check-broken-pr-links.yml

name: Check Links in Pull Requests

on:
  pull_request:           # changed from pull_request
    branches:
      - main
    paths:
      - '**/*.md'

jobs:
  check-links:
    runs-on: ubuntu-latest

    steps:
      # 1️⃣ Checkout repository
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          fetch-depth: 0

      # 2️⃣ Get changed Markdown files in the PR
      - name: Get changed Markdown files
        id: changed-files
        run: |
          # Fetch the base branch to ensure we have the latest
          git fetch origin ${{ github.event.pull_request.base.ref }}
          
          # Compare against the merge base (the point where your branch diverged)
          BASE_SHA=$(git merge-base origin/${{ github.event.pull_request.base.ref }} ${{ github.sha }})
          
          echo "Base SHA: $BASE_SHA"
          echo "Head SHA: ${{ github.sha }}"
          
          # Get only files changed in THIS PR (comparing merge-base to HEAD)
          CHANGED_FILES=$(git diff --name-only $BASE_SHA ${{ github.sha }} | grep '\.md$' || true)
          
          echo "Changed files:"
          echo "$CHANGED_FILES"
          
          CHANGED_FILES="${CHANGED_FILES//$'\n'/ }"
          echo "CHANGED_FILES=$CHANGED_FILES" >> $GITHUB_ENV
          
          if [ -z "$CHANGED_FILES" ]; then
            echo "No Markdown files changed"
          else
            echo "Changed Markdown files: $CHANGED_FILES"
          fi

      # 3️⃣ Skip if no Markdown files changed
      - name: Skip if no Markdown files changed
        if: env.CHANGED_FILES == ''
        run: |
          echo "No Markdown files changed. Skipping link check."
          exit 0

      # 4️⃣ Run Lychee on changed files
      - name: Run Lychee
        id: run-lychee
        if: env.CHANGED_FILES != ''
        uses: lycheeverse/lychee-action@v2
        with:
          args: |
            --no-progress
            --include-fragments
            --format detailed
            ${{ env.CHANGED_FILES }}
          output: lychee/out_raw.md
          fail: false  # ✅ don't fail yet, let us capture output

      # 5️⃣ Format Lychee output (user-friendly, relative paths)
      - name: Format Lychee report
        id: format-report
        if: always()
        run: |
          mkdir -p lychee
          : > lychee/comment.md   # start with an empty file
      
          awk '
            /^Errors in / {
              file=$3
              gsub("^/home/runner/work/UmbracoDocs/UmbracoDocs/", "", file)
              current_file = file
              has_errors = 0
              next
            }

            /Error: ERROR\]/ {
              # Skip anchor errors for #umbraco-[number] patterns
              if ($0 ~ /#umbraco-[0-9]+/) next
              
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^Error: ERROR\] /, "", msg)
              gsub("^file:///home/runner/work/UmbracoDocs/UmbracoDocs/", "", msg)
              gsub(/\|/, "\\|", msg)   # escape Markdown pipe
              print "\n⚓ Anchor not found → " msg "\n" >> "lychee/comment.md"
              next
            }

            /Cannot find file/ {
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^Error: ERROR\] /, "", msg)
              gsub("file:///home/runner/work/UmbracoDocs/UmbracoDocs/", "", msg)
              gsub(/\|/, "\\|", msg)
              print "\n🗒️ File not found → " msg "\n" >> "lychee/comment.md"
              next
            }

            /\[404\]/ {
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^- \[ \] /, "", msg)
              sub(/^\[404\] /, "", msg)
              gsub(/\|/, "\\|", msg)   # escape pipe
              print "\n❌ 404 Not Found → " msg "\n" >> "lychee/comment.md"
              next
            }

            /\[301\]|\[302\]/ {
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^- \[ \] /, "", msg)
              sub(/^\[(301|302)\] /, "", msg)
              gsub(/\|/, "\\|", msg)   # escape pipe
              print "\n🔀 Redirect → " msg "\n" >> "lychee/comment.md"
              next
            }

            /\[522\]/ {
              # Skip 522 errors from issues.umbraco.org
              if ($0 ~ /issues\.umbraco\.org/) next
              
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^- \[ \] /, "", msg)
              sub(/^\[522\] /, "", msg)
              gsub(/\|/, "\\|", msg)
              print "\n⚠ Server error (522) → " msg "\n" >> "lychee/comment.md"
              next
            }

            /\[502\]/ {
              # Skip 502 errors from umbraco.com
              if ($0 ~ /umbraco\.com/) next
              
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^- \[ \] /, "", msg)
              sub(/^\[502\] /, "", msg)
              gsub(/\|/, "\\|", msg)
              print "\n⚠ Server error (502) → " msg "\n" >> "lychee/comment.md"
              next
            }

            /Timeout/ && !/Timeouts/ {
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^- \[ \] /, "", msg)
              gsub(/\|/, "\\|", msg)   # escape pipe just in case
              print "\n⏳ Timeout → " msg "\n" >> "lychee/comment.md"
              next
            }

            # catch-all for any other errors
            /^\- \[ \] \[[0-9]+\]/ {
              # Print file header if this is the first error for this file
              if (current_file != "" && has_errors == 0) {
                print "\n**Broken links found in: " current_file "**" >> "lychee/comment.md"
                has_errors = 1
              }
              
              msg = $0
              sub(/^- \[ \] /, "", msg)
              gsub(/\|/, "|", msg)
              print "\n⚠ Unknown error → " msg "\n" >> "lychee/comment.md"
              next
            }
            
          ' lychee/out_raw.md
      
          # Add header only if we found content
          if [ -s lychee/comment.md ]; then
            sed -i '1i **The Link Checker found broken links in your PR**.\n Please review the following list:\n' lychee/comment.md
            echo "has_content=true" >> $GITHUB_OUTPUT
          else
            echo "has_content=false" >> $GITHUB_OUTPUT
          fi

      # 6️⃣ Comment broken links on PR (if present)
      - name: Comment broken links
        if: always() && (env.CHANGED_FILES != '') && (steps.format-report.outputs.has_content == 'true')
        uses: marocchino/sticky-pull-request-comment@v2
        with:
          path: lychee/comment.md
          recreate: true

      # 7️⃣ Fail workflow if broken links exist
      - name: Fail workflow if broken links
        if: steps.format-report.outputs.has_content == 'true'
        run: |
          echo "❌ Broken links detected. Please review the PR comment for details."
          exit 1