Files
scrapers/Working/govt/or.portland.py

74 lines
2.2 KiB
Python
Raw Permalink Normal View History

2026-01-09 22:49:19 -05:00
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
import events.digitools as digitools
venue, created = Organization.objects.get_or_create(
name="Portland City Council",
city="Portand",
website="https://www.portland.gov",
is_venue=True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'pdx')
event_type = "Gv"
# Time Signatures
tz = pytz.timezone("US/Eastern")
DATETIME_FORMAT = '%B %d, %Y %I:%M %p'
DATETIME_FORMAT_2 = '%B %d, %Y'
def get_events(ps, event_type):
print("Getting events ...")
contents = ps.xpath('.//*/div[@class="row position-relative"]')
ppr(contents)
for c in contents:
try:
event = {}
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event['title'] = c.xpath('.//*/h2[@class="h4"]/a/span/text()')[0]
event['link'] = venue.website + c.xpath('.//*/h2[@class="h4"]/a/@href')[0]
event['date'] = c.xpath('.//*/time/text()')[0]
event['time'] = c.xpath('.//*/span[@class="pe-4"]/text()')[0].replace("\n", "").strip()
# event['time2'] = c.xpath('.//*/span[@class="pe-4"]/text()')
try:
event['dateStamp'] = datetime.strptime(event['date']+" "+event['time'], DATETIME_FORMAT)
except:
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT_2)
# 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()
for n in range(3):
link = venue.website + "/auditor/council-clerk/events?page=" + str(n)
ps = digitools.getSource(br, link)
get_events(ps, "Gv")
sleep(2)
digitools.updateScraper(scraper, item_count_start)
br.close()