reverse-sync-push.yml

name: Reverse Sync on Push

on:
  workflow_dispatch:  # Allow manual triggering
    inputs:
      # target repository to be synced to, org/repo format
      target_repo:
        required: true
        type: string
      # paths to be ignored when reverse syncing
      ignore_paths:
        required: false
        type: string
        default: '.github/,README.md'
      # paths to be synced, comma separated
      synced_paths:
        required: false
        type: string
        default: docs/,.yarn/,doom.config.yml,yarn.lock,tsconfig.json,package.json,sites.yaml

  workflow_call:  # Allow other workflows to reuse this workflow
    inputs:
      # source repository to be synced to, org/repo format
      target_repo:
        required: true
        type: string
      # paths to be ignored when reverse syncing
      ignore_paths:
        required: false
        type: string
        default: '.github/,README.md'
      # paths to be synced, comma separated
      synced_paths:
        required: false
        type: string
        default: docs/,.yarn/,doom.config.yml,yarn.lock,tsconfig.json,package.json,sites.yaml
    secrets:
      # token with access privilege to both repos
      GH_TOKEN:
        required: true

# Limit token capabilities to what the job really needs
permissions:
  contents: read          # checkout / git diff
  pull-requests: write    # create PR in target repo

jobs:
  reverse-sync:
    runs-on: ubuntu-latest

    steps:
    - name: Checkout repository
      uses: actions/checkout@v4
      with:
        token: ${{ secrets.GH_TOKEN }}
        fetch-depth: 0

    - name: Check if commit is from merged PR
      id: check_pr_commit
      run: |
        # Get the latest commit
        commit_sha="${{ github.sha }}"
        echo "commit_sha=$commit_sha" >> $GITHUB_OUTPUT

        # Get commit message
        commit_message=$(git log -1 --pretty=format:"%s" $commit_sha)
        echo "commit_message=$commit_message" >> $GITHUB_OUTPUT

        # Get commit author
        commit_author=$(git log -1 --pretty=format:"%an" $commit_sha)
        commit_author_email=$(git log -1 --pretty=format:"%ae" $commit_sha)
        echo "commit_author=$commit_author" >> $GITHUB_OUTPUT
        echo "commit_author_email=$commit_author_email" >> $GITHUB_OUTPUT

        echo "=> Commit: $commit_sha"
        echo "=> Message: $commit_message"
        echo "=> Author: $commit_author <$commit_author_email>"

        # Check if this is a merge commit from GitHub (squash merge creates a single commit)
        # Look for PR number in commit message (GitHub automatically adds this)
        if [[ "$commit_message" =~ \(#([0-9]+)\)$ ]]; then
          pr_number="${BASH_REMATCH[1]}"
          echo "pr_number=$pr_number" >> $GITHUB_OUTPUT
          echo "is_pr_commit=true" >> $GITHUB_OUTPUT
          echo "✅ Detected commit from PR #$pr_number"
        else
          echo "is_pr_commit=false" >> $GITHUB_OUTPUT
          echo "ℹ️ Not a PR commit - skipping reverse sync"
        fi

        # Skip if the commit is from our sync bot
        if [[ "$commit_message" == *"[reverse-sync]"* ]]; then
          echo "skip_sync=true" >> $GITHUB_OUTPUT
          echo "🤖 Commit is from sync bot - skipping reverse sync"
        elif [[ "$commit_message" == *"Sync documentation"* ]] || [[ "$commit_message" == *"sync-docs"* ]] || [[ "$commit_message" == *"[skip-sync]"* ]]; then
          echo "skip_sync=true" >> $GITHUB_OUTPUT
          echo "🤖 Commit appears to be from sync process - skipping reverse sync"
        else
          echo "skip_sync=false" >> $GITHUB_OUTPUT
          echo "👥 Commit is from external contributor - proceeding with reverse sync"
        fi

    - name: Get PR information
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false'
      id: get_pr_info
      run: |
        pr_number="${{ steps.check_pr_commit.outputs.pr_number }}"

        echo "👀 Looking for PR $pr_number ..."

