def test_ffill_bfill_dask(method):
da, _ = make_interpolate_example_data((40, 40), 0.5)
da = da.chunk({"x": 5})
dask_method = getattr(da, method)
numpy_method = getattr(da.compute(), method)
# unchunked axis
with raise_if_dask_computes():
actual = dask_method("time")
expected = numpy_method("time")
assert_equal(actual, expected)
# chunked axis
with raise_if_dask_computes():
actual = dask_method("x")
expected = numpy_method("x")
assert_equal(actual, expected)
# with limit
with raise_if_dask_computes():
actual = dask_method("time", limit=3)
expected = numpy_method("time", limit=3)
assert_equal(actual, expected)
# limit < axis size
with pytest.raises(NotImplementedError):
actual = dask_method("x", limit=2)
# limit > axis size
with raise_if_dask_computes():
actual = dask_method("x", limit=41)
expected = numpy_method("x", limit=41)
assert_equal(actual, expected)
def test_skipna_broadcast_weights_assignment_destination(
a_rand_nan, b_rand_nan, weights_lonlat, metric):
"""Tests that 'assignment destination is read-only' is not raised
https://github.com/xarray-contrib/xskillscore/issues/79"""
with raise_if_dask_computes():
metric(a_rand_nan,
b_rand_nan, ["lat", "lon"],
weights=weights_lonlat,
skipna=True)
def test_skipna_returns_same_value_as_dropped_pairwise_nans(
a_1d_fixed_nan, b_1d_fixed_nan, metric):
"""Tests that DataArrays with pairwise nans return the same result
as the same two with those nans dropped."""
a_dropped, b_dropped, _ = drop_nans(a_1d_fixed_nan, b_1d_fixed_nan)
with raise_if_dask_computes():
res_with_nans = metric(a_1d_fixed_nan,
b_1d_fixed_nan,
"time",
skipna=True)
res_dropped_nans = metric(a_dropped, b_dropped, "time")
assert_allclose(res_with_nans, res_dropped_nans)
def test_skipna_returns_nan_when_false(a_1d_fixed_nan, b_1d_fixed_nan, metric):
"""Tests that nan is returned if there's any nans in the time series
and skipna is False."""
with raise_if_dask_computes():
res = metric(a_1d_fixed_nan, b_1d_fixed_nan, "time", skipna=False)
assert np.isnan(res).all()
def test_nan_skipna(a, b):
# Randomly add some nans to a
a = a.where(np.random.random(a.shape) < 0.5)
with raise_if_dask_computes():
pearson_r(a, b, dim="lat", skipna=True)