prefdpt: possibilité d'exclure des titres de cartes

Attrap_prefdpt.py

@@ -21,7 +21,8 @@ class Attrap_prefdpt(Attrap):
         },
         'link_to_raa': False,
         'autodetect_links_to_raa': True,
-        'follow_link_on_unrecognised_date': True
+        'follow_link_on_unrecognised_date': True,
+        'exclude': []
     }
     white_card = {
         'regex': {
@@ -35,7 +36,8 @@ class Attrap_prefdpt(Attrap):
         },
         'link_to_raa': False,
         'autodetect_links_to_raa': True,
-        'follow_link_on_unrecognised_date': True
+        'follow_link_on_unrecognised_date': True,
+        'exclude': []
     }
 
     # Liste des widgets à analyser (sera remplie au moment de l'initialisation, mais peut être surchargée par la classe de préfecture de département)
@@ -54,13 +56,14 @@ class Attrap_prefdpt(Attrap):
     class DptWidget:
         """Une classe représentant un widget sur le site d'une préfecture de département."""
 
-        def __init__(self, name, regex=None, css_path=None, link_to_raa=False, autodetect_links_to_raa=True, follow_link_on_unrecognised_date=True):
+        def __init__(self, name, regex=None, css_path=None, link_to_raa=False, autodetect_links_to_raa=True, follow_link_on_unrecognised_date=True, exclude=[]):
             self.name = name
             self.regex = regex
             self.css_path = css_path
             self.link_to_raa = link_to_raa
             self.autodetect_links_to_raa = autodetect_links_to_raa
             self.follow_link_on_unrecognised_date = follow_link_on_unrecognised_date
+            self.exclude = exclude
 
         def has_css_path(self, key):
             return self.css_path and self.css_path.get(key, None) is not None
@@ -100,7 +103,8 @@ class Attrap_prefdpt(Attrap):
                     css_path=self.grey_card['css_path'],
                     link_to_raa=self.grey_card['link_to_raa'],
                     autodetect_links_to_raa=self.grey_card['autodetect_links_to_raa'],
-                    follow_link_on_unrecognised_date=self.grey_card['follow_link_on_unrecognised_date']
+                    follow_link_on_unrecognised_date=self.grey_card['follow_link_on_unrecognised_date'],
+                    exclude=self.grey_card['exclude']
                 )
             )
 
@@ -112,7 +116,8 @@ class Attrap_prefdpt(Attrap):
                     css_path=self.white_card['css_path'],
                     link_to_raa=self.white_card['link_to_raa'],
                     autodetect_links_to_raa=self.white_card['autodetect_links_to_raa'],
-                    follow_link_on_unrecognised_date=self.white_card['follow_link_on_unrecognised_date']
+                    follow_link_on_unrecognised_date=self.white_card['follow_link_on_unrecognised_date'],
+                    exclude=self.white_card['exclude']
                 )
             )
 
@@ -149,7 +154,7 @@ class Attrap_prefdpt(Attrap):
                     False
                 )
             for card in cards:
-                if card['url'] not in self.page_urls_to_parse:
+                if card['url'] not in self.page_urls_to_parse and card['name'].strip() not in widget.exclude:
                     date = None
                     date_is_correct = False