test.yml

name: Tests

on:
  push:
    branches: [main, "release/**"]
  pull_request:
  schedule:
    # Nightly quarantine run at 02:00 UTC — exercises quarantined tests to
    # detect when they become stable again.
    - cron: "0 2 * * *"
  workflow_dispatch:
    inputs:
      quarantine_mode:
        description: 'Set to "run" to execute quarantined tests instead of skipping them'
        required: false
        default: "skip"

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: ${{ github.event_name == 'pull_request' }}

env:
  NODE_VERSION: "20"
  # Default: quarantined tests are skipped. The nightly job and workflow_dispatch
  # override this to "run" so flaky tests can be checked in isolation.
  QUARANTINE_MODE: ${{ github.event.inputs.quarantine_mode || (github.event_name == 'schedule' && 'run') || 'skip' }}

jobs:
  # ---------------------------------------------------------------------------
  # 1. Validate quarantine registry before anything else runs
  # ---------------------------------------------------------------------------
  validate-quarantine:
    name: Validate quarantine registry
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Check jq is available
        run: jq --version

      - name: Validate quarantine.json
        run: bash scripts/manage-flaky-tests.sh validate

      - name: Print quarantine report
        if: always()
        run: bash scripts/manage-flaky-tests.sh report

  # ---------------------------------------------------------------------------
  # 2. Backend tests (sharded, mirrors the existing backend-tests.yml approach)
  # ---------------------------------------------------------------------------
  backend-test:
    name: Backend – shard ${{ matrix.shard }}/${{ matrix.total }}
    needs: validate-quarantine
    runs-on: ubuntu-latest
    strategy:
      fail-fast: false
      matrix:
        shard: [1, 2, 3, 4]
        total: [4]

    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-node@v4
        with:
          node-version: ${{ env.NODE_VERSION }}
          cache: npm
          cache-dependency-path: backend/package-lock.json

      - name: Install backend dependencies
        working-directory: backend
        run: npm ci

      - name: Run backend tests (shard ${{ matrix.shard }}/${{ matrix.total }})
        working-directory: backend
        env:
          QUARANTINE_MODE: ${{ env.QUARANTINE_MODE }}
          NODE_ENV: test
        run: |
          npm run test:shard -- \
            --shard=${{ matrix.shard }}/${{ matrix.total }} \
            --reporter=blob \
            --reporter=github-actions

      - name: Upload blob report
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: backend-blob-${{ matrix.shard }}-of-${{ matrix.total }}
          path: backend/.vitest-reports/
          retention-days: 7

  # ---------------------------------------------------------------------------
  # 3. Merge coverage from all shards + enforce thresholds
  # ---------------------------------------------------------------------------
  backend-coverage:
    name: Backend – merge coverage & check thresholds
    needs: backend-test
    runs-on: ubuntu-latest
    if: always()

    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-node@v4
        with:
          node-version: ${{ env.NODE_VERSION }}
          cache: npm
          cache-dependency-path: backend/package-lock.json

      - name: Install backend dependencies
        working-directory: backend
        run: npm ci

      - name: Download all blob reports
        uses: actions/download-artifact@v4
        with:
          pattern: backend-blob-*
          path: backend/.vitest-reports/
          merge-multiple: true

      - name: Merge coverage reports
        working-directory: backend
        run: npm run test:merge-coverage

      - name: Upload merged coverage
        uses: actions/upload-artifact@v4
        with:
          name: backend-coverage
          path: backend/coverage/
          retention-days: 14

      - name: Summarise flaky-test status in job summary
        if: always()
        run: |
          echo "## Quarantine Status" >> $GITHUB_STEP_SUMMARY
          bash scripts/manage-flaky-tests.sh report >> $GITHUB_STEP_SUMMARY

  # ---------------------------------------------------------------------------
  # 4. Frontend tests
  # ---------------------------------------------------------------------------
  frontend-test:
    name: Frontend unit tests
    needs: validate-quarantine
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-node@v4
        with:
          node-version: ${{ env.NODE_VERSION }}
          cache: npm
          cache-dependency-path: frontend/package-lock.json

      - name: Install frontend dependencies
        working-directory: frontend
        run: npm ci

      - name: Run frontend unit tests
        working-directory: frontend
        env:
          NODE_ENV: test
        run: npm test

