pref59: détecte l'URL de l'année voulue automatiquement

Attrap_pref59.py

@@ -15,14 +15,7 @@ class Attrap_pref59(Attrap):
 
     # Config
     __HOST = 'https://www.nord.gouv.fr'
-    __RAA_PAGE = {
-        '2024': f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord/2024',
-        '2023': f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord/2023',
-        '2022': f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord/2022',
-        '2021': f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord/2021',
-        '2020': f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord/2020',
-        '2019': f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord/2019'
-    }
+    __RAA_PAGE = f'{__HOST}/Publications/Recueils-des-actes-administratifs/RAA-du-departement-du-Nord'
     __USER_AGENT = 'Mozilla/5.0 (Windows NT 10.0; rv:109.0) Gecko/20100101 Firefox/115.0'
     full_name = 'Préfecture du Nord'
     short_code = 'pref59'
@@ -32,22 +25,21 @@ class Attrap_pref59(Attrap):
         self.set_sleep_time(30)
 
     def get_raa(self, keywords):
-        pages_to_parse = []
-        if self.not_before.year <= 2024:
-            pages_to_parse.append(self.__RAA_PAGE['2024'])
-        if self.not_before.year <= 2023:
-            pages_to_parse.append(self.__RAA_PAGE['2023'])
-        if self.not_before.year <= 2022:
-            pages_to_parse.append(self.__RAA_PAGE['2022'])
-        if self.not_before.year <= 2021:
-            pages_to_parse.append(self.__RAA_PAGE['2021'])
-        if self.not_before.year <= 2020:
-            pages_to_parse.append(self.__RAA_PAGE['2020'])
-        if self.not_before.year <= 2019:
-            pages_to_parse.append(self.__RAA_PAGE['2019'])
+        # On récupère les pages d'années
+        year_pages = []
+        page_content = self.get_page(self.__RAA_PAGE, 'get').content
+        for card in self.get_sub_pages(
+            page_content,
+            'div.fr-card__body div.fr-card__content h2.fr-card__title a',
+            self.__HOST,
+            False
+        ):
+            if Attrap.guess_date(card['name'], '([0-9]{4})').year >= self.not_before.year:
+                year_pages.append(card['url'])
 
+        pages_to_parse = []
         elements = []
-        for raa_page in pages_to_parse:
+        for raa_page in year_pages:
             page_content = self.get_page(raa_page, 'get').content
             sub_pages = self.get_sub_pages(
                 page_content,