Parse nmap output to print host port

parse-nmap.sh

#!/bin/bash
# TODO
# - Consolidate ports for the same host from different files

ONE_PER_LINE=0
AS_URLS=0
AS_IPS=0
NO_VERSION=0
SEP=" "
SUBSEP=","
FILES=()
HTTPS_RE="https|http.*(ssl|tls)|(ssl|tls).*http"
HTTP_RE="http|web"
HTTP_PORTS="80,8080,8000,8001"
HTTPS_PORTS="443,8443"

usage() {
  cat <<EOF
Print open ports in nmap scan.

Usage: $(basename ${0}) [<options>] <.gnmap file> ...

Option:
  -1                    Display one open port per line. Default is one host per line with a comma-separated list of ports.
  -s|--sep <sep>        Separator to use between host and port. Default is '${SEP}'.
  -S|--subsep <sep>     Separator to use between ports. Default is '${SUBSEP}'. Ignored if -1 is given.
  -u|--url              Display HTTP services as URLs (this enables -1). Non-matching ports will not be displayed.
  -n|--ip               Do not use the domain name, only IP address.
  -N|--no-version       Do not show the version info.
  --http-re <regex>     Regex to use to detect HTTP services (checked last). Set to empty to disable and check only port. Default is '${HTTP_RE}'.
  --https-re <regex>    Regex to use to detect HTTPS services (checked first). Set to empty to disable and check only port. Default is '${HTTPS_RE}'.
  --http-ports <ports>  Always treat those ports (comma-separated) as HTTP services. Default is '${HTTP_PORTS}'.
  --https-ports <ports> Always treat those ports (comma-separated) as HTTPS services. Default is '${HTTPS_PORTS}'.

Multiple files can be given. Comparison is case-insensitive.
EOF
  exit 1
}

while [[ $# -gt 0 ]] ; do
  case "${1}" in
    -1)
      ONE_PER_LINE=1
      ;;
    -u|--url)
      AS_URLS=1
      ONE_PER_LINE=1
      ;;
    -n|--ip)
      AS_IPS=1
      ;;
    -N|--no-version)
      NO_VERSION=1
      ;;
    -s|--sep)
      SEP="${2}"
      shift
      ;;
    -S|--subsep)
      SUBSEP="${2}"
      shift
      ;;
    --http-re)
      HTTP_RE="${2}"
      shift
      ;;
    --https-re)
      HTTPS_RE="${2}"
      shift
      ;;
    --http-ports)
      HTTP_PORTS="${2}"
      shift
      ;;
    --https-ports)
      HTTPS_PORTS="${2}"
      shift
      ;;
    -*)
      usage
      ;;
    *)
      FILES+=("${1}")
      ;;
  esac
  shift
done

awk -F '\t' \
  -v one_per_line="${ONE_PER_LINE}" \
  -v as_urls="${AS_URLS}" \
  -v as_ips="${AS_IPS}" \
  -v no_version="${NO_VERSION}" \
  -v http_re="${HTTP_RE}" \
  -v https_re="${HTTPS_RE}" \
  -v http_ports="${HTTP_PORTS}" \
  -v https_ports="${HTTPS_PORTS}" \
  -v sep="${SEP}" \
  -v subsep="${SUBSEP}" '
function show_host_compact(host, ports, _p, _first, _pnum, _proto, _srv) {
  _first = 1
  for (_p in ports) {
    if (! is_open(ports[_p])) { continue }
    if (_first) {
      printf "%s%s", hostname(host, 1), sep
    }

    _pnum = port_num(ports[_p])
    _proto = port_proto(ports[_p])
    _srv = port_srv(ports[_p])
    printf "%s%s/%s", (_first ? "" : subsep), _pnum, _proto
    if (! no_version) { printf " (%s)", _srv }
    _first = 0
  }
  print ""
}

function show_host(host, ports, _p, _http, _pnum, _proto, _srv) {
  for (_p in ports) {
    if (! is_open(ports[_p])) { continue }

    _pnum = port_num(ports[_p])
    _proto = port_proto(ports[_p])
    _srv = port_srv(ports[_p])
    if (as_urls) {
      _http = http_proto(ports[_p])
      if (!_http) { continue }
      printf "%s://%s:%s", _http, hostname(host), _pnum
    } else {
      printf "%s%s%s/%s", hostname(host, 1), sep, _pnum, _proto
    }
    if (! no_version) { printf " (%s)", _srv }
    printf "\n"
  }
}

function hostname(host, full, _f) {
  if (full) { _f = 0 }
  else { _f = 2 }
  match(host, /([0-9.]+) \((.*)\)$/, m)
  if (m[2] && ! as_ips) { return m[_f] }
  return m[1]
}

function http_proto(port, _pnum, _srv) {
  _pnum = port_num(port)
  _srv = port_srv(port)
  if (_port_or_service_matches(_srv, _pnum, https_re, https_ports)) {
    return "https"
  }
  if (_port_or_service_matches(_srv, _pnum, http_re, http_ports)) {
    return "http"
  }
}

function _port_or_service_matches(srv, port, srv_re, port_list, _tmp, _i) {
  if (srv_re && srv ~ srv_re) { return 1 }
  split(port_list, _tmp, / *, */) 
  for (_i in _tmp) {
    if (port == _tmp[_i]) { return 1 }
  }
}

function is_open(port) {
  return (port_field(port, 2) == "open")
}

function port_num(port) {
  return port_field(port, 1)
}

function port_proto(port) {
  return port_field(port, 3)
}

function port_srv(port, _srv) {
  _srv = port_field(port, 5) " " port_field(port, 7)
  return (_srv == " " ? "unknown" : _srv)
}

function port_field(port, i, _tmp) {
  split(port, _tmp, "/") 
  return _tmp[i]
}

BEGIN { IGNORE_CASE = 1 }

# ($0, /^Ports: *(.*)/, m) {
($2 ~ /^Ports:/) {
  host = gensub(/Host: /, "", 1, $1)
  split(gensub(/Ports: /, "", 1, $2), ports, /, /)
  if (one_per_line) {
    show_host(host, ports)
  } else {
    show_host_compact(host, ports)
  }
}
' "${FILES[@]}"