name: 'Evals: PR Impact'

on:
  pull_request:
    paths:
      - 'packages/core/src/prompts/**'
      - 'packages/core/src/tools/**'
      - 'packages/core/src/agents/**'
      - 'evals/**'

permissions:
  contents: 'read'
  checks: 'write'
  pull-requests: 'write'

jobs:
  eval-run:
    name: 'Eval Run (${{ matrix.model }}, attempt ${{ matrix.run_attempt }})'
    runs-on: 'gemini-cli-ubuntu-16-core'
    if: "github.repository == 'google-gemini/gemini-cli'"
    strategy:
      fail-fast: false
      matrix:
        run_attempt: [1, 2, 3]
        model: ["gemini-3.1-pro-preview-customtools", "gemini-3-flash-preview"]
    steps:
      - name: 'Checkout'
        uses: 'actions/checkout@v4'

      - name: 'Set up Node.js'
        uses: 'actions/setup-node@v4'
        with:
          node-version-file: '.nvmrc'
          cache: 'npm'

      - name: 'Install dependencies'
        run: 'npm ci'

      - name: 'Build project'
        run: 'npm run build'

      - name: 'Run Evals'
        env:
          GEMINI_API_KEY: '${{ secrets.GEMINI_API_KEY }}'
          RUN_EVALS: 'true'
          GEMINI_MODEL: '${{ matrix.model }}'
        run: |
          # Use a unique directory name for this matrix leg
          DIR_NAME="eval-logs-${{ matrix.model }}-${{ matrix.run_attempt }}"
          mkdir -p "evals/logs/$DIR_NAME"
          npm run test:all_evals -- --outputFile.json="evals/logs/$DIR_NAME/report.json" || true

      - name: 'Upload Logs'
        uses: 'actions/upload-artifact@v4'
        with:
          name: 'eval-logs-${{ matrix.model }}-${{ matrix.run_attempt }}'
          path: 'evals/logs'
          retention-days: 1

  aggregate-impact:
    name: 'Aggregate Impact'
    needs: ['eval-run']
    if: 'always()'
    runs-on: 'gemini-cli-ubuntu-16-core'
    steps:
      - name: 'Checkout'
        uses: 'actions/checkout@v4'

      - name: 'Download Logs'
        uses: 'actions/download-artifact@v4'
        with:
          path: 'artifacts'

      - name: 'Generate Impact Report'
        id: 'generate-report'
        env:
          GH_TOKEN: '${{ secrets.GITHUB_TOKEN }}'
        run: |
          if [ ! -d "artifacts" ]; then
            echo "No logs found, skipping report generation."
            exit 0
          fi
          echo "<!-- eval-impact-report -->" > report.md
          node scripts/aggregate_evals.js artifacts --compare-main --pr-comment >> report.md
          cat report.md >> "$GITHUB_STEP_SUMMARY"

      - name: 'Comment on PR'
        env:
          GH_TOKEN: '${{ secrets.GITHUB_TOKEN }}'
        run: |
          if [ ! -f "report.md" ]; then
            exit 0
          fi
          PR_NUMBER=${{ github.event.pull_request.number }}
          EXISTING_COMMENT=$(gh pr view $PR_NUMBER --json comments --jq '.comments[] | select(.body | contains("eval-impact-report")) | .id' | head -n 1)
          
          if [ -n "$EXISTING_COMMENT" ]; then
            gh pr comment $PR_NUMBER --body-file report.md
          else
            gh pr comment $PR_NUMBER --body-file report.md
          fi