[pull] main from pydata:main

.pre-commit-config.yaml

@@ -59,6 +59,21 @@ repos:
             typing-extensions>=4.1.0,
             numpy,
           ]
+  # keep the mypy version used in CI (pixi.toml) in sync with the hook above, so that the
+  # pre-commit.ci autoupdate also updates mypy in CI
+  - repo: local
+    hooks:
+      - id: sync-mypy-version
+        name: sync the mypy version in pixi.toml
+        language: python
+        files: ^\.pre-commit-config\.yaml$
+        pass_filenames: false
+        entry: >-
+          python -c "import pathlib, re;
+          config = pathlib.Path('.pre-commit-config.yaml').read_text();
+          version = re.search(r'mirrors-mypy\s+rev: v?([0-9.]+)', config).group(1);
+          pixi = pathlib.Path('pixi.toml');
+          pixi.write_text(re.sub(r'(?m)^mypy = \"==[0-9.]+\"', f'mypy = \"=={version}\"', pixi.read_text()))"
   - repo: https://github.com/citation-file-format/cff-converter-python
     rev: 5295f87c0e261da61a7b919fc754e3a77edd98a7
     hooks:

_stubtest/allowlist.txt

@@ -6,167 +6,33 @@
 # Format: Each line is a regex pattern matching fully qualified names to ignore.
 # See: https://mypy.readthedocs.io/en/stable/stubtest.html#allowlist
 
-# =============================================================================
-# typing module re-exports (metaclass/signature differences)
-# =============================================================================
-
-# Any has metaclass differences between stub and runtime
-xarray\.core\.(dataarray|dataset|variable)\.Any$
-xarray\.core\.(dataarray|dataset|variable)\.Any\.__new__$
-
-# Callable also has metaclass differences
-xarray\.core\.(dataarray|dataset|variable)\.Callable$
-
-# Self is a TYPE_CHECKING import
-xarray\.core\.(dataarray|variable)\.Self$
-
-# TypeVar has __mro_entries__ at runtime but not in stubs
-xarray\.core\.dataarray\.TypeVar\.__mro_entries__$
-
 # =============================================================================
 # TYPE_CHECKING imports - not present at runtime
 # =============================================================================
 
-# Type aliases from xarray.core.types
-xarray\.core\.(dataarray|dataset|variable)\.Dims$
-xarray\.core\.(dataarray|dataset)\.DatetimeLike$
-xarray\.core\.(dataarray|dataset)\.DatetimeUnitOptions$
-xarray\.core\.(dataarray|dataset)\.ErrorOptions$
-xarray\.core\.(dataarray|dataset|variable)\.ErrorOptionsWithWarn$
-xarray\.core\.(dataarray|dataset)\.GroupIndices$
-xarray\.core\.(dataarray|dataset)\.GroupInput$
-xarray\.core\.(dataarray|dataset)\.Grouper$
-xarray\.core\.(dataarray|dataset)\.InterpOptions$
-xarray\.core\.(dataarray|dataset)\.NetcdfWriteModes$
-xarray\.core\.(dataarray|dataset|variable)\.PadModeOptions$
-xarray\.core\.(dataarray|dataset|variable)\.PadReflectOptions$
-xarray\.core\.(dataarray|dataset|variable)\.QuantileMethods$
-xarray\.core\.(dataarray|dataset)\.QueryEngineOptions$
-xarray\.core\.(dataarray|dataset)\.QueryParserOptions$
-xarray\.core\.(dataarray|dataset)\.ReindexMethodOptions$
-xarray\.core\.(dataarray|dataset)\.ResampleCompatible$
-xarray\.core\.(dataarray|dataset)\.Resampler$
-xarray\.core\.(dataarray|dataset)\.SideOptions$
-xarray\.core\.(dataarray|dataset)\.CoarsenBoundaryOptions$
-xarray\.core\.(dataarray|dataset)\.ZarrWriteModes$
-xarray\.core\.(dataarray|dataset)\.ZarrStore$
-xarray\.core\.(dataarray|dataset)\.ZarrStoreLike$
-xarray\.core\.(dataarray|dataset)\.ArrayLike$
-xarray\.core\.dataset\.CFCalendar$
-xarray\.core\.dataset\.CombineAttrsOptions$
-xarray\.core\.dataset\.CompatOptions$
-xarray\.core\.dataset\.JoinOptions$
-xarray\.core\.dataset\.CoercibleMapping$
-xarray\.core\.dataset\.CoercibleValue$
-xarray\.core\.dataset\.DataVars$
-xarray\.core\.dataset\.DsCompatible$
-
 # TypeVars
-xarray\.core\.(dataarray|dataset)\.T_ChunkDimFreq$
-xarray\.core\.(dataarray|dataset)\.T_ChunksFreq$
-xarray\.core\.(dataarray|dataset|variable)\.T_Chunks$
-xarray\.core\.(dataarray|dataset)\.T_NetcdfEngine$
-xarray\.core\.(dataarray|dataset)\.T_NetcdfTypes$
-xarray\.core\.(dataarray|dataset)\.T_Xarray$
 xarray\.core\.dataarray\.T_XarrayOther$
-xarray\.core\.dataset\.T_DatasetPadConstantValues$
-xarray\.core\.variable\.T_DuckArray$
-xarray\.core\.variable\.T_VarPadConstantValues$
 
