Fix issues with MultiIndex in dropna, stack & reset_index (#8753

)

Fixes: #3583 

This PR contains fixes for :

- [x] `stack`: Where the MultiIndex names are not being assigned correctly in `from_table` call.
- [x] `dropna`: Where the MultiIndex names are not being preserved after a `libcudf` API call.
- [x] `reset_index`: Where the MultiIndex level names are not being materialized correctly when the index is reset.

Authors:
  - GALI PREM SAGAR (https://github.com/galipremsagar)

Approvers:
  - Michael Wang (https://github.com/isVoid)
  - Ashwin Srinath (https://github.com/shwina)

URL: #8753

python/cudf/cudf/core/dataframe.py

@@ -2987,17 +2987,22 @@ class max_speed
             result = self
         else:
             result = self.copy()
-        if all(name is None for name in self.index.names):
+
+        if not drop:
             if isinstance(self.index, cudf.MultiIndex):
                 names = tuple(
-                    f"level_{i}" for i, _ in enumerate(self.index.names)
+                    name if name is not None else f"level_{i}"
+                    for i, name in enumerate(self.index.names)
                 )
             else:
-                names = ("index",)
-        else:
-            names = self.index.names
+                if self.index.name is None:
+                    if "index" in self._data.names:
+                        names = ("level_0",)
+                    else:
+                        names = ("index",)
+                else:
+                    names = (self.index.name,)
 
-        if not drop:
             index_columns = self.index._data.columns
             for name, index_column in zip(
                 reversed(names), reversed(index_columns)
@@ -7493,8 +7498,13 @@ def stack(self, level=-1, dropna=True):
         repeated_index = self.index.repeat(self.shape[1])
         name_index = Frame({0: self._column_names}).tile(self.shape[0])
         new_index = list(repeated_index._columns) + [name_index._columns[0]]
+        if isinstance(self._index, cudf.MultiIndex):
+            index_names = self._index.names + [None]
+        else:
+            index_names = [None] * len(new_index)
         new_index = cudf.core.multiindex.MultiIndex.from_frame(
-            DataFrame(dict(zip(range(0, len(new_index)), new_index)))
+            DataFrame(dict(zip(range(0, len(new_index)), new_index))),
+            names=index_names,
         )
 
         # Collect datatypes and cast columns as that type

python/cudf/cudf/core/frame.py

@@ -1380,6 +1380,10 @@ def _drop_na_rows(
             )
         )
         result._copy_type_metadata(frame)
+        if self._index is not None:
+            result._index.name = self._index.name
+            if isinstance(self._index, cudf.MultiIndex):
+                result._index.names = self._index.names
         return result
 
     def _drop_na_columns(self, how="any", subset=None, thresh=None):

python/cudf/cudf/tests/test_reshape.py

@@ -103,16 +103,33 @@ def test_df_stack(nulls, num_cols, num_rows, dtype):
     gdf = cudf.from_pandas(pdf)
 
     got = gdf.stack()
-
     expect = pdf.stack()
-    if {None} == set(expect.index.names):
-        expect.rename_axis(
-            list(range(0, len(expect.index.names))), inplace=True
-        )
 
     assert_eq(expect, got)
 
 
+def test_df_stack_reset_index():
+    df = cudf.DataFrame(
+        {
+            "a": [1, 2, 3, 4],
+            "b": [10, 11, 12, 13],
+            "c": ["ab", "cd", None, "gh"],
+        }
+    )
+    df = df.set_index(["a", "b"])
+    pdf = df.to_pandas()
+
+    expected = pdf.stack()
+    actual = df.stack()
+
+    assert_eq(expected, actual)
+
+    expected = expected.reset_index()
+    actual = actual.reset_index()
+
+    assert_eq(expected, actual)
+
+
 @pytest.mark.parametrize("num_rows", [1, 2, 10, 1000])
 @pytest.mark.parametrize("num_cols", [1, 2, 10])
 @pytest.mark.parametrize(