upgrade log parser

This commit is contained in:
Usman Nasir
2020-02-14 21:24:53 +05:00
parent 24792a3489
commit 4b64c84b97

View File

@@ -5,41 +5,77 @@
import os
import re
import sys
import time
from collections import Counter
from datetime import datetime, date, timedelta
from datetime import date, timedelta
from datetime import datetime
from os.path import join, isfile
# print('version is', sys.version)
def detectcontrolpanel():
global controlpanel
try:
if os.path.isfile('/usr/local/cpanel/cpanel'):
controlpanel = 'cpanel'
except:
controlpanel = 'Control Panel not found'
try:
if os.path.isfile('/usr/bin/cyberpanel'):
controlpanel = 'cyberpanel'
except:
controlpanel = 'Control Panel not found'
return controlpanel
def main():
script = sys.argv[0]
filename = sys.argv[2]
# filename = sys.argv[2]
# filenametest = "/home/example.com.access_log"
# username = 'server'
username = str(sys.argv[1])
# Define the day of interest in the Apache common log format.
try:
daysAgo = int(sys.argv[1])
# daysAgo = 2
daysago = int(sys.argv[2])
# daysago = 4
except:
daysAgo = 1
theDay = date.today() - timedelta(daysAgo)
apacheDay = theDay.strftime('[%d/%b/%Y:')
daysago = 0
the_day = date.today() - timedelta(daysago)
apache_day = the_day.strftime('[%d/%b/%Y:')
dcpumon_day = the_day.strftime('%Y/%b/%d')
# Set variables to empty
controlpanel = ''
domlogs_path = ''
try:
if os.path.isfile('/usr/local/cpanel/cpanel') | os.path.isfile(os.getcwd() + '/cpanel'):
controlpanel = 'cpanel'
datetime_dcpumon = date.today().strftime('%Y/%b/%d') # 2020/Feb/10
# Current Dcpumon file
dcpumon_current_log = "/var/log/dcpumon/" + datetime_dcpumon # /var/log/dcpumon/2019/Feb/15
acesslog_sed = "-ssl_log"
if username == 'server':
domlogs_path = '/usr/local/apache/domlogs/'
else:
user_homedir = "/home/" + username
user_accesslogs = "/home/" + username + "/logs/"
domlogs_path = "/usr/local/apache/domlogs/" + username
elif os.path.isfile('/usr/bin/cyberpanel') | os.path.isfile(os.getcwd() + '/cyberpanel'):
controlpanel = 'cyberpanel'
acesslog_sed = ".access_log"
if username == 'server':
# Needs updated to glob all /home/*/logs/
domlogs_path = '/home/username/Desktop/domlogs'
else:
# Get users homedir path
user_homedir = os.path.expanduser("~" + username)
domlogs_path = user_homedir + "/logs/"
except:
controlpanel = 'Control Panel not found'
# Define Output file
stats_output = open(os.getcwd() + '/stats.txt', "w")
# Define log path directory
path = domlogs_path
# path = "/home/username/Desktop/domlogs"
# Get list of dir contents
logs_path_contents = os.listdir(path)
# Get list of files only from this directory
logs = filter(lambda f: isfile(join(path, f)), logs_path_contents)
# Regex for the Apache common log format.
parts = [ # host %h :ip/hostname of the client 172.68.142.138
@@ -75,13 +111,6 @@ def main():
uptimeroboturl = re.compile(r'https?://(www\.)?uptimerobot\..*')
uptimerobot = re.compile(r'UptimeRobot')
# WordPress CMS Regex
wordpresslogin = re.compile(r'wp-login\.php.*')
wordpressadmin = re.compile(r'wp-admin')
wordpresscron = re.compile(r'wp-cron\.php.*')
wordpressxmlrpc = re.compile(r'xmlrpc\.php')
wordpressajax = re.compile(r'admin-ajax\.php')
# Change Apache log items into Python types.
def pythonized(d):
@@ -158,73 +187,155 @@ def main():
else:
return False
# Is the request a Wordpress related login event?
def wordpressbrute(hit):
if hit['request']:
return (wordpresslogin.search(hit['request'])
or wordpressadmin.search(hit['request']))
else:
return False
# create a function which returns the value of a dictionary
def keyfunction(k):
return d[k]
# Initialize.
# Initialize pages for top IP's
pages = []
# Initialize dictionaries for hit counters
wp_login_dict = {}
wp_cron_dict = {}
wp_xmlrpc_dict = {}
wp_admin_ajax_dict = {}
# Parse all the lines associated with the day of interest.
# Open file
log = open(filename)
for line in log:
if apacheDay in line:
for log in logs:
file = os.path.join(path, log)
text = open(file, "r")
wp_login_hit_count = 0
wp_cron_hit_count = 0
wp_xmlrpc_hit_count = 0
wp_admin_ajax_hit_count = 0
for line in text:
if apache_day in line:
if re.match("(.*)(wp-login.php)(.*)", line):
wp_login_hit_count = wp_login_hit_count + 1
if re.match("(.*)(wp-cron.php)(.*)", line):
wp_cron_hit_count = wp_cron_hit_count + 1
if re.match("(.*)(xmlrpc.php)(.*)", line):
wp_xmlrpc_hit_count = wp_xmlrpc_hit_count + 1
if re.match("(.*)(admin-ajax.php)(.*)", line):
wp_admin_ajax_hit_count = wp_admin_ajax_hit_count + 1
m = pattern.match(line)
hit = m.groupdict()
if ispage(hit):
pages.append(pythonized(hit))
else:
continue
log.close()
