mdw@git.distorted.org.uk Git - tripe/blob - peerdb/tripe-newpeers.in

   1 #! @PYTHON@
   2 ### -*-python-*-
   3 ###
   4 ### Build a CDB file from configuration file
   5 ###
   6 ### (c) 2007 Straylight/Edgeware
   7 ###
   8
   9 ###----- Licensing notice ---------------------------------------------------
  10 ###
  11 ### This file is part of Trivial IP Encryption (TrIPE).
  12 ###
  13 ### TrIPE is free software: you can redistribute it and/or modify it under
  14 ### the terms of the GNU General Public License as published by the Free
  15 ### Software Foundation; either version 3 of the License, or (at your
  16 ### option) any later version.
  17 ###
  18 ### TrIPE is distributed in the hope that it will be useful, but WITHOUT
  19 ### ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  20 ### FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  21 ### for more details.
  22 ###
  23 ### You should have received a copy of the GNU General Public License
  24 ### along with TrIPE.  If not, see <https://www.gnu.org/licenses/>.
  25
  26 VERSION = '@VERSION@'
  27
  28 ###--------------------------------------------------------------------------
  29 ### External dependencies.
  30
  31 import mLib as M
  32 from optparse import OptionParser
  33 import cdb as CDB
  34 from sys import stdin, stdout, exit, argv
  35 import re as RX
  36 import os as OS
  37 from cStringIO import StringIO
  38
  39 ###--------------------------------------------------------------------------
  40 ### Utilities.
  41
  42 class CDBFake (object):
  43   """Like cdbmake, but just outputs data suitable for cdb-map."""
  44   def __init__(me, file = stdout):
  45     me.file = file
  46   def add(me, key, value):
  47     me.file.write('%s:%s\n' % (key, value))
  48   def finish(me):
  49     pass
  50
  51 class ExpectedError (Exception): pass
  52
  53 ###--------------------------------------------------------------------------
  54 ### A bulk DNS resolver.
  55
  56 class ResolverFailure (ExpectedError):
  57   def __init__(me, host, msg):
  58     me.host = host
  59     me.msg = msg
  60   def __str__(me):
  61     return "failed to resolve `%s': %s" % (me.host, me.msg)
  62
  63 class BulkResolver (object):
  64   """
  65   Resolve a number of DNS names in parallel.
  66
  67   The BulkResovler resolves a number of hostnames in parallel.  Using it
  68   works in three phases:
  69
  70     1. You call prepare(HOSTNAME) a number of times, to feed in the hostnames
  71        you're interested in.
  72
  73     2. You call run() to actually drive the resolver.
  74
  75     3. You call lookup(HOSTNAME) to get the address you wanted.  This will
  76        fail with KeyError if the resolver couldn't resolve the HOSTNAME.
  77   """
  78
  79   def __init__(me):
  80     """Initialize the resolver."""
  81     me._resolvers = {}
  82     me._namemap = {}
  83
  84   def prepare(me, host):
  85     """Prime the resolver to resolve the name HOST."""
  86     if host not in me._resolvers:
  87       me._resolvers[host] = M.SelResolveByName \
  88                             (host,
  89                              lambda name, alias, addr:
  90                                me._resolved(host, addr[0]),
  91                              lambda: me._resolved(host, None))
  92
  93   def run(me):
  94     """Run the background DNS resolver until it's finished."""
  95     while me._resolvers:
  96       M.select()
  97
  98   def lookup(me, host):
  99     """
 100     Fetch the address corresponding to HOST.
 101     """
 102     addr = me._namemap[host]
 103     if addr is None:
 104       raise ResolverFailure(host, '(unknown failure)')
 105     return addr
 106
 107   def _resolved(me, host, addr):
 108     """Callback function: remember that ADDR is the address for HOST."""
 109     me._namemap[host] = addr
 110     del me._resolvers[host]
 111
 112 ###--------------------------------------------------------------------------
 113 ### The configuration parser.
 114
 115 ## Match a comment or empty line.
