def test_subscriber_review_ads_drops_old_ads_for_that_city():
mySubscriber = Subscriber(4711)
ad1 = Ad.from_dict(nettenMenschenDict)
ad2 = Ad.from_dict(mitbewohnerinFuer21qm)
mySubscriber.review_ads({ad1}, 'ber') == set()
assert mySubscriber.already_had(ad1)
assert mySubscriber.already_had(ad2) == False
def test_subscriber_review_ads_returns_new_ads():
mySubscriber = Subscriber(4711)
ad1 = Ad.from_dict(nettenMenschenDict)
ad2 = Ad.from_dict(mitbewohnerinFuer21qm)
mySubscriber.review_ads({}, 'ber')
assert mySubscriber.review_ads({ad1}, 'ber') == {ad1}
assert mySubscriber.review_ads({ad1, ad2}, 'ber') == {ad2}
assert mySubscriber.review_ads({ad1, ad2}, 'ber') == set()
assert mySubscriber.review_ads({ad1, ad2}, 'ber') == set()
def test_subscriber_review_ads_drops_old_ads():
mySubscriber = Subscriber(4711)
ad1 = Ad.from_dict(nettenMenschenDict)
ad2 = Ad.from_dict(mitbewohnerinFuer21qm)
mySubscriber.review_ads({ad1, ad2}, 'ber')
# empty, because ad2 was in the list before
assert mySubscriber.review_ads({ad2}, 'ber') == set()
# now, this behaviour happens if an ad was updated after a day
# it was not scraped the last time, but now it should be resent
assert mySubscriber.review_ads({ad1, ad2}, 'ber') == {ad1}
def test_subscriber_review_ads_drops_old_ads_for_that_city():
mySubscriber = Subscriber(4711)
ad1 = Ad.from_dict(nettenMenschenDict)
ad2 = Ad.from_dict(mitbewohnerinFuer21qm)
assert mySubscriber.review_ads({ad1}, 'ber') == set()
assert mySubscriber.review_ads({ad2}, 'muc') == set()
assert mySubscriber.review_ads({ad1}, 'ber') == set()
assert mySubscriber.review_ads({}, 'ber') == set()
assert mySubscriber.review_ads({ad1}, 'ber') == {ad1}
# these two sets are completely independent of each other
assert mySubscriber.review_ads({ad2}, 'muc') == set()
def get_ads_from_listings(listings: List[BeautifulSoup],
city: str,
first_run=False) -> set:
new_ads = set()
for listing in listings:
links = listing.find_all('a', class_='detailansicht')
link_to_offer = 'https://www.wg-gesucht.de/{}'.format(
links[0].get_attribute_list('href')[0])
# logging.info('new offer: {}'.format(link_to_offer))
price_wrapper = listing.find(class_="detail-size-price-wrapper")
link_named_price = price_wrapper.find(class_="detailansicht")
# print(list(link_named_price.children))
size, rent = next(link_named_price.children).replace(' ', '').replace(
'\n', '').replace('€', '').split('|')
mates = link_named_price.find('span').get_attribute_list('title')[0]
# print(mates)
searching_for = link_named_price.find_all(
'img')[-1].get_attribute_list('alt')[0].replace(
'Mitbewohnerin',
'🚺').replace('Mitbwohner',
'🚹').replace('Mitbewohner', '🚹')
headline = listing.find(class_='headline-list-view')
# mates = headline.find('span').get_attribute_list('title')[0]
# emojis read faster -- also note the typo from the page missing the first e in mitbewohner
# searching_for = headline.find_all('img')[-1].get_attribute_list('alt')[0].replace('Mitbewohnerin', '🚺')
# searching_for = searching_for.replace('Mitbwohner', '🚹').replace('Mitbewohner', '🚹')
title = headline.find('a').text.replace('\n', '').strip()
location_and_availability = listing.find('p')
location_and_availability_split = location_and_availability.text[
location_and_availability.text.index('in'):].replace('\n',
'').split()
index_avail = location_and_availability_split.index('Verfügbar:')
location = ' '.join(location_and_availability_split[:index_avail])
availability = ' '.join(location_and_availability_split[index_avail:])
wg_details = '{} {}'.format(mates, location)
info = {
'city': city,
'url': link_to_offer,
'title': title,
'size': size,
'rent': rent,
'availability': availability,
'wg_details': wg_details,
'searching_for': searching_for,
}
ad = Ad.from_dict(info)
new_ads.add(ad)
return new_ads
def test_filter_available_3months_is_ok_forever():
ad = Ad.from_dict(mitbewohnerinFuer21qm)
mySubscriber = Subscriber(4711)
mySubscriber.add_filter(FilterAvailability, datetime.timedelta(weeks=12))
assert mySubscriber.is_interested_in(ad)
def test_parsing_of_availability():
ad = Ad.from_dict(mitbewohnerinFuer21qm)
assert ad.available_from() == datetime.datetime(day=25, month=2, year=2018)
assert ad.available_to() == None
def test_parsing_of_availability():
ad = Ad.from_dict(nettenMenschenDict)
assert ad.available_from() == datetime.datetime(day=1, month=3, year=2018)
assert ad.available_to() == datetime.datetime(day=31, month=3, year=2018)
def test_ad_to_chat_message():
ad = Ad.from_dict(nettenMenschenDict)
assert ad.to_chat_message() == nettenMenschenString
def test_filter_gender_ok():
mySubscriber = Subscriber(4711)
mySubscriber.add_filter(FilterGender, 'w')
ad = Ad.from_dict(mitbewohnerinFuer21qm)
assert mySubscriber.is_interested_in(ad)
def test_filter_rent_ok():
mySubscriber = Subscriber(4711)
mySubscriber.add_filter(FilterRent, 600)
ad = Ad.from_dict(mitbewohnerinFuer21qm)
assert mySubscriber.is_interested_in(ad)
def test_filter_rent_too_expensive():
mySubscriber = Subscriber(4711)
mySubscriber.add_filter(FilterRent, 500)
ad = Ad.from_dict(mitbewohnerinFuer21qm)
assert mySubscriber.is_interested_in(ad) == False
def test_filter_city_not_ok():
mySubscriber = Subscriber(4711)
mySubscriber.add_filter(FilterCity, ['muc', 'stuggi'])
ad = Ad.from_dict(mitbewohnerinFuer21qm)
assert mySubscriber.is_interested_in(ad) == False
def test_subscriber_review_ads_returns_empty_set_on_first_run():
mySubscriber = Subscriber(4711)
ad1 = Ad.from_dict(nettenMenschenDict)
ad2 = Ad.from_dict(mitbewohnerinFuer21qm)
# this behaviour will prevent current ads to be sent to the user
assert mySubscriber.review_ads({ad1, ad2}, 'ber') == set()
def test_filter_available_2months_ok():
ad = Ad.from_dict(nettenMenschenDict)
mySubscriber = Subscriber(4711)
mySubscriber.add_filter(FilterAvailability, datetime.timedelta(weeks=4))
assert mySubscriber.is_interested_in(ad)