nmap2csv

#!/usr/bin/env python2

###########
# IMPORTS #
###########

from __future__ import print_function
import sys
import argparse
import os
import socket
try:
    import xml.etree.cElementTree as ElementTree
except ImportError:
    import xml.etree.ElementTree as ElementTree


#############
# FUNCTIONS #
#############

def error(*objects):
    print(*objects, file=sys.stderr)


def info(*objects):
    print(*objects, file=sys.stdout)


def extract(xml_files, hosts_up_only=True, open_ports_only=True, ip_address_filter=None):
    hosts = list()
    for xml_file in xml_files:
        if not os.path.exists(xml_file):
            error("File '%s' does not exist." % xml_file)
            exit(1)

        if not os.access(xml_file, os.R_OK):
            error("File '%s' is not readable." % xml_file)
            exit(1)

        xml = ElementTree.parse(xml_file)
        try:
            for host in xml.findall('host'):
                address = host.find('address').get('addr', "")
                status_xml = host.find('status', "")
                status = status_xml.get('state') if status_xml is not None else None
                hostnames_xml = host.find('hostnames')
                hostname_xml = hostnames_xml.find('hostname') if hostnames_xml is not None else None
                hostname = hostname_xml.get('name', "") if hostname_xml is not None else ""
                if ip_address_filter is not None:
                    if address not in ip_address_filter:
                        continue
                if status == 'up':
                    for port in host.find('ports').findall('port'):
                        state = port.find('state').get('state', "")
                        if state == 'open':
                            service_xml = port.find('service')
                            service = service_xml.get('name', "") if service_xml is not None else None
                            tunnel = service_xml.get('tunnel', "") if service_xml is not None else None
                            # rename some of the standard nmap service names so that they are a bit easier to deal with
                            if service is None:
                                service = 'unknown'
                            if service.startswith('http-'):
                                service = 'http'
                            if service.startswith('https-'):
                                service = 'https'
                            if tunnel and tunnel == 'ssl':
                                if service == 'http' or service == 'upnp' or service == 'socks5':
                                    service = 'https'
                            hosts.append(dict(ip_address=address,
                                              host=hostname,
                                              port=port.get('portid', ""),
                                              state=state,
                                              protocol=port.get('protocol', ""),
                                              service=service,
                                              tunnel=tunnel,
                                              product=service_xml.get('product', "") if service_xml is not None else None,
                                              version=service_xml.get('version', "") if service_xml is not None else None,
                                              info=service_xml.get('extrainfo', "") if service_xml is not None else None))
                        else:
                            if not open_ports_only:
                                hosts.append(dict(ip_address=address,
                                                  host=hostname,
                                                  port=port.get('portid', ""),
                                                  state=state))
            else:
                if not hosts_up_only:
                   hosts.append(dict(ip_address=address,
                                     host=hostname))
        except AttributeError as e:
            error("Error parsing file '%s', %s" % (xml, e))
            exit(1)
    return hosts


def print_csv(xml_files, hosts_up_only, open_ports_only, headers=False, delim=","):
    hosts = extract(xml_files, hosts_up_only, open_ports_only)
    lines = set()
    for host in hosts:
        lines.add("%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s" % (
            host.get('ip_address'), delim,
            host.get('host', ""), delim,
            host.get('port', ""), delim,
            host.get('state', ""), delim,
            host.get('protocol', ""), delim,
            host.get('service', ""), delim,
            host.get('tunnel', ""), delim,
            host.get('product', ""), delim,
            host.get('version', ""), delim,
            host.get('info', "")))
    if headers:
        print("IP ADDRESS%sHOSTNAME%sPORT%sSTATE%sPROTOCOL%sSERVICE%sTUNNEL%sPRODUCT%sVERSION%sINFO" % (
            delim, delim, delim, delim, delim, delim, delim, delim, delim))
    for line in sorted(lines, key=lambda item: socket.inet_aton(item.split(',', 1)[0])):
        print(line)

########
# MAIN #
########

if __name__ == '__main__':
    desc = 'Parse nmap xml output and print out all discovered hosts and their open ports in CSV format.'

    parser = argparse.ArgumentParser(description=desc)
    parser.add_argument('files',
                        action='store',
                        nargs='+',
                        help='nmap xml file(s) to parse',
                        metavar='INPUT')
    parser.add_argument('-a', '--hosts',
                        action='store_false',
                        help='print all hosts regardless of status, by default only hosts deemed \'up\' are printed')
    parser.add_argument('-p', '--ports',
                        action='store_false',
                        help='print all ports regardless of status, by default only ports deemed \'open\' are printed')
    parser.add_argument('-t', '--tabs',
                        action='store_true',
                        help='use tab as a field delimiter, by default a comma is used')
    parser.add_argument('-d', '--description',
                        action='store_true',
                        help='print column headers, by default no headers are printed')
    args = parser.parse_args()

    for xml in args.files:
        if not os.path.isfile(xml):
            error("File '%s' does not exist." % xml)
            exit(1)
        if not os.access(xml, os.R_OK):
            error("File '%s' is not readable." % xml)
            exit(1)

    if not args.tabs:
        print_csv(args.files, args.hosts, args.ports, args.description, ",")
    else:
        print_csv(args.files, args.hosts, args.ports, args.description, "\t")