# GitHub Actions Workflow for DSPy Evaluations

## Overview

This document describes the GitHub Actions workflow for automated DSPy evaluations on every PR and merge.

## Workflow Configuration

```yaml
# .github/workflows/dspy-eval.yml
name: DSPy RAG Evaluation

on:
  push:
    branches: [main]
    paths:
      - 'backend/rag/**'
      - 'tests/dspy_gitops/**'
      - 'src/glam_extractor/api/**'
  pull_request:
    branches: [main]
    paths:
      - 'backend/rag/**'
      - 'tests/dspy_gitops/**'
      - 'src/glam_extractor/api/**'
  workflow_dispatch:
    inputs:
      evaluation_level:
        description: 'Evaluation depth'
        required: true
        default: 'standard'
        type: choice
        options:
          - smoke
          - standard
          - comprehensive
  schedule:
    # Nightly comprehensive evaluation at 2 AM UTC
    - cron: '0 2 * * *'

env:
  PYTHON_VERSION: '3.11'
  OXIGRAPH_URL: 'http://91.98.224.44:7878'  # Production Oxigraph

jobs:
  # ==========================================================================
  # Layer 1: Fast Unit Tests (no LLM calls)
  # ==========================================================================
  unit-tests:
    name: Unit Tests
    runs-on: ubuntu-latest
    timeout-minutes: 5
    
    steps:
      - uses: actions/checkout@v4
      
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: ${{ env.PYTHON_VERSION }}
          cache: 'pip'
      
      - name: Install dependencies
        run: |
          pip install -e ".[dev]"
          pip install pytest pytest-cov rapidfuzz
      
      - name: Run unit tests
        run: |
          pytest tests/dspy_gitops/test_layer1_unit.py \
            -v --tb=short \
            --cov=backend/rag \
            --cov-report=xml
      
      - name: Upload coverage
        uses: codecov/codecov-action@v4
        with:
          files: ./coverage.xml
          flags: dspy-unit

  # ==========================================================================
  # Layer 2: DSPy Module Tests (with LLM)
  # ==========================================================================
  dspy-module-tests:
    name: DSPy Module Tests
    runs-on: ubuntu-latest
    timeout-minutes: 15
    needs: unit-tests
    
    # Only run on PRs or scheduled runs
    if: github.event_name == 'pull_request' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
    
    steps:
      - uses: actions/checkout@v4
      
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: ${{ env.PYTHON_VERSION }}
          cache: 'pip'
      
      - name: Install dependencies
        run: |
          pip install -e ".[dev]"
          pip install dspy-ai pytest httpx rapidfuzz
      
      - name: Run DSPy module tests
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
          OXIGRAPH_ENDPOINT: ${{ env.OXIGRAPH_URL }}
        run: |
          pytest tests/dspy_gitops/test_layer2_dspy.py \
            -v --tb=short \
            --junit-xml=dspy-results.xml
      
      - name: Upload test results
        uses: actions/upload-artifact@v4
        if: always()
        with:
          name: dspy-test-results
          path: dspy-results.xml
      
      - name: Comment PR with results
        if: github.event_name == 'pull_request'
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            // Parse JUnit XML and create comment
            // (simplified - use proper XML parser in production)
            const results = fs.readFileSync('dspy-results.xml', 'utf8');
            const passed = (results.match(/tests="(\d+)"/)?.[1] || 0);
            const failed = (results.match(/failures="(\d+)"/)?.[1] || 0);
            
            const body = `## DSPy Evaluation Results
            
            | Metric | Value |
            |--------|-------|
            | Tests Passed | ${passed} |
            | Tests Failed | ${failed} |
            | Status | ${failed > 0 ? '❌ FAILED' : '✅ PASSED'} |
            `;
            
            github.rest.issues.createComment({
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
              body: body
            });

  # ==========================================================================
  # Layer 3: Integration Tests (live API)
  # ==========================================================================
  integration-tests:
    name: Integration Tests
    runs-on: ubuntu-latest
    timeout-minutes: 10
    needs: unit-tests
    
    steps:
      - uses: actions/checkout@v4
      
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: ${{ env.PYTHON_VERSION }}
          cache: 'pip'
      
      - name: Install dependencies
        run: |
          pip install -e ".[dev]"
          pip install pytest httpx
      
      - name: Check API health
        run: |
          # Verify Oxigraph is accessible
          curl -sf "${OXIGRAPH_URL}/query" \
            -H "Accept: application/sparql-results+json" \
            --data-urlencode "query=SELECT (COUNT(*) as ?c) WHERE { ?s ?p ?o } LIMIT 1" \
            || (echo "Oxigraph not accessible" && exit 1)
      
      - name: Run integration tests
        env:
          OXIGRAPH_ENDPOINT: ${{ env.OXIGRAPH_URL }}
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          pytest tests/dspy_gitops/test_layer3_integration.py \
            -v --tb=short \
