test-deduplicator

Test Deduplicator

Overview

This skill analyzes test suites to identify redundant or duplicate tests by examining code coverage, semantic similarity, and execution behavior. It groups equivalent tests, explains deduplication rationale, and recommends which tests can be safely removed while preserving overall test effectiveness.

Workflow

1. Collect Test Coverage Data

First, gather coverage information for each test in the suite.

For Python projects with pytest:

# Run tests with coverage for each test individually
pytest --cov=src --cov-report=json tests/

# Or use the coverage analyzer script with pre-collected data
python scripts/coverage_analyzer.py coverage_data.json --output coverage_analysis.json

Coverage data format:

{
  "test_module::test_function_1": {
    "coverage": {
      "src/module.py": [10, 11, 12, 15, 20],
      "src/utils.py": [5, 6, 7]
    }
  },
  "test_module::test_function_2": {
    "coverage": {
      "src/module.py": [10, 11, 12, 15, 20],
      "src/utils.py": [5, 6, 7]
    }
  }
}

2. Analyze Semantic Similarity

Parse test code to identify semantic similarities:

python scripts/semantic_analyzer.py tests/*.py --output semantic_analysis.json --threshold 0.7

This analyzes:

• Assertion patterns
• Function call sequences
• Code structure similarity
• Variable usage patterns

3. Generate Deduplication Recommendations

Combine coverage and semantic analysis to produce recommendations:

python scripts/deduplicator.py \
  --coverage coverage_analysis.json \
  --semantic semantic_analysis.json \
  --output recommendations.json \
  --threshold 0.8

4. Review Recommendations

Examine the generated recommendations:

{
  "redundant_groups": [
    {
      "type": "identical_coverage",
      "tests": ["test_a", "test_b", "test_c"],
      "reason": "Tests have identical code coverage",
      "confidence": 1.0,
      "keep": "test_a",
      "remove": ["test_b", "test_c"],
      "rationale": "test_a has descriptive name, fast execution"
    }
  ],
  "removal_candidates": [...],
  "coverage_impact": {
    "coverage_preserved": true,
    "tests_removed": 15
  }
}