source: hybrid/branches/releng-9.0/nanobsd/files/usr/local/www/wlportal/index.cgi@ 10725

#!/usr/bin/env python
#
# Wrap me around tcpserver or inetd, example usage for tcpserver (debug):
# tcpserver -HRl localhost 172.31.255.1 /root/wlportal.py
#
# Or put me in a CGI script in for example thttpd server:
#
# = Usage =
# This is a wrapper script which does very basic HTML parsing and altering of
# pfctl tables rules to build a basic Captive Portal, with basic sanity
# checking. The ACL is IP based (this is a poor mans solution, layer2
# ACL would be much better), so don't take security very seriously.
# 
# To get traffic by default to the portal iI requires a few special rules in
# pfctl to work properly:
#   no rdr on { $captive_ifs } proto tcp from <wlportal> to !$wl_net port 80
#   rdr on { $captive_ifs } proto tcp from $wl_net to !$wl_net port 80 -> 127.0.0.1 port 8082
# 
# Enties older than 5 minutes not being used will be removed if the (hidden)
# argument action=cleanup is given as GET variable. So having this in cron (would fix it):
# */5 * * * * /usr/bin/fetch -q http://172.31.255.1/wlportal?action=cleanup
#
# XXX: The whitelist entries first needs to contact the wlportal.py to get
# added to the whitelist, this may cause issues during initial setup and hence
# it might be advised to create a block of static whitelist IP addresses which
# get added during boot and will never disappear.
#
# State   : v0.6.0
# Version : $Id: index.cgi 10725 2012-05-08 18:11:40Z rick $
# Author  : Rick van der Zwet <info@rickvanderzwet.nl>
# Licence : BSDLike http://wirelessleiden.nl/LICENSE
import cgitb
cgitb.enable(logdir="/tmp")


import os
import signal
import subprocess
import socket
import sys
import time
import traceback
import urlparse
import yaml

from jinja2 import Template

# XXX: Make me dynamic for example put me in the conf file
cfg = { 
  'autologin'     : False,
  'cmd_arp'       : '/usr/sbin/arp',
  'pfctl'         : '/sbin/pfctl',
  'portal_sponsor': 'Sponsor van Stichting Wireless Leiden',
  'portal_url'    : 'http://wirelessleiden.nl/welkom?connected_to=%s' % socket.gethostname(),
  'portalroot'    : '172.31.255.1',
  'refresh_delay' : 3,
  'tmpl_autologin': '/usr/local/etc/wlportal/autologin.tmpl',
  'tmpl_login'    : '/usr/local/etc/wlportal/login.tmpl',
  'whitelist'     : [],
  'config_files'  : ['/usr/local/etc/wlportal/config.yaml','/etc/wleiden.yaml'],
  'expire_time'   : None,
  'accessdb'      : '/var/db/clients',
  'net_status'    : '/tmp/network.status',
}


# No failback if config does not exist, to really make sure the user knows if
# the config file failed to parse properly or is non-existing
for config_file in cfg['config_files']:
 if os.path.isfile(config_file):
   cfg.update(yaml.load(open(config_file)))

internet_up = True
if os.path.isfile(cfg['net_status']):
  internet_up = 'internet=up' in open(cfg['net_status'], 'r').read().lower()

if not internet_up:
  cfg['warning_msg'] = "<b>Internet Problemen</b>: De laatste 15 minuten zijn er problemen met de (internet) verbinding geconstateerd, de gebruikers ervaring kan dus niet optimaal zijn. Onze excuses voor het eventuele ongemak. Bij aanhoudende problemen kunt u contact opnemen met gebruikers@lijst.wirelessleiden.nl" 

def log_registered_host(remote_mac, remote_host):
  """
   Write statistics file, used for (nagios) monitoring purposes 
  """
  with open(cfg['accessdb'],"a") as fh:
   epoch = int(time.time())
   fh.write("%s %s %s \n" % (epoch, remote_mac, remote_host) )

class MACnotFound(Exception):
  pass

def get_mac(ipaddr):
  """ Find out the MAC for a certain IP address """
  try:
    return subprocess.check_output(['/usr/sbin/arp', '-n' ,ipaddr], shell=False).split()[3][1:-1]
  except subprocess.CalledProcessError:
    raise MACnotFound

