dify-docs/.github/workflows/sync_docs_update.yml

# Workflow for updating translation PRs on sync events
# Triggered by Analyze workflow for security validation
name: Update Translation PR

on:
  workflow_run:
    workflows: ["Analyze Documentation Changes"]
    types: [completed]
    branches-ignore:
      - 'docs-sync-pr-*'

permissions:
  contents: write
  pull-requests: write
  actions: read

concurrency:
  group: docs-translation-${{ github.event.workflow_run.head_branch }}
  cancel-in-progress: false

jobs:
  update-translation:
    runs-on: ubuntu-latest
    # Only run if analyze workflow succeeded
    if: github.event.workflow_run.conclusion == 'success'
    steps:
      - name: Download analysis artifacts
        uses: actions/download-artifact@v4
        with:
          name: docs-sync-analysis-${{ github.event.workflow_run.id }}
          path: /tmp/analysis
          github-token: ${{ secrets.GITHUB_TOKEN }}
          run-id: ${{ github.event.workflow_run.id }}

      - name: Load and validate analysis
        id: load-analysis
        run: |
          echo "Loading validated analysis from secure workflow..."

          # Check if sync_plan.json exists (created by analyze workflow)
          if [ ! -f "/tmp/analysis/sync_plan.json" ]; then
            echo "❌ No sync plan found - analyze workflow may have skipped this PR"
            echo "should_proceed=false" >> $GITHUB_OUTPUT
            exit 0
          fi

          # Load analysis metadata
          PR_NUMBER=$(jq -r '.metadata.pr_number' /tmp/analysis/sync_plan.json)
          PR_TYPE=$(jq -r '.metadata.pr_type' /tmp/analysis/sync_plan.json)
          IS_INCREMENTAL=$(jq -r '.metadata.is_incremental' /tmp/analysis/sync_plan.json)
          BASE_SHA=$(jq -r '.metadata.base_sha' /tmp/analysis/sync_plan.json)
          HEAD_SHA=$(jq -r '.metadata.head_sha' /tmp/analysis/sync_plan.json)

          # Verify this is a source-language-only PR (already validated by analyze workflow)
          if [ "$PR_TYPE" != "source" ]; then
            echo "ℹ️ Not a source-language-only PR (type: $PR_TYPE) - skipping translation update"
            echo "should_proceed=false" >> $GITHUB_OUTPUT
            exit 0
          fi

          echo "✅ Validated analysis loaded from secure workflow"
          echo "pr_number=$PR_NUMBER" >> $GITHUB_OUTPUT
          echo "is_incremental=$IS_INCREMENTAL" >> $GITHUB_OUTPUT
          echo "base_sha=$BASE_SHA" >> $GITHUB_OUTPUT
          echo "head_sha=$HEAD_SHA" >> $GITHUB_OUTPUT
          echo "should_proceed=true" >> $GITHUB_OUTPUT

          # Display summary
          FILE_COUNT=$(jq -r '.metadata.file_count // 0' /tmp/analysis/sync_plan.json)
          echo "📊 Analysis Summary:"
          echo "  - PR: #$PR_NUMBER"
          echo "  - Type: $PR_TYPE"
          echo "  - Files: $FILE_COUNT"
          echo "  - Incremental: $IS_INCREMENTAL"
          echo "  - Range: ${BASE_SHA:0:8}...${HEAD_SHA:0:8}"

      - name: Checkout PR
        if: steps.load-analysis.outputs.should_proceed == 'true'
        uses: actions/checkout@v4
        with:
          ref: ${{ steps.load-analysis.outputs.head_sha }}
          fetch-depth: 0
          token: ${{ secrets.GITHUB_TOKEN }}

      - name: Set up Python
        if: steps.load-analysis.outputs.should_proceed == 'true'
        uses: actions/setup-python@v4
        with:
          python-version: '3.9'

      - name: Find associated translation PR
        if: steps.load-analysis.outputs.should_proceed == 'true'
        id: find-translation-pr
        env:
          GH_TOKEN: ${{ github.token }}
        run: |
          PR_NUMBER="${{ steps.load-analysis.outputs.pr_number }}"
          echo "Looking for translation PR associated with PR #${PR_NUMBER}..."