            --junit-xml=integration-results.xml
      
      - name: Upload test results
        uses: actions/upload-artifact@v4
        if: always()
        with:
          name: integration-test-results
          path: integration-results.xml

  # ==========================================================================
  # Layer 4: Comprehensive Evaluation (nightly only)
  # ==========================================================================
  comprehensive-eval:
    name: Comprehensive Evaluation
    runs-on: ubuntu-latest
    timeout-minutes: 60
    needs: [unit-tests, dspy-module-tests, integration-tests]
    
    # Only run on schedule or manual trigger with 'comprehensive'
    if: github.event_name == 'schedule' || (github.event_name == 'workflow_dispatch' && github.event.inputs.evaluation_level == 'comprehensive')
    
    steps:
      - uses: actions/checkout@v4
      
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: ${{ env.PYTHON_VERSION }}
          cache: 'pip'
      
      - name: Install dependencies
        run: |
          pip install -e ".[dev]"
          pip install dspy-ai pytest httpx rapidfuzz pandas
      
      - name: Run comprehensive evaluation
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
          OXIGRAPH_ENDPOINT: ${{ env.OXIGRAPH_URL }}
        run: |
          pytest tests/dspy_gitops/test_layer4_comprehensive.py \
            -v --tb=short \
            --junit-xml=comprehensive-results.xml \
            --json-report \
            --json-report-file=eval-report.json
      
      - name: Save evaluation metrics
        run: |
          python -c "
          import json
          with open('eval-report.json') as f:
              report = json.load(f)
          
          # Extract key metrics
          metrics = {
              'timestamp': '$(date -u +%Y-%m-%dT%H:%M:%SZ)',
              'commit': '${{ github.sha }}',
              'total_tests': report.get('summary', {}).get('total', 0),
              'passed': report.get('summary', {}).get('passed', 0),
              'failed': report.get('summary', {}).get('failed', 0),
          }
          
          with open('metrics.json', 'w') as f:
              json.dump(metrics, f, indent=2)
          "
      
      - name: Upload evaluation artifacts
        uses: actions/upload-artifact@v4
        with:
          name: comprehensive-eval-results
          path: |
            comprehensive-results.xml
            eval-report.json
            metrics.json
      
      - name: Post to Slack (optional)
        if: failure()
        uses: slackapi/slack-github-action@v1
        with:
          payload: |
            {
              "text": "🚨 DSPy Comprehensive Evaluation Failed",
              "blocks": [
                {
                  "type": "section",
                  "text": {
                    "type": "mrkdwn",
                    "text": "*DSPy Nightly Evaluation Failed*\n<${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}|View Run>"
                  }
                }
              ]
            }
        env:
          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}

  # ==========================================================================
  # Quality Gate Check
  # ==========================================================================
  quality-gate:
    name: Quality Gate
    runs-on: ubuntu-latest
    needs: [unit-tests, dspy-module-tests, integration-tests]
    if: always()
    
    steps:
      - name: Check all tests passed
        run: |
          if [[ "${{ needs.unit-tests.result }}" != "success" ]]; then
            echo "Unit tests failed"
            exit 1
          fi
          
          # DSPy module tests are optional for push, required for PR
          if [[ "${{ github.event_name }}" == "pull_request" ]]; then
            if [[ "${{ needs.dspy-module-tests.result }}" != "success" ]]; then
              echo "DSPy module tests failed"
              exit 1
            fi
          fi
          
          if [[ "${{ needs.integration-tests.result }}" != "success" ]]; then
            echo "Integration tests failed"
            exit 1
          fi
          
          echo "All quality gates passed!"
```

## Required Secrets

Configure these in GitHub repository settings:

| Secret | Description |
|--------|-------------|
| `ANTHROPIC_API_KEY` | API key for Claude LLM |
| `SLACK_WEBHOOK_URL` | (Optional) Slack notifications |

## Test Markers for Selective Execution

```python
# pytest.ini or pyproject.toml
[tool.pytest.ini_options]
markers = [
    "layer1: fast unit tests without LLM",
    "layer2: DSPy module tests with LLM",
    "layer3: integration tests with live API",
    "layer4: comprehensive evaluation",
    "smoke: quick smoke tests",
    "slow: tests that take > 10 seconds",
    "requires_oxigraph: tests that need Oxigraph connection",
    "requires_llm: tests that need LLM API access",
]
```

## Local Development

Run specific layers locally:

```bash
# Layer 1 only (fast, no LLM)
pytest tests/dspy_gitops/test_layer1_unit.py -v

# Layer 2 only (needs ANTHROPIC_API_KEY)
pytest tests/dspy_gitops/test_layer2_dspy.py -v

# Smoke tests only
pytest tests/dspy_gitops/ -m smoke -v

# All except slow tests
pytest tests/dspy_gitops/ -m "not slow" -v
```