def _get_matches_info(self, match_elements, championship_urls):
matches = []
base_len = len(OneXBetScraper._BASE_URL)
for match_element in list(match_elements)[:]:
try:
url = match_element.get_attribute('href')[base_len:]
except StaleElementReferenceException:
print('Caught StaleElementReferenceException')
continue
if url in championship_urls or match_element.get_attribute(
'class') != 'link' or url.endswith('-Special-bets/'):
continue
match_title_text = match_element.find_element_by_class_name(
'gname').text
match_title = MatchTitle.from_str(match_title_text)
date_time_str = match_element.find_element_by_class_name(
'date').text
try:
date_time = DateTime.from_1xbet_str(date_time_str)
except ValueError:
print(url)
continue
match = Match(match_title, self._BASE_URL + url, date_time, self)
matches.append(match)
return matches
def _get_match_basic_data(self, event):
date = event.find_element_by_class_name('event--date').text
time = event.find_element_by_class_name('event--time').text
date_time_str = date + time
date_time = DateTime.from_favorit_str(date_time_str)
name = event.find_element_by_class_name('long--name').text.lower()
button = event.find_element_by_class_name('event--more')
self.renderer.click(button)
url = self.renderer.current_url
return Match(MatchTitle(name.split(' - ')), url, date_time, self)
def _get_match_basic_data(self, match):
date_time_str = match.find_element_by_class_name('date').text
date_time = DateTime.from_marathon_str(date_time_str)
url = match.find_element_by_class_name('member-link').get_attribute(
'href')
teams = [
el.text for el in match.find_elements_by_tag_name('span')
if el.get_attribute('data-member-link')
]
return Match(MatchTitle(teams), url, date_time, self)
def test_good_init_3(self):
dt = DateTime.from_dict({
'year': 2005,
'month': 7,
'day': 14,
'hour': 12,
'minute': 30,
'second': 10
})
self.assertIsInstance(dt, DateTime)
def create_programme(self, programme):
cer = programme["cod_evento_rejilla"]
if cer in self.cache:
return self.cache.get(cer)
info = {
"channel": programme["cod_cadena_tv"],
"title": programme["des_evento_rejilla"],
"category": programme["des_genero"],
"start": DateTime().parse(programme["f_evento_rejilla"]),
"stop": DateTime().parse(programme["f_fin_evento_rejilla"])
}
cee = programme["cod_elemento_emision"]
if cee and DOWNLOAD_EXTRA_INFO:
info.update(Movistar.get_extra_info(cee))
# DAYS_TO_DOWNLOAD + 1 we add the day before data | 86400 seconds in a day
expire_time = (DAYS_TO_DOWNLOAD + 1) * 86400
self.cache.set(cer, info, expire=expire_time)
return info
def _get_championship_matches_info(self, url):
soup = self._get_soup(url)
matches = []
bks = self._get_bks(soup)
for bk in bks:
match_title = ParimatchScraper._get_match_title(bk)
date_time_str = ParimatchScraper._get_match_date_time_str(bk)
date_time = DateTime.from_parimatch_str(date_time_str)
match = Match(match_title, url, date_time, self)
match.id = bk.find(class_='no').text
matches.append(match)
return matches
def from_dict(cls, match_dict):
url = None
date_time = None
title = None
key = list(match_dict.keys())[0]
found = re.search(r'^(https://.+?) (.+?): (.+?)$', key)
if found:
url = found.group(1)
date_time = DateTime.fromisoformat(found.group(2))
title = MatchTitle.from_str(found.group(3))
bets_dict = list(match_dict.values())[0][0]
bets = [
Bet.from_dict({bet_title: odds})
for bet_title, odds in bets_dict.items()
]
return cls(title, url, date_time, None, bets)
def set_post(self,
title="",
category="",
describes="",
id="",
tag="",
content="",
datetime=DateTime()):
self.title = title
self.category = category
self.describes = describes
self.id = id
if type(tag) == str:
self.tag = convert_str_to_array(tag).copy()
elif type(tag) == list:
self.tag = tag[:]
self.content = content
self.date = datetime
def __init__(self,
title="",
category="",
describes="",
id="",
tag="",
content="",
times=0,
datetime=DateTime()):
self.title = title
self.category = category
self.describes = describes
self.id = id
if type(tag) is str:
self.tag = convert_str_to_array(tag).copy()
elif type(tag) is list:
self.tag = tag.copy()
self.content = content
self.date = datetime
self.times = times
def get_matches_info_sport(self, sport_name):
matches = []
subsections = self.get_tournaments(sport_name)
for subsection in subsections:
# if not LIVE:
try:
self.renderer.click(subsection)
except StaleElementReferenceException:
continue
# print(' ', subsections.index(subsection) + 1)
time.sleep(2)
events = self.renderer.find_elements_by_class_name(
'sportEventRow__body___3Ywcg')
