-
Notifications
You must be signed in to change notification settings - Fork 15
/
Copy pathtest_merge.py
176 lines (146 loc) · 5.88 KB
/
test_merge.py
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
from functools import partial
import dask
import numpy as np
import pytest
import xarray as xr
from openeo_pg_parser_networkx.pg_schema import ParameterReference
from openeo_processes_dask.process_implementations import merge_cubes
from openeo_processes_dask.process_implementations.cubes.merge import (
NEW_DIM_COORDS,
NEW_DIM_NAME,
)
from openeo_processes_dask.process_implementations.exceptions import (
OverlapResolverMissing,
)
from tests.mockdata import create_fake_rastercube
@pytest.mark.parametrize("size", [(6, 5, 4, 4)])
@pytest.mark.parametrize("dtype", [np.float64])
def test_merge_cubes_type_1(temporal_interval, bounding_box, random_raster_data):
"""See Example 1 from https://processes.openeo.org/#merge_cubes."""
origin_cube = create_fake_rastercube(
data=random_raster_data,
spatial_extent=bounding_box,
temporal_extent=temporal_interval,
bands=["B02", "B03", "B04", "--324"],
backend="dask",
)
cube_1 = origin_cube.drop_sel({"bands": ["B04", "--324"]})
cube_2 = origin_cube.drop_sel({"bands": ["B02", "B03"]})
merged_cube = merge_cubes(cube_1, cube_2)
assert isinstance(merged_cube.data, dask.array.Array)
xr.testing.assert_equal(merged_cube, origin_cube)
@pytest.mark.parametrize("size", [(6, 5, 4, 3)])
@pytest.mark.parametrize("dtype", [np.float64])
def test_merge_cubes_type_2(
temporal_interval, bounding_box, random_raster_data, process_registry
):
origin_cube = create_fake_rastercube(
data=random_raster_data,
spatial_extent=bounding_box,
temporal_extent=temporal_interval,
bands=["B01", "B02", "B03"],
backend="dask",
)
cube_1 = origin_cube.drop_sel({"bands": "B03"})
cube_2 = origin_cube.drop_sel({"bands": "B01"})
with pytest.raises(OverlapResolverMissing):
merge_cubes(cube_1, cube_2)
overlap_resolver = partial(
process_registry["add"].implementation,
x=ParameterReference(from_parameter="x"),
y=ParameterReference(from_parameter="y"),
)
merged_cube = merge_cubes(cube_1, cube_2, overlap_resolver=overlap_resolver)
assert isinstance(merged_cube.data, dask.array.Array)
xr.testing.assert_equal(
merged_cube.sel({"bands": "B02"}) / 2, origin_cube.sel({"bands": "B02"})
)
@pytest.mark.parametrize("size", [(6, 5, 4, 3)])
@pytest.mark.parametrize("dtype", [np.float64])
def test_merge_cubes_type_3(
temporal_interval, bounding_box, random_raster_data, process_registry
):
# This is basically broadcasting the smaller datacube and then applying the overlap resolver.
origin_cube = create_fake_rastercube(
data=random_raster_data,
spatial_extent=bounding_box,
temporal_extent=temporal_interval,
bands=["B01", "B02", "B03"],
backend="dask",
)
cube_1 = origin_cube
cube_2 = origin_cube
# If no overlap reducer is provided, then simply concatenate along a new dimension
merged_cube = merge_cubes(cube_1, cube_2)
expected_result = xr.concat([cube_1, cube_2], dim=NEW_DIM_NAME).reindex(
{NEW_DIM_NAME: NEW_DIM_COORDS}
)
xr.testing.assert_equal(merged_cube, expected_result)
# If an overlap reducer is provided, then reduce per pixel
merged_cube = merge_cubes(
cube_1,
cube_2,
partial(
process_registry["add"].implementation,
x=ParameterReference(from_parameter="x"),
y=ParameterReference(from_parameter="y"),
),
)
assert isinstance(merged_cube.data, dask.array.Array)
xr.testing.assert_equal(merged_cube, cube_1 * 2)
@pytest.mark.parametrize("size", [(6, 5, 4, 3)])
@pytest.mark.parametrize("dtype", [np.float64])
def test_merge_cubes_type_4(
temporal_interval, bounding_box, random_raster_data, process_registry
):
# This is basically broadcasting the smaller datacube and then applying the overlap resolver.
cube_1 = create_fake_rastercube(
data=random_raster_data,
spatial_extent=bounding_box,
temporal_extent=temporal_interval,
bands=["B01", "B02", "B03"],
backend="dask",
)
cube_2 = xr.DataArray(
np.ones((len(cube_1["x"]), len(cube_1["y"]))),
dims=["x", "y"],
coords={"x": cube_1.coords["x"], "y": cube_1.coords["y"]},
)
with pytest.raises(OverlapResolverMissing):
merge_cubes(cube_1, cube_2)
overlap_resolver = partial(
process_registry["add"].implementation,
x=ParameterReference(from_parameter="x"),
y=ParameterReference(from_parameter="y"),
)
merged_cube_1 = merge_cubes(cube_1, cube_2, overlap_resolver=overlap_resolver)
merged_cube_2 = merge_cubes(cube_2, cube_1, overlap_resolver=overlap_resolver)
assert isinstance(merged_cube_1.data, dask.array.Array)
xr.testing.assert_equal(merged_cube_1, cube_1 + 1)
assert isinstance(merged_cube_2.data, dask.array.Array)
xr.testing.assert_equal(merged_cube_2, cube_1 + 1)
@pytest.mark.parametrize("size", [(6, 5, 4, 1)])
@pytest.mark.parametrize("dtype", [np.float64])
def test_conflicting_coords(
temporal_interval, bounding_box, random_raster_data, process_registry
):
# See https://github.com/Open-EO/openeo-processes-dask/pull/148 for why is is necessary
# This is basically broadcasting the smaller datacube and then applying the overlap resolver.
cube_1 = create_fake_rastercube(
data=random_raster_data,
spatial_extent=bounding_box,
temporal_extent=temporal_interval,
bands=["B01"],
backend="dask",
)
cube_1["s2:processing_baseline"] = "05.8"
cube_2 = create_fake_rastercube(
data=random_raster_data,
spatial_extent=bounding_box,
temporal_extent=temporal_interval,
bands=["B02"],
backend="dask",
)
cube_2["s2:processing_baseline"] = "05.9"
merged_cube_1 = merge_cubes(cube_1, cube_2)
assert isinstance(merged_cube_1.data, dask.array.Array)