src/python/location-importer.in

   1 #!/usr/bin/python3
   2 ###############################################################################
   3 #                                                                             #
   4 # libloc - A library to determine the location of someone on the Internet     #
   5 #                                                                             #
   6 # Copyright (C) 2020 IPFire Development Team <info@ipfire.org>                #
   7 #                                                                             #
   8 # This library is free software; you can redistribute it and/or               #
   9 # modify it under the terms of the GNU Lesser General Public                  #
  10 # License as published by the Free Software Foundation; either                #
  11 # version 2.1 of the License, or (at your option) any later version.          #
  12 #                                                                             #
  13 # This library is distributed in the hope that it will be useful,             #
  14 # but WITHOUT ANY WARRANTY; without even the implied warranty of              #
  15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU           #
  16 # Lesser General Public License for more details.                             #
  17 #                                                                             #
  18 ###############################################################################
  19
  20 import argparse
  21 import ipaddress
  22 import logging
  23 import math
  24 import re
  25 import sys
  26 import telnetlib
  27
  28 # Load our location module
  29 import location
  30 import location.database
  31 import location.importer
  32 from location.i18n import _
  33
  34 # Initialise logging
  35 log = logging.getLogger("location.importer")
  36 log.propagate = 1
  37
  38 class CLI(object):
  39         def parse_cli(self):
  40                 parser = argparse.ArgumentParser(
  41                         description=_("Location Importer Command Line Interface"),
  42                 )
  43                 subparsers = parser.add_subparsers()
  44
  45                 # Global configuration flags
  46                 parser.add_argument("--debug", action="store_true",
  47                         help=_("Enable debug output"))
  48
  49                 # version
  50                 parser.add_argument("--version", action="version",
  51                         version="%(prog)s @VERSION@")
  52
  53                 # Database
  54                 parser.add_argument("--database-host", required=True,
  55                         help=_("Database Hostname"), metavar=_("HOST"))
  56                 parser.add_argument("--database-name", required=True,
  57                         help=_("Database Name"), metavar=_("NAME"))
  58                 parser.add_argument("--database-username", required=True,
  59                         help=_("Database Username"), metavar=_("USERNAME"))
  60                 parser.add_argument("--database-password", required=True,
  61                         help=_("Database Password"), metavar=_("PASSWORD"))
  62
  63                 # Write Database
  64                 write = subparsers.add_parser("write", help=_("Write database to file"))
  65                 write.set_defaults(func=self.handle_write)
  66                 write.add_argument("file", nargs=1, help=_("Database File"))
  67                 write.add_argument("--signing-key", nargs="?", type=open, help=_("Signing Key"))
  68                 write.add_argument("--vendor", nargs="?", help=_("Sets the vendor"))
  69                 write.add_argument("--description", nargs="?", help=_("Sets a description"))
  70                 write.add_argument("--license", nargs="?", help=_("Sets the license"))
  71
  72                 # Update WHOIS
  73                 update_whois = subparsers.add_parser("update-whois", help=_("Update WHOIS Information"))
  74                 update_whois.set_defaults(func=self.handle_update_whois)
  75
  76                 # Update announcements
  77                 update_announcements = subparsers.add_parser("update-announcements",
  78                         help=_("Update BGP Annoucements"))
  79                 update_announcements.set_defaults(func=self.handle_update_announcements)
  80                 update_announcements.add_argument("server", nargs=1,
  81                         help=_("Route Server to connect to"), metavar=_("SERVER"))
  82
  83                 # Update overrides
  84                 update_overrides = subparsers.add_parser("update-overrides",
  85                         help=_("Update overrides"),
  86                 )
  87                 update_overrides.add_argument(
  88                         "files", nargs="+", help=_("Files to import"),
  89                 )
  90                 update_overrides.set_defaults(func=self.handle_update_overrides)
  91
  92                 args = parser.parse_args()
  93
  94                 # Enable debug logging
  95                 if args.debug:
  96                         log.setLevel(logging.DEBUG)
  97
  98                 # Print usage if no action was given
  99                 if not "func" in args:
 100                         parser.print_usage()
 101                         sys.exit(2)
 102
 103                 return args
 104
 105         def run(self):
 106                 # Parse command line arguments
