SF Python Project Night 4/19/2017: Working with Binary Data

#!/bin/env python3

import sys
fh = open('/Users/Derek/Downloads/net.cap', 'rb')
fh.seek(24)  # Skip the header

loop_count = 0
final_data = dict()
while True:
    timestamp_sec_bytes = fh.read(4)
    if not timestamp_sec_bytes:
        break
    loop_count += 1
    timestamp_ms_bytes = fh.read(4)
    packet_size_bytes = fh.read(4)
    packet_len_bytes = fh.read(4)

    packet_size = int.from_bytes(packet_size_bytes, byteorder="little")
    ethernet_frame_bytes = fh.read(packet_size)

    ip_datagram_bytes = ethernet_frame_bytes[14:]
    ip_datagram_header_length = 4*(ip_datagram_bytes[0] & 0x0F)
    #print("Header length: {}".format(ip_datagram_header_length))

    source_ip_address = ip_datagram_bytes[12:16]
    formatted_ip_source = ["{:d}".format(b)for b in source_ip_address]
    #print("Source IP: {}".format(".".join(formatted_ip_source)))

    dest_ip_address = ip_datagram_bytes[16:20]
    formatted_ip_dest = ["{:d}".format(b)for b in dest_ip_address]
    #print("Dest IP:   {}".format(".".join(formatted_ip_dest)))
    if ".".join(formatted_ip_dest) == "192.30.252.154":
        continue

    tcp_payload = ip_datagram_bytes[ip_datagram_header_length:]
    seq_num = int.from_bytes(tcp_payload[4:8], byteorder="little")
    print("Sequence: {}".format(seq_num))

    tcp_data_offset = tcp_payload[12] >> 4
    print("TCP Data Offset: {}".format(tcp_data_offset))
    final_data[seq_num] = tcp_payload[tcp_data_offset*4:]

    #print(ethernet_frame_bytes[0:22])

    # print(packet_size)
    print("")
    # break

print("Found " + str(loop_count) + " packets")
print("There are {} items in our dictionary".format(len(final_data)))