moved scrapers into app_dir events
This commit is contained in:
80
events/scrapers/Working/venues/MplsVFW.py
Normal file
80
events/scrapers/Working/venues/MplsVFW.py
Normal file
@@ -0,0 +1,80 @@
|
||||
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, Event
|
||||
import events.digitools as digitools
|
||||
|
||||
|
||||
from selenium.webdriver.common.by import By
|
||||
from lxml import html
|
||||
|
||||
venue, created = Organization.objects.get_or_create(
|
||||
name="Uptown VFW",
|
||||
city="Minneapolis",
|
||||
website="https://noboolpresents.com/venues/uptown-vfw/",
|
||||
is_venue = True
|
||||
)
|
||||
|
||||
scraper,item_count_start = digitools.getScraper(venue)
|
||||
|
||||
tz = pytz.timezone("US/Central")
|
||||
|
||||
DATETIME_FORMAT = '%a %B %d @ %I:%M %p %Y'
|
||||
DATETIME_FORMAT_2 = '%b %d %I:%M%p %Y'
|
||||
DATETIME_FORMAT_3 = '%b %d %Y'
|
||||
# Set initial variables for City, etc
|
||||
calendar_url = 'https://noboolpresents.com/venues/uptown-vfw/'
|
||||
current_year = str(datetime.now().year)
|
||||
|
||||
if len(sys.argv) >= 2:
|
||||
arg1 = sys.argv[1]
|
||||
br = digitools.getBrowser(arg1)
|
||||
else:
|
||||
print("No run_env")
|
||||
quit()
|
||||
|
||||
br.get(calendar_url)
|
||||
sleep(30)
|
||||
|
||||
def getEvents(br):
|
||||
ps = html.fromstring(br.page_source)
|
||||
events = ps.xpath('.//*/article')
|
||||
for event in events:
|
||||
deets = {}
|
||||
dateTime = event.xpath('.//*/span[@class="tribe-event-date-start"]/text()')[0].replace("•", "").strip() + " " + current_year
|
||||
title = event.xpath('.//*/h2[@class="alt-font"]/a/text()')[0].replace("\n", "").replace("\t", "")
|
||||
link = event.xpath('.//*/h2[@class="alt-font"]/a/@href')[0]
|
||||
deets["tickets"] = event.xpath('.//*/span[@class="tribe-events-c-small-cta__price"]/strong/text()')[0]
|
||||
try:
|
||||
new_event = Event.objects.update_or_create(
|
||||
calendar = scraper.calendar,
|
||||
scraper = scraper,
|
||||
event_type = 'Mu',
|
||||
show_title = title,
|
||||
show_link = link,
|
||||
show_date = datetime.strptime(dateTime, DATETIME_FORMAT),
|
||||
show_day = datetime.strptime(dateTime, DATETIME_FORMAT).date(),
|
||||
more_details = deets["tickets"],
|
||||
venue = venue
|
||||
)
|
||||
scraper.items+=1
|
||||
except Exception as e:
|
||||
print("oops", e)
|
||||
|
||||
getEvents(br)
|
||||
br.find_element(By.XPATH, './/*/li[@class="tribe-events-c-nav__list-item tribe-events-c-nav__list-item--next"]/a').click()
|
||||
sleep(5)
|
||||
getEvents(br)
|
||||
br.close()
|
||||
|
||||
digitools.updateScraper(scraper, item_count_start)
|
||||
Reference in New Issue
Block a user