sync-upstream.yml

name: Sync Upstream

# Checks anomalyco/models.dev (upstream) for new commits three times daily
# and merges them into our fork's dev branch automatically when safe to do so.
#
# Schedule (UTC):
#   - 13:00 UTC = 08:00 EST / 05:00 PST
#   - 17:00 UTC = 12:00 EST / 09:00 PST
#   - 01:00 UTC = 20:00 EST / 17:00 PST
#
# Conflict handling:
#   - Fast-forward and clean merges are committed and pushed automatically.
#   - If a merge conflict is detected the workflow writes a detailed report
#     to the job summary (visible in the Actions tab) and exits without
#     pushing anything broken. The job is marked as failed so GitHub sends
#     the usual failure notification email/alert.
#   - Same for validation failures: report written to summary, push skipped.

on:
  schedule:
    - cron: '0 13 * * *'   # 08:00 EST
    - cron: '0 17 * * *'   # 12:00 EST
    - cron: '0 1  * * *'   # 20:00 EST
  workflow_dispatch:         # Allow manual trigger at any time

concurrency:
  group: sync-upstream
  cancel-in-progress: false  # Never cancel an in-progress sync

permissions:
  contents: write   # push merged commits to dev

jobs:
  sync:
    name: Merge upstream into dev
    runs-on: ubuntu-latest

    steps:
      - name: Checkout fork (dev branch, full history)
        uses: actions/checkout@v4
        with:
          ref: dev
          fetch-depth: 0
          token: ${{ secrets.GITHUB_TOKEN }}

      - name: Configure git identity
        run: |
          git config user.name  "github-actions[bot]"
          git config user.email "github-actions[bot]@users.noreply.github.com"

      - name: Add upstream remote
        run: |
          git remote add upstream https://github.com/anomalyco/models.dev.git
          git remote set-url --push upstream DISABLED  # never push to upstream

      - name: Fetch upstream dev
        run: git fetch upstream dev --no-tags

      - name: Check whether sync is needed
        id: check
        run: |
          UPSTREAM_SHA=$(git rev-parse upstream/dev)
          FORK_SHA=$(git rev-parse HEAD)

          echo "upstream_sha=$UPSTREAM_SHA" >> "$GITHUB_OUTPUT"
          echo "fork_sha=$FORK_SHA"         >> "$GITHUB_OUTPUT"

          if git merge-base --is-ancestor upstream/dev HEAD; then
            echo "status=up_to_date" >> "$GITHUB_OUTPUT"
            echo "✅ Fork is already up to date with upstream."
          else
            NEW=$(git rev-list HEAD..upstream/dev --count)
            echo "status=behind"         >> "$GITHUB_OUTPUT"
            echo "new_commit_count=$NEW" >> "$GITHUB_OUTPUT"
            echo "📦 Fork is $NEW commit(s) behind upstream."
          fi

      - name: Attempt upstream merge
        if: steps.check.outputs.status == 'behind'
        id: merge
        run: |
          # Attempt a no-commit merge first to detect conflicts cleanly
          if git merge --no-commit --no-ff upstream/dev 2>&1; then
            # No conflicts — complete the merge
            git merge --abort 2>/dev/null || true

            UPSTREAM_SHA="${{ steps.check.outputs.upstream_sha }}"
            SHORT_SHA="${UPSTREAM_SHA:0:8}"
            COUNT="${{ steps.check.outputs.new_commit_count }}"

            git merge upstream/dev \
              --no-edit \
              -m "chore: sync $COUNT commit(s) from upstream (${SHORT_SHA})"

            echo "result=success" >> "$GITHUB_OUTPUT"
          else
            git merge --abort 2>/dev/null || true

            # Capture the list of conflicting files for the summary
            git diff --name-only --diff-filter=U 2>/dev/null > /tmp/conflict-files.txt || true

            echo "result=conflict" >> "$GITHUB_OUTPUT"
            echo "⚠️  Merge conflict detected."
          fi

      - name: Set up Bun (for validation)
        if: steps.merge.outputs.result == 'success'
        uses: oven-sh/setup-bun@v1
        with:
          bun-version: latest

      - name: Cache Bun dependencies
        if: steps.merge.outputs.result == 'success'
        uses: actions/setup-node@v4
        with:
          cache: bun
          cache-dependency-path: ./bun.lock

      - name: Validate data after merge
        if: steps.merge.outputs.result == 'success'
        id: validate
        run: |
          bun install --frozen-lockfile

          if bun validate > /dev/null 2>&1; then
            echo "result=pass" >> "$GITHUB_OUTPUT"
            echo "✅ bun validate passed."
          else
            echo "result=fail" >> "$GITHUB_OUTPUT"
            bun validate 2>&1 | tail -40 > /tmp/validate-errors.txt || true
            echo "❌ bun validate failed."
          fi

      - name: Push merged commits
        if: |
          steps.merge.outputs.result == 'success' &&
          steps.validate.outputs.result == 'pass'
        run: |
          git push origin dev
          echo "🚀 Pushed merged commits to origin/dev."

