name: Reachability Benchmark

# Sprint: SPRINT_3500_0003_0001
# Task: CORPUS-009 - Create Gitea workflow for reachability benchmark
# Task: CORPUS-010 - Configure nightly + per-PR benchmark runs

on:
  workflow_dispatch:
    inputs:
      baseline_version:
        description: 'Baseline version to compare against'
        required: false
        default: 'latest'
      verbose:
        description: 'Enable verbose output'
        required: false
        type: boolean
        default: false
  push:
    branches: [ main ]
    paths:
      - 'datasets/reachability/**'
      - 'src/Scanner/__Libraries/StellaOps.Scanner.Benchmarks/**'
      - 'bench/reachability-benchmark/**'
      - '.gitea/workflows/reachability-bench.yaml'
  pull_request:
    paths:
      - 'datasets/reachability/**'
      - 'src/Scanner/__Libraries/StellaOps.Scanner.Benchmarks/**'
      - 'bench/reachability-benchmark/**'
  schedule:
    # Nightly at 02:00 UTC
    - cron: '0 2 * * *'

jobs:
  benchmark:
    runs-on: ubuntu-22.04
    env:
      DOTNET_NOLOGO: 1
      DOTNET_CLI_TELEMETRY_OPTOUT: 1
      DOTNET_SYSTEM_GLOBALIZATION_INVARIANT: 1
      TZ: UTC
      STELLAOPS_OFFLINE: 'true'
      STELLAOPS_DETERMINISTIC: 'true'
    outputs:
      precision: ${{ steps.metrics.outputs.precision }}
      recall: ${{ steps.metrics.outputs.recall }}
      f1: ${{ steps.metrics.outputs.f1 }}
      pr_auc: ${{ steps.metrics.outputs.pr_auc }}
      regression: ${{ steps.compare.outputs.regression }}
      
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          fetch-depth: 0

      - name: Setup .NET 10
        uses: actions/setup-dotnet@v4
        with:
          dotnet-version: 10.0.100
          include-prerelease: true

      - name: Cache NuGet packages
        uses: actions/cache@v4
        with:
          path: ~/.nuget/packages
          key: ${{ runner.os }}-nuget-${{ hashFiles('**/*.csproj') }}
          restore-keys: |
            ${{ runner.os }}-nuget-

      - name: Restore benchmark project
        run: |
          dotnet restore src/Scanner/__Libraries/StellaOps.Scanner.Benchmarks/StellaOps.Scanner.Benchmarks.csproj \
            --configfile nuget.config

      - name: Build benchmark project
        run: |
          dotnet build src/Scanner/__Libraries/StellaOps.Scanner.Benchmarks/StellaOps.Scanner.Benchmarks.csproj \
            -c Release \
            --no-restore

      - name: Validate corpus integrity
        run: |
          echo "::group::Validating corpus index"
          if [ ! -f datasets/reachability/corpus.json ]; then
            echo "::error::corpus.json not found"
            exit 1
          fi
          python3 -c "import json; data = json.load(open('datasets/reachability/corpus.json')); print(f'Corpus contains {len(data.get(\"samples\", []))} samples')"
          echo "::endgroup::"

      - name: Run benchmark
        id: benchmark
        run: |
          echo "::group::Running reachability benchmark"
          mkdir -p bench/results
          
          # Run the corpus benchmark
          dotnet run \
            --project src/Scanner/__Libraries/StellaOps.Scanner.Benchmarks/StellaOps.Scanner.Benchmarks.csproj \
            -c Release \
            --no-build \
            -- corpus run \
            --corpus datasets/reachability/corpus.json \
            --output bench/results/benchmark-${{ github.sha }}.json \
            --format json \
            ${{ inputs.verbose == 'true' && '--verbose' || '' }}
          
          echo "::endgroup::"

