4 Script that constructs a graph in which hosts are nodes.
5 An edge between two hosts indicate that the hosts communicate.
6 Hosts are labeled and identified by their IPs.
7 The graph is written to a file in Graph Exchange XML format for later import and visual inspection in Gephi.
9 Update per February 2, 2018:
10 Extension of base_gefx_generator.py.
11 This script constructs a bipartite graph with IoT devices on one side and Internet hosts on the other side.
12 As a result, this graph does NOT show inter IoT device communication.
14 The input to this script is the JSON output by extract_from_tshark.py by Anastasia Shuba.
16 This script is a simplification of Milad Asgari's parser_data_to_gephi.py script.
17 It serves as a baseline for future scripts that want to include more information in the graph.
27 import parser.parse_dns
29 from networkx.algorithms import bipartite
32 DEVICE_MAC_LIST = "devicelist.dat"
33 EXCLUSION_MAC_LIST = "exclusion.dat"
34 COLUMN_MAC = "MAC_address"
35 COLUMN_DEVICE_NAME = "device_name"
37 JSON_KEY_SOURCE = "_source"
38 JSON_KEY_LAYERS = "layers"
39 JSON_KEY_FRAME = "frame"
40 JSON_KEY_FRAME_PROTOCOLS = "frame.protocols"
41 JSON_KEY_FRAME_TIME_EPOCH = "frame.time_epoch"
42 JSON_KEY_FRAME_LENGTH = "frame.len"
44 JSON_KEY_ETH_SRC = "eth.src"
45 JSON_KEY_ETH_DST = "eth.dst"
47 JSON_KEY_IP_SRC = "ip.src"
48 JSON_KEY_IP_DST = "ip.dst"
52 # List of checked protocols
53 listchkprot = [ "arp",
61 # Switch to generate graph that only shows local communication
62 ONLY_INCLUDE_LOCAL_COMMUNICATION = False
65 def create_device_list(dev_list_file):
66 """ Create list for smart home devices from a CSV file
68 dev_list_file: CSV file path that contains list of device MAC addresses
70 # Open the device MAC list file
71 with open(dev_list_file) as csvfile:
72 mac_list = csv.DictReader(csvfile, (COLUMN_MAC, COLUMN_DEVICE_NAME))
75 crude_list.append(item)
76 # Create key-value dictionary
78 for item in crude_list:
79 dev_list[item[COLUMN_MAC]] = item[COLUMN_DEVICE_NAME]
80 #print item["MAC_address"] + " => " + item["device_name"]
81 #for key, value in devlist.iteritems():
82 # print key + " => " + value
87 def traverse_and_merge_nodes(G, dev_list_file):
88 """ Merge nodes that have similar properties, e.g. same protocols
89 But, we only do this for leaves (outer nodes), and not for
90 nodes that are in the middle/have many neighbors.
91 The pre-condition is that the node:
92 (1) only has one neighbor, and
93 (2) not a smarthome device.
94 then we compare the edges, whether they use the same protocols
95 or not. If yes, then we collapse that node and we attach
96 it to the very first node that uses that set of protocols.
