Resizerful

This is a serverless image handler wrapper

Aws setup

To deploy this follow the following guide.

https://aws.amazon.com/solutions/serverless-image-handler/

This will set up a an api gateway endpoint, a lambda, and a cloudfront distribution. At this point the image handler will serve images.

To secure the images as suggested in the Security section of the guide we are using signed URLs. Make sure you also secure the following:

• Make sure the S3 bucket is set to be private
• Make sure the api gateway enpoint that calls the lambda is also private

Follow this section to secure the cloudfront distribution:

Serving Private Content with Signed URLs and Signed Cookies

there are two main steps,

• creating a key value pair. Make sure you get the PEM file and make note of the key paid ID

• restricting access under Cloudfront > Cloud Front distributions > the distribution > Behaviors tab > Edit > "Restrict Viewer Access" => Yes

Install

pipenv install -e "git+https://github.com/wndrfl/resizerful.git#egg=resizerful"

Usage

Upload images to an S3 bucket using your favorite method. e.g.

client = boto3.client('s3', region_name='us-west-2')
client.upload_file('escape.jpg', 'wonderful-memes', 'escape.jpg')

Then keep track of the bucket name and image name in the bucket. Now you can create a Resizerful object and call signed_resize_url in this fashion

key = '''  ...  '''
resizer = Resizerful(
    bucket_name='wonderful-memes',
    cf_base_url='https://dzzplw659h3bu.cloudfront.net/',
    key_pair_id='APKAJXOKVLCWKI726JTA',
    private_key=key
)
signed_url = resizer.signed_resize_url('escape.jpg')
print(signed_url)

In this case, the private key in the variable key is the contents of the PEM file from the cloudfront distribution.

Resizerful takes 4 required parameters

argument	Notes
bucket_name	S3 bucket
cf_base_url	base path of the clouldfront distribution
key_pair_id	key pair id of the cloudfront distribution
private_key	contents of PEM file of the key pair being used

signed_resize_url takes many parameters, including:

width	width
height	height
fit	One of ('cover', 'contain', 'fill', 'inside', 'outside')
grayscale	grayscale filter
flatten	flattens the image, whatever that means
flip	flips image vertically
flop	flops image horizontally
negate	some sort of retro filter
normalise	another filter