fix TestColumnNumberOfDifferentNulls (#312)

* fix TestColumnNumberOfDifferentNulls

* style fix

* small unit test update

* fix defaults for TestColumnNumberOfDifferentNulls

* add unittests for TestColumnNumberOfDifferentNulls + style fixes with black

src/evidently/tests/data_integrity_tests.py

@@ -548,14 +548,19 @@ def get_condition(self) -> TestValueCondition:
         reference_null_values = self.metric.get_result().reference_null_values
 
         if reference_null_values is not None:
-            ref_value = len(reference_null_values.different_nulls_by_column[self.column_name])
+            if self.column_name not in reference_null_values.number_of_different_nulls_by_column:
+                raise ValueError(
+                    f"Cannot define test default conditions: no column '{self.column_name}' in reference dataset."
+                )
+
+            ref_value = reference_null_values.number_of_different_nulls_by_column[self.column_name]
             return TestValueCondition(lte=ref_value)
 
         raise ValueError("Neither required test parameters nor reference data has been provided.")
 
     def calculate_value_for_test(self) -> Numeric:
         metric_data = self.metric.get_result().current_null_values
-        return len(metric_data.different_nulls_by_column[self.column_name])
+        return metric_data.number_of_different_nulls_by_column[self.column_name]
 
     def get_description(self, value: Numeric) -> str:
         return (

tests/tests/test_data_integrity_tests.py

@@ -549,6 +549,34 @@ def test_data_integrity_test_different_nulls_one_column() -> None:
     assert suite
 
 
+def test_data_integrity_test_different_nulls_one_column_no_nulls() -> None:
+    test_dataset = pd.DataFrame({"feature1": [1, 2, 3], "feature2": ["b", "", None]})
+
+    suite = TestSuite(
+        tests=[
+            TestColumnNumberOfDifferentNulls(column_name="feature1", null_values=["null", "n/a"], replace=False, eq=0)
+        ]
+    )
+    suite.run(current_data=test_dataset, reference_data=test_dataset, column_mapping=ColumnMapping())
+    assert suite
+
+
+def test_data_integrity_test_different_nulls_one_column_with_defaults() -> None:
+    test_dataset = pd.DataFrame({"feature": ["null", "", None]})
+    reference_dataset = pd.DataFrame({"feature": ["n/a", "test", None]})
+    data_mapping = ColumnMapping()
+
+    suite = TestSuite(
+        tests=[TestColumnNumberOfDifferentNulls(column_name="feature", null_values=["null", "n/a"], replace=False)]
+    )
+    suite.run(current_data=test_dataset, reference_data=reference_dataset, column_mapping=data_mapping)
+    assert not suite, suite.json()
+
+    another_test_dataset = pd.DataFrame({"feature": ["null", "test", None]})
+    suite.run(current_data=another_test_dataset, reference_data=reference_dataset, column_mapping=data_mapping)
+    assert suite, suite.json()
+
+
 def test_data_integrity_test_number_of_nulls_one_column() -> None:
     test_dataset = pd.DataFrame({"feature1": ["", None, "null", "a"], "feature2": ["b", "null", None, None]})