import tldextract
import networkx as nx
import sys
+import csv
+import re
from decimal import *
import parser.parse_dns
+# List of devices
+DEVICE_MAC_LIST = "devicelist.dat"
+EXCLUSION_MAC_LIST = "exclusion.dat"
+COLUMN_MAC = "MAC_address"
+COLUMN_DEVICE_NAME = "device_name"
+# Fields
+JSON_KEY_SOURCE = "_source"
+JSON_KEY_LAYERS = "layers"
+JSON_KEY_FRAME = "frame"
+JSON_KEY_FRAME_PROTOCOLS = "frame.protocols"
+JSON_KEY_FRAME_TIME_EPOCH = "frame.time_epoch"
+JSON_KEY_ETH = "eth"
JSON_KEY_ETH_SRC = "eth.src"
JSON_KEY_ETH_DST = "eth.dst"
+JSON_KEY_IP = "ip"
+JSON_KEY_IP_SRC = "ip.src"
+JSON_KEY_IP_DST = "ip.dst"
+# Checked protocols
+JSON_KEY_UDP = "udp"
+JSON_KEY_TCP = "tcp"
+# List of checked protocols
+listchkprot = [ "arp",
+ "bootp",
+ "dhcpv6",
+ "dns",
+ "llmnr",
+ "mdns",
+ "ssdp" ]
+
+
+def create_device_list(dev_list_file):
+ """ Create list for smart home devices from a CSV file
+ Args:
+ dev_list_file: CSV file path that contains list of device MAC addresses
+ """
+ # Open the device MAC list file
+ with open(dev_list_file) as csvfile:
+ mac_list = csv.DictReader(csvfile, (COLUMN_MAC, COLUMN_DEVICE_NAME))
+ crude_list = list()
+ for item in mac_list:
+ crude_list.append(item)
+ # Create key-value dictionary
+ dev_list = dict()
+ for item in crude_list:
+ dev_list[item[COLUMN_MAC]] = item[COLUMN_DEVICE_NAME]
+ #print item["MAC_address"] + " => " + item["device_name"]
+ #for key, value in devlist.iteritems():
+ # print key + " => " + value
+
+ return dev_list
+
def parse_json(file_path):
- device_dns_mappings = parser.parse_dns.parse_json_dns("./json/dns.json")
+ # Create a smart home device list
+ dev_list = create_device_list(DEVICE_MAC_LIST)
+ # Create an exclusion list
+ exc_list = create_device_list(EXCLUSION_MAC_LIST)
+ # First parse the file once, constructing a map that contains information about individual devices' DNS resolutions.
+ device_dns_mappings = parser.parse_dns.parse_json_dns(file_path) # "./json/eth1.dump.json"
+
# Init empty graph
- G = nx.DiGraph()
+ G = nx.DiGraph()
+ # Mapping from node to a set of protocols
+ edge_to_prot = dict()
+
+ # Parse file again, this time constructing a graph of device<->server and device<->device communication.
with open(file_path) as jf:
# Read JSON.
# data becomes reference to root JSON object (or in our case json array)
data = json.load(jf)
- # Loop through json objects in data
- for k in data:
- # Fetch timestamp of packet
- packet_timestamp = Decimal(data[k]["ts"])
- # Fetch eth source and destination info
- eth_src = data[k][JSON_KEY_ETH_SRC]
- eth_dst = data[k][JSON_KEY_ETH_DST]
- # Traffic can be both outbound and inbound.
- # Determine which one of the two by looking up device MAC in DNS map.
- iot_device = None
- if eth_src in device_dns_mappings:
- iot_device = eth_src
- elif eth_dst in device_dns_mappings:
- iot_device = eth_dst
- else:
- print "[ WARNING: DNS mapping not found for device with MAC", eth_src, "OR", eth_dst, "]"
- # This must be local communication between two IoT devices OR an IoT device talking to a hardcoded IP.
- # For now let's assume local communication.
- # Add a node for each device and an edge between them.
- G.add_node(eth_src)
- G.add_node(eth_dst)
- G.add_edge(eth_src, eth_dst)
- # TODO add regex check on src+dst IP to figure out if hardcoded server IP (e.g. check if one of the two are NOT a 192.168.x.y IP)
+
+ # Loop through json objects (packets) in data
+ for p in data:
+ # p is a JSON object, not an index
+ # Drill down to object containing data from the different layers
+ layers = p[JSON_KEY_SOURCE][JSON_KEY_LAYERS]
+
+ iscontinue = False
+ for prot in listchkprot:
+ if prot in layers:
+ iscontinue = True
+ if iscontinue:
+ continue
+
+ # Skip any non udp/non tcp traffic
+ if JSON_KEY_UDP not in layers and JSON_KEY_TCP not in layers:
continue
- # It is outbound traffic if iot_device matches src, otherwise it must be inbound traffic.
- outbound_traffic = iot_device == eth_src
- ''' Graph construction '''
- # No need to check if the Nodes and/or Edges we add already exist:
- # NetworkX won't add already existing nodes/edges (except in the case of a MultiGraph or MultiDiGraph (see NetworkX doc)).
-
- # Add a node for each host.
- # First add node for IoT device.
- G.add_node(iot_device)
- # Then add node for the server.
- # For this we need to distinguish between outbound and inbound traffic so that we look up the proper IP in our DNS map.
- # For outbound traffic, the server's IP is the destination IP.
- # For inbound traffic, the server's IP is the source IP.
