tykayn/oedb-backend
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

add extractor viparis

Browse source
This commit is contained in:
Tykayn 2025-10-10 15:11:05 +02:00 • committed by tykayn
parent fd2d51b662
commit ee48a3c665
9 changed files with 2221 additions and 4443 deletions
Download patch file Download diff file Expand all files Collapse all files

15
extractors/ccpl_agenda.py
View file

@ -106,7 +106,7 @@ class CCPLAgendaScraper:
cached_hash = self.cache_data.get("content_hash")
return cached_hash != new_hash
def fetch_agenda_data(self) -> Optional[str]:
def fetch_agenda_data(self, force_refresh: bool = False) -> Optional[str]:
"""Récupère les données de l'agenda CCPL"""
try:
logger.info(f"🌐 Récupération de l'agenda CCPL: {self.agenda_url}")
@ -121,9 +121,12 @@ class CCPLAgendaScraper:
content = response.text
content_hash = self.get_content_hash(content)
# Vérifier si le contenu a changé
if self.is_content_changed(content_hash):
logger.info("🔄 Nouveau contenu détecté, mise à jour du cache")
# Vérifier si le contenu a changé ou si on force le rafraîchissement
if self.is_content_changed(content_hash) or force_refresh:
if force_refresh:
logger.info("🔄 Rafraîchissement forcé, mise à jour du cache")
else:
logger.info("🔄 Nouveau contenu détecté, mise à jour du cache")
self.cache_data["content_hash"] = content_hash
self.cache_data["last_fetch"] = datetime.now().isoformat()
self.save_cache_data()
@ -282,7 +285,7 @@ class CCPLAgendaScraper:
"label": title,
"description": description,
"type": "scheduled",
"what": "culture.community",
"what": "culture.community.ccpl",
"where": location,
"start": self.parse_date(date_text),
"stop": self.parse_date(date_text, end=True),
@ -894,7 +897,7 @@ class CCPLAgendaScraper:
try:
# Récupérer les données de l'agenda
html_content = self.fetch_agenda_data()
html_content = self.fetch_agenda_data(force_refresh)
if html_content is None and not force_refresh:
logger.info("ℹ️ Utilisation du cache (pas de nouveau contenu)")