 116 RX_COMMENT = RX.compile(r'(?x) ^ \s* (?: $ | [;#])')
 117
 118 ## Match a section group header.
 119 RX_GRPHDR = RX.compile(r'(?x) ^ \s* \[ (.*) \] \s* $')
 120
 121 ## Match an assignment line.
 122 RX_ASSGN = RX.compile(r'''(?x) ^
 123         ([^\s:=] (?: [^:=]* [^\s:=])?)
 124         \s* [:=] \s*
 125         (| \S | \S.*\S)
 126         \s* $''')
 127
 128 ## Match a continuation line.
 129 RX_CONT = RX.compile(r'''(?x) ^ \s+
 130         (| \S | \S.*\S)
 131         \s* $''')
 132
 133 ## Match a $(VAR) configuration variable reference; group 1 is the VAR.
 134 RX_REF = RX.compile(r'(?x) \$ \( ([^)]+) \)')
 135
 136 ## Match a $[HOST] name resolution reference; group 1 is the HOST.
 137 RX_RESOLVE = RX.compile(r'(?x) \$ \[ ([^]]+) \]')
 138
 139 class ConfigSyntaxError (ExpectedError):
 140   def __init__(me, fname, lno, msg):
 141     me.fname = fname
 142     me.lno = lno
 143     me.msg = msg
 144   def __str__(me):
 145     return '%s:%d: %s' % (me.fname, me.lno, me.msg)
 146
 147 def _fmt_path(path):
 148   return ' -> '.join(["`%s'" % hop for hop in path])
 149
 150 class AmbiguousOptionError (ExpectedError):
 151   def __init__(me, key, patha, vala, pathb, valb):
 152     me.key = key
 153     me.patha, me.vala = patha, vala
 154     me.pathb, me.valb = pathb, valb
 155   def __str__(me):
 156     return "Ambiguous answer resolving key `%s': " \
 157         "path %s yields `%s' but %s yields `%s'" % \
 158         (me.key, _fmt_path(me.patha), me.vala, _fmt_path(me.pathb), me.valb)
 159
 160 class InheritanceCycleError (ExpectedError):
 161   def __init__(me, key, path):
 162     me.key = key
 163     me.path = path
 164   def __str__(me):
 165     return "Found a cycle %s looking up key `%s'" % \
 166         (_fmt_path(me.path), me.key)
 167
 168 class MissingSectionException (ExpectedError):
 169   def __init__(me, sec):
 170     me.sec = sec
 171   def __str__(me):
 172     return "Section `%s' not found" % (me.sec)
 173
 174 class MissingKeyException (ExpectedError):
 175   def __init__(me, sec, key):
 176     me.sec = sec
 177     me.key = key
 178   def __str__(me):
 179     return "Key `%s' not found in section `%s'" % (me.key, me.sec)
 180
 181 class ConfigSection (object):
 182   """
 183   A section in a configuration parser.
 184
 185   This is where a lot of the nitty-gritty stuff actually happens.  The
 186   `MyConfigParser' knows a lot about the internals of this class, which saves
 187   on building a complicated interface.
 188   """
 189
 190   def __init__(me, name, cp):
 191     """Initialize a new, empty section with a given NAME and parent CP."""
 192
 193     ## The cache maps item keys to entries, which consist of a pair of
 194     ## objects.  There are four possible states for a cache entry:
 195     ##
 196     ##   * missing -- there is no entry at all with this key, so we must
 197     ##     search for it;
 198     ##
 199     ##   * None, None -- we are actively trying to resolve this key, so if we
 200     ##     encounter this state, we have found a cycle in the inheritance
 201     ##     graph;
 202     ##
 203     ##   * None, [] -- we know that this key isn't reachable through any of
 204     ##     our parents;
 205     ##
 206     ##   * VALUE, PATH -- we know that the key resolves to VALUE, along the
 207     ##     PATH from us (exclusive) to the defining parent (inclusive).
