def test_hunt_flats(self):
config = Config(string=self.DUMMY_CONFIG)
hunter = Hunter(config,
[CrawlImmowelt(Config(string=self.DUMMY_CONFIG))],
IdMaintainer(":memory:"))
exposes = hunter.hunt_flats()
self.assertTrue(count(exposes) > 0, "Expected to find exposes")
def test_is_processed_works(mocker):
config = Config(string=IdMaintainerTest.DUMMY_CONFIG)
id_watch = IdMaintainer(":memory:")
hunter = Hunter(config, [DummyCrawler()], id_watch)
exposes = hunter.hunt_flats()
assert count(exposes) > 4
for expose in exposes:
assert id_watch.is_processed(expose['id'])
def test_is_processed_works(id_watch):
config = Config(string=CONFIG_WITH_FILTERS)
config.set_searchers([DummyCrawler()])
hunter = Hunter(config, id_watch)
exposes = hunter.hunt_flats()
assert count(exposes) > 4
for expose in exposes:
assert id_watch.is_processed(expose['id'])
def test_ids_are_added_to_maintainer(mocker):
config = Config(string=IdMaintainerTest.DUMMY_CONFIG)
id_watch = IdMaintainer(":memory:")
spy = mocker.spy(id_watch, "mark_processed")
hunter = Hunter(config, [DummyCrawler()], id_watch)
exposes = hunter.hunt_flats()
assert count(exposes) > 4
assert spy.call_count == 24
def test_exposes_are_returned_with_limit():
config = Config(string=IdMaintainerTest.CONFIG_WITH_FILTERS)
id_watch = IdMaintainer(":memory:")
hunter = Hunter(config, [DummyCrawler()], id_watch)
hunter.hunt_flats()
saved = id_watch.get_recent_exposes(10)
assert len(saved) == 10
expose = saved[0]
assert expose['title'] is not None
def test_exposes_are_returned_with_limit(id_watch):
config = Config(string=CONFIG_WITH_FILTERS)
config.set_searchers([DummyCrawler()])
hunter = Hunter(config, id_watch)
hunter.hunt_flats()
saved = id_watch.get_recent_exposes(10)
assert len(saved) == 10
expose = saved[0]
assert expose['title'] is not None
def test_exposes_are_saved_to_maintainer(id_watch):
config = Config(string=CONFIG_WITH_FILTERS)
config.set_searchers([DummyCrawler()])
hunter = Hunter(config, id_watch)
exposes = hunter.hunt_flats()
assert count(exposes) > 4
saved = id_watch.get_exposes_since(datetime.datetime.now() - datetime.timedelta(seconds=10))
assert len(saved) > 0
assert count(exposes) < len(saved)
def test_exposes_are_saved_to_maintainer():
config = Config(string=IdMaintainerTest.CONFIG_WITH_FILTERS)
id_watch = IdMaintainer(":memory:")
hunter = Hunter(config, [DummyCrawler()], id_watch)
exposes = hunter.hunt_flats()
assert count(exposes) > 4
saved = id_watch.get_exposes_since(datetime.datetime.now() - datetime.timedelta(seconds=10))
assert len(saved) > 0
assert count(exposes) < len(saved)
def test_exposes_are_returned_as_dictionaries(id_watch):
config = Config(string=CONFIG_WITH_FILTERS)
config.set_searchers([DummyCrawler()])
hunter = Hunter(config, id_watch)
hunter.hunt_flats()
saved = id_watch.get_exposes_since(datetime.datetime.now() - datetime.timedelta(seconds=10))
assert len(saved) > 0
expose = saved[0]
assert expose['title'] is not None
assert expose['created_at'] is not None
def test_exposes_are_returned_as_dictionaries():
config = Config(string=IdMaintainerTest.CONFIG_WITH_FILTERS)
id_watch = IdMaintainer(":memory:")
hunter = Hunter(config, [DummyCrawler()], id_watch)
hunter.hunt_flats()
saved = id_watch.get_exposes_since(datetime.datetime.now() - datetime.timedelta(seconds=10))
assert len(saved) > 0
expose = saved[0]
assert expose['title'] is not None
assert expose['created_at'] is not None
def test_exposes_are_returned_filtered(id_watch):
config = Config(string=CONFIG_WITH_FILTERS)
config.set_searchers([DummyCrawler()])
hunter = Hunter(config, id_watch)
hunter.hunt_flats()
hunter.hunt_flats()
filter = Filter.builder().max_size_filter(70).build()
saved = id_watch.get_recent_exposes(10, filter=filter)
assert len(saved) == 10
for expose in saved:
assert int(re.match(r'\d+', expose['size'])[0]) <= 70
def test_exposes_are_returned_filtered():
config = Config(string=IdMaintainerTest.CONFIG_WITH_FILTERS)
id_watch = IdMaintainer(":memory:")
hunter = Hunter(config, [DummyCrawler()], id_watch)
hunter.hunt_flats()
hunter.hunt_flats()
