def generate_flow_plot_data(input_file,
home_ip,
website='youtube.com',
is_incoming=True):
stream_set = set()
closed_set = set()
start_timestamp = None
connection_list = GaplessList()
with open(input_file, 'rb') as csv_file:
data_reader = csv.DictReader(csv_file, delimiter=',')
for row in data_reader:
# set start_timestamp if not set
start_timestamp = float(
row['timestamp']
) if start_timestamp is None else start_timestamp
if website in row['website'] and row['protocol'] == 'TCP':
if row['dst' if is_incoming else 'src'] == home_ip:
stream_id = make_stream_id(row)
if row['is_rst'] == 'False' and row[
'is_fin'] == 'False' and stream_id not in closed_set:
stream_set.add(stream_id)
else:
stream_set.discard(stream_id)
closed_set.add(stream_id)
packet_time = int(
float(row['timestamp']) - start_timestamp)
connection_list.set_element(packet_time, len(stream_set))
return connection_list.get_list()
def generate_bitrate_plot_data(input_file, home_ip, website='youtube.com', is_incoming=True, protocols=None):
if protocols is None:
protocols = ['TCP']
start_timestamp = None
byte_rate_list = GaplessList(fill=0)
with open(input_file, 'rb') as csv_file:
data_reader = csv.DictReader(csv_file, delimiter=',')
for row in data_reader:
# set start_timestamp if not set
start_timestamp = float(row['timestamp']) if start_timestamp is None else start_timestamp
if website in row['website'] and (row['protocol'] in protocols):
if row['dst' if is_incoming else 'src'] == home_ip:
packet_time = int(float(row['timestamp']) - start_timestamp)
byte_rate_list.increment_element(packet_time, int(row['len']))
return byte_rate_list.get_list()
def generate_flow_plot_data(input_file, home_ip, website="youtube.com", is_incoming=True):
stream_set = set()
closed_set = set()
start_timestamp = None
connection_list = GaplessList()
with open(input_file, "rb") as csv_file:
data_reader = csv.DictReader(csv_file, delimiter=",")
for row in data_reader:
# set start_timestamp if not set
start_timestamp = float(row["timestamp"]) if start_timestamp is None else start_timestamp
if website in row["website"] and row["protocol"] == "TCP":
if row["dst" if is_incoming else "src"] == home_ip:
stream_id = make_stream_id(row)
if row["is_rst"] == "False" and row["is_fin"] == "False" and stream_id not in closed_set:
stream_set.add(stream_id)
else:
stream_set.discard(stream_id)
closed_set.add(stream_id)
packet_time = int(float(row["timestamp"]) - start_timestamp)
connection_list.set_element(packet_time, len(stream_set))
return connection_list.get_list()
def generate_bitrate_plot_data(input_file,
home_ip,
website='youtube.com',
is_incoming=True,
protocols=None):
if protocols is None:
protocols = ['TCP']
start_timestamp = None
byte_rate_list = GaplessList(fill=0)
with open(input_file, 'rb') as csv_file:
data_reader = csv.DictReader(csv_file, delimiter=',')
for row in data_reader:
# set start_timestamp if not set
start_timestamp = float(
row['timestamp']
) if start_timestamp is None else start_timestamp
if website in row['website'] and (row['protocol'] in protocols):
if row['dst' if is_incoming else 'src'] == home_ip:
packet_time = int(
float(row['timestamp']) - start_timestamp)
byte_rate_list.increment_element(packet_time,
int(row['len']))
return byte_rate_list.get_list()
