[BUGFIX] fix incorrect pandas top rows usage (#3091)

* fix a bug in the usage of Pandas for subscripting to get select number of rows
* Added a test of pandas slicing in MapMetrics functions.

great_expectations/expectations/metrics/map_metric.py

@@ -518,7 +518,7 @@ def _pandas_column_map_condition_values(
 ):
     """Return values from the specified domain that match the map-style metric in the metrics dictionary."""
     (
-        boolean_map_unexpected_values,
+        boolean_mapped_unexpected_values,
         compute_domain_kwargs,
         accessor_domain_kwargs,
     ) = metrics["unexpected_condition"]
@@ -553,15 +553,14 @@ def _pandas_column_map_condition_values(
 
     domain_values = df[column_name]
 
+    domain_values = domain_values[boolean_mapped_unexpected_values == True]
+
     result_format = metric_value_kwargs["result_format"]
+
     if result_format["result_format"] == "COMPLETE":
-        return list(domain_values[boolean_map_unexpected_values == True])
+        return list(domain_values)
     else:
-        return list(
-            domain_values[boolean_map_unexpected_values == True][
-                : result_format["partial_unexpected_count"]
-            ]
-        )
+        return list(domain_values[: result_format["partial_unexpected_count"]])
 
 
 def _pandas_column_map_series_and_domain_values(
@@ -574,7 +573,7 @@ def _pandas_column_map_series_and_domain_values(
 ):
     """Return values from the specified domain that match the map-style metric in the metrics dictionary."""
     (
-        boolean_map_unexpected_values,
+        boolean_mapped_unexpected_values,
         compute_domain_kwargs,
         accessor_domain_kwargs,
     ) = metrics["unexpected_condition"]
@@ -620,24 +619,20 @@ def _pandas_column_map_series_and_domain_values(
 
     domain_values = df[column_name]
 
+    domain_values = domain_values[boolean_mapped_unexpected_values == True]
+    map_series = map_series[boolean_mapped_unexpected_values == True]
+
     result_format = metric_value_kwargs["result_format"]
+
     if result_format["result_format"] == "COMPLETE":
         return (
-            list(domain_values[boolean_map_unexpected_values == True]),
-            list(map_series[boolean_map_unexpected_values == True]),
+            list(domain_values),
+            list(map_series),
         )
     else:
         return (
-            list(
-                domain_values[boolean_map_unexpected_values == True][
-                    : result_format["partial_unexpected_count"]
-                ]
-            ),
-            list(
-                map_series[boolean_map_unexpected_values == True][
-                    : result_format["partial_unexpected_count"]
-                ]
-            ),
+            list(domain_values[: result_format["partial_unexpected_count"]]),
+            list(map_series[: result_format["partial_unexpected_count"]]),
         )
 
 
@@ -682,14 +677,12 @@ def _pandas_map_condition_index(
 
     result_format = metric_value_kwargs["result_format"]
 
+    df = df[boolean_mapped_unexpected_values]
+
     if result_format["result_format"] == "COMPLETE":
-        return list(df[boolean_mapped_unexpected_values].index)
+        return list(df.index)
 
-    return list(
-        df[boolean_mapped_unexpected_values].index[
-            : result_format["partial_unexpected_count"]
-        ]
-    )
+    return list(df.index[: result_format["partial_unexpected_count"]])
 
 
 def _pandas_column_map_condition_value_counts(
@@ -803,12 +796,12 @@ def _pandas_map_condition_rows(
 
     result_format = metric_value_kwargs["result_format"]
 
+    df = df[boolean_mapped_unexpected_values]
+
     if result_format["result_format"] == "COMPLETE":
-        return df[boolean_mapped_unexpected_values]
+        return df
 
-    return df[boolean_mapped_unexpected_values][
-        result_format["partial_unexpected_count"]
-    ]
+    return df.iloc[: result_format["partial_unexpected_count"]]
 
 
 def _sqlalchemy_map_condition_unexpected_count_aggregate_fn(
@@ -1298,7 +1291,6 @@ def _register_metric_functions(cls):
                             metric_provider=_pandas_column_map_condition_value_counts,
                             metric_fn_type=MetricFunctionTypes.VALUE,
                         )
-
                 elif issubclass(engine, SqlAlchemyExecutionEngine):
                     register_metric(
                         metric_name=metric_name + ".condition",
@@ -1437,7 +1429,6 @@ def _register_metric_functions(cls):
                             metric_provider=_spark_column_map_condition_value_counts,
                             metric_fn_type=MetricFunctionTypes.VALUE,
                         )
-
             elif metric_fn_type in [
                 MetricPartialFunctionTypes.MAP_SERIES,
                 MetricPartialFunctionTypes.MAP_FN,

tests/expectations/metrics/test_core.py

@@ -563,6 +563,25 @@ def test_map_unique_pd_column_exists():
     assert list(metrics[condition_metric.id][0]) == [False, False, True, True, False]
     assert metrics[unexpected_count_metric.id] == 2
 
+    unexpected_rows_metric = MetricConfiguration(
+        metric_name="column_values.unique.unexpected_rows",
+        metric_domain_kwargs={"column": "a"},
+        metric_value_kwargs={
+            "result_format": {"result_format": "SUMMARY", "partial_unexpected_count": 1}
+        },
+        metric_dependencies={
+            "unexpected_condition": condition_metric,
+            "table.columns": table_columns_metric,
+        },
+    )
+    results = engine.resolve_metrics(
+        metrics_to_resolve=(unexpected_rows_metric,), metrics=metrics
+    )
+    metrics.update(results)
+
+    assert metrics[unexpected_rows_metric.id]["a"].index == [2]
+    assert metrics[unexpected_rows_metric.id]["a"].values == [3]
+
 
 def test_map_unique_pd_column_does_not_exist():
     engine = build_pandas_engine(pd.DataFrame({"a": [1, 2, 3, 3, None]}))