 107                 args = self.parse_cli()
 108
 109                 # Initialise database
 110                 self.db = self._setup_database(args)
 111
 112                 # Call function
 113                 ret = args.func(args)
 114
 115                 # Return with exit code
 116                 if ret:
 117                         sys.exit(ret)
 118
 119                 # Otherwise just exit
 120                 sys.exit(0)
 121
 122         def _setup_database(self, ns):
 123                 """
 124                         Initialise the database
 125                 """
 126                 # Connect to database
 127                 db = location.database.Connection(
 128                         host=ns.database_host, database=ns.database_name,
 129                         user=ns.database_username, password=ns.database_password,
 130                 )
 131
 132                 with db.transaction():
 133                         db.execute("""
 134                                 -- announcements
 135                                 CREATE TABLE IF NOT EXISTS announcements(network inet, autnum bigint,
 136                                         first_seen_at timestamp without time zone DEFAULT CURRENT_TIMESTAMP,
 137                                         last_seen_at timestamp without time zone DEFAULT CURRENT_TIMESTAMP);
 138                                 CREATE UNIQUE INDEX IF NOT EXISTS announcements_networks ON announcements(network);
 139                                 CREATE INDEX IF NOT EXISTS announcements_family ON announcements(family(network));
 140
 141                                 -- autnums
 142                                 CREATE TABLE IF NOT EXISTS autnums(number bigint, name text NOT NULL);
 143                                 CREATE UNIQUE INDEX IF NOT EXISTS autnums_number ON autnums(number);
 144
 145                                 -- networks
 146                                 CREATE TABLE IF NOT EXISTS networks(network inet, country text);
 147                                 CREATE UNIQUE INDEX IF NOT EXISTS networks_network ON networks(network);
 148                                 CREATE INDEX IF NOT EXISTS networks_search ON networks USING GIST(network inet_ops);
 149
 150                                 -- overrides
 151                                 CREATE TABLE IF NOT EXISTS autnum_overrides(
 152                                         number bigint NOT NULL,
 153                                         name text,
 154                                         is_anonymous_proxy boolean DEFAULT FALSE,
 155                                         is_satellite_provider boolean DEFAULT FALSE,
 156                                         is_anycast boolean DEFAULT FALSE
 157                                 );
 158                                 CREATE UNIQUE INDEX IF NOT EXISTS autnum_overrides_number
 159                                         ON autnum_overrides(number);
 160
 161                                 CREATE TABLE IF NOT EXISTS network_overrides(
 162                                         network inet NOT NULL,
 163                                         country text,
 164                                         is_anonymous_proxy boolean DEFAULT FALSE,
 165                                         is_satellite_provider boolean DEFAULT FALSE,
 166                                         is_anycast boolean DEFAULT FALSE
 167                                 );
 168                                 CREATE UNIQUE INDEX IF NOT EXISTS network_overrides_network
 169                                         ON network_overrides(network);
 170                         """)
 171
 172                 return db
 173
 174         def handle_write(self, ns):
 175                 """
 176                         Compiles a database in libloc format out of what is in the database
 177                 """
 178                 print(ns)
 179
 180                 # Allocate a writer
 181                 writer = location.Writer(ns.signing_key)
 182
 183                 # Set all metadata
 184                 if ns.vendor:
 185                         writer.vendor = ns.vendor
 186
 187                 if ns.description:
 188                         writer.description = ns.description
 189
 190                 if ns.license:
 191                         writer.license = ns.license
 192
 193                 # Add all Autonomous Systems
 194                 log.info("Writing Autonomous Systems...")
 195
 196                 # Select all ASes with a name
 197                 rows = self.db.query("""
 198                         SELECT
 199                                 autnums.number AS number,
 200                                 COALESCE(
 201                                         (SELECT overrides.name FROM autnum_overrides overrides
 202                                                 WHERE overrides.number = autnums.number),
 203                                         autnums.name
 204                                 ) AS name
 205                                 FROM autnums
 206                                 WHERE name <> %s ORDER BY number
 207                         """, "")
 208
 209                 for row in rows:
 210                         a = writer.add_as(row.number)
 211                         a.name = row.name
 212
 213                 # Add all networks
 214                 log.info("Writing networks...")
