Skip to content

Latest commit

 

History

History
29 lines (21 loc) · 1.8 KB

infrastructure.md

File metadata and controls

29 lines (21 loc) · 1.8 KB
Raw

A11ygator

An application and Twitter bot that bites websites to taste their accessibility.

Infrastructure

Infrastructure overview

  • The app (static HTML+JS+CSS files) is stored on a bucket and is served via CloudFront.
  • Informations about the generated reports are saved in a DynamoDB table, used for de-duplication and statistics.
  • The screenshots are stored on another bucket, along with JSON reports, and are served via CloudFront—/screenshots/* path pattern.
  • The API is configured via API Gateway, and proxied via CloudFront—/api/* path pattern.

API Endpoints

  • GET /api/webhooks/twitter: endpoint required by Twitter for webhooks, to pass Challenge Response Check (CRC).
  • POST /api/webhooks/twitter: Twitter webhook that will receive payloads and selectively enqueue messages into the SQS Queue for reports to be generated.
  • POST /api/reports: endpoint to manually enqueue a report generation.
  • GET /api/reports/{id}: endpoint to obtain a generated report in either JSON or HTML format — it will return 404 until the report is ready.

Background processing

  • Report generation requests are saved in DynamoDB and sent to a SNS Queue.
  • Enqueued messages are consumed by a Lambda function that loads the URL, waits for the configured amount of time, and runs Pa11y with the configured standard, then takes two screenshots (viewport and full-page), uploads the screenshots and the report JSON to an S3 bucket, updates the information saved in DynamoDB, and sends a notification to a SNS Topic.
  • A Lambda function is listening for notifications of generated reports that were requested via Twitter, and tweets a reply that contains aggregate data and a link to the HTML report.