      - name: Extract metrics
        id: metrics
        run: |
          echo "::group::Extracting metrics"
          RESULT_FILE="bench/results/benchmark-${{ github.sha }}.json"
          
          if [ -f "$RESULT_FILE" ]; then
            PRECISION=$(jq -r '.metrics.precision // 0' "$RESULT_FILE")
            RECALL=$(jq -r '.metrics.recall // 0' "$RESULT_FILE")
            F1=$(jq -r '.metrics.f1 // 0' "$RESULT_FILE")
            PR_AUC=$(jq -r '.metrics.pr_auc // 0' "$RESULT_FILE")
            
            echo "precision=$PRECISION" >> $GITHUB_OUTPUT
            echo "recall=$RECALL" >> $GITHUB_OUTPUT
            echo "f1=$F1" >> $GITHUB_OUTPUT
            echo "pr_auc=$PR_AUC" >> $GITHUB_OUTPUT
            
            echo "Precision: $PRECISION"
            echo "Recall: $RECALL"
            echo "F1: $F1"
            echo "PR-AUC: $PR_AUC"
          else
            echo "::error::Benchmark result file not found"
            exit 1
          fi
          echo "::endgroup::"

      - name: Get baseline
        id: baseline
        run: |
          echo "::group::Loading baseline"
          BASELINE_VERSION="${{ inputs.baseline_version || 'latest' }}"
          
          if [ "$BASELINE_VERSION" = "latest" ]; then
            BASELINE_FILE=$(ls -t bench/baselines/*.json 2>/dev/null | head -1)
          else
            BASELINE_FILE="bench/baselines/$BASELINE_VERSION.json"
          fi
          
          if [ -f "$BASELINE_FILE" ]; then
            echo "baseline_file=$BASELINE_FILE" >> $GITHUB_OUTPUT
            echo "Using baseline: $BASELINE_FILE"
          else
            echo "::warning::No baseline found, skipping comparison"
            echo "baseline_file=" >> $GITHUB_OUTPUT
          fi
          echo "::endgroup::"

      - name: Compare to baseline
        id: compare
        if: steps.baseline.outputs.baseline_file != ''
        run: |
          echo "::group::Comparing to baseline"
          BASELINE_FILE="${{ steps.baseline.outputs.baseline_file }}"
          RESULT_FILE="bench/results/benchmark-${{ github.sha }}.json"
          
          # Extract baseline metrics
          BASELINE_PRECISION=$(jq -r '.metrics.precision // 0' "$BASELINE_FILE")
          BASELINE_RECALL=$(jq -r '.metrics.recall // 0' "$BASELINE_FILE")
          BASELINE_PR_AUC=$(jq -r '.metrics.pr_auc // 0' "$BASELINE_FILE")
          
          # Extract current metrics
          CURRENT_PRECISION=$(jq -r '.metrics.precision // 0' "$RESULT_FILE")
          CURRENT_RECALL=$(jq -r '.metrics.recall // 0' "$RESULT_FILE")
          CURRENT_PR_AUC=$(jq -r '.metrics.pr_auc // 0' "$RESULT_FILE")
          
          # Calculate deltas
          PRECISION_DELTA=$(echo "$CURRENT_PRECISION - $BASELINE_PRECISION" | bc -l)
          RECALL_DELTA=$(echo "$CURRENT_RECALL - $BASELINE_RECALL" | bc -l)
          PR_AUC_DELTA=$(echo "$CURRENT_PR_AUC - $BASELINE_PR_AUC" | bc -l)
          
          echo "Precision delta: $PRECISION_DELTA"
          echo "Recall delta: $RECALL_DELTA"
          echo "PR-AUC delta: $PR_AUC_DELTA"
          
          # Check for regression (PR-AUC drop > 2%)
          REGRESSION_THRESHOLD=-0.02
          if (( $(echo "$PR_AUC_DELTA < $REGRESSION_THRESHOLD" | bc -l) )); then
            echo "::error::PR-AUC regression detected: $PR_AUC_DELTA (threshold: $REGRESSION_THRESHOLD)"
            echo "regression=true" >> $GITHUB_OUTPUT
          else
            echo "regression=false" >> $GITHUB_OUTPUT
          fi
          echo "::endgroup::"

