Ejemplo n.º 1
0
def blacklist_moving_wifis(ago=1, offset=0, batch=1000):
    # TODO: this doesn't take into account wifi AP's which have
    # permanently moved after a certain date

    # maximum difference of two decimal places, ~1km at equator
    # or 500m at 67 degrees north
    max_difference = 100000
    day, max_day = daily_task_days(ago)
    try:
        with blacklist_moving_wifis.db_session() as session:
            query = session.query(distinct(
                WifiMeasure.key)).filter(WifiMeasure.created < max_day).filter(
                    WifiMeasure.created >= day).order_by(
                        WifiMeasure.id).limit(batch).offset(offset)
            new_wifis = [w[0] for w in query.all()]
            if not new_wifis:  # pragma: no cover
                # nothing to be done
                return []
            # check min/max lat/lon
            query = session.query(WifiMeasure.key, func.max(WifiMeasure.lat),
                                  func.min(WifiMeasure.lat),
                                  func.max(WifiMeasure.lon),
                                  func.min(WifiMeasure.lon)).filter(
                                      WifiMeasure.key.in_(new_wifis)).group_by(
                                          WifiMeasure.key)
            results = query.all()
            moving_keys = set()
            for result in results:
                wifi_key, max_lat, min_lat, max_lon, min_lon = result
                diff_lat = abs(max_lat - min_lat)
                diff_lon = abs(max_lon - min_lon)
                if diff_lat >= max_difference or diff_lon >= max_difference:
                    moving_keys.add(wifi_key)
            if moving_keys:
                utcnow = datetime.utcnow()
                query = session.query(WifiBlacklist.key).filter(
                    WifiBlacklist.key.in_(moving_keys))
                already_blocked = set([a[0] for a in query.all()])
                moving_keys = moving_keys - already_blocked
                if not moving_keys:
                    return []
                for key in moving_keys:
                    # TODO: on duplicate key ignore
                    session.add(WifiBlacklist(key=key, created=utcnow))
                remove_wifi.delay(list(moving_keys))
                session.commit()
            return moving_keys
    except IntegrityError as exc:  # pragma: no cover
        # TODO log error
        return []
    except Exception as exc:  # pragma: no cover
        raise blacklist_moving_wifis.retry(exc=exc)
Ejemplo n.º 2
0
    def test_wifi_blacklist(self):
        from ichnaea.service.submit.tasks import insert_wifi_measure
        session = self.db_master_session
        bad_key = "ab1234567890"
        good_key = "cd1234567890"
        black = WifiBlacklist(key=bad_key)
        session.add(black)
        session.flush()
        measure = dict(id=0, lat=10000000, lon=20000000)
        entries = [{"key": good_key}, {"key": good_key}, {"key": bad_key}]

        result = insert_wifi_measure.delay(measure, entries)
        self.assertEqual(result.get(), 2)

        measures = session.query(WifiMeasure).all()
        self.assertEqual(len(measures), 2)
        self.assertEqual(set([m.key for m in measures]), set([good_key]))
Ejemplo n.º 3
0
    def test_blacklist_time_used_as_creation_time(self):
        now = util.utcnow()
        last_week = now - TEMPORARY_BLACKLIST_DURATION - timedelta(days=1)
        session = self.db_master_session

        wifi_key = "ab1234567890"

        session.add(WifiBlacklist(time=last_week, count=1, key=wifi_key))
        session.flush()

        # add a new entry for the previously blacklisted wifi
        obs = dict(lat=PARIS_LAT, lon=PARIS_LON, key=wifi_key)
        insert_measures_wifi.delay([obs]).get()

        # the wifi was inserted again
        wifis = session.query(Wifi).all()
        self.assertEqual(len(wifis), 1)

        # and the creation date was set to the date of the blacklist entry
        self.assertEqual(wifis[0].created, last_week)
Ejemplo n.º 4
0
    def test_blacklist(self):
        session = self.db_master_session
        bad_key = "ab1234567890"
        good_key = "cd1234567890"
        black = WifiBlacklist(key=bad_key)
        session.add(black)
        session.flush()
        measure = dict(lat=1, lon=2)
        entries = [{"key": good_key}, {"key": good_key}, {"key": bad_key}]
        for e in entries:
            e.update(measure)

        result = insert_measures_wifi.delay(entries)
        self.assertEqual(result.get(), 2)

        measures = session.query(WifiMeasure).all()
        self.assertEqual(len(measures), 2)
        self.assertEqual(set([m.key for m in measures]), set([good_key]))

