def station_count(self):
"""
Return data point's stations length
:return: list of stations size per hour
"""
stations = [0] * 24
for hour in self.data_file.buckets:
for pt in self.data_file.buckets[hour]:
try:
pt_data = json.loads(pt['data'])
if 'stations' in pt_data:
stations[hour] = len(pt_data['stations'])
except ValueError:
logger.error(f"Invalid json string: {pt['data']}")
raise ValueError
logger.info(f"stations: {stations}")
return stations
def read_data(filename):
"""
Read data in the file
:param filename:
:return: list of data sorted by timestamp
"""
result = []
with open(filename, 'r') as json_file:
json_list = list(json_file)
for json_str in json_list:
try:
result.append(json.loads(json_str))
except ValueError:
logger.error(f"Invalid json string in {filename}: {json_str}")
raise ValueError
sorted_data = sorted(result, key=lambda x: x['timestamp'])
return sorted_data
def avg_upd_not_dropped(self):
"""
Calculates average time difference of updates that are not dropped
:return:
"""
sum_upd = 0
count = 0
for hour in self.data_file.buckets:
for i in range(len(self.data_file.buckets[hour]) - 1):
if not is_dropped(self.data_file.buckets[hour][i + 1]['timestamp'],
self.data_file.buckets[hour][i]['timestamp']):
sum_upd += self.data_file.buckets[hour][i + 1]['timestamp'] - \
self.data_file.buckets[hour][i]['timestamp']
count += 1
if count == 0:
logger.error("every update is dropped")
return 1.0
average = round(sum_upd / count, 2)
logger.info(f"average upd duration if not dropped:{average}s")
return average
def latest_dr_ver(self):
"""
Latest driver version per hour
:return: list of driver versions per hour
"""
dr_ver = [None] * 24
no_dr_ver = []
for hour in self.data_file.buckets:
for pt in self.data_file.buckets[hour]:
try:
pt_data = json.loads(pt['data'])
if 'dr_ver' in pt_data:
dr_ver[hour] = pt_data['dr_ver']
except ValueError:
logger.error(f"Invalid json string: {pt['data']}")
raise ValueError
if dr_ver[hour] is None:
no_dr_ver.append(hour)
if len(no_dr_ver) != 0:
logger.info(f"no driver version in hours: {no_dr_ver}")
logger.info(f"driver version: {dr_ver}")
return dr_ver
def neighbor_count(self):
"""
Return data point's neighbors length
:return: list of neighbors size per hour
"""
neighbors = [0] * 24
prev = 0
for hour in self.data_file.buckets:
for pt in self.data_file.buckets[hour]:
if pt['type'] == "slow":
try:
pt_data = json.loads(pt['data'])
neighbors_count = len(pt_data['neighbors'])
neighbors[hour] = neighbors_count
prev = neighbors_count
except ValueError:
logger.error(f"Invalid json string: {pt['data']}")
raise ValueError
except KeyError:
logger.error(f"slow update at {pt['timestamp']} does not have neighbors")
if neighbors[hour] == 0:
neighbors[hour] = prev
logger.info(f"neighbors: {neighbors}")
return neighbors