Update tools used in pre-commit testing. (#1741)

This updates us to the latest versions of Black and Ruff, which results
in some minor non-behavioral changes to the code.

.pre-commit-config.yaml

@@ -1,16 +1,16 @@
 exclude: ^doc/source/
 repos:
-  - repo: https://github.com/ambv/black
-    rev: 22.12.0
+  - repo: https://github.com/psf/black
+    rev: "23.9.1"
     hooks:
     - id: black
-  - repo: https://github.com/charliermarsh/ruff-pre-commit
-    rev: v0.0.226
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.0.291
     hooks:
     - id: ruff
-      args: ["--force-exclude", "--config=apis/python/pyproject.toml"]
+      args: ["--config=apis/python/pyproject.toml"]
   - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.0.0
+    rev: v1.5.1
     hooks:
     - id: mypy
       additional_dependencies:

apis/python/src/tiledbsoma/_collection.py

@@ -64,7 +64,7 @@ class _CachedElement:
     """The reified object, if it has been opened."""
 
 
-class CollectionBase(
+class CollectionBase(  # type: ignore[misc]  # __eq__ false positive
     TileDBObject[_tdb_handles.GroupWrapper],
     somacore.collection.BaseCollection[CollectionElementType],
 ):
@@ -628,7 +628,7 @@ def _check_allows_child(cls, key: str, child_cls: type) -> None:
 AnyTileDBCollection = CollectionBase[Any]
 
 
-class Collection(
+class Collection(  # type: ignore[misc]  # __eq__ false positive
     CollectionBase[CollectionElementType], somacore.Collection[CollectionElementType]
 ):
     """:class:`Collection` is a persistent container of named SOMA objects, stored as

apis/python/src/tiledbsoma/_experiment.py

@@ -17,7 +17,7 @@
 from ._tiledb_object import AnyTileDBObject
 
 
-class Experiment(
+class Experiment(  # type: ignore[misc]  # __eq__ false positive
     CollectionBase[AnyTileDBObject],
     experiment.Experiment[  # type: ignore[type-var]
         DataFrame,

apis/python/src/tiledbsoma/_measurement.py

@@ -17,7 +17,7 @@
 from ._tiledb_object import AnyTileDBObject
 
 
-class Measurement(
+class Measurement(  # type: ignore[misc]  # __eq__ false positive
     CollectionBase[AnyTileDBObject],
     measurement.Measurement[  # type: ignore[type-var]
         DataFrame,

apis/python/src/tiledbsoma/io/_registration/signatures.py

@@ -214,7 +214,6 @@ def from_soma_experiment(
         """
 
         with tiledbsoma.Experiment.open(uri, context=context) as exp:
-
             obs_schema = _string_dict_from_arrow_schema(exp.obs.schema)
 
             var_schema = _string_dict_from_arrow_schema(

apis/python/src/tiledbsoma/io/ingest.py

@@ -79,6 +79,7 @@
 _NDArr = TypeVar("_NDArr", bound=NDArray)
 _TDBO = TypeVar("_TDBO", bound=TileDBObject[RawHandle])
 
+
 # ----------------------------------------------------------------
 class IngestionParams:
     """Maps from user-level ingest modes to a set of implementation-level boolean flags."""
@@ -93,7 +94,6 @@ def __init__(
         ingest_mode: str,
         label_mapping: Optional[ExperimentAmbientLabelMapping],
     ) -> None:
-
         if ingest_mode == "schema_only":
             self.write_schema_no_data = True
             self.error_if_already_exists = True
@@ -1125,7 +1125,6 @@ def _write_dataframe(
     context: Optional[SOMATileDBContext] = None,
     axis_mapping: AxisIDMapping,
 ) -> DataFrame:
-
     _util.get_start_stamp()
     logging.log_io(None, f"START  WRITING {df_uri}")
 
@@ -1468,7 +1467,6 @@ def _update_dataframe(
     with DataFrame.open(
         sdf.uri, mode="r", context=context, platform_config=platform_config
     ) as sdf_r:
-
         # Until we someday support deletes, this is the correct check on the existing,
         # contiguous soma join IDs compared to the new contiguous ones about to be created.
         old_jids = sorted(
@@ -1831,7 +1829,6 @@ def _coo_to_table(
         axis: int = 0,
         base: int = 0,
     ) -> pa.Table:
-
         soma_dim_0 = mat_coo.row + base if base > 0 and axis == 0 else mat_coo.row
         soma_dim_1 = mat_coo.col + base if base > 0 and axis == 1 else mat_coo.col
 
@@ -2058,7 +2055,6 @@ def _ingest_uns_dict(
     ingestion_params: IngestionParams,
     use_relative_uri: Optional[bool],
 ) -> None:
-
     with _create_or_open_collection(
         Collection,
         _util.uri_joinpath(parent.uri, parent_key),
@@ -2092,7 +2088,6 @@ def _ingest_uns_node(
     ingestion_params: IngestionParams,
     use_relative_uri: Optional[bool],
 ) -> None:
-
     if isinstance(value, np.generic):
         # This is some kind of numpy scalar value. Metadata entries
         # only accept native Python types, so unwrap it.

