#!/usr/bin/python
import sys
import re
import os.path

USAGE = """\
USAGE: {} [[-s] SOURCE_IP | -d DEST_IP | SOURCE_IP DEST_IP] FILENAME
Display counts for each unique source-destination-protocol triplet.""".format(sys.argv[0])


"""Basically, this script takes a csv-like file, and gives a count of how many adjacent (e.g. line 44 and line 45) duplicate lines it finds.
Additionally, it will ignore any lines that do not include sys.argv[2] in their 3rd field (i.e. line[2])."""

IP_PATTERN = re.compile(r'^(\d|[1-9]\d|1\d\d|2[0-4]\d|25[0-5])\.'
                       +r'(\d|[1-9]\d|1\d\d|2[0-4]\d|25[0-5])\.'
                       +r'(\d|[1-9]\d|1\d\d|2[0-4]\d|25[0-5])\.'
                       +r'(\d|[1-9]\d|1\d\d|2[0-4]\d|25[0-5])$')

def check_ips(*ips):
   for ip in ips:
        if IP_PATTERN.match(ip) is None:
            print ip + ' is not a valid IPv4 address'
            sys.exit()


filename = os.path.abspath(sys.argv[-1])
if not is.path.exists(filename):
    print USAGE
    sys.exit()
elif os.path.exists(filename) and not os.path.isfile(filename):
    print arg1 + ' is not a file'
    sys.exit()

with open(filename) as f:
    lines = [','.join(line.strip().replace('"', '').split(',')[2:5]) for line in f]

if len(sys.argv) == 3:
    check_ips(sys.argv[1]
    lines = [line for line in lines if sys.argv[1] in line.split(',')[0]]
elif len(sys.argv) == 4 and sys.argv[2] == '-s':
    check_ips(sys.argv[2])
    lines = [line for line in lines if sys.argv[2] in line.split(',')[0]]
elif len(sys.argv) == 4 and sys.argv[2] == '-d':
    check_ips(sys.argv[2])
    lines = [line for line in lines if sys.argv[2] in line.split(',')[1]]
elif len(sys.argv) == 4:
    check_ips(sys.argv[1], sys.argv[2])
    lines = [line for line in lines if sys.argv[1] in line.split(',')[0] and sys.argv[2] in line.split(',')[1]]
else:
    print USAGE
    sys.exit()

counts = {}

for line in lines:
    if line in counts:
        counts[line] += 1
    else:
        counts[line] = 1

print '       Source IP      Destination IP  Protocol           Count'

last = ['', '', '']
for line in sorted(counts.keys()):
    line = line.split(',')
    if [last[0], last[1]] == line[0:2]:
        print (' '*37 + '{:>9} {:>15}'
                        .format(line[2], counts[','.join(line)]))
    elif last[0] == line[0]:
        print (' '*20 + '{:<16} {:>9} {:>15}'
                        .format(line[1], line[2], counts[','.join(line)]))
    else:
        print ('{:<16}    {:<16} {:>9} {count:>15}'
               .format(*line, count = counts[','.join(line)]))
    last = line

print 'Total:  %20s' % sum(counts.values())