import scrape import helper url = "https://www.landkreis-ansbach.de/Quicknavigation/Startseite/Aktuelle-Informationen-zum-Coronavirus.php?object=tx,1503.10.1&ModID=7&FID=2238.2854.1&NavID=2150.1" regex = "und damit insgesamt [0-9]+" date_regex = "Stand: .*?\)" date_format = "Stand: %d.%m.%Y)" case_func = lambda bs: helper.extract_case_num_directregex(bs.text, regex, 0) date_func = lambda bs: helper.extract_status_date_directregex( bs.text, date_regex, date_format, 0, "%Y-%m-%d") scrape.scrape(url, "09571", case_func, date_func)
from bs4 import BeautifulSoup import re import scrape import helper from database_interface import * url = "https://www.kreis-warendorf.de/aktuelles/startseite/" date_regex = r"\(Stand: .*?\)" req = scrape.request_url(url) bs = BeautifulSoup(req.text,"html.parser") text=bs.getText() status = helper.get_status(re.findall(date_regex,text)[0].replace("(","").replace(")","")) case_regex = r"zählen wir \d+ Infizierte im Kreis" cases = helper.extract_case_num_directregex(text, case_regex, 0) add_to_database("05570", status, cases, "Kreis Warendorf")