Files
scrapers/Working/cals/MplStpMag.mn.py

102 lines
3.8 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
from events.digitools import getBrowser, createURL, createBasicEvent, getSource
venue, created = Organization.objects.get_or_create(
name="Mpls Stp Mag",
city="Minneapolis",
website="https://calendar.mspmag.com/calendars/all-events/",
)
event_type = ""
# Time Signatures
tz = pytz.timezone("US/Central")
td = relativedelta.relativedelta(days=1)
fortnight = relativedelta.relativedelta(days=14)
odt = datetime.now() + fortnight
# DATETIME_FORMAT = '%b %d %Y %I:%M %p'
DATETIME_FORMAT = '%A, %B %d %Y %I:%M%p'
DATETIME_FORMAT_ALT = '%A, %B %d %Y'
def get_events(ps, event_type):
contents = ps.xpath('.//*/div[@class="css-card js-card day-card type-smad expandable"]')
for c in contents:
try:
event = {}
event['calendar'] = venue.calendar
event_block = c.xpath('.//*/li[@class="card-listings-item event-element"]')
date = c.xpath('.//div[@class="day-card__header day-card__header--daily"]/text()')[0].replace("\n", "").strip()
if date == "Today":
date = datetime.today()
elif date == "Tomorrow":
date = datetime.today() + td
# month = c.xpath('.//*/span[@class="mth"]/text()')[0]
year = datetime.now().year
# if month == "Jan":
# year = int(year) + 1
dateTime = datetime.strptime(date + " " + str(year), DATETIME_FORMAT_ALT)
if dateTime > odt:
print("DATE TIME ", dateTime)
break
for ev in event_block:
time = ev.xpath('.//*/span[@class="card-listing-item-time"]/text()')[0].replace("@", "").strip()
if time == "All day":
time = "12:00pm"
event['title'] = ev.xpath('.//*/div[@class="card-listing-item-title"]/text()')[0] + " (Check link for times.)"
elif "-" in time:
time = time.split("-")[0]
event['title'] = ev.xpath('.//*/div[@class="card-listing-item-title"]/text()')[0]
else:
event['title'] = ev.xpath('.//*/div[@class="card-listing-item-title"]/text()')[0]
event['location'] = ev.xpath('.//*/span[@class="card-listing-item-location"]/text()')[0]
if event['location'] == '7th St. Entry':
event['location'] = '7th St Entry'
elif event['location'] == '7th Street Entry':
event['location'] = '7th St Entry'
elif event['location'] == 'Amsterdam Bar and Hall':
event['location'] = 'Amsterdam Bar & Hall'
new_venue, created = Organization.objects.get_or_create(name=event['location'])
print("V: ", new_venue, created)
event['dateTime'] = date + " " + str(year) + " " + time
event['link'] = venue.website + c.xpath('.//@data-event')[0]
event['dateStamp'] = datetime.strptime(event['dateTime'], DATETIME_FORMAT)
createBasicEvent(event, event_type, new_venue)
except Exception as e:
print(e)
ppr(event)
print("\n\n+++\n\n")
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = getBrowser(arg1)
else:
print("No run_env")
quit()
# Get Event Page Link(s)
# links = createURL("https://acmecomedycompany.com/the-club/calendar/")
ps = getSource(br, venue.website)
get_events(ps, "Ed")
sleep(3)
br.close()