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("SELECT * FROM autnums \
 198                         WHERE name <> %s ORDER BY number", "")
 199
 200                 for row in rows:
 201                         a = writer.add_as(row.number)
 202                         a.name = row.name
 203
 204                 # Add all networks
 205                 log.info("Writing networks...")
 206
 207                 # Select all known networks
 208                 rows = self.db.query("""
 209                         SELECT
 210                                 announcements.network AS network,
 211                                 announcements.autnum AS autnum,
 212                                 (
 213                                         SELECT networks.country FROM networks
 214                                                 WHERE announcements.network <<= networks.network
 215                                 ) AS country,
 216
 217                                 -- Flags
 218                                 FALSE AS is_anonymous_proxy,
 219                                 FALSE AS is_satellite_provider,
 220                                 FALSE AS is_anycast
 221                         FROM announcements
 222                 """)
 223
 224                 for row in rows:
 225                         network = writer.add_network(row.network)
 226
 227                         # Save AS & country
 228                         network.asn, network.country_code = row.autnum, row.country
 229
 230                         # Set flags
 231                         if row.is_anonymous_proxy:
 232                                 network.set_flag(location.NETWORK_FLAG_ANONYMOUS_PROXY)
 233
 234                         if row.is_satellite_provider:
 235                                 network.set_flag(location.NETWORK_FLAG_SATELLITE_PROVIDER)
 236
 237                         if row.is_anycast:
 238                                 network.set_flag(location.NETWORK_FLAG_ANYCAST)
 239
 240                 # Write everything to file
 241                 log.info("Writing database to file...")
 242                 for file in ns.file:
 243                         writer.write(file)
 244
 245         def handle_update_whois(self, ns):
 246                 downloader = location.importer.Downloader()
 247
 248                 # Download all sources
 249                 with self.db.transaction():
 250                         # Create some temporary tables to store parsed data
 251                         self.db.execute("""
 252                                 CREATE TEMPORARY TABLE _autnums(number integer, organization text)
 253                                         ON COMMIT DROP;
 254                                 CREATE UNIQUE INDEX _autnums_number ON _autnums(number);
 255
 256                                 CREATE TEMPORARY TABLE _organizations(handle text, name text)
 257                                         ON COMMIT DROP;
 258                                 CREATE UNIQUE INDEX _organizations_handle ON _organizations(handle);
 259                         """)
 260
 261                         for source in location.importer.WHOIS_SOURCES:
 262                                 with downloader.request(source, return_blocks=True) as f:
 263                                         for block in f:
 264                                                 self._parse_block(block)
 265
 266                         self.db.execute("""
 267                                 INSERT INTO autnums(number, name)
 268                                         SELECT _autnums.number, _organizations.name FROM _autnums
 269                                                 LEFT JOIN _organizations ON _autnums.organization = _organizations.handle
 270                                 ON CONFLICT (number) DO UPDATE SET name = excluded.name;
 271                         """)
 272
 273                 # Download all extended sources
 274                 for source in location.importer.EXTENDED_SOURCES:
 275                         with self.db.transaction():
 276                                 # Download data
 277                                 with downloader.request(source) as f:
 278                                         for line in f:
 279                                                 self._parse_line(line)
 280
 281         def _parse_block(self, block):
 282                 # Get first line to find out what type of block this is
 283                 line = block[0]
 284
 285                 # aut-num
 286                 if line.startswith("aut-num:"):
 287                         return self._parse_autnum_block(block)
 288
 289                 # organisation
 290                 elif line.startswith("organisation:"):
 291                         return self._parse_org_block(block)
 292
 293         def _parse_autnum_block(self, block):
 294                 autnum = {}
 295                 for line in block:
 296                         # Split line
 297                         key, val = split_line(line)
 298
 299                         if key == "aut-num":
 300                                 m = re.match(r"^(AS|as)(\d+)", val)
 301                                 if m:
 302                                         autnum["asn"] = m.group(2)
 303
 304                         elif key == "org":
 305                                 autnum[key] = val
 306
 307                 # Skip empty objects
 308                 if not autnum:
 309                         return
 310
 311                 # Insert into database
 312                 self.db.execute("INSERT INTO _autnums(number, organization) \
 313                         VALUES(%s, %s) ON CONFLICT (number) DO UPDATE SET \
 314                                 organization = excluded.organization",
 315                         autnum.get("asn"), autnum.get("org"),
 316                 )
 317
 318         def _parse_org_block(self, block):
 319                 org = {}
 320                 for line in block:
 321                         # Split line
 322                         key, val = split_line(line)
 323
 324                         if key in ("organisation", "org-name"):
 325                                 org[key] = val
 326
 327                 # Skip empty objects
 328                 if not org:
 329                         return
 330
 331                 self.db.execute("INSERT INTO _organizations(handle, name) \
 332                         VALUES(%s, %s) ON CONFLICT (handle) DO \
 333                         UPDATE SET name = excluded.name",
 334                         org.get("organisation"), org.get("org-name"),
 335                 )
 336
 337         def _parse_line(self, line):
 338                 # Skip version line
 339                 if line.startswith("2"):
 340                         return
 341
 342                 # Skip comments
