Enforce ruff/flake8-pyi rules (PYI) (pydata#10359)

DimitriPapadopoulos · web-flow · commit 83ccfff46464 · 2025-05-29T10:16:29.000-07:00
* Apply ruff/flake8-pyi rule PYI018 PYI018 Private TypeVar is never used * Apply ruff/flake8-pyi rule PYI024 PYI024 Use `typing.NamedTuple` instead of `collections.namedtuple` * Apply ruff/flake8-pyi rule PYI025 PYI025 Use `from collections.abc import Set as AbstractSet` to avoid confusion with the `set` builtin * Apply ruff/flake8-pyi rule PYI032 PYI032 Prefer `object` to `Any` for the second parameter * Apply ruff/flake8-pyi rule PYI034 PYI034 such methods usually return `self` at runtime * Apply ruff/flake8-pyi rule PYI046 PYI046 Private protocol is never used * Apply ruff/flake8-pyi rule PYI051 PYI051 `Literal["..."]` is redundant in a union with `str` * Apply ruff/flake8-pyi rule PYI055 PYI055 Multiple `type` members in a union. Combine them into one * Apply ruff/flake8-pyi rule PYI063 PYI063 Use PEP 570 syntax for positional-only parameters * Enforce ruff/flake8-pyi rules (PYI) * Manual typing fix By definition `T_Chunks` and `T_Engine` could actually be `None`. * Partially revert "Apply ruff/flake8-pyi rule PYI051" This partially reverts commit 4bc134a.
diff --git a/pyproject.toml b/pyproject.toml
@@ -254,6 +254,7 @@ extend-select = [
   "ISC",  # flake8-implicit-str-concat
   "PIE",  # flake8-pie
   "TID",  # flake8-tidy-imports (absolute imports)
+  "PYI",  # flake8-pyi
   "I",    # isort
   "PERF", # Perflint
   "W",    # pycodestyle warnings
@@ -269,6 +270,8 @@ extend-safe-fixes = [
 ignore = [
   "C40",     # unnecessary generator, comprehension, or literal
   "PIE790",  # unnecessary pass statement
+  "PYI019",  # use `Self` instead of custom TypeVar
+  "PYI041",  # use `float` instead of `int | float`
   "PERF203", # try-except within a loop incurs performance overhead
   "E402",    # module level import not at top of file
   "E731",    # do not assign a lambda expression, use a def
@@ -284,6 +287,9 @@ ignore = [
 [tool.ruff.lint.per-file-ignores]
 # don't enforce absolute imports
 "asv_bench/**" = ["TID252"]
+# looks like ruff bugs
+"xarray/core/_typed_ops.py" = ["PYI034"]
+"xarray/namedarray/_typing.py" = ["PYI018", "PYI046"]
 
 [tool.ruff.lint.isort]
 known-first-party = ["xarray"]
diff --git a/xarray/backends/api.py b/xarray/backends/api.py
@@ -71,7 +71,7 @@
     T_NetcdfEngine = Literal["netcdf4", "scipy", "h5netcdf"]
     T_Engine = Union[
         T_NetcdfEngine,
-        Literal["pydap", "zarr"],
+        Literal["pydap", "zarr"],  # noqa: PYI051
         type[BackendEntrypoint],
         str,  # no nice typing support for custom backends
         None,
@@ -710,8 +710,8 @@ def open_dataset(
 def open_dataarray(
     filename_or_obj: str | os.PathLike[Any] | ReadBuffer | AbstractDataStore,
     *,
-    engine: T_Engine | None = None,
-    chunks: T_Chunks | None = None,
+    engine: T_Engine = None,
+    chunks: T_Chunks = None,
     cache: bool | None = None,
     decode_cf: bool | None = None,
     mask_and_scale: bool | None = None,
@@ -1394,7 +1394,7 @@ def open_mfdataset(
     | os.PathLike
     | ReadBuffer
     | NestedSequence[str | os.PathLike | ReadBuffer],
-    chunks: T_Chunks | None = None,
+    chunks: T_Chunks = None,
     concat_dim: (
         str
         | DataArray
@@ -1406,7 +1406,7 @@ def open_mfdataset(
     ) = None,
     compat: CompatOptions = "no_conflicts",
     preprocess: Callable[[Dataset], Dataset] | None = None,
-    engine: T_Engine | None = None,
+    engine: T_Engine = None,
     data_vars: Literal["all", "minimal", "different"] | list[str] = "all",
     coords="different",
     combine: Literal["by_coords", "nested"] = "by_coords",
diff --git a/xarray/computation/apply_ufunc.py b/xarray/computation/apply_ufunc.py
@@ -16,16 +16,14 @@
     Iterator,
     Mapping,
     Sequence,
-    Set,
 )
-from typing import TYPE_CHECKING, Any, Literal, TypeVar, Union
+from collections.abc import (
+    Set as AbstractSet,
+)
+from typing import TYPE_CHECKING, Any, Literal
 
