irkerd

   1 #!/usr/bin/env python
   2 """
   3 irkerd - a simple IRC multiplexer daemon
   4
   5 Listens for JSON objects of the form {'to':<irc-url>, 'privmsg':<text>}
   6 and relays messages to IRC channels. Each request must be followed by
   7 a newline.
   8
   9 The <text> must be a string.  The value of the 'to' attribute can be a
  10 string containing an IRC URL (e.g. 'irc://chat.freenet.net/botwar') or
  11 a list of such strings; in the latter case the message is broadcast to
  12 all listed channels.  Note that the channel portion of the URL need
  13 *not* have a leading '#' unless the channel name itself does.
  14
  15 Options: -d sets the debug-message level (probably only of interest to
  16 developers). The -V option prints the program version and exits.
  17
  18 Design and code by Eric S. Raymond <esr@thyrsus.com>. See the project
  19 resource page at <http://www.catb.org/~esr/irker/>.
  20
  21 Requires Python 2.6 or 2.5 with the simplejson library installed, and
  22 the irc client library at version >= 2.0.2: see
  23
  24 http://pypi.python.org/pypi/irc/
  25 """
  26 # These things might need tuning
  27
  28 HOST = "localhost"
  29 PORT = 6659
  30
  31 NAMESTYLE = "irker%03d"         # IRC nick template - must contain '%d'
  32 XMIT_TTL = (3 * 60 * 60)        # Time to live, seconds from last transmit
  33 PING_TTL = (15 * 60)            # Time to live, seconds from last PING
  34 HANDSHAKE_TTL = 60              # Time to live, seconds from nick transmit
  35 CHANNEL_TTL = (3 * 60 * 60)     # Time to live, seconds from last transmit
  36 DISCONNECT_TTL = (24 * 60 * 60) # Time to live, seconds from last connect
  37 UNSEEN_TTL = 60                 # Time to live, seconds since first request
  38 CHANNEL_MAX = 18                # Max channels open per socket (default)
  39 ANTI_FLOOD_DELAY = 0.5          # Anti-flood delay after transmissions, seconds
  40 ANTI_BUZZ_DELAY = 0.09          # Anti-buzz delay after queue-empty check
  41 CONNECTION_MAX = 200            # To avoid hitting a thread limit
  42
  43 # No user-serviceable parts below this line
  44
  45 version = "1.10"
  46
  47 import sys, getopt, urlparse, time, random, socket
  48 import threading, Queue, SocketServer
  49 import irc.client, logging
  50 try:
  51     import simplejson as json   # Faster, also makes us Python-2.4-compatible
  52 except ImportError:
  53     import json
  54
  55 # Sketch of implementation:
  56 #
  57 # One Irker object manages multiple IRC sessions.  It holds a map of
  58 # Dispatcher objects, one per (server, port) combination, which are
  59 # responsible for routing messages to one of any number of Connection
  60 # objects that do the actual socket conversations.  The reason for the
  61 # Dispatcher layer is that IRC daemons limit the number of channels a
  62 # client (that is, from the daemon's point of view, a socket) can be
  63 # joined to, so each session to a server needs a flock of Connection
  64 # instances each with its own socket.
  65 #
  66 # Connections are timed out and removed when either they haven't seen a
  67 # PING for a while (indicating that the server may be stalled or down)
  68 # or there has been no message traffic to them for a while, or
  69 # even if the queue is nonempty but efforts to connect have failed for
  70 # a long time.
  71 #
  72 # There are multiple threads. One accepts incoming traffic from all servers.
  73 # Each Connection also has a consumer thread and a thread-safe message queue.
  74 # The program main appends messages to queues as JSON requests are received;
  75 # the consumer threads try to ship them to servers.  When a socket write
  76 # stalls, it only blocks an individual consumer thread; if it stalls long
  77 # enough, the session will be timed out.
  78 #
  79 # Message delivery is thus not reliable in the face of network stalls,
  80 # but this was considered acceptable because IRC (notoriously) has the
  81 # same problem - there is little point in reliable delivery to a relay
  82 # that is down or unreliable.
