Optim

Automatically optimize your images on S3 with the magic of AWS Lambda.

imagemin-on-aws-lambda is a fork of Optim.

It is a super-simple Lambda function that can listen to an S3 bucket for uploads, and runs everything it can through imagemin.

Currently supported:

PNG
GIF
JPG
SVG
WEBP

Setup

Clone this repo:

git clone https://github.com/zeroasterisk/optim.git

Install Node packages:

npm install

Configure AWS credientials if they are not already set in your environment:

aws configure

Create an IAM role for Optim to use. It needs the following permissions on all the S3 buckets you want to use (allowing these operations on ARN * is easiest to start with):

getObject
putObject
putObjectAcl

Find the ARN for this role. It looks something like arn:aws:iam::111111111111:role/YOURROLEHERE.

Create with claudia for the first time:

npm run create --role arn:aws:iam::111111111111:role/YOURROLEHERE --region us-east-1 --memory 1536 --timeout 300 --version development --handler bin.default --name image-optim

You can now hook this function up to any S3 bucket you like in the management console. Easiest way is to follow AWS's guide

Scripts

Extract ImageMagick:

npm run extract-imagemagick

Update with claudia:

npm run update

Test:

npm run test

Triggers

The script expects an s3 trigger. You can invoke this directly and send in your own trigger event/data if you make it look like this:

{
  "Records": [ {
    "s3": {
      "object": {
        "key": "somefolder/somefile.png"
      },
      "bucket": {
        "name": "c.eltoro.com"
      }
    }
  } ]
}

Building on OSX or Windows?

AWS lambda runs on linux. Sometimes packages rely on binaries which have to be for the runtime. There are other ways to accomplish this, but here is one of the easiest.

rm -rf node_modules
docker run -v "${PWD}:/var/task" "/Users/<name>/.aws:/root/.aws" lambci/lambda:build "npm i && npm run update"

NOTE: this should work, but I have had issues with it. WIP

Acknowledgements

Big credit goes to Optim the original project, which is 80% of the code.

Also, bigger credit to the imagemin project which does the heavy lifting.

zeroasterisk / optim

readme