This repository was archived by the owner on May 7, 2026. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 68
Expand file tree
/
Copy pathwindow_spec.py
More file actions
315 lines (272 loc) · 9.71 KB
/
Copy pathwindow_spec.py
File metadata and controls
315 lines (272 loc) · 9.71 KB
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
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
# Copyright 2023 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import annotations
from dataclasses import dataclass, replace
import datetime
import itertools
from typing import Callable, Literal, Mapping, Optional, Sequence, Set, Tuple, Union
import numpy as np
import pandas as pd
import bigframes.core.expression as ex
import bigframes.core.identifiers as ids
import bigframes.core.ordering as orderings
# Unbound Windows
def unbound(
grouping_keys: Tuple[str, ...] = (),
min_periods: int = 0,
ordering: Tuple[orderings.OrderingExpression, ...] = (),
) -> WindowSpec:
"""
Create an unbound window.
Args:
grouping_keys:
Columns ids of grouping keys
min_periods (int, default 0):
Minimum number of input rows to generate output.
ordering:
Orders the rows within the window.
Returns:
WindowSpec
"""
return WindowSpec(
grouping_keys=tuple(map(ex.deref, grouping_keys)),
min_periods=min_periods,
ordering=ordering,
)
### Rows-based Windows
def rows(
grouping_keys: Tuple[str, ...] = (),
start: Optional[int] = None,
end: Optional[int] = None,
min_periods: int = 0,
ordering: Tuple[orderings.OrderingExpression, ...] = (),
) -> WindowSpec:
"""
Create a row-bounded window.
Args:
grouping_keys:
Columns ids of grouping keys
start:
The window's starting boundary relative to the current row. For example, "-1" means one row prior
"1" means one row after, and "0" means the current row. If None, the window is unbounded from the start.
following:
The window's ending boundary relative to the current row. For example, "-1" means one row prior
"1" means one row after, and "0" means the current row. If None, the window is unbounded until the end.
min_periods (int, default 0):
Minimum number of input rows to generate output.
ordering:
Ordering to apply on top of based dataframe ordering
Returns:
WindowSpec
"""
bounds = RowsWindowBounds(
start=start,
end=end,
)
return WindowSpec(
grouping_keys=tuple(map(ex.deref, grouping_keys)),
bounds=bounds,
min_periods=min_periods,
ordering=ordering,
)
def cumulative_rows(
grouping_keys: Tuple[str, ...] = (), min_periods: int = 0
) -> WindowSpec:
"""
Create a expanding window that includes all preceding rows
Args:
grouping_keys:
Columns ids of grouping keys
min_periods (int, default 0):
Minimum number of input rows to generate output.
Returns:
WindowSpec
"""
bounds = RowsWindowBounds(end=0)
return WindowSpec(
grouping_keys=tuple(map(ex.deref, grouping_keys)),
bounds=bounds,
min_periods=min_periods,
)
def inverse_cumulative_rows(
grouping_keys: Tuple[str, ...] = (), min_periods: int = 0
) -> WindowSpec:
"""
Create a shrinking window that includes all following rows
Args:
grouping_keys:
Columns ids of grouping keys
min_periods (int, default 0):
Minimum number of input rows to generate output.
