Files
scrapers/Working/workshop/cabooze.mn.py

74 lines
2.2 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'] = '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
venue, created = Organization.objects.get_or_create(
name="Cabooze",
city="Minneapolis",
website="https://www.cabooze.com/#/events",
is_venue=True
)
scraper,item_count_start = digitools.getScraper(venue)
event_type = ""
# Time Signatures
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%b %d %I:%M %p %Y'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'
def get_events(ps, event_type):
print("Getting events ...")
contents = ps.xpath('.//*/div[@class="vp-event-row vp-widget-reset vp-venue-thecabooze"]')
ppr(contents)
for c in contents:
try:
event = {}
date = c.xpath('.//*/span[@class="vp-month-n-day"]/text()')[0]
print(date)
month = date.split(" ")[0]
time = c.xpath('.//*/span[@class="vp-time"]/text()')[0]
year = datetime.now().year
if month == "Jan":
year = int(year) + 1
event['calendar'] = scraper.calendar
event['title'] = c.xpath('.//*/div[@class="vp-event-name"]/text()')[0]
event['date'] = [date, time, str(year)]
event['date'] = " ".join(event['date'])
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['link'] = "https://www.cabooze.com/" + c.xpath('.//a[@class="vp-event-link"]/@href')[0]
print("Event Dict Created")
ppr(event)
digitools.createBasicEvent(event, event_type, venue)
scraper.items+=1
except Exception as e:
print(e)
ppr(event)
print("\n\n+++\n\n")
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, "Mu")
sleep(3)
digitools.updateScraper(scraper, item_count_start)
br.close()