Add unknown_analysis/tests/test_core.py

unknown_analysis/tests/test_core.py

@@ -0,0 +1,61 @@
+import pytest
+from unknown_analysis import core
+
+@pytest.fixture
+def sample_log_data():
+    return [
+        {"artifact_key": "a1", "status": "PASS", "cause": "", "path": "p1", "error": ""},
+        {"artifact_key": "a2", "status": "UNKNOWN", "cause": "missing", "path": "p2", "error": "FileNotFoundError"},
+        {"artifact_key": "a3", "status": "UNKNOWN", "cause": "schema", "path": "p3", "error": "SchemaMismatch"},
+        {"artifact_key": "a4", "status": "PASS", "cause": "", "path": "p4", "error": ""},
+        {"artifact_key": "a5", "status": "UNKNOWN", "cause": "missing", "path": "p5", "error": "FileNotFoundError"}
+    ]
+
+@pytest.fixture
+def empty_log_data():
+    return []
+
+def test_calculate_unknown_rates_basic(sample_log_data):
+    result = core.calculate_unknown_rates(sample_log_data)
+    assert isinstance(result, dict)
+    assert set(result.keys()) == {"unknown_artifact_missing_rate", "unknown_schema_rate"}
+    total_unknowns = 3
+    missing_count = 2
+    schema_count = 1
+    assert pytest.approx(result['unknown_artifact_missing_rate'], 0.1) == missing_count / total_unknowns
+    assert pytest.approx(result['unknown_schema_rate'], 0.1) == schema_count / total_unknowns
+
+def test_calculate_unknown_rates_empty(empty_log_data):
+    result = core.calculate_unknown_rates(empty_log_data)
+    assert result['unknown_artifact_missing_rate'] == 0.0
+    assert result['unknown_schema_rate'] == 0.0
+
+def test_calculate_unknown_rates_invalid_input():
+    with pytest.raises((TypeError, KeyError)):
+        core.calculate_unknown_rates([{"invalid": True}])
+
+def test_get_top_pass_unknown_switches_basic(sample_log_data):
+    result = core.get_top_pass_unknown_switches(sample_log_data)
+    assert isinstance(result, list)
+    assert all(isinstance(item, dict) for item in result)
+    if result:
+        for entry in result:
+            assert set(entry.keys()) == {"cause", "path", "error"}
+
+def test_get_top_pass_unknown_switches_empty(empty_log_data):
+    result = core.get_top_pass_unknown_switches(empty_log_data)
+    assert result == []
+
+def test_get_top_pass_unknown_switches_structure(sample_log_data):
+    top_switches = core.get_top_pass_unknown_switches(sample_log_data)
+    # Ensure all required fields exist and are strings
+    for s in top_switches:
+        assert isinstance(s['cause'], str)
+        assert isinstance(s['path'], str)
+        assert isinstance(s['error'], str)
+
+def test_integration_rates_and_switches(sample_log_data):
+    rates = core.calculate_unknown_rates(sample_log_data)
+    switches = core.get_top_pass_unknown_switches(sample_log_data)
+    assert 'unknown_artifact_missing_rate' in rates
+    assert isinstance(switches, list)