import os, sys from datetime import datetime from dateutil import relativedelta sys.path.append('/var/www/digisnaxx.ado/scrapers') import dtss dtss.getReady() from time import sleep from pprint import pprint as ppr import pytz from events.models import Organization, Scraper, Calendar import events.digitools as digitools venue, created = Organization.objects.get_or_create( name="Amsterdam Bar & Hall", city="St. Paul", website="https://www.amsterdambarandhall.com/events-new/", is_venue=True ) scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp') DATETIME_FORMAT = '%B %d %Y %I:%M%p' DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y' def get_events(ps): contents = ps.xpath('.//*/ul[@class="events-list"]/li') for c in contents: try: event = {} day = c.xpath('.//*/div[@class="date-day"]/text()')[0] month = c.xpath('.//*/div[@class="date-month"]/text()')[0] year = datetime.now().year event['scraper'] = scraper event['calendars'] = [scraper.calendar] event['title'] = c.xpath('.//div/h4/a/text()')[0] event['date'] = [month, day, str(year), c.xpath('.//div[@class="event-info"]/p/text()')[0].split(" ")[0]] event['date'] = " ".join(event['date']) event['dateStamp'] =datetime.strptime(event['date'], DATETIME_FORMAT) event['link'] = c.xpath('.//div[@class="event-info"]/h4/a/@href')[0] if " presents" in event['title']: event['title'] = event['title'].split("presents")[1][1:].strip() if event['title'].startswith('.'): print("BLAHH\n") event['title'] = event['title'][1:].strip() digitools.createBasicEvent(event, "Mu", venue) scraper.items+=1 except Exception as e: print(e) if len(sys.argv) >= 2: arg1 = sys.argv[1] br = digitools.getBrowser(arg1) else: print("No run_env") quit() ps = digitools.getSource(br, venue.website) get_events(ps) sleep(3) br.close() digitools.updateScraper(scraper, item_count_start)