 343                 if line.startswith("#"):
 344                         return
 345
 346                 try:
 347                         registry, country_code, type, line = line.split("|", 3)
 348                 except:
 349                         log.warning("Could not parse line: %s" % line)
 350                         return
 351
 352                 # Skip any lines that are for stats only
 353                 if country_code == "*":
 354                         return
 355
 356                 if type in ("ipv6", "ipv4"):
 357                         return self._parse_ip_line(country_code, type, line)
 358
 359         def _parse_ip_line(self, country, type, line):
 360                 try:
 361                         address, prefix, date, status, organization = line.split("|")
 362                 except ValueError:
 363                         organization = None
 364
 365                         # Try parsing the line without organization
 366                         try:
 367                                 address, prefix, date, status = line.split("|")
 368                         except ValueError:
 369                                 log.warning("Unhandled line format: %s" % line)
 370                                 return
 371
 372                 # Skip anything that isn't properly assigned
 373                 if not status in ("assigned", "allocated"):
 374                         return
 375
 376                 # Cast prefix into an integer
 377                 try:
 378                         prefix = int(prefix)
 379                 except:
 380                         log.warning("Invalid prefix: %s" % prefix)
 381
 382                 # Fix prefix length for IPv4
 383                 if type == "ipv4":
 384                         prefix = 32 - int(math.log(prefix, 2))
 385
 386                 # Try to parse the address
 387                 try:
 388                         network = ipaddress.ip_network("%s/%s" % (address, prefix), strict=False)
 389                 except ValueError:
 390                         log.warning("Invalid IP address: %s" % address)
 391                         return
 392
 393                 self.db.execute("INSERT INTO networks(network, country) \
 394                         VALUES(%s, %s) ON CONFLICT (network) DO \
 395                         UPDATE SET country = excluded.country",
 396                         "%s" % network, country,
 397                 )
 398
 399         def handle_update_announcements(self, ns):
 400                 server = ns.server[0]
 401
 402                 # Pre-compile regular expression for routes
 403                 #route = re.compile(b"^\*>?\s[\si]?([^\s]+)[.\s]*?(\d+)\si$", re.MULTILINE)
 404                 route = re.compile(b"^\*[\s\>]i([^\s]+).+?(\d+)\si\r\n", re.MULTILINE|re.DOTALL)
 405
 406                 with telnetlib.Telnet(server) as t:
 407                         # Enable debug mode
 408                         #if ns.debug:
 409                         #       t.set_debuglevel(10)
 410
 411                         # Wait for console greeting
 412                         greeting = t.read_until(b"> ")
 413                         log.debug(greeting.decode())
 414
 415                         # Disable pagination
 416                         t.write(b"terminal length 0\n")
 417
 418                         # Wait for the prompt to return
 419                         t.read_until(b"> ")
 420
 421                         # Fetch the routing tables
 422                         with self.db.transaction():
 423                                 for protocol in ("ipv6", "ipv4"):
 424                                         log.info("Requesting %s routing table" % protocol)
 425
 426                                         # Request the full unicast routing table
 427                                         t.write(b"show bgp %s unicast\n" % protocol.encode())
 428
 429                                         # Read entire header which ends with "Path"
 430                                         t.read_until(b"Path\r\n")
 431
 432                                         while True:
 433                                                 # Try reading a full entry
 434                                                 # Those might be broken across multiple lines but ends with i
 435                                                 line = t.read_until(b"i\r\n", timeout=5)
 436                                                 if not line:
 437                                                         break
 438
 439                                                 # Show line for debugging
 440                                                 #log.debug(repr(line))
 441
 442                                                 # Try finding a route in here
 443                                                 m = route.match(line)
 444                                                 if m:
 445                                                         network, autnum = m.groups()
 446
 447                                                         # Convert network to string
 448                                                         network = network.decode()
 449
 450                                                         # Convert AS number to integer
 451                                                         autnum = int(autnum)
 452
 453                                                         log.info("Found announcement for %s by %s" % (network, autnum))
 454
 455                                                         self.db.execute("INSERT INTO announcements(network, autnum) \
 456                                                                 VALUES(%s, %s) ON CONFLICT (network) DO \
 457                                                                 UPDATE SET autnum = excluded.autnum, last_seen_at = CURRENT_TIMESTAMP",
 458                                                                 network, autnum,
 459                                                         )
 460
 461                                         log.info("Finished reading the %s routing table" % protocol)
 462
 463                                 # Purge anything we never want here
 464                                 self.db.execute("""
 465                                         -- Delete default routes
 466                                         DELETE FROM announcements WHERE network = '::/0' OR network = '0.0.0.0/0';
 467
 468                                         -- Delete anything that is not global unicast address space
 469                                         DELETE FROM announcements WHERE family(network) = 6 AND NOT network <<= '2000::/3';
 470
 471                                         -- DELETE RFC1918 address space
 472                                         DELETE FROM announcements WHERE family(network) = 4 AND network <<= '10.0.0.0/8';
