image-processor

Image Processor example

This is a demo that uses Python Pillow within AWS Lambda to resize an image. The Pillow library is built and bundled via Serephem.

Requirements

• An AWS account
• Docker

Get started

1. Once inside this example's directory, install Serephem.

   $ npm i

2. Configure the serverless.yml file with your own settings (i.e. region, stage, profile). Defaults are:

   • region: us-east-1
   • stage: dev
   • profile: Ommited, thus points to default

Resources

S3 Bucket

An S3 bucket named srphm-image-processor is created on deployment. Here you will upload your images to process. You must upload them to the original folder. Either you can create it or will be created for you when running the test Lambda (see below).

Lambdas

This example features 2 Lambdas:

• processor: This processor takes an image uploaded to S3 and resizes it to 128, 192 and 320 pixels, keeping the width/height ratio. It then uploads each to a different folder in the S3 bucket.
• test: Uploads a default provided image (tokyo.jpg) to the original directory (creating it if it doesn't exist). This, as a result, triggers the processor Lambda.

Deploy

$ sls deploy -v

The first time you run this, you will see how Docker builds the image-processor.zip file. This file contains the Pillow library. On subsequent deployments, the zip file will be pulled from local cache unless it is deleted or you decide to rebuild via the nocache option.

Remember to run sls remove when done with the demo.

Test

1. Log in to the AWS console

2. Here you can choose 2 paths:

   • Upload an image manually

     a. Go to AWS S3 and open the srphm-image-processor bucket

     b. Create the original folder if it doesn't exist

     c. Open the folder and upload an image

   • Use the test Lambda

     a. Go to the test Lambda in the AWS console

     b. Click the Test button to run it. Use {} as test event.

3. In the bucket, you will notice 3 new folders: w128, w192 and w320. Each will contain resized versions of the uploaded images.