diff --git a/src/agenda_culturel/import_tasks/custom_extractors/lacomedie.py b/src/agenda_culturel/import_tasks/custom_extractors/lacomedie.py index 1c1241b..c33d796 100644 --- a/src/agenda_culturel/import_tasks/custom_extractors/lacomedie.py +++ b/src/agenda_culturel/import_tasks/custom_extractors/lacomedie.py @@ -9,6 +9,12 @@ class CExtractor(TwoStepsExtractor): nom_lieu = "La Comédie de Clermont" url_referer = "https://lacomediedeclermont.com/saison24-25/" + def is_to_import_from_url(self, url): + if "on-debriefe" in url: + return False + else: + return True + def category_comedie2agenda(self, category): mapping = { "Théâtre": "Spectacles", @@ -50,33 +56,35 @@ class CExtractor(TwoStepsExtractor): e_url = ( e.select("a")[0]["href"] + "#" + d ) # a "fake" url specific for each day of this show - self.add_event_url(e_url) - self.add_event_start_day(e_url, d) - t = ( - str(e.select("div#datecal")[0]) - .split(" ")[-1] - .split("<")[0] - ) - self.add_event_start_time(e_url, t) - title = e.select("a")[0].contents[0] - self.add_event_title(e_url, title) - category = e.select("div#lieuevtcal span") - if len(category) > 0: - category, tag = self.category_comedie2agenda( - category[-1].contents[0] + + if self.is_to_import_from_url(e_url): + self.add_event_url(e_url) + self.add_event_start_day(e_url, d) + t = ( + str(e.select("div#datecal")[0]) + .split(" ")[-1] + .split("<")[0] ) - if category: - self.add_event_category(e_url, category) - if tag: - self.add_event_tag(e_url, tag) - location = ( - e.select("div#lieuevtcal")[0] - .contents[-1] - .split("•")[-1] - ) - if location.replace(" ", "") == "": - location = self.nom_lieu - self.add_event_location(e_url, location) + self.add_event_start_time(e_url, t) + title = e.select("a")[0].contents[0] + self.add_event_title(e_url, title) + category = e.select("div#lieuevtcal span") + if len(category) > 0: + category, tag = self.category_comedie2agenda( + category[-1].contents[0] + ) + if category: + self.add_event_category(e_url, category) + if tag: + self.add_event_tag(e_url, tag) + location = ( + e.select("div#lieuevtcal")[0] + .contents[-1] + .split("•")[-1] + ) + if location.replace(" ", "") == "": + location = self.nom_lieu + self.add_event_location(e_url, location) def add_event_from_content( self,