      - name: Generate markdown report
        run: |
          echo "::group::Generating report"
          RESULT_FILE="bench/results/benchmark-${{ github.sha }}.json"
          REPORT_FILE="bench/results/benchmark-${{ github.sha }}.md"
          
          cat > "$REPORT_FILE" << 'EOF'
          # Reachability Benchmark Report
          
          **Commit:** ${{ github.sha }}
          **Run:** ${{ github.run_number }}
          **Date:** $(date -u +"%Y-%m-%dT%H:%M:%SZ")
          
          ## Metrics
          
          | Metric | Value |
          |--------|-------|
          | Precision | ${{ steps.metrics.outputs.precision }} |
          | Recall | ${{ steps.metrics.outputs.recall }} |
          | F1 Score | ${{ steps.metrics.outputs.f1 }} |
          | PR-AUC | ${{ steps.metrics.outputs.pr_auc }} |
          
          ## Comparison
          
          ${{ steps.compare.outputs.regression == 'true' && '⚠️ **REGRESSION DETECTED**' || '✅ No regression' }}
          EOF
          
          echo "Report generated: $REPORT_FILE"
          echo "::endgroup::"

      - name: Upload results
        uses: actions/upload-artifact@v4
        with:
          name: benchmark-results-${{ github.sha }}
          path: |
            bench/results/benchmark-${{ github.sha }}.json
            bench/results/benchmark-${{ github.sha }}.md
          retention-days: 90

      - name: Fail on regression
        if: steps.compare.outputs.regression == 'true' && github.event_name == 'pull_request'
        run: |
          echo "::error::Benchmark regression detected. PR-AUC dropped below threshold."
          exit 1

  update-baseline:
    needs: benchmark
    if: github.event_name == 'push' && github.ref == 'refs/heads/main' && needs.benchmark.outputs.regression != 'true'
    runs-on: ubuntu-22.04
    steps:
      - name: Checkout
        uses: actions/checkout@v4

      - name: Download results
        uses: actions/download-artifact@v4
        with:
          name: benchmark-results-${{ github.sha }}
          path: bench/results/

      - name: Update baseline (nightly only)
        if: github.event_name == 'schedule'
        run: |
          DATE=$(date +%Y%m%d)
          cp bench/results/benchmark-${{ github.sha }}.json bench/baselines/baseline-$DATE.json
          echo "Updated baseline to baseline-$DATE.json"

  notify-pr:
    needs: benchmark
    if: github.event_name == 'pull_request'
    runs-on: ubuntu-22.04
    permissions:
      pull-requests: write
    steps:
      - name: Comment on PR
        uses: actions/github-script@v7
        with:
          script: |
            const precision = '${{ needs.benchmark.outputs.precision }}';
            const recall = '${{ needs.benchmark.outputs.recall }}';
            const f1 = '${{ needs.benchmark.outputs.f1 }}';
            const prAuc = '${{ needs.benchmark.outputs.pr_auc }}';
            const regression = '${{ needs.benchmark.outputs.regression }}' === 'true';
            
            const status = regression ? '⚠️ REGRESSION' : '✅ PASS';
            
            const body = `## Reachability Benchmark Results ${status}
            
            | Metric | Value |
            |--------|-------|
            | Precision | ${precision} |
            | Recall | ${recall} |
            | F1 Score | ${f1} |
            | PR-AUC | ${prAuc} |
            
            ${regression ? '### ⚠️ Regression Detected\nPR-AUC dropped below threshold. Please review changes.' : ''}
            
            <details>
            <summary>Details</summary>
            
            - Commit: \`${{ github.sha }}\`
            - Run: [#${{ github.run_number }}](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }})
            
            </details>`;
            
            github.rest.issues.createComment({
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
              body: body
            });