Move validation check to a separate function for reuse

Author: lajohn4747

sdmetrics/single_table/privacy/dcr_overfitting_protection.py

@@ -8,6 +8,7 @@
 from sdmetrics.goal import Goal
 from sdmetrics.single_table.base import SingleTableMetric
 from sdmetrics.single_table.privacy.dcr_utils import calculate_dcr
+from sdmetrics.single_table.privacy.util import validate_num_samples_num_iteration
 
 
 class DCROverfittingProtection(SingleTableMetric):
@@ -32,20 +33,7 @@ def _validate_inputs(
         num_rows_subsample,
         num_iterations,
     ):
-        if num_rows_subsample is not None:
-            if not isinstance(num_rows_subsample, int) or num_rows_subsample < 1:
-                raise ValueError(
-                    f'num_rows_subsample ({num_rows_subsample}) must be an integer greater than 1.'
-                )
-        elif num_rows_subsample is None and num_iterations > 1:
-            raise ValueError(
-                'num_iterations should not be greater than 1 if there is no subsampling.'
-            )
-
-        if not isinstance(num_iterations, int) or num_iterations < 1:
-            raise ValueError(
-                f'num_iterations ({num_iterations}) must be an integer greater than 1.'
-            )
+        validate_num_samples_num_iteration(num_rows_subsample, num_iterations)
 
         if len(real_training_data) * 0.5 > len(real_validation_data):
             warnings.warn(

sdmetrics/single_table/privacy/util.py

@@ -148,3 +148,16 @@ def allow_nan_array(attributes):
             ret.append(entry)
 
     return ret
+
+
+def validate_num_samples_num_iteration(num_rows_subsample, num_iterations):
+    if num_rows_subsample is not None:
+        if not isinstance(num_rows_subsample, int) or num_rows_subsample < 1:
+            raise ValueError(
+                f'num_rows_subsample ({num_rows_subsample}) must be an integer greater than 1.'
+            )
+    elif num_rows_subsample is None and num_iterations > 1:
+        raise ValueError('num_iterations should not be greater than 1 if there is no subsampling.')
+
+    if not isinstance(num_iterations, int) or num_iterations < 1:
+        raise ValueError(f'num_iterations ({num_iterations}) must be an integer greater than 1.')

tests/unit/single_table/privacy/test_util.py

@@ -1,4 +1,11 @@
-from sdmetrics.single_table.privacy.util import closest_neighbors
+import re
+
+import pytest
+
+from sdmetrics.single_table.privacy.util import (
+    closest_neighbors,
+    validate_num_samples_num_iteration,
+)
 
 
 def test_closest_neighbors_exact():
@@ -30,3 +37,20 @@ def test_closest_neighbors_non_exact():
     assert ('a', '1') in results
     assert ('a', '3') in results
     assert ('b', '2') in results
+
+
+def test_validate_num_samples_num_iteration():
+    # Run and Assert
+    zero_subsample_msg = re.escape('num_rows_subsample (0) must be an integer greater than 1.')
+    with pytest.raises(ValueError, match=zero_subsample_msg):
+        validate_num_samples_num_iteration(0, 1)
+
+    subsample_none_msg = re.escape(
+        'num_iterations should not be greater than 1 if there is no subsampling.'
+    )
+    with pytest.raises(ValueError, match=subsample_none_msg):
+        validate_num_samples_num_iteration(None, 2)
+
+    zero_iteration_msg = re.escape('num_iterations (0) must be an integer greater than 1.')
+    with pytest.raises(ValueError, match=zero_iteration_msg):
+        validate_num_samples_num_iteration(1, 0)