Lambda Packager

Another lambda packager. Native tooling for python and node lambda packages (no docker).

Why? Python and nodejs are interpreted languages that don't always need to be built on the same architecture targetting the deployment. Lambda functions are (typically) small but not straightforward to build. Numerous tools exist to help you package lambdas but they often have prerequisites that are cumbersome or may be blocked by corporate policies (docker availability).

Lambda Packager does some things differently that solve common problems. It does the minimum steps for packaging and plugs into existing CI/CD workflows for user's just getting started with AWS Lambda.

Features

Fast, runs on your machine and doesn't require any other tools
Creates a unique name for your package containing the git sha and a distinct fingerprint to solve the lambda update problem with cloudformation
Can easily run in CI/CD pipelines
Installs production dependencies only

Quick Start

npm i -g @mneil/lambda-packager
lambda-packager

Usage

Running lambda-package might be enough. By default lambda-package will try to detect your project and "Just Work". However, not all projects are created equal.

See Default Project Structure

Default Project Structure

You can structure your Lambda project any way you wish. However, some structures lend themselves better to different programming languages. Because of this the recommended structure for you project is as follows:

Root: The root of the project
- <manifest>: A manifest file, requirements.txt, package.json, etc...
- lambda: A directory named lambda
  - src: A directory containing source code

Options

Usage: lambda-packager [options]

Options:
  -V, --version              output the version number
  -d, --debug                output extra debugging. Equivilent of setting DEBUG=lp*
  --base-dir <dir>           root folder to use for manifest discovery. Defaults to cwd (default:
                             "/path/to/project")
  -p, --package-dir <dir>    source folder of your lambda package (default: "lambda")
  -i, --install-dir <dir>    relative path in package directory to install dependencies (default:
                             "src")
  -m, --manifest <filepath>  path to your manifest file
  -p, --output <dir>         name of the archive to create (default: "dist.zip")
  -b, --bucket <s3_bucket>   upload package to the s3 bucket
  --prefix <s3_path>         a path prefix on s3 to place the archive
  -h, --help                 display help for command

Why?

This folder structure has advantages for any target language. It can support multiple lambdas with shared or separate dependencies.

Import paths in Python, in particular, can be difficult or confusing to debug. What might work locally doesn't work in Lambda. When using Python it is suggested that you put your code in the lambda/src directory and use import paths like import function from src.module. When running tests with tools like pytest add ./lambda to your PYTHONPATH.

#tests/conftest.py
import sys
import pathlib
sys.path.insert(0, pathlib.Path(__file__).parent.joinpath("lambda").resolve())

Directory names are arbitrary but you should nest it 2 folders deep

Debugging

This package uses debug. Enable debug logging with

export DEBUG=lp:*

or pass the --debug flag (-d) to the cli.