 import numpy as np
 
-_T = TypeVar("_T", bound=Union["Dataset", "DataArray"])
-_U = TypeVar("_U", bound=Union["Dataset", "DataArray"])
-_V = TypeVar("_V", bound=Union["Dataset", "DataArray"])
-
 from xarray.core import duck_array_ops, utils
 from xarray.core.formatting import limit_lines
 from xarray.core.indexes import Index, filter_indexes_from_coords
@@ -200,7 +198,7 @@ def _get_coords_list(args: Iterable[Any]) -> list[Coordinates]:
 def build_output_coords_and_indexes(
     args: Iterable[Any],
     signature: _UFuncSignature,
-    exclude_dims: Set = frozenset(),
+    exclude_dims: AbstractSet = frozenset(),
     combine_attrs: CombineAttrsOptions = "override",
 ) -> tuple[list[dict[Any, Variable]], list[dict[Any, Index]]]:
     """Build output coordinates and indexes for an operation.
@@ -615,7 +613,7 @@ def apply_groupby_func(func, *args):
 
 
 def unified_dim_sizes(
-    variables: Iterable[Variable], exclude_dims: Set = frozenset()
+    variables: Iterable[Variable], exclude_dims: AbstractSet = frozenset()
 ) -> dict[Hashable, int]:
     dim_sizes: dict[Hashable, int] = {}
 
@@ -894,7 +892,7 @@ def apply_ufunc(
     *args: Any,
     input_core_dims: Sequence[Sequence] | None = None,
     output_core_dims: Sequence[Sequence] | None = ((),),
-    exclude_dims: Set = frozenset(),
+    exclude_dims: AbstractSet = frozenset(),
     vectorize: bool = False,
     join: JoinOptions = "exact",
     dataset_join: str = "exact",
diff --git a/xarray/core/accessor_dt.py b/xarray/core/accessor_dt.py
@@ -5,6 +5,7 @@
 
 import numpy as np
 import pandas as pd
+from typing_extensions import Self
 
 from xarray.coding.calendar_ops import _decimal_year
 from xarray.coding.times import infer_calendar_name
@@ -650,7 +651,7 @@ def total_seconds(self) -> T_DataArray:
 class CombinedDatetimelikeAccessor(
     DatetimeAccessor[T_DataArray], TimedeltaAccessor[T_DataArray]
 ):
-    def __new__(cls, obj: T_DataArray) -> CombinedDatetimelikeAccessor:
+    def __new__(cls, obj: T_DataArray) -> Self:
         # CombinedDatetimelikeAccessor isn't really instantiated. Instead
         # we need to choose which parent (datetime or timedelta) is
         # appropriate. Since we're checking the dtypes anyway, we'll just
diff --git a/xarray/core/datatree_render.py b/xarray/core/datatree_render.py
@@ -8,15 +8,18 @@
 
 from __future__ import annotations
 
-from collections import namedtuple
 from collections.abc import Iterable, Iterator
 from math import ceil
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, NamedTuple
 
 if TYPE_CHECKING:
     from xarray.core.datatree import DataTree
 
-Row = namedtuple("Row", ("pre", "fill", "node"))
+
+class Row(NamedTuple):
+    pre: str
+    fill: str
+    node: DataTree | str
 
 
 class AbstractStyle:
diff --git a/xarray/core/indexes.py b/xarray/core/indexes.py
@@ -1652,7 +1652,7 @@ class Indexes(collections.abc.Mapping, Generic[T_PandasOrXarrayIndex]):
 
     """
 
-    _index_type: type[Index] | type[pd.Index]
+    _index_type: type[Index | pd.Index]
     _indexes: dict[Any, T_PandasOrXarrayIndex]
     _variables: dict[Any, Variable]
 