filter = Filter.builder().max_size_filter(70).build()
saved = id_watch.get_recent_exposes(10, filter_set=filter)
assert len(saved) == 10
for expose in saved:
assert int(re.match(r'\d+', expose['size'])[0]) <= 70
def test_filter_min_price(self):
min_price = 700
config = Config(string=self.FILTER_MIN_PRICE_CONFIG)
hunter = Hunter(config, [DummyCrawler()], IdMaintainer(":memory:"))
exposes = hunter.hunt_flats()
self.assertTrue(count(exposes) > 4, "Expected to find exposes")
unfiltered = list(
filter(
lambda expose: float(
re.search(r'\d+([\.,]\d+)?', expose['price'])[0]) <
min_price, exposes))
if len(unfiltered) > 0:
for expose in unfiltered:
print("Got unfiltered expose: ", expose)
self.assertTrue(
len(unfiltered) == 0, "Expected cheap flats to be filtered")
def client():
app.config['TESTING'] = True
with tempfile.NamedTemporaryFile(mode='w+') as temp_db:
app.config['HUNTER'] = Hunter(Config(string=DUMMY_CONFIG),
[CrawlImmowelt()],
IdMaintainer(temp_db.name))
with app.test_client() as client:
yield client
def launch_flat_hunt(config):
id_watch = IdMaintainer('%s/processed_ids.db' % config.database_location())
hunter = Hunter(config, id_watch)
hunter.hunt_flats()
while config.get('loop', dict()).get('active', False):
time.sleep(config.get('loop', dict()).get('sleeping_time', 60 * 10))
hunter.hunt_flats()
def launch_flat_hunt(config):
id_watch = IdMaintainer('%s/processed_ids.db' %
os.path.dirname(os.path.abspath(__file__)))
hunter = Hunter(config, id_watch)
hunter.hunt_flats()
while config.get('loop', dict()).get('active', False):
time.sleep(config.get('loop', dict()).get('sleeping_time', 60 * 10))
hunter.hunt_flats()
def launch_flat_hunt(config):
searchers = [CrawlImmobilienscout(), CrawlWgGesucht(),CrawlEbayKleinanzeigen()]
id_watch = IdMaintainer('%s/processed_ids.db' % os.path.dirname(os.path.abspath(__file__)))
hunter = Hunter()
hunter.hunt_flats(config, searchers, id_watch)
while config.get('loop', dict()).get('active', False):
time.sleep(config.get('loop', dict()).get('sleeping_time',60*10))
hunter.hunt_flats(config, searchers, id_watch)
def launch_flat_hunt(config):
"""Start the crawler loop"""
id_watch = IdMaintainer('%s/processed_ids.db' % config.database_location())
hunter = Hunter(config, all_searchers(config), id_watch,
RedisPubsub(config))
hunter.hunt_flats()
while config.get('loop', dict()).get('active', False):
time.sleep(config.get('loop', dict()).get('sleeping_time', 60 * 10))
hunter.hunt_flats()
class HunterTest(unittest.TestCase):
DUMMY_CONFIG = """
urls:
- https://www.immowelt.de/liste/berlin/wohnungen/mieten?roomi=2&prima=1500&wflmi=70&sort=createdate%2Bdesc
google_maps_api:
key: SOME_KEY
url: https://maps.googleapis.com/maps/api/distancematrix/json?origins={origin}&destinations={dest}&mode={mode}&sensor=true&key={key}&arrival_time={arrival}
enable: true
"""
def setUp(self):
self.hunter = Hunter(Config(string=self.DUMMY_CONFIG),
[CrawlImmowelt()], IdMaintainer(":memory:"))
def test_hunt_flats(self):
exposes = self.hunter.hunt_flats()
self.assertTrue(len(exposes) > 0, "Expected to find exposes")
from flathunter.crawl_immobilienscout import CrawlImmobilienscout
from flathunter.crawl_wggesucht import CrawlWgGesucht
from flathunter.crawl_immowelt import CrawlImmowelt
from flathunter.idmaintainer import IdMaintainer
from flathunter.googlecloud_idmaintainer import GoogleCloudIdMaintainer
from flathunter.hunter import Hunter
from flathunter.config import Config
from flathunter.web import app
searchers = [
CrawlImmobilienscout(),
CrawlWgGesucht(),
CrawlEbayKleinanzeigen(),
CrawlImmowelt()
]
if __name__ == '__main__':
# Use the SQLite DB file if we are running locally
id_watch = IdMaintainer('%s/processed_ids.db' %
os.path.dirname(os.path.abspath(__file__)))
else:
# Use Google Cloud DB if we run on the cloud
id_watch = GoogleCloudIdMaintainer()
hunter = Hunter(Config(), searchers, id_watch)
app.config["HUNTER"] = hunter
if __name__ == '__main__':
app.run(host='127.0.0.1', port=8080, debug=True)
def setUp(self):
self.hunter = Hunter(Config(string=self.DUMMY_CONFIG),
[CrawlImmowelt()], IdMaintainer(":memory:"))