Backport PR #1806: Fix backed sparse matrix compat with scipy 1.15 (#…

…1809)

Co-authored-by: Philipp A <[email protected]>

docs/release-notes/1806.bugfix.md

@@ -0,0 +1 @@
+Add {mod}`scipy` 1.5 compatibility {user}`flying-sheep`

src/anndata/_core/sparse_dataset.py

@@ -23,7 +23,9 @@
 
 import h5py
 import numpy as np
+import scipy
 import scipy.sparse as ss
+from packaging.version import Version
 from scipy.sparse import _sparsetools
 
 from .. import abc
@@ -39,11 +41,14 @@
 
     from .._types import GroupStorageType
     from ..compat import H5Array
-    from .index import Index
+    from .index import Index, Index1D
 else:
     from scipy.sparse import spmatrix as _cs_matrix
 
 
+SCIPY_1_15 = Version(scipy.__version__) >= Version("1.15rc0")
+
+
 class BackedFormat(NamedTuple):
     format: Literal["csr", "csc"]
     backed_type: type[BackedSparseMatrix]
@@ -353,7 +358,9 @@ def _get_group_format(group: GroupStorageType) -> str:
 
 
 # Check for the overridden few methods above in our BackedSparseMatrix subclasses
-def is_sparse_indexing_overridden(format: Literal["csr", "csc"], row, col):
+def is_sparse_indexing_overridden(
+    format: Literal["csr", "csc"], row: Index1D, col: Index1D
+):
     major_indexer, minor_indexer = (row, col) if format == "csr" else (col, row)
     return isinstance(minor_indexer, slice) and (
         (isinstance(major_indexer, int | np.integer))
@@ -362,6 +369,13 @@ def is_sparse_indexing_overridden(format: Literal["csr", "csc"], row, col):
     )
 
 
+def validate_indices(
+    mtx: BackedSparseMatrix, indices: tuple[Index1D, Index1D]
+) -> tuple[Index1D, Index1D]:
+    res = mtx._validate_indices(indices)
+    return res[0] if SCIPY_1_15 else res
+
+
 class BaseCompressedSparseDataset(abc._AbstractCSDataset, ABC):
     _group: GroupStorageType
 
@@ -424,8 +438,8 @@ def __getitem__(
         indices = self._normalize_index(index)
         row, col = indices
         mtx = self._to_backed()
-        row_sp_matrix_validated, col_sp_matrix_validated = mtx._validate_indices(
-            (row, col)
+        row_sp_matrix_validated, col_sp_matrix_validated = validate_indices(
+            mtx, indices
         )
 
         # Handle masked indexing along major axis

src/anndata/compat/__init__.py

@@ -47,7 +47,7 @@ class Empty:
     pass
 
 
-Index1D = slice | int | str | np.int64 | np.ndarray
+Index1D = slice | int | str | np.int64 | np.ndarray | pd.Series
 IndexRest = Index1D | EllipsisType
 Index = (
     IndexRest

tests/test_backed_sparse.py

@@ -615,17 +615,28 @@ def test_backed_sizeof(
     assert csr_mem.__sizeof__() > csc_disk.__sizeof__()
 
 
+sparray_scipy_bug_marks = (
+    [pytest.mark.skip(reason="scipy bug causes view to be allocated")]
+    if CAN_USE_SPARSE_ARRAY
+    else []
+)
+
+
 @pytest.mark.parametrize(
     "group_fn",
     [
         pytest.param(lambda _: zarr.group(), id="zarr"),
         pytest.param(lambda p: h5py.File(p / "test.h5", mode="a"), id="h5py"),
     ],
 )
-@pytest.mark.parametrize("sparse_class", [sparse.csr_matrix, sparse.csr_array])
+@pytest.mark.parametrize(
+    "sparse_class",
+    [
+        sparse.csr_matrix,
+        pytest.param(sparse.csr_array, marks=[*sparray_scipy_bug_marks]),
+    ],
+)
 def test_append_overflow_check(group_fn, sparse_class, tmpdir):
-    if CAN_USE_SPARSE_ARRAY and issubclass(sparse_class, SpArray):
-        pytest.skip("scipy bug causes view to be allocated")
     group = group_fn(tmpdir)
     typemax_int32 = np.iinfo(np.int32).max
     orig_mtx = sparse_class(np.ones((1, 1), dtype=bool))

tests/test_concatenate.py

@@ -1444,7 +1444,7 @@ def test_concat_outer_aligned_mapping(elem):
     del b.obsm[elem]
 
     concated = concat({"a": a, "b": b}, join="outer", label="group")
-    result = concated.obsm[elem][concated.obs["group"] == "b"]
+    result = concated[concated.obs["group"] == "b"].obsm[elem]
 
     check_filled_like(result, elem_name=f"obsm/{elem}")