Backport PR #1318 on branch 0.12.x (fix: use standard version API) (#2132)

Co-authored-by: Philipp A <flying-sheep@web.de>

Author: meeseeksmachine

.github/workflows/test-cpu.yml

@@ -120,11 +120,6 @@ jobs:
           python -m build --sdist --wheel .
           twine check dist/*
 
-      - name: Check runtime version
-        run: |
-          pip install dist/*.whl
-          python -c 'import anndata; print(anndata.__version__)'
-
   check:
     if: always()
     needs:

docs/release-notes/1318.fix.md

@@ -0,0 +1 @@
+Deprecate `__version__` and use standard {func}`~importlib.metadata.version` API {user}`flying-sheep`

pyproject.toml

@@ -57,8 +57,6 @@ Home-page = "https://github.com/scverse/anndata"
 
 [project.optional-dependencies]
 dev = [
-    # runtime dev version generation
-    "hatch-vcs",
     "anndata[dev-doc]",
 ]
 doc = [
@@ -146,7 +144,9 @@ addopts = [
 filterwarnings = [
     "ignore::anndata._warnings.OldFormatWarning",
     "ignore::anndata._warnings.ExperimentalFeatureWarning",
-    "ignore:.*first_column_names:FutureWarning:scanpy",     # scanpy 1.10.x
+    "ignore:.*first_column_names:FutureWarning:scanpy",                           # scanpy 1.10.x
+    "ignore:Importing read_.* from `anndata` is deprecated:FutureWarning:scanpy",
+    "ignore:`__version__` is deprecated:FutureWarning:scanpy",
 ]
 # When `--strict-warnings` is used, all warnings are treated as errors, except those:
 filterwarnings_when_strict = [

src/anndata/__init__.py

@@ -12,7 +12,6 @@
 from ._core.merge import concat
 from ._core.raw import Raw
 from ._settings import settings
-from ._version import __version__
 from ._warnings import (
     ExperimentalFeatureWarning,
     ImplicitModificationWarning,
@@ -28,30 +27,13 @@
 # We use these in tests by attribute access
 from . import logging  # noqa: F401  # isort: skip
 
-_DEPRECATED_IO = (
-    "read_loom",
-    "read_hdf",
-    "read_excel",
-    "read_umi_tools",
-    "read_csv",
-    "read_text",
-    "read_mtx",
-)
-_DEPRECATED = {method: f"io.{method}" for method in _DEPRECATED_IO}
-
-
-def __getattr__(attr_name: str) -> Any:
-    return module_get_attr_redirect(attr_name, deprecated_mapping=_DEPRECATED)
-
-
 __all__ = [
     "AnnData",
     "ExperimentalFeatureWarning",
     "ImplicitModificationWarning",
     "OldFormatWarning",
     "Raw",
     "WriteWarning",
-    "__version__",
     "abc",
     "concat",
     "experimental",
@@ -63,3 +45,26 @@ def __getattr__(attr_name: str) -> Any:
     "types",
     "typing",
 ]
+
+_DEPRECATED_IO = (
+    "read_loom",
+    "read_hdf",
+    "read_excel",
+    "read_umi_tools",
+    "read_csv",
+    "read_text",
+    "read_mtx",
+)
+_DEPRECATED = {method: f"io.{method}" for method in _DEPRECATED_IO}
+
+
+def __getattr__(attr_name: str) -> Any:
+    if attr_name == "__version__":
+        import warnings
+        from importlib.metadata import version
+
+        msg = "`__version__` is deprecated, use `importlib.metadata.version('anndata')` instead."
+        warnings.warn(msg, FutureWarning, stacklevel=2)
+        return version("anndata")
+
+    return module_get_attr_redirect(attr_name, deprecated_mapping=_DEPRECATED)

src/anndata/_core/merge.py

@@ -14,9 +14,7 @@
 
 import numpy as np
 import pandas as pd
-import scipy
 from natsort import natsorted
-from packaging.version import Version
 from scipy import sparse
 
 from anndata._core.file_backing import to_memory
@@ -30,7 +28,6 @@
     CupyCSRMatrix,
     CupySparseMatrix,
     DaskArray,
-    _map_cat_to_str,
 )
 from ..utils import asarray, axis_len, warn_once
 from .anndata import AnnData
@@ -185,15 +182,6 @@ def equal_sparse(a, b) -> bool:
             # Comparison broken for CSC matrices
             # https://github.com/cupy/cupy/issues/7757
             a, b = CupyCSRMatrix(a), CupyCSRMatrix(b)
-        if Version(scipy.__version__) >= Version("1.16.0rc1"):
-            # TODO: https://github.com/scipy/scipy/issues/23068
-            return bool(
-                a.format == b.format
-                and (a.shape == b.shape)
-                and np.all(a.indptr == b.indptr)
-                and np.all(a.indices == b.indices)
-                and np.all((a.data == b.data) | (np.isnan(a.data) & np.isnan(b.data)))
-            )
         comp = a != b
         if isinstance(comp, bool):
             return not comp
@@ -1646,7 +1634,7 @@ def concat(  # noqa: PLR0912, PLR0913, PLR0915
     )
     if index_unique is not None:
         concat_indices = concat_indices.str.cat(
-            _map_cat_to_str(label_col), sep=index_unique
+            label_col.map(str, na_action="ignore"), sep=index_unique
         )
     concat_indices = pd.Index(concat_indices)
 

src/anndata/_core/sparse_dataset.py

@@ -16,14 +16,14 @@
 from abc import ABC
 from collections.abc import Iterable
 from functools import cached_property
+from importlib.metadata import version
 from itertools import accumulate, chain, pairwise
 from math import floor
 from pathlib import Path
 from typing import TYPE_CHECKING, NamedTuple
 
 import h5py
 import numpy as np
-import scipy
 import scipy.sparse as ss
 from packaging.version import Version
 from scipy.sparse import _sparsetools
@@ -54,7 +54,7 @@
     from scipy.sparse import spmatrix as _cs_matrix
 