Returns:
WindowSpec
"""
bounds = RowsWindowBounds(start=0)
return WindowSpec(
grouping_keys=tuple(map(ex.deref, grouping_keys)),
bounds=bounds,
min_periods=min_periods,
)
### Struct Classes
@dataclass(frozen=True)
class RowsWindowBounds:
start: Optional[int] = None
end: Optional[int] = None
@classmethod
def from_window_size(
cls, window: int, closed: Literal["right", "left", "both", "neither"]
) -> RowsWindowBounds:
if closed == "right":
return cls(-(window - 1), 0)
elif closed == "left":
return cls(-window, -1)
elif closed == "both":
return cls(-window, 0)
elif closed == "neither":
return cls(-(window - 1), -1)
else:
raise ValueError(f"Unsupported value for 'closed' parameter: {closed}")
def __post_init__(self):
if self.start is None:
return
if self.end is None:
return
if self.start > self.end:
raise ValueError(
f"Invalid window: start({self.start}) is greater than end({self.end})"
)
@dataclass(frozen=True)
class RangeWindowBounds:
"""Represents a time range window, inclusively bounded by start and end"""
start: pd.Timedelta | None = None
end: pd.Timedelta | None = None
@classmethod
def from_timedelta_window(
cls,
window: pd.Timedelta | np.timedelta64 | datetime.timedelta,
closed: Literal["right", "left", "both", "neither"],
) -> RangeWindowBounds:
window = pd.Timedelta(window)
tick = pd.Timedelta("1us")
zero = pd.Timedelta(0)
if closed == "right":
return cls(-(window - tick), zero)
elif closed == "left":
return cls(-window, -tick)
elif closed == "both":
return cls(-window, zero)
elif closed == "neither":
return cls(-(window - tick), -tick)
else:
raise ValueError(f"Unsupported value for 'closed' parameter: {closed}")
def __post_init__(self):
if self.start is None:
return
if self.end is None:
return
if self.start > self.end:
raise ValueError(
f"Invalid window: start({self.start}) is greater than end({self.end})"
)
@dataclass(frozen=True)
class WindowSpec:
"""
Specifies a window over which aggregate and analytic function may be applied.
Attributes:
grouping_keys: A set of columns to group on
bounds: The window boundaries
ordering: A list of columns ids and ordering direction to override base ordering
min_periods: The minimum number of observations in window required to have a value
"""
grouping_keys: Tuple[ex.Expression, ...] = tuple()
ordering: Tuple[orderings.OrderingExpression, ...] = tuple()
bounds: Union[RowsWindowBounds, RangeWindowBounds, None] = None
min_periods: int = 0
@property
def is_row_bounded(self):
"""
Whether the window is bounded by row offsets.
This is relevant for determining whether the window requires a total order
to calculate deterministically.
"""
return isinstance(self.bounds, RowsWindowBounds) and (
(self.bounds.start is not None) or (self.bounds.end is not None)
)
@property
def is_range_bounded(self):
"""
Whether the window is bounded by range offsets.
This is relevant for determining whether the window requires a total order
to calculate deterministically.
"""
return isinstance(self.bounds, RangeWindowBounds)
@property
def is_unbounded(self):
"""
Whether the window is unbounded.
This is relevant for determining whether the window requires a total order
to calculate deterministically.
"""
return self.bounds is None or (
self.bounds.start is None and self.bounds.end is None
)
@property
def expressions(self) -> Sequence[ex.Expression]:
ordering_exprs = (item.scalar_expression for item in self.ordering)
return (*self.grouping_keys, *ordering_exprs)
@property
def all_referenced_columns(self) -> Set[ids.ColumnId]:
"""
Return list of all variables reference ind the window.
"""
ordering_vars = itertools.chain.from_iterable(
item.scalar_expression.column_references for item in self.ordering
)
grouping_vars = itertools.chain.from_iterable(
item.column_references for item in self.grouping_keys
)
return set(itertools.chain(grouping_vars, ordering_vars))
def without_order(self, force: bool = False) -> WindowSpec:
"""Removes ordering clause if ordering isn't required to define bounds."""
if self.is_row_bounded and not force:
raise ValueError("Cannot remove order from row-bounded window")
return replace(self, ordering=())
def remap_column_refs(
self,
mapping: Mapping[ids.ColumnId, ids.ColumnId],
allow_partial_bindings: bool = False,
) -> WindowSpec:
return WindowSpec(
grouping_keys=tuple(
key.remap_column_refs(mapping, allow_partial_bindings)
for key in self.grouping_keys
),
ordering=tuple(
order_part.remap_column_refs(mapping, allow_partial_bindings)
for order_part in self.ordering
),
bounds=self.bounds,
min_periods=self.min_periods,
)
def transform_exprs(
self: WindowSpec, t: Callable[[ex.Expression], ex.Expression]
) -> WindowSpec:
return WindowSpec(
grouping_keys=tuple(t(key) for key in self.grouping_keys),
ordering=tuple(
order_part.transform_exprs(t) for order_part in self.ordering
),
bounds=self.bounds,
min_periods=self.min_periods,
)