        # Get PR information using GitHub API
        pr_info=$(curl -s -H "Authorization: token ${{ secrets.GITHUB_TOKEN }}" \
          "https://api.github.com/repos/${{ github.repository }}/pulls/$pr_number")


        echo "👀 PR data $pr_info ..."

        pr_title=$(echo "$pr_info" | jq -r '.title')
        pr_author=$(echo "$pr_info" | jq -r '.user.login')
        pr_url=$(echo "$pr_info" | jq -r '.html_url')
        pr_base_ref=$(echo "$pr_info" | jq -r '.base.ref')

        echo "pr_title=$pr_title" >> $GITHUB_OUTPUT
        echo "pr_author=$pr_author" >> $GITHUB_OUTPUT
        echo "pr_url=$pr_url" >> $GITHUB_OUTPUT
        echo "pr_base_ref=$pr_base_ref" >> $GITHUB_OUTPUT

        echo "=> PR #$pr_number: $pr_title"
        echo "=> Author: $pr_author"
        echo "=> URL: $pr_url"

    - name: Get commit changes
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false'
      id: get_changes
      run: |
        commit_sha="${{ steps.check_pr_commit.outputs.commit_sha }}"

        # Get the parent commit (previous commit before this one)
        parent_commit=$(git rev-parse ${commit_sha}^)
        echo "parent_commit=$parent_commit" >> $GITHUB_OUTPUT

        # Get list of changed files in the commit, excluding ignored paths
        ignore_pattern=$(echo "${{inputs.ignore_paths}}" | sed 's/,/|/g' | sed 's|/$||g')
        echo "🙈 Ignored paths: $ignore_pattern"

        git diff --name-only $parent_commit $commit_sha | grep -v -E "^($ignore_pattern)" > changed_files.txt || true

        echo "📋 Changed files in commit:"
        cat changed_files.txt

        # Check if any relevant files were changed
        if [ -s changed_files.txt ]; then
          echo "has_doc_changes=true" >> $GITHUB_OUTPUT
          echo "✅ Documentation changes detected"
        else
          echo "has_doc_changes=false" >> $GITHUB_OUTPUT
          echo "ℹ️ No documentation changes detected"
        fi

    - name: Checkout target repository
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false' && steps.get_changes.outputs.has_doc_changes == 'true'
      uses: actions/checkout@v4
      with:
        repository: ${{inputs.target_repo}}
        token: ${{ secrets.GH_TOKEN }}
        path: target-docs
        fetch-depth: 0
        ref: ${{ steps.get_pr_info.outputs.pr_base_ref }}

    - name: Create reverse sync branch
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false' && steps.get_changes.outputs.has_doc_changes == 'true'
      id: create_branch
      run: |
        cd target-docs

        # Create a unique branch name
        pr_number="${{ steps.check_pr_commit.outputs.pr_number }}"
        branch_name="reverse-sync/pr-$pr_number-$(date +%s)"
        echo "branch_name=$branch_name" >> $GITHUB_OUTPUT

        git checkout -b "$branch_name"

        # Configure git
        git config user.name "${{github.event.pusher.name}}"
        git config user.email "${{github.event.pusher.email}}"

        echo "📝 Created branch: $branch_name"

    - name: Apply changes from sonarqube-docs
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false' && steps.get_changes.outputs.has_doc_changes == 'true'
      run: |
        commit_sha="${{ steps.check_pr_commit.outputs.commit_sha }}"
        parent_commit="${{ steps.get_changes.outputs.parent_commit }}"

        # Create a patch with only the synced paths
        echo "📑 Will only sync these paths: ${{inputs.synced_paths}}"
        git format-patch $parent_commit..$commit_sha --stdout -- ${{inputs.synced_paths}} > changes.patch

        echo "📋 Generated Patch: "
        cat changes.patch

        cd target-docs

        # Apply the patch
        if [ -s ../changes.patch ]; then
          echo "📦 Applying changes from sonarqube-docs..."
          git apply ../changes.patch || {
            echo "⚠️ Patch application failed, trying manual copy..."