        wifis = session.query(Wifi).all()
        self.assertEqual(len(wifis), 1)
        self.assertEqual(set([w.key for w in wifis]), set([good_key]))
Ejemplo n.º 5
0
    def test_blacklist_moving_wifis(self):
        from ichnaea.tasks import blacklist_moving_wifis
        session = self.db_master_session
        k1 = "ab1234567890"
        k2 = "cd1234567890"
        k3 = "ef1234567890"
        k4 = "b01234567890"
        k5 = "d21234567890"
        measures = [
            WifiMeasure(lat=10010000, lon=10010000, key=k1),
            WifiMeasure(lat=10020000, lon=10050000, key=k1),
            WifiMeasure(lat=10030000, lon=10090000, key=k1),
            WifiMeasure(lat=20100000, lon=20000000, key=k2),
            WifiMeasure(lat=20200000, lon=20000000, key=k2),
            WifiMeasure(lat=30000000, lon=30000000, key=k3),
            WifiMeasure(lat=-30000000, lon=30000000, key=k3),
            WifiMeasure(lat=-41000000, lon=40000000, key=k4),
            WifiMeasure(lat=-41100000, lon=40000000, key=k4),
            WifiMeasure(lat=50000000, lon=50000000, key=k5),
            WifiMeasure(lat=51000000, lon=50000000, key=k5),
        ]
        session.add_all(measures)
        session.add(WifiBlacklist(key=k5))
        session.commit()

        result = blacklist_moving_wifis.delay(ago=0)
        self.assertEqual(sorted(result.get()), sorted([k2, k3, k4]))

        measures = session.query(WifiBlacklist).all()
        self.assertEqual(len(measures), 4)
        self.assertEqual(set([m.key for m in measures]), set([k2, k3, k4, k5]))

        measures = session.query(WifiMeasure).all()
        self.assertEqual(len(measures), 5)
        self.assertEqual(set([m.key for m in measures]), set([k1, k5]))

        # test duplicate call
        result = blacklist_moving_wifis.delay(ago=0)
        self.assertEqual(result.get(), [])
Ejemplo n.º 6
0
    def test_blacklist(self):
        utcnow = util.utcnow()
        session = self.db_master_session
        bad_key = "ab1234567890"
        good_key = "cd1234567890"
        black = WifiBlacklist(time=utcnow, count=1, key=bad_key)
        session.add(black)
        session.flush()
        obs = dict(lat=1, lon=2)
        entries = [{"key": good_key}, {"key": good_key}, {"key": bad_key}]
        for e in entries:
            e.update(obs)

        result = insert_measures_wifi.delay(entries)
        self.assertEqual(result.get(), 2)

        observations = session.query(WifiObservation).all()
        self.assertEqual(len(observations), 2)
        self.assertEqual(set([o.key for o in observations]), set([good_key]))

        wifis = session.query(Wifi).all()
        self.assertEqual(len(wifis), 1)
        self.assertEqual(set([w.key for w in wifis]), set([good_key]))
Ejemplo n.º 7
0
 def _make_one(self, **kw):
     from ichnaea.models import WifiBlacklist
     return WifiBlacklist(**kw)
Ejemplo n.º 8
0
    def test_blacklist_moving_wifis(self):
        now = util.utcnow()
        long_ago = now - timedelta(days=40)
        session = self.db_master_session
        k1 = "ab1234567890"
        k2 = "bc1234567890"
        k3 = "cd1234567890"
        k4 = "de1234567890"
        k5 = "ef1234567890"
        k6 = "fa1234567890"

        keys = set([k1, k2, k3, k4, k5, k6])

        # keys k2, k3 and k4 are expected to be detected as moving
        data = [
            # a wifi with an entry but no prior position
            Wifi(key=k1, new_measures=3, total_measures=0),
            WifiMeasure(lat=1.001, lon=1.001, key=k1),
            WifiMeasure(lat=1.002, lon=1.005, key=k1),
            WifiMeasure(lat=1.003, lon=1.009, key=k1),
            # a wifi with a prior known position
            Wifi(lat=2.0, lon=2.0, key=k2, new_measures=2, total_measures=1),
            WifiMeasure(lat=2.01, lon=2, key=k2),
            WifiMeasure(lat=2.07, lon=2, key=k2),
            # a wifi with a very different prior position
            Wifi(lat=1.0, lon=1.0, key=k3, new_measures=2, total_measures=1),
            WifiMeasure(lat=3.0, lon=3.0, key=k3),
            WifiMeasure(lat=-3.0, lon=3.0, key=k3),
            # another wifi with a prior known position (and negative lat)
            Wifi(lat=-4.0, lon=4.0, key=k4, new_measures=2, total_measures=1),
            WifiMeasure(lat=-4.1, lon=4, key=k4),
            WifiMeasure(lat=-4.16, lon=4, key=k4),
            # an already blacklisted wifi
            WifiBlacklist(key=k5),
            WifiMeasure(lat=5.0, lon=5.0, key=k5),
            WifiMeasure(lat=5.1, lon=5.0, key=k5),
            # a wifi with an old different record we ignore, position
            # estimate has been updated since
            Wifi(lat=6.0, lon=6.0, key=k6, new_measures=2, total_measures=1),
            WifiMeasure(lat=6.9, lon=6.9, key=k6, time=long_ago),
            WifiMeasure(lat=6.0, lon=6.0, key=k6),
            WifiMeasure(lat=6.001, lon=6.0, key=k6),
        ]
        session.add_all(data)
        session.commit()

