pref83: réduit le nombre de requêtes

Attrap_pref83.py

@@ -11,14 +11,7 @@ class Attrap_pref83(Attrap):
 
     # Config
     hostname = 'https://www.var.gouv.fr'
-    raa_page = {
-        '2024': f'{hostname}/Publications/RAA-Recueil-des-actes-administratifs/Recueil-des-actes-administratifs-2024',
-        '2023': f'{hostname}/Publications/RAA-Recueil-des-actes-administratifs/Recueil-des-actes-administratifs-2023',
-        '2022': f'{hostname}/Publications/RAA-Recueil-des-actes-administratifs/Recueil-des-actes-administratifs-2022',
-        '2021': f'{hostname}/Publications/RAA-Recueil-des-actes-administratifs/Recueil-des-actes-administratifs-2021',
-        '2020': f'{hostname}/Publications/RAA-Recueil-des-actes-administratifs/Recueil-des-actes-administratifs-2020',
-        '2019': f'{hostname}/Publications/RAA-Recueil-des-actes-administratifs/Recueil-des-actes-administratifs-2019'
-    }
+    raa_page = f'{hostname}/Publications/Recueil-des-actes-administratifs'
     user_agent = 'Mozilla/5.0 (Windows NT 10.0; rv:109.0) Gecko/20100101 Firefox/115.0'
     full_name = 'Préfecture du Var'
     short_code = 'pref83'
@@ -29,38 +22,39 @@ class Attrap_pref83(Attrap):
         self.set_sleep_time(30)
 
     def get_raa(self, keywords):
+        year_pages_to_parse = []
+
+        # On détermine quelles pages d'année parser
+        page_content = self.get_page(self.raa_page, 'get').content
+        for year_page in self.get_sub_pages(
+            page_content,
+            'div.fr-card__body div.fr-card__content h2.fr-card__title a',
+            self.hostname,
+            False
+        ):
+            year = Attrap.guess_date(year_page['name'].strip(), 'Recueil des actes administratifs ([0-9]{4})').year
+            if year < 9999 and year >= self.not_before.year:
+                year_pages_to_parse.append(year_page['url'])
+
         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'])
-
-        sub_pages_to_parse = []
 
         # Pour chaque année, on cherche les sous-pages de mois
-        for raa_page in pages_to_parse:
-            sub_pages_to_parse.append(raa_page)
+        for raa_page in year_pages_to_parse:
+            pages_to_parse.append(raa_page)
             page_content = self.get_page(raa_page, 'get').content
-            month_pages = self.get_sub_pages(
+            for month_page in self.get_sub_pages(
                 page_content,
                 '.fr-card.fr-card--sm.fr-card--grey.fr-enlarge-link div.fr-card__body div.fr-card__content h2.fr-card__title a',
                 self.hostname,
                 False
-            )[::-1]
-            for month_page in month_pages:
-                sub_pages_to_parse.append(month_page['url'])
+            )[::-1]:
+                card_date = Attrap.guess_date(month_page['name'].strip(), '(.*)').replace(day=1)
+                if card_date >= self.not_before.replace(day=1):
+                    pages_to_parse.append(month_page['url'])
 
         # On parse les pages contenant des RAA
         elements = self.get_raa_with_pager(
-            sub_pages_to_parse[::-1],
+            pages_to_parse[::-1],
             '.fr-pagination__link.fr-pagination__link--next',
             self.hostname
         )