  # ---------------------------------------------------------------------------
  # 5. Nightly quarantine run — surfaces newly-stable tests
  #    Only runs on schedule or when quarantine_mode=run is passed manually.
  # ---------------------------------------------------------------------------
  quarantine-run:
    name: Quarantine run (flaky tests in isolation)
    # Run on schedule OR when explicitly triggered with quarantine_mode=run
    if: >
      github.event_name == 'schedule' ||
      (github.event_name == 'workflow_dispatch' &&
       github.event.inputs.quarantine_mode == 'run')
    needs: validate-quarantine
    runs-on: ubuntu-latest
    continue-on-error: true # never block main CI; result is informational

    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-node@v4
        with:
          node-version: ${{ env.NODE_VERSION }}
          cache: npm
          cache-dependency-path: backend/package-lock.json

      - name: Install backend dependencies
        working-directory: backend
        run: npm ci

      - name: Run quarantined tests
        working-directory: backend
        env:
          QUARANTINE_MODE: run
          NODE_ENV: test
        # Use a custom reporter so failures are clearly labeled as quarantined.
        run: |
          npm run test:ci -- \
            --reporter=verbose \
            --reporter=github-actions \
            2>&1 | tee quarantine-run.log || true

      - name: Upload quarantine run log
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: quarantine-run-log
          path: backend/quarantine-run.log
          retention-days: 14

      - name: Summarise quarantine run
        if: always()
        run: |
          echo "## Nightly Quarantine Run" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Tests below are currently quarantined. Passing tests here are" >> $GITHUB_STEP_SUMMARY
          echo "candidates to be re-enabled via \`scripts/manage-flaky-tests.sh remove\`." >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          bash scripts/manage-flaky-tests.sh report >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY

          echo "### Run output (tail)" >> $GITHUB_STEP_SUMMARY
          echo '```' >> $GITHUB_STEP_SUMMARY
          tail -80 backend/quarantine-run.log >> $GITHUB_STEP_SUMMARY 2>/dev/null || echo "(no log)" >> $GITHUB_STEP_SUMMARY
          echo '```' >> $GITHUB_STEP_SUMMARY

      - name: Alert if quarantined tests consistently pass (candidate for re-enable)
        if: always()
        run: |
          # Exit 0 from the test run means all quarantined tests passed — signal to maintainers.
          if grep -q "Tests  [0-9]* passed" backend/quarantine-run.log 2>/dev/null && \
             ! grep -q "failed" backend/quarantine-run.log 2>/dev/null; then
            echo "::notice title=Quarantine::All quarantined tests PASSED in the nightly run. Consider re-enabling them via scripts/manage-flaky-tests.sh remove."
          fi

  # ---------------------------------------------------------------------------
  # 6. PR gate — summarise quarantine impact on this PR
  # ---------------------------------------------------------------------------
  quarantine-pr-check:
    name: Quarantine check (PR)
    if: github.event_name == 'pull_request'
    needs: validate-quarantine
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0

      - name: Check if quarantine.json was modified in this PR
        id: quarantine_diff
        run: |
          if git diff --name-only origin/${{ github.base_ref }}...HEAD | grep -q "quarantine.json"; then
            echo "modified=true" >> $GITHUB_OUTPUT
          else
            echo "modified=false" >> $GITHUB_OUTPUT
          fi

      - name: Fail if quarantine.json grew without a linked issue
        if: steps.quarantine_diff.outputs.modified == 'true'
        run: |
          echo "quarantine.json was modified in this PR. Running validation..."
          bash scripts/manage-flaky-tests.sh validate
          echo ""
          echo "Current quarantine state:"
          bash scripts/manage-flaky-tests.sh list

      - name: Add quarantine summary to PR step summary
        if: always()
        run: |
          echo "## Flaky Test Quarantine" >> $GITHUB_STEP_SUMMARY
          bash scripts/manage-flaky-tests.sh report >> $GITHUB_STEP_SUMMARY

  slack-notification:
    name: Notify Slack on failure
    needs: [validate-quarantine, backend-test, backend-coverage, frontend-test]
    if: always() && contains(needs.*.result, 'failure') && github.ref == 'refs/heads/main'
    runs-on: ubuntu-latest
    steps:
      - name: Send Slack Notification
        run: |
          curl -X POST -H 'Content-type: application/json' \
          --data "{\"text\":\"🚨 CI Failure on main!\n*Repo*: ${{ github.repository }}\n*Branch*: ${{ github.ref_name }}\n*Failing Step*: ${{ github.workflow }}\n*Link*: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}\"}" \
          ${{ secrets.SLACK_WEBHOOK_URL }}