            # Fallback: manual copy of changed files
            while IFS= read -r file; do
              if [ -f "../$file" ]; then
                mkdir -p "$(dirname "$file")"
                cp "../$file" "$file"
                echo "✅ Copied: $file"
              fi
            done < ../changed_files.txt
          }
        else
          echo "⚠️ No patch generated, using manual copy..."

          # Manual copy approach
          while IFS= read -r file; do
            if [ -f "../$file" ]; then
              mkdir -p "$(dirname "$file")"
              cp "../$file" "$file"
              echo "✅ Copied: $file"
            fi
          done < ../changed_files.txt
        fi

    - name: Commit changes
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false' && steps.get_changes.outputs.has_doc_changes == 'true'
      id: commit_changes
      run: |
        cd target-docs

        echo "🔍 Checking for changes..."

        git status

        echo "=> Adding changes..."

        git add .

        echo "==> Checking if there are staged changes..."
        if git diff --staged --quiet; then
          echo "has_changes=false" >> $GITHUB_OUTPUT
          echo "ℹ️ No changes to commit"
        else
          echo "=> Changes detected..."
          echo "has_changes=true" >> $GITHUB_OUTPUT

          pr_number="${{ steps.check_pr_commit.outputs.pr_number }}"
          pr_title="${{ steps.get_pr_info.outputs.pr_title }}"
          pr_author="${{ steps.get_pr_info.outputs.pr_author }}"
          pr_url="${{ steps.get_pr_info.outputs.pr_url }}"
          commit_sha="${{ steps.check_pr_commit.outputs.commit_sha }}"

          echo "=> Creating commit message..."
          echo "  - PR Number: $pr_number"
          echo "  - PR Title: $pr_title"
          echo "  - PR Author: $pr_author"
          echo "  - PR URL: $pr_url"
          echo "  - Commit SHA: $commit_sha"

          # Create commit message with reverse sync marker

          echo "[reverse-sync] Sync documentation changes from sonarqube-docs PR #$pr_number" > commit_message.txt
          echo "" >> commit_message.txt
          echo "- Title: $pr_title" >> commit_message.txt
          echo "- Author: $pr_author" >> commit_message.txt
          echo "- URL: $pr_url" >> commit_message.txt
          echo "- Commit: $commit_sha" >> commit_message.txt

          echo "=> Commit message:"
          cat commit_message.txt

          echo "=> Committing changes..."

          git commit -F commit_message.txt
          rm commit_message.txt

          echo "✅ Changes committed successfully"
        fi

    - name: Push branch and create PR
      if: steps.check_pr_commit.outputs.is_pr_commit == 'true' && steps.check_pr_commit.outputs.skip_sync == 'false' && steps.get_changes.outputs.has_doc_changes == 'true' && steps.commit_changes.outputs.has_changes == 'true'
      run: |
        cd target-docs
        branch_name="${{ steps.create_branch.outputs.branch_name }}"
        pr_base_ref="${{ steps.get_pr_info.outputs.pr_base_ref }}"
        pr_number="${{ steps.check_pr_commit.outputs.pr_number }}"
        pr_title="${{ steps.get_pr_info.outputs.pr_title }}"
        pr_author="${{ steps.get_pr_info.outputs.pr_author }}"
        pr_url="${{ steps.get_pr_info.outputs.pr_url }}"

        # Push the branch
        git push origin "$branch_name"

        # Create PR body with proper JSON escaping
        pr_body=$(cat << 'EOF'
        ### 🔄 Reverse Sync from sonarqube-docs

        This PR incorporates documentation changes from external contributors to the sonarqube-docs repository.