time.sleep(2)
for event in events:
try:
event.find_element_by_class_name(
'matchDateTime__isLive___8f4IP')
continue # match is live
except NoSuchElementException:
pass
except StaleElementReferenceException:
continue
date_text = event.find_element_by_class_name(
'matchDateTime__date___2Hw-c').text
teams_webel = event.find_elements_by_class_name(
'__app-LogoTitle-wrapper')
teams = [el.text for el in teams_webel]
url = teams_webel[0].find_element_by_tag_name(
'a').get_attribute('href')
date = DateTime.from_ggbet_str(date_text)
matches.append(Match(MatchTitle(teams), url, date, self))
# if not LIVE:
self.renderer.click(subsection)
print(len(matches))
return Sport(sport_name, matches)
def test_not_equal_day(self):
assert DateTime(0, 0, 0) != DateTime(1, 0, 0)
def test_not_equal(self):
assert DateTime(0, 0, 0) != DateTime(0, 1, 0)
def test_equal_day(self):
assert DateTime(1, 0, 0) == DateTime(1, 0, 0)
def test_greater_minute(self):
assert DateTime(0, 0, 0) < DateTime(0, 0, 1)
def test_good7(self):
self.stamp = DateTime.check_timestamp("25/12/2013")
correct_stamp = datetime(2013, 12, 25, 23, 59)
self.assertEqual( correct_stamp, self.stamp)
def test_not_greater_equal_day(self):
assert not DateTime(1, 0, 0) <= DateTime(0, 0, 0)
def test_in_range(self):
sut = DateRange(DateTime(3, 22, 0), DateTime(3, 24, 0))
assert sut.in_range(DateTime(3, 23, 0))
def test_bad2(self):
self.stamp = DateTime.check_timestamp("2014wuityhwo")
correct_stamp = datetime(2014, 1, 1, 00, 00)
self.assertEqual( correct_stamp, self.stamp)
def test_bad1(self):
self.stamp = DateTime.check_timestamp("$£%^H2ivoy0")
correct_stamp = datetime(1111, 1, 1, 00, 00)
self.assertEqual( correct_stamp, self.stamp)
def test_good_not_equal(self):
dt_1 = DateTime()
dt_2 = DateTime(second=800)
self.assertNotEqual(dt_1, dt_2)
def test_bad4(self):
self.stamp = DateTime.check_timestamp("11/23/08")
correct_stamp = datetime(2008, 11, 23, 1, 00)
self.assertEqual( correct_stamp, self.stamp)
def test_bad_not_equal(self):
dt_1 = DateTime()
dt_2 = DateTime()
self.assertEqual(dt_1, dt_2)
def test_good_equal(self):
dt_1 = DateTime()
dt_2 = DateTime()
self.assertEqual(dt_1, dt_2)
def test_middle5(self):
self.stamp = DateTime.check_timestamp("2013/02/30")
correct_stamp = datetime(2013, 2, 28, 23, 59)
self.assertEqual( correct_stamp, self.stamp)
def test_greater_day(self):
assert DateTime(0, 0, 0) < DateTime(1, 0, 0)
def test_vbad0(self):
self.stamp = DateTime.check_timestamp("iutoeht")
correct_stamp = datetime(1111, 1, 1, 00, 00)
self.assertEqual( correct_stamp, self.stamp)
def test_greater_hour(self):
assert DateTime(0, 0, 0) < DateTime(0, 1, 0)
def test_not_greater_equal_hour(self):
assert not DateTime(0, 2, 0) <= DateTime(0, 1, 0)
def test_not_greater_day(self):
assert not DateTime(2, 0, 0) < DateTime(1, 0, 0)
def test_not_greater_equal_minute(self):
assert not DateTime(0, 0, 2) <= DateTime(0, 0, 1)
def test_not_in_range(self):
sut = DateRange(DateTime(0, 0, 0), DateTime(0, 1, 0))
assert not sut.in_range(DateTime(0, 2, 0))
def test_equal(self):
assert DateTime(0, 0, 0) == DateTime(0, 0, 0)
while True:
s = SequenceMatcher(None, first_team, second_team)
substring = s.find_longest_match(0, len(first_team), 0,
len(second_team))
if substring.size < min(3, min_initial_length):
break
substrings_total_length += substring.size
# print(first_team[substring.a:substring.b])
first_team = first_team[:substring.a] + first_team[substring.a +
substring.size:]
second_team = second_team[:substring.
b] + second_team[substring.b +
substring.size:]
similarity = substrings_total_length / min_initial_length + \
(substrings_total_length - max_initial_length) / (10 * max_initial_length)
return similarity
if __name__ == '__main__':
certainty = 0.5
comparator = MatchComparator()
m1 = Match(MatchTitle(['jd gaming', 'team we']), '123',
DateTime(2020, 12, 26, 12, 30), 1, [])
m2 = Match(MatchTitle(['jd', 'we']), '321', DateTime(2020, 12, 26, 13), 2,
[])
similarity = comparator.calculate_matches_similarity(m1, m2, certainty)
print(similarity)
print(comparator.similar(m1, m2, certainty))
print(comparator.similarities)
def test_good8(self):
self.stamp = DateTime.check_timestamp("25-12-2013")
correct_stamp = datetime(2013, 12, 25, 00, 59)
self.assertEqual( correct_stamp, self.stamp)
def set_date(self, datetime=DateTime()):
self.date = datetime
def test_less_hour(self):
assert DateTime(0, 1, 0) > DateTime(0, 0, 0)
def test_less_minute(self):
assert DateTime(0, 0, 1) > DateTime(0, 0, 0)
def setUp(self):
self.stamp = DateTime.check_timestamp("")
