ImageMagick layer builder for AWS lambda

Introduction

Run build_zip.sh to build the Docker image and copy a zip file with the layer out of the image.

Usage

Move the image_magick_layer.zip into a directory for your layer in the SAM source tree. Add a LayerVersion to the Resources section of your SAM template.yaml:

Resources:
  ImageMagickLayer:
    Type: AWS::Serverless::LayerVersion
    Properties:
      ContentUri: image_magick_layer/image_magick_layer.zip
      CompatibleRuntimes:
        - python3.7

Add the layer to lambdas that need it in template.yaml:

  ExtractPageImages:
    Type: AWS::Serverless::Function
    Properties:
      CodeUri: extract_images_from_pdf/
      Handler: extract_images_from_pdf.lambda_handler
      Policies:
        - AWSLambdaExecute
        - Version: "2012-10-17"
          Statement:
            - Effect: Allow
              Action:
                - "s3:GetObject"
              Resource: "arn:aws:s3:::*"
      Runtime: python3.8
      Layers:
        - !Ref ImageMagickLayer
      Environment:
        Variables:
          LD_LIBRARY_PATH: "$LAMBDA_TASK_ROOT/lib:$LAMBDA_TASK_ROOT/lib64:$LAMBDA_RUNTIME_DIR:$LAMBDA_RUNTIME_DIR/lib:$LAMBDA_TASK_ROOT:/opt/lib64:/opt/lib:/usr/lib64:/usr/lib:/lib64:/lib"

You need to add /opt/lib64 to LD_LIBRARY_PATH for libraries to be found.

Verification

One stop script

Run the script test.sh and it will use the docker image to convert test/test.pdf to image-0.png and image-1.png.

Example output

> ./test.sh
docker cp test/test.pdf 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99:/tmp/.
docker cp test/convert.sh 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99:/tmp/.
docker start 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99
27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99
docker exec 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99 sh /tmp/convert.sh
docker cp 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99:/tmp/image-0.png .
docker cp 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99:/tmp/image-1.png .
docker stop 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99
27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99
docker rm 27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99
27d5e0bd67f93a957ea6c311660d4f04d7f3b4b8dc2b3ef884145e1eca5d9b99

Manual verification

If you want to manually test the image , run

docker run -it image_magick

This will open up a bash shell in the image.

Run docker ps in a different shell and note the container id.

Copy a pdf file into it like this:

docker cp  <pdf filename> <container id>:/tmp/test.pdf

Replace and <container_id> with the actual name of the test file and the container id you found with docker ps.

Run the following inside the docker image:

export GS_LIB=/opt/share/ghostscript/9.06/Resource/Init
export MAGICK_HOME=/opt
export MAGICK_CONFIGURE_PATH=/opt/etc/ImageMagick
export MAGICK_CODER_MODULE_PATH=/opt/lib64/ImageMagick-6.7.8/modules-Q16/coders
export LD_LIBRARY_PATH=/opt/lib64:/opt/lib:/lib64:/usr/lib64:/usr/lib
/opt/bin/convert -density 300 -background white -alpha remove /tmp/test.pdf -quality 100 /tmp/image.png

Copy the resulting image out of the container to view it:

docker cp  <container id>:/tmp/image.png .
open /tmp/image.png

Replace open with your application of choice if either your OS doesn't supporting open or you want to open the image in another application.

Test Lambda application

First build the new layer with build_zip.sh. Go to the example directory (cd example). Build and deploy the application:

sam build
sam package --output-template-file package.yaml --s3-bucket <deployment bucket name>
sam deploy --template-file package.yaml --region eu-central-1 --capabilities CAPABILITY_IAM --stack-name image-magick-example

Test the application on an image.

aws s3 cp ../test/test.pdf s3://imagemagick-test-bucket/pdf/test.pdf

After a few seconds, images of the pages will appear in the S3 bucket:

aws s3 ls s3://imagemagick-test-bucket/images/

Python3.8

Seems like necessary libraries are not avilable in python3.8 lambda environments so this will only work for 3.7.