ipmapgen

#!/usr/bin/python3

import json
import urllib.request
import re
from collections import defaultdict
import os
import stat
import argparse

DEFAULT_GEOIP_URL='http://freegeoip.net/json/%s'

def get_geoip(geoip_url, ipaddress):
  content = urllib.request.urlopen(geoip_url % ipaddress, None, 20).read()
  return json.loads(content.decode('utf8'))

def fill_json(initjson, infos, count):
  initjson['infos'] = infos
  initjson['count'] = count
  return initjson

def generate_geoip(in_fp, out_fp, regexp, capture_ip_pos, capture_nb, geoip_url):
  try:
    countarray = defaultdict(int)
    match_obj = re.compile(regexp)
    for line in in_fp:
      match = match_obj.search(line)
      if match != None:
        infosarr = []
        for i in range(1, capture_nb):
          infosarr.append(match.group(i))

        infos = ', '.join(infosarr)
        ipaddress = match.group(capture_ip_pos)
        countarray[(infos, ipaddress)] += 1

    jsonarr = []
    for infos, ipaddress in countarray:
      initjson   = get_geoip(geoip_url, ipaddress)
      filledjson = fill_json(initjson, infos, countarray[infos, ipaddress])
      jsonarr.append(filledjson)

    json.dump(jsonarr, out_fp)

  except Exception as e:
    print(e)

def main():
  parser = argparse.ArgumentParser(description='Generate a JSON file of geographical coordinates based on a log file')
  parser.add_argument('-I', '--in', metavar = 'infile',
                      help     = 'Input log file to parse',
                      type     = argparse.FileType('r'),
                      required = True,
                      dest     = 'in_fp')
  parser.add_argument('-O', '--out', metavar = 'outfile',
                      help     = 'Output file to store generated JSON (overwritten)',
                      type     = argparse.FileType('w'),
                      required = True,
                      dest     = 'out_fp')
  parser.add_argument('-r', '--regexp', metavar = 'regexp',
                      help     =   'A Python 3 valid regexp holding at least one capture group for '
                                 + 'an IP address. If more than one capture group are provided, '
                                 + 'then you must specify \'-c\' and \'-i\'. Supplementary capture '
                                 + 'groups will be use to group matching IP addresses',
                      required = True,
                      dest     = 'regexp')
  parser.add_argument('-c', '--capture-count', metavar = 'count',
                      help     =  'Count of capture group in the provided \'regexp\'',
                      type     = int,
                      default  = 1,
                      dest     = 'capture_gp_nb')
  parser.add_argument('-i', '--ip-capture-pos', metavar = 'pos',
                      help     =  'Position of the IP address capture group in \'regexp\'',
                      type     = int,
                      default  = 1,
                      dest     = 'capture_ip_pos')
  parser.add_argument('-u', '--url', metavar = 'geoip_url',
                      help     =   'Override default URL of the GeoIP service to use. Use \'%%\' '
                                 + 'to specify IP position',
                      default  = DEFAULT_GEOIP_URL,
                      dest     = 'geoip_url')
  namespace = parser.parse_args()

  generate_geoip(namespace.in_fp, namespace.out_fp, namespace.regexp,
                 namespace.capture_gp_nb, namespace.capture_ip_pos,
                 namespace.geoip_url)


if __name__ == '__main__':
  main()