@@ -1670,7 +1670,7 @@ def __init__(
         self,
         indexes: Mapping[Any, T_PandasOrXarrayIndex] | None = None,
         variables: Mapping[Any, Variable] | None = None,
-        index_type: type[Index] | type[pd.Index] = Index,
+        index_type: type[Index | pd.Index] = Index,
     ):
         """Constructor not for public consumption.
 
diff --git a/xarray/core/types.py b/xarray/core/types.py
@@ -214,7 +214,7 @@ def copy(
 
 # FYI in some cases we don't allow `None`, which this doesn't take account of.
 # FYI the `str` is for a size string, e.g. "16MB", supported by dask.
-T_ChunkDim: TypeAlias = str | int | Literal["auto"] | None | tuple[int, ...]
+T_ChunkDim: TypeAlias = str | int | Literal["auto"] | None | tuple[int, ...]  # noqa: PYI051
 T_ChunkDimFreq: TypeAlias = Union["TimeResampler", T_ChunkDim]
 T_ChunksFreq: TypeAlias = T_ChunkDim | Mapping[Any, T_ChunkDimFreq]
 # We allow the tuple form of this (though arguably we could transition to named dims only)
@@ -329,7 +329,7 @@ def mode(self) -> str:
         # for _get_filepath_or_buffer
         ...
 
-    def seek(self, __offset: int, __whence: int = ...) -> int:
+    def seek(self, offset: int, whence: int = ..., /) -> int:
         # with one argument: gzip.GzipFile, bz2.BZ2File
         # with two arguments: zip.ZipFile, read_sas
         ...
@@ -345,7 +345,7 @@ def tell(self) -> int:
 
 @runtime_checkable
 class ReadBuffer(BaseBuffer, Protocol[AnyStr_co]):
-    def read(self, __n: int = ...) -> AnyStr_co:
+    def read(self, n: int = ..., /) -> AnyStr_co:
         # for BytesIOWrapper, gzip.GzipFile, bz2.BZ2File
         ...
 
diff --git a/xarray/core/utils.py b/xarray/core/utils.py
@@ -61,9 +61,11 @@
     MutableMapping,
     MutableSet,
     Sequence,
-    Set,
     ValuesView,
 )
+from collections.abc import (
+    Set as AbstractSet,
+)
 from enum import Enum
 from pathlib import Path
 from types import EllipsisType, ModuleType
@@ -1055,7 +1057,7 @@ def parse_ordered_dims(
         )
 
 
-def _check_dims(dim: Set[Hashable], all_dims: Set[Hashable]) -> None:
+def _check_dims(dim: AbstractSet[Hashable], all_dims: AbstractSet[Hashable]) -> None:
     wrong_dims = (dim - all_dims) - {...}
     if wrong_dims:
         wrong_dims_str = ", ".join(f"'{d}'" for d in wrong_dims)
diff --git a/xarray/groupers.py b/xarray/groupers.py
@@ -902,7 +902,7 @@ def factorize(self, group: T_Group) -> EncodedGroups:
         first_items = agged["first"]
         counts = agged["count"]
 
-        index_class: type[CFTimeIndex] | type[pd.DatetimeIndex]
+        index_class: type[CFTimeIndex | pd.DatetimeIndex]
         if _contains_cftime_datetimes(group.data):
             index_class = CFTimeIndex
             datetime_class = type(first_n_items(group.data, 1).item())
diff --git a/xarray/namedarray/_typing.py b/xarray/namedarray/_typing.py
@@ -39,7 +39,6 @@ class Default(Enum):
 _default = Default.token
 
 # https://stackoverflow.com/questions/74633074/how-to-type-hint-a-generic-numpy-array
-_T = TypeVar("_T")
 _T_co = TypeVar("_T_co", covariant=True)
 
 _dtype = np.dtype
@@ -79,7 +78,7 @@ def dtype(self) -> _DType_co: ...
 _NormalizedChunks = tuple[tuple[int, ...], ...]
 # FYI in some cases we don't allow `None`, which this doesn't take account of.
 # # FYI the `str` is for a size string, e.g. "16MB", supported by dask.
-T_ChunkDim: TypeAlias = str | int | Literal["auto"] | None | tuple[int, ...]
+T_ChunkDim: TypeAlias = str | int | Literal["auto"] | None | tuple[int, ...]  # noqa: PYI051
 # We allow the tuple form of this (though arguably we could transition to named dims only)
 T_Chunks: TypeAlias = T_ChunkDim | Mapping[Any, T_ChunkDim]
 
diff --git a/xarray/namedarray/dtypes.py b/xarray/namedarray/dtypes.py
@@ -13,19 +13,19 @@
 