def get_active_MACs():
  """ Return dictionary with active IPs as keys """
  output = subprocess.check_output(['/usr/sbin/arp', '-n' ,'-a'], shell=False)
  db = {}
  for line in output.strip().split('\n'):
    i = line.split()
    mac = i[3][1:-1]
    ip = i[5]
    db[ip] = mac
  return db
    

class PacketFilterControl():
  """ Manage an Packet Filter using pfctl and table wlportal"""
  def add(self, ipaddr):
    """ Add Allow Entry in Firewall"""
    output = subprocess.Popen([cfg['pfctl'],'-t','wlportal', '-T', 'add', ipaddr], stderr=subprocess.PIPE).communicate()[1]
    is_added = '1/1 addresses added.' in output
    return is_added
  def delete(self, ipaddr):
    """ Delete one Allow Entry to Firewall"""
    output = subprocess.Popen([cfg['pfctl'],'-t','wlportal', '-T', 'delete', ipaddr], stderr=subprocess.PIPE).communicate()[1]
  def flush(self):
    """ Delete all Allow Entries from Firewall"""
    output = subprocess.Popen([cfg['pfctl'],'-t','wlportal', '-T', 'flush'], stderr=subprocess.PIPE).communicate()[1]
    #0 addresses deleted.
    return int(output.strip().split('\n')[-1].split()[0])
  def cleanup(self, expire_time=None):
    """ Delete obsolete entries and expired entries from the Firewall"""
    deleted_entries = 0
    # Delete entries older than certain time
    if expire_time:
      output = subprocess.Popen([cfg['pfctl'],'-t','wlportal', '-T', 'expire', expire_time], stdout=subprocess.PIPE).communicate()[0]
      # 0/0 addresses expired.
      deleted_entries += int(output.strip.split()[-1].split('/')[0])

    # Delete entries which the MAC<>IP mapping does no longer hold. The
    # ``rogue'' clients, commonly seen when DHCP scope is small and IPs get
    # re-used frequently, are wipped and require an re-connect.
    stored_mac = {}
    if os.path.isfile(cfg['accessdb']):
      for line in open(cfg['accessdb'],'r'):
        (epoch, mac, ipaddr) = line.split()
        stored_mac[ipaddr] = mac
    # Live configuration
    active_mac = get_active_MACs()
    # Process all active ip addresses from firewall and compare changes   
    output = subprocess.Popen([cfg['pfctl'],'-t','wlportal', '-T', 'show'], stdout=subprocess.PIPE).communicate()[0]
    for ip in output.split():
      if ip in cfg['whitelist']:
        # IP is whitelisted
        continue
      elif active_mac.has_key(ip) and active_mac[ip] in cfg['whitelist']:
        # MAC is whitelisted
        continue
      elif stored_mac.has_key(ip) and active_mac[ip] == stored_mac[ip]:
        # previous record found
        # Stored v.s. Active happy
        continue
      else:
        self.delete(ip)
        deleted_entries =+ 1 
    return deleted_entries
      
      
# Call from crontab
if sys.argv[1:]:
  if sys.argv[1] == 'cleanup':
    fw = PacketFilterControl()
    fw.cleanup()
    sys.exit(0)

### BEGIN STANDALONE/CGI PARSING ###
#
# Query String Dictionaries
qs_post = None
qs = None
header = []
if not os.environ.has_key('REQUEST_METHOD'):
  # a) We are not wrapped around in a HTTP server, so this _is_ the
  #    HTTP server, so act like one.
  class TimeoutException(Exception):
    """ Helper for alarm signal handling"""
    pass
  
  def handler(signum, frame):
    """ Helper for alarm signal handling"""
    raise TimeoutException
  
  
  # Parse the HTTP/1.1 Content-Header (partially)
  signal.signal(signal.SIGALRM,handler)
  us = None
  method = None
  hostname = None
  content_length = None
  remote_host = None
  while True:
    try:
      signal.alarm(1)
      line = sys.stdin.readline().strip()
      if not line:
        break
      header.append(line)
      signal.alarm(0)
      if line.startswith('GET '):
        us = urlparse.urlsplit(line.split()[1])
        method = 'GET'
      elif line.startswith('POST '):
        method = 'POST'
        us = urlparse.urlsplit(line.split()[1])
      elif line.startswith('Host: '):
        hostname = line.split()[1]
      elif line.startswith('Content-Length: '):
        content_length = int(line.split()[1])
    except TimeoutException:
      break
  
