Fix problème import (tiret manquant parfois)

This commit is contained in:
Jean-Marie Favreau 2025-03-28 16:04:49 +01:00
parent 9f7204a35f
commit 83bcc43e35

View File

@ -1,4 +1,5 @@
from bs4 import BeautifulSoup from bs4 import BeautifulSoup
import re
from ..extractor import Extractor from ..extractor import Extractor
from ..twosteps_extractor import TwoStepsExtractor from ..twosteps_extractor import TwoStepsExtractor
@ -51,10 +52,17 @@ class CExtractor(TwoStepsExtractor):
else: else:
image = None image = None
desc = soup.select("head meta[name=description]")[0]["content"] desc = soup.select("head meta[name=description]")[0]["content"]
start_day = Extractor.parse_french_date(desc.split("-")[0]) res = re.match("(.*202[5-9])[ -]*([0-9h:]+)[ -]*([0-9h:]*)", desc)
start_time = Extractor.parse_french_time(desc.split("-")[1]) if res:
end_time = Extractor.parse_french_time(desc.split("-")[2]) start_day = Extractor.parse_french_date(res[1])
end_day = Extractor.guess_end_day(start_day, start_time, end_time) start_time = Extractor.parse_french_time(res[2])
end_time = Extractor.parse_french_time(res[3])
end_day = Extractor.guess_end_day(start_day, start_time, end_time)
else:
start_day = None
start_time = None
end_time = None
end_day = None
location = self.nom_lieu location = self.nom_lieu
descriptions = soup.select("div.vce-col-content") descriptions = soup.select("div.vce-col-content")