Add audit_data_processing/tests/test_core.py

audit_data_processing/tests/test_core.py

@@ -0,0 +1,69 @@
+import pytest
+import pandas as pd
+from audit_data_processing import core
+
+
+@pytest.fixture
+def sample_data():
+    # Beispielhafte Audit-Run-Daten mit warn_rate und unknown_rate
+    class RunData:
+        def __init__(self, run_id, warn_rate, unknown_rate, pinned):
+            self.run_id = run_id
+            self.warn_rate = warn_rate
+            self.unknown_rate = unknown_rate
+            self.pinned = pinned
+
+    return [
+        RunData("r1", 0.1, 0.05, True),
+        RunData("r2", 0.2, 0.10, False),
+        RunData("r3", 0.3, 0.15, True),
+        RunData("r4", 0.4, 0.20, False),
+        RunData("r5", 0.5, 0.25, True)
+    ]
+
+
+def test_calculate_percentiles_basic(sample_data):
+    percentiles = [50, 75, 90, 95]
+    result = core.calculate_percentiles(sample_data, percentiles)
+
+    assert isinstance(result, dict)
+    assert set(result.keys()) == {"warn_rate", "unknown_rate"}
+
+    # Prüfe, dass alle verlangten Prozentile berechnet wurden
+    for metric in result.values():
+        for p in percentiles:
+            assert p in metric
+            assert isinstance(metric[p], float)
+
+
+def test_calculate_percentiles_values(sample_data):
+    # Erwartete Werte via Pandas für Vergleich berechnen
+    df = pd.DataFrame([
+        {"warn_rate": d.warn_rate, "unknown_rate": d.unknown_rate}
+        for d in sample_data
+    ])
+    percentiles = [50, 75, 90, 95]
+    expected_warn = {
+        p: float(df["warn_rate"].quantile(p / 100)) for p in percentiles
+    }
+    expected_unknown = {
+        p: float(df["unknown_rate"].quantile(p / 100)) for p in percentiles
+    }
+
+    result = core.calculate_percentiles(sample_data, percentiles)
+
+    for p in percentiles:
+        assert pytest.approx(result["warn_rate"][p], rel=1e-9) == expected_warn[p]
+        assert pytest.approx(result["unknown_rate"][p], rel=1e-9) == expected_unknown[p]
+
+
+def test_calculate_percentiles_empty():
+    # Fehlerfall: leere Liste sollte leer oder Exception liefern
+    with pytest.raises((ValueError, ZeroDivisionError, KeyError, IndexError)):
+        core.calculate_percentiles([], [50])
+
+
+def test_calculate_percentiles_invalid_input():
+    # Fehlerfall: Daten ohne erforderliche Attribute
+    with pytest.raises(Exception):
+        core.calculate_percentiles([{"warn_rate": 0.1}], [50])