 208     me.name = name
 209     me._itemmap = dict()
 210     me._cache = dict()
 211     me._cp = cp
 212
 213   def _expand(me, string, resolvep):
 214     """
 215     Expands $(...) and (optionally) $[...] placeholders in STRING.
 216
 217     RESOLVEP is a boolean switch: do we bother to tax the resolver or not?
 218     This is turned off by MyConfigParser's resolve() method while it's
 219     collecting hostnames to be resolved.
 220     """
 221     string = RX_REF.sub \
 222              (lambda m: me.get(m.group(1), resolvep), string)
 223     if resolvep:
 224       string = RX_RESOLVE.sub(lambda m: me._cp._resolver.lookup(m.group(1)),
 225                               string)
 226     return string
 227
 228   def _parents(me):
 229     """Yield this section's parents."""
 230     try: names = me._itemmap['@inherit']
 231     except KeyError: return
 232     for name in names.replace(',', ' ').split():
 233       yield me._cp.section(name)
 234
 235   def _get(me, key, path = None):
 236     """
 237     Low-level option-fetching method.
 238
 239     Fetch the value for the named KEY in this section, or maybe (recursively)
 240     a section which it inherits from.
 241
 242     Returns a pair VALUE, PATH.  The value is not expanded; nor do we check
 243     for the special `name' key.  The caller is expected to do these things.
 244     Returns None if no value could be found.
 245     """
 246
 247     ## If we weren't given a path, then we'd better make one.
 248     if path is None: path = []
 249
 250     ## Extend the path to cover us, but remember to remove us again when
 251     ## we've finished.  If we need to pass the current path back upwards,
 252     ## then remember to take a copy.
 253     path.append(me.name)
 254     try:
 255
 256       ## If we've been this way before on another pass through then return the
 257       ## value we found then.  If we're still thinking about it then we've
 258       ## found a cycle.
 259       try: v, p = me._cache[key]
 260       except KeyError: pass
 261       else:
 262         if p is None: raise InheritanceCycleError(key, path[:])
 263         else: return v, path + p
 264
 265       ## See whether the answer is ready waiting for us.
 266       try: v = me._itemmap[key]
 267       except KeyError: pass
 268       else:
 269         p = path[:]
 270         me._cache[key] = v, []
 271         return v, p
 272
 273       ## Initially we have no idea.
 274       value = None
 275       winner = []
 276
 277       ## Go through our parents and ask them what they think.
 278       me._cache[key] = None, None
 279       for p in me._parents():
 280
 281         ## See whether we get an answer.  If not, keep on going.
 282         v, pp = p._get(key, path)
 283         if v is None: continue
 284
 285         ## If we got an answer, check that it matches any previous ones.
 286         if value is None:
 287           value = v
 288           winner = pp
 289         elif value != v:
 290           raise AmbiguousOptionError(key, winner, value, pp, v)
 291
 292       ## That's the best we could manage.
 293       me._cache[key] = value, winner[len(path):]
 294       return value, winner
 295
 296     finally:
 297       ## Remove us from the path again.
 298       path.pop()
 299
 300   def get(me, key, resolvep = True):
 301     """
 302     Retrieve the value of KEY from this section.
 303     """
 304
 305     ## Special handling for the `name' key.
 306     if key == 'name':
 307       value = me._itemmap.get('name', me.name)
 308     elif key == '@inherits':
 309       try: return me._itemmap['@inherits']
 310       except KeyError: raise MissingKeyException(me.name, key)
 311     else:
 312       value, _ = me._get(key)
 313       if value is None:
 314         raise MissingKeyException(me.name, key)
 315
 316     ## Expand the value and return it.
 317     return me._expand(value, resolvep)
 318
 319   def items(me, resolvep = True):
 320     """
 321     Yield a list of item names in the section.
 322     """
 323
 324     ## Initialize for a depth-first walk of the inheritance graph.
 325     seen = { 'name': True }
 326     visiting = { me.name: True }
 327     stack = [me]
 328
 329     ## Visit nodes, collecting their keys.  Don't believe the values:
 330     ## resolving inheritance is too hard to do like this.
