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 from events.models import Event as DSEvent, Organization from digitools import getBrowser, createDashURL, createBasicEvent, getSource venue, created = Organization.objects.get_or_create( name="Palmer's Bar", city="Minneapolis", website="https://palmers-bar.com", ) calendar_url = "https://palmers-bar.com" DATETIME_FORMAT = '%d %m %Y %I:%M%p' def get_events(ps, link): contents = ps.xpath('.//*/td') for c in contents: try: # day_num = c.xpath('.//*/div[@class="marker-daynum"]/text()')[0] events = c.xpath('.//*/li') # print(events) for e in events: event = {} event_link = calendar_url + e.xpath('.//a/@href')[0] ps = getSource(br, event_link) new_event = ps.xpath('.//*/h1[@class="eventitem-column-meta"]') event['title'] = new_event.xpath('.//*/h1[@class="event-title"]') event['date'] = new_event.xpath('.//*/time[@class="event-date"]') event['time'] = new_event.xpath('.//*/time[@class="event-time-12hr-start"]') event['link'] = event_link # event['month'] = link.split("month=")[1].split("-")[0] # event['year'] = link.split("month=")[1].split("-")[1] # event['title'] = e.xpath('.//h1/a[@class="flyoutitem-link"]/text()') # event['time'] = e.xpath('.//div[@class="flyoutitem-datetime flyoutitem-datetime--12hr"]/text()') # event['date'] = str(day_num) + ' ' + 'time' # event['dateStamp'] = datetime.strptime(dateStamp, DATETIME_FORMAT) ppr(event) if len(event['title']): nevent = {} nevent['title'] = "".join(event['title']).strip() event['time'] = event['time'][0].strip().split(" –")[0] nevent['link'] = "https://palmers-bar.com" + e.xpath('.//a/@href')[0] event['dateStamp'] = str(day_num) + ' ' + event['month'] + ' ' + event['year'] + ' ' + event['time'] nevent['dateStamp'] = datetime.strptime(event['dateStamp'], DATETIME_FORMAT) ppr(nevent) # createBasicEvent(nevent, 'Mu', venue) except Exception as e: continue if len(sys.argv) >= 2: arg1 = sys.argv[1] br = getBrowser(arg1) else: print("No run_env") quit() links = createDashURL("https://palmers-bar.com/?view=calendar&month=") for link in links: ps = getSource(br, link) get_events(ps, link) sleep(5) # ppr(events) br.close()