Un événement fb peut correspondre à plusieurs événements

experimentations/get_facebook_events.py

@@ -29,8 +29,8 @@ from src.agenda_culturel.import_tasks.custom_extractors import *
 if __name__ == "__main__":
 
     u2e = URL2Events(ChromiumHeadlessDownloader(), fbevents.CExtractor())
-    url = "https://www.facebook.com/laJeteeClermont/events"
-    url_human = "https://www.facebook.com/laJeteeClermont/events"
+    url = "https://www.facebook.com/laJeteeClermont/upcoming_hosted_events"
+    url_human = "https://www.facebook.com/laJeteeClermont/upcoming_hosted_events"
 
     try:
         events = u2e.process(url, url_human, cache = "cache-lajetee-fb.html", default_values = {"location": "La Jetée"}, published = True)

src/agenda_culturel/celery.py

@@ -239,8 +239,6 @@ def import_events_from_url(self, url, cat):
                 url, published=False, default_values=values
             )
 
-            # TODO: use celery to import the associated events
-
             # convert it to json
             json_events = json.dumps(events, default=str)
 

src/agenda_culturel/import_tasks/custom_extractors/fbevents.py

@@ -38,11 +38,10 @@ class CExtractor(TwoStepsExtractor):
             )
 
         if fevent is not None:
-            event = fevent.build_event(event_url)
-            event["published"] = published
-            print([e.elements for e in fevent.neighbor_events])
+            for event in fevent.build_events(event_url):
+                event["published"] = published
 
-            if "category" in default_values:
-                event["category"] = default_values["category"]
-            self.add_event(**event)
+                if "category" in default_values:
+                    event["category"] = default_values["category"]
+                self.add_event(**event)
 

src/agenda_culturel/import_tasks/extractor_facebook.py

@@ -32,6 +32,7 @@ class FacebookEvent:
             "price_info",
             "cover_media_renderer",
             "id",
+            "parent_if_exists_or_self",
             "day_time_sentence",
             "event_place",
             "comet_neighboring_siblings",
@@ -52,12 +53,20 @@ class FacebookEvent:
         "event_place": {"event_place_name": ["name"]},
     }
 
-    def __init__(self, i, event):
-        self.fragments = {}
-        self.elements = {}
-        self.neighbor_events = None
-        self.possible_end_timestamp = []
-        self.add_fragment(i, event)
+    def __init__(self, *args):
+        if len(args) == 1:
+            other = args[0]
+            self.fragments = other.fragments
+            self.elements = other.elements
+            self.neighbor_events = None
+        else:
+            i = args[0]
+            event = args[1]
+            self.fragments = {}
+            self.elements = {}
+            self.neighbor_events = None
+            self.possible_end_timestamp = []
+            self.add_fragment(i, event)
 
     def get_element(self, key):
         return self.elements[key] if key in self.elements else None
@@ -177,8 +186,6 @@ class FacebookEvent:
         return event
 
     def build_event(self, url):
-        self.get_element("image")
-
         return {
             "title": self.get_element("name"),
             "category": None,
@@ -195,6 +202,26 @@ class FacebookEvent:
             "image_alt": self.get_element("image"),
         }
 
+    def get_parent_id(self):
+        return self.get_element("parent_if_exists_or_self")["id"]
+
+
+    def build_events(self, url):
+
+        if self.neighbor_events is None or len(self.neighbor_events) == 0:
+            return [self.build_event(url)]
+        else:
+            url_base = "https://www.facebook.com/events/" + self.get_parent_id() + "/"
+            result = []
+            for nb_e in self.neighbor_events:
+                # we create a copy of the event
+                clone = FacebookEvent(self)
+                # we set start and end timestamp accordnig to the neighbor
+                clone.elements["start_timestamp"] = nb_e.elements["start_timestamp"]
+                clone.elements["end_timestamp"] = nb_e.elements["end_timestamp"]
+                ## we generate the event
+                result.append(clone.build_event(url_base + nb_e.elements["id"] + "/"))
+            return result
 
 class FacebookEventExtractor(Extractor):
 
@@ -230,13 +257,13 @@ class FacebookEventExtractor(Extractor):
 
         if fevent is not None:
             self.set_header(url)
-            event = fevent.build_event(url)
-            logger.warning("published: " + str(published))
-            event["published"] = published
+            for event in fevent.build_events(url):
+                logger.warning("published: " + str(published))
+                event["published"] = published
 
-            if "category" in default_values:
-                event["category"] = default_values["category"]
-            self.add_event(**event)
+                if "category" in default_values:
+                    event["category"] = default_values["category"]
+                self.add_event(**event)
             return self.get_structure()