mdw@git.distorted.org.uk Git - hippotat/blob - hippotatlib/__init__.py

   1 # -*- python -*-
   2 #
   3 # Hippotat - Asinine IP Over HTTP program
   4 # hippotatlib/__init__.py - common library code
   5 #
   6 # Copyright 2017 Ian Jackson
   7 #
   8 # GPLv3+
   9 #
  10 #    This program is free software: you can redistribute it and/or modify
  11 #    it under the terms of the GNU General Public License as published by
  12 #    the Free Software Foundation, either version 3 of the License, or
  13 #    (at your option) any later version.
  14 #
  15 #    This program is distributed in the hope that it will be useful,
  16 #    but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 #    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 #    GNU General Public License for more details.
  19 #
  20 #    You should have received a copy of the GNU General Public License
  21 #    along with this program, in the file GPLv3.  If not,
  22 #    see <http://www.gnu.org/licenses/>.
  23
  24
  25 import signal
  26 signal.signal(signal.SIGINT, signal.SIG_DFL)
  27
  28 import sys
  29 import os
  30
  31 from zope.interface import implementer
  32
  33 import twisted
  34 from twisted.internet import reactor
  35 import twisted.internet.endpoints
  36 import twisted.logger
  37 from twisted.logger import LogLevel
  38 import twisted.python.constants
  39 from twisted.python.constants import NamedConstant
  40
  41 import ipaddress
  42 from ipaddress import AddressValueError
  43
  44 from optparse import OptionParser
  45 import configparser
  46 from configparser import ConfigParser
  47 from configparser import NoOptionError
  48
  49 from functools import partial
  50
  51 import collections
  52 import time
  53 import codecs
  54 import traceback
  55
  56 import re as regexp
  57
  58 import hippotatlib.slip as slip
  59
  60 class DBG(twisted.python.constants.Names):
  61   INIT = NamedConstant()
  62   CONFIG = NamedConstant()
  63   ROUTE = NamedConstant()
  64   DROP = NamedConstant()
  65   OWNSOURCE = NamedConstant()
  66   FLOW = NamedConstant()
  67   HTTP = NamedConstant()
  68   TWISTED = NamedConstant()
  69   QUEUE = NamedConstant()
  70   HTTP_CTRL = NamedConstant()
  71   QUEUE_CTRL = NamedConstant()
  72   HTTP_FULL = NamedConstant()
  73   CTRL_DUMP = NamedConstant()
  74   SLIP_FULL = NamedConstant()
  75   DATA_COMPLETE = NamedConstant()
  76
  77 _hex_codec = codecs.getencoder('hex_codec')
  78
  79 #---------- logging ----------
  80
  81 org_stderr = sys.stderr
  82
  83 log = twisted.logger.Logger()
  84
  85 debug_set = set()
  86 debug_def_detail = DBG.HTTP
  87
  88 def log_debug(dflag, msg, idof=None, d=None):
  89   if dflag not in debug_set: return
  90   #print('---------------->',repr((dflag, msg, idof, d)), file=sys.stderr)
  91   if idof is not None:
  92     msg = '[%#x] %s' % (id(idof), msg)
  93   if d is not None:
  94     trunc = ''
  95     if not DBG.DATA_COMPLETE in debug_set:
  96       if len(d) > 64:
  97         d = d[0:64]
  98         trunc = '...'
