pref66: détecte l'URL de l'année voulue automatiquement

Attrap_pref66.py

@@ -15,14 +15,7 @@ class Attrap_pref66(Attrap):
 
     # Config
     __HOST = 'https://www.pyrenees-orientales.gouv.fr'
-    __RAA_PAGE = {
-        '2024': f'{__HOST}/Publications/Le-recueil-des-actes-administratifs/Annee-2024',
-        '2023': f'{__HOST}/Publications/Le-recueil-des-actes-administratifs/Annee-2023',
-        '2022': f'{__HOST}/Publications/Le-recueil-des-actes-administratifs/Annee-2022',
-        '2021': f'{__HOST}/Publications/Le-recueil-des-actes-administratifs/Annee-2021',
-        '2020': f'{__HOST}/Publications/Le-recueil-des-actes-administratifs/Annee-2020',
-        '2019': f'{__HOST}/Publications/Le-recueil-des-actes-administratifs/Annee-2019'
-    }
+    __RAA_PAGE = f'{__HOST}/Publications/Le-recueil-des-actes-administratifs'
     __USER_AGENT = 'Mozilla/5.0 (Windows NT 10.0; rv:109.0) Gecko/20100101 Firefox/115.0'
     full_name = 'Préfecture des Pyrénées-Orientales'
     short_code = 'pref66'
@@ -32,6 +25,19 @@ class Attrap_pref66(Attrap):
         self.set_sleep_time(30)
 
     def get_raa(self, keywords):
+        # On détermine quelles pages d'année parser
+        year_pages = []
+        page_content = self.get_page(self.__RAA_PAGE, 'get').content
+        for year_page in self.get_sub_pages(
+            page_content,
+            '.fr-table table tr td h3 a.fr-link',
+            self.__HOST,
+            False
+        ):
+            year = Attrap.guess_date(year_page['name'].strip(), '.* ([0-9]{4})').year
+            if year < 9999 and year >= self.not_before.year:
+                year_pages.append([year_page['url'], year])
+
         elements = []
 
         # La préfecture des Pyrénées-Orientales est une originale : avant 2024,
@@ -43,24 +49,16 @@ class Attrap_pref66(Attrap):
         # n'est pas exhaustif. On doit donc parser toutes les sous-pages de
         # 2024 puisqu'on ne peut se fier au tableau récapitulatif.
         # Grrr.
-        if self.not_before.year <= 2024:
-            for element in self.get_raa_elements_since_2024(self.__RAA_PAGE['2024']):
-                elements.append(element)
-        if self.not_before.year <= 2023:
-            for element in self.get_raa_elements_before_2024(self.__RAA_PAGE['2023']):
-                elements.append(element)
-        if self.not_before.year <= 2022:
-            for element in self.get_raa_elements_before_2024(self.__RAA_PAGE['2022']):
-                elements.append(element)
-        if self.not_before.year <= 2021:
-            for element in self.get_raa_elements_before_2024(self.__RAA_PAGE['2021']):
-                elements.append(element)
-        if self.not_before.year <= 2020:
-            for element in self.get_raa_elements_before_2024(self.__RAA_PAGE['2020']):
-                elements.append(element)
-        if self.not_before.year <= 2019:
-            for element in self.get_raa_elements_before_2024(self.__RAA_PAGE['2019']):
-                elements.append(element)
+        for year_page in year_pages:
+            url = year_page[0]
+            year = year_page[1]
+
+            if year >= 2024:
+                for element in self.get_raa_elements_since_2024(url):
+                    elements.append(element)
+            else:
+                for element in self.get_raa_elements_before_2024(url):
+                    elements.append(element)
 
         self.parse_raa(elements, keywords)
         self.mailer()