Add outlier_analysis/tests/test_core.py

outlier_analysis/tests/test_core.py

@@ -0,0 +1,66 @@
+import pytest
+import math
+from typing import List, Dict
+from outlier_analysis import core
+
+
+@pytest.fixture
+def sample_log_data() -> List[Dict]:
+    return [
+        {"run_id": "18", "latency_ms": 45.0, "stratum": "near-expiry-unpinned", "job_parallelism": 4, "retry_total_overhead_ms": 0.0},
+        {"run_id": "18", "latency_ms": 120.0, "stratum": "near-expiry-unpinned", "job_parallelism": 4, "retry_total_overhead_ms": 10.0},
+        {"run_id": "18", "latency_ms": 95.0, "stratum": "pinned", "job_parallelism": 4, "retry_total_overhead_ms": 5.0},
+        {"run_id": "19", "latency_ms": 80.0, "stratum": "pinned", "job_parallelism": 8, "retry_total_overhead_ms": 0.0},
+        {"run_id": "19", "latency_ms": 101.0, "stratum": "unpinned", "job_parallelism": 8, "retry_total_overhead_ms": 3.5},
+        {"run_id": "19", "latency_ms": 200.0, "stratum": "unpinned", "job_parallelism": 8, "retry_total_overhead_ms": 15.0},
+        {"run_id": "20", "latency_ms": 49.0, "stratum": "pinned", "job_parallelism": 2, "retry_total_overhead_ms": 0.0},
+        {"run_id": "20", "latency_ms": 52.0, "stratum": "pinned", "job_parallelism": 2, "retry_total_overhead_ms": 0.0}
+    ]
+
+
+def test_analyze_outliers_returns_valid_structure(sample_log_data):
+    result = core.analyze_outliers(sample_log_data)
+
+    assert isinstance(result, dict)
+    assert all(isinstance(v, dict) for v in result.values())
+
+    for run_id, summary in result.items():
+        assert set(summary.keys()) == {"run_id", "outlier_count", "latency_distribution"}
+        assert isinstance(summary["run_id"], str)
+        assert isinstance(summary["outlier_count"], int)
+        dist = summary["latency_distribution"]
+        assert isinstance(dist, dict)
+        for key in ("p50", "p95", "p99", "max"):
+            assert key in dist
+            assert isinstance(dist[key], (int, float))
+
+
+def test_outlier_detection_counts(sample_log_data):
+    result = core.analyze_outliers(sample_log_data)
+    # Run 18 has 2 entries >90ms, expect 2 outliers
+    assert result["18"]["outlier_count"] == 2
+    # Run 19 has 2 entries >90ms as well
+    assert result["19"]["outlier_count"] == 2
+    # Run 20 has none >90ms
+    assert result["20"]["outlier_count"] == 0
+
+
+def test_latency_distribution_values(sample_log_data):
+    result = core.analyze_outliers(sample_log_data)
+    for run_id, summary in result.items():
+        dist = summary["latency_distribution"]
+        for k, v in dist.items():
+            assert v >= 0.0, f"Latency metric {k} should be non-negative"
+        # Monotonic: p50 <= p95 <= p99 <= max
+        assert dist["p50"] <= dist["p95"] <= dist["p99"] <= dist["max"]
+
+
+def test_empty_input_returns_empty():
+    result = core.analyze_outliers([])
+    assert isinstance(result, dict)
+    assert result == {}
+
+
+def test_invalid_input_type_raises():
+    with pytest.raises((AssertionError, TypeError)):
+        core.analyze_outliers(None)  # type: ignore