      - name: Summary
        if: always()
        run: |
          echo "### Sync Upstream — Summary" >> "$GITHUB_STEP_SUMMARY"
          echo "" >> "$GITHUB_STEP_SUMMARY"
          echo "| | |" >> "$GITHUB_STEP_SUMMARY"
          echo "|---|---|" >> "$GITHUB_STEP_SUMMARY"
          echo "| Upstream SHA | \`${{ steps.check.outputs.upstream_sha }}\` |" >> "$GITHUB_STEP_SUMMARY"
          echo "| Fork SHA     | \`${{ steps.check.outputs.fork_sha }}\` |" >> "$GITHUB_STEP_SUMMARY"
          echo "" >> "$GITHUB_STEP_SUMMARY"

          STATUS="${{ steps.check.outputs.status }}"
          MERGE="${{ steps.merge.outputs.result }}"
          VALIDATE="${{ steps.validate.outputs.result }}"

          if [ "$STATUS" = "up_to_date" ]; then
            echo "✅ **Already up to date** — no new upstream commits." >> "$GITHUB_STEP_SUMMARY"

          elif [ "$MERGE" = "success" ] && [ "$VALIDATE" = "pass" ]; then
            echo "🚀 **Synced successfully** — merged ${{ steps.check.outputs.new_commit_count }} commit(s) and pushed to \`dev\`." >> "$GITHUB_STEP_SUMMARY"

          elif [ "$MERGE" = "conflict" ]; then
            echo "⚠️ **Merge conflict — manual resolution required**" >> "$GITHUB_STEP_SUMMARY"
            echo "" >> "$GITHUB_STEP_SUMMARY"
            echo "The fork is ${{ steps.check.outputs.new_commit_count }} commit(s) behind upstream but could not be merged automatically." >> "$GITHUB_STEP_SUMMARY"
            echo "" >> "$GITHUB_STEP_SUMMARY"

            # Show conflicting files if captured
            if [ -f /tmp/conflict-files.txt ] && [ -s /tmp/conflict-files.txt ]; then
              echo "**Conflicting files:**" >> "$GITHUB_STEP_SUMMARY"
              echo '```' >> "$GITHUB_STEP_SUMMARY"
              cat /tmp/conflict-files.txt >> "$GITHUB_STEP_SUMMARY"
              echo '```' >> "$GITHUB_STEP_SUMMARY"
              echo "" >> "$GITHUB_STEP_SUMMARY"
            fi

            # Show pending upstream commits
            PENDING=$(git log HEAD..upstream/dev --oneline 2>/dev/null | head -20 || echo "(unavailable)")
            echo "**Pending upstream commits:**" >> "$GITHUB_STEP_SUMMARY"
            echo '```' >> "$GITHUB_STEP_SUMMARY"
            echo "$PENDING" >> "$GITHUB_STEP_SUMMARY"
            echo '```' >> "$GITHUB_STEP_SUMMARY"
            echo "" >> "$GITHUB_STEP_SUMMARY"
            echo "**To resolve locally:**" >> "$GITHUB_STEP_SUMMARY"
            echo '```bash' >> "$GITHUB_STEP_SUMMARY"
            echo 'git fetch upstream' >> "$GITHUB_STEP_SUMMARY"
            echo 'git checkout dev' >> "$GITHUB_STEP_SUMMARY"
            echo 'git merge upstream/dev' >> "$GITHUB_STEP_SUMMARY"
            echo '# resolve conflicts, then:' >> "$GITHUB_STEP_SUMMARY"
            echo 'git push origin dev' >> "$GITHUB_STEP_SUMMARY"
            echo '```' >> "$GITHUB_STEP_SUMMARY"

            # Exit non-zero so GitHub sends a failure notification
            exit 1

          elif [ "$VALIDATE" = "fail" ]; then
            echo "❌ **Validation failed — push skipped**" >> "$GITHUB_STEP_SUMMARY"
            echo "" >> "$GITHUB_STEP_SUMMARY"
            echo "Upstream merged cleanly but \`bun validate\` failed on the merged state." >> "$GITHUB_STEP_SUMMARY"
            echo "The push was skipped to avoid introducing invalid data." >> "$GITHUB_STEP_SUMMARY"
            echo "" >> "$GITHUB_STEP_SUMMARY"
            if [ -f /tmp/validate-errors.txt ]; then
              echo "**Validation errors:**" >> "$GITHUB_STEP_SUMMARY"
              echo '```' >> "$GITHUB_STEP_SUMMARY"
              cat /tmp/validate-errors.txt >> "$GITHUB_STEP_SUMMARY"
              echo '```' >> "$GITHUB_STEP_SUMMARY"
            fi

            # Exit non-zero so GitHub sends a failure notification
            exit 1
          fi