  99     d = _hex_codec(d)[0].decode('ascii')
 100     msg += ' ' + d + trunc
 101   log.info('{dflag} {msgcore}', dflag=dflag, msgcore=msg)
 102
 103 def logevent_is_boringtwisted(event):
 104   try:
 105     if event.get('log_level') != LogLevel.info:
 106       return False
 107     dflag = event.get('dflag')
 108     if dflag is False                            : return False
 109     if dflag                         in debug_set: return False
 110     if dflag is None and DBG.TWISTED in debug_set: return False
 111     return True
 112   except Exception:
 113     print('EXCEPTION (IN BORINGTWISTED CHECK)',
 114           traceback.format_exc(), file=org_stderr)
 115     return False
 116
 117 @implementer(twisted.logger.ILogFilterPredicate)
 118 class LogNotBoringTwisted:
 119   def __call__(self, event):
 120     return (
 121       twisted.logger.PredicateResult.no
 122       if logevent_is_boringtwisted(event) else
 123       twisted.logger.PredicateResult.yes
 124     )
 125
 126 #---------- default config ----------
 127
 128 defcfg = '''
 129 [DEFAULT]
 130 max_batch_down = 65536
 131 max_queue_time = 10
 132 target_requests_outstanding = 3
 133 http_timeout = 30
 134 http_timeout_grace = 5
 135 max_requests_outstanding = 6
 136 max_batch_up = 4000
 137 http_retry = 5
 138 port = 80
 139 vroutes = ''
 140 ifname_client = hippo%%d
 141 ifname_server = shippo%%d
 142
 143 #[server] or [<client>] overrides
 144 ipif = userv root ipif %(local)s,%(peer)s,%(mtu)s,slip,%(ifname)s %(rnets)s
 145
 146 # relating to virtual network
 147 mtu = 1500
 148
 149 # addrs = 127.0.0.1 ::1
 150 # url
 151
 152 # relating to virtual network
 153 vvnetwork = 172.24.230.192
 154 # vnetwork = <prefix>/<len>
 155 # vaddr    = <ipaddr>
 156 # vrelay   = <ipaddr>
 157
 158
 159 # [<client-ip4-or-ipv6-address>]
 160 # password = <password>    # used by both, must match
 161
 162 [LIMIT]
 163 max_batch_down = 262144
 164 max_queue_time = 121
 165 http_timeout = 121
 166 target_requests_outstanding = 10
 167 '''
 168
 169 # these need to be defined here so that they can be imported by import *
 170 cfg = ConfigParser(strict=False)
 171 optparser = OptionParser()
 172
 173 _mimetrans = bytes.maketrans(b'-'+slip.esc, slip.esc+b'-')
 174 def mime_translate(s):
 175   # SLIP-encoded packets cannot contain ESC ESC.
 176   # Swap `-' and ESC.  The result cannot contain `--'
 177   return s.translate(_mimetrans)
 178
 179 class ConfigResults:
 180   def __init__(self):
 181     pass
 182   def __repr__(self):
 183     return 'ConfigResults('+repr(self.__dict__)+')'
 184
 185 def log_discard(packet, iface, saddr, daddr, why):
 186   log_debug(DBG.DROP,
 187             'discarded packet [%s] %s -> %s: %s' % (iface, saddr, daddr, why),
 188             d=packet)
 189
 190 #---------- packet parsing ----------
 191
 192 def packet_addrs(packet):
 193   version = packet[0] >> 4
 194   if version == 4:
 195     addrlen = 4
 196     saddroff = 3*4
 197     factory = ipaddress.IPv4Address
 198   elif version == 6:
 199     addrlen = 16
 200     saddroff = 2*4
 201     factory = ipaddress.IPv6Address
 202   else:
 203     raise ValueError('unsupported IP version %d' % version)
 204   saddr = factory(packet[ saddroff           : saddroff + addrlen   ])
 205   daddr = factory(packet[ saddroff + addrlen : saddroff + addrlen*2 ])
 206   return (saddr, daddr)
 207
 208 #---------- address handling ----------
 209
 210 def ipaddr(input):
 211   try:
 212     r = ipaddress.IPv4Address(input)
 213   except AddressValueError:
 214     r = ipaddress.IPv6Address(input)
 215   return r
 216
 217 def ipnetwork(input):
 218   try:
 219     r = ipaddress.IPv4Network(input)
 220   except NetworkValueError:
 221     r = ipaddress.IPv6Network(input)
 222   return r
 223
 224 #---------- ipif (SLIP) subprocess ----------
 225
 226 class SlipStreamDecoder():
 227   def __init__(self, desc, on_packet):
 228     self._buffer = b''
 229     self._on_packet = on_packet
 230     self._desc = desc
 231     self._log('__init__')
 232
 233   def _log(self, msg, **kwargs):
 234     log_debug(DBG.SLIP_FULL, 'slip %s: %s' % (self._desc, msg), **kwargs)
 235
 236   def inputdata(self, data):
 237     self._log('inputdata', d=data)
 238     data = self._buffer + data
 239     self._buffer = b''
 240     packets = slip.decode(data, True)
 241     self._buffer = packets.pop()
 242     for packet in packets:
 243       self._maybe_packet(packet)
 244     self._log('bufremain', d=self._buffer)
 245
 246   def _maybe_packet(self, packet):
 247     self._log('maybepacket', d=packet)
 248     if len(packet):
 249       self._on_packet(packet)
 250
 251   def flush(self):
 252     self._log('flush')
 253     data = self._buffer
 254     self._buffer = b''
 255     packets = slip.decode(data)
 256     assert(len(packets) == 1)
 257     self._maybe_packet(packets[0])
 258
 259 class _IpifProcessProtocol(twisted.internet.protocol.ProcessProtocol):
 260   def __init__(self, router):
 261     self._router = router
 262     self._decoder = SlipStreamDecoder('ipif', self.slip_on_packet)
 263   def connectionMade(self): pass
 264   def outReceived(self, data):
 265     self._decoder.inputdata(data)
 266   def slip_on_packet(self, packet):
 267     (saddr, daddr) = packet_addrs(packet)
 268     if saddr.is_link_local or daddr.is_link_local:
 269       log_discard(packet, 'ipif', saddr, daddr, 'link-local')
 270       return
 271     self._router(packet, saddr, daddr)
 272   def processEnded(self, status):
 273     status.raiseException()
 274
 275 def start_ipif(command, router):
 276   ipif = _IpifProcessProtocol(router)
 277   reactor.spawnProcess(ipif,
 278                        '/bin/sh',['sh','-xc', command],
 279                        childFDs={0:'w', 1:'r', 2:2},
 280                        env=None)
 281   return ipif
 282
 283 def queue_inbound(ipif, packet):
 284   log_debug(DBG.FLOW, "queue_inbound", d=packet)
 285   ipif.transport.write(slip.delimiter)
 286   ipif.transport.write(slip.encode(packet))
 287   ipif.transport.write(slip.delimiter)
 288
 289 #---------- packet queue ----------
 290
 291 class PacketQueue():
 292   def __init__(self, desc, max_queue_time):
 293     self._desc = desc
 294     assert(desc + '')
 295     self._max_queue_time = max_queue_time
 296     self._pq = collections.deque() # packets
 297
 298   def _log(self, dflag, msg, **kwargs):
 299     log_debug(dflag, self._desc+' pq: '+msg, **kwargs)
 300
 301   def append(self, packet):
 302     self._log(DBG.QUEUE, 'append', d=packet)
 303     self._pq.append((time.monotonic(), packet))
 304
 305   def nonempty(self):
 306     self._log(DBG.QUEUE, 'nonempty ?')
 307     while True:
 308       try: (queuetime, packet) = self._pq[0]
 309       except IndexError:
 310         self._log(DBG.QUEUE, 'nonempty ? empty.')
 311         return False
 312
 313       age = time.monotonic() - queuetime
 314       if age > self._max_queue_time:
 315         # strip old packets off the front
 316         self._log(DBG.QUEUE, 'dropping (old)', d=packet)
 317         self._pq.popleft()
 318         continue
 319
 320       self._log(DBG.QUEUE, 'nonempty ? nonempty.')
 321       return True
 322
 323   def process(self, sizequery, moredata, max_batch):
 324     # sizequery() should return size of batch so far
 325     # moredata(s) should add s to batch
 326     self._log(DBG.QUEUE, 'process...')