  83 #
  84 # This code uses only NICK, JOIN, PART, MODE, and PRIVMSG. It is strictly
  85 # compliant to RFC1459, except for the interpretation and use of the
  86 # DEAF and CHANLIMIT and (obsolete) MAXCHANNELS features.  CHANLIMIT
  87 # is as described in the Internet RFC draft
  88 # draft-brocklesby-irc-isupport-03 at <http://www.mirc.com/isupport.html>.
  89 # The ",isnick" feature is as described in
  90 # <http://ftp.ics.uci.edu/pub/ietf/uri/draft-mirashi-url-irc-01.txt>.
  91
  92 class Connection:
  93     def __init__(self, irkerd, servername, port):
  94         self.irker = irkerd
  95         self.servername = servername
  96         self.port = port
  97         self.nick_trial = None
  98         self.connection = None
  99         self.status = None
 100         self.last_xmit = time.time()
 101         self.last_ping = time.time()
 102         self.channels_joined = {}
 103         self.channel_limits = {}
 104         # The consumer thread
 105         self.queue = Queue.Queue()
 106         self.thread = None
 107     def nickname(self, n=None):
 108         "Return a name for the nth server connection."
 109         if n is None:
 110             n = self.nick_trial
 111         return (NAMESTYLE % n)
 112     def handle_ping(self):
 113         "Register the fact that the server has pinged this connection."
 114         self.last_ping = time.time()
 115     def handle_welcome(self):
 116         "The server says we're OK, with a non-conflicting nick."
 117         self.status = "ready"
 118         self.irker.debug(1, "nick %s accepted" % self.nickname())
 119     def handle_badnick(self):
 120         "The server says our nick has a conflict."
 121         self.irker.debug(1, "nick %s rejected" % self.nickname())
 122         # Randomness prevents a malicious user or bot from antcipating the
 123         # next trial name in order to block us from completing the handshake.
 124         self.nick_trial += random.randint(1, 3)
 125         self.last_xmit = time.time()
 126         self.connection.nick(self.nickname())
 127     def handle_disconnect(self):
 128         "Server disconnected us for flooding or some other reason."
 129         self.connection = None
 130         self.status = "disconnected"
 131     def handle_kick(self, outof):
 132         "We've been kicked."
 133         self.status = "handshaking"
 134         try:
 135             del self.channels_joined[outof]
 136         except KeyError:
 137             self.irker.logerr("kicked by %s from %s that's not joined"
 138                               % (self.servername, outof))
 139         qcopy = []
 140         while not self.queue.empty():
 141             (channel, message) = self.queue.get()
 142             if channel != outof:
 143                 qcopy.append((channel, message))
 144         for (channel, message) in qcopy:
 145             self.queue.put((channel, message))
 146         self.status = "ready"
 147     def enqueue(self, channel, message):
 148         "Enque a message for transmission."
 149         if self.thread is None or not self.thread.is_alive():
 150             self.status = "unseen"
 151             self.thread = threading.Thread(target=self.dequeue)
 152             self.thread.setDaemon(True)
 153             self.thread.start()
 154         self.queue.put((channel, message))
 155     def dequeue(self):
 156         "Try to ship pending messages from the queue."
 157         try:
 158             while True:
 159                 # We want to be kind to the IRC servers and not hold unused
 160                 # sockets open forever, so they have a time-to-live.  The
 161                 # loop is coded this particular way so that we can drop
 162                 # the actual server connection when its time-to-live
 163                 # expires, then reconnect and resume transmission if the
 164                 # queue fills up again.
 165                 if self.queue.empty():
 166                     # Queue is empty, at some point we want to time out
 167                     # the connection rather than holding a socket open in
 168                     # the server forever.
 169                     now = time.time()
 170                     xmit_timeout = now > self.last_xmit + XMIT_TTL
 171                     ping_timeout = now > self.last_ping + PING_TTL
 172                     if self.status == "disconnected":
 173                         # If the queue is empty, we can drop this connection.
