def _hw_ai_callback(self):
# TODO: Get lock?
for name, b in self._buffers['hw_ai'].items():
t0_sample = b._total_samples_read
samples = b.read()[0] / b._sf
if b._discarded < b._discard:
to_discard = min(samples.shape[-1], b._discard)
b._discarded += to_discard
t0_sample -= to_discard
log.info(
'Discarding %.0f samples to compensate for AI filter delay',
to_discard)
samples = samples[to_discard:]
if len(samples):
b._total_samples_read += len(samples)
data = InputData(samples,
metadata={
't0_sample': t0_sample,
'fs': b.fs
})
for channel_name, cb in self._callbacks.get('ai', []):
if channel_name == name:
cb(data)
def test_input_data():
expected_array = np.concatenate((
np.zeros(shape=5),
np.ones(shape=5),
np.random.uniform(size=5),
))
expected_metadata = {'t0': 0}
expected_input = InputData(expected_array, expected_metadata)
input_data = [
InputData(expected_array[0:5], metadata=expected_metadata),
InputData(expected_array[5:10], metadata=expected_metadata),
InputData(expected_array[10:15], metadata=expected_metadata),
]
result = concatenate(input_data)
assert expected_input.metadata == result.metadata
assert np.array_equal(expected_input, result)
def test_pipeline(data, pipeline):
expected = []
for i in range(8):
d = InputData(np.random.uniform(size=5), {'n': 5})
pipeline.send(d)
expected.append(d[np.newaxis])
expected = concatenate(expected, axis=-1).reshape((-1, 10))
assert len(data) == 1
assert data[0].shape == (4, 10)
assert np.array_equal(expected, data[0])
assert data[0].metadata == {'n': 5, 'block_size': 10}
def test_slice():
r = np.random.uniform(size=100)
d = InputData(r, {'t0_sample': 10, 'fs': 100})
assert d[::2].metadata == {'t0_sample': 10, 'fs': 50}
assert d[::4].metadata == {'t0_sample': 10, 'fs': 25}
assert d[:4:].metadata == {'t0_sample': 10, 'fs': 100}
assert d[1::2].metadata == {'t0_sample': 11, 'fs': 50}
assert d[10::2].metadata == {'t0_sample': 20, 'fs': 50}
assert d[10] == r[10]
assert np.all(d[:10] == r[:10])
assert d[-10::2].metadata == {'t0_sample': 100, 'fs': 50}
assert d[-20::2].metadata == {'t0_sample': 90, 'fs': 50}
assert d[-20::].metadata == {'t0_sample': 90, 'fs': 100}
assert d[-20::-2].metadata == {'t0_sample': 90, 'fs': -50}
def test_pipeline_flush(data, pipeline):
expected = []
for i in range(16):
d = InputData(np.random.uniform(size=5), {'n': 5})
pipeline.send(d)
expected.append(d[np.newaxis])
if i == 6:
pipeline.send(Ellipsis)
expected = []
expected = concatenate(expected[:8], axis=-1).reshape((-1, 10))
assert len(data) == 1
assert data[0].shape == (4, 10)
assert np.array_equal(expected, data[0])
assert data[0].metadata == expected.metadata