source download plumbed in

[hippotat] / hippotatd

#!/usr/bin/python3

from hippotatlib import *

import os
import tempfile
import atexit
import shutil

import twisted.internet
from twisted.web.server import NOT_DONE_YET

import twisted.web.static

import hippotatlib.ownsource
from hippotatlib.ownsource import SourceShipmentPreparer

#import twisted.web.server import Site
#from twisted.web.resource import Resource

import syslog

cleanups = [ ]

clients = { }

#---------- "router" ----------

def route(packet, iface, saddr, daddr):
  def lt(dest):
    log_debug(DBG.ROUTE, 'route: %s -> %s: %s' % (saddr,daddr,dest), d=packet)
  try: dclient = clients[daddr]
  except KeyError: dclient = None
  if dclient is not None:
    lt('client')
    dclient.queue_outbound(packet)
  elif daddr == c.vaddr or daddr not in c.vnetwork:
    lt('inbound')
    queue_inbound(ipif, packet)
  elif daddr == c.relay:
    lt('discard relay')
    log_discard(packet, iface, saddr, daddr, 'relay')
  else:
    lt('discard no-client')
    log_discard(packet, iface, saddr, daddr, 'no-client')

#---------- client ----------

class Client():
  def __init__(self, ip, cc):
    # instance data members
    self._ip = ip
    self.cc = cc
    self._rq = collections.deque() # requests
    self._pq = PacketQueue(str(ip), self.cc.max_queue_time)

    if ip not in c.vnetwork:
      raise ValueError('client %s not in vnetwork' % ip)

    if ip in clients:
      raise ValueError('multiple client cfg sections for %s' % ip)
    clients[ip] = self

    self._log(DBG.INIT, 'new')

  def _log(self, dflag, msg, **kwargs):
    log_debug(dflag, ('client %s: ' % self._ip)+msg, **kwargs)

  def process_arriving_data(self, d):
    self._log(DBG.FLOW, "req data (enc'd)", d=d)
    if not len(d): return
    for packet in slip.decode(d):
      (saddr, daddr) = packet_addrs(packet)
      if saddr != self._ip:
        raise ValueError('wrong source address %s' % saddr)
      route(packet, self._ip, saddr, daddr)

  def _req_cancel(self, request):
    self._log(DBG.HTTP_CTRL, 'cancel', idof=request)
    request.finish()

  def _req_error(self, err, request):
    self._log(DBG.HTTP_CTRL, 'error %s' % err, idof=request)
    self._req_cancel(request)

  def queue_outbound(self, packet):
    self._pq.append(packet)
    self._check_outbound()

  def _req_fin(self, dummy, request, cl):
    self._log(DBG.HTTP_CTRL, '_req_fin ' + repr(dummy), idof=request)
    try: cl.cancel()
    except twisted.internet.error.AlreadyCalled: pass

  def new_request(self, request):
    request.setHeader('Content-Type','application/octet-stream')
    cl = reactor.callLater(self.cc.http_timeout, self._req_cancel, request)
    nf = request.notifyFinish()
    nf.addErrback(self._req_error, request)
    nf.addCallback(self._req_fin, request, cl)
    self._rq.append(request)
    self._check_outbound()

  def _req_write(self, req, d):
    self._log(DBG.HTTP, 'req_write ', idof=req, d=d)
    req.write(d)

  def _check_outbound(self):
    log_debug(DBG.HTTP_CTRL, 'CHKO')
    while True:
      try: request = self._rq[0]
      except IndexError: request = None
      if request and request.finished:
        self._log(DBG.HTTP_CTRL, 'CHKO req finished, discard', idof=request)
        self._rq.popleft()
        continue

      if not self._pq.nonempty():
        # no packets, oh well
        self._log(DBG.HTTP_CTRL, 'CHKO no packets, OUT-DONE', idof=request)
        break

      if request is None:
        # no request
        self._log(DBG.HTTP_CTRL, 'CHKO no request, OUT-DONE', idof=request)
        break

      self._log(DBG.HTTP_CTRL, 'CHKO processing', idof=request)
      # request, and also some non-expired packets
      self._pq.process((lambda: request.sentLength),
                       (lambda d: self._req_write(request, d)),
                       self.cc.max_batch_down)

      assert(request.sentLength)
      self._rq.popleft()
      request.finish()
      self._log(DBG.HTTP, 'complete', idof=request)
      # round again, looking for more to do

    while len(self._rq) > self.cc.target_requests_outstanding:
      request = self._rq.popleft()
      self._log(DBG.HTTP, 'CHKO above target, returning empty', idof=request)
      request.finish()