 331     while stack:
 332       sec = stack.pop()
 333       for p in sec._parents():
 334         if p.name not in visiting:
 335           stack.append(p); visiting[p.name] = True
 336
 337       for key in sec._itemmap.iterkeys(): seen[key] = None
 338
 339     ## And we're done.
 340     return seen.iterkeys()
 341
 342 class MyConfigParser (object):
 343   """
 344   A more advanced configuration parser.
 345
 346   This has four major enhancements over the standard ConfigParser which are
 347   relevant to us.
 348
 349     * It recognizes `@inherits' keys and follows them when expanding a
 350       value.
 351
 352     * It recognizes `$(VAR)' references to configuration variables during
 353       expansion and processes them correctly.
 354
 355     * It recognizes `$[HOST]' name-resolver requests and handles them
 356       correctly.
 357
 358     * Its parsing behaviour is well-defined.
 359
 360   Use:
 361
 362     1. Call parse(FILENAME) to slurp in the configuration data.
 363
 364     2. Call resolve() to collect the hostnames which need to be resolved and
 365        actually do the name resolution.
 366
 367     3. Call sections() to get a list of the configuration sections, or
 368        section(NAME) to find a named section.
 369
 370     4. Call get(ITEM) on a section to collect the results, or items() to
 371        iterate over them.
 372   """
 373
 374   def __init__(me):
 375     """
 376     Initialize a new, empty configuration parser.
 377     """
 378     me._sectmap = dict()
 379     me._resolver = BulkResolver()
 380
 381   def parse(me, f):
 382     """
 383     Parse configuration from a file F.
 384     """
 385
 386     ## Initial parser state.
 387     sect = None
 388     key = None
 389     val = None
 390     lno = 0
 391
 392     ## An unpleasant hack.  Python makes it hard to capture a value in a
 393     ## variable and examine it in a single action, and this is the best that
 394     ## I came up with.
 395     m = [None]
 396     def match(rx): m[0] = rx.match(line); return m[0]
 397
 398     ## Commit a key's value when we've determined that there are no further
 399     ## continuation lines.
 400     def flush():
 401       if key is not None: sect._itemmap[key] = val.getvalue()
 402
 403     ## Work through all of the input lines.
 404     for line in f:
 405       lno += 1
 406
 407       if match(RX_COMMENT):
 408         ## A comment or a blank line.  Nothing doing.  (This means that we
 409         ## leave out blank lines which look like they might be continuation
 410         ## lines.)
 411
 412         pass
 413
 414       elif match(RX_GRPHDR):
 415         ## A section header.  Flush out any previous value and set up the new
 416         ## group.
 417
 418         flush()
 419         name = m[0].group(1)
 420         try: sect = me._sectmap[name]
 421         except KeyError: sect = me._sectmap[name] = ConfigSection(name, me)
 422         key = None
 423
 424       elif match(RX_ASSGN):
 425         ## A new assignment.  Flush out the old one, and set up to store this
 426         ## one.
 427
 428         if sect is None:
 429           raise ConfigSyntaxError(f.name, lno, 'no active section to update')
 430         flush()
 431         key = m[0].group(1)
 432         val = StringIO(); val.write(m[0].group(2))
 433
 434       elif match(RX_CONT):
 435         ## A continuation line.  Accumulate the value.
 436
 437         if key is None:
 438           raise ConfigSyntaxError(f.name, lno, 'no config value to continue')
 439         val.write('\n'); val.write(m[0].group(1))
 440
 441       else:
 442         ## Something else.
 443
 444         raise ConfigSyntaxError(f.name, lno, 'incomprehensible line')
 445
 446     ## Don't forget to commit any final value material.
 447     flush()
 448
 449   def section(me, name):
 450     """Return a ConfigSection with the given NAME."""
 451     try: return me._sectmap[name]
 452     except KeyError: raise MissingSectionException(name)
 453
 454   def sections(me):
 455     """Yield the known sections."""