        #### Original PR Details
        - **Repository**: alauda/sonarqube-docs
        - **PR**: #%PR_NUMBER% - %PR_TITLE%
        - **Author**: @%PR_AUTHOR%
        - **URL**: %PR_URL%

        #### Changes
        This PR includes changes to documentation files that were contributed by external contributors to the public sonarqube-docs repository.

        #### Important Notes
        - ⚠️ This PR contains the `[reverse-sync]` marker to prevent infinite sync loops
        - ✅ Once merged, this will NOT trigger a forward sync back to sonarqube-docs
        - 🔍 Please review the changes to ensure they align with internal documentation standards

        ---
        *This PR was automatically created by the reverse sync workflow.*
        EOF)

        # Replace placeholders in the PR body
        pr_body=$(echo "$pr_body" | sed "s/%PR_NUMBER%/$pr_number/g")
        pr_body=$(echo "$pr_body" | sed "s/%PR_TITLE%/$pr_title/g")
        pr_body=$(echo "$pr_body" | sed "s/%PR_AUTHOR%/$pr_author/g")
        pr_body=$(echo "$pr_body" | sed "s|%PR_URL%|$pr_url|g")

        # Create JSON payload with proper escaping
        json_payload=$(jq -n \
          --arg title "[reverse-sync] Documentation changes from sonarqube-docs PR #$pr_number" \
          --arg head "$branch_name" \
          --arg base "$pr_base_ref" \
          --arg body "$pr_body" \
          '{title: $title, head: $head, base: $base, body: $body}')

        # Create the PR using GitHub API
        curl -X POST \
          -H "Authorization: token ${{ secrets.GH_TOKEN }}" \
          -H "Accept: application/vnd.github.v3+json" \
          -H "Content-Type: application/json" \
          https://api.github.com/repos/${{ inputs.target_repo }}/pulls \
          -d "$json_payload"

    - name: Create workflow summary
      run: |
        if [ "${{ steps.check_pr_commit.outputs.is_pr_commit }}" == "false" ]; then
          echo "## ℹ️ Not a PR Commit" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "This commit was not created from a merged PR, so no reverse sync was performed." >> $GITHUB_STEP_SUMMARY
        elif [ "${{ steps.check_pr_commit.outputs.skip_sync }}" == "true" ]; then
          echo "## 🤖 Sync Bot Commit - Reverse Sync Skipped" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "This commit was created by the sync bot, so reverse sync was skipped to prevent loops." >> $GITHUB_STEP_SUMMARY
        elif [ "${{ steps.get_changes.outputs.has_doc_changes }}" == "false" ]; then
          echo "## ℹ️ No Documentation Changes" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "This commit didn't contain any documentation changes, so no reverse sync was needed." >> $GITHUB_STEP_SUMMARY
        elif [ "${{ steps.commit_changes.outputs.has_changes }}" == "false" ]; then
          echo "## ℹ️ No Changes to Sync" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "The documentation changes were already present in target-docs." >> $GITHUB_STEP_SUMMARY
        else
          echo "## 🎉 Reverse Sync Completed" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Successfully created a PR in target-docs with the documentation changes." >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "### Details:" >> $GITHUB_STEP_SUMMARY
          echo "- **Source PR**: #${{ steps.check_pr_commit.outputs.pr_number }} by @${{ steps.get_pr_info.outputs.pr_author }}" >> $GITHUB_STEP_SUMMARY
          echo "- **Branch**: ${{ steps.create_branch.outputs.branch_name }}" >> $GITHUB_STEP_SUMMARY
          echo "- **Base ref**: ${{ steps.get_pr_info.outputs.pr_base_ref }}" >> $GITHUB_STEP_SUMMARY
          echo "- **Target repository**: ${{ inputs.target_repo }}" >> $GITHUB_STEP_SUMMARY
        fi