simplify

narwhals/_pandas_like/group_by.py

@@ -11,7 +11,6 @@
 from narwhals._expression_parsing import is_simple_aggregation
 from narwhals._expression_parsing import parse_into_exprs
 from narwhals._pandas_like.utils import native_series_from_iterable
-from narwhals._pandas_like.utils import reset_index_no_copy
 from narwhals.utils import Implementation
 from narwhals.utils import remove_prefix
 
@@ -189,13 +188,17 @@ def agg_pandas(  # noqa: PLR0915
             result_simple_aggs = result_simple_aggs.rename(
                 columns=name_mapping, copy=False
             )
-            reset_index_no_copy(result_simple_aggs, native_namespace)
+            # Keep inplace=True to avoid making a redundant copy.
+            # This one will likely stay in pandas https://github.com/pandas-dev/pandas/pull/51466/files
+            result_simple_aggs.reset_index(inplace=True)  # noqa: PD002
         if nunique_aggs:
             result_nunique_aggs = grouped[list(nunique_aggs.values())].nunique(
                 dropna=False
             )
             result_nunique_aggs.columns = list(nunique_aggs.keys())
-            reset_index_no_copy(result_nunique_aggs, native_namespace)
+            # Keep inplace=True to avoid making a redundant copy.
+            # This one will likely stay in pandas https://github.com/pandas-dev/pandas/pull/51466/files
+            result_nunique_aggs.reset_index(inplace=True)  # noqa: PD002
         if simple_aggs and nunique_aggs:
             if (
                 set(result_simple_aggs.columns)
@@ -261,6 +264,8 @@ def func(df: Any) -> Any:
     else:  # pragma: no cover
         result_complex = grouped.apply(func)
 
-    reset_index_no_copy(result_complex, native_namespace)
+    # Keep inplace=True to avoid making a redundant copy.
+    # This one will likely stay in pandas https://github.com/pandas-dev/pandas/pull/51466/files
+    result_complex.reset_index(inplace=True)  # noqa: PD002
 
     return from_dataframe(result_complex.loc[:, output_names])

narwhals/_pandas_like/series.py

@@ -598,7 +598,10 @@ def value_counts(
             dropna=False,
             sort=False,
             normalize=normalize,
-        ).reset_index()
+        )
+        # Keep inplace=True to avoid making a redundant copy.
+        # This one will likely stay in pandas https://github.com/pandas-dev/pandas/pull/51466/files
+        val_count.reset_index(inplace=True)  # noqa: PD002
 
         val_count.columns = [index_name_, value_name_]
 

narwhals/_pandas_like/utils.py

@@ -590,22 +590,3 @@ def calculate_timestamp_date(s: pd.Series, time_unit: str) -> pd.Series:
     else:
         result = s * 1_000
     return result
-
-
-def reset_index_no_copy(native_dataframe: pd.DataFrame, native_namespace: Any) -> None:
-    """
-    Reset index without triggering a copy.
-
-    pandas' reset_index isn't free, and creates a copy. To avoid that, and keep
-    overhead as low as possible, we introduce a utility to use instead of that.
-    This should be used internally in Narwhals whenever you need to reset the index
-    of an object which is not the original object passed by the user. We should
-    never mutate the user's object, but if we create intermediate objects ourselves,
-    then it's fine to mutate them.
-    """
-    if set(native_dataframe.index.names).intersection(native_dataframe.columns):
-        msg = f"Cannot insert column with name {native_dataframe.index.name} into dataframe with columns {native_dataframe.columns}"
-        raise ValueError(msg)
-    for i, name in enumerate(native_dataframe.index.names):
-        native_dataframe.insert(0, name, native_dataframe.index.get_level_values(i))
-    native_dataframe.index = native_namespace.RangeIndex(len(native_dataframe))

tests/utils_test.py

@@ -8,7 +8,6 @@
 from pandas.testing import assert_series_equal
 
 import narwhals.stable.v1 as nw
-from narwhals._pandas_like.utils import reset_index_no_copy
 from tests.utils import PANDAS_VERSION
 from tests.utils import get_module_version_as_tuple
 
@@ -148,14 +147,3 @@ def test_maybe_convert_dtypes_polars() -> None:
 def test_get_trivial_version_with_uninstalled_module() -> None:
     result = get_module_version_as_tuple("non_existent_module")
     assert result == (0, 0, 0)
-
-
-def test_reset_index_no_copy() -> None:
-    df = pd.DataFrame({"a": [1, 2, 3], "b": [4.0, 5.0, 6.0]}).set_index("a")
-    reset_index_no_copy(df, pd)
-    pd.testing.assert_frame_equal(
-        df, pd.DataFrame({"a": [1, 2, 3], "b": [4.0, 5.0, 6.0]})
-    )
-    df.index.name = "b"
-    with pytest.raises(ValueError, match="Cannot insert column"):
-        reset_index_no_copy(df, pd)