suivi et exclusion de pages fr des réunions

This commit is contained in:
Tykayn 2025-09-01 12:38:43 +02:00 committed by tykayn
parent 471eab4cd0
commit 466f9c773b
5 changed files with 231 additions and 4 deletions

View file

@ -63,14 +63,15 @@ NUM_WIKI_PAGES = 1
# 3. Pages with FR: prefix (e.g., "FR:Tag:leisure%3Dchildren_club")
SPECIFIC_PAGES = [
"Anatomie_des_étiquettes_osm",
"FR:Tag:leisure=children_club",
"FR:Tag:harassment_prevention=Dask_angela",
"Tag:leisure=children_club",
"Tag:harassment_prevention=ask_angela",
"Key:harassment_prevention",
"Proposal process",
"Automated_Edits_code_of_conduct",
"Key:cuisine",
"Libre_Charge_Map",
"OSM_Mon_Commerce"
"OSM_Mon_Commerce",
"Tag:amenity=charging_station"
]
def fetch_top_keys(limit=NUM_WIKI_PAGES):
@ -225,6 +226,10 @@ def fetch_wiki_page(key, language='en', is_specific_page=False):
Returns:
dict: Dictionary with page information or None if page doesn't exist
"""
# Skip pages with "FR:User:" or "FR:Réunions"
if "FR:User:" in key or "FR:Réunions" in key:
logger.info(f"Skipping excluded page: {key}")
return None
# Handle different URL formats
if is_specific_page:
# Case 1: Full URL