The Python Lambda Logistics (pll
) library is designed to assist with an automated deployment for Python 3
AWS Lambda functions.
As you will no doubt discover, as soon as you require more than the basic libraries, you need to start creating what are known as 'deployment packages' (zip files with all dependencies) for pushing even basic scripts up to AWS.
These packages require that the libraries are compiled for amazon's version of linux (which is executing the code on your behalf).
Not helpful if, like me, your development machine isn't running AWS Linux...
Some of the main libraries this has helped me so far with packaging:
- SQLAlchemy
- Pandas
- NumPy
pll
aims to simplify the above by automating the build of your requirements.txt
file in a Docker
image of
awslinux
.
- Create working directories in your current project
Build
-- contains compiledsite-packages
for future re-useDist
-- outputted lambdazip
for upload to AWS
- Check for a local docker image of
saracen9/python-lambda
or pull down - Mount the following locations into the docker image
Build
--> /tmp/app/psite-packagesDist
--> /tmp/app/zipped-package- YourAppplicationRoot --> /app
- Run the container and compile
- Shutdown and £profit
- (Opt) push code to AWS (Coming soon)
The original Dockerfile used to create the image is contained within the package source code for those who want to hack their own.
To manually run and configure the docker container use the below command.
docker run --rm --name py-lamb-compiler \
-v ~/my/app:/app \
-v ~/my/app/dist:/tmp/app/dist \
-v ~/my/app/build:/tmp/app/build \
saracen9/python-lambda
Alternatively install the library and put a pll.yaml
in the root of your application.
packaging:
app_root: '/my/app'
build_path: '/my/app/build'
dist_path: '/my/app/dist'
You can then run python -m pll
and package the application from any script. As an additional option pass in the
--config
switch to define your own config file.
In addition, relative paths to the cwd
are accepted. These are then converted to absolute paths at runtime E.g.
packaging:
app_root: '' # current directory
build_path: 'build/' # current directory/build
dist_path: 'dist/' # current directory/dist