performance-test.yml

name: Performance Test

on:
  push:
    branches: [ main, dev ]
  pull_request:
    branches: [ main, dev ]
  workflow_dispatch:  # Enables manual triggering
  schedule:
    - cron: '0 2 * * *'  # Runs daily at 2 AM UTC

jobs:
  performance:
    name: "Performance Test"
    runs-on: ubuntu-latest
    timeout-minutes: 60  # Set a timeout to prevent jobs from running indefinitely
    
    steps:
      - uses: actions/checkout@v4

      - name: Set up Python
        uses: actions/setup-python@v4
        with:
          python-version: '3.10'
          
      - name: Install dependencies
        run: |
          python -m pip install --upgrade pip
          python -m pip install --upgrade -r requirements.txt
          python -m pip install -e .  # Editable install ensures we test the actual source code
      
      - name: Test Owlery Connectivity
        run: |
          echo "Testing basic connectivity to Owlery server..."
          curl -v --max-time 30 "http://owl.virtualflybrain.org/kbs/vfb/subclasses?object=%3Chttp%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FFBbt_00005106%3E&direct=false&includeDeprecated=false&includeEquivalent=true" | head -20 || echo "Simple query failed or timed out"
          echo ""
          echo "Testing if server responds at all..."
          curl -v --max-time 10 "http://owl.virtualflybrain.org/kbs/vfb" || echo "Server unreachable"
          
      - name: Run Performance Test
        run: |
          python -m unittest src.test.test_query_performance -v 2>&1 | tee performance_test_output.log
        
      - name: Run Legacy Performance Test  
        env:
          VFBQUERY_CACHE_ENABLED: 'true'
          MPLBACKEND: 'Agg'
          VISPY_GL_LIB: 'osmesa'
          VISPY_USE_EGL: '0'
        run: |
          python -m unittest -v src.test.term_info_queries_test.TermInfoQueriesTest.test_term_info_performance 2>&1 | tee -a performance_test_output.log
        
      - name: Create Performance Report
        if: always()  # Always run this step, even if the test fails
        run: |
          # Create performance.md file
          cat > performance.md << EOF
          # VFBquery Performance Test Results
          
          **Test Date:** $(date -u '+%Y-%m-%d %H:%M:%S UTC')
          **Git Commit:** ${{ github.sha }}
          **Branch:** ${{ github.ref_name }}
          **Workflow Run:** [${{ github.run_id }}](https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }})
          
          ## Test Overview
          
          This performance test measures the execution time of all implemented VFB queries organized by functionality:
          
          ### 1. Term Information Queries
          
          - **Term Info**: Comprehensive term information retrieval with preview data
          
          ### 2. Neuron Part & Synaptic Queries
          
          - **NeuronsPartHere**: Neurons with parts overlapping anatomical regions
          - **NeuronsSynaptic**: Neurons with synapses in a region
          - **NeuronsPresynapticHere**: Neurons with presynaptic terminals in a region
          - **NeuronsPostsynapticHere**: Neurons with postsynaptic terminals in a region
          
          ### 3. Anatomical Hierarchy Queries
          
          - **ComponentsOf**: Anatomical components of a structure
          - **PartsOf**: Parts of an anatomical structure
          - **SubclassesOf**: Subclasses of anatomical terms (can be very slow for complex terms)
          
          ### 4. Tract/Nerve & Lineage Queries
          
          - **NeuronClassesFasciculatingHere**: Neurons fasciculating with tracts
          - **TractsNervesInnervatingHere**: Tracts/nerves innervating neuropils
          - **LineageClonesIn**: Lineage clones in neuropils (complex OWL reasoning)
          
          ### 5. Image & Developmental Queries
          
          - **ImagesNeurons**: Neuron images in anatomical regions
          - **ImagesThatDevelopFrom**: Developmental lineage images
          - **epFrag**: Expression pattern fragments
          - **ListAllAvailableImages**: All available images for a term
          
          ### 6. Connectivity Queries
          
          - **NeuronNeuronConnectivity**: Neuron-to-neuron connectivity
          - **NeuronRegionConnectivity**: Neuron-to-region connectivity
          - **NeuronInputsTo**: Individual neuron inputs
          
          ### 7. Similarity Queries (NBLAST & NeuronBridge)
          
          - **SimilarMorphologyTo**: NBLAST morphological similarity
          - **SimilarMorphologyToPartOf**: NBLAST to expression patterns (NBLASTexp)
          - **SimilarMorphologyToPartOfexp**: Reverse NBLASTexp
          - **SimilarMorphologyToNB**: NeuronBridge matches
          - **SimilarMorphologyToNBexp**: NeuronBridge for expression patterns
          
          ### 8. Expression & Transcriptomics Queries
          
          - **ExpressionOverlapsHere**: Expression patterns overlapping regions
          - **anatScRNAseqQuery**: scRNAseq clusters in anatomy
          - **clusterExpression**: Genes expressed in clusters
          - **expressionCluster**: Clusters expressing genes
          - **scRNAdatasetData**: Cluster data from scRNAseq datasets
          