 215
 216                 # Select all known networks
 217                 rows = self.db.query("""
 218                         SELECT
 219                                 announcements.network AS network,
 220                                 announcements.autnum AS autnum,
 221                                 (
 222                                         SELECT networks.country FROM networks
 223                                                 WHERE announcements.network <<= networks.network
 224                                 ) AS country,
 225
 226                                 -- Flags
 227                                 FALSE AS is_anonymous_proxy,
 228                                 FALSE AS is_satellite_provider,
 229                                 FALSE AS is_anycast
 230                         FROM announcements
 231                 """)
 232
 233                 for row in rows:
 234                         network = writer.add_network(row.network)
 235
 236                         # Save AS & country
 237                         network.asn, network.country_code = row.autnum, row.country
 238
 239                         # Set flags
 240                         if row.is_anonymous_proxy:
 241                                 network.set_flag(location.NETWORK_FLAG_ANONYMOUS_PROXY)
 242
 243                         if row.is_satellite_provider:
 244                                 network.set_flag(location.NETWORK_FLAG_SATELLITE_PROVIDER)
 245
 246                         if row.is_anycast:
 247                                 network.set_flag(location.NETWORK_FLAG_ANYCAST)
 248
 249                 # Write everything to file
 250                 log.info("Writing database to file...")
 251                 for file in ns.file:
 252                         writer.write(file)
 253
 254         def handle_update_whois(self, ns):
 255                 downloader = location.importer.Downloader()
 256
 257                 # Download all sources
 258                 with self.db.transaction():
 259                         # Create some temporary tables to store parsed data
 260                         self.db.execute("""
 261                                 CREATE TEMPORARY TABLE _autnums(number integer, organization text)
 262                                         ON COMMIT DROP;
 263                                 CREATE UNIQUE INDEX _autnums_number ON _autnums(number);
 264
 265                                 CREATE TEMPORARY TABLE _organizations(handle text, name text)
 266                                         ON COMMIT DROP;
 267                                 CREATE UNIQUE INDEX _organizations_handle ON _organizations(handle);
 268                         """)
 269
 270                         for source in location.importer.WHOIS_SOURCES:
 271                                 with downloader.request(source, return_blocks=True) as f:
 272                                         for block in f:
 273                                                 self._parse_block(block)
 274
 275                         self.db.execute("""
 276                                 INSERT INTO autnums(number, name)
 277                                         SELECT _autnums.number, _organizations.name FROM _autnums
 278                                                 LEFT JOIN _organizations ON _autnums.organization = _organizations.handle
 279                                 ON CONFLICT (number) DO UPDATE SET name = excluded.name;
 280                         """)
 281
 282                 # Download all extended sources
 283                 for source in location.importer.EXTENDED_SOURCES:
 284                         with self.db.transaction():
 285                                 # Download data
 286                                 with downloader.request(source) as f:
 287                                         for line in f:
 288                                                 self._parse_line(line)
 289
 290         def _parse_block(self, block):
 291                 # Get first line to find out what type of block this is
 292                 line = block[0]
 293
 294                 # aut-num
 295                 if line.startswith("aut-num:"):
 296                         return self._parse_autnum_block(block)
 297
 298                 # organisation
 299                 elif line.startswith("organisation:"):
 300                         return self._parse_org_block(block)
 301
 302         def _parse_autnum_block(self, block):
 303                 autnum = {}
 304                 for line in block:
 305                         # Split line
 306                         key, val = split_line(line)
 307
 308                         if key == "aut-num":
 309                                 m = re.match(r"^(AS|as)(\d+)", val)
 310                                 if m:
 311                                         autnum["asn"] = m.group(2)
 312
 313                         elif key == "org":
 314                                 autnum[key] = val
 315
 316                 # Skip empty objects
 317                 if not autnum:
 318                         return
 319
 320                 # Insert into database
 321                 self.db.execute("INSERT INTO _autnums(number, organization) \
 322                         VALUES(%s, %s) ON CONFLICT (number) DO UPDATE SET \
 323                                 organization = excluded.organization",
 324                         autnum.get("asn"), autnum.get("org"),
 325                 )
 326
 327         def _parse_org_block(self, block):
 328                 org = {}
 329                 for line in block:
 330                         # Split line
 331                         key, val = split_line(line)
 332
 333                         if key in ("organisation", "org-name"):
 334                                 org[key] = val
 335
 336                 # Skip empty objects
 337                 if not org:
 338                         return
 339
 340                 self.db.execute("INSERT INTO _organizations(handle, name) \
 341                         VALUES(%s, %s) ON CONFLICT (handle) DO \
 342                         UPDATE SET name = excluded.name",
 343                         org.get("organisation"), org.get("org-name"),
 344                 )
 345
 346         def _parse_line(self, line):
 347                 # Skip version line
 348                 if line.startswith("2"):
 349                         return
 350
 351                 # Skip comments
 352                 if line.startswith("#"):
 353                         return
 354
 355                 try:
 356                         registry, country_code, type, line = line.split("|", 3)
 357                 except:
 358                         log.warning("Could not parse line: %s" % line)