# print >> stats_output, log + "|" + line,
# print(log + "|" + line, end="", file=stats_output)
log = log.replace('-ssl_log', '', 1)
log = log.replace('.access_log', '', 1)
wp_login_dict[log] = int(wp_login_hit_count)
wp_cron_dict[log] = int(wp_cron_hit_count)
wp_xmlrpc_dict[log] = int(wp_xmlrpc_hit_count)
wp_admin_ajax_dict[log] = int(wp_admin_ajax_hit_count)
# print(log)
# print("Wordpress Logins => " + str(wp_login_hit_count))
# print("Wordpress wp-cron => " + str(wp_cron_hit_count))
# print("Wordpress xmlrpc => " + str(wp_xmlrpc_hit_count))
# print("Wordpress admin-ajax => " + str(wp_admin_ajax_hit_count))
# print("===============================================================")
text.close()
# print(pages, file=stats_output)
print(' ')
print('============================================')
print('Snapshot for ' + username)
print(time.strftime('%H:%M%p %Z on %b %d, %Y'))
if controlpanel == 'cpanel' or controlpanel == 'cyberpanel':
print(controlpanel + " detected")
else:
print('No control Panel detected')
print('Accesslog path used: ' + path)
# print(dcpumon_current_log)
# Show the top five pages and the total.
print ('Show top 10 pages %s' % theDay.strftime('%b %d, %Y'))
print('''
Show top 10 pages %s''' % the_day.strftime('%b %d, %Y'))
pageviews = Counter(x['request'] for x in pages if goodagent(x))
pagestop10 = pageviews.most_common(10)
for p in pagestop10:
print (' %5d %s' % p[::-1])
print (' %5d total' % len(pages))
print(' %5d %s' % p[::-1])
print(' %5d total' % len(pages))
print('============================================')
# Show the top five referrers.
print ('''
Show top 10 referrers %s''' % theDay.strftime('%b %d, %Y'))
print('''
Show top 10 referrers %s''' % the_day.strftime('%b %d, %Y'))
referrers = Counter(x['referrer'] for x in pages if goodref(x))
referrerstop10 = referrers.most_common(10)
for r in referrerstop10:
print (' %5d %s' % r[::-1])
print (' %5d total' % sum(referrers.values()))
print(' %5d %s' % r[::-1])
print(' %5d total' % sum(referrers.values()))
print('============================================')
# Show the top 10 IPs.
print ('''
Show Top 10 IPs %s''' % theDay.strftime('%b %d, %Y'))
print('''
Show Top 10 IPs %s''' % the_day.strftime('%b %d, %Y'))
iphits = Counter(x['host'] for x in pages if goodagent(x))
iptop10 = iphits.most_common(10)
for p in iptop10:
print (' %5d %s' % p[::-1])
print (' %5d total hits' % sum(iphits.values()))
print(' %5d %s' % p[::-1])
print(' %5d total hits' % sum(iphits.values()))
print('============================================')
# CMS Checks
# Wordpress Checks
# Wordpress Login Bruteforcing checks for wp-login.php
print ('''
Wordpress Bruteforce Logins for wp-login.php %s''' % theDay.strftime('%b %d, %Y'))
wordpressloginhits = Counter(x['request'] for x in pages if wordpressbrute(x))
# wordpresslogintop10 = wordpressloginhits.most_common(10)
# for p in wordpresslogintop10:
# print ' %5d %s' % p[::-1]
print (' %5d total' % sum(wordpressloginhits.values()))
print(' ')
print('CMS Checks')
print(' ')
print('Wordpress Checks')
print('============================================')
d = wp_login_dict
# print(d)
print('''Wordpress Bruteforce Logins for wp-login.php %s''' % the_day.strftime('%b %d, %Y'))
print(' ')
# sort by dictionary by the values and print top 10 {key, value} pairs
for key in sorted(d, key=keyfunction, reverse=True)[:10]:
print(' %5d %s' % (d[key], key))
print(' %5d total hits' % sum(dict.values(d)))
print(' ')
d = wp_cron_dict
print('''Wordpress Cron wp-cron.php(virtual cron) checks for %s''' % the_day.strftime('%b %d, %Y'))
print(' ')
# sort by dictionary by the values and print top 10 {key, value} pairs
for key in sorted(d, key=keyfunction, reverse=True)[:10]:
print(' %5d %s' % (d[key], key))
print(' %5d total hits' % sum(dict.values(d)))
print(' ')
d = wp_xmlrpc_dict
print('''Wordpress XMLRPC Attacks checks for xmlrpc.php for %s''' % the_day.strftime('%b %d, %Y'))
print(' ')
# sort by dictionary by the values and print top 10 {key, value} pairs
for key in sorted(d, key=keyfunction, reverse=True)[:10]:
print(' %5d %s' % (d[key], key))
print(' %5d total hits' % sum(dict.values(d)))
print(' ')
d = wp_admin_ajax_dict
print('''Wordpress Heartbeat API checks for admin-ajax.php for %s''' % the_day.strftime('%b %d, %Y'))
print(' ')
# sort by dictionary by the values and print top 10 {key, value} pairs
for key in sorted(d, key=keyfunction, reverse=True)[:10]:
print(' %5d %s' % (d[key], key))
print(' %5d total hits' % sum(dict.values(d)))
print('============================================')
if __name__ == '__main__':
# detectcontrolpanel()
main()