web-crawler-test.yml

name: Build Run and Test Website Crawl

on:
  workflow_dispatch:
    inputs:
      A11Y_ENGINE_BRANCH:
        description: 'Oobee Branch (master, other)'
        required: true
        default: 'master'
      A11Y_SCAN_URL:
        description: 'Website URL'
        required: true
        default: 'https://'
      A11Y_SCANNER:
        description: 'Scanner (website, sitemap)'
        required: true
        default: 'website'
      A11Y_SCAN_MAX_NUM_PAGES:
        description: 'Maximum Number of Pages to Scan'
        required: true
        default: '10'
      A11Y_SCAN_MAX_CONCURRENCY:
        description: 'Maximum Number of Concurrent Page Scans'
        required: true
        default: '10'
      A11Y_SCAN_SCREENSHOTS:
        description: 'Include screenshots (screenshots, none)'
        required: true
        default: 'screenshots'
      A11Y_SCAN_FOLLOW_ROBOTS:
        description: 'Crawler adhere to robots.txt (yes, no)'
        required: true
        default: 'no'
      A11Y_SCAN_STRATEGY:
        description: 'Crawler adhere to pages matching exact hostname or same parent domain (same-hostname, same-domain)'
        required: true
        default: 'same-domain'
      USERNAME:
        description: "Your name (for telemetry purposes)"
        required: false
      EMAIL:
        description: "Email address (for telemetry purposes)"
        required: false

jobs:
  build:
    runs-on: ubuntu-latest

    steps:
      - name: Get Public IP Address
        run:
          curl https://ipinfo.io
      
      - name: Check if email is valid or empty
        run: |
          email="${{ inputs.EMAIL }}"
          echo "Email provided: $email"
          valid_email_regex='^.+@.+\..+$'
          echo "Validating with regex: $valid_email_regex"
          status="$(if [ -z "$email" ]; then echo "Empty"; elif [[ $email =~ $valid_email_regex ]]; then echo "Valid"; else echo "Invalid"; fi)"

          if [[ "$status" == "Invalid" ]]; then
            echo "Invalid email provided. Workflow failed."
            exit 1
          fi

        # Proceed with steps if EMAIL is valid or empty
      - name: Use provided name and email or defaults
        run: |
          if [[ -z "${{ inputs.EMAIL }}" ]]; then
            final_email="${{ github.actor }}@users.noreply.github.com"
          else
            final_email="${{ inputs.EMAIL }}"
          fi
          
          username="${{ inputs.USERNAME || github.actor }}"

          echo "final_email=$final_email" >> $GITHUB_ENV
          echo "username=$username" >> $GITHUB_ENV
          # Use $final_email and $username in subsequent steps
      
      - name: Checkout code
        uses: actions/checkout@master
        with:
          repository: "GovTechSG/oobee"
          ref: "${{ github.event.inputs.A11Y_ENGINE_BRANCH }}"
          
      - name: Build and run Docker container
        run: |
          # Build your Docker image (replace 'oobee' and 'Dockerfile' with your image name and Dockerfile path)
          docker build -t oobee .
          # Run the Docker container (adjust any options as needed)
          docker run -dit --name oobee-container --add-host=docs.google.com:0.0.0.0 --add-host=assets.wogaa.sg:0.0.0.0 oobee
        # Ensure the Docker container is running before running the next step
        continue-on-error: true
  
      - name: Start accessibility scan in Docker container
        run: |
          # Execute a webisite crawl
          docker exec -e OOBEE_VERBOSE=true oobee-container npm run cli -- -c "${{ github.event.inputs.A11Y_SCANNER }}" -u "${{ github.event.inputs.A11Y_SCAN_URL }}" -p "${{ github.event.inputs.A11Y_SCAN_MAX_NUM_PAGES }}" -t "${{ github.event.inputs.A11Y_SCAN_MAX_CONCURRENCY }}" -a "${{ github.event.inputs.A11Y_SCAN_SCREENSHOTS }}" -r "${{ github.event.inputs.A11Y_SCAN_FOLLOW_ROBOTS }}" -s "${{ github.event.inputs.A11Y_SCAN_STRATEGY}}" -k "${{ env.username }}:${{ env.final_email }}" -b chromium || true
         
      - name: Print errors
        run: |
          docker exec oobee-container cat errors.txt || true
        
      - name: Print logs
        run: |
          docker exec oobee-container cat logs/*/*.txt || true
      - name: Copy Results from Docker Container
        run: docker cp oobee-container:/app/results ./results
  
      - name: Zip Results
        run: zip -r results.zip ./results
  
      - name: Upload Zip File
        uses: actions/upload-artifact@v4
        with:
          name: results
          path: results.zip
            
      - name: Cleanup Docker container
        run: |
          # Stop and remove the Docker container
          docker stop oobee-container
          docker rm oobee-container