        result = location_update_wifi.delay(min_new=1)
        self.assertEqual(result.get(), (5, 3))

        black = session.query(WifiBlacklist).all()
        self.assertEqual(set([b.key for b in black]), set([k2, k3, k4, k5]))

        measures = session.query(WifiMeasure).all()
        self.assertEqual(len(measures), 14)
        self.assertEqual(set([m.key for m in measures]), keys)

        # test duplicate call
        result = location_update_wifi.delay(min_new=1)
        self.assertEqual(result.get(), 0)

        self.check_stats(
            total=6,
            timer=[
                # We made duplicate calls
                ('task.data.location_update_wifi', 2),
                # One of those would've scheduled a remove_wifi task
                ('task.data.remove_wifi', 1)
            ],
            gauge=[
                ('task.data.location_update_wifi.new_measures_1_100', 2),
            ])
Ejemplo n.º 9
0
    def test_blacklist_moving_wifis(self):
        from ichnaea.tasks import wifi_location_update
        now = datetime.utcnow()
        long_ago = now - timedelta(days=40)
        session = self.db_master_session
        k1 = "ab1234567890"
        k2 = "bc1234567890"
        k3 = "cd1234567890"
        k4 = "de1234567890"
        k5 = "ef1234567890"
        k6 = "fa1234567890"

        keys = set([k1, k2, k3, k4, k5, k6])

        # keys k2, k3 and k4 are expected to be detected as moving
        data = [
            # a wifi with an entry but no prior position
            Wifi(key=k1, new_measures=3, total_measures=0),
            WifiMeasure(lat=10010000, lon=10010000, key=k1),
            WifiMeasure(lat=10020000, lon=10050000, key=k1),
            WifiMeasure(lat=10030000, lon=10090000, key=k1),
            # a wifi with a prior known position
            Wifi(lat=20000000,
                 lon=20000000,
                 key=k2,
                 new_measures=2,
                 total_measures=1),
            WifiMeasure(lat=20100000, lon=20000000, key=k2),
            WifiMeasure(lat=20700000, lon=20000000, key=k2),
            # a wifi with a very different prior position
            Wifi(lat=10000000,
                 lon=10000000,
                 key=k3,
                 new_measures=2,
                 total_measures=1),
            WifiMeasure(lat=30000000, lon=30000000, key=k3),
            WifiMeasure(lat=-30000000, lon=30000000, key=k3),
            # another wifi with a prior known position (and negative lat)
            Wifi(lat=-40000000,
                 lon=40000000,
                 key=k4,
                 new_measures=2,
                 total_measures=1),
            WifiMeasure(lat=-41000000, lon=40000000, key=k4),
            WifiMeasure(lat=-41600000, lon=40000000, key=k4),
            # an already blacklisted wifi
            WifiBlacklist(key=k5),
            WifiMeasure(lat=50000000, lon=50000000, key=k5),
            WifiMeasure(lat=51000000, lon=50000000, key=k5),
            # a wifi with an old different record we ignore, position
            # estimate has been updated since
            Wifi(lat=60000000,
                 lon=60000000,
                 key=k6,
                 new_measures=2,
                 total_measures=1),
            WifiMeasure(lat=69000000, lon=69000000, key=k6, created=long_ago),
            WifiMeasure(lat=60000000, lon=60000000, key=k6),
            WifiMeasure(lat=60010000, lon=60000000, key=k6),
        ]
        session.add_all(data)
        session.commit()

        result = wifi_location_update.delay(min_new=1)
        self.assertEqual(result.get(), (5, 3))

        black = session.query(WifiBlacklist).all()
        self.assertEqual(set([b.key for b in black]), set([k2, k3, k4, k5]))

        measures = session.query(WifiMeasure).all()
        self.assertEqual(len(measures), 14)
        self.assertEqual(set([m.key for m in measures]), keys)

        # test duplicate call
        result = wifi_location_update.delay(min_new=1)
        self.assertEqual(result.get(), 0)

        self.check_expected_heka_messages(
            total=4,
            timer=[
                # We made duplicate calls
                ('task.wifi_location_update', 2),

                # One of those would've scheduled a remove_wifi task
                ('task.remove_wifi', 1)
            ])