import os, sys from datetime import datetime from dateutil import relativedelta import django sys.path.append('../../../') os.environ['DJANGO_SETTINGS_MODULE'] = 'ds_events.settings' 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 current_year = str(datetime.now().year) venue, created = Organization.objects.get_or_create( name="Piller Forum", city="Minneapolis", website="https://www.pilllar.com/pages/events", is_venue = True ) scraper,item_count_start = digitools.getScraper(venue) event_type = "Mu" # Time Signatures tz = pytz.timezone("US/Central") DATETIME_FORMAT = '%b. %d %Y %I:%M %p' DATETIME_FORMAT_night = '%b. %d %Y %I:%M %p' DATETIME_FORMAT_2 = '%b. %d %Y %I:%Mam' def get_events(ps, event_type): contents = ps.xpath('.//*/div[@class="sse-column sse-half sse-center"]') for c in contents: try: event = {} event['scraper'] = scraper event['calendar'] = scraper.calendar event['link'] = venue.website # time = c.xpath('.//*/span[@class="vp-time"]/text()')[0].strip() date = c.xpath('.//h1[@class="sse-size-64"]/text()')[0] if len(date) > 1: print(date) year = datetime.now().year event_date = date + " " + str(year) event['title'] = c.xpath('.//p/span/b/text()')[0] details = c.xpath('.//p/text()') if 'Music' in details[-1]: event_time = c.xpath('.//p/text()')[-1].split("Music")[1].strip() event_type = "Mu" event_dt = event_date + " " + event_time + " PM" event['dateStamp'] = datetime.strptime(event_dt, DATETIME_FORMAT_night) elif len(details) == 1: try: event_time = details[0].split("-")[0].strip() event_dt = event_date + " " + event_time + ' PM' event['dateStamp'] = datetime.strptime(event_dt, DATETIME_FORMAT_night) event_type = "Ot" except Exception as e: event_time = details[0].split("-")[0].strip() event_dt = event_date + " " + event_time event['dateStamp'] = datetime.strptime(event_dt, DATETIME_FORMAT_2) event_type = "Ot" digitools.createBasicEvent(event, event_type, venue) 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, event_type) sleep(3) br.close() digitools.updateScraper(scraper, item_count_start)