src/backend/mirrors.py

   1 #!/usr/bin/python
   2
   3 from __future__ import division
   4
   5 import datetime
   6 import logging
   7 import math
   8 import os.path
   9 import random
  10 import socket
  11 import time
  12 import tornado.httpclient
  13 import tornado.netutil
  14 import urlparse
  15
  16 from misc import Object
  17
  18 class Downloads(Object):
  19         @property
  20         def total(self):
  21                 ret = self.db.get("SELECT COUNT(*) AS total FROM log_download")
  22
  23                 return ret.total
  24
  25         @property
  26         def today(self):
  27                 ret = self.db.get("SELECT COUNT(*) AS today FROM log_download WHERE date::date = NOW()::date")
  28
  29                 return ret.today
  30
  31         @property
  32         def yesterday(self):
  33                 ret = self.db.get("SELECT COUNT(*) AS yesterday FROM log_download WHERE date::date = (NOW() - INTERVAL '1 day')::date")
  34
  35                 return ret.yesterday
  36
  37         @property
  38         def daily_map(self):
  39                 ret = self.db.query("WITH downloads AS (SELECT * FROM log_download \
  40                         WHERE DATE(date) BETWEEN (NOW()::date - INTERVAL '30 days') AND DATE(NOW())) \
  41                         SELECT DATE(date) AS date, COUNT(*) AS count FROM downloads \
  42                         GROUP BY DATE(date) ORDER BY date")
  43
  44                 return ret
  45
  46         def get_countries(self, duration="all"):
  47                 query = "SELECT country_code, count(country_code) AS count FROM log_download"
  48
  49                 if duration == "today":
  50                         query += " WHERE date::date = NOW()::date"
  51
  52                 query += " GROUP BY country_code ORDER BY count DESC"
  53
  54                 results = self.db.query(query)
  55                 ret = []
  56
  57                 count = sum([o.count for o in results])
  58                 if count:
  59                         for res in results:
  60                                 ret.append((res.country_code, res.count / count))
  61
  62                 return ret
  63
  64         def get_mirror_load(self, duration="all"):
  65                 query = "SELECT mirror, COUNT(mirror) AS count FROM log_download"
  66
  67                 if duration == "today":
  68                         query += " WHERE date::date = NOW()::date"
  69
  70                 query += " GROUP BY mirror ORDER BY count DESC"
  71
  72                 results = self.db.query(query)
  73                 ret = {}
  74
  75                 count = sum([o.count for o in results])
  76                 if count:
  77                         for res in results:
  78                                 mirror = self.mirrors.get(res.mirror)
  79                                 ret[mirror.hostname] = res.count / count
  80
  81                 return ret
  82
  83
  84 class Mirrors(Object):
  85         def check_all(self):
  86                 for mirror in self.get_all():
  87                         mirror.check()
  88
  89         def get(self, id):
  90                 return Mirror(self.backend, id)
  91
  92         def get_all(self):
  93                 res = self.db.query("SELECT * FROM mirrors WHERE enabled = %s", True)
  94
  95                 mirrors = []
  96                 for row in res:
  97                         mirror = Mirror(self.backend, row.id, row)
  98                         mirrors.append(mirror)
  99
 100                 return MirrorSet(self.backend, sorted(mirrors))
 101
 102         def get_by_hostname(self, hostname):
 103                 ret = self.db.get("SELECT * FROM mirrors WHERE hostname = %s", hostname)
 104
 105                 if ret:
 106                         return Mirror(self.backend, ret.id, ret)
 107
 108         def get_with_file(self, filename, country=None):
 109                 # XXX quick and dirty solution - needs a performance boost
 110                 mirror_ids = [m.mirror for m in self.db.query("SELECT mirror FROM mirror_files WHERE filename=%s", filename)]
 111
 112                 #if country:
 113                 #       # Sort out all mirrors that are not preferred to the given country
 114                 #       for mirror in self.get_for_country(country):
 115                 #               if not mirror.id in mirror_ids:
 116                 #                       mirror_ids.remove(mirror.id)
 117
 118                 mirrors = []
 119                 for mirror_id in mirror_ids:
 120                         mirror = self.get(mirror_id)
 121                         if not mirror.state == "UP":
 122                                 continue
 123                         mirrors.append(mirror)
 124
 125                 return mirrors
 126
 127         def get_for_location(self, location, max_distance=4000, filename=None):
 128                 if not location:
 129                         return []
 130
 131                 if filename:
 132                         res = self.db.query("\