 
-SCIPY_1_15 = Version(scipy.__version__) >= Version("1.15rc0")
+SCIPY_1_15 = Version(version("scipy")) >= Version("1.15rc0")
 
 
 class BackedFormat(NamedTuple):

src/anndata/_io/h5ad.py

@@ -27,7 +27,6 @@
 from .specs import read_elem, write_elem
 from .specs.registry import IOSpec, write_spec
 from .utils import (
-    H5PY_V3,
     _read_legacy_raw,
     idx_chunks_along_axis,
     no_write_dataset_2d,
@@ -324,16 +323,12 @@ def read_dataframe_legacy(dataset: h5py.Dataset) -> pd.DataFrame:
         "Consider rewriting it."
     )
     warn(msg, OldFormatWarning, stacklevel=2)
-    if H5PY_V3:
-        df = pd.DataFrame(
-            _decode_structured_array(
-                _from_fixed_length_strings(dataset[()]), dtype=dataset.dtype
-            )
+    df = pd.DataFrame(
+        _decode_structured_array(
+            _from_fixed_length_strings(dataset[()]), dtype=dataset.dtype
         )
-    else:
-        df = pd.DataFrame(_from_fixed_length_strings(dataset[()]))
-    df.set_index(df.columns[0], inplace=True)
-    return df
+    )
+    return df.set_index(df.columns[0])
 
 
 def read_dataframe(group: h5py.Group | h5py.Dataset) -> pd.DataFrame:
@@ -346,10 +341,9 @@ def read_dataframe(group: h5py.Group | h5py.Dataset) -> pd.DataFrame:
 
 @report_read_key_on_error
 def read_dataset(dataset: h5py.Dataset):
-    if H5PY_V3:
-        string_dtype = h5py.check_string_dtype(dataset.dtype)
-        if (string_dtype is not None) and (string_dtype.encoding == "utf-8"):
-            dataset = dataset.asstr()
+    string_dtype = h5py.check_string_dtype(dataset.dtype)
+    if (string_dtype is not None) and (string_dtype.encoding == "utf-8"):
+        dataset = dataset.asstr()
     value = dataset[()]
     if not hasattr(value, "dtype"):
         return value
@@ -362,10 +356,9 @@ def read_dataset(dataset: h5py.Dataset):
             return value[0]
     elif len(value.dtype.descr) > 1:  # Compound dtype
         # For backwards compat, now strings are written as variable length
-        dtype = value.dtype
-        value = _from_fixed_length_strings(value)
-        if H5PY_V3:
-            value = _decode_structured_array(value, dtype=dtype)
+        value = _decode_structured_array(
+            _from_fixed_length_strings(value), dtype=value.dtype
+        )
     if value.shape == ():
         value = value[()]
     return value

src/anndata/_io/specs/methods.py

@@ -4,6 +4,7 @@
 from collections.abc import Mapping
 from copy import copy
 from functools import partial
+from importlib.metadata import version
 from itertools import product
 from types import MappingProxyType
 from typing import TYPE_CHECKING
@@ -21,7 +22,7 @@
 from anndata._core.index import _normalize_indices
 from anndata._core.merge import intersect_keys
 from anndata._core.sparse_dataset import _CSCDataset, _CSRDataset, sparse_dataset
-from anndata._io.utils import H5PY_V3, check_key, zero_dim_array_as_scalar
+from anndata._io.utils import check_key, zero_dim_array_as_scalar
 from anndata._warnings import OldFormatWarning
 from anndata.compat import (
     NULLABLE_NUMPY_STRING_TYPE,
@@ -609,7 +610,7 @@ def write_vlen_string_array_zarr(
     if is_zarr_v2():
         import numcodecs
 
-        if Version(numcodecs.__version__) < Version("0.13"):
+        if Version(version("numcodecs")) < Version("0.13"):
             msg = "Old numcodecs version detected. Please update for improved performance and stability."
             warnings.warn(msg, UserWarning, stacklevel=2)
             # Workaround for https://github.com/zarr-developers/numcodecs/issues/514
@@ -665,10 +666,9 @@ def _to_hdf5_vlen_strings(value: np.ndarray) -> np.ndarray:
 @_REGISTRY.register_read(ZarrArray, IOSpec("rec-array", "0.2.0"))
 def read_recarray(d: ArrayStorageType, *, _reader: Reader) -> np.recarray | npt.NDArray:
     value = d[()]
-    dtype = value.dtype
-    value = _from_fixed_length_strings(value)
-    if H5PY_V3:
-        value = _decode_structured_array(value, dtype=dtype)
+    value = _decode_structured_array(
+        _from_fixed_length_strings(value), dtype=value.dtype
+    )
     return value
 
 

src/anndata/_io/utils.py

@@ -1,13 +1,11 @@
 from __future__ import annotations
 
+from collections.abc import Callable
 from functools import WRAPPER_ASSIGNMENTS, wraps
 from itertools import pairwise
-from typing import TYPE_CHECKING, cast
+from typing import TYPE_CHECKING, Literal, cast
 from warnings import warn
 
-import h5py
-from packaging.version import Version
-
 from .._core.sparse_dataset import BaseCompressedSparseDataset
 
 if TYPE_CHECKING:
@@ -21,9 +19,6 @@
 
     Storage = StorageType | BaseCompressedSparseDataset
 
-# For allowing h5py v3
-# https://github.com/scverse/anndata/issues/442
-H5PY_V3 = Version(h5py.__version__).major >= 3
 
 # -------------------------------------------------------------------------------
 # Type conversion