 359                         return
 360
 361                 # Skip any lines that are for stats only
 362                 if country_code == "*":
 363                         return
 364
 365                 if type in ("ipv6", "ipv4"):
 366                         return self._parse_ip_line(country_code, type, line)
 367
 368         def _parse_ip_line(self, country, type, line):
 369                 try:
 370                         address, prefix, date, status, organization = line.split("|")
 371                 except ValueError:
 372                         organization = None
 373
 374                         # Try parsing the line without organization
 375                         try:
 376                                 address, prefix, date, status = line.split("|")
 377                         except ValueError:
 378                                 log.warning("Unhandled line format: %s" % line)
 379                                 return
 380
 381                 # Skip anything that isn't properly assigned
 382                 if not status in ("assigned", "allocated"):
 383                         return
 384
 385                 # Cast prefix into an integer
 386                 try:
 387                         prefix = int(prefix)
 388                 except:
 389                         log.warning("Invalid prefix: %s" % prefix)
 390
 391                 # Fix prefix length for IPv4
 392                 if type == "ipv4":
 393                         prefix = 32 - int(math.log(prefix, 2))
 394
 395                 # Try to parse the address
 396                 try:
 397                         network = ipaddress.ip_network("%s/%s" % (address, prefix), strict=False)
 398                 except ValueError:
 399                         log.warning("Invalid IP address: %s" % address)
 400                         return
 401
 402                 self.db.execute("INSERT INTO networks(network, country) \
 403                         VALUES(%s, %s) ON CONFLICT (network) DO \
 404                         UPDATE SET country = excluded.country",
 405                         "%s" % network, country,
 406                 )
 407
 408         def handle_update_announcements(self, ns):
 409                 server = ns.server[0]
 410
 411                 # Pre-compile regular expression for routes
 412                 #route = re.compile(b"^\*>?\s[\si]?([^\s]+)[.\s]*?(\d+)\si$", re.MULTILINE)
 413                 route = re.compile(b"^\*[\s\>]i([^\s]+).+?(\d+)\si\r\n", re.MULTILINE|re.DOTALL)
 414
 415                 with telnetlib.Telnet(server) as t:
 416                         # Enable debug mode
 417                         #if ns.debug:
 418                         #       t.set_debuglevel(10)
 419
 420                         # Wait for console greeting
 421                         greeting = t.read_until(b"> ")
 422                         log.debug(greeting.decode())
 423
 424                         # Disable pagination
 425                         t.write(b"terminal length 0\n")
 426
 427                         # Wait for the prompt to return
 428                         t.read_until(b"> ")
 429
 430                         # Fetch the routing tables
 431                         with self.db.transaction():
 432                                 for protocol in ("ipv6", "ipv4"):
 433                                         log.info("Requesting %s routing table" % protocol)
 434
 435                                         # Request the full unicast routing table
 436                                         t.write(b"show bgp %s unicast\n" % protocol.encode())
 437
 438                                         # Read entire header which ends with "Path"
 439                                         t.read_until(b"Path\r\n")
 440
 441                                         while True:
 442                                                 # Try reading a full entry
 443                                                 # Those might be broken across multiple lines but ends with i
 444                                                 line = t.read_until(b"i\r\n", timeout=5)
 445                                                 if not line:
 446                                                         break
 447
 448                                                 # Show line for debugging
 449                                                 #log.debug(repr(line))
 450
 451                                                 # Try finding a route in here
 452                                                 m = route.match(line)
 453                                                 if m:
 454                                                         network, autnum = m.groups()
 455
 456                                                         # Convert network to string
 457                                                         network = network.decode()
 458
 459                                                         # Convert AS number to integer
 460                                                         autnum = int(autnum)
 461
 462                                                         log.info("Found announcement for %s by %s" % (network, autnum))
 463
 464                                                         self.db.execute("INSERT INTO announcements(network, autnum) \
 465                                                                 VALUES(%s, %s) ON CONFLICT (network) DO \
 466                                                                 UPDATE SET autnum = excluded.autnum, last_seen_at = CURRENT_TIMESTAMP",
 467                                                                 network, autnum,
 468                                                         )
 469
 470                                         log.info("Finished reading the %s routing table" % protocol)
 471
 472                                 # Purge anything we never want here
 473                                 self.db.execute("""
 474                                         -- Delete default routes
 475                                         DELETE FROM announcements WHERE network = '::/0' OR network = '0.0.0.0/0';
 476
 477                                         -- Delete anything that is not global unicast address space
 478                                         DELETE FROM announcements WHERE family(network) = 6 AND NOT network <<= '2000::/3';
 479
 480                                         -- DELETE RFC1918 address space
 481                                         DELETE FROM announcements WHERE family(network) = 4 AND network <<= '10.0.0.0/8';
 482                                         DELETE FROM announcements WHERE family(network) = 4 AND network <<= '172.16.0.0/12';
