While coding the session monitor a couple of weeks ago I developed a quick script which could query ntop for session information. Jaime started using it for graphing now, so I thought it might be useful to soembody.
import sgmllib, re, sys
import socket
from sets import Set
class MyParser(sgmllib.SGMLParser):
"A simple parser class."
def parse(self, s):
"Parse the given string 's'."
self.feed(s)
self.close()
def __init__(self, verbose=0):
"Initialise an object, passing 'verbose' to the superclass."
sgmllib.SGMLParser.__init__(self, verbose)
self.starting_description = 0
self.inside_table_element = 0
self.inside_table_row = 0
self.in_session = 0
self.td_count = 0
self.seen = 0
self.sessions = []
self.session_num = 0
self.queried_ip = ""
self.dst_port_group = {}
self.src_port_group = {}
self.source_ip = ""
self.source_port = ""
self.dest_ip = ""
self.dest_port = ""
self.data_sent = 0
self.data_rcvd = 0
self.active_since = ""
self.last_seen = ""
self.duration = 0
self.inactive = 0
def tosecs(self, data):
# example: 27 sec
if data.endswith('sec'):
data = data.split(' ', 1)[0]
# examples: 1 day 1:52:27
# 2 days 4:56:23
elif data.__contains__('day'):
result = re.findall('(d+) days? (S+)', data)
(days, time) = result[0]
seconds = int(days) * 86400
dt = time.split(':')
seconds += int(dt.pop())
try:
seconds += seconds + int(dt.pop()) * 60
seconds += seconds + int(dt.pop()) * 3600
except IndexError:
pass
data = str(seconds)
# example: 1:52:27
else:
dt = data.split(':')
seconds = int(dt.pop())
try:
seconds += int(dt.pop()) * 60
if len(dt) == 3:
seconds += int(dt.pop()) * 3600
except IndexError:
pass
data = str(seconds)
return data
def start_td(self, attributes):
self.td_count += 1
self.inside_table_element = 1
def end_td(self):
self.inside_table_element = 0
self.seen = 0
def start_tr(self, attributes):
self.inside_table_row = 1
def end_tr(self):
self.inside_table_row = 0
self.td_count = 0
if self.in_session and self.source_ip:
self.session_num += 1
tmp = "%s:%s --> %s:%s (%f %f) duration: %s" %
(self.source_ip,self.source_port,self.dest_ip,self.dest_port, self.data_sent, self.data_rcvd, self.duration)
self.sessions.append(tmp)
def start_a(self, attributes):
self.inside_table_row = 1
for name, value in attributes:
if name == "href" and self.in_session:
matches = re.findall("d+.d+.d+.d+",value)
if len(matches) > 0:
if self.td_count is 1:
self.source_ip = matches[0]
elif self.td_count is 2:
self.dest_ip = matches[0]
def handle_data(self, data):
if self.queried_ip is "":
matches = re.findall("(d+.d+.d+.d+)",data)
if len(matches) > 0:
self.queried_ip = matches[0]
if data.__contains__("Active TCP/UDP Sessions"):
self.in_session = 1
if data.__contains__("The color of the host"):
self.sessions.append("NumSessions:%d" % int(self.session_num))
self.in_session = 0
source_sessions = 0
dest_sessions = 0
for sess in self.sessions:
src_str = "^%s:.*" % self.queried_ip
dst_str = "^d+.d+.d+.d+:d+s+-->s+%s:.*" % self.queried_ip
src_sess = "^%s:d+s+-->s+(d+.d+.d+.d+):(d+)" % self.queried_ip
dst_sess = "^(d+.d+.d+.d+):d+s+-->s+%s:(d+)" % self.queried_ip
if re.findall(src_str, sess):
source_sessions += 1
if re.findall(dst_str, sess):
dest_sessions += 1
matches = re.findall(src_sess, sess)
if len(matches) > 0:
if matches[0][1] in self.src_port_group:
self.src_port_group[matches[0][1]].add(matches[0][0])
else:
self.src_port_group[matches[0][1]] = Set()
self.src_port_group[matches[0][1]].add(matches[0][0])
matches = re.findall(dst_sess, sess)
if len(matches) > 0:
if matches[0][1] in self.dst_port_group:
self.dst_port_group[matches[0][1]].add(matches[0][0])
else:
self.dst_port_group[matches[0][1]] = Set()
self.dst_port_group[matches[0][1]].add(matches[0][0])
