res = requests.get(urlstring) soup = bs4.BeautifulSoup(res.text, 'lxml') span = soup.find_all('span', {'class': 'actual-price'}) vehicles = soup.find_all('li', {'class': 'vehicle'}) for i in vehicles: Car_Price = CleanUp_Car_Price( str(i.find_all('span', {'class': 'actual-price'})[0])) Car_info_tab = i.find_all('div', {'class': 'small-7 columns'}) first_thing = Car_info_tab[0] Car_Make = CleanUp_Car_Name(str(first_thing.select('h3')[0])).replace( ' ', '_') Ref = str(i.find_all('div', {'class': 'ref-views'})[0])[36:43] Car_Age_Milage = CleanUp_Notes(str(first_thing.select('p')[1])) Year = Car_Age_Milage[0] Mileage = Car_Age_Milage[1] print('Reff:', Ref) print('Car Price:', Car_Price) print('Car Make:', Car_Make) print('Year:', Year) print('Mileage:', Mileage) #Car_Age_Milage, is notes. ## Put this stuff in a db conn = sqlite3.connect('Cars.db') cursor = conn.cursor()
def test_CleanUp_Notes(): list = CleanUp_Notes( '<p>MPV, Manual, Petrol, Grey, 2011 (11), 53,104 miles</p>') assert list[0] == 2011 assert list[1] == 53104