          # Search for translation PR by branch name pattern
          TRANSLATION_PR_DATA=$(gh pr list \
            --search "head:docs-sync-pr-${PR_NUMBER}" \
            --json number,title,url,state \
            --jq '.[0] // empty' 2>/dev/null || echo "")

          if [ -n "$TRANSLATION_PR_DATA" ] && [ "$TRANSLATION_PR_DATA" != "null" ]; then
            TRANSLATION_PR_NUMBER=$(echo "$TRANSLATION_PR_DATA" | jq -r '.number')
            TRANSLATION_PR_STATE=$(echo "$TRANSLATION_PR_DATA" | jq -r '.state')
            TRANSLATION_PR_URL=$(echo "$TRANSLATION_PR_DATA" | jq -r '.url')

            if [ "$TRANSLATION_PR_STATE" = "OPEN" ]; then
              echo "✅ Found active translation PR #${TRANSLATION_PR_NUMBER}"
              echo "translation_pr_number=$TRANSLATION_PR_NUMBER" >> $GITHUB_OUTPUT
              echo "translation_pr_url=$TRANSLATION_PR_URL" >> $GITHUB_OUTPUT
              echo "found_translation_pr=true" >> $GITHUB_OUTPUT
            else
              echo "ℹ️ Found translation PR #${TRANSLATION_PR_NUMBER} but it's ${TRANSLATION_PR_STATE} - skipping update"
              echo "found_translation_pr=false" >> $GITHUB_OUTPUT
            fi
          else
            echo "ℹ️ No translation PR found for PR #${PR_NUMBER} - this might be the first update"
            echo "found_translation_pr=false" >> $GITHUB_OUTPUT
          fi

      - name: Determine update range
        if: steps.find-translation-pr.outputs.found_translation_pr == 'true'
        id: update-range
        env:
          GH_TOKEN: ${{ github.token }}
        run: |
          PR_NUMBER="${{ steps.load-analysis.outputs.pr_number }}"
          HEAD_SHA="${{ steps.load-analysis.outputs.head_sha }}"
          BASE_SHA="${{ steps.load-analysis.outputs.base_sha }}"

          echo "Determining incremental update range..."

          # Get last processed commit from translation branch commit messages
          SYNC_BRANCH="docs-sync-pr-${PR_NUMBER}"
          git fetch origin "$SYNC_BRANCH" 2>/dev/null || true

          LAST_PROCESSED=$(git log "origin/$SYNC_BRANCH" --format=%B -1 \
            | grep -oP 'Last-Processed-Commit: \K[a-f0-9]+' \
            | head -1 || echo "")

          if [ -n "$LAST_PROCESSED" ]; then
            echo "✅ Found last processed commit: $LAST_PROCESSED"
            COMPARE_BASE="$LAST_PROCESSED"
          else
            echo "⚠️ No last processed commit found, using analysis base SHA"
            COMPARE_BASE="$BASE_SHA"
          fi

          COMPARE_HEAD="$HEAD_SHA"

          echo "compare_base=$COMPARE_BASE" >> $GITHUB_OUTPUT
          echo "compare_head=$COMPARE_HEAD" >> $GITHUB_OUTPUT

          echo "📊 Incremental update range: $COMPARE_BASE...$COMPARE_HEAD"

      - name: Install dependencies
        if: steps.find-translation-pr.outputs.found_translation_pr == 'true'
        run: |
          cd tools/translate
          pip install httpx aiofiles python-dotenv

      - name: Run translation and commit
        if: steps.find-translation-pr.outputs.found_translation_pr == 'true'
        id: update-translations
        env:
          DIFY_API_KEY: ${{ secrets.DIFY_API_KEY }}
          GH_TOKEN: ${{ github.token }}
        run: |
          echo "Running incremental translation update with validated inputs..."

