def TfExDataSourceConstructor(*args):
"""Select a EEG or ECG/EKG TfExampleDataSource instance, given by FLAGS.file_type.
Args:
*args: arguments passed to the constructor
Returns:
TfExampleEegDataSource or TfExampleEkgDataSource instance
"""
if FLAGS.file_type == 'EEG':
return data_source.TfExampleEegDataSource(*args)
elif FLAGS.file_type == 'ECG' or FLAGS.file_type == 'EKG':
return data_source.TfExampleEkgDataSource(*args)
else:
logging.warning('Unknown file type %s, using EEG', FLAGS.file_type)
return data_source.TfExampleEegDataSource(*args)
def setUp(self):
super(WaveformDataServiceTest, self).setUp()
tf_ex = tf.train.Example()
feature = tf_ex.features.feature
feature['eeg_channel/num_samples'].int64_list.value.append(20)
feature['eeg_channel/sampling_frequency_hz'].float_list.value.append(1.0)
for i in range(20):
feature['test_feature'].float_list.value.append(i*2)
for i in range(20):
feature[('eeg_channel/EEG '
'test_sub-REF/samples')].float_list.value.append(i+1)
for i in range(20):
feature[('eeg_channel/EEG '
'test_min-REF/samples')].float_list.value.append(1)
self.waveform_data_source = data_source.TfExampleEegDataSource(tf_ex,
'test_key')
def setUp(self):
super(PredictionDataServiceTest, self).setUp()
tf_ex = tf.train.Example()
feature = tf_ex.features.feature
feature['eeg_channel/num_samples'].int64_list.value.append(20)
feature['eeg_channel/sampling_frequency_hz'].float_list.value.append(
1.0)
for i in range(20):
feature['test_feature'].float_list.value.append(i * 2)
for i in range(20):
feature[('eeg_channel/EEG '
'test_sub-REF/samples')].float_list.value.append(i + 1)
for i in range(20):
feature[('eeg_channel/EEG '
'test_min-REF/samples')].float_list.value.append(1)
time = timestamp_pb2.Timestamp()
time.seconds = 10
time.nanos = 800000000
feature['start_time'].bytes_list.value.append(time.SerializeToString())
feature['segment/patient_id'].bytes_list.value.append(b'test patient')
waveform_data_source = data_source.TfExampleEegDataSource(
tf_ex, 'test key')
pred_outputs = prediction_output_pb2.PredictionOutputs()
pred_output = pred_outputs.prediction_output.add()
pred_output.chunk_info.chunk_id = 'test chunk'
pred_output.chunk_info.chunk_start_time.seconds = 10
pred_output.chunk_info.chunk_start_time.nanos = 800000000
pred_output.chunk_info.chunk_size_sec = 2
label = pred_output.label.add()
label.name = 'test label'
pred_output_2 = pred_outputs.prediction_output.add()
pred_output_2.chunk_info.chunk_id = 'test chunk 2'
pred_output_2.chunk_info.chunk_start_time.seconds = 12
pred_output_2.chunk_info.chunk_start_time.nanos = 800000000
pred_output_2.chunk_info.chunk_size_sec = 2
label = pred_output_2.label.add()
label.name = 'test label'
self._pred = prediction_data_service.PredictionDataService(
pred_outputs, waveform_data_source, 100)