self.sessions.append("SessionsAsSource:%s" % source_sessions)
self.sessions.append("SessionsAsDest:%s" % dest_sessions)
for port in self.src_port_group:
self.sessions.append("UniqPort%sAsSourceSessions:%d" % (port, len(self.src_port_group[port])))
for port in self.dst_port_group:
self.sessions.append("UniqPort%sAsDestSessions:%d" % (port, len(self.dst_port_group[port])))
if self.inside_table_element and self.in_session:
if self.td_count <= 2:
matches = re.findall(":(w+)",data)
if len(matches) > 0:
try:
port = socket.getservbyname(matches[0])
except:
port = matches[0]
if self.td_count is 1:
self.source_port = port
elif self.td_count is 2:
self.dest_port = port
elif self.td_count is 3:
if self.seen == 0:
self.data_sent = float(data)
self.seen = 1
else:
if data.__contains__("KB"):
self.data_sent *= 1024
elif data.__contains__("MB"):
self.data_sent *= 1024 * 1024
elif data.__contains__("GB"):
self.data_sent *= 1024 * 1024 * 1024
self.seen = 0
elif self.td_count is 4:
if self.seen == 0:
self.data_rcvd = float(data)
self.seen = 1
else:
if data.__contains__("KB"):
self.data_rcvd *= 1024
elif data.__contains__("MB"):
self.data_rcvd *= 1024 * 1024
elif data.__contains__("GB"):
self.data_rcvd *= 1024 * 1024 * 1024
self.seen = 0
elif self.td_count is 5:
self.active_since = data
elif self.td_count is 6:
self.last_seen = data
elif self.td_count is 7:
self.duration = self.tosecs(data)
elif self.td_count is 8:
self.inactive = self.tosecs(data)
def get_sessions(self):
return self.sessions
# Real start
import urllib, sgmllib
argv = sys.argv
try:
ntop = argv[1]
ip = argv[2]
except:
print "Prints out session information about a given host"
print "Usage: " + argv[0] + " ntop_ip host_ip"
sys.exit()
# Get something to work with.
f = None
try:
f = urllib.urlopen("http://" + ntop + ":3000/" + ip + ".html")
except:
print "Unable to connect to " + ntop
sys.exit()
s = f.read()
# Try and process the page.
# The class should have been defined first, remember.
myparser = MyParser()
myparser.parse(s)
a = myparser.get_sessions()
print "Sessions for " + ip
print "--------------------------"
for b in a:
print b
The output looks somewhat like this:
Gestalt:ntop dk$ python session_query.py 192.168.1.2 192.168.1.2 Sessions for 192.168.1.2 -------------------------- 192.168.1.43:2409 --> 192.168.1.2:3000 (567.000000 2662.400000) duration: 0 192.168.1.2:39492 --> 195.77.186.178:33800 (74240.000000 75264.000000) duration: 406 192.168.1.2:45206 --> 195.55.170.126:33800 (140595.200000 143974.400000) duration: 644 192.168.1.2:54151 --> 207.158.15.208:443 (1153433.600000 1153433.600000) duration: 637 192.168.1.35:59334 --> 192.168.1.2:3000 (258.000000 112.000000) duration: 0 NumSessions:5 SessionsAsSource:3 SessionsAsDest:2 UniqPort443AsSourceSessions:1 UniqPort33800AsSourceSessions:2 UniqPort3000AsDestSessions:2 Gestalt:ntop dk$
Aaah, and with a little trick also the main session page can be parsed:
Gestalt:ntop dk$ python session_query.py 192.168.1.2 NetNetstat Sessions for NetNetstat -------------------------- 192.168.1.43:2409 --> 192.168.1.2:3000 (567.000000 2662.400000) duration: 0 192.168.1.2:39492 --> 195.77.186.178:33800 (75366.400000 76288.000000) duration: 459 192.168.1.2:45206 --> 195.55.170.126:33800 (141721.600000 144896.000000) duration: 706 192.168.1.2:54151 --> 207.158.15.208:443 (1153433.600000 1153433.600000) duration: 695 192.168.1.35:59337 --> 192.168.1.2:3000 (257.000000 112.000000) duration: 0 NumSessions:5 SessionsAsSource:0 SessionsAsDest:1 UniqPort33800AsDestSessions:1 Gestalt:ntop dk$
As a finishing note, we did setup another blog with ossim-related info: Santi’s blog. As an opener he features a demo video with a sample presentation done to a customer.