 133                                 WITH client AS (SELECT point(%s, %s) AS location) \
 134                                 SELECT * FROM mirrors WHERE mirrors.state = %s \
 135                                         AND mirrors.id IN ( \
 136                                                 SELECT mirror FROM mirror_files WHERE filename = %s \
 137                                         ) AND mirrors.id IN ( \
 138                                         SELECT id FROM mirrors_locations, client \
 139                                                 WHERE geodistance(mirrors_locations.location, client.location) <= %s \
 140                                 )",
 141                                 location.latitude, location.longitude, "UP", filename, max_distance)
 142                 else:
 143                         res = self.db.query("\
 144                                 WITH client AS (SELECT point(%s, %s) AS location) \
 145                                 SELECT * FROM mirrors WHERE mirrors.state = %s AND mirrors.id IN ( \
 146                                         SELECT id FROM mirrors_locations, client \
 147                                                 WHERE geodistance(mirrors_locations.location, client.location) <= %s \
 148                                 )",
 149                                 location.latitude, location.longitude, "UP", max_distance)
 150
 151                 mirrors = []
 152                 for row in res:
 153                         mirror = Mirror(self.backend, row.id, row)
 154                         mirrors.append(mirror)
 155
 156                 return sorted(mirrors, reverse=True)
 157
 158         def get_all_files(self):
 159                 files = []
 160
 161                 for mirror in self.get_all():
 162                         if not mirror.state == "UP":
 163                                 continue
 164
 165                         for file in mirror.filelist:
 166                                 if not file in files:
 167                                         files.append(file)
 168
 169                 return files
 170
 171         def get_random(self, filename=None):
 172                 if filename:
 173                         ret = self.db.get("SELECT * FROM mirrors WHERE state = %s \
 174                                 AND mirrors.id IN (SELECT mirror FROM mirror_files \
 175                                         WHERE filename = %s) ORDER BY RANDOM() LIMIT 1", "UP", filename)
 176                 else:
 177                         ret = self.db.get("SELECT * FROM mirrors WHERE state = %s \
 178                                 ORDER BY RANDOM() LIMIT 1", "UP")
 179
 180                 if ret:
 181                         return Mirror(self.backend, ret.id, ret)
 182
 183         def file_exists(self, filename):
 184                 ret = self.db.get("SELECT 1 FROM mirror_files \
 185                         WHERE filename = %s LIMIT 1", filename)
 186
 187                 if ret:
 188                         return True
 189
 190                 return False
 191
 192
 193 class MirrorSet(Object):
 194         def __init__(self, backend, mirrors):
 195                 Object.__init__(self, backend)
 196
 197                 self._mirrors = mirrors
 198
 199         def __add__(self, other):
 200                 mirrors = []
 201
 202                 for mirror in self._mirrors + other._mirrors:
 203                         if mirror in mirrors:
 204                                 continue
 205
 206                         mirrors.append(mirror)
 207
 208                 return MirrorSet(self.backend, mirrors)
 209
 210         def __sub__(self, other):
 211                 mirrors = self._mirrors[:]
 212
 213                 for mirror in other._mirrors:
 214                         if mirror in mirrors:
 215                                 mirrors.remove(mirror)
 216
 217                 return MirrorSet(self.backend, mirrors)
 218
 219         def __iter__(self):
 220                 return iter(self._mirrors)
 221
 222         def __len__(self):
 223                 return len(self._mirrors)
 224
 225         def __str__(self):
 226                 return "<MirrorSet %s>" % ", ".join([m.hostname for m in self._mirrors])
 227
 228         def get_with_file(self, filename):
 229                 with_file = [m.mirror for m in self.db.query("SELECT mirror FROM mirror_files WHERE filename=%s", filename)]
 230
 231                 mirrors = []
 232                 for mirror in self._mirrors:
 233                         if mirror.id in with_file:
 234                                 mirrors.append(mirror)
 235
 236                 return MirrorSet(self.backend, mirrors)
 237
 238         def get_random(self):
 239                 mirrors = []
 240                 for mirror in self._mirrors:
 241                         for i in range(0, mirror.priority):
 242                                 mirrors.append(mirror)
 243
 244                 return random.choice(mirrors)
 245
 246         def get_for_location(self, location):
