10.2The Correlation Framework

# test_production_correlation.py
"""
Correlate test suite results with production incident data to answer:
- Which tests, if they had existed, would have caught recent incidents?
- Which tests are "load-bearing" (their failure predicts production issues)?
- Which tests are "dead weight" (always pass, never catch real bugs)?
"""

def analyze_test_production_correlation(test_results: list, incidents: list) -> dict:
    """
    For each production incident, determine if any test in the suite
    covers the affected component and failure mode.
    """
    coverage_gaps = []
    load_bearing_tests = set()

    for incident in incidents:
        affected_component = incident["component"]
        failure_mode = incident["failure_mode"]

        # Find tests that cover this component
        covering_tests = [
            t for t in test_results
            if t["component"] == affected_component
        ]

        if not covering_tests:
            coverage_gaps.append({
                "incident": incident["id"],
                "component": affected_component,
                "failure_mode": failure_mode,
                "gap_type": "no_test_coverage",
                "recommendation": f"Add tests for {affected_component} {failure_mode}",
            })
            continue

        # Check if any covering test was failing before the incident
        pre_incident_failures = [
            t for t in covering_tests
            if t["last_failure"] and t["last_failure"] < incident["detected_at"]
            and (incident["detected_at"] - t["last_failure"]).days < 7
        ]

        if pre_incident_failures:
            for test in pre_incident_failures:
                load_bearing_tests.add(test["name"])
        else:
            coverage_gaps.append({
                "incident": incident["id"],
                "component": affected_component,
                "failure_mode": failure_mode,
                "gap_type": "tests_pass_but_incident_occurred",
                "recommendation": (
                    f"Tests for {affected_component} may not cover "
                    f"the {failure_mode} failure mode"
                ),
            })

    return {
        "coverage_gaps": coverage_gaps,
        "load_bearing_tests": list(load_bearing_tests),
        "gap_count": len(coverage_gaps),
        "incident_count": len(incidents),
        "coverage_percentage": (
            (len(incidents) - len(coverage_gaps)) / len(incidents) * 100
            if incidents else 100
        ),
    }