more scrapers

This commit is contained in:
2026-01-24 19:01:47 -05:00
parent 7013d8327a
commit 3c4a41ae2c
58 changed files with 1614 additions and 2988 deletions

View File

@@ -0,0 +1,73 @@
import os, sys
from datetime import datetime
from dateutil import relativedelta
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
import pytz
from events.models import Organization, Scraper, Calendar, Event
import events.digitools as digitools
tz = pytz.timezone("US/Central")
tz_str = "-0600 UTC"
DATETIME_FORMAT = '%a %b %d %Y %I:%M %p %z %Z'
# DATETIME_FORMAT_2 = '%a, %b %d, %Y %I %p %z %Z'
venue, created = Organization.objects.get_or_create(
name="Subterranean",
city="Chicago",
website="https://subt.net/",
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'chi')
scraper.items = 0
scraper.save()
def get_events(ps, event_type):
contents = ps.xpath('.//*/div[@class="event-info-block"]')
for c in contents:
try:
event = {}
date = c.xpath('.//p[@class="fs-18 bold mt-1r event-date"]/text()')[0]
time = c.xpath('.//*/span[@class="door-time"]/text()')[0].replace("Doors: ", "").replace("/", "").strip()
year = datetime.now().year
event['date'] = (' ').join([date, str(year), time, tz_str])
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event['title'] = c.xpath('.//p[@class="fs-18 bold mb-12 event-title"]/a/text()')[0]
try:
event['support'] = c.xpath('.//p[@class="fs-12 supporting-talent"]/text()')[0]
except:
pass
event['venue'] = c.xpath('.//p[@class="fs-12 venue"]/text()')[0]
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['link'] = venue.website
if len(event['support']) > 0:
event['new_title'] = event['title'] + " w/ " + event['support']
digitools.createBasicEvent(event, "Mu", venue)
scraper.items+=1
except Exception as e:
print("\nError: ", e)
# ppr(event)
pass
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
else:
print("No run_env")
br.close()
quit()
ps = digitools.getSource(br, venue.website)
get_events(ps, "Mu")
sleep(3)
digitools.updateScraper(scraper, item_count_start)
br.close()