def test_windowed_stateful_pass_callable():
data = np.array([2, 4, 6], dtype='float')
assert triple(data, 3, 3, data) == (9, data)
call = windowed_stateful_pass(value=triple, return_callable=True)
ret = call(data, 2)
assert_allclose(ret, [np.nan, 12, 18])
def test_windowed_stateful_pass_val_out():
data = np.array(range(95), dtype='float')**3
assert out_flip(data, 1, 1, data) == (-1, data)
ret = windowed_stateful_pass(data, 9, value=out_flip)
assert np.isnan(ret[0:8]).all()
assert_allclose(ret[9:], data[:-9] * -1)
def test_windowed_stateful_pass_respects_value():
data = np.array(range(101), dtype='float')**3
assert double(data, 1, 2, data) == (2, data)
ret = windowed_stateful_pass(data, 7, value=double)
assert np.isnan(ret[0:6]).all()
assert_allclose(ret[6:], data[6:] * 2)
def test_windowed_stateful_pass_raw():
data = np.array(range(11), dtype='float')**3
# not using jit, falling back to "value" function
# defined in windowed_stateful_pass module
raw = windowed_stateful_pass(data, 3)
assert np.isnan(raw[0:2]).all()
assert_allclose(raw[2:], data[2:])
def test_windowed_stateful_pass_constant():
data = np.array(range(50), dtype='float')
mean_result = mean_range(data, 66, 66, data)
assert mean_result[0] == 3.0
assert_allclose(mean_result[1], np.array([1, 2, 3, 4, 5], dtype='float'))
ret = windowed_stateful_pass(data, 4, value=mean_range)
assert np.isnan(ret[0:3]).all()
assert_allclose(ret[3:], 3.0)
def test_windowed_stateful_pass_rolling_mean():
data = np.array([1, 2, 3, 5, 8, 13], dtype='float')
# testing rolling_mean state handling
empty_state = np.empty(0, dtype=data.dtype)
mean, state = rolling_mean(data[:-1], 0, 0, empty_state)
expected_state = np.array(data[:-1].sum())
assert mean, state == (data[:-1].mean(), expected_state)
mean2, state2 = rolling_mean(data[1:], 13, 1, state)
expected_state_2 = np.array(data[1:].sum())
assert mean2, state2 == (data[1:].mean(), expected_state_2)
ret = windowed_stateful_pass(data, 2, value=rolling_mean)
assert np.isnan(ret[0])
assert_allclose(ret[1:], np.array([1.5, 2.5, 4.0, 6.5, 10.5]))