 # External library types (dask, delayed, etc.)
-xarray\.core\.(dataarray|dataset|variable)\.ChunkManagerEntrypoint$
-xarray\.core\.(dataarray|dataset)\.DaskDataFrame$
-xarray\.core\.(dataarray|dataset)\.Delayed$
-xarray\.core\.dataset\.AbstractDataStore$
 xarray\.core\.dataarray\.iris_Cube$
 
-# DataArray TYPE_CHECKING import in dataset module
-xarray\.core\.dataset\.DataArray$
-
-# NamedArray TYPE_CHECKING import
-xarray\.core\.variable\.NamedArray$
-
-# =============================================================================
-# GroupBy/Rolling/Coarsen/Weighted/Resample classes (TYPE_CHECKING imports)
-# =============================================================================
-
-xarray\.core\.dataarray\.DataArrayGroupBy$
-xarray\.core\.dataarray\.DataArrayCoarsen$
-xarray\.core\.dataarray\.DataArrayRolling$
-xarray\.core\.dataarray\.DataArrayWeighted$
-xarray\.core\.dataarray\.DataArrayResample$
-xarray\.core\.dataset\.DatasetGroupBy$
-xarray\.core\.dataset\.DatasetCoarsen$
-xarray\.core\.dataset\.DatasetRolling$
-xarray\.core\.dataset\.DatasetWeighted$
-xarray\.core\.dataset\.DatasetResample$
-
-# =============================================================================
-# CFTimeIndex properties - read-only at runtime
-# =============================================================================
-
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.date_type$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.day$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.day_of_week$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.dayofweek$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.day_of_year$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.dayofyear$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.days_in_month$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.hour$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.microsecond$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.minute$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.month$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.second$
-xarray\.core\.(dataarray|dataset)\.CFTimeIndex\.year$
-
 # =============================================================================
 # Plot accessors and methods
 # =============================================================================
 
 xarray\.core\.dataarray\.DataArray\.plot$
 xarray\.core\.(dataarray|dataset)\.Dataset\.plot$
-xarray\.core\.dataarray\.DataArrayPlotAccessor\.(contour|contourf|imshow|line|pcolormesh|scatter|step|surface)$
-xarray\.core\.dataset\.DatasetPlotAccessor\.(quiver|scatter|streamplot)$
 
 # =============================================================================
 # __array__ method - complex numpy protocol
 # =============================================================================
 
 xarray\.core\.(dataarray|dataset)\.Dataset\.__array__$
 
-# =============================================================================
-# Mapping/MutableMapping/Sequence/Iterable protocol methods
-# =============================================================================
-
-xarray\.core\.(dataarray|dataset|variable)\.Mapping\.get$
-xarray\.core\.(dataarray|dataset|variable)\.Mapping\.__reversed__$
-xarray\.core\.(dataarray|dataset|variable)\.Sequence\.index$
-xarray\.core\.(dataarray|dataset)\.Iterable\.__class_getitem__$
-xarray\.core\.(dataarray|dataset)\.MutableMapping\.(pop|setdefault)$
-xarray\.core\.(dataarray|dataset)\.PathLike\.__class_getitem__$
-xarray\.core\.dataset\.FrozenMappingWarningOnValuesAccess\.get$
-
-# Number.__hash__
-xarray\.core\.dataset\.Number\.__hash__$
-
-# =============================================================================
-# IO protocol (typing.IO)
-# =============================================================================
-
-xarray\.core\.dataset\.IO\.(closed|mode|name|read|readline|readlines|seek|truncate|write|writelines|__iter__|__next__)$
-
 # =============================================================================
 # Variable/IndexVariable methods inherited from numpy-like interface
 # =============================================================================
 
 xarray\.core\.(dataarray|dataset|variable)\.Variable\.(item|searchsorted)$
 xarray\.core\.(dataarray|dataset|variable)\.IndexVariable\.(item|searchsorted)$
 xarray\.core\.dataarray\.DataArray\.(item|searchsorted)$
-xarray\.core\.dataarray\.DataArrayArithmetic\.(item|searchsorted)$
-xarray\.core\.variable\.VariableArithmetic\.(item|searchsorted)$
-
-# =============================================================================
-# Subclass pattern for Variable
-# =============================================================================
-
-xarray\.core\.variable\.<subclass.*

pixi.toml

@@ -221,7 +221,7 @@ linkcheck = { cmd = "make linkcheck", cwd = "doc", description = "Check URLs in
 
