The purpose of this repository is to demonstrate a PDF generator inside AWS lambda with chrome-aws-lambda, serverless, pug and knex. This has been done as part of the article in dev.to
- Initialize serverless either inside project or globally (after installing package globally) with
serverless
- Modify PUG template inside src/
- Fetch data for the template. In this repository we use knex and postgresSQL. For production use I would suggest to add database username and password to environmental variables either inside deployment platform or in AWS.
- Deploy with
npm run deploy:dev
Deploying this to AWS will generate you an url like https://xxxx.execute-api.*area*.amazonaws.com/development/pdf/{yearMonth} in development and https://xxxx.execute-api.*area*.amazonaws.com/production/pdf/{yearMonth} in production.
When running this setup with 1024 MB memory, it takes rougly 4 seconds to execute. With more complex pages or data fetching default timeout may not be enough so you may need to increase the timeout either with serverless.yaml or straight from AWS console.
Puppeteer is pretty versatile so you can create PDF from your own html file like in this example with PUG template language or calling URL.
More information regarding this repository can found from Generate a PDF in AWS Lambda with NodeJS and Puppeteer dev.to post.
Since this deployment setup exposes your PDF generator straight to open the internet, it can be called just by knowing the URL. This can lead to the abuse of your application resulting in either your data to leak or getting a huge bill from AWS.
To mitigate the risk, there are several ways to increase security and reduce/block the abuse:
- Enabling the API key to API Gateway to limit access and monitor the usage. In this option, a client needs to send an API key in the request to be able to use the lambda function so you need to create a way to get the API key to the client. It won't fully block the abuse since the API key is exposed when making the request and the same key can be used again but it will give you a chance to limit the usage amount. This works pretty well when you don't have authorization in your application.
Serverless has API key plugin to create this for your lambda function.
-
Enable authorizer to API Gateway (either AWS_IAM or custom). In this option, you need to have a user authorization set up (either provided by AWS or using a custom one). This method blocks the abuse from unauthorized use but if you have open registration, the user can still make an unlimited amount of request.
-
Checking authorization before starting up puppeteer. This is almost the same as the previous option but the authorization check is done inside the lambda and not in the API gateway. The difference is that you don't have to have a custom authorization function the Lambda function handles itself. It won't block the abuse since the endpoint will respond every time but it makes execution time for unauthorized use a lot smaller.