def test_iter_chunk_keys_with_more_base_dims(self):
actual = sorted(core.iter_chunk_keys({'x': (3, 3)}, base={'x': 30, 'y': 0}))
expected = [
xarray_beam.ChunkKey({'x': 30, 'y': 0}),
xarray_beam.ChunkKey({'x': 33, 'y': 0}),
]
self.assertEqual(actual, expected)
def test_chunks_to_zarr(self):
dataset = xarray.Dataset(
{'foo': ('x', np.arange(0, 60, 10))},
coords={'x': np.arange(6)},
)
chunked = dataset.chunk()
inputs = [
(xarray_beam.ChunkKey({'x': 0}), dataset),
]
with self.subTest('no template'):
temp_dir = self.create_tempdir().full_path
inputs | xarray_beam.ChunksToZarr(temp_dir)
result = xarray.open_zarr(temp_dir, consolidated=True)
xarray.testing.assert_identical(dataset, result)
with self.subTest('with template'):
temp_dir = self.create_tempdir().full_path
inputs | xarray_beam.ChunksToZarr(temp_dir, chunked)
result = xarray.open_zarr(temp_dir, consolidated=True)
xarray.testing.assert_identical(dataset, result)
with self.subTest('with zarr_chunks and with template'):
temp_dir = self.create_tempdir().full_path
zarr_chunks = {'x': 3}
inputs | xarray_beam.ChunksToZarr(temp_dir, chunked, zarr_chunks)
result = xarray.open_zarr(temp_dir, consolidated=True)
xarray.testing.assert_identical(dataset, result)
self.assertEqual(result.chunks, {'x': (3, 3)})
with self.subTest('with zarr_chunks and no template'):
temp_dir = self.create_tempdir().full_path
zarr_chunks = {'x': 3}
inputs | xarray_beam.ChunksToZarr(temp_dir, zarr_chunks=zarr_chunks)
result = xarray.open_zarr(temp_dir, consolidated=True)
xarray.testing.assert_identical(dataset, result)
self.assertEqual(result.chunks, {'x': (3, 3)})
temp_dir = self.create_tempdir().full_path
with self.assertRaisesRegex(
ValueError,
'template does not have any variables chunked with Dask',
):
xarray_beam.ChunksToZarr(temp_dir, dataset)
temp_dir = self.create_tempdir().full_path
template = chunked.assign_coords(x=np.zeros(6))
with self.assertRaisesRegex(
ValueError,
'template and chunk indexes do not match',
):
inputs | xarray_beam.ChunksToZarr(temp_dir, template)
inputs2 = [
(xarray_beam.ChunkKey({'x': 0}),
dataset.expand_dims(z=[1, 2])),
]
temp_dir = self.create_tempdir().full_path
with self.assertRaisesRegex(
ValueError,
'unexpected new indexes found in chunk',
):
inputs2 | xarray_beam.ChunksToZarr(temp_dir, template)
def test_comparison(self):
key = xarray_beam.ChunkKey({'x': 0, 'y': 10})
with self.assertRaises(TypeError):
key < 'foo'
with self.assertRaisesRegex(ValueError, 'Dimensions must match'):
key < xarray_beam.ChunkKey({'x': 0})
other = xarray_beam.ChunkKey({'x': 0, 'y': 20})
self.assertLess(key, other)
self.assertGreater(other, key)
def test_use_as_beam_key(self):
inputs = [
(xarray_beam.ChunkKey({'x': 0, 'y': 1}), 1),
(xarray_beam.ChunkKey({'x': 0, 'y': 2}), 2),
(xarray_beam.ChunkKey({'y': 1, 'x': 0}), 3),
]
expected = [
(xarray_beam.ChunkKey({'x': 0, 'y': 1}), [1, 3]),
(xarray_beam.ChunkKey({'x': 0, 'y': 2}), [2]),
]
actual = inputs | beam.GroupByKey()