 
 [feature.typing.dependencies]
-mypy = "==1.19.1"
+mypy = "==2.3.1"
 pyright = "*"
 hypothesis = "*"
 lxml = "*"

pyproject.toml

@@ -132,14 +132,12 @@ exclude = ['build', 'xarray/util/generate_.*\.py']
 files = "xarray"
 show_error_context = true
 warn_redundant_casts = true
-warn_unused_configs = true
 warn_unused_ignores = true
 
 # Much of the numerical computing stack doesn't have type annotations yet.
 [[tool.mypy.overrides]]
 ignore_missing_imports = true
 module = [
-  "affine.*",
   "bottleneck.*",
   "cartopy.*",
   "cf_units.*",
@@ -155,7 +153,6 @@ module = [
   "mpl_toolkits.*",
   "nc_time_axis.*",
   "netCDF4.*",
-  "netcdftime.*",
   "numcodecs.*",
   "opt_einsum.*",
   "pint.*",

xarray/computation/computation.py

@@ -296,10 +296,7 @@ def _cov_corr(
         # Adjust covariance for degrees of freedom
         valid_count = valid_values.sum(dim)
         adjust = valid_count / (valid_count - ddof)
-        # I think the cast is required because of `T_DataArray` + `T_Xarray` (would be
-        # the same with `T_DatasetOrArray`)
-        # https://github.com/pydata/xarray/pull/8384#issuecomment-1784228026
-        return cast(T_DataArray, cov * adjust)
+        return cov * adjust
 
     else:
         # Compute std and corr

xarray/core/common.py

@@ -457,7 +457,7 @@ def squeeze(
         numpy.squeeze
         """
         dims = get_squeeze_dims(self, dim, axis)
-        return self.isel(drop=drop, **dict.fromkeys(dims, 0))
+        return self.isel(dict.fromkeys(dims, 0), drop=drop)
 
     def clip(
         self,

xarray/core/dataset.py

@@ -9268,6 +9268,7 @@ def pad(
         """
         pad_width = either_dict_or_kwargs(pad_width, pad_width_kwargs, "pad")
 
+        coord_pad_mode: PadModeOptions
         if mode in ("edge", "reflect", "symmetric", "wrap"):
             coord_pad_mode = mode
             coord_pad_options = {

xarray/indexes/nd_point_index.py

@@ -2,7 +2,7 @@
 
 import abc
 from collections.abc import Hashable, Iterable, Mapping
-from typing import TYPE_CHECKING, Any, Generic, TypeVar
+from typing import TYPE_CHECKING, Any, Generic, TypeVar, cast
 
 import numpy as np
 
@@ -268,7 +268,7 @@ def from_variables(
 
         tree_adapter_cls: type[T_TreeAdapter] = opts.pop("tree_adapter_cls", None)
         if tree_adapter_cls is None:
-            tree_adapter_cls = ScipyKDTreeAdapter
+            tree_adapter_cls = cast(type[T_TreeAdapter], ScipyKDTreeAdapter)
 
         points = get_points(variables.values())
 

xarray/tests/test_dataarray.py

@@ -1328,8 +1328,7 @@ def test_thin(self) -> None:
     def test_loc(self) -> None:
         self.ds["x"] = ("x", np.array(list("abcdefghij")))
         da = self.ds["foo"]
-        # typing issue: see https://github.com/python/mypy/issues/2410
-        assert_identical(da[:3], da.loc[:"c"])  # type: ignore[misc]
+        assert_identical(da[:3], da.loc[:"c"])
         assert_identical(da[1], da.loc["b"])
         assert_identical(da[1], da.loc[{"x": "b"}])
         assert_identical(da[1], da.loc["b", ...])
@@ -1347,8 +1346,7 @@ def test_loc_assign(self) -> None:
         self.ds["x"] = ("x", np.array(list("abcdefghij")))
         da = self.ds["foo"]
         # assignment
-        # typing issue: see https://github.com/python/mypy/issues/2410
-        da.loc["a":"j"] = 0  # type: ignore[misc]
+        da.loc["a":"j"] = 0
         assert np.all(da.values == 0)
         da.loc[{"x": slice("a", "j")}] = 2
         assert np.all(da.values == 2)
@@ -2921,6 +2919,11 @@ def test_transpose(self) -> None:
     def test_squeeze(self) -> None:
         assert_equal(self.dv.variable.squeeze(), self.dv.squeeze().variable)
 