 247                 distance = 2500
 248                 mirrors = []
 249
 250                 if location:
 251                         while len(mirrors) <= 3 and distance <= 8000:
 252                                 for mirror in self._mirrors:
 253                                         if mirror in mirrors:
 254                                                 continue
 255
 256                                         mirror_distance = mirror.distance_to(location)
 257                                         if mirror_distance is None:
 258                                                 continue
 259
 260                                         if mirror_distance <= distance:
 261                                                 mirrors.append(mirror)
 262
 263                                 distance *= 1.2
 264
 265                 return MirrorSet(self.backend, mirrors)
 266
 267         def get_with_state(self, state):
 268                 mirrors = []
 269
 270                 for mirror in self._mirrors:
 271                         if mirror.state == state:
 272                                 mirrors.append(mirror)
 273
 274                 return MirrorSet(self.backend, mirrors)
 275
 276
 277 class Mirror(Object):
 278         def __init__(self, backend, id, data=None):
 279                 Object.__init__(self, backend)
 280
 281                 self.id = id
 282
 283                 if data:
 284                         self._info = data
 285                 else:
 286                         self._info = self.db.get("SELECT * FROM mirrors WHERE id = %s", self.id)
 287                 self._info["url"] = self.generate_url()
 288
 289                 self.__location = None
 290                 self.__country_name = None
 291
 292         def __repr__(self):
 293                 return "<%s %s>" % (self.__class__.__name__, self.url)
 294
 295         def __cmp__(self, other):
 296                 ret = cmp(self.country_code, other.country_code)
 297
 298                 if not ret:
 299                         ret = cmp(self.hostname, other.hostname)
 300
 301                 return ret
 302
 303         def generate_url(self):
 304                 url = "%s://%s" % ("https" if self.supports_https else "http", self.hostname)
 305                 if not self.path.startswith("/"):
 306                         url += "/"
 307                 url += "%s" % self.path
 308                 if not self.path.endswith("/"):
 309                         url += "/"
 310                 return url
 311
 312         @property
 313         def hostname(self):
 314                 return self._info.hostname
 315
 316         @property
 317         def path(self):
 318                 return self._info.path
 319
 320         @property
 321         def supports_https(self):
 322                 return self._info.supports_https
 323
 324         @property
 325         def address(self):
 326                 for addr in self.addresses4:
 327                         return addr
 328
 329                 for addr in self.addresses6:
 330                         return addr
 331
 332         @property
 333         def owner(self):
 334                 return self._info.owner
 335
 336         @property
 337         def location(self):
 338                 if self.__location is None:
 339                         self.__location = self.geoip.get_location(self.address)
 340
 341                 return self.__location
 342
 343         @property
 344         def latitude(self):
 345                 if self.location:
 346                         return self.location.latitude
 347
 348         @property
 349         def longitude(self):
 350                 if self.location:
 351                         return self.location.longitude
 352
 353         @property
 354         def coordinates(self):
 355                 return (self.latitude, self.longitude)
 356
 357         @property
 358         def coordiante_str(self):
 359                 coordinates = []
 360
 361                 for i in self.coordinates:
 362                         coordinates.append("%s" % i)
 363
 364                 return ",".join(coordinates)
 365
 366         @property
 367         def country_code(self):
 368                 if self.location:
 369                         return self.location.country
 370
 371         @property
 372         def country_name(self):
 373                 if self.__country_name is None:
 374                         self.__country_name = self.geoip.get_country_name(self.country_code)
 375
 376                 return self.__country_name
 377
 378         @property
 379         def location_str(self):
 380                 location = []
 381
 382                 if self._info.location:
 383                         location.append(self._info.location)
 384
 385                 elif self.location:
 386                         location.append(self.location.city)
 387                         location.append(self.country_name)
 388
 389                 return ", ".join([s for s in location if s])