          PR_NUMBER="${{ steps.load-analysis.outputs.pr_number }}"
          HEAD_SHA="${{ steps.update-range.outputs.compare_head }}"
          BASE_SHA="${{ steps.update-range.outputs.compare_base }}"

          # Get PR title from workflow run event
          PR_TITLE="${{ github.event.workflow_run.pull_requests[0].title }}"

          echo "PR: #${PR_NUMBER}"
          echo "Comparison: ${BASE_SHA:0:8}...${HEAD_SHA:0:8}"
          echo "Using validated sync plan from analyze workflow"

          # Call the Python script for incremental translation
          cd tools/translate
          python translate_pr.py \
            --pr-number "$PR_NUMBER" \
            --head-sha "$HEAD_SHA" \
            --base-sha "$BASE_SHA" \
            --pr-title "$PR_TITLE" \
            --is-incremental \
            2>&1 | tee /tmp/translation_output.log

          SCRIPT_EXIT_CODE=${PIPESTATUS[0]}

          # Extract JSON result
          RESULT_JSON=$(grep -A 1000 "RESULT_JSON:" /tmp/translation_output.log | tail -n +2 | grep -B 1000 "^========" | head -n -1)

          if [ -n "$RESULT_JSON" ]; then
            echo "$RESULT_JSON" > /tmp/translation_result.json

            # Parse outputs
            SUCCESS=$(echo "$RESULT_JSON" | jq -r '.success')
            HAS_CHANGES=$(echo "$RESULT_JSON" | jq -r '.has_changes // false')

            echo "has_changes=$HAS_CHANGES" >> $GITHUB_OUTPUT
            echo "commit_successful=$([ "$SUCCESS" = "true" ] && echo true || echo false)" >> $GITHUB_OUTPUT

            # Extract translation results
            echo "$RESULT_JSON" | jq -r '.translation_results' > /tmp/update_results.json 2>/dev/null || echo '{"translated":[],"failed":[],"skipped":[]}' > /tmp/update_results.json

            echo "✅ Translation update completed"
          else
            echo "❌ Could not parse result JSON"
            echo "has_changes=false" >> $GITHUB_OUTPUT
            echo "commit_successful=false" >> $GITHUB_OUTPUT
            exit 1
          fi

          exit $SCRIPT_EXIT_CODE

      - name: Comment on original PR about update
        if: steps.update-translations.outputs.has_changes == 'true' && steps.update-translations.outputs.commit_successful == 'true'
        uses: actions/github-script@v7
        continue-on-error: true
        with:
          script: |
            const fs = require('fs');
            const prNumber = ${{ steps.load-analysis.outputs.pr_number }};
            const translationPrNumber = '${{ steps.find-translation-pr.outputs.translation_pr_number }}';
            const translationPrUrl = '${{ steps.find-translation-pr.outputs.translation_pr_url }}';

            // Load update results
            let results = { translated: [], failed: [], skipped: [] };
            try {
              results = JSON.parse(fs.readFileSync('/tmp/update_results.json', 'utf8'));
            } catch (e) {
              console.log('Could not load update results');
            }

            let comment = `## 🌐 Multi-language Sync\n\n`;
            comment += `✅ Updated sync PR [#${translationPrNumber}](${translationPrUrl})\n\n`;

            if (results.translated && results.translated.length > 0) {
              comment += `**Synced ${results.translated.length} file${results.translated.length > 1 ? 's' : ''}** to cn + jp\n\n`;
            }

            if (results.failed && results.failed.length > 0) {
              comment += `⚠️ **${results.failed.length} file${results.failed.length > 1 ? 's' : ''} failed**\n\n`;
            }

            comment += `_Future commits will auto-update the sync PR._`;

            try {
              await github.rest.issues.createComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                issue_number: prNumber,
                body: comment
              });
            } catch (error) {
              console.log('Could not comment on original PR:', error.message);
            }

