roles/apache/files/log-anon

   1 #!/usr/bin/python
   2 import sys, re
   3 from netaddr import IPAddress
   4
   5 if len(sys.argv) != 2:
   6         print >>sys.stderr, "Usage: %s filename" % sys.argv[0]
   7         sys.exit(1)
   8
   9 log = open(sys.argv[1], "a")
  10
  11 # group 1 must be the prefix, group 2 the IP, group 3 the suffix
  12 ipmatch = r"([0-9a-f.:]+)"
  13 accesslog = re.compile(r"^(\S+ )"+ipmatch+r"( .*)$")
  14 errorlog = re.compile(r"^(.* \[client )"+ipmatch+r"(\] .*)$")
  15
  16 while True:
  17         line = sys.stdin.readline()
  18         if not line: break
  19         # process this line
  20         match = accesslog.search(line)
  21         if match is None:
  22                 match = errorlog.search(line)
  23         # check what we got
  24         if match is None:
  25                 # unknown line
  26                 log.write(line)
  27         else:
  28                 prefix = match.group(1)
  29                 ip = match.group(2)
  30                 suffix = match.group(3)
  31                 try:
  32                         ip = IPAddress(ip) # parse the addres
  33                         ip = ip & (IPAddress('255.255.255.0') if ip.version == 4 else IPAddress('ffff:ffff:ffff::')) # mask out a bunch of bits
  34                         # now we have a parsed representation of the IP address we want to print
  35                         log.write(prefix+str(ip)+suffix+"\n")
  36                 except (ValueError, netaddr.core.AddrFormatError):
  37                         # not actually an IP address...
  38                         log.write(line)
  39         log.flush()