          ### 9. Dataset & Template Queries
          
          - **PaintedDomains**: Template painted anatomy domains
          - **DatasetImages**: Images in datasets
          - **AllAlignedImages**: Images aligned to templates
          - **AlignedDatasets**: Datasets aligned to templates
          - **AllDatasets**: All available datasets
          
          ### 10. Publication & Transgene Queries
          
          - **TermsForPub**: Terms referencing publications
          - **TransgeneExpressionHere**: Transgene expression patterns in regions
          
          ## Performance Thresholds
          
          - **Fast queries**: < 1 second (SOLR lookups)
          - **Medium queries**: < 3 seconds (Owlery + SOLR)
          - **Slow queries**: < 10 seconds (Neo4j + complex processing)
          - **Very Slow queries**: < 31 seconds (Complex OWL reasoning - over 30 seconds)
          
          ## Test Results
          
          \`\`\`
          $(cat performance_test_output.log)
          \`\`\`
          
          ## Summary
          
          EOF
          
          # Check overall test status
          if grep -q "OK" performance_test_output.log || grep -q "Ran.*test" performance_test_output.log; then
            echo "✅ **Test Status**: Performance tests completed" >> performance.md
            echo "" >> performance.md
            
            # Count successes and failures
            TOTAL_TESTS=$(grep -c "^test_" performance_test_output.log || echo "0")
            FAILED_TESTS=$(grep -c "FAIL:" performance_test_output.log || echo "0")
            ERROR_TESTS=$(grep -c "ERROR:" performance_test_output.log || echo "0")
            PASSED_TESTS=$((TOTAL_TESTS - FAILED_TESTS - ERROR_TESTS))
            
            echo "### Test Statistics" >> performance.md
            echo "" >> performance.md
            echo "- **Total Tests**: ${TOTAL_TESTS}" >> performance.md
            echo "- **Passed**: ${PASSED_TESTS} ✅" >> performance.md
            echo "- **Failed**: ${FAILED_TESTS} ❌" >> performance.md
            echo "- **Errors**: ${ERROR_TESTS} ⚠️" >> performance.md
            echo "" >> performance.md
            
            # Extract timing information for key queries
            echo "### Query Performance Details" >> performance.md
            echo "" >> performance.md
            
            # Extract all timing lines
            if grep -q "seconds" performance_test_output.log; then
              echo "| Query | Duration | Status |" >> performance.md
              echo "|-------|----------|--------|" >> performance.md
              
              # Parse timing information
              grep -E "^(get_term_info|NeuronsPartHere|NeuronsSynaptic|NeuronsPresynapticHere|NeuronsPostsynapticHere|ComponentsOf|PartsOf|SubclassesOf|NeuronClassesFasciculatingHere|TractsNervesInnervatingHere|LineageClonesIn|ListAllAvailableImages):" performance_test_output.log | while read line; do
                QUERY=$(echo "$line" | sed 's/:.*//') 
                DURATION=$(echo "$line" | sed 's/.*: \([0-9.]*\)s.*/\1/')
                if echo "$line" | grep -q "✅"; then
                  STATUS="✅ Pass"
                else
                  STATUS="❌ Fail"
                fi
                echo "| $QUERY | ${DURATION}s | $STATUS |" >> performance.md
              done
            fi
            
            echo "" >> performance.md
            
            # Overall result
            if [ "$FAILED_TESTS" -eq "0" ] && [ "$ERROR_TESTS" -eq "0" ]; then
              echo "🎉 **Result**: All performance thresholds met!" >> performance.md
            else
              echo "⚠️ **Result**: Some performance thresholds exceeded or tests failed" >> performance.md
              echo "" >> performance.md
              echo "Please review the failed tests above. Common causes:" >> performance.md
              echo "- Network latency to VFB services" >> performance.md
              echo "- SOLR/Neo4j/Owlery server load" >> performance.md
              echo "- First-time cache population (expected to be slower)" >> performance.md
            fi
          else
            echo "❌ **Test Status**: Performance tests failed to run properly" >> performance.md
            echo "" >> performance.md
            echo "Please check the test output above for errors." >> performance.md
          fi
          
          echo "" >> performance.md
          echo "---" >> performance.md
          echo "" >> performance.md
          echo "## Historical Performance" >> performance.md
          echo "" >> performance.md
          echo "Track performance trends across commits:" >> performance.md
          echo "- [GitHub Actions History](https://github.com/${{ github.repository }}/actions/workflows/performance-test.yml)" >> performance.md
          echo "" >> performance.md
          echo "---" >> performance.md
          echo "*Last updated: $(date -u '+%Y-%m-%d %H:%M:%S UTC')*" >> performance.md
          
          # Also add to GitHub step summary
          echo "## Performance Test Report" >> $GITHUB_STEP_SUMMARY
          echo "Performance results have been saved to performance.md" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          cat performance.md >> $GITHUB_STEP_SUMMARY
          
      - name: Commit and Push Performance Report
        if: always() && github.ref == 'refs/heads/main'
        run: |
          git config --local user.email "action@github.com"
          git config --local user.name "GitHub Action"
          git add performance.md
          git diff --staged --quiet || git commit -m "Update performance test results [skip ci]"
          git push origin HEAD:main