apis/python/tests/test_collection.py

@@ -17,7 +17,6 @@
 
 # ----------------------------------------------------------------
 def create_and_populate_dataframe(path: str) -> soma.DataFrame:
-
     arrow_schema = pa.schema(
         [
             ("foo", pa.int32()),
@@ -45,7 +44,6 @@ def create_and_populate_sparse_nd_array(path: str) -> soma.SparseNDArray:
     with soma.SparseNDArray.create(
         path, type=pa.int64(), shape=(nr, nc)
     ) as sparse_nd_array:
-
         tensor = pa.SparseCOOTensor.from_numpy(
             data=np.asarray([7, 8, 9]),
             coords=[[0, 1], [2, 3], [3, 4]],

apis/python/tests/test_dataframe.py

@@ -164,7 +164,6 @@ def simple_data_frame(tmp_path):
     with soma.DataFrame.create(
         tmp_path.as_posix(), schema=schema, index_column_names=index_column_names
     ) as sdf:
-
         data = {
             "index": [0, 1, 2, 3],
             "soma_joinid": [10, 11, 12, 13],
@@ -1039,7 +1038,6 @@ def test_create_platform_config_overrides(
 @pytest.mark.parametrize("allows_duplicates", [False, True])
 @pytest.mark.parametrize("consolidate", [False, True])
 def test_timestamped_ops(tmp_path, allows_duplicates, consolidate):
-
     uri = tmp_path.as_posix()
 
     platform_config = {"tiledb": {"create": {"allows_duplicates": allows_duplicates}}}

apis/python/tests/test_dense_nd_array.py

@@ -74,7 +74,6 @@ def test_dense_nd_array_read_write_tensor(tmp_path, shape: Tuple[int, ...]):
 
     # check multiple read paths
     with soma.DenseNDArray.open(tmp_path.as_posix()) as b:
-
         t = b.read((slice(None),) * ndim, result_order="row-major")
         assert t.equals(pa.Tensor.from_numpy(data))
 
@@ -313,7 +312,6 @@ def test_dense_nd_array_indexing_errors(tmp_path, io):
     with soma.DenseNDArray.create(
         tmp_path.as_posix(), type=pa.int64(), shape=shape
     ) as a:
-
         npa = np.random.default_rng().standard_normal(np.prod(shape)).reshape(shape)
 
         write_coords = tuple(slice(0, dim_len) for dim_len in shape)

apis/python/tests/test_experiment_basic.py

@@ -10,7 +10,6 @@
 
 # ----------------------------------------------------------------
 def create_and_populate_obs(uri: str) -> soma.DataFrame:
-
     obs_arrow_schema = pa.schema(
         [
             ("foo", pa.int32()),
@@ -21,7 +20,6 @@ def create_and_populate_obs(uri: str) -> soma.DataFrame:
 
     # TODO: indexing option ...
     with soma.DataFrame.create(uri, schema=obs_arrow_schema) as obs:
-
         pydict = {}
         pydict["soma_joinid"] = [0, 1, 2, 3, 4]
         pydict["foo"] = [10, 20, 30, 40, 50]
@@ -35,7 +33,6 @@ def create_and_populate_obs(uri: str) -> soma.DataFrame:
 
 # ----------------------------------------------------------------
 def create_and_populate_var(uri: str) -> soma.DataFrame:
-
     var_arrow_schema = pa.schema(
         [
             ("quux", pa.large_string()),
@@ -44,7 +41,6 @@ def create_and_populate_var(uri: str) -> soma.DataFrame:
     )
 
     with soma.DataFrame.create(uri, schema=var_arrow_schema) as var:
-
         pydict = {}
         pydict["soma_joinid"] = [0, 1, 2, 3]
         pydict["quux"] = ["zebra", "yak", "xylophone", "wapiti"]
@@ -70,7 +66,6 @@ def create_and_populate_sparse_nd_array(uri: str) -> soma.SparseNDArray:
     with soma.SparseNDArray.create(
         uri, type=pa.int64(), shape=[nr, nc]
     ) as sparse_nd_array:
-
         tensor = pa.SparseCOOTensor.from_numpy(
             data=np.asarray([7, 8, 9]),
             coords=[[0, 2], [3, 1], [4, 2]],

apis/python/tests/test_metadata.py

@@ -195,7 +195,7 @@ def test_metadata_marshalling_OK(soma_object, test_value):
     assert "test_value" in soma_object.metadata
 
     val = soma_object.metadata["test_value"]
-    if type(test_value) is float and math.isnan(test_value):
+    if isinstance(test_value, float) and math.isnan(test_value):
         # By definition, NaN != NaN, so we can't just compare
         assert math.isnan(val)
     else:

apis/python/tests/test_registration_signatures.py

@@ -45,7 +45,6 @@ def test_signature_serdes(canned_h5ad_file, canned_anndata):
 
 
 def test_compatible(canned_anndata):
-
     # Check that zero inputs result in zero incompatibility
     signatures.Signature.check_compatible({})
 

apis/python/tests/test_tiledbobject.py

@@ -3,7 +3,6 @@
 
 import tiledbsoma as soma
 
-
 # Checking that objects _do_ exist is already done (thoroughly) in other tests. Here
 # we primarily focus on the negative cases.
 
@@ -109,7 +108,6 @@ def test_tiledbobject_exists_cross_types(tmp_path, name1, name2):
         assert cls1.exists(uri1)
 
     else:
-
         uri1 = (tmp_path / name1).as_posix()
         uri2 = (tmp_path / name2).as_posix()
 

apis/python/tests/test_util.py

@@ -10,7 +10,6 @@
 
 
 def test_uri_joinpath_file():
-
     assert uri_joinpath("/A/", "B") == "/A/B"
     assert uri_joinpath("/A/", "/B") == "/B"
     assert uri_joinpath("/A/B", "C") == "/A/B/C"