 473                                         DELETE FROM announcements WHERE family(network) = 4 AND network <<= '172.16.0.0/12';
 474                                         DELETE FROM announcements WHERE family(network) = 4 AND network <<= '192.168.0.0/16';
 475
 476                                         -- Delete networks that are too small to be in the global routing table
 477                                         DELETE FROM announcements WHERE family(network) = 6 AND masklen(network) > 48;
 478                                         DELETE FROM announcements WHERE family(network) = 4 AND masklen(network) > 24;
 479
 480                                         -- Delete any non-public or reserved ASNs
 481                                         DELETE FROM announcements WHERE NOT (
 482                                                 (autnum >= 1 AND autnum <= 23455)
 483                                                 OR
 484                                                 (autnum >= 23457 AND autnum <= 64495)
 485                                                 OR
 486                                                 (autnum >= 131072 AND autnum <= 4199999999)
 487                                         );
 488
 489                                         -- Delete everything that we have not seen for 14 days
 490                                         DELETE FROM announcements WHERE last_seen_at <= CURRENT_TIMESTAMP - INTERVAL '14 days';
 491                                 """)
 492
 493         def handle_update_overrides(self, ns):
 494                 with self.db.transaction():
 495                         # Drop all data that we have
 496                         self.db.execute("""
 497                                 TRUNCATE TABLE autnum_overrides;
 498                                 TRUNCATE TABLE network_overrides;
 499                         """)
 500
 501                         for file in ns.files:
 502                                 log.info("Reading %s..." % file)
 503
 504                                 with open(file, "rb") as f:
 505                                         for type, block in location.importer.read_blocks(f):
 506                                                 if type == "net":
 507                                                         network = block.get("net")
 508                                                         # Try to parse and normalise the network
 509                                                         try:
 510                                                                 network = ipaddress.ip_network(network, strict=False)
 511                                                         except ValueError as e:
 512                                                                 log.warning("Invalid IP network: %s: %s" % (network, e))
 513                                                                 continue
 514
 515                                                         self.db.execute("""
 516                                                                 INSERT INTO network_overrides(
 517                                                                         network,
 518                                                                         country,
 519                                                                         is_anonymous_proxy,
 520                                                                         is_satellite_provider,
 521                                                                         is_anycast
 522                                                                 ) VALUES (%s, %s, %s, %s)
 523                                                                 ON CONFLICT (network) DO NOTHING""",
 524                                                                 "%s" % network,
 525                                                                 block.get("country"),
 526                                                                 block.get("is-anonymous-proxy") == "yes",
 527                                                                 block.get("is-satellite-provider") == "yes",
 528                                                                 block.get("is-anycast") == "yes",
 529                                                         )
 530
 531                                                 elif type == "autnum":
 532                                                         autnum = block.get("autnum")
 533
 534                                                         # Check if AS number begins with "AS"
 535                                                         if not autnum.startswith("AS"):
 536                                                                 log.warning("Invalid AS number: %s" % autnum)
 537                                                                 continue
 538
 539                                                         # Strip "AS"
 540                                                         autnum = autnum[2:]
 541
 542                                                         self.db.execute("""
 543                                                                 INSERT INTO autnum_overrides(
 544                                                                         number,
 545                                                                         name,
 546                                                                         is_anonymous_proxy,
 547                                                                         is_satellite_provider,
 548                                                                         is_anycast
 549                                                                 ) VALUES(%s, %s, %s, %s, %s)
 550                                                                 ON CONFLICT DO NOTHING""",
 551                                                                 autnum, block.get("name"),
 552                                                                 block.get("is-anonymous-proxy") == "yes",
 553                                                                 block.get("is-satellite-provider") == "yes",
 554                                                                 block.get("is-anycast") == "yes",
 555                                                         )
 556
 557                                                 else:
 558                                                         log.warning("Unsupport type: %s" % type)
 559
 560
 561 def split_line(line):
 562         key, colon, val = line.partition(":")
 563
 564         # Strip any excess space
 565         key = key.strip()
 566         val = val.strip()
 567
 568         return key, val
 569
 570 def main():
 571         # Run the command line interface
 572         c = CLI()
 573         c.run()
 574
 575 main()