def process_request(request, desca):
  # find client, update config, etc.
  metadata = request.args[b'm'][0]
  metadata = metadata.split(b'\r\n')
  (ci_s, pw, tro, cto) = metadata[0:4]
  desca['m[0,2:3]'] = [ci_s, tro, cto]
  ci_s = ci_s.decode('utf-8')
  tro = int(tro); desca['tro']= tro
  cto = int(cto); desca['cto']= cto
  ci = ipaddr(ci_s)
  desca['ci'] = ci
  cl = clients[ci]
  if pw != cl.cc.password: raise ValueError('bad password')
  desca['pwok']=True

  if tro != cl.cc.target_requests_outstanding:
    raise ValueError('tro must be %d' % cl.cc.target_requests_outstanding)

  if cto < cl.cc.http_timeout:
    raise ValueError('cto must be >= %d' % cl.cc.http_timeout)

  try:
    d = request.args[b'd'][0]
    desca['d'] = d
    desca['dlen'] = len(d)
  except KeyError:
    d = b''
    desca['dlen'] = None

  log_http(desca, 'processing', idof=id(request), d=d)

  d = mime_translate(d)

  cl.process_arriving_data(d)
  cl.new_request(request)

def log_http(desca, msg, **kwargs):
  try:
    kwargs['d'] = desca['d']
    del desca['d']
  except KeyError:
    pass
  log_debug(DBG.HTTP, msg + repr(desca), **kwargs)

class NotStupidResource(twisted.web.resource.Resource):
  # why this is not the default is a mystery!
  def getChild(self, name, request):
    if name == b'': return self
    else: return twisted.web.resource.Resource.getChild(name, request)

class IphttpResource(NotStupidResource):
  def render_POST(self, request):
    log_debug(DBG.HTTP_FULL,
              'req recv: ' + repr(request) + ' ' + repr(request.args),
              idof=id(request))
    desca = {'d': None}
    try: process_request(request, desca)
    except Exception as e:
      emsg = traceback.format_exc()
      log_http(desca, 'RETURNING EXCEPTION ' + emsg)
      request.setHeader('Content-Type','text/plain; charset="utf-8"')
      request.setResponseCode(400)
      return (emsg + ' # ' + repr(desca) + '\r\n').encode('utf-8')
    log_debug(DBG.HTTP_CTRL, '...', idof=id(request))
    return NOT_DONE_YET

  def render_GET(self, request):
    log_debug(DBG.HTTP, 'GET request')
    return b'''
<html><body>
hippotat
<p>
<a href="source">source</a> available
</body></html>
'''

class SourceResource(twisted.web.static.File):
  def __init__(self):
    td = tempfile.mkdtemp()

    def cleanup():
      try: shutil.rmtree(td)
      except FileNotFoundError: pass

    cleanups.append(cleanup)

    self._ssp = SourceShipmentPreparer(td)
    self._ssp.logger = self.log
    self._ssp.generate()

    super().__init__(self._ssp.output_path)

  def log(self, m):
    log_debug(DBG.OWNSOURCE, m)

def start_http():
  resource = IphttpResource()
  resource.putChild(b'source',SourceResource())
  site = twisted.web.server.Site(resource)
  for sa in c.saddrs:
    ep = sa.make_endpoint()
    crash_on_defer(ep.listen(site))
    log_debug(DBG.INIT, 'listening on %s' % sa)

#---------- config and setup ----------

def process_cfg(putative_servers, putative_clients):
  global c
  c = ConfigResults()
  c.server = cfg.get('SERVER','server')

  cfg_process_common(c, c.server)
  cfg_process_saddrs(c, c.server)
  cfg_process_vnetwork(c, c.server)
  cfg_process_vaddr(c, c.server)

  for (ci,cs) in putative_clients.items():
    cc = ConfigResults()
    sections = cfg_process_client_common(cc,c.server,cs,ci)
    if not sections: continue
    cfg_process_client_limited(cc,c.server,sections, 'max_batch_down')
    cfg_process_client_limited(cc,c.server,sections, 'max_queue_time')
    Client(ci, cc)

  try:
    c.vrelay = cfg.get(c.server, 'vrelay')
  except NoOptionError:
    for search in c.vnetwork.hosts():
      if search == c.vaddr: continue
      c.vrelay = search
      break

  cfg_process_ipif(c,
                   [c.server, 'DEFAULT'],
                   (('local','vaddr'),
                    ('peer', 'vrelay'),
                    ('rnets','vnetwork')))

def catch_termination():
  def run_cleanups():
    for cleanup in cleanups:
      cleanup()

  atexit.register(run_cleanups)

  def signal_handler(name, sig, *args):
    signal.signal(sig, signal.SIG_DFL)
    print('exiting due to %s' % name, file=sys.stderr)
    run_cleanups()
    os.kill(os.getpid(), sig)
    raise RuntimeError('did not die due to signal %s !' % name)

  for sig in (signal.SIGINT, signal.SIGTERM):
    signal.signal(sig, partial(signal_handler, sig.name))

common_startup(process_cfg)
catch_termination()
ipif = start_ipif(c.ipif_command, (lambda p,s,d: route(p,"[ipif]",s,d)))
start_http()
common_run()