Python-lambda is a toolset for developing and deploying serverless Python code in AWS Lambda.

A call for contributors

With python-lambda and pytube both continuing to gain momentum, I'm calling for contributors to help build out new features, review pull requests, fix bugs, and maintain overall code quality. If you're interested, please email me at nficano[at]gmail.com.

Description

AWS Lambda is a service that allows you to write Python, Java, or Node.js code that gets executed in response to events like http requests or files uploaded to S3.

Working with Lambda is relatively easy, but the process of bundling and deploying your code is not as simple as it could be.

The Python-Lambda library takes away the guess work of developing your Python-Lambda services by providing you a toolset to streamline the annoying parts.

Requirements

• Python 2.7, >= 3.13 (At the time of writing this, these are the Python runtimes supported by AWS Lambda).
• Pip (~8.1.1)
• Virtualenv (~15.0.0)
• Virtualenvwrapper (~4.7.1)

Getting Started

First, you must create an IAM Role on your AWS account called lambda_basic_execution with the LambdaBasicExecution policy attached.

On your computer, create a new virtualenv and project folder.

$ mkvirtualenv pylambda
(pylambda) $ mkdir pylambda

Next, download Python-Lambda using pip via pypi.

(pylambda) $ pip install python-lambda

From your pylambda directory, run the following to bootstrap your project.

(pylambda) $ lambda init

This will create the following files: event.json, __init__.py, service.py, and config.yaml.

Let's begin by opening config.yaml in the text editor of your choice. For the purpose of this tutorial, the only required information is aws_access_key_id and aws_secret_access_key. You can find these by logging into the AWS management console.

Next let's open service.py, in here you'll find the following function:

def handler(event, context):
    # Your code goes here!
    e = event.get('e')
    pi = event.get('pi')
    return e + pi

This is the handler function; this is the function AWS Lambda will invoke in response to an event. You will notice that in the sample code e and pi are values in a dict. AWS Lambda uses the event parameter to pass in event data to the handler.

So if, for example, your function is responding to an http request, event will be the POST JSON data and if your function returns something, the contents will be in your http response payload.

Next let's open the event.json file:

{
  "pi": 3.14,
  "e": 2.718
}

Here you'll find the values of e and pi that are being referenced in the sample code.

If you now try and run:

(pylambda) $ lambda invoke -v

You will get:

# 5.858
# execution time: 0.00000310s
# function execution timeout: 15s

As you probably put together, the lambda invoke command grabs the values stored in the event.json file and passes them to your function.

The event.json file should help you develop your Lambda service locally. You can specify an alternate event.json file by passing the --event-file=<filename>.json argument to lambda invoke.

When you're ready to deploy your code to Lambda simply run:

(pylambda) $ lambda deploy

The deploy script will evaluate your virtualenv and identify your project dependencies. It will package these up along with your handler function to a zip file that it then uploads to AWS Lambda.

You can now log into the AWS Lambda management console to verify the code deployed successfully.

Wiring to an API endpoint

If you're looking to develop a simple microservice you can easily wire your function up to an http endpoint.

Begin by navigating to your AWS Lambda management console and clicking on your function. Click the API Endpoints tab and click "Add API endpoint".

Under API endpoint type select "API Gateway".

Next change Method to POST and Security to "Open" and click submit (NOTE: you should secure this for use in production, open security is used for demo purposes).

At last you need to change the return value of the function to comply with the standard defined for the API Gateway endpoint, the function should now look like this:

def handler(event, context):
    # Your code goes here!
    e = event.get('e')
    pi = event.get('pi')
    return {
        "statusCode": 200,
        "headers": { "Content-Type": "application/json"},
        "body": e + pi
    }

Now try and run:

$ curl --header "Content-Type:application/json" \
       --request POST \
       --data '{"pi": 3.14, "e": 2.718}' \
       https://<API endpoint URL>
# 5.8580000000000005

Environment Variables

Lambda functions support environment variables. In order to set environment variables for your deployed code to use, you can configure them in config.yaml. To load the value for the environment variable at the time of deployment (instead of hard coding them in your configuration file), you can use local environment values (see 'env3' in example code below).

environment_variables:
  env1: foo
  env2: baz
  env3: ${LOCAL_ENVIRONMENT_VARIABLE_NAME}

This would create environment variables in the lambda instance upon deploy. If your functions don't need environment variables, simply leave this section out of your config.

