import urllib2
import re
import sys

# parsetor.py script to detect TOR IPs in access_log files
# www.securitybydefault.com

url="https://exitlist.torproject.org/exit-addresses"
page =urllib2.urlopen(url)
data=page.read()

arraydata = data.split("\n")

ips = []

for text in arraydata:

    regex = re.findall( r'[0-9]+(?:\.[0-9]+){3}', text )

    strdata = ', '.join(regex)

    if strdata is not None and strdata not in ips:
        ips.append(strdata)


LogFile = sys.argv[1]

file = open(LogFile, "r")

for LOG in file.readlines():

    regex = re.findall( r'^[0-9]+(?:\.[0-9]+){3}', LOG )

    strdata = ', '.join(regex)

    if strdata in ips:

        print LOG