IP2ASN IP2host

ip address enrichments.py

df['ip']=df['column'].str.extract('([0-9]+\.[0-9]+\.[0-9]+\.[0-9]+)')

##########################
########### IP to hostname
import socket
import pandas as pd

def ip2hostname (df_series_ip):
    df_output = pd.DataFrame()
    for ip in df_series_ip:
        try:
            df_output = df_output.append({'ip':ip,
                                          'host':socket.gethostbyaddr(ip)[0]},ignore_index=True)
        except:
            df_output = df_output.append({'ip':ip,
                                          'host':''},ignore_index=True)
    return df_output.reset_index(drop=True)

##########################
########### IP to ASN country

import subprocess
import os.path
import random
import time
import pandas as pd

def ip2asn(pandaseries_ips):
    iptoasn_response = open('output_ip2asn_teamcymru.txt', 'w')

    #Creating the request file containing a list of IPs
    #For more information, please access: http://www.team-cymru.com/IP-ASN-mapping.html
    iptoasn_request = open('input_ip2asn_teamcymru.txt', 'w')
    iptoasn_request.write('begin\nverbose\n')
    pd.Series(pandaseries_ips.dropna().unique()).to_csv(iptoasn_request,header=False,index=False,sep="\t") 
    iptoasn_request.write('end')
    iptoasn_request.close()

    #Performing the bulk request
    cat = subprocess.Popen(['cat', 'input_ip2asn_teamcymru.txt'], stdout=subprocess.PIPE)
    print(cat)
    netcat = subprocess.Popen(['netcat', 'whois.cymru.com', '43'], stdin=cat.stdout, stdout=iptoasn_response)
    print (netcat)
    time.sleep(3) #for some reason the poll does not work! This was the way to overcome the waiting time.
    #Closing the output file
    iptoasn_response.close()

    df = pd.read_csv('output_ip2asn_teamcymru.txt',\
                     skiprows=1,\
                     delimiter="\s+\|\s",\
                     names = ['asn', 'ip', 'bgp_prefix', 'country','registry','info_date','as_name'])

    ### CAN BE USEFUL: 
    ###df['asn'] = df['asn'].astype(str).apply(lambda x: 'AS'+x.split('.')[0])
    
    !rm input_ip2asn_teamcymru.txt
    !rm output_ip2asn_teamcymru.txt
    
    return df