MassScan SAX XML Parser

#!/usr/bin/env python

# sax parser for massscan

import xml.sax


class XML_Parser(xml.sax.ContentHandler):

    def __init__(self):
        self.system_name = None
        self.port_number = None
        self.protocol = None
        self.masscan = False
        self.nmap = False
        self.nessus = False
        self.url_list = []
        self.port_open = False
        self.rdp_list = []
        self.vnc_list = []
        self.http_ports = ['80', '8080']
        self.https_ports = ['443', '8443']

    def startElement(self, tag, attributes):
        # Determine the Scanner being used
        if tag == "nmaprun" and attributes['scanner'] == "masscan":
            self.masscan = True
        elif tag == "nmaprun" and attributes['scanner'] == "nmap":
            self.nmap = True
        elif tag == "nessusclientdata_v2":
            self.nessus = True

        if self.masscan:
            if tag == "address":
                self.system_name = attributes['addr']
            elif tag == "port":
                self.port_number = attributes['portid']
            elif tag == "service":
                if "http" in attributes['name']:
                    self.protocol = "http"
                elif "ssl" in attributes['name']:
                    self.protocol = "https"
                elif "vnc" in attributes['name']:
                    self.protocol = "vnc"
            elif tag == "state":
                if attributes['state'] == "open":
                    self.port_open = True

        elif self.nmap:
            pass

        elif self.nessus:
            pass

    def endElement(self, tag):
        if self.masscan:
            if tag == "host":
                if (self.system_name is not None) and (self.port_number is not None) and self.port_open:
                    if self.protocol == "http" or self.protocol == "https":
                        built_url = self.protocol + "://" + self.system_name + ":" + self.port_number
                        if built_url not in self.url_list:
                            self.url_list.append(built_url)
                    elif self.protocol is None and self.port_number in self.http_ports:
                        built_url = "http://" + self.system_name + ":" + self.port_number
                        if built_url not in self.url_list:
                            self.url_list.append(built_url)
                    elif self.protocol is None and self.port_number in self.https_ports:
                        built_url = "https://" + self.system_name + ":" + self.port_number
                        if built_url not in self.url_list:
                            self.url_list.append(built_url)
                    elif self.protocol == "vnc":
                        if self.system_name not in self.vnc_list:
                            self.vnc_list.append(self.system_name)
                    elif self.port_number == "3389":
                        if self.system_name not in self.rdp_list:
                            self.rdp_list.append(self.system_name)

                self.system_name = None
                self.port_number = None
                self.protocol = None
                self.port_open = False
            elif tag == "nmaprun":
                for url in self.url_list:
                    print url

    def characters(self, content):
        pass

if __name__ == "__main__":

    # Time to start parsing
    # Create parser
    parser = xml.sax.make_parser()
    # Turn off namespaces
    parser.setFeature(xml.sax.handler.feature_namespaces, 0)

    # Override the parser
    Handler = XML_Parser()
    parser.setContentHandler(Handler)

    parser.parse('filename.xml')