 483                                         DELETE FROM announcements WHERE family(network) = 4 AND network <<= '192.168.0.0/16';
 484
 485                                         -- Delete networks that are too small to be in the global routing table
 486                                         DELETE FROM announcements WHERE family(network) = 6 AND masklen(network) > 48;
 487                                         DELETE FROM announcements WHERE family(network) = 4 AND masklen(network) > 24;
 488
 489                                         -- Delete any non-public or reserved ASNs
 490                                         DELETE FROM announcements WHERE NOT (
 491                                                 (autnum >= 1 AND autnum <= 23455)
 492                                                 OR
 493                                                 (autnum >= 23457 AND autnum <= 64495)
 494                                                 OR
 495                                                 (autnum >= 131072 AND autnum <= 4199999999)
 496                                         );
 497
 498                                         -- Delete everything that we have not seen for 14 days
 499                                         DELETE FROM announcements WHERE last_seen_at <= CURRENT_TIMESTAMP - INTERVAL '14 days';
 500                                 """)
 501
 502         def handle_update_overrides(self, ns):
 503                 with self.db.transaction():
 504                         # Drop all data that we have
 505                         self.db.execute("""
 506                                 TRUNCATE TABLE autnum_overrides;
 507                                 TRUNCATE TABLE network_overrides;
 508                         """)
 509
 510                         for file in ns.files:
 511                                 log.info("Reading %s..." % file)
 512
 513                                 with open(file, "rb") as f:
 514                                         for type, block in location.importer.read_blocks(f):
 515                                                 if type == "net":
 516                                                         network = block.get("net")
 517                                                         # Try to parse and normalise the network
 518                                                         try:
 519                                                                 network = ipaddress.ip_network(network, strict=False)
 520                                                         except ValueError as e:
 521                                                                 log.warning("Invalid IP network: %s: %s" % (network, e))
 522                                                                 continue
 523
 524                                                         self.db.execute("""
 525                                                                 INSERT INTO network_overrides(
 526                                                                         network,
 527                                                                         country,
 528                                                                         is_anonymous_proxy,
 529                                                                         is_satellite_provider,
 530                                                                         is_anycast
 531                                                                 ) VALUES (%s, %s, %s, %s)
 532                                                                 ON CONFLICT (network) DO NOTHING""",
 533                                                                 "%s" % network,
 534                                                                 block.get("country"),
 535                                                                 block.get("is-anonymous-proxy") == "yes",
 536                                                                 block.get("is-satellite-provider") == "yes",
 537                                                                 block.get("is-anycast") == "yes",
 538                                                         )
 539
 540                                                 elif type == "autnum":
 541                                                         autnum = block.get("autnum")
 542
 543                                                         # Check if AS number begins with "AS"
 544                                                         if not autnum.startswith("AS"):
 545                                                                 log.warning("Invalid AS number: %s" % autnum)
 546                                                                 continue
 547
 548                                                         # Strip "AS"
 549                                                         autnum = autnum[2:]
 550
 551                                                         self.db.execute("""
 552                                                                 INSERT INTO autnum_overrides(
 553                                                                         number,
 554                                                                         name,
 555                                                                         is_anonymous_proxy,
 556                                                                         is_satellite_provider,
 557                                                                         is_anycast
 558                                                                 ) VALUES(%s, %s, %s, %s, %s)
 559                                                                 ON CONFLICT DO NOTHING""",
 560                                                                 autnum, block.get("name"),
 561                                                                 block.get("is-anonymous-proxy") == "yes",
 562                                                                 block.get("is-satellite-provider") == "yes",
 563                                                                 block.get("is-anycast") == "yes",
 564                                                         )
 565
 566                                                 else:
 567                                                         log.warning("Unsupport type: %s" % type)
 568
 569
 570 def split_line(line):
 571         key, colon, val = line.partition(":")
 572
 573         # Strip any excess space
 574         key = key.strip()
 575         val = val.strip()
 576
 577         return key, val
 578
 579 def main():
 580         # Run the command line interface
 581         c = CLI()
 582         c.run()
 583
 584 main()