      - name: Comment on translation PR about update
        if: steps.update-translations.outputs.has_changes == 'true' && steps.update-translations.outputs.commit_successful == 'true'
        uses: actions/github-script@v7
        continue-on-error: true
        with:
          script: |
            const fs = require('fs');
            const prNumber = ${{ steps.load-analysis.outputs.pr_number }};
            const translationPrNumber = '${{ steps.find-translation-pr.outputs.translation_pr_number }}';

            // Load update results
            let results = { translated: [], failed: [], skipped: [] };
            try {
              results = JSON.parse(fs.readFileSync('/tmp/update_results.json', 'utf8'));
            } catch (e) {
              console.log('Could not load update results');
            }

            const fileCount = results.translated ? results.translated.length : 0;
            const updateComment = `✅ Synced ${fileCount} file${fileCount !== 1 ? 's' : ''} from PR #${prNumber}` +
              (results.failed && results.failed.length > 0 ? `\n\n⚠️ ${results.failed.length} file${results.failed.length !== 1 ? 's' : ''} failed` : '');

            try {
              await github.rest.issues.createComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                issue_number: translationPrNumber,
                body: updateComment
              });
            } catch (error) {
              console.log('Could not comment on translation PR:', error.message);
            }

      - name: Handle no updates needed
        if: steps.find-translation-pr.outputs.found_translation_pr == 'true' && steps.update-translations.outputs.has_changes != 'true'
        uses: actions/github-script@v7
        continue-on-error: true
        with:
          script: |
            const prNumber = ${{ steps.load-analysis.outputs.pr_number }};
            const translationPrNumber = '${{ steps.find-translation-pr.outputs.translation_pr_number }}';

            const comment = `✅ Sync PR [#${translationPrNumber}](https://github.com/${{ github.repository }}/pull/${translationPrNumber}) is already up to date.`;

            try {
              await github.rest.issues.createComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                issue_number: prNumber,
                body: comment
              });
            } catch (error) {
              console.log('Could not comment on original PR:', error.message);
            }

  handle-cancellation:
    runs-on: ubuntu-latest
    needs: update-translation
    if: always() && needs.update-translation.result == 'cancelled'
    steps:
      - name: Notify about cancelled workflow
        uses: actions/github-script@v7
        continue-on-error: true
        with:
          script: |
            console.log('⚠️ Update workflow was cancelled - likely due to newer commit');

            // Try to get PR number from workflow run artifacts
            const workflowRunId = context.payload.workflow_run.id;
            const headBranch = context.payload.workflow_run.head_branch;

            try {
              // List artifacts from the analyze workflow
              const artifacts = await github.rest.actions.listWorkflowRunArtifacts({
                owner: context.repo.owner,
                repo: context.repo.repo,
                run_id: workflowRunId
              });

              // Find analysis artifact
              const analysisArtifact = artifacts.data.artifacts.find(a =>
                a.name.startsWith('docs-sync-analysis-')
              );

              if (!analysisArtifact) {
                console.log('No analysis artifact found - cannot determine PR number');
                return;
              }

              // Extract PR number from artifact name
              const prNumber = analysisArtifact.name.split('-').pop();

              console.log(`Found PR #${prNumber} for cancelled workflow`);

              // Get repository info for workflow dispatch link
              const repoUrl = `https://github.com/${context.repo.owner}/${context.repo.repo}`;
              const workflowDispatchUrl = `${repoUrl}/actions/workflows/sync_docs_execute.yml`;

              const comment = '## ⚠️ Sync Update Skipped\n\n' +
                'This commit was not synced because a newer commit arrived. **Your latest commit will be synced automatically.**\n\n' +
                '**If you need this specific commit synced:**\n' +
                `Go to [Actions → Execute Documentation Sync](${workflowDispatchUrl}) and manually run with PR number **${prNumber}**\n\n` +
                '_When you push multiple commits quickly, only the first and last get synced to avoid backlog._';

              await github.rest.issues.createComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                issue_number: parseInt(prNumber),
                body: comment
              });

              console.log(`✅ Posted cancellation notice to PR #${prNumber}`);

            } catch (error) {
              console.log(`Failed to notify PR: ${error.message}`);
            }