 390
 391         @property
 392         def asn(self):
 393                 if not hasattr(self, "__asn"):
 394                         self.__asn = self.geoip.get_asn(self.address)
 395
 396                 return self.__asn
 397
 398         @property
 399         def filelist(self):
 400                 filelist = self.db.query("SELECT filename FROM mirror_files WHERE mirror=%s ORDER BY filename", self.id)
 401                 return [f.filename for f in filelist]
 402
 403         @property
 404         def prefix(self):
 405                 return ""
 406
 407         @property
 408         def url(self):
 409                 return self._info.url
 410
 411         def build_url(self, filename):
 412                 return urlparse.urljoin(self.url, filename)
 413
 414         @property
 415         def last_update(self):
 416                 return self._info.last_update
 417
 418         @property
 419         def state(self):
 420                 return self._info.state
 421
 422         def set_state(self, state):
 423                 logging.info("Setting state of %s to %s" % (self.hostname, state))
 424
 425                 if self.state == state:
 426                         return
 427
 428                 self.db.execute("UPDATE mirrors SET state = %s WHERE id = %s", state, self.id)
 429
 430                 # Reload changed settings
 431                 if hasattr(self, "_info"):
 432                         self._info["state"] = state
 433
 434         @property
 435         def enabled(self):
 436                 return self._info.enabled
 437
 438         @property
 439         def disabled(self):
 440                 return not self.enabled
 441
 442         def check(self):
 443                 logging.info("Running check for mirror %s" % self.hostname)
 444
 445                 self.db.execute("UPDATE mirrors SET address = %s WHERE id = %s",
 446                         self.address, self.id)
 447
 448                 self.check_timestamp()
 449                 self.check_filelist()
 450
 451         def check_state(self):
 452                 logging.debug("Checking state of mirror %s" % self.id)
 453
 454                 if not self.enabled:
 455                         self.set_state("DOWN")
 456                         return
 457
 458                 now = datetime.datetime.utcnow()
 459
 460                 time_delta = now - self.last_update
 461                 time_diff = time_delta.total_seconds()
 462
 463                 time_down = self.settings.get_int("mirrors_time_down", 3*24*60*60)
 464                 if time_diff >= time_down:
 465                         self.set_state("DOWN")
 466                         return
 467
 468                 time_outofsync = self.settings.get_int("mirrors_time_outofsync", 6*60*60)
 469                 if time_diff >= time_outofsync:
 470                         self.set_state("OUTOFSYNC")
 471                         return
 472
 473                 self.set_state("UP")
 474
 475         def check_timestamp(self):
 476                 http = tornado.httpclient.AsyncHTTPClient()
 477
 478                 http.fetch(self.url + ".timestamp",
 479                         headers={ "Pragma" : "no-cache" },
 480                         callback=self.__check_timestamp_response)
 481
 482         def __check_timestamp_response(self, response):
 483                 if response.error:
 484                         logging.debug("Error getting timestamp from %s" % self.hostname)
 485                         self.set_state("DOWN")
 486                         return
 487
 488                 try:
 489                         timestamp = int(response.body.strip())
 490                 except ValueError:
 491                         timestamp = 0
 492
 493                 timestamp = datetime.datetime.utcfromtimestamp(timestamp)
 494
 495                 self.db.execute("UPDATE mirrors SET last_update = %s WHERE id = %s",
 496                         timestamp, self.id)
 497
 498                 # Reload changed settings
 499                 if hasattr(self, "_info"):
 500                         self._info["timestamp"] = timestamp
 501
 502                 self.check_state()
 503
 504                 logging.info("Successfully updated timestamp from %s" % self.hostname)
 505
 506         def check_filelist(self):
 507                 # XXX need to remove data from disabled mirrors
 508                 if not self.enabled:
 509                         return
 510
 511                 http = tornado.httpclient.AsyncHTTPClient()
 512
 513                 http.fetch(self.url + ".filelist",
 514                         headers={ "Pragma" : "no-cache" },
 515                         callback=self.__check_filelist_response)
 516
 517         def __check_filelist_response(self, response):
 518                 if response.error:
 519                         logging.debug("Error getting timestamp from %s" % self.hostname)
 520                         return
 521
 522                 files = self.filelist
 523
 524                 for file in response.body.splitlines():
