Add stability_analysis/tests/test_core.py

stability_analysis/tests/test_core.py

@@ -0,0 +1,57 @@
+import pytest
+import pandas as pd
+import numpy as np
+from stability_analysis.core import analyze_data
+
+@pytest.fixture
+def sample_data():
+    return [
+        {"expires_at_dist_hours": 1.5, "retry_total_overhead_ms": 10},
+        {"expires_at_dist_hours": 1.6, "retry_total_overhead_ms": 11},
+        {"expires_at_dist_hours": 5.2, "retry_total_overhead_ms": 150},  # outlier
+        {"expires_at_dist_hours": 1.4, "retry_total_overhead_ms": 9},
+        {"expires_at_dist_hours": 1.5, "retry_total_overhead_ms": 10}
+    ]
+
+
+def test_analyze_data_returns_expected_structure(sample_data):
+    result = analyze_data(sample_data)
+    assert isinstance(result, dict)
+    for key in ("stable_cluster", "outlier_counts", "patterns"):
+        assert key in result, f"Missing key: {key}"
+
+    assert isinstance(result["stable_cluster"], list)
+    assert isinstance(result["outlier_counts"], dict)
+    assert isinstance(result["patterns"], list)
+
+
+def test_outlier_detection(sample_data):
+    result = analyze_data(sample_data)
+    assert any(count > 0 for count in result["outlier_counts"].values()), "Outliers should be detected"
+
+
+def test_stable_cluster_identification(sample_data):
+    result = analyze_data(sample_data)
+    assert len(result["stable_cluster"]) >= 1, "Should identify at least one stable cluster"
+
+
+def test_pattern_recognition(sample_data):
+    result = analyze_data(sample_data)
+    # Should return some patterns based on grouped similarities
+    assert isinstance(result["patterns"], list)
+    assert all(isinstance(p, str) for p in result["patterns"])
+
+
+def test_invalid_input_raises():
+    with pytest.raises((TypeError, ValueError)):
+        analyze_data(None)
+    with pytest.raises((TypeError, ValueError)):
+        analyze_data([{"expires_at_dist_hours": "NaN"}])
+
+
+def test_edge_case_empty_input():
+    result = analyze_data([])
+    assert isinstance(result, dict)
+    assert result["stable_cluster"] == []
+    assert result["outlier_counts"] == {}
+    assert result["patterns"] == []