Masscan XML Parser

scanparser.py

__author__ = '@awhitehatter'
__version__ = '0.1'

import xml.etree.ElementTree as ET
import argparse
import sys


'''
###########################################################################
Quick and Dirty Masscan XML parser. 
Can yank out hosts, ports or both (colon seperator)

usage: scanparser.py [-h] -f FILE [-p] [-i] [--pretty] [--nospace]

Command line arguments

optional arguments:
  -h, --help            show this help message and exit
  -f FILE, --file FILE  XML File to parse
  -p, --ports           Lists ports
  -i, --ips             Lists host ip addresses
  --pretty              Outputs a comma seperated list (instead of every line)
  --nospace             Sets pretty output to no spaces, usefule for copy
                        pasta into other tools
###########################################################################
'''
def parseAll(root):
    # Parses out host address and port number, prints them host:port
    for item in root.findall('./host'):
        for child in item:
            if child.tag == 'address':
                address = child.attrib['addr']
            elif child.tag == 'ports':
                for grandchild in child:
                    print(address + ":" + grandchild.attrib['portid'])


def parsePorts(root):
    # Parses and prints out ports
    for item in root.findall('./host'):
        for child in item:
            if child.tag == 'ports':
                for grandchild in child:
                    print(grandchild.attrib['portid'])


def parseIPs(root):
    # Parses and prints out IP addresses
        for item in root.findall('./host'):
            for child in item:
                if child.tag == 'address':
                    address = child.attrib['addr']
                    print(address)


def prettyPorts(root):
    # Parses and prints out comma separated ports with spaces

    #create an empty list
    plist = []

    #parse ports and add to list
    for item in root.findall('./host'):
        for child in item:
            if child.tag == 'ports':
                for grandchild in child:
                    if grandchild.attrib['portid'] not in plist:
                        plist.append(grandchild.attrib['portid'])

    # Join comma list 
    pstring = ", ".join(plist)

    # Print the list
    print(pstring)


def prettyPortsNoSpaces(root):
    # Parses and prints out comma seperated ports with no spaces

    #create an empty list
    plist = []

    #parse ports and add to list
    for item in root.findall('./host'):
        for child in item:
            if child.tag == 'ports':
                for grandchild in child:
                    if grandchild.attrib['portid'] not in plist:
                        plist.append(grandchild.attrib['portid'])

    # Join comma list 
    plist.sort()
    pstring = ",".join(plist)

    # Print the list
    print(pstring)



def prettyHosts(root):
    # parses and prints out comma separted IP addresses with spaces
        
    #create an empty list
    phosts = []    
    
    # Parses and prints out IP addresses
    for item in root.findall('./host'):
        for child in item:
            if child.tag == 'address':
                if child.attrib['addr'] not in phosts:
                    phosts.append(child.attrib['addr'])
    
    #Join comma list
    phosts.sort()
    pstring = ", ".join(phosts)

    print(pstring)


def prettyHostsNoSpaces(root):
    # parses and prints out comma separted IP addresses with no spaces
        
    #create an empty list
    phosts = []    
    
    # Parses and prints out IP addresses
    for item in root.findall('./host'):
        for child in item:
            if child.tag == 'address':
                if child.attrib['addr'] not in phosts:
                    phosts.append(child.attrib['addr'])
    
    #Join comma list
    phosts.sort()
    pstring = ",".join(phosts)

    print(pstring)


if __name__ == "__main__":
    # parse the command line arguments
    parser = argparse.ArgumentParser(description="Command line arguments")
    parser.add_argument('-f', '--file', action='store', help='XML File to parse', required=True)
    parser.add_argument('-p', '--ports', action='store_true',  default=False, help='Lists ports')
    parser.add_argument('-i', '--ips', action='store_true', default=False, help='Lists host ip addresses')
    parser.add_argument('--pretty', action='store_true', default=False, help='Outputs a comma seperated list')
    parser.add_argument('--nospace', action='store_true', default=False, help='Sets pretty output to no spaces, usefule for copy pasta into other tools')
    args = parser.parse_args()

    xmldata = args.file
    ports = args.ports
    ips = args.ips
    pretty = args.pretty
    spaces = args.nospace
    
    if pretty == False and spaces == True:
        print("You must specify --pretty with --nospace")
        sys.exit(1)

    if pretty and ports == False and ips == False:
        print("You must specify ports (-p) or ips (-i) for comma list")
        sys.exit(1)

    #Parse XML 
    tree = ET.parse(xmldata)
    root = tree.getroot()

    #Parse according to specified options
    if ips == False and ports == False:
        parseAll(root)
    elif pretty:
        if spaces:
            if ips:
                prettyHostsNoSpaces(root)
            
            if ports:
                prettyPortsNoSpaces(root)
        else:
            if ips:
                prettyHosts(root)
            
            if ports:
                prettyPorts(root)
    else:
        if ips:
            parseIPs(root)

        if ports:
            parsePorts(root)