Uploading to S3

You may find that you do not need the toolkit to fully deploy your Lambda or that your code bundle is too large to upload via the API. You can use the upload command to send the bundle to an S3 bucket of your choosing. Before doing this, you will need to set the following variables in config.yaml:

role: basic_s3_upload
bucket_name: 'example-bucket'
s3_key_prefix: 'path/to/file/'

Your role must have s3:PutObject permission on the bucket/key that you specify for the upload to work properly. Once you have that set, you can execute lambda upload to initiate the transfer.

Deploying via S3

You can also choose to use S3 as your source for Lambda deployments. This can be done by issuing lambda deploy-s3 with the same variables/AWS permissions you'd set for executing the upload command.

Building Docker Images for Lambda

lambda build-image

You will need to provide your own Dockerfile compatible with Python Lambda for builds, and also Docker of course. For builds build variables can be provided in the nested dictionary like so. Do not use shorthand flages (like -t instead of --tag). For flags that normally don't take in a value please provide a boolean. Use quotes around the variable names and values for best results. --provenance=false is automatically provided in the build commmand, no need to add it in yaml. If you wish to build and immediately push to ECR you can add "--push": "true" and provide the full ECR URI in the --tag

image_build_variables:
  "--tag": "my_lambda"
  "--platform": linux/amd64
  "--no-cache": "false"
  "build_path": "."

Tagging Images for Lambda

lambda tag-image

For images to be used in AWS Lambda they will need to be tagged with the ECR name and port.

For tagging you'll need to add these folllowing variables to your yaml.

aws_account_id: 000123456789
ecr_repository: trainer/training-things/test-repo
lambda_image_tag: v0.0.1

These will be used to build the full ECR URI. lambda_image_tag can be provided via command line with --lambda-image-tag When done like this it will overwrite the value in the YAML.

Pushing Images to ECR for Lambda

lambda push-image

For pushing to ECR you will likely need to authenticate your AWS account with ECR first. There are 3 ways the URI can be built here and will follow this order.

1. Providing --lambda-image-uri while calling lambda push-image
2. Adding "lambda_image_uri": "000123456789..." in your yaml.
3. As long as you have aws_account_id and ecr_repository in your yaml, you can add --lambda-image-tag in the lambda deploy-image command.

Deploying Docker Images to Lambda

lambda deploy-image

Finally after your image is in ECR you can deploy it to Lambda. There 3 ways the URI can be built here and will follow this order.

1. Providing --lambda-image-uri while calling lambda push-image
2. Adding "lambda_image_uri": "000123456789..." in your yaml.
3. As long as you have aws_account_id and ecr_repository in your yaml, you can add --lambda-image-tag in the lambda deploy-image command.

Shortcutting Docker Images for Lambda

You condense the building, tagging, and pushing into one step by adding additional variables in your yaml.

image_build_variables:
  "--tag": "00012345678.dkr.ecr.us-east.9.amazonaws.com/trainer/training-things/my-lambda:latest"
  "--platform": linux/amd64
  "--no-cache": "false"
  "build_path": "."
  "--push": "true

Development

Development of "python-lambda" is facilitated exclusively on GitHub. Contributions in the form of patches, tests and feature creation and/or requests are very welcome and highly encouraged. Please open an issue if this tool does not function as you'd expect.

Environment Setup

1. Install pipenv
2. Install direnv
3. Install Precommit (optional but preferred)
4. cd into the project and enter "direnv allow" when prompted. This will begin installing all the development dependancies.
5. If you installed pre-commit, run pre-commit install inside the project directory to setup the githooks.

Releasing to Pypi

Once you pushed your chances to master, run one of the following:

# If you're installing a major release:
make deploy-major

# If you're installing a minor release:
make deploy-minor

# If you're installing a patch release:
make deploy-patch