check-links-in-prod.yaml

# This workflow verifies hyperlinks using the deployed
# docs site. It has to be triggered manually.
name: Check links

on:
  workflow_dispatch: {}
  schedule:
    - cron: '39 0 20 * *' # every 20th of the month at 00:39

env:
  # space-separated list of the top-level sections to check in https://docs.giantswarm.io/{section}/
  SECTIONS: 'getting-started overview reference support tutorials'

jobs:
  check-links:
    runs-on: ubuntu-latest
    steps:
      - name: Pull image
        run: docker pull ghcr.io/linkchecker/linkchecker:latest

      - name: Create output folder
        run: |
          mkdir -p output
          chmod 777 output

      - name: check-links-in-overview-pages
        shell: python {0}
        continue-on-error: true
        
        run: |
          import os
          import subprocess

          pwd = os.getcwd()
          sections = os.getenv('SECTIONS').split()

          for i, section in enumerate(sections):
            # call docker run command
            cmd = ['docker', 'run', '--rm', '--name', 'linkchecker',
                   '--volume', f'{pwd}/output:/workdir', '--workdir', '/workdir',
                   'ghcr.io/linkchecker/linkchecker:latest',
                   f'https://docs.giantswarm.io/{section}/',
                   '--check-extern',
                   '--threads 1',
                   '--recursion-level 2',
                   '--no-status',
                   '--file-output html/utf8/{section}.html']
            
            result = subprocess.run(cmd)
            if result.returncode > 0:
              env_file = os.getenv('GITHUB_ENV')
              with open(env_file, 'a') as f:
                f.write(f"failed{i}=true\n")

      - name: Store reports as artifacts
        uses: actions/upload-artifact@5d5d22a31266ced268874388b861e4b58bb5c2f3 # v4.3.2
        with:
          name: link-check-reports
          path: ./output
          if-no-files-found: warn
          retention-days: 14

      - name: Report errors
        shell: python {0}
        run: |
          import sys
          import os

          found_error = False
          sections = os.getenv('SECTIONS').split()

          for i, section in enumerate(sections):
            failed = os.getenv(f"failed{i}")
            if failed == "true":
              print(f"Found link errors in the {section} section. Please download the report artifact and fix.")
              found_error = True

          if found_error:
            sys.exit(1)