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 +1,53 @@
import requests, os, sys
from icalendar import Calendar as iCalendar, Event
from datetime import datetime, timedelta
from dateutil import relativedelta
td = relativedelta.relativedelta(hours=5)
from pprint import pprint as ppr
import pytz
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from events.models import Event as DSEvent, Organization, Scraper, Calendar
import events.digitools as digitools
venue, created = Organization.objects.get_or_create(
name="Yoga MDE",
city="Medellin",
website="https://digisnaxx.com/",
)
try:
scraper, created = Scraper.objects.get_or_create(
name="Yoga MDE",
website="https://calendar.google.com/calendar/ical/721a61e06e9a96b7bd30cdb7439aa7791829c316bb8b77237ea93cf41d5396b4%40group.calendar.google.com/public/basic.ics",
calendar = Calendar.objects.get(shortcode='mde'),
items = 0,
new_items = 0,
last_ran = datetime.now(),
)
except Exception as e:
print(e)
scraper = Scraper.objects.get(name=venue.name)
item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text.strip())
counter = 0
digitools.getiCalRepeateEvents(gcal, scraper, venue, "Ex", "mde")
digitools.updateScraper(scraper, item_count_start)
new_time = datetime.now() - timedelta(days=1)
right_bound_time = datetime.now() + timedelta(days=45)
events = DSEvent.objects.filter(show_date__lte=new_time)
events1 = DSEvent.objects.filter(show_date__gte=right_bound_time)
for e in events:
e.delete()
for e in events1:
e.delete()

View File

@@ -0,0 +1,55 @@
import requests, os, sys
from icalendar import Calendar as iCalendar, Event
from datetime import datetime, timedelta
from dateutil import relativedelta
td = relativedelta.relativedelta(hours=5)
from pprint import pprint as ppr
import pytz
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from events.models import Event as DSEvent, Organization, Scraper, Calendar
import events.digitools as digitools
venue, created = Organization.objects.get_or_create(
name="idioki",
city="Medellin",
website="https://idioki.com/",
)
try:
scraper, created = Scraper.objects.get_or_create(
name="Yoga MDE",
website="",
calendar = Calendar.objects.get(shortcode='mde'),
items = 0,
new_items = 0,
last_ran = datetime.now(),
)
except Exception as e:
print(e)
scraper = Scraper.objects.get(name=venue.name)
event_type = "Mu"
item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
counter = 0
digitools.getiCalRepeateEvents(gcal, scraper, venue, "Ed", "mde")
digitools.updateScraper(scraper, item_count_start)
new_time = datetime.now() - timedelta(days=1)
right_bound_time = datetime.now() + timedelta(days=45)
events = DSEvent.objects.filter(show_date__lte=new_time)
events1 = DSEvent.objects.filter(show_date__gte=right_bound_time)
for e in events:
e.delete()
for e in events1:
e.delete()

View File

@@ -1,7 +1,7 @@
import requests, os, sys
from icalendar import Calendar as iCalendar, Event
from datetime import datetime
from datetime import datetime, timedelta
from dateutil import relativedelta
td = relativedelta.relativedelta(hours=5)
@@ -34,13 +34,11 @@ except Exception as e:
print(e)
scraper = Scraper.objects.get(name=venue.name)
event_type = "Mu"
item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
counter = 0
digitools.getiCalRepeateEvents(gcal, scraper, venue, "Ed", cal)
digitools.getiCalRepeateEvents(gcal, scraper, venue, "Ed", "mde")
digitools.updateScraper(scraper, item_count_start)
new_time = datetime.now() - timedelta(days=1)
@@ -52,4 +50,4 @@ for e in events:
e.delete()
for e in events1:
e.delete()
e.delete()

View File

@@ -1,7 +1,7 @@
import requests, os, sys
from icalendar import Calendar as iCalendar, Event
from datetime import datetime
from datetime import datetime, timedelta
from dateutil import relativedelta
td = relativedelta.relativedelta(hours=5)
@@ -40,17 +40,18 @@ objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
counter = 0
digitools.getiCalRepeateEvents(gcal, scraper, venue, "Ed", cal)
print("Got Gcal")
digitools.getiCalRepeateEvents(gcal, scraper, venue, "Ed", "000")
print("Got EVents")
digitools.updateScraper(scraper, item_count_start)
print("Updated")
# new_time = datetime.now() - timedelta(days=1)
# right_bound_time = datetime.now() + timedelta(days=45)
# events = DSEvent.objects.filter(show_date__lte=new_time)
# events1 = DSEvent.objects.filter(show_date__gte=right_bound_time)
new_time = datetime.now() - timedelta(days=1)
right_bound_time = datetime.now() + timedelta(days=45)
events = DSEvent.objects.filter(show_date__lte=new_time)
events1 = DSEvent.objects.filter(show_date__gte=right_bound_time)
# for e in events:
# e.delete()
for e in events:
e.delete()
for e in events1:
e.delete()
# for e in events1:
# e.delete()

View File

@@ -21,13 +21,19 @@ venue, created = Organization.objects.get_or_create(
website="https://chicityclerkelms.chicago.gov/Meetings/",
)
website="https://calendar.google.com/calendar/ical/chicagolegislativereference%40gmail.com/public/basic.ics",
website="https://calendar.google.com/calendar/ical/chicagolegislativereference%40gmail.com/public/basic.ics"
# website="https://calendar.google.com/calendar/ical/chicagolegislativereference%40gmail.com/public/basic.ics"
scraper,item_count_start,virtcal = digitools.getScraper(venue, website, 'chi')
print("Got Scraper")
event_type = "Gv"
item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
objIcalData = requests.get(website)
print("Got Obj")
gcal = iCalendar.from_ical(objIcalData.text)
print("Got Gcal")
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
print("Got Events")
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -32,5 +32,5 @@ objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -28,5 +28,5 @@ item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -28,5 +28,5 @@ item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -28,5 +28,5 @@ item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
events = digitools.getiCalEvents(gcal, scraper, venue, "Mu")
digitools.buildiCalEvents(events, event_type)
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -29,5 +29,5 @@ item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -28,5 +28,5 @@ item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.text)
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
digitools.buildiCalEvents(events, event_type, scraper, venue)
digitools.updateScraper(scraper, item_count_start)

View File

@@ -28,6 +28,9 @@ scraper,item_count_start,virtcal = digitools.getScraper(venue, website, 'msp')
item_count_start = scraper.items
objIcalData = requests.get(scraper.website)
gcal = iCalendar.from_ical(objIcalData.content)
print("Got gcal")
events = digitools.getiCalEvents(gcal, scraper, venue, event_type)
digitools.buildiCalEvents(events, event_type)
print("Got Events")
digitools.buildiCalEvents(events, event_type, scraper, venue)
print("Built Events")
digitools.updateScraper(scraper, item_count_start)