1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
|
from __future__ import annotations
from collections.abc import Callable, Iterable, Sequence
from typing import TYPE_CHECKING, Any
import numpy as np
from xarray.core.indexing import ImplicitToExplicitIndexingAdapter
from xarray.namedarray.parallelcompat import ChunkManagerEntrypoint, T_ChunkedArray
from xarray.namedarray.utils import is_duck_dask_array, module_available
if TYPE_CHECKING:
from xarray.namedarray._typing import (
T_Chunks,
_DType_co,
_NormalizedChunks,
duckarray,
)
try:
from dask.array import Array as DaskArray
except ImportError:
DaskArray = np.ndarray[Any, Any]
dask_available = module_available("dask")
class DaskManager(ChunkManagerEntrypoint["DaskArray"]):
array_cls: type[DaskArray]
available: bool = dask_available
def __init__(self) -> None:
# TODO can we replace this with a class attribute instead?
from dask.array import Array
self.array_cls = Array
def is_chunked_array(self, data: duckarray[Any, Any]) -> bool:
return is_duck_dask_array(data)
def chunks(self, data: Any) -> _NormalizedChunks:
return data.chunks # type: ignore[no-any-return]
def normalize_chunks(
self,
chunks: T_Chunks | _NormalizedChunks,
shape: tuple[int, ...] | None = None,
limit: int | None = None,
dtype: _DType_co | None = None,
previous_chunks: _NormalizedChunks | None = None,
) -> Any:
"""Called by open_dataset"""
from dask.array.core import normalize_chunks
return normalize_chunks(
chunks,
shape=shape,
limit=limit,
dtype=dtype,
previous_chunks=previous_chunks,
) # type: ignore[no-untyped-call]
def from_array(
self, data: Any, chunks: T_Chunks | _NormalizedChunks, **kwargs: Any
) -> DaskArray | Any:
import dask.array as da
if isinstance(data, ImplicitToExplicitIndexingAdapter):
# lazily loaded backend array classes should use NumPy array operations.
kwargs["meta"] = np.ndarray
return da.from_array(
data,
chunks,
**kwargs,
) # type: ignore[no-untyped-call]
def compute(
self, *data: Any, **kwargs: Any
) -> tuple[np.ndarray[Any, _DType_co], ...]:
from dask.array import compute
return compute(*data, **kwargs) # type: ignore[no-untyped-call, no-any-return]
def persist(self, *data: Any, **kwargs: Any) -> tuple[DaskArray | Any, ...]:
from dask import persist
return persist(*data, **kwargs) # type: ignore[no-untyped-call, no-any-return]
@property
def array_api(self) -> Any:
from dask import array as da
return da
def reduction(
self,
arr: T_ChunkedArray,
func: Callable[..., Any],
combine_func: Callable[..., Any] | None = None,
aggregate_func: Callable[..., Any] | None = None,
axis: int | Sequence[int] | None = None,
dtype: _DType_co | None = None,
keepdims: bool = False,
) -> DaskArray | Any:
from dask.array import reduction
return reduction(
arr,
chunk=func,
combine=combine_func,
aggregate=aggregate_func,
axis=axis,
dtype=dtype,
keepdims=keepdims,
) # type: ignore[no-untyped-call]
def scan(
self,
func: Callable[..., Any],
binop: Callable[..., Any],
ident: float,
arr: T_ChunkedArray,
axis: int | None = None,
dtype: _DType_co | None = None,
**kwargs: Any,
) -> DaskArray | Any:
from dask.array.reductions import cumreduction
return cumreduction(
func,
binop,
ident,
arr,
axis=axis,
dtype=dtype,
**kwargs,
) # type: ignore[no-untyped-call]
def apply_gufunc(
self,
func: Callable[..., Any],
signature: str,
*args: Any,
axes: Sequence[tuple[int, ...]] | None = None,
axis: int | None = None,
keepdims: bool = False,
output_dtypes: Sequence[_DType_co] | None = None,
output_sizes: dict[str, int] | None = None,
vectorize: bool | None = None,
allow_rechunk: bool = False,
meta: tuple[np.ndarray[Any, _DType_co], ...] | None = None,
**kwargs: Any,
) -> Any:
from dask.array.gufunc import apply_gufunc
return apply_gufunc(
func,
signature,
*args,
axes=axes,
axis=axis,
keepdims=keepdims,
output_dtypes=output_dtypes,
output_sizes=output_sizes,
vectorize=vectorize,
allow_rechunk=allow_rechunk,
meta=meta,
**kwargs,
) # type: ignore[no-untyped-call]
def map_blocks(
self,
func: Callable[..., Any],
*args: Any,
dtype: _DType_co | None = None,
chunks: tuple[int, ...] | None = None,
drop_axis: int | Sequence[int] | None = None,
new_axis: int | Sequence[int] | None = None,
**kwargs: Any,
) -> Any:
from dask.array import map_blocks
# pass through name, meta, token as kwargs
return map_blocks(
func,
*args,
dtype=dtype,
chunks=chunks,
drop_axis=drop_axis,
new_axis=new_axis,
**kwargs,
) # type: ignore[no-untyped-call]
def blockwise(
self,
func: Callable[..., Any],
out_ind: Iterable[Any],
*args: Any,
# can't type this as mypy assumes args are all same type, but dask blockwise args alternate types
name: str | None = None,
token: Any | None = None,
dtype: _DType_co | None = None,
adjust_chunks: dict[Any, Callable[..., Any]] | None = None,
new_axes: dict[Any, int] | None = None,
align_arrays: bool = True,
concatenate: bool | None = None,
meta: tuple[np.ndarray[Any, _DType_co], ...] | None = None,
**kwargs: Any,
) -> DaskArray | Any:
from dask.array import blockwise
return blockwise(
func,
out_ind,
*args,
name=name,
token=token,
dtype=dtype,
adjust_chunks=adjust_chunks,
new_axes=new_axes,
align_arrays=align_arrays,
concatenate=concatenate,
meta=meta,
**kwargs,
) # type: ignore[no-untyped-call]
def unify_chunks(
self,
*args: Any, # can't type this as mypy assumes args are all same type, but dask unify_chunks args alternate types
**kwargs: Any,
) -> tuple[dict[str, _NormalizedChunks], list[DaskArray]]:
from dask.array.core import unify_chunks
return unify_chunks(*args, **kwargs) # type: ignore[no-any-return, no-untyped-call]
def store(
self,
sources: Any | Sequence[Any],
targets: Any,
**kwargs: Any,
) -> Any:
from dask.array import store
return store(
sources=sources,
targets=targets,
**kwargs,
)
def shuffle(
self, x: DaskArray, indexer: list[list[int]], axis: int, chunks: T_Chunks
) -> DaskArray:
import dask.array
if not module_available("dask", minversion="2024.08.1"):
raise ValueError(
"This method is very inefficient on dask<2024.08.1. Please upgrade."
)
if chunks is None:
chunks = "auto"
if chunks != "auto":
raise NotImplementedError("Only chunks='auto' is supported at present.")
return dask.array.shuffle(x, indexer, axis, chunks="auto")
|