 174                         self.status = "expired"
 175                         break
 176                     elif xmit_timeout or ping_timeout:
 177                         self.irker.debug(1, "timing out connection to %s at %s (ping_timeout=%s, xmit_timeout=%s)" % (self.servername, time.asctime(), ping_timeout, xmit_timeout))
 178                         with self.irker.library_lock:
 179                             self.connection.context = None
 180                             self.connection.quit("transmission timeout")
 181                             self.connection = None
 182                         self.status = "disconnected"
 183                     else:
 184                         # Prevent this thread from hogging the CPU by pausing
 185                         # for just a little bit after the queue-empty check.
 186                         # As long as this is less that the duration of a human
 187                         # reflex arc it is highly unlikely any human will ever
 188                         # notice.
 189                         time.sleep(ANTI_BUZZ_DELAY)
 190                 elif self.status == "disconnected" \
 191                          and time.time() > self.last_xmit + DISCONNECT_TTL:
 192                     # Queue is nonempty, but the IRC server might be
 193                     # down. Letting failed connections retain queue
 194                     # space forever would be a memory leak.
 195                     self.status = "expired"
 196                     break
 197                 elif not self.connection:
 198                     # Queue is nonempty but server isn't connected.
 199                     with self.irker.library_lock:
 200                         self.connection = self.irker.irc.server()
 201                         self.connection.context = self
 202                         # Try to avoid colliding with other instances
 203                         self.nick_trial = random.randint(1, 990)
 204                         self.channels_joined = {}
 205                         try:
 206                             # This will throw
 207                             # irc.client.ServerConnectionError on failure
 208                             self.connection.connect(self.servername,
 209                                                 self.port,
 210                                                 nickname=self.nickname(),
 211                                                 username="irker",
 212                                                 ircname="irker relaying client")
 213                             self.status = "handshaking"
 214                             self.irker.debug(1, "XMIT_TTL bump (%s connection) at %s" % (self.servername, time.asctime()))
 215                             self.last_xmit = time.time()
 216                             self.last_ping = time.time()
 217                         except irc.client.ServerConnectionError:
 218                             self.status = "disconnected"
 219                 elif self.status == "handshaking":
 220                     if time.time() > self.last_xmit + HANDSHAKE_TTL:
 221                         self.status = "expired"
 222                         break
 223                     else:
 224                         # Don't buzz on the empty-queue test while we're
 225                         # handshaking
 226                         time.sleep(ANTI_BUZZ_DELAY)
 227                 elif self.status == "unseen" \
 228                          and time.time() > self.last_xmit + UNSEEN_TTL:
 229                     # Nasty people could attempt a denial-of-service
 230                     # attack by flooding us with requests with invalid
 231                     # servernames. We guard against this by rapidly
 232                     # expiring connections that have a nonempty queue but
 233                     # have never had a successful open.
 234                     self.status = "expired"
 235                     break
 236                 elif self.status == "ready":
 237                     with self.irker.library_lock:
 238                         (channel, message) = self.queue.get()
 239                         if channel not in self.channels_joined:
 240                             self.connection.join(channel)
 241                             self.irker.debug(1, "joining %s on %s." % (channel, self.servername))
 242                         for segment in message.split("\n"):
 243                             self.connection.privmsg(channel, segment)
 244                             time.sleep(ANTI_FLOOD_DELAY)
 245                         self.last_xmit = self.channels_joined[channel] = time.time()
 246                         self.irker.debug(1, "XMIT_TTL bump (%s transmission) at %s" % (self.servername, time.asctime()))
 247                         self.queue.task_done()
 248         except:
 249             (exc_type, _exc_value, exc_traceback) = sys.exc_info()
 250             self.irker.logerr("exception %s in thread for %s" % \
 251                               (exc_type, self.servername))
 252             # This is so we can see tracebacks for errors inside the thread
 253             # when we need to be able to for debugging purposes.