+    def test_squeeze_non_str_dim_name(self) -> None:
+        array = DataArray(np.zeros((1, 3)), dims=[0, "y"])
+        expected = DataArray(np.zeros(3), dims=["y"])
+        assert_identical(array.squeeze(), expected)
+
     def test_squeeze_drop(self) -> None:
         array = DataArray([1], [("x", [0])])
         expected = DataArray(1)

xarray/tests/test_dataarray_typing.yml

@@ -12,15 +12,15 @@
 
     da = DataArray().pipe(lambda data, arg: arg, "foo")
 
-    reveal_type(da)  # N: Revealed type is "builtins.str"
+    reveal_type(da)  # N: Revealed type is "str"
 
 - case: test_mypy_pipe_lambda_chaining_return_type
   main: |
     from xarray import DataArray
 
     answer = DataArray().pipe(lambda data, arg: arg, "foo").count("o")
 
-    reveal_type(answer)  # N: Revealed type is "builtins.int"
+    reveal_type(answer)  # N: Revealed type is "int"
 
 - case: test_mypy_pipe_lambda_missing_arg
   main: |
@@ -31,7 +31,7 @@
   out: |
     main:4: error: No overload variant of "pipe" of "DataWithCoords" matches argument type "Callable[[Any, Any], Any]"  [call-overload]
     main:4: note: Possible overload variants:
-    main:4: note:     def [P`2, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:4: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:4: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_lambda_extra_arg
@@ -43,7 +43,7 @@
   out: |
     main:4: error: No overload variant of "pipe" of "DataWithCoords" matches argument types "Callable[[Any], Any]", "str"  [call-overload]
     main:4: note: Possible overload variants:
-    main:4: note:     def [P`2, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:4: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:4: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_function_missing_posarg
@@ -58,7 +58,7 @@
   out: |
     main:7: error: No overload variant of "pipe" of "DataWithCoords" matches argument type "Callable[[DataArray, int], DataArray]"  [call-overload]
     main:7: note: Possible overload variants:
-    main:7: note:     def [P`2, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:7: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:7: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_function_extra_posarg
@@ -73,7 +73,7 @@
   out: |
     main:7: error: No overload variant of "pipe" of "DataWithCoords" matches argument types "Callable[[DataArray, int], DataArray]", "int", "str"  [call-overload]
     main:7: note: Possible overload variants:
-    main:7: note:     def [P`2, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:7: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:7: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_function_missing_kwarg
@@ -88,7 +88,7 @@
   out: |
     main:7: error: No overload variant of "pipe" of "DataWithCoords" matches argument types "def f(da: DataArray, arg: int, *, kwonly: int) -> DataArray", "int"  [call-overload]
     main:7: note: Possible overload variants:
-    main:7: note:     def [P`2, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:7: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:7: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_function_missing_keyword
@@ -103,7 +103,7 @@
   out: |
     main:7: error: No overload variant of "pipe" of "DataWithCoords" matches argument types "def f(da: DataArray, arg: int, *, kwonly: int) -> DataArray", "int", "int"  [call-overload]
     main:7: note: Possible overload variants:
-    main:7: note:     def [P`2, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:7: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:7: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_function_unexpected_keyword
@@ -141,7 +141,7 @@
 
     answer = DataArray().pipe((f, "da"), 42)
 
-    reveal_type(answer)  # N: Revealed type is "builtins.int"
+    reveal_type(answer)  # N: Revealed type is "int"
 
 - case: test_mypy_pipe_tuple_missing_arg
   main: |
@@ -165,7 +165,7 @@
   out: |
     main:17: error: No overload variant of "pipe" of "DataWithCoords" matches argument type "Callable[[Any, Any], DataArray]"  [call-overload]
     main:17: note: Possible overload variants:
-    main:17: note:     def [P`9, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:17: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:17: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T
 
 - case: test_mypy_pipe_tuple_extra_arg
@@ -190,5 +190,5 @@
   out: |
     main:17: error: No overload variant of "pipe" of "DataWithCoords" matches argument types "Callable[[Any, Any], DataArray]", "int", "str"  [call-overload]
     main:17: note: Possible overload variants:
-    main:17: note:     def [P`9, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
+    main:17: note:     def [P, T] pipe(self, func: Callable[[DataArray, **P], T], *args: P.args, **kwargs: P.kwargs) -> T
     main:17: note:     def [T] pipe(self, func: tuple[Callable[..., T], str], *args: Any, **kwargs: Any) -> T