lots of updates :/

This commit is contained in:
2025-11-30 15:34:08 -05:00
parent 61e198ff55
commit e06a5d49f8
84 changed files with 6614 additions and 2297 deletions

View File

@@ -1,12 +1,14 @@
import re, os, sys
from datetime import datetime
from datetime import datetime, timedelta
import requests
import json
import django
sys.path.append('../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'ds_events.settings'
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
from events.models import Event, Organization, Scraper
from events.models import Event, Organization, Scraper, Calendar
from time import sleep
from pprint import pprint as ppr
@@ -17,13 +19,17 @@ from selenium.webdriver.common.by import By
from xvfbwrapper import Xvfb
from lxml import html
import pytz
from events.digitools import getBrowser, createURL, createBasicEvent, getSource
from events.digitools import getBrowser, createURL, createBasicEvent, getSource, add_calendar
import events.digitools as digitools
try:
scraper, created = Scraper.objects.get_or_create(
name="Mpls City Council",
website="https://lims.minneapolismn.gov/Calendar/citycouncil/upcoming",
calendar = Calendar.objects.get(shortcode='msp'),
items = 0,
new_items = 0,
last_ran = datetime.now(),
)
except Exception as e:
@@ -31,11 +37,24 @@ except Exception as e:
scraper = Scraper.objects.get(name="Mpls City Council")
print("Scraper: ", scraper)
DATETIME_FORMAT = '%Y-%m-%dT%H:%M:%S'
tz = pytz.timezone("US/Central")
td = timedelta(days=2)
odt = datetime.now() - td
DATETIME_FORMAT = '%A, %b %d, %Y %I:%M %p'
month = odt.strftime("%b")
day = int(datetime.now().day)
calendar_url = 'https://lims.minneapolismn.gov/Calendar/citycouncil/upcoming'
if day == 1:
day = 30
year = int(datetime.now().year)
cal_url = 'https://lims.minneapolismn.gov/Calendar/GetCalenderList?fromDate={}%20{},%20{}&toDate=null&meetingType=0&committeeId=null&pageCount=100&offsetStart=0&abbreviation=undefined&keywords=&sortOrder=1'.format(month, day, year)
# print("URL: ", cal_url)
# cal_url = 'https://lims.minneapolismn.gov/Calendar/GetCalenderList?fromDate=Nov%2015,%202025&toDate=null&meetingType=0&committeeId=null&pageCount=50&offsetStart=0&abbreviation=undefined&keywords=&sortOrder=1'
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
@@ -44,56 +63,60 @@ else:
print("No run_env")
quit()
br.get(calendar_url)
sleep(25)
# br.find_element(By.XPATH, '//*/li[@class="tab-header-small"]/a').click()
# sleep(15)
# all_entries = Select(br.find_element(By.XPATH, '//*/select'))
# all_entries.select_by_value('50')
# sleep(15)
br.get(cal_url)
sleep(2)
ps = html.fromstring(br.page_source)
events = ps.xpath('.//body/pre/text()')[0]
new_events = json.loads(events)
# ppr(dict(new_events[0]))
event_list = []
dayBlocks = ps.xpath('.//*/div[@class="ng-scope"]')
meetings = []
for dB in dayBlocks:
date = dB.xpath('.//div[@class="row"]/div/span[@class="ng-binding"]/text()')[0]
events = dB.xpath('.//div[@class="upcoming ng-scope"]/div')
for event in events:
time = event.xpath('.//div/text()')[0]
title = event.xpath('.//div/a/text()')[0].strip()
if not len(title) > 0:
title = event.xpath('.//div/span/a/text()')[0].strip()
link = event.xpath('.//div/a/@href')[0]
if link.startswith("/Download/"):
link = calendar_url
else:
link = "https://lims.minneapolismn.gov" + link
location = title.split(',')[-1].strip()
mtg_title = title.split(',')[:-1]
if len(mtg_title) > 1:
mtg_title = (' -').join(mtg_title).strip()
else:
mtg_title = mtg_title[0].strip()
dateTime = datetime.strptime(date + " " + time, DATETIME_FORMAT)
if location == "City Hall":
location = "Mpls City Hall"
print(dateTime, location, mtg_title, link)
print('\n\n++++\n\n')
venue, created = Organization.objects.get_or_create(name=location, city="Minneapolis")
new_event = Event.objects.update_or_create(
calendar = 'msp'
event_type = 'Gv',
show_title = mtg_title,
show_link = link,
show_date = dateTime,
show_day = dateTime,
venue = venue
)
scraper.items+=1
for event in new_events:
e = {}
e['title'] = event['CommitteeName']
e['link'] = scraper.website
e['dateStamp'] = datetime.strptime(event['MeetingTime'], DATETIME_FORMAT)
e['agendaStatus'] = event['AgendaStatus']
e['address'] = event['Address']
e['description'] = event['Description']
e['scraper'] = scraper
e['calendars'] = [scraper.calendar]
try:
# print(event['Location'].split(",")[1])
loc = event['Location'].split(",")[1]
if "City Hall" in loc:
e['location'] = "Mpls City Hall"
venue = Organization.objects.get(
name="Mpls City Hall",
)
digitools.createBasicEvent(e, 'Gv', venue)
elif "Public Service Building" in loc:
e['location'] = "Mpls Public Service Building"
venue = Organization.objects.get(
name="Mpls Public Service Building",
)
digitools.createBasicEvent(e, 'Gv', venue)
except:
# print(event['Location'].split("-")[0])
e['location'] = event['Location'].split("-")[0].strip()
try:
venue, created = Organization.objects.get_or_create(
name=e['location'] ,
city="Minneapolis",
website=scraper.website,
is_venue=False,
address_complete=e['address']
)
except:
venue = Organization.objects.get(
name=e['location'] ,
city="Minneapolis",
)
digitools.createBasicEvent(e, 'Gv', venue)
event_list.append(e)
br.close()
scraper.save()
# br.find_element_by_class_name('fc-btn_allCalendars-button').click()