Files
scrapers/Working/venues/cedar.mpls.py

74 lines
2.3 KiB
Python
Raw Normal View History

2025-11-30 16:29:50 -05:00
import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
from time import sleep
from pprint import pprint as ppr
import pytz
from events.models import Organization, Scraper
import events.digitools as digitools
venue, created = Organization.objects.get_or_create(
name="Cedar Cultural Center",
city="Minneapolis",
website="https://www.thecedar.org",
is_venue=True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%A, %B %d, %Y %I:%M %p'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'
DATETIME_FORMAT_3 = '%A, %B %d at %I:%M%p %Y'
DATETIME_FORMAT_4 = '%A, %B %d at %I%p %Y'
DATETIME_FORMAT_5 = '%A, %B %d @%I%p %Y'
def get_events(ps):
links = ps.xpath('.//*/div[@class="summary-title"]/a/@href')
for l in links:
if "cedar-news-blog" in l:
continue
pse = digitools.getSource(br, "https://www.thecedar.org" + l)
event = {}
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event['link'] = "https://www.thecedar.org" + l
try:
time = pse.xpath('.//*/time[@class="event-time-localized-start"]/text()')[0]
date = pse.xpath('.//*/time[@class="event-date"]/text()')[0]
event['title'] = pse.xpath('.//*/h1[@class="eventitem-title"]/text()')[0]
except:
try:
time = pse.xpath('.//*/time[@class="event-time-localized"]/text()')[0]
date = pse.xpath('.//*/time[@class="event-date"]/text()')[0]
event['title'] = pse.xpath('.//*/h1[@class="eventitem-title"]/text()')[0]
except Exception as e:
print(e)
print("failed event: ", event)
dateStamp = date + " " + time
event['dateStamp'] = datetime.strptime(dateStamp, DATETIME_FORMAT)
digitools.createBasicEvent(event, "Mu", venue)
scraper.items+=1
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
else:
print("No run_env")
quit()
calendar_url = 'https://www.thecedar.org/listing'
ps = digitools.getSource(br, calendar_url)
get_events(ps)
br.close()
digitools.updateScraper(scraper, item_count_start)