broken-links.yml

name: broken-links

on:
  schedule:
    # Nightly at 09:00 UTC (around 03:00 MT / 04:00 CT).
    - cron: '0 9 * * *'
  workflow_dispatch: {}

permissions:
  issues: write
  contents: read

jobs:
  crawl:
    name: Crawl site for broken links
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@v4

      - name: Pick target URL
        id: target
        # We feed Lychee the sitemap rather than the homepage. Lychee's
        # CLI doesn't support recursive crawling — it checks every link
        # on every input. Pointing it at /sitemap.xml gives it the full
        # URL list (33 static pages + 162 blog posts) and it then checks
        # every link on every one of those pages. That's the "test all
        # internal pages too" coverage the homepage-only run lacked.
        run: |
          if [ -n "${{ secrets.STAGING_URL }}" ]; then
            base="${{ secrets.STAGING_URL }}"
          else
            base="https://beta.ontargetaba.com/"
          fi
          # Trim trailing slash, append sitemap.xml
          base="${base%/}"
          echo "url=${base}/sitemap.xml" >> "$GITHUB_OUTPUT"

      - name: Run Lychee
        id: lychee
        uses: lycheeverse/lychee-action@v2
        with:
          args: >-
            --verbose
            --no-progress
            --max-retries 2
            --retry-wait-time 5
            --timeout 20
            --accept 200,206,429
            --include-verbatim
            --exclude '^https://ontargetaba\.com/'
            --exclude '^https://static\.hotjar\.com/c/hotjar-$'
            --exclude 'search_term_string'
            --exclude '^https://(www\.)?linkedin\.com/'
            --exclude '^https://(www\.)?instagram\.com/'
            --exclude '^https://(www\.)?facebook\.com/'
            ${{ steps.target.outputs.url }}
          fail: false
          output: ./lychee-report.md

      - name: Verify Lychee actually ran
        # `fail: false` above lets the step continue when broken links
        # are found, but it also masks the case where Lychee itself
        # blew up (bad CLI args, sitemap unreachable, etc.) — in which
        # case no report is produced and we'd silently open a misleading
        # tracking issue. Treat "no report or empty report" as a hard
        # workflow failure so red CI flags the real problem.
        run: |
          if [ ! -s ./lychee-report.md ]; then
            echo "::error::Lychee produced no report — it likely failed to run. Check the 'Run Lychee' step logs above for CLI arg errors or network failures."
            exit 1
          fi

      - name: Find existing tracking issue
        id: existing
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          num=$(gh issue list --state open --label broken-links \
                  --search "Broken links found" \
                  --json number --jq '.[0].number // empty')
          echo "number=$num" >> "$GITHUB_OUTPUT"

      - name: Open or update tracking issue
        if: steps.lychee.outputs.exit_code != '0'
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          TARGET_URL: ${{ steps.target.outputs.url }}
          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
        run: |
          date_str=$(date -u +%Y-%m-%d)
          title="Broken links found ${date_str}"

          # Build the body in a file. We previously stitched it with
          # bash $(...) + +=, but command substitution strips trailing
          # newlines, so when lychee-report.md didn't end with one the
          # closing ``` ended up smushed against the last content line.
          # Writing through printf to a real file gives us deterministic
          # line endings — closing fence is always on its own line.
          {
            printf '## Broken links\n\n'
            printf 'Target: %s\n' "$TARGET_URL"
            printf 'Run: %s\n\n' "$RUN_URL"
            printf '```\n'
            if [ -s ./lychee-report.md ]; then
              cat ./lychee-report.md
            else
              printf 'See run logs for details.\n'
            fi
            # Force a final newline before the closing fence regardless
            # of whether the report ended with one.
            printf '\n```\n'
          } > /tmp/issue-body.md

          if [ -n "${{ steps.existing.outputs.number }}" ]; then
            gh issue edit "${{ steps.existing.outputs.number }}" \
              --title "$title" \
              --body-file /tmp/issue-body.md
          else
            gh issue create \
              --title "$title" \
              --body-file /tmp/issue-body.md \
              --label broken-links
          fi

      - name: Close tracking issue when clean
        if: steps.lychee.outputs.exit_code == '0' && steps.existing.outputs.number != ''
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          gh issue close "${{ steps.existing.outputs.number }}" \
            --comment "All previously broken links are now resolving. Closing."