CI workflow (lockfale/OSINT-Framework)

name: CI
 
on:
  push:
    branches: [master]
  pull_request:
    branches: [master]
  schedule:
    # Run link check weekly on Saturday at 08:00 UTC
    - cron: '0 8 * * 6'
 
concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true
 
jobs:
  json-lint:
    timeout-minutes: 30
    name: JSON Lint
    runs-on: latchkey-small
    steps:
      - uses: actions/checkout@v4
 
      - name: Validate arf.json is well-formed
        run: python3 -m json.tool public/arf.json > /dev/null
 
  smoke-test:
    timeout-minutes: 30
    name: Smoke Test
    runs-on: latchkey-small
    steps:
      - uses: actions/checkout@v4
 
      - name: Check index.html local asset references exist
        run: |
          check_asset() {
            local path="public/$1"
            if [ ! -f "$path" ]; then
              echo "MISSING: $path"
              exit 1
            else
              echo "OK: $path"
            fi
          }
          check_asset "css/arf.css"
          check_asset "js/d3.min.js"
          check_asset "js/arf.js"
          check_asset "arf.json"
 
  link-check:
    timeout-minutes: 30
    name: Link Check
    runs-on: latchkey-small
    # Only run on schedule (weekly Saturday) - avoid hammering external sites on every push/PR
    if: github.event_name == 'schedule'
    steps:
      - uses: actions/checkout@v4
 
      - name: Extract URLs from arf.json
        run: |
          python3 -c "
          import json, re, sys
 
          with open('public/arf.json') as f:
              data = json.load(f)
 
          def extract_urls(node):
              urls = []
              url = node.get('url', '')
              if url and url.startswith('http'):
                  urls.append(url)
              for child in node.get('children', []):
                  urls.extend(extract_urls(child))
              return urls
 
          urls = extract_urls(data)
          print(f'Found {len(urls)} URLs', file=sys.stderr)
          with open('urls.txt', 'w') as f:
              f.write('\n'.join(urls))
          "
 
      - name: Run lychee link checker
        uses: lycheeverse/lychee-action@v2
        with:
          args: |
            --verbose
            --no-progress
            --timeout 20
            --max-retries 2
            --retry-wait-time 5
            --accept 200,201,204,206,301,302,307,308,403,429
            urls.txt
          fail: true
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}