agenda_culturel/src/agenda_culturel/import_tasks/custom_extractors/lefotomat.py

73 lines
2.5 KiB
Python

from ..generic_extractors import *
import re
import json5
from datetime import timedelta
# A class dedicated to get events from Le Fotomat'
# URL: https://www.lefotomat.com/
class CExtractor(TwoStepsExtractor):
nom_lieu = "Le Fotomat'"
def category_fotomat2agenda(self, category):
if not category:
return None
mapping = { "Concerts": "Concert"}
if category in mapping:
return mapping[category]
else:
return None
def build_event_url_list(self, content):
soup = BeautifulSoup(content, "xml")
events = soup.select("item")
for e in events:
e_url = e.find("link").contents[0]
self.add_event_url(e_url)
title = e.find("title").contents[0]
self.add_event_title(e_url, title)
category = self.category_fotomat2agenda(e.find("category").contents[0])
if category:
self.add_event_category(e_url, category)
def add_event_from_content(self, event_content, event_url, url_human = None, default_values = None, published = False):
soup = BeautifulSoup(event_content, "html.parser")
image = soup.select("div.post-content img.wp-post-image")
if image:
image = image[0]["src"]
else:
image = None
desc = soup.select("head meta[name=description]")[0]["content"]
start_day = self.parse_french_date(desc.split("-")[0])
start_time = self.parse_french_time(desc.split("-")[1])
end_time = self.parse_french_time(desc.split("-")[2])
end_day = self.guess_end_day(start_day, start_time, end_time)
location = self.nom_lieu
descriptions = soup.select("div.vce-col-content")
if descriptions:
descriptions = [d.get_text() for d in descriptions]
description = max(descriptions, key=len)
else:
description = None
article = soup.select("article.post")
tags = []
for c in article[0]["class"]:
if c.startswith("category-"):
tag = '-'.join(c.split("-")[1:])
if tag != "concerts":
tags.append(tag)
url_human = event_url
self.add_event_with_props(event_url, None, None, start_day, location, description, tags, recurrences=None, uuids=[event_url], url_human=url_human, start_time=start_time, end_day=end_day, end_time=end_time, published=published, image=image)