weblog_helper

#!/usr/bin/env python

"""
Notes:
weblog-helper script requires netaddr python library. Please install it beforehand:
pip install netaddr

TODO:
- check if local log is present, else download from S3. Read from file
"""

import sys
if sys.version_info > (3, 0):
	sys.stderr.write('This script requires Python 2.7')
	sys.exit(1)

import netaddr
import argparse
import json
import logging
import requests
logging.basicConfig(format='%(asctime)s %(levelname)-8s %(message)s',
					level=logging.ERROR,
					datefmt='%Y-%m-%d %H:%M:%S'
					)


def validate_ip(arg_source_ip):
	"""Return True if *arg_source_ip* is a valid ip"""
	try:
		ip = netaddr.IPNetwork(arg_source_ip)
	except (netaddr.core.AddrFormatError, ValueError):
		raise ValueError("Argument (%s) is not in correct presentation format" % arg_source_ip)
	return True

def get_access_log(target_url):
	"""Return access log"""
	return requests.get(target_url).content

def print_if_interesting(log_line, arg_source_ip):
	"""Print *log_line* if it is interesting """
	log_source_ip = get_source_ip(log_line)
	if is_in_range(log_source_ip, arg_source_ip):
		print(log_line)

def get_source_ip(log_line):
	"""Return source ip field from *log_line* """
	return log_line.split(' - - ')[0]

def is_in_range(log_source_ip, arg_source_ip):
	"""Return True if *log_line* contains the source ip we are looking for """
	if is_cidr:
		return netaddr.IPAddress(log_source_ip) in netaddr.IPNetwork(arg_source_ip)
	else:
		return log_source_ip == arg_source_ip

def is_cidr(arg_source_ip):
	"""Return True if provided *arg_source_ip* represents a CIDR, not a host"""
	if netaddr.IPNetwork(arg_source_ip).prefixlen < 32:
		return True
	return False

def main():
	logging.info("Execution started.")
	
	# Read command-line parameters
	parser = argparse.ArgumentParser()
	parser.add_argument("--ip", action="store", required=True, dest="ip", help="Source IP or CIDR to search for.")

	args = parser.parse_args()
	arg_source_ip = args.ip

	# Check if IP has valid format (host or cidr)
	if validate_ip(arg_source_ip):

		# Check if ip argument is host or cidr
		# Set varible *is_cidr* as global to only check once if *arg_source_ip* is host or cidr
		global is_cidr
		if is_cidr(arg_source_ip):
			is_cidr = True
		else:
			is_cidr = False

		# Read http access log
		# Not ideal if log file is big, would be better to use a file handler. Also downloads file every time script is run
		data = get_access_log("https://s3.amazonaws.com/syseng-challenge/public_access.log.txt")
		logging.info("Log file downloaded.")

		for line in data.splitlines():
			print_if_interesting(line, arg_source_ip)

	logging.info("Execution finished.")


if __name__ == "__main__":
	main()