, I did not want to publish the package to any package repository.

First, I tried PythonLayerVersion, but it did not work as I intended; more preceisely, I could not figure out how to achieve what I wanted to do with it. As far as I looked into the source code, it looked that it just downloads packages listed in requirements.txt and copies them under a python folder. It did not look that it handles any scripts in an entry folder.

Thus, I had to somehow make a Lambda layer from my private package.

If the platform of your machine running Docker is different from the target platform (compatibleArchitectures) of the layer, you may face an error message similar to the following:

WARNING: The requested image's platform (linux/arm64) does not match the detected host platform (linux/amd64/v3) and no specific platform was requested
exec /usr/bin/bash: exec format error

/home/ubuntu/cdk-python-library-layer/example/node_modules/aws-cdk-lib/core/lib/asset-staging.ts:395
      throw new Error(`Failed to bundle asset ${this.node.path}, bundle output is located at ${bundleErrorDir}: ${err}`);

If you are building a layer compatible with multiple platforms, change the order of compatibleArchitectures so that the first item matches your machine's platform; e.g., suppose your machine is based on x86_64:

compatibleArchitectures: [
    lambda.Architecture.X86_64,
    lambda.Architecture.ARM_64,
]

Or allow Docker to build a cross-platform image. How to do it depends on your environment, though, this page would be helpful. On Ubuntu 22.04, I was able to solve this issue by installing qemu-user-static.

sudo apt-get install qemu-user-static