BUG: Fix wrong reading sparse matrix (#31991)

Author: m7142yosuke

doc/source/whatsnew/v1.1.0.rst

@@ -623,7 +623,7 @@ Reshaping
 Sparse
 ^^^^^^
 - Creating a :class:`SparseArray` from timezone-aware dtype will issue a warning before dropping timezone information, instead of doing so silently (:issue:`32501`)
--
+- Bug in :meth:`arrays.SparseArray.from_spmatrix` wrongly read scipy sparse matrix (:issue:`31991`)
 -
 
 ExtensionArray

pandas/core/arrays/sparse/array.py

@@ -439,11 +439,10 @@ def from_spmatrix(cls, data):
 
         # our sparse index classes require that the positions be strictly
         # increasing. So we need to sort loc, and arr accordingly.
+        data = data.tocsc()
+        data.sort_indices()
         arr = data.data
-        idx, _ = data.nonzero()
-        loc = np.argsort(idx)
-        arr = arr.take(loc)
-        idx.sort()
+        idx = data.indices
 
         zero = np.array(0, dtype=arr.dtype).item()
         dtype = SparseDtype(arr.dtype, zero)

pandas/tests/arrays/sparse/test_accessor.py

@@ -41,6 +41,18 @@ def test_from_spmatrix(self, format, labels, dtype):
         ).astype(sp_dtype)
         tm.assert_frame_equal(result, expected)
 
+    @pytest.mark.parametrize("format", ["csc", "csr", "coo"])
+    @td.skip_if_no_scipy
+    def test_from_spmatrix_including_explicit_zero(self, format):
+        import scipy.sparse
+
+        mat = scipy.sparse.random(10, 2, density=0.5, format=format)
+        mat.data[0] = 0
+        result = pd.DataFrame.sparse.from_spmatrix(mat)
+        dtype = SparseDtype("float64", 0.0)
+        expected = pd.DataFrame(mat.todense()).astype(dtype)
+        tm.assert_frame_equal(result, expected)
+
     @pytest.mark.parametrize(
         "columns",
         [["a", "b"], pd.MultiIndex.from_product([["A"], ["a", "b"]]), ["a", "a"]],

pandas/tests/arrays/sparse/test_array.py

@@ -208,6 +208,19 @@ def test_from_spmatrix(self, size, format):
         expected = mat.toarray().ravel()
         tm.assert_numpy_array_equal(result, expected)
 
+    @pytest.mark.parametrize("format", ["coo", "csc", "csr"])
+    @td.skip_if_no_scipy
+    def test_from_spmatrix_including_explicit_zero(self, format):
+        import scipy.sparse
+
+        mat = scipy.sparse.random(10, 1, density=0.5, format=format)
+        mat.data[0] = 0
+        result = SparseArray.from_spmatrix(mat)
+
+        result = np.asarray(result)
+        expected = mat.toarray().ravel()
+        tm.assert_numpy_array_equal(result, expected)
+
     @td.skip_if_no_scipy
     def test_from_spmatrix_raises(self):
         import scipy.sparse