diff --git a/Attrap_pref39.py b/Attrap_pref39.py index 0ac61986173f82b77f1746c5b726963107b4be0c..45b565481802e457cf445e98545044ff95fb6460 100644 --- a/Attrap_pref39.py +++ b/Attrap_pref39.py @@ -77,9 +77,9 @@ class Attrap_pref39(Attrap): # On charge le parser soup = BeautifulSoup(page_content, "html.parser") - # On récupère chaque balise a - for a in soup.select("a.fr-card__link"): - + # On récupère chaque block de card + for block in soup.select(".fr-card__content"): + a = block.select("a.fr-card__link")[0] if a.get("href") and a["href"].endswith(".pdf"): if a["href"].startswith("/"): url = f"{self.__HOST}{a['href']}" @@ -89,9 +89,8 @@ class Attrap_pref39(Attrap): url = unquote(url) name = a.get_text().strip() - date_str = soup.select("p.fr-card__detail").get_text().replace("Publié le ", "") - print(date_str) - date = datetime.datetime.strptime(date_str, "%Y-%m-0%d") + date_str = block.select("p.fr-card__detail")[0].get_text().replace("Publié le ", "") + date = datetime.datetime.strptime(date_str, "%d/%m/%Y") raa = Attrap.RAA(url, date, name) elements.append(raa)