seoLogs_spotSpoofedSEBots.py

#Define the function

def reverse_dns(ip_address):
    '''
    This method returns the true host name for a 
    given IP address
    '''
    host_name = socket.gethostbyaddr(ip_address)
    reversed_dns = host_name[0]
    return reversed_dns

def forward_dns(reversed_dns):
    '''
    This method returns the first IP address string
    that responds as the given domain name
    '''
    try:
        data = socket.gethostbyname(reversed_dns)
        ip = str(data)
        return ip
    except Exception:
        print('error')
        return False

def ip_match(ip, true_ip):
    '''
    This method takes an ip address used for a reverse dns lookup
    and an ip address returned from a forward dns lookup
    and determines if they match.
    '''
    if ip == true_ip:
        ip_match = True
    else:
        ip_match = False
    return ip_match

def confirm_googlebot(host, ip_match):
    '''
    This method takes a hostname and the results of the ip_match() method
    and determines if an ip address from a log file is truly googlebot
    '''
    googlebot = False
    if host != False:
        if host.endswith('.googlebot.com') or host.endswith('.google.com'):
            if ip_match == True:
                #googlebot = 'Yes'
                googlebot = True
    return googlebot
                
def run(ip):
    try:
        host = reverse_dns(ip)
        true_ip = forward_dns(host)
        is_match = ip_match(ip, true_ip)
        return confirm_googlebot(host, is_match)
    except:
        #return 'No'
        return False

#Run the function against the IP addresses listed in the dataframe

df['isRealGbot?'] = df['IP'].apply(run)