98 G: a complete networkx graph
99 dev_list_file: CSV file path that contains list of device MAC addresses
102 #print "Nodes: ", nodes
103 node_to_merge = dict()
104 # Create list of smarthome devices
105 dev_list = create_device_list(DEVICE_MAC_LIST)
106 # Traverse every node
107 # Check that the node is not a smarthome device
109 neighbors = G[node] #G.neighbors(node)
110 #print "Neighbors: ", neighbors, "\n"
111 # Skip if the node is a smarthome device
114 # Skip if the node has many neighbors (non-leaf) or no neighbor at all
115 if len(neighbors) is not 1:
117 #print "Node: ", node
118 neighbor = neighbors.keys()[0] #neighbors[0]
119 #print "Neighbor: ", neighbors
120 protocols = G[node][neighbor]['Protocol']
121 #print "Protocol: ", protocols
122 # Store neighbor-protocol as key in dictionary
123 neigh_proto = neighbor + "-" + protocols
124 if neigh_proto not in node_to_merge:
125 node_to_merge[neigh_proto] = node
127 # Merge this node if there is already an entry
130 node_to_merge_with = node_to_merge[neigh_proto]
131 merged_nodes = G.node[node_to_merge_with]['Merged']
132 # Check if this is the first node
133 if merged_nodes is '':
136 # Put comma if there is already one or more nodes
137 merged_nodes += ", " + node
138 # Then attach as attribute
139 G.node[node_to_merge_with]['Merged'] = merged_nodes
144 def place_in_graph(G, eth_src, eth_dst, device_dns_mappings, dev_list, layers,
145 edge_to_prot, edge_to_vol):
146 """ Place nodes and edges on the graph
148 G: the complete graph
149 eth_src: MAC address of source
150 eth_dst: MAC address of destination
151 device_dns_mappings: device to DNS mappings (data structure)
152 dev_list: list of existing smarthome devices
153 layers: layers of JSON file structure
154 edge_to_prot: edge to protocols mappings
155 edge_to_vol: edge to traffic volume mappings
157 # Get timestamp of packet (router's timestamp)
158 timestamp = Decimal(layers[JSON_KEY_FRAME][JSON_KEY_FRAME_TIME_EPOCH])
160 packet_len = Decimal(layers[JSON_KEY_FRAME][JSON_KEY_FRAME_LENGTH])
161 # Get the protocol and strip just the name of it
162 long_protocol = layers[JSON_KEY_FRAME][JSON_KEY_FRAME_PROTOCOLS]
163 # Split once starting from the end of the string and get it
164 split_protocol = long_protocol.split(':')
166 if len(split_protocol) < 5:
167 last_index = len(split_protocol) - 1
168 protocol = split_protocol[last_index]
170 protocol = split_protocol[3] + ":" + split_protocol[4]
171 #print "timestamp: ", timestamp, " - new protocol added: ", protocol, "\n"
172 # Store protocol into the set (source)
174 # Key to search in the dictionary is <src-mac-address>-<dst-mac_address>
175 dict_key = eth_src + "-" + eth_dst
176 if dict_key not in edge_to_prot:
177 edge_to_prot[dict_key] = set()
178 protocols = edge_to_prot[dict_key]
179 protocols.add(protocol)
180 protocols_str = ', '.join(protocols)
181 #print "protocols: ", protocols_str, "\n"
182 # Check packet length and accumulate to get traffic volume
183 if dict_key not in edge_to_vol:
184 edge_to_vol[dict_key] = 0;
185 edge_to_vol[dict_key] = edge_to_vol[dict_key] + packet_len
186 volume = str(edge_to_vol[dict_key])
187 # And source and destination IPs
188 ip_src = layers[JSON_KEY_IP][JSON_KEY_IP_SRC]
189 ip_dst = layers[JSON_KEY_IP][JSON_KEY_IP_DST]
190 # Categorize source and destination IP addresses: local vs. non-local
191 ip_re = re.compile(r'\b192.168.[0-9.]+')
192 src_is_local = ip_re.search(ip_src)
193 dst_is_local = ip_re.search(ip_dst)
195 # Skip device to cloud communication if we are interested in the local graph.
196 # TODO should this go before the protocol dict is changed?
197 if ONLY_INCLUDE_LOCAL_COMMUNICATION and not (src_is_local and dst_is_local):
200 #print "ip.src =", ip_src, "ip.dst =", ip_dst, "\n"
201 # Place nodes and edges
204 # Integer values used for tagging nodes, indicating to Gephi if they are local IoT devices or web servers.
207 # Values for the 'bipartite' attribute of a node when constructing the bipartite graph
209 bipartite_web_server = 1
211 G.add_node(eth_src, Name=dev_list[eth_src], islocal=local_node, bipartite=bipartite_iot)
215 # Check first if the key (eth_dst) exists in the dictionary
216 if eth_dst in device_dns_mappings:
217 # If the source is not local, then it's inbound traffic, and hence the eth_dst is the MAC of the IoT device.