 254             if debuglvl > 0:
 255                 raise exc_type, _exc_value, exc_traceback
 256     def live(self):
 257         "Should this connection not be scavenged?"
 258         return self.status != "expired"
 259     def joined_to(self, channel):
 260         "Is this connection joined to the specified channel?"
 261         return channel in self.channels_joined
 262     def accepting(self, channel):
 263         "Can this connection accept a join of this channel?"
 264         if self.channel_limits:
 265             match_count = 0
 266             for already in self.channels_joined:
 267                 # This obscure code is because the RFCs allow separate limits
 268                 # by channel type (indicated by the first character of the name)
 269                 # a feature that is almost never actually used.
 270                 if already[0] == channel[0]:
 271                     match_count += 1
 272             return match_count < self.channel_limits.get(channel[0], CHANNEL_MAX)
 273         else:
 274             return len(self.channels_joined) < CHANNEL_MAX
 275
 276 class Target():
 277     "Represent a transmission target."
 278     def __init__(self, url):
 279         parsed = urlparse.urlparse(url)
 280         irchost, _, ircport = parsed.netloc.partition(':')
 281         if not ircport:
 282             ircport = 6667
 283         self.servername = irchost
 284         # IRC channel names are case-insensitive.  If we don't smash
 285         # case here we may run into problems later. There was a bug
 286         # observed on irc.rizon.net where an irkerd user specified #Channel,
 287         # got kicked, and irkerd crashed because the server returned
 288         # "#channel" in the notification that our kick handler saw.
 289         self.channel = parsed.path.lstrip('/').lower()
 290         isnick = self.channel.endswith(",isnick")
 291         if isnick:
 292             self.channel = self.channel[:-7]
 293         if self.channel and not isnick and self.channel[0] not in "#&+":
 294             self.channel = "#" + self.channel
 295         self.port = int(ircport)
 296     def valid(self):
 297         "Both components must be present for a valid target."
 298         return self.servername and self.channel
 299     def server(self):
 300         "Return a hashable tuple representing the destination server."
 301         return (self.servername, self.port)
 302
 303 class Dispatcher:
 304     "Manage connections to a particular server-port combination."
 305     def __init__(self, irkerd, servername, port):
 306         self.irker = irkerd
 307         self.servername = servername
 308         self.port = port
 309         self.connections = []
 310     def dispatch(self, channel, message):
 311         "Dispatch messages for our server-port combination."
 312         # First, check if there is room for another channel
 313         # on any of our existing connections.
 314         connections = [x for x in self.connections if x.live()]
 315         eligibles = [x for x in connections if x.joined_to(channel)] \
 316                     or [x for x in connections if x.accepting(channel)]
 317         if eligibles:
 318             eligibles[0].enqueue(channel, message)
 319             return
 320         # All connections are full up. Look for one old enough to be
 321         # scavenged.
 322         ancients = []
 323         for connection in connections:
 324             for (chan, age) in connections.channels_joined.items():
 325                 if age < time.time() - CHANNEL_TTL:
 326                     ancients.append((connection, chan, age))
 327         if ancients:
 328             ancients.sort(key=lambda x: x[2])
 329             (found_connection, drop_channel, _drop_age) = ancients[0]
 330             found_connection.part(drop_channel, "scavenged by irkerd")
 331             del found_connection.channels_joined[drop_channel]
 332             #time.sleep(ANTI_FLOOD_DELAY)
 333             found_connection.enqueue(channel, message)
 334             return
 335         # Didn't find any channels with no recent activity
 336         newconn = Connection(self.irker,
 337                              self.servername,
 338                              self.port)
 339         self.connections.append(newconn)
 340         newconn.enqueue(channel, message)
 341     def live(self):
 342         "Does this server-port combination have any live connections?"
 343         self.connections = [x for x in self.connections if x.live()]
 344         return len(self.connections) > 0
 345     def last_xmit(self):
 346         "Return the time of the most recent transmission."
 347         return max(x.last_xmit for x in self.connections)
 348
 349 class Irker:
 350     "Persistent IRC multiplexer."