self.assertEqual(actual, expected)
def test_operators(self):
key = xarray_beam.ChunkKey({'x': 0, 'y': 10})
expected = xarray_beam.ChunkKey({'x': 0, 'y': 10, 'z': 100})
actual = key | {'z': 100}
self.assertEqual(actual, expected)
expected = xarray_beam.ChunkKey({'y': 10})
actual = key - {'x'}
self.assertEqual(actual, expected)
with self.assertRaises(TypeError):
key - 'x'
with self.assertRaisesRegex(ValueError, 'not found'):
key - {'z'}
def test_mapping(self):
key = xarray_beam.ChunkKey({'x': 0, 'y': 10})
self.assertEqual(list(key.keys()), ['x', 'y'])
self.assertEqual(list(key.values()), [0, 10])
self.assertIn('x', key)
self.assertNotIn('z', key)
self.assertEqual(key['x'], 0)
def test_immutability(self):
key = xarray_beam.ChunkKey({'x': 0, 'y': 10})
with self.assertRaises(TypeError):
key['z'] = 100
dict_ = {key: 'foo'}
self.assertEqual(dict_[key], 'foo')
def test_dataset_to_chunks_whole(self):
dataset = xarray.Dataset({'foo': ('x', np.arange(6))})
expected = [(xarray_beam.ChunkKey({'x': 0}), dataset)]
actual = (
test_util.EagerPipeline()
| xarray_beam.DatasetToChunks(dataset, chunks={'x': -1})
)
self.assertIdenticalChunks(actual, expected)
actual = (
test_util.EagerPipeline()
| xarray_beam.DatasetToChunks(dataset, chunks={})
)
self.assertIdenticalChunks(actual, expected)
def test_dataset_to_chunks_multiple(self):
dataset = xarray.Dataset({'foo': ('x', np.arange(6))})
expected = [
(xarray_beam.ChunkKey({'x': 0}), dataset.head(x=3)),
(xarray_beam.ChunkKey({'x': 3}), dataset.tail(x=3)),
]
actual = (
test_util.EagerPipeline()
| xarray_beam.DatasetToChunks(dataset.chunk({'x': 3}))
)
self.assertIdenticalChunks(actual, expected)
actual = (
test_util.EagerPipeline()
| xarray_beam.DatasetToChunks(dataset.chunk({'x': 3}), num_threads=2)
)
self.assertIdenticalChunks(actual, expected)
actual = (
test_util.EagerPipeline()
| xarray_beam.DatasetToChunks(dataset, chunks={'x': 3})
)
self.assertIdenticalChunks(actual, expected)
def test_to_slices(self):
key = xarray_beam.ChunkKey({'x': 0, 'y': 10})
expected = {'x': slice(0, 5, 1), 'y': slice(10, 20, 1)}
slices = key.to_slices({'x': 5, 'y': 10})
self.assertEqual(slices, expected)
slices = key.to_slices({'x': 5, 'y': 10, 'extra_key': 100})
self.assertEqual(slices, expected)
expected = {'x': slice(None), 'y': slice(10, 20, 1)}
slices = key.to_slices({'y': 10})
self.assertEqual(slices, expected)
with self.assertRaisesRegex(ValueError, 'non-zero offset'):
key.to_slices({'x': 5})
def test_iter_chunk_keys(self):
actual = sorted(core.iter_chunk_keys({'x': (3, 3), 'y': (2, 2, 2)}))
expected = [
xarray_beam.ChunkKey({'x': 0, 'y': 0}),
xarray_beam.ChunkKey({'x': 0, 'y': 2}),
xarray_beam.ChunkKey({'x': 0, 'y': 4}),
xarray_beam.ChunkKey({'x': 3, 'y': 0}),
xarray_beam.ChunkKey({'x': 3, 'y': 2}),
xarray_beam.ChunkKey({'x': 3, 'y': 4}),
]
self.assertEqual(actual, expected)
def test_repr(self):
key = xarray_beam.ChunkKey({'x': 0, 'y': 10})
expected = "ChunkKey({'x': 0, 'y': 10})"
self.assertEqual(repr(key), expected)
