first commit
This commit is contained in:
67
event_scrapers/Templates/TemplateScraper.py.bak
Normal file
67
event_scrapers/Templates/TemplateScraper.py.bak
Normal file
@@ -0,0 +1,67 @@
|
||||
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="Venue Name",
|
||||
city="Minneapolis",
|
||||
website="Event Website",
|
||||
)
|
||||
|
||||
event_type = ""
|
||||
|
||||
# Time Signatures
|
||||
tz = pytz.timezone("US/Central")
|
||||
DATETIME_FORMAT = '%b %d %Y %I:%M %p'
|
||||
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'
|
||||
|
||||
def get_events(ps, event_type):
|
||||
contents = ps.xpath('.//*/li[@class="event"]')
|
||||
for c in contents:
|
||||
try:
|
||||
event = {}
|
||||
day = c.xpath('.//*/span[@class="day"]/text()')[0]
|
||||
month = c.xpath('.//*/span[@class="mth"]/text()')[0]
|
||||
year = datetime.now().year
|
||||
if month == "Jan":
|
||||
year = int(year) + 1
|
||||
event['title'] = c.xpath('.//*/span[@class="event_title"]/a/text()')[0]
|
||||
event['date'] = [month, day, str(year), c.xpath('.//*/span[@class="event_time"]/text()')[0].strip()]
|
||||
event['date'] = " ".join(event['date'])
|
||||
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
|
||||
event['link'] = c.xpath('.//*/span[@class="event_title"]/a/@href')[0]
|
||||
print("Event Dict Created")
|
||||
createBasicEvent(event, event_type, 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/")
|
||||
|
||||
for link in links:
|
||||
ps = getSource(br, link)
|
||||
get_events(ps, "Ed")
|
||||
sleep(3)
|
||||
|
||||
br.close()
|
||||
Reference in New Issue
Block a user