 327     while True:
 328       try: (dummy, packet) = self._pq[0]
 329       except IndexError:
 330         self._log(DBG.QUEUE, 'process... empty')
 331         break
 332
 333       self._log(DBG.QUEUE_CTRL, 'process... packet', d=packet)
 334
 335       encoded = slip.encode(packet)
 336       sofar = sizequery()
 337
 338       self._log(DBG.QUEUE_CTRL,
 339                 'process... (sofar=%d, max=%d) encoded' % (sofar, max_batch),
 340                 d=encoded)
 341
 342       if sofar > 0:
 343         if sofar + len(slip.delimiter) + len(encoded) > max_batch:
 344           self._log(DBG.QUEUE_CTRL, 'process... overflow')
 345           break
 346         moredata(slip.delimiter)
 347
 348       moredata(encoded)
 349       self._pq.popleft()
 350
 351 #---------- error handling ----------
 352
 353 _crashing = False
 354
 355 def crash(err):
 356   global _crashing
 357   _crashing = True
 358   print('========== CRASH ==========', err,
 359         '===========================', file=sys.stderr)
 360   try: reactor.stop()
 361   except twisted.internet.error.ReactorNotRunning: pass
 362
 363 def crash_on_defer(defer):
 364   defer.addErrback(lambda err: crash(err))
 365
 366 def crash_on_critical(event):
 367   if event.get('log_level') >= LogLevel.critical:
 368     crash(twisted.logger.formatEvent(event))
 369
 370 #---------- config processing ----------
 371
 372 def _cfg_process_putatives():
 373   servers = { }
 374   clients = { }
 375   # maps from abstract object to canonical name for cs's
 376
 377   def putative(cmap, abstract, canoncs):
 378     try:
 379       current_canoncs = cmap[abstract]
 380     except KeyError:
 381       pass
 382     else:
 383       assert(current_canoncs == canoncs)
 384     cmap[abstract] = canoncs
 385
 386   server_pat = r'[-.0-9A-Za-z]+'
 387   client_pat = r'[.:0-9a-f]+'
 388   server_re = regexp.compile(server_pat)
 389   serverclient_re = regexp.compile(server_pat + r' ' + client_pat)
 390
 391   for cs in cfg.sections():
 392     if cs == 'LIMIT':
 393       # plan A "[LIMIT]"
 394       continue
 395
 396     try:
 397       # plan B "[<client>]" part 1
 398       ci = ipaddr(cs)
 399     except AddressValueError:
 400
 401       if server_re.fullmatch(cs):
 402         # plan C "[<servername>]"
 403         putative(servers, cs, cs)
 404         continue
 405
 406       if serverclient_re.fullmatch(cs):
 407         # plan D "[<servername> <client>]" part 1
 408         (pss,pcs) = cs.split(' ')
 409
 410         if pcs == 'LIMIT':
 411           # plan E "[<servername> LIMIT]"
 412           continue
 413
 414         try:
 415           # plan D "[<servername> <client>]" part 2
 416           ci = ipaddr(pc)
 417         except AddressValueError:
 418           # plan F "[<some thing we do not understand>]"
 419           # well, we ignore this
 420           print('warning: ignoring config section %s' % cs, file=sys.stderr)
 421           continue
 422
 423         else: # no AddressValueError
 424           # plan D "[<servername> <client]" part 3
 425           putative(clients, ci, pcs)
 426           putative(servers, pss, pss)
 427           continue
 428
 429     else: # no AddressValueError
 430       # plan B "[<client>" part 2
 431       putative(clients, ci, cs)
 432       continue
 433
 434   return (servers, clients)
 435
 436 def cfg_process_common(c, ss):
 437   c.mtu = cfg.getint(ss, 'mtu')
 438
 439 def cfg_process_saddrs(c, ss):
 440   class ServerAddr():
 441     def __init__(self, port, addrspec):
 442       self.port = port
 443       # also self.addr
 444       try:
 445         self.addr = ipaddress.IPv4Address(addrspec)
 446         self._endpointfactory = twisted.internet.endpoints.TCP4ServerEndpoint
 447         self._inurl = b'%s'
 448       except AddressValueError:
 449         self.addr = ipaddress.IPv6Address(addrspec)
 450         self._endpointfactory = twisted.internet.endpoints.TCP6ServerEndpoint
 451         self._inurl = b'[%s]'
 452     def make_endpoint(self):
 453       return self._endpointfactory(reactor, self.port,
 454                                    interface= '%s' % self.addr)
 455     def url(self):