- server_ip = data[k]["dst_ip"] if outbound_traffic else data[k]["src_ip"]
- hostname = device_dns_mappings[iot_device].hostname_for_ip_at_time(server_ip, packet_timestamp)
- if hostname is None:
- # TODO this can occur when two local devices communicate OR if IoT device has hardcoded server IP.
- # However, we only get here for the DNS that have not performed any DNS lookups
- # We should use a regex check early in the loop to see if it is two local devices communicating.
- # This way we would not have to consider these corner cases later on.
- print "[ WARNING: no ip-hostname mapping found for ip", server_ip, " -- adding eth.src->eth.dst edge, but note that this may be incorrect if IoT device has hardcoded server IP ]"
- G.add_node(eth_src)
- G.add_node(eth_dst)
- G.add_edge(eth_src, eth_dst)
+ # Fetch source and destination MACs
+ eth = layers.get(JSON_KEY_ETH, None)
+ if eth is None:
+ print "[ WARNING: eth data not found ]"
+ continue
+ eth_src = eth.get(JSON_KEY_ETH_SRC, None)
+ eth_dst = eth.get(JSON_KEY_ETH_DST, None)
+ # Exclude devices in the exclusion list
+ if eth_src in exc_list:
+ print "[ WARNING: Source ", eth_src, " is excluded from graph! ]"
continue
- G.add_node(hostname)
- # Connect the two nodes we just added.
- if outbound_traffic:
- G.add_edge(iot_device, hostname)
+ if eth_dst in exc_list:
+ print "[ WARNING: Destination ", eth_dst, " is excluded from graph! ]"
+ continue
+
+ # Fetch timestamp of packet (router's timestamp)
+ timestamp = Decimal(layers[JSON_KEY_FRAME][JSON_KEY_FRAME_TIME_EPOCH])
+ # Get the protocol and strip just the name of it
+ long_protocol = layers[JSON_KEY_FRAME][JSON_KEY_FRAME_PROTOCOLS]
+ # Split once starting from the end of the string and get it
+ #protocol = long_protocol.rsplit(':', 1)[1]
+ split_protocol = long_protocol.split(':')
+ protocol = None
+ if len(split_protocol) < 5:
+ last_index = len(split_protocol) - 1
+ protocol = split_protocol[last_index]
+ else:
+ protocol = split_protocol[3] + ":" + split_protocol[4]
+ print "timestamp: ", timestamp, " - new protocol added: ", protocol, "\n"
+
+ # Store protocol into the set (source)
+ protocols = None
+ # Key to search for protocol list in the dictionary is
+ # <src-mac-address>-<dst-mac_address>
+ protocol_key = eth_src + "-" + eth_dst
+ if protocol_key not in edge_to_prot:
+ edge_to_prot[protocol_key] = set()
+ protocols = edge_to_prot[protocol_key]
+ protocols.add(protocol)
+ protocols_str = ', '.join(protocols)
+ print "protocols: ", protocols_str, "\n"
+ # And source and destination IPs
+ ip_src = layers[JSON_KEY_IP][JSON_KEY_IP_SRC]
+ ip_dst = layers[JSON_KEY_IP][JSON_KEY_IP_DST]
+
+ # Categorize source and destination IP addresses: local vs. non-local
+ ipre = re.compile(r'\b192.168.[0-9.]+')
+ src_is_local = ipre.search(ip_src)
+ dst_is_local = ipre.search(ip_dst)
+ print "ip.src =", ip_src, "ip.dst =", ip_dst, "\n"
+
+ src_node = None
+ dst_node = None
+ if src_is_local:
+ G.add_node(eth_src, Name=dev_list[eth_src])
+ src_node = eth_src
else:
- G.add_edge(hostname, iot_device)
+ hostname = None
+ # Check first if the key (eth_dst) exists in the dictionary
+ if eth_dst in device_dns_mappings:
+ # If the source is not local, then it's inbound traffic, and hence the eth_dst is the MAC of the IoT device.
+ hostname = device_dns_mappings[eth_dst].hostname_for_ip_at_time(ip_src, timestamp)
+ if hostname is None:
+ # Use IP if no hostname mapping
+ hostname = ip_src
+ G.add_node(hostname)
+ src_node = hostname
+
+ if dst_is_local:
+ G.add_node(eth_dst, Name=dev_list[eth_dst])
+ dst_node = eth_dst
+ else:
+ hostname = None
+ # Check first if the key (eth_dst) exists in the dictionary
+ if eth_src in device_dns_mappings:
+ # If the destination is not local, then it's outbound traffic, and hence the eth_src is the MAC of the IoT device.
+ hostname = device_dns_mappings[eth_src].hostname_for_ip_at_time(ip_dst, timestamp)
+ if hostname is None:
+ # Use IP if no hostname mapping
+ hostname = ip_dst
+ G.add_node(hostname)
+ dst_node = hostname
+ G.add_edge(src_node, dst_node, Protocol=protocols_str)
+
+ # Print DNS mapping for reference
+ for mac in device_dns_mappings:
+ ddm = device_dns_mappings[mac]
+ ddm.print_mappings()
+
return G
+
# ------------------------------------------------------
# Not currently used.
# Might be useful later on if we wish to resolve IPs.
return False
# ------------------------------------------------------
+
if __name__ == '__main__':
if len(sys.argv) < 3:
print "Usage:", sys.argv[0], "input_file output_file"