 351     def __init__(self, debuglevel=0):
 352         self.debuglevel = debuglevel
 353         self.irc = irc.client.IRC()
 354         self.irc.add_global_handler("ping", self._handle_ping)
 355         self.irc.add_global_handler("welcome", self._handle_welcome)
 356         self.irc.add_global_handler("erroneusnickname", self._handle_badnick)
 357         self.irc.add_global_handler("nicknameinuse", self._handle_badnick)
 358         self.irc.add_global_handler("nickcollision", self._handle_badnick)
 359         self.irc.add_global_handler("unavailresource", self._handle_badnick)
 360         self.irc.add_global_handler("featurelist", self._handle_features)
 361         self.irc.add_global_handler("disconnect", self._handle_disconnect)
 362         self.irc.add_global_handler("kick", self._handle_kick)
 363         self.library_lock = threading.Lock()
 364         thread = threading.Thread(target=self._process_forever)
 365         thread.setDaemon(True)
 366         self.irc._thread = thread
 367         thread.start()
 368         self.servers = {}
 369     def logerr(self, errmsg):
 370         "Log a processing error."
 371         sys.stderr.write("irkerd: " + errmsg + "\n")
 372     def debug(self, level, errmsg):
 373         "Debugging information."
 374         if self.debuglevel >= level:
 375             sys.stderr.write("irkerd: %s\n" % errmsg)
 376     def _process_forever(self):
 377         "IRC library process_forever with mutex."
 378         self.debug(1, "process_forever()")
 379         while True:
 380             with self.library_lock:
 381                 self.irc.process_once(ANTI_BUZZ_DELAY)
 382     def _handle_ping(self, connection, _event):
 383         "PING arrived, bump the last-received time for the connection."
 384         if connection.context:
 385             connection.context.handle_ping()
 386     def _handle_welcome(self, connection, _event):
 387         "Welcome arrived, nick accepted for this connection."
 388         if connection.context:
 389             connection.context.handle_welcome()
 390     def _handle_badnick(self, connection, _event):
 391         "Nick not accepted for this connection."
 392         if connection.context:
 393             connection.context.handle_badnick()
 394     def _handle_features(self, connection, event):
 395         "Determine if and how we can set deaf mode."
 396         if connection.context:
 397             cxt = connection.context
 398             for lump in event.arguments():
 399                 if lump.startswith("DEAF="):
 400                     connection.mode(cxt.nickname(), "+"+lump[5:])
 401                 elif lump.startswith("MAXCHANNELS="):
 402                     m = int(lump[12:])
 403                     for pref in "#&+":
 404                         cxt.channel_limits[pref] = m
 405                     self.debug(1, "%s maxchannels is %d"
 406                                % (connection.server, m))
 407                 elif lump.startswith("CHANLIMIT=#:"):
 408                     limits = lump[10:].split(",")
 409                     try:
 410                         for token in limits:
 411                             (prefixes, limit) = token.split(":")
 412                             limit = int(limit)
 413                             for c in prefixes:
 414                                 cxt.channel_limits[c] = limit
 415                         self.debug(1, "%s channel limit map is %s"
 416                                    % (connection.server, cxt.channel_limits))
 417                     except ValueError:
 418                         self.logerr("ill-formed CHANLIMIT property")
 419     def _handle_disconnect(self, connection, _event):
 420         "Server hung up the connection."
 421         self.debug(1, "server %s disconnected" % connection.server)
 422         connection.close()
 423         if connection.context:
 424             connection.context.handle_disconnect()
 425     def _handle_kick(self, connection, event):
 426         "Server hung up the connection."
 427         self.debug(1, "irker has been kicked from %s on %s" % (event.target(), connection.server))
 428         if connection.context:
 429             connection.context.handle_kick(event.target())
 430     def handle(self, line):
 431         "Perform a JSON relay request."