218 hostname = device_dns_mappings[eth_dst].hostname_for_ip_at_time(ip_src, timestamp)
220 # Use IP if no hostname mapping
222 # Non-smarthome devices can be merged later
223 G.add_node(hostname, Merged='', islocal=remote_node, bipartite=bipartite_web_server)
227 G.add_node(eth_dst, Name=dev_list[eth_dst], islocal=local_node, bipartite=bipartite_iot)
231 # Check first if the key (eth_dst) exists in the dictionary
232 if eth_src in device_dns_mappings:
233 # If the destination is not local, then it's outbound traffic, and hence the eth_src is the MAC of the IoT device.
234 hostname = device_dns_mappings[eth_src].hostname_for_ip_at_time(ip_dst, timestamp)
236 # Use IP if no hostname mapping
238 # Non-smarthome devices can be merged later
239 G.add_node(hostname, Merged='', islocal=remote_node, bipartite=bipartite_web_server)
241 G.add_edge(src_node, dst_node, Protocol=protocols_str, Volume=volume)
244 def parse_json(file_path):
245 """ Parse JSON file and create graph
247 file_path: path to the JSON file
249 # Create a smart home device list
250 dev_list = create_device_list(DEVICE_MAC_LIST)
251 # Create an exclusion list
252 exc_list = create_device_list(EXCLUSION_MAC_LIST)
253 # First parse the file once, constructing a map that contains information about individual devices' DNS resolutions.
254 device_dns_mappings = parser.parse_dns.parse_json_dns(file_path) # "./json/eth1.dump.json"
257 # Mapping from edge to a set of protocols
258 edge_to_prot = dict()
259 # Mapping from edge to traffic volume
261 # Parse file again, this time constructing a graph of device<->server and device<->device communication.
262 with open(file_path) as jf:
263 # Read JSON; data becomes reference to root JSON object (or in our case json array)
265 # Loop through json objects (packets) in data
267 # p is a JSON object, not an index - drill down to object containing data from the different layers
268 layers = p[JSON_KEY_SOURCE][JSON_KEY_LAYERS]
271 for prot in listchkprot:
277 # Skip any non udp/non tcp traffic
278 if JSON_KEY_UDP not in layers and JSON_KEY_TCP not in layers:
281 # Fetch source and destination MACs
282 eth = layers.get(JSON_KEY_ETH, None)
284 print "[ WARNING: eth data not found ]"
286 eth_src = eth.get(JSON_KEY_ETH_SRC, None)
287 eth_dst = eth.get(JSON_KEY_ETH_DST, None)
288 # Exclude devices in the exclusion list
289 if eth_src in exc_list:
290 print "[ WARNING: Source ", eth_src, " is excluded from graph! ]"
292 if eth_dst in exc_list:
293 print "[ WARNING: Destination ", eth_dst, " is excluded from graph! ]"
296 # Place nodes and edges in graph
297 place_in_graph(G, eth_src, eth_dst, device_dns_mappings, dev_list, layers,
298 edge_to_prot, edge_to_vol)
300 # Print DNS mapping for reference
301 #for mac in device_dns_mappings:
302 # ddm = device_dns_mappings[mac]
303 # ddm.print_mappings()
308 # ------------------------------------------------------
309 # Not currently used.
310 # Might be useful later on if we wish to resolve IPs.
311 def get_domain(host):
312 ext_result = tldextract.extract(str(host))
313 # Be consistent with ReCon and keep suffix
314 domain = ext_result.domain + "." + ext_result.suffix
319 socket.inet_aton(addr)
323 # ------------------------------------------------------
326 if __name__ == '__main__':
327 if len(sys.argv) < 3:
328 print "Usage:", sys.argv[0], "input_file output_file"
329 print "outfile_file should end in .gexf"
331 # Input file: Path to JSON file generated from tshark JSON output using Anastasia's script (extract_from_tshark.py).
332 input_file = sys.argv[1]
333 print "[ input_file =", input_file, "]"
334 # Output file: Path to file where the Gephi XML should be written.
335 output_file = sys.argv[2]
336 print "[ output_file =", output_file, "]"
337 # Construct graph from JSON
338 G = parse_json(input_file)
339 # Contract nodes that have the same properties, i.e. same protocols
340 G = traverse_and_merge_nodes(G, DEVICE_MAC_LIST)
341 # Write Graph in Graph Exchange XML format
342 nx.write_gexf(G, output_file)