 456       url = b'http://' + (self._inurl % str(self.addr).encode('ascii'))
 457       if self.port != 80: url += b':%d' % self.port
 458       url += b'/'
 459       return url
 460     def __repr__(self):
 461       return 'ServerAddr'+repr((self.port,self.addr))
 462
 463   c.port = cfg.getint(ss,'port')
 464   c.saddrs = [ ]
 465   for addrspec in cfg.get(ss, 'addrs').split():
 466     sa = ServerAddr(c.port, addrspec)
 467     c.saddrs.append(sa)
 468
 469 def cfg_process_vnetwork(c, ss):
 470   c.vnetwork = ipnetwork(cfg.get(ss,'vnetwork'))
 471   if c.vnetwork.num_addresses < 3 + 2:
 472     raise ValueError('vnetwork needs at least 2^3 addresses')
 473
 474 def cfg_process_vaddr(c, ss):
 475   try:
 476     c.vaddr = cfg.get(ss,'vaddr')
 477   except NoOptionError:
 478     cfg_process_vnetwork(c, ss)
 479     c.vaddr = next(c.vnetwork.hosts())
 480
 481 def cfg_search_section(key,sections):
 482   for section in sections:
 483     if cfg.has_option(section, key):
 484       return section
 485   raise NoOptionError(key, repr(sections))
 486
 487 def cfg_search(getter,key,sections):
 488   section = cfg_search_section(key,sections)
 489   return getter(section, key)
 490
 491 def cfg_process_client_limited(cc,ss,sections,key):
 492   val = cfg_search(cfg.getint, key, sections)
 493   lim = cfg_search(cfg.getint, key, ['%s LIMIT' % ss, 'LIMIT'])
 494   cc.__dict__[key] = min(val,lim)
 495
 496 def cfg_process_client_common(cc,ss,cs,ci):
 497   # returns sections to search in, iff password is defined, otherwise None
 498   cc.ci = ci
 499
 500   sections = ['%s %s' % (ss,cs),
 501               cs,
 502               ss,
 503               'DEFAULT']
 504
 505   try: pwsection = cfg_search_section('password', sections)
 506   except NoOptionError: return None
 507
 508   pw = cfg.get(pwsection, 'password')
 509   cc.password = pw.encode('utf-8')
 510
 511   cfg_process_client_limited(cc,ss,sections,'target_requests_outstanding')
 512   cfg_process_client_limited(cc,ss,sections,'http_timeout')
 513
 514   return sections
 515
 516 def cfg_process_ipif(c, sections, varmap):
 517   for d, s in varmap:
 518     try: v = getattr(c, s)
 519     except AttributeError: continue
 520     setattr(c, d, v)
 521
 522   #print('CFGIPIF',repr((varmap, sections, c.__dict__)),file=sys.stderr)
 523
 524   section = cfg_search_section('ipif', sections)
 525   c.ipif_command = cfg.get(section,'ipif', vars=c.__dict__)
 526
 527 #---------- startup ----------
 528
 529 def log_debug_config(m):
 530   if not DBG.CONFIG in debug_set: return
 531   print('DBG.CONFIG:', m)
 532
 533 def common_startup(process_cfg):
 534   # calls process_cfg(putative_clients, putative_servers)
 535
 536   # ConfigParser hates #-comments after values
 537   trailingcomments_re = regexp.compile(r'#.*')
 538   cfg.read_string(trailingcomments_re.sub('', defcfg))
 539   need_defcfg = True
 540
 541   def readconfig(pathname, mandatory=True):
 542     def log(m, p=pathname):
 543       if not DBG.CONFIG in debug_set: return
 544       log_debug_config('%s: %s' % (m, p))
 545
 546     try:
 547       files = os.listdir(pathname)
 548
 549     except FileNotFoundError:
 550       if mandatory: raise
 551       log('skipped')
 552       return
 553
 554     except NotADirectoryError:
 555       cfg.read(pathname)
 556       log('read file')
 557       return
 558
 559     # is a directory
 560     log('directory')
 561     re = regexp.compile('[^-A-Za-z0-9_]')
 562     for f in os.listdir(pathname):
 563       if re.search(f): continue
 564       subpath = pathname + '/' + f
 565       try:
 566         os.stat(subpath)
 567       except FileNotFoundError:
 568         log('entry skipped', subpath)
 569         continue
 570       cfg.read(subpath)
 571       log('entry read', subpath)
 572
 573   def oc_config(od,os, value, op):
 574     nonlocal need_defcfg
 575     need_defcfg = False
 576     readconfig(value)
 577
 578   def oc_extra_config(od,os, value, op):
 579     readconfig(value)
 580
 581   def read_defconfig():
 582     readconfig('/etc/hippotat/config.d', False)
 583     readconfig('/etc/hippotat/passwords.d', False)
 584     readconfig('/etc/hippotat/master.cfg',   False)