 525                         file = os.path.join(self.prefix, file)
 526
 527                         if file in files:
 528                                 files.remove(file)
 529                                 continue
 530
 531                         self.db.execute("INSERT INTO mirror_files(mirror, filename) VALUES(%s, %s)",
 532                                 self.id, file)
 533
 534                 for file in files:
 535                         self.db.execute("DELETE FROM mirror_files WHERE mirror=%s AND filename=%s",
 536                                 self.id, file)
 537
 538                 logging.info("Successfully updated mirror filelist from %s" % self.hostname)
 539
 540         @property
 541         def prefer_for_countries(self):
 542                 countries = self._info.get("prefer_for_countries", "")
 543                 if countries:
 544                         return sorted(countries.split(", "))
 545
 546                 return []
 547
 548         @property
 549         def prefer_for_countries_names(self):
 550                 countries = [self.geoip.get_country_name(c.upper()) for c in self.prefer_for_countries]
 551
 552                 return sorted(countries)
 553
 554         def distance_to(self, location, ignore_preference=False):
 555                 if not location:
 556                         return None
 557
 558                 country_code = None
 559                 if location.country:
 560                         country_code = location.country.lower()
 561
 562                 if not ignore_preference and country_code in self.prefer_for_countries:
 563                         return 0
 564
 565                 # http://www.movable-type.co.uk/scripts/latlong.html
 566
 567                 if self.latitude is None:
 568                         return None
 569
 570                 if self.longitude is None:
 571                         return None
 572
 573                 earth = 6371 # km
 574                 delta_lat = math.radians(self.latitude - location.latitude)
 575                 delta_lon = math.radians(self.longitude - location.longitude)
 576
 577                 lat1 = math.radians(self.latitude)
 578                 lat2 = math.radians(location.latitude)
 579
 580                 a = math.sin(delta_lat / 2) ** 2
 581                 a += math.cos(lat1) * math.cos(lat2) * (math.sin(delta_lon / 2) ** 2)
 582
 583                 b1 = math.sqrt(a)
 584                 b2 = math.sqrt(1 - a)
 585
 586                 c = 2 * math.atan2(b1, b2)
 587
 588                 return c * earth
 589
 590         def traffic(self, since):
 591                 # XXX needs to be done better
 592
 593                 files = {}
 594                 for entry in self.db.query("SELECT filename, filesize FROM files"):
 595                         files[entry.filename] = entry.filesize
 596
 597                 query = "SELECT COUNT(filename) as count, filename FROM log_download WHERE mirror = %s"
 598                 query += " AND date >= %s GROUP BY filename"
 599
 600                 traffic = 0
 601                 for entry in self.db.query(query, self.id, since):
 602                         if files.has_key(entry.filename):
 603                                 traffic += entry.count * files[entry.filename]
 604
 605                 return traffic
 606
 607         @property
 608         def priority(self):
 609                 return self._info.get("priority", 10)
 610
 611         @property
 612         def development(self):
 613                 return self._info.get("mirrorlist_devel", False)
 614
 615         @property
 616         def mirrorlist(self):
 617                 return self._info.get("mirrorlist", False)
 618
 619         @property
 620         def addresses(self):
 621                 if not hasattr(self, "__addresses"):
 622                         try:
 623                                 addrinfo = socket.getaddrinfo(self.hostname, 0, socket.AF_UNSPEC, socket.SOCK_STREAM)
 624                         except:
 625                                 raise Exception("Could not resolve %s" % self.hostname)
 626
 627                         ret = []
 628                         for family, socktype, proto, canonname, address in addrinfo:
 629                                 if family == socket.AF_INET:
 630                                         address, port = address
 631                                 elif family == socket.AF_INET6:
 632                                         address, port, flowid, scopeid = address
 633                                 ret.append((family, address))
 634
 635                         self.__addresses = ret
 636
 637                 return self.__addresses
 638
 639         @property
 640         def addresses6(self):
 641                 return [address for family, address in self.addresses if family == socket.AF_INET6]
 642
 643         @property
 644         def addresses4(self):
 645                 return [address for family, address in self.addresses if family == socket.AF_INET]