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