 @functools.total_ordering
 class AlwaysGreaterThan:
-    def __gt__(self, other: Any) -> Literal[True]:
+    def __gt__(self, other: object) -> Literal[True]:
         return True
 
-    def __eq__(self, other: Any) -> bool:
+    def __eq__(self, other: object) -> bool:
         return isinstance(other, type(self))
 
 
 @functools.total_ordering
 class AlwaysLessThan:
-    def __lt__(self, other: Any) -> Literal[True]:
+    def __lt__(self, other: object) -> Literal[True]:
         return True
 
-    def __eq__(self, other: Any) -> bool:
+    def __eq__(self, other: object) -> bool:
         return isinstance(other, type(self))
 
 
diff --git a/xarray/structure/chunks.py b/xarray/structure/chunks.py
@@ -167,15 +167,15 @@ def _maybe_chunk(
 
 
 @overload
-def unify_chunks(__obj: _T) -> tuple[_T]: ...
+def unify_chunks(obj: _T, /) -> tuple[_T]: ...
 
 
 @overload
-def unify_chunks(__obj1: _T, __obj2: _U) -> tuple[_T, _U]: ...
+def unify_chunks(obj1: _T, obj2: _U, /) -> tuple[_T, _U]: ...
 
 
 @overload
-def unify_chunks(__obj1: _T, __obj2: _U, __obj3: _V) -> tuple[_T, _U, _V]: ...
+def unify_chunks(obj1: _T, obj2: _U, obj3: _V, /) -> tuple[_T, _U, _V]: ...
 
 
 @overload
diff --git a/xarray/structure/merge.py b/xarray/structure/merge.py
@@ -1,7 +1,8 @@
 from __future__ import annotations
 
 from collections import defaultdict
-from collections.abc import Hashable, Iterable, Mapping, Sequence, Set
+from collections.abc import Hashable, Iterable, Mapping, Sequence
+from collections.abc import Set as AbstractSet
 from typing import TYPE_CHECKING, Any, NamedTuple, Union
 
 import pandas as pd
@@ -396,7 +397,7 @@ def collect_from_coordinates(
 def merge_coordinates_without_align(
     objects: list[Coordinates],
     prioritized: Mapping[Any, MergeElement] | None = None,
-    exclude_dims: Set = frozenset(),
+    exclude_dims: AbstractSet = frozenset(),
     combine_attrs: CombineAttrsOptions = "override",
 ) -> tuple[dict[Hashable, Variable], dict[Hashable, Index]]:
     """Merge variables/indexes from coordinates without automatic alignments.
diff --git a/xarray/tests/test_backends.py b/xarray/tests/test_backends.py
@@ -6213,9 +6213,7 @@ def test_h5netcdf_entrypoint(tmp_path: Path) -> None:
 
 @requires_netCDF4
 @pytest.mark.parametrize("str_type", (str, np.str_))
-def test_write_file_from_np_str(
-    str_type: type[str] | type[np.str_], tmpdir: str
-) -> None:
+def test_write_file_from_np_str(str_type: type[str | np.str_], tmpdir: str) -> None:
     # https://github.com/pydata/xarray/pull/5264
     scenarios = [str_type(v) for v in ["scenario_a", "scenario_b", "scenario_c"]]
     years = range(2015, 2100 + 1)
diff --git a/xarray/util/generate_aggregations.py b/xarray/util/generate_aggregations.py
@@ -13,9 +13,9 @@
 
 """
 
-import collections
 import textwrap
 from dataclasses import dataclass, field
+from typing import NamedTuple
 
 MODULE_PREAMBLE = '''\
 """Mixin classes with reduction operations."""
@@ -227,7 +227,14 @@ def {method}(
 and better supported. ``cumsum`` and ``cumprod`` may be deprecated
 in the future."""
 
-ExtraKwarg = collections.namedtuple("ExtraKwarg", "docs kwarg call example")
+
+class ExtraKwarg(NamedTuple):
+    docs: str
+    kwarg: str
+    call: str
+    example: str
+
+
 skipna = ExtraKwarg(
     docs=_SKIPNA_DOCSTRING,
     kwarg="skipna: bool | None = None,",