 432         try:
 433             request = json.loads(line.strip())
 434             if not isinstance(request, dict):
 435                 self.logerr("request is not a JSON dictionary: %r" % request)
 436             elif "to" not in request or "privmsg" not in request:
 437                 self.logerr("malformed request - 'to' or 'privmsg' missing: %r" % request)
 438             else:
 439                 channels = request['to']
 440                 message = request['privmsg']
 441                 if not isinstance(channels, (list, basestring)):
 442                     self.logerr("malformed request - unexpected channel type: %r" % channels)
 443                 if not isinstance(message, basestring):
 444                     self.logerr("malformed request - unexpected message type: %r" % message)
 445                 else:
 446                     if not isinstance(channels, list):
 447                         channels = [channels]
 448                     for url in channels:
 449                         if not isinstance(url, basestring):
 450                             self.logerr("malformed request - URL has unexpected type: %r" % url)
 451                         else:
 452                             target = Target(url)
 453                             if not target.valid():
 454                                 return
 455                             if target.server() not in self.servers:
 456                                 self.servers[target.server()] = Dispatcher(self, target.servername, target.port)
 457                             self.servers[target.server()].dispatch(target.channel, message)
 458                             # GC dispatchers with no active connections
 459                             servernames = self.servers.keys()
 460                             for servername in servernames:
 461                                 if not self.servers[servername].live():
 462                                     del self.servers[servername]
 463                             # If we might be pushing a resource limit
 464                             # even after garbage collection, remove a
 465                             # session.  The goal here is to head off
 466                             # DoS attacks that aim at exhausting
 467                             # thread space or file descriptors.  The
 468                             # cost is that attempts to DoS this
 469                             # service will cause lots of join/leave
 470                             # spam as we scavenge old channels after
 471                             # connecting to new ones. The particular
 472                             # method used for selecting a session to
 473                             # be terminated doesn't matter much; we
 474                             # choose the one longest idle on the
 475                             # assumption that message activity is likely
 476                             # to be clumpy.
 477                             if len(self.servers) >= CONNECTION_MAX:
 478                                 oldest = min(self.servers.keys(), key=lambda name: self.servers[name].last_xmit())
 479                                 del self.servers[oldest]
 480         except ValueError:
 481             self.logerr("can't recognize JSON on input: %r" % line)
 482         except RuntimeError:
 483             self.logerr("wildly malformed JSON blew the parser stack.")
 484
 485 class IrkerTCPHandler(SocketServer.StreamRequestHandler):
 486     def handle(self):
 487         while True:
 488             line = self.rfile.readline()
 489             if not line:
 490                 break
 491             irker.handle(line.strip())
 492
 493 class IrkerUDPHandler(SocketServer.BaseRequestHandler):
 494     def handle(self):
 495         data = self.request[0].strip()
 496         #socket = self.request[1]
 497         irker.handle(data)
 498
 499 if __name__ == '__main__':
 500     debuglvl = 0
 501     (options, arguments) = getopt.getopt(sys.argv[1:], "d:V")
 502     for (opt, val) in options:
 503         if opt == '-d':         # Enable debug/progress messages
 504             debuglvl = int(val)
 505             if debuglvl > 1:
 506                 logging.basicConfig(level=logging.DEBUG)
 507         elif opt == '-V':       # Emit version and exit
 508             sys.stdout.write("irkerd version %s\n" % version)
 509             sys.exit(0)
 510     irker = Irker(debuglevel=debuglvl)
 511     irker.debug(1, "irkerd version %s" % version)
 512     try:
 513         tcpserver = SocketServer.TCPServer((HOST, PORT), IrkerTCPHandler)
 514         udpserver = SocketServer.UDPServer((HOST, PORT), IrkerUDPHandler)
 515         for server in [tcpserver, udpserver]:
 516             server = threading.Thread(target=server.serve_forever)
 517             server.setDaemon(True)
 518             server.start()
 519         try:
 520             while True:
 521                 time.sleep(10)
 522         except KeyboardInterrupt:
 523             raise SystemExit(1)
 524     except socket.error, e:
 525         sys.stderr.write("irkerd: server launch failed: %r\n" % e)
 526
 527 # end