 585
 586   def oc_defconfig(od,os, value, op):
 587     nonlocal need_defcfg
 588     need_defcfg = False
 589     read_defconfig(value)
 590
 591   def dfs_less_detailed(dl):
 592     return [df for df in DBG.iterconstants() if df <= dl]
 593
 594   def ds_default(od,os,dl,op):
 595     global debug_set
 596     debug_set.clear
 597     debug_set |= set(dfs_less_detailed(debug_def_detail))
 598
 599   def ds_select(od,os, spec, op):
 600     for it in spec.split(','):
 601
 602       if it.startswith('-'):
 603         mutator = debug_set.discard
 604         it = it[1:]
 605       else:
 606         mutator = debug_set.add
 607
 608       if it == '+':
 609         dfs = DBG.iterconstants()
 610
 611       else:
 612         if it.endswith('+'):
 613           mapper = dfs_less_detailed
 614           it = it[0:len(it)-1]
 615         else:
 616           mapper = lambda x: [x]
 617
 618           try:
 619             dfspec = DBG.lookupByName(it)
 620           except ValueError:
 621             optparser.error('unknown debug flag %s in --debug-select' % it)
 622
 623         dfs = mapper(dfspec)
 624
 625       for df in dfs:
 626         mutator(df)
 627
 628   optparser.add_option('-D', '--debug',
 629                        nargs=0,
 630                        action='callback',
 631                        help='enable default debug (to stdout)',
 632                        callback= ds_default)
 633
 634   optparser.add_option('--debug-select',
 635                        nargs=1,
 636                        type='string',
 637                        metavar='[-]DFLAG[+]|[-]+,...',
 638                        help=
 639 '''enable (`-': disable) each specified DFLAG;
 640 `+': do same for all "more interesting" DFLAGSs;
 641 just `+': all DFLAGs.
 642   DFLAGS: ''' + ' '.join([df.name for df in DBG.iterconstants()]),
 643                        action='callback',
 644                        callback= ds_select)
 645
 646   optparser.add_option('-c', '--config',
 647                        nargs=1,
 648                        type='string',
 649                        metavar='CONFIGFILE',
 650                        dest='configfile',
 651                        action='callback',
 652                        callback= oc_config)
 653
 654   optparser.add_option('--extra-config',
 655                        nargs=1,
 656                        type='string',
 657                        metavar='CONFIGFILE',
 658                        dest='configfile',
 659                        action='callback',
 660                        callback= oc_extra_config)
 661
 662   optparser.add_option('--default-config',
 663                        action='callback',
 664                        callback= oc_defconfig)
 665
 666   (opts, args) = optparser.parse_args()
 667   if len(args): optparser.error('no non-option arguments please')
 668
 669   if need_defcfg:
 670     read_defconfig()
 671
 672   try:
 673     (pss, pcs) = _cfg_process_putatives()
 674     process_cfg(opts, pss, pcs)
 675   except (configparser.Error, ValueError):
 676     traceback.print_exc(file=sys.stderr)
 677     print('\nInvalid configuration, giving up.', file=sys.stderr)
 678     sys.exit(12)
 679
 680
 681   #print('X', debug_set, file=sys.stderr)
 682
 683   log_formatter = twisted.logger.formatEventAsClassicLogText
 684   stdout_obs = twisted.logger.FileLogObserver(sys.stdout, log_formatter)
 685   stderr_obs = twisted.logger.FileLogObserver(sys.stderr, log_formatter)
 686   pred = twisted.logger.LogLevelFilterPredicate(LogLevel.error)
 687   stdsomething_obs = twisted.logger.FilteringLogObserver(
 688     stderr_obs, [pred], stdout_obs
 689   )
 690   global file_log_observer
 691   file_log_observer = twisted.logger.FilteringLogObserver(
 692     stdsomething_obs, [LogNotBoringTwisted()]
 693   )
 694   #log_observer = stdsomething_obs
 695   twisted.logger.globalLogBeginner.beginLoggingTo(
 696     [ file_log_observer, crash_on_critical ]
 697     )
 698
 699 def common_run():
 700   log_debug(DBG.INIT, 'entering reactor')
 701   if not _crashing: reactor.run()
 702   print('ENDED', file=sys.stderr)
 703   sys.exit(16)