BUG: to_coo raising Systemerror for EA (#51007)

phofl · web-flow · commit e4df678be828 · 2023-01-30T11:39:48.000-08:00
diff --git a/doc/source/whatsnew/v2.0.0.rst b/doc/source/whatsnew/v2.0.0.rst
@@ -1160,7 +1160,7 @@ Sparse
 ^^^^^^
 - Bug in :meth:`Series.astype` when converting a ``SparseDtype`` with ``datetime64[ns]`` subtype to ``int64`` dtype raising, inconsistent with the non-sparse behavior (:issue:`49631`,:issue:`50087`)
 - Bug in :meth:`Series.astype` when converting a from ``datetime64[ns]`` to ``Sparse[datetime64[ns]]`` incorrectly raising (:issue:`50082`)
--
+- Bug in :meth:`Series.sparse.to_coo` raising ``SystemError`` when :class:`MultiIndex` contains a ``ExtensionArray`` (:issue:`50996`)
 
 ExtensionArray
 ^^^^^^^^^^^^^^
diff --git a/pandas/core/arrays/sparse/scipy_sparse.py b/pandas/core/arrays/sparse/scipy_sparse.py
@@ -73,7 +73,7 @@ def _levels_to_axis(
 
     else:
         levels_values = lib.fast_zip(
-            [ss.index.get_level_values(lvl).values for lvl in levels]
+            [ss.index.get_level_values(lvl).to_numpy() for lvl in levels]
         )
         codes, ax_labels = factorize(levels_values, sort=sort_labels)
         ax_coords = codes[valid_ilocs]
diff --git a/pandas/tests/arrays/sparse/test_accessor.py b/pandas/tests/arrays/sparse/test_accessor.py
@@ -183,6 +183,25 @@ def test_to_coo_nonzero_fill_val_raises(self, fill_value):
         with pytest.raises(ValueError, match="fill value must be 0"):
             df.sparse.to_coo()
 
+    @td.skip_if_no_scipy
+    def test_to_coo_midx_categorical(self):
+        # GH#50996
+        import scipy.sparse
+
+        midx = pd.MultiIndex.from_arrays(
+            [
+                pd.CategoricalIndex(list("ab"), name="x"),
+                pd.CategoricalIndex([0, 1], name="y"),
+            ]
+        )
+
+        ser = pd.Series(1, index=midx, dtype="Sparse[int]")
+        result = ser.sparse.to_coo(row_levels=["x"], column_levels=["y"])[0]
+        expected = scipy.sparse.coo_matrix(
+            (np.array([1, 1]), (np.array([0, 1]), np.array([0, 1]))), shape=(2, 2)
+        )
+        assert (result != expected).nnz == 0
+
     def test_to_dense(self):
         df = pd.DataFrame(
             {

Original file line number	Diff line number	Diff line change
`@@ -73,7 +73,7 @@ def _levels_to_axis(`
`73`	`73`
`74`	`74`	`else:`
`75`	`75`	`levels_values = lib.fast_zip(`
`76`		`- [ss.index.get_level_values(lvl).values for lvl in levels]`
	`76`	`+ [ss.index.get_level_values(lvl).to_numpy() for lvl in levels]`
`77`	`77`	`)`
`78`	`78`	`codes, ax_labels = factorize(levels_values, sort=sort_labels)`
`79`	`79`	`ax_coords = codes[valid_ilocs]`