 456     return me._sectmap.itervalues()
 457
 458   def resolve(me):
 459     """
 460     Works out all of the hostnames which need resolving and resolves them.
 461
 462     Until you call this, attempts to fetch configuration items which need to
 463     resolve hostnames will fail!
 464     """
 465     for sec in me.sections():
 466       for key in sec.items():
 467         value = sec.get(key, resolvep = False)
 468         for match in RX_RESOLVE.finditer(value):
 469           me._resolver.prepare(match.group(1))
 470     me._resolver.run()
 471
 472 ###--------------------------------------------------------------------------
 473 ### Command-line handling.
 474
 475 def inputiter(things):
 476   """
 477   Iterate over command-line arguments, returning corresponding open files.
 478
 479   If none were given, or one is `-', assume standard input; if one is a
 480   directory, scan it for files other than backups; otherwise return the
 481   opened files.
 482   """
 483
 484   if not things:
 485     if OS.isatty(stdin.fileno()):
 486       M.die('no input given, and stdin is a terminal')
 487     yield stdin
 488   else:
 489     for thing in things:
 490       if thing == '-':
 491         yield stdin
 492       elif OS.path.isdir(thing):
 493         for item in OS.listdir(thing):
 494           if item.endswith('~') or item.endswith('#'):
 495             continue
 496           name = OS.path.join(thing, item)
 497           if not OS.path.isfile(name):
 498             continue
 499           yield file(name)
 500       else:
 501         yield file(thing)
 502
 503 def parse_options(argv = argv):
 504   """
 505   Parse command-line options, returning a pair (OPTS, ARGS).
 506   """
 507   M.ego(argv[0])
 508   op = OptionParser(usage = '%prog [-c CDB] INPUT...',
 509                     version = '%%prog (tripe, version %s)' % VERSION)
 510   op.add_option('-c', '--cdb', metavar = 'CDB',
 511                 dest = 'cdbfile', default = None,
 512                 help = 'Compile output into a CDB file.')
 513   opts, args = op.parse_args(argv)
 514   return opts, args
 515
 516 ###--------------------------------------------------------------------------
 517 ### Main code.
 518
 519 def getconf(args):
 520   """
 521   Read the configuration files and return the accumulated result.
 522
 523   We make sure that all hostnames have been properly resolved.
 524   """
 525   conf = MyConfigParser()
 526   for f in inputiter(args):
 527     conf.parse(f)
 528   conf.resolve()
 529   return conf
 530
 531 def output(conf, cdb):
 532   """
 533   Output the configuration information CONF to the database CDB.
 534
 535   This is where the special `user' and `auto' database entries get set.
 536   """
 537   auto = []
 538   for sec in sorted(conf.sections(), key = lambda sec: sec.name):
 539     if sec.name.startswith('@'):
 540       continue
 541     elif sec.name.startswith('$'):
 542       label = sec.name
 543     else:
 544       label = 'P%s' % sec.name
 545       try: a = sec.get('auto')
 546       except MissingKeyException: pass
 547       else:
 548         if a in ('y', 'yes', 't', 'true', '1', 'on'): auto.append(sec.name)
 549       try: u = sec.get('user')
 550       except MissingKeyException: pass
 551       else: cdb.add('U%s' % u)
 552     url = M.URLEncode(semip = True)
 553     for key in sorted(sec.items()):
 554       if not key.startswith('@'):
 555         url.encode(key, sec.get(key))
 556     cdb.add(label, url.result)
 557   cdb.add('%AUTO', ' '.join(auto))
 558   cdb.finish()
 559
 560 def main():
 561   """Main program."""
 562   opts, args = parse_options()
 563   if opts.cdbfile:
 564     cdb = CDB.cdbmake(opts.cdbfile, opts.cdbfile + '.new')
 565   else:
 566     cdb = CDBFake()
 567   try:
 568     conf = getconf(args[1:])
 569     output(conf, cdb)
 570   except ExpectedError, e:
 571     M.moan(str(e))
 572     exit(2)
 573
 574 if __name__ == '__main__':
 575   main()
 576
 577 ###----- That's all, folks --------------------------------------------------