Deploy Python3 function with Serverless Framework on AWS Lambda

• I misunderstood the release date, and it was the next day.

Introduction

--This article is Python Part 3 Advent Calendar 2019 is the 16th day. --This is a memo until you deploy and publish a Python3 function with Serverless Framework on AWS Lambda.

Premise

--The npm command can be used --There is an IAM user for serverless framework --You can refer to the official this area (Administrator Access is recommended). --Docker can be used

Serverless Framework installation and configuration

Installation is one shot of npm install.

$ npm install serverless

Make sure the path is in --version.

$ serverless --version
Framework Core: 1.59.3
Plugin: 3.2.5
SDK: 2.2.1
Components Core: 1.1.2
Components CLI: 1.4.0

Set the IAM account for Serverless Framewrok created in advance with serverless config credentials.

serverless config credentials --provider aws --key foo --secret bar

This completes the preparations.

Template generation and deploy

Generate a template project for Python3.

serverless create --template aws-python3 --name aws-lambda-hello-python --path aws-lambda-hello-python

The following files are actually generated.

$ cd aws-lambda-hello-python

$ tree
.
├── handler.py
└── serverless.yml

0 directories, 2 files

In the above command, the main method is not generated, but this time I wanted to make it work locally, so I added it.

handler.py

import json


def hello(event, context):
    body = {
        "message": "Go Serverless v1.0! Your function executed successfully!",
        "input": event
    }

    response = {
        "statusCode": 200,
        "body": json.dumps(body)
    }

    return response

    # Use this code if you don't use the http event with the LAMBDA-PROXY
    # integration
    """
    return {
        "message": "Go Serverless v1.0! Your function executed successfully!",
        "event": event
    }
    """

if __name__  == "__main__":
    print(json.dumps(hello("","")))

There are a lot of comments, but the minimum required is: (Since you haven't specified a region, it will be deployed to the default us-east-1)

serverless.yml

service: aws-lambda-hello-python

provider:
  name: aws
  runtime: python3.8

functions:
  hello:
    handler: handler.hello

Let's move it a little with Local.

$ pyenv local 3.8.0
$ python3 handler.py
{"statusCode": 200, "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": \"\"}"}

It worked. Now, let's deploy and check the operation.

$ serverless deploy -v
(Message omitted)

$ serverless invoke -f hello
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}

It seems that it was able to deploy properly.

Use of external libraries

--I would like to be able to use an external library. ――I would like to introduce numpy, but the native library needs to be built for each environment, and it is a little troublesome when using it with Lambda normally, but using Serverless Framework makes it relatively easy.

Make the venv environment available, install numpy and freeze it.

$ python3 -m venv .venv
$ .venv/bin/pip install numpy
$ .venv/bin/pip freeze > requirements.txt

Add the process that uses numpy appropriately. (The mean will change).

handler.py

import json
import numpy as np


def hello(event, context):
    arr = range(1, 100+1)
    mean = np.mean(arr)

    body = {
        "message": "Go Serverless v1.0! Your function executed successfully!",
        "input": event,
        "mean" : mean
    }

    response = {
        "statusCode": 200,
        "body": json.dumps(body)
    }

    return response

    # Use this code if you don't use the http event with the LAMBDA-PROXY
    # integration
    """
    return {
        "message": "Go Serverless v1.0! Your function executed successfully!",
        "event": event
    }
    """

if __name__  == "__main__":
    print(json.dumps(hello("","")))

When executed locally, it looks like this.

$ .venv/bin/python3 handler.py
{"statusCode": 200, "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": \"\", \"mean\": 50.5}"}

Install a plugin that allows serverless to use requirements.

npm install serverless-python-requirements

Add to serverless.yml (add plugins and below)

service: aws-lambda-hello-python

provider:
  name: aws
  runtime: python3.8

functions:
  hello:
    handler: handler.hello

plugins:
  - serverless-python-requirements

custom:
  pythonRequirements:
    dockerizePip: true

package:
  include:
    - handler.py
  exclude:
    - '**'

Try deploying (docker will run a numpy build and deploy it)

$ serverless deploy -v

Let's check the operation.

$ serverless invoke -f hello
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}, \"mean\": 50.5}"
}

Numpy is working properly! Let's have a fun Python & Serverless life!