  # Capture Portal, make sure to redirect all to portal
  if hostname != cfg['portalroot']:
    print "HTTP/1.1 302 Moved Temponary\r\n",
    print "Location: http://%(portalroot)s/\r\n" % cfg,
    sys.exit(0)
  
  
  # Handle potential POST
  if method == 'POST' and content_length:
    body = sys.stdin.read(content_length)
    qs_post = urlparse.parse_qs(body)
  
  # Parse Query String
  if us and us.path == "/wlportal" and us.query:
    qs = urlparse.parse_qs(us.query)

  remote_host = os.environ['REMOTEHOST']
else:
  # b) CGI Script: Parse the CGI Variables if present
  if os.environ['REQUEST_METHOD'] == "POST":
    content_length = int(os.environ['CONTENT_LENGTH'])
    body = sys.stdin.read(content_length)
    qs_post = urlparse.parse_qs(body)

  if os.environ.has_key('QUERY_STRING'):
    qs = urlparse.parse_qs(os.environ['QUERY_STRING'])

  remote_host = os.environ['REMOTE_ADDR']
#
### END STANDALONE/CGI PARSING ###


# Helpers for HTML 'templates'
content = cfg.copy()
content.update(extra_header='')

# IP or MAC on the whitelist does not need to authenticate, used for devices
# which need to connect to the internet, but has no 'buttons' to press OK.
#
# This assumes that devices will re-connect if they are not able to connect 
# to their original host, as we do not preserve the original URI.
remote_mac = get_mac(remote_host)
if cfg['autologin'] or remote_host in cfg['whitelist'] or remote_mac in cfg['whitelist']:
  qs_post = { 'action' : 'login' }

try:
  fw = PacketFilterControl()
  
  # Put authenticate use and process response
  if qs and qs.has_key('action'):
    if 'flush' in qs['action']:
      retval = fw.flush()
      content['status_msg'] += "# [INFO] Deleted %s entries\n" % retval
    elif 'update' in qs['action']:
      tech_footer = "# [INFO] Update timestamp of all entries\n"
      fw.update()
      content['status_msg'] += fw.get_log()
    elif 'cleanup' in qs['action']:
      retval = fw.cleanup(cfg['expire_time'])
      content['status_msg'] += "# [INFO] Deleted %s entries\n" % retval
  elif qs_post and qs_post.has_key('action'):
    if 'login' in qs_post['action']:
      if fw.add(remote_host):
        content['extra_header'] = "Refresh: %(refresh_delay)s; url=%(portal_url)s\r" % content
        content['status_msg'] = "Sucessfully Logged In! || " +\
        """ Will redirect you in %(refresh_delay)s seconds to <a href="%(portal_url)s">%(portal_url)s</a> """ % content
        log_registered_host(remote_mac, remote_host)
      else:
        content['status_msg'] = "ERROR! Already Logged On"
    elif 'logout' in qs_post['action']:
      fw.delete(remote_host)
      content['status_msg'] = "Succesfully logged out!"

except Exception, e:
  content['tech_footer'] += traceback.format_exc()
  content['status_msg'] = "<div class='error'>Internal error!<pre>%s</pre></div>" % traceback.format_exc()
  pass

  # Present Main Screen
print """\
HTTP/1.1 200 OK\r
Content-Type: text/html\r
%(extra_header)s
""" % content

try:
  tmpl_file = cfg['tmpl_autologin'] if cfg['autologin'] else cfg['tmpl_login']
  page = open(tmpl_file,'r').read()
except IOError:
  page = """
<html><head></head><body>
<h2>%(status_msg)s</h2>

<h3>Wireless Leiden - Internet Portal</h3>
<form action="http://%(portalroot)s/wlportal/" method="POST">
<input name="action" type="hidden" value="login" />
<input type="submit" value="OK, agreed" />
</form>

<h3>More options</h3>
<form action="http://%(portalroot)s/wlportal/" method="POST">
<input name="action" type="hidden" value="logout" />
<input type="submit" value="Cancel and/or Logout" />
</form>
<hr /><em>Technical Details:</em><pre>
%(tech_footer)s
</pre>
</body></html>
"""

print Template(page).render(content)