def test_wrong_shape(tmp_path, sparse_format, a_shape, b_shape):
h5_path = tmp_path / "base.h5"
a_mem = sparse.random(*a_shape, format=sparse_format)
b_mem = sparse.random(*b_shape, format=sparse_format)
with h5py.File(h5_path, "a") as f:
ad._io.h5ad.write_attribute(f, "a", a_mem)
ad._io.h5ad.write_attribute(f, "b", b_mem)
a_disk = SparseDataset(f["a"])
b_disk = SparseDataset(f["b"])
with pytest.raises(AssertionError):
a_disk.append(b_disk)
def test_dataset_append_memory(tmp_path, sparse_format, append_method):
h5_path = tmp_path / "test.h5"
a = sparse_format(sparse.random(100, 100))
b = sparse_format(sparse.random(100, 100))
with h5py.File(h5_path, "a") as f:
ad._io.h5ad.write_attribute(f, "mtx", a)
diskmtx = SparseDataset(f["mtx"])
diskmtx.append(b)
fromdisk = diskmtx.to_memory()
frommem = append_method([a, b])
assert_equal(fromdisk, frommem)
def read_basic_zarr(elem):
from anndata._io import zarr
warn(
f"Element '{elem.name}' was written without encoding metadata.",
OldFormatWarning,
stacklevel=3,
)
if isinstance(elem, Mapping):
# Backwards compat sparse arrays
if "h5sparse_format" in elem.attrs:
return SparseDataset(elem).to_memory()
return {k: read_elem(v) for k, v in elem.items()}
elif isinstance(elem, ZarrArray):
return zarr.read_dataset(elem) # TODO: Handle legacy
def test_dataset_append_disk(tmp_path, sparse_format, append_method):
h5_path = tmp_path / "test.h5"
a = sparse_format(sparse.random(10, 10))
b = sparse_format(sparse.random(10, 10))
with h5py.File(h5_path, "a") as f:
ad._io.h5ad.write_attribute(f, "a", a)
ad._io.h5ad.write_attribute(f, "b", b)
a_disk = SparseDataset(f["a"])
b_disk = SparseDataset(f["b"])
a_disk.append(b_disk)
fromdisk = a_disk.to_memory()
frommem = append_method([a, b])
assert_equal(fromdisk, frommem)
def test_wrong_formats(tmp_path):
h5_path = tmp_path / "base.h5"
base = sparse.random(100, 100, format="csr")
with h5py.File(h5_path, "a") as f:
ad._io.h5ad.write_attribute(f, "base", base)
disk_mtx = SparseDataset(f["base"])
pre_checks = disk_mtx.to_memory()
with pytest.raises(ValueError):
disk_mtx.append(sparse.random(100, 100, format="csc"))
with pytest.raises(ValueError):
disk_mtx.append(sparse.random(100, 100, format="coo"))
with pytest.raises(NotImplementedError):
disk_mtx.append(np.random.random((100, 100)))
disk_dense = f.create_dataset("dense",
data=np.random.random((100, 100)))
with pytest.raises(NotImplementedError):
disk_mtx.append(disk_dense)
post_checks = disk_mtx.to_memory()
# Check nothing changed
assert not np.any((pre_checks != post_checks).toarray())
def read_sparse_partial(elem,
*,
items=None,
indices=(slice(None), slice(None))):
return SparseDataset(elem)[indices]
def read_sparse(elem):
return SparseDataset(elem).to_memory()
