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_addresses_are_processed_by_hunter(self):
config = Config(string=self.DUMMY_CONFIG)
hunter = Hunter(config, [DummyCrawler(addresses_as_links=True)], IdMaintainer(":memory:"))
exposes = hunter.hunt_flats()
self.assertTrue(count(exposes) > 4, "Expected to find exposes")
for expose in exposes:
self.assertFalse(expose['address'].startswith('http'), "Expected addresses to be processed by default")
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_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_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_filter_min_size(self):
min_size = 80
config = Config(string=self.FILTER_MIN_SIZE_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['size'])[0]) <
min_size, exposes))
if len(unfiltered) > 0:
for expose in unfiltered:
print("Got unfiltered expose: ", expose)
self.assertTrue(
len(unfiltered) == 0, "Expected small flats to be filtered")
def test_filter_max_price(self):
max_price = 1000
config = Config(string=self.FILTER_MAX_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]) >
max_price, exposes))
if len(unfiltered) > 0:
for expose in unfiltered:
print("Got unfiltered expose: ", expose)
self.assertTrue(
len(unfiltered) == 0, "Expected expensive flats to be filtered")
def test_filter_titles(self):
titlewords = ["wg", "tausch", "flat", "ruhig", "gruen"]
filteredwords = [
"wg", "tausch", "wochenendheimfahrer", "pendler", "zwischenmiete"
]
config = Config(string=self.FILTER_TITLES_CONFIG)
hunter = Hunter(config, [DummyCrawler(titlewords)],
IdMaintainer(":memory:"))
exposes = hunter.hunt_flats()
self.assertTrue(count(exposes) > 4, "Expected to find exposes")
unfiltered = list(
filter(
lambda expose: any(word in expose['title']
for word in filteredwords), exposes))
if len(unfiltered) > 0:
for expose in unfiltered:
print("Got unfiltered expose: ", expose)
self.assertTrue(len(unfiltered) == 0, "Expected words to be filtered")
def test_resolve_durations(self, m):
config = Config(string=self.DUMMY_CONFIG)
hunter = Hunter(config, [DummyCrawler()], IdMaintainer(":memory:"))
matcher = re.compile(
'maps.googleapis.com/maps/api/distancematrix/json')
m.get(
matcher,
text=
'{"status": "OK", "rows": [ { "elements": [ { "distance": { "text": "far", "value": 123 }, "duration": { "text": "days", "value": 123 } } ] } ]}'
)
exposes = hunter.hunt_flats()
self.assertTrue(count(exposes) > 4, "Expected to find exposes")
without_durations = list(
filter(lambda expose: 'durations' not in expose, exposes))
if len(without_durations) > 0:
for expose in without_durations:
print("Got expose: ", expose)
self.assertTrue(
len(without_durations) == 0, "Expected durations to be calculated")
def test_dont_crawl_other_urls(crawler):
exposes = crawler.crawl("https://www.example.com")
assert count(exposes) == 0