From: rtrimana <rtrimana@uci.edu>
Date: Fri, 3 Nov 2017 15:33:55 +0000 (-0700)
Subject: A simple script to analyze the distribution of packet frequencies against time
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=bceb30b9aba7fa2d4df53e8196ee981f27d3cba4;p=pingpong.git

A simple script to analyze the distribution of packet frequencies against time
---

diff --git a/parse_packet_frequency.py b/parse_packet_frequency.py
new file mode 100644
index 0000000..813471a
--- /dev/null
+++ b/parse_packet_frequency.py
@@ -0,0 +1,106 @@
+#!/usr/bin/python
+
+"""
+Script that takes a file (output by wireshark/tshark, in JSON format) and analyze
+the traffic frequency of a certain device at a certain time.
+"""
+
+import sys
+import json
+from collections import defaultdict
+from dateutil import parser
+
+JSON_KEY_SOURCE = "_source"
+JSON_KEY_LAYERS = "layers"
+
+JSON_KEY_ETH = "eth"
+JSON_KEY_ETH_DST = "eth.dst"
+JSON_KEY_ETH_SRC = "eth.src"
+JSON_KEY_FRAME = "frame"
+JSON_KEY_FRAME_TIME = "frame.time"
+
+
+def save_to_file(tbl_header, dictionary, filename_out):
+    """ Show summary of statistics of PCAP file
+        Args:
+            tbl_header: header for the saved table
+            dictionary: dictionary to be saved
+            filename_out: file name to save
+    """
+    # Appending, not overwriting!
+    f = open(filename_out, 'a')
+    # Write the table header
+    f.write("\n\n" + str(tbl_header) + "\n");
+    # Iterate over dictionary and write (key, value) pairs
+    #for key, value in dictionary.iteritems():
+    for key in sorted(dictionary):
+        f.write(str(key) + ", " + str(dictionary[key]) + "\n")
+
+    f.close()
+    print "Writing output to file: ", filename_out
+
+
+def main():
+    """ Main function
+    """
+	if len(sys.argv) < 5:
+		print "Usage: python", sys.argv[0], "<input_file> <output_file> <device_name> <mac_address>"
+		return
+	# Parse the file for the specified MAC address
+	time_freq = parse_json(sys.argv[1], sys.argv[4])
+	# Write statistics into file
+	save_to_file(sys.argv[3], time_freq, sys.argv[2])
+	print "====================================================================="
+	for time in time_freq.keys():
+		print time, " => ", time_freq[time]
+	print "====================================================================="
+
+
+# Convert JSON file containing DNS traffic to a map in which a hostname points to its set of associated IPs.
+def parse_json(file_path, mac_address):
+    """ Show summary of statistics of PCAP file
+        Args:
+            file_path: path of the read file
+            mac_address: MAC address of a device to analyze
+    """
+	# Maps timestamps to frequencies of packets
+	time_freq = dict()
+	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
+		# Each entry is a pcap entry (request/response (packet) and associated metadata)
+		for p in data:
+			# p is a JSON object, not an index
+			layers = p[JSON_KEY_SOURCE][JSON_KEY_LAYERS]
+			# Get timestamp
+			frame = layers.get(JSON_KEY_FRAME, None)
+			date_time = frame.get(JSON_KEY_FRAME_TIME, None)
+     		# Get into the Ethernet address part
+			eth = layers.get(JSON_KEY_ETH, None)
+			# Skip any non DNS traffic
+			if eth is None:
+				print "[ WARNING: Packet has no ethernet address! ]"
+				continue
+		    # Get source and destination MAC addresses
+			src = eth.get(JSON_KEY_ETH_SRC, None)
+			dst = eth.get(JSON_KEY_ETH_DST, None)
+			# Get just the time part
+			date_time_obj = parser.parse(date_time)
+			# Remove the microsecond part
+			time_str = str(date_time_obj.time())[:8]
+			print str(time_str) + " - src:" + str(src) + " - dest:" + str(dst)
+			# Get and count the traffic for the specified MAC address
+			if src == mac_address or dst == mac_address:
+			    # Check if timestamp already exists in the map
+			    # If yes, then just increment the frequency value...
+			    if time_str in time_freq:
+			        time_freq[time_str] = time_freq[time_str] + 1
+			    else: # If not, then put the value one there
+			        time_freq[time_str] = 1
+	return time_freq
+
+if __name__ == '__main__':
+	main()
+