def register_device():
patient_id = request.json['patient_id']
spimo_id = request.json['spimo_id']
pedal_id = request.json['pedal_id']
date_time = datetime.datetime.now()
connection = get_connection()
table_name = 'dp_pair'
data = {'device_id': spimo_id}
row = read_data(table_name, data)
#scenario 2: No such device_id
if row is None:
close_connection(connection)
return str(2)
#scenario 1: Successful
elif row.patient_id is None:
data = {'patient_id': patient_id, 'updated_date': date_time}
condition = {'device_id': spimo_id}
update_data(table_name, data, condition)
condition = {'device_id': pedal_id}
update_data(table_name, data, condition)
close_connection(connection)
return str(1)
#scenario 2: duplicated id
else:
close_connection(connection)
return str(0)
def return_device():
patient_id = request.json['patient_id']
device_id = request.json['device_id']
date_time = datetime.datetime.now()
table_name = 'dp_pair'
row = read_data(table_name, data)
# device has been returned
if row.patient_id is None:
return str(0)
else:
data = {'patient_id': None}
condition = {'device_id': device_id}
update_data(table_name, data, condition)
def handle_audio():
api_result = json.loads(request.form.get('data'))
seq = api_result["seq"]
device_id = api_result["device_id"]
session_id = api_result["session_id"]
audiofile = request.files['audio']
filename = werkzeug.utils.secure_filename(audiofile.filename)
audiofile.save(filename)
if filename is not None:
signal, rate = sf.read(filename)
wav_length = len(signal) / rate #second
window_length = 1
# for index in range(sample_per_wav): # take ten windows in an audio file
window_seed = random.uniform(0, abs(wav_length - window_length))
window_range = (window_seed, window_seed + window_length)
window_signal = signal[int(window_range[0] *
rate):int(window_range[1] * rate)]
mfcc_feat = mfcc(window_signal)
mfcc_feat_scale = mfcc_feat / np.linalg.norm(mfcc_feat)
mfcc_feat_scale = mfcc_feat_scale[np.newaxis, :]
print(mfcc_feat_scale.shape)
# result_array=(mfcc_feat-np.min(mfcc_feat))/np.ptp(mfcc_feat)
model = tensorflow.keras.models.load_model('./model')
predictions = model.predict(mfcc_feat_scale)
print(predictions)
into_sad_prob = predictions[0][0]
into_angry_prob = predictions[0][1]
into_happy_prob = predictions[0][2]
into_neutral_prob = predictions[0][3]
table_name = 'transaction_table'
data = {
'session_id': session_id,
'seq': seq,
'into_sad_prob': str(into_sad_prob),
'into_angry_prob': str(into_angry_prob),
'into_happy_prob': str(into_happy_prob),
'into_neutral_prob': str(into_neutral_prob)
}
r_data = {'session_id': session_id, 'seq': seq}
connection = get_connection()
row, number_rows = read_data(table_name, r_data, connection)
if number_rows == 0:
connection = insert_data(table_name, data, connection)
else:
condition = {'session_id': session_id, 'seq': seq}
connection = update_data(table_name, data, condition, connection)
return str(1)
else:
return str(0)
def device_refresh():
patient_id = request.args.get('patient_id')
print(patient_id)
device_id = request.args.get('device_id')
print(device_id)
connection = get_connection()
table_name = 'dp_pair'
data = {'device_id': device_id, 'patient_id': patient_id}
row = read_data(table_name, data)
start_time = row.start_time
current_time = datetime.datetime.now()
if (device_id[0:3] == 'SPI'):
table_name = 'spimo'
elif (device_id[0:3] == 'PED'):
table_name = 'pedal'
row = read_count(table_name, start_time, current_time, device_id)
result = row.total_count
close_connection(connection)
return str(result)
def handle_ml():
session_id = request.json['session_id']
seq = request.json['seq']
table_name = 'transaction_table'
connection = get_connection()
text_senti_avg, text_senti_std, text_senti_min, text_senti_max = calculate_features(
table_name, session_id, seq, connection)
condition = {'session_id': session_id, 'seq': seq}
features_data = {
'text_senti_avg': text_senti_avg,
'text_senti_std': text_senti_std,
'text_senti_min': text_senti_min,
'text_senti_max': text_senti_max
}
update_data(table_name, features_data, condition, connection)
r_data = {'session_id': session_id, 'seq': seq}
rows, number_rows = read_data(table_name, r_data, connection)
emotion_label = send_request_to_ml(rows)
features_data = {'result': emotion_label}
update_data(table_name, features_data, condition, connection)
close_connection(connection)
return emotion_label
def return_device():
patient_id = request.json['patient_id']
spimo_id = request.json['spimo_id']
pedal_id = request.json['pedal_id']
date_time = datetime.datetime.now()
connection = get_connection()
table_name = 'dp_pair'
data = {'device_id': spimo_id}
row = read_data(table_name, data)
# device has been returned
if row.patient_id is None:
close_connection(connection)
return str(0)
else:
data = {'patient_id': None, 'updated_date': None}
condition = {'device_id': spimo_id}
update_data(table_name, data, condition)
condition = {'device_id': pedal_id}
update_data(table_name, data, condition)
close_connection(connection)
return str(1)