Install pip in Serverless Framework and AWS Lambda with Python environment

How to deploy AWS Lambda Python scripts that require pip installation with Serverless Framework.

I wrote about the AWS Lambda Ruby script that requires gem installation in the Last time article.

Procedure outline

You can easily do it by inserting a plugin.

1. serverless plugin install -n serverless-python-requirements
2. Create requirements.txt
3. After that, if you deploy normally, it will do various things without permission

Detailed procedure

Serverless Framework service creation

$ serverless create --template aws-python3
Serverless: Generating boilerplate...
 _______                             __
|   _   .-----.----.--.--.-----.----|  .-----.-----.-----.
|   |___|  -__|   _|  |  |  -__|   _|  |  -__|__ --|__ --|
|____   |_____|__|  \___/|_____|__| |__|_____|_____|_____|
|   |   |             The Serverless Application Framework
|       |                           serverless.com, v2.16.1
 -------'

Serverless: Successfully generated boilerplate for template: "aws-python3"
Serverless: NOTE: Please update the "service" property in serverless.yml with your service name

Three files will be generated.

.gitignore
handler.py
serverless.yml

Plugin installation

Install the plugin serverless-python-requirements.

$ serverless plugin install -n serverless-python-requirements

The following files and directories will increase.

node_modules
package.json
package-lock.json

Since the Serverless Framework is implemented in Node.js, it seems that node_modules and package.json exist even though it is a Python project.

Source code

serverless.yml

serverless.yml has the following contents. The description of plugins is added without permission when the plugin is installed.

serverless.yml

service: sample

frameworkVersion: '2'

provider:
  name: aws
  runtime: python3.8
  region: ap-northeast-1

functions:
  hello:
    handler: handler.hello

plugins:
  - serverless-python-requirements

requirements.txt

Let's use a library called jpholiday as a sample library. A library for determining Japanese holidays.

Create requirements.txt and make it as follows. This is a file with only one line of library name.

jpholiday

Let's use a library called jpholiday as a sample. A library for determining Japanese holidays.

Python source code

handler.py

import datetime
import jpholiday

def hello(event, context):
    holidayName = jpholiday.is_holiday_name(datetime.date(2021, 8, 8))
    print(holidayName) #To CloudWatch"Mountain day"Is written out

Deploy

If you create it so far and then deploy it with the serverless command, not only Lambda itself but also the serverless command will automatically create an image with the library installed and upload it as a layer of AWS Lambda.

$ serverless deploy -v

Run

If you look at Lambda in the AWS Management Console after deployment, it looks like this:

Unlike Ruby, it seems that the library is stored directly in Lambda instead of Layer.

Link

-(Ruby version) Gem installation in Serverless Framework and AWS Lambda with Ruby environment