Jura

Attrap_pref39.py

@@ -77,8 +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']}"
@@ -87,11 +88,14 @@ class Attrap_pref39(Attrap):
 
                 url = unquote(url)
                 name = a.get_text().strip()
-                date_str = name.strip("RAA-39-").split(" ")[0]
-                try:
-                    date = datetime.datetime.strptime(date_str.strip("_"), "%Y-%m-0%d")
-                except ValueError:
-                    date = datetime.datetime.strptime(date_str.strip("_"), "%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)
         return elements