def run(self) -> None: driver = Driver() driver = driver.get_driver() driver.get( 'https://domclick.ru/search?offset=10&limit=10&sort_dir=desc') html = driver.page_source soup = BeautifulSoup(html) max_ads = re.search( '(\d*)', soup.find('div', { 'class': 'listing-title' }).text.replace(' ', '')).group(1) for i in range(0, int(max_ads), 100): soup = BeautifulSoup( requests.get(self._ads_list_url_pattern.format(i)).text) hrefs = list(map(lambda a: a.get('href', ''), soup.find_all('a'))) for href in hrefs: if href.find('card') != -1: self._data.put(href)
def get_page(self, url): d = Driver().get_driver() d.get(url) return BeautifulSoup(d.page_source)