#!/usr/bin/env python

import re

log_file = 'access.log'
number_of_top_urls = 2
number_of_top_clients = 5

regexp = re.compile(
    r'^(?P<ipaddr>172\.16\.'     # 1st and 2nd octets
    r'(1\.'                      # case when 3d octet is "1"
    r'([1-9][0-9]|1[0-9]{1,2}|'  # (1) 0-99,100-199
    r'2[0-4][0-9]|25[0-5])'      # (1) 200-249,250-255
    r'|(2\.'                     # case when 3d octet is "2"
    r'([0-9]|[1-9][0-9]|'        # (2) 0-9,10-99,
    r'1[0-1][0-9]|120)'          # (2) 100-119,120
    r')))\s+'                    # whitespace(s)
    r'\S+\s+\S+\s+'              # '- - ' some stuff
    r'(?P<time>\[[^\]]+\])\s+'   # timestamp
    r'"(?P<request>[^"]+)"'      # request - 'METHOD /url RESPONSE_CODE bytes'
    r'\s+.*'                     # rest of log line
)

filtered_requests = dict()

with open(log_file) as logs:
    for line in logs.readlines():
        m = re.match(regexp, line)
        if not m:
            continue
        ip = m.group("ipaddr")
        request = m.group("request")
        url = request.split()[1]
        if url not in filtered_requests:
            filtered_requests[url] = [ip]
        else:
            filtered_requests[url].append(ip)

top_requests = sorted(
    filtered_requests,
    key=lambda x: len(filtered_requests[x]),
    reverse=True
)[0:number_of_top_urls]

top_requests_ips = []

for req in top_requests:
    top_requests_ips += filtered_requests[req]

top_ips_of_top_requests = sorted(
    set(top_requests_ips),
    key=lambda x: top_requests_ips.count(x),
    reverse=True
)[0:number_of_top_clients]

print ('{0} most active users of {1} most popular URLs: {2}'
       ''.format(number_of_top_clients, number_of_top_urls,
                 ', '.join(top_ips_of_top_requests)))