DigiSnaxx/scrapers
3
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

DoD Contracts, Maps and Austria

Browse Source
This commit is contained in:
canin carlos
2026-02-19 22:49:47 -05:00
parent 3c4a41ae2c
commit 3eda7647e9
13 changed files with 469 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

0
Working/venues/au.vi.arenawien.py → Working/venues/au.vie.arenawien.py
View File

72
Working/venues/getMaps.py Normal file
View File

@@ -0,0 +1,72 @@
import os, sys
from datetime import datetime, timedelta
from dateutil import relativedelta
from pprint import pprint as ppr
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 selenium.webdriver.common.by import By
from events.models import Organization, Scraper, Calendar, Event
import events.digitools as digitools
import re
def getSite(br, website, org):
ps = digitools.getSource(br, website)
links = ps.xpath('.//*/div[@class="luibr"]/div/div/a/@href')
ppr(links)
nsite = "https://www.google.com"
if len(links) > 0:
nlink = nsite + links[0]
org.barrio = nlink
org.save()
sleep(5)
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
else:
print("No run_env")
br.close()
quit()
orgs = Organization.objects.all()
# for org in orgs:
# try:
# if len(org.city) > 0:
# msg = org.name.split(" ") + org.city.split(" ")
# site = "https://www.google.com/search?q=google+maps+" + "+".join(msg)
# print(site)
# getSite(br, site, org)
# except Exception as e:
# ppr(org)
# print(e)
for org in orgs[5:]:
try:
br.get(org.barrio)
org.gmap_link = org.barrio
sleep(5)
current_url = br.current_url
match = re.search(r'@(-?\d+\.\d+),(-?\d+\.\d+)', current_url)
print(match[1], match[0].split(",")[1])
org.latitude = match[1].strip()
org.longitude = match[0].split(",")[1].strip()
org.save()
except Exception as e:
ppr(org)
print(e)
br.close()

2
Working/venues/il.chi.subterranean.py
View File

@@ -44,7 +44,7 @@ def get_events(ps, event_type):
try:
event['support'] = c.xpath('.//p[@class="fs-12 supporting-talent"]/text()')[0]
except:
pass
continue
event['venue'] = c.xpath('.//p[@class="fs-12 venue"]/text()')[0]
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['link'] = venue.website

51
Working/venues/manual/eventario.py
View File

@@ -49,31 +49,33 @@ def getLinks(br, url, links):
div = br.find_element(By.XPATH, ".//*/div[@class='feed-pagination flexify']/a")
div.click()
sleep(2)
br.execute_script("window.scrollTo(0, window.scrollY + 1100)")
br.execute_script("window.scrollTo(0, window.scrollY + 1375)")
sleep(2)
except:
x = 0
ps = html.fromstring(br.page_source)
newLinks = []
# newlinks = ps.xpath('.//*/div[@class="e-con-inner"]/*/a/@href')
events = ps.xpath('.//*/div[@class="e-con-inner"]')
events = ps.xpath('.//*/div[@class="ts-preview"]')
for event in events:
ev = {}
try:
ev['link'] = event.xpath('.//*/a/@href')[0]
ev['link'] = event.xpath('.//*/h3/a/@href')[0]
ev['title'] = event.xpath('.//*/h3/a/text()')[0]
ev['venue'] = event.xpath('.//*/ul/li/a/text()')[-1:][0].replace('\n', '').replace('\t', '')
# e['venue'] = event.xpath('.//*/ul/li/a/text()')[-1:][0]
ev['venueLink'] = event.xpath('.//*/ul/li/a/@href')[1]
ev['venue'] = event.xpath('.//*/ul/li[@class="elementor-repeater-item-46edd7d flexify ts-action"]/a[@class="ts-action-con"]/text()')[-1:][0].replace('\n', '').replace('\t', '')
ev['venueLink'] = event.xpath('.//*/ul/li[@class="elementor-repeater-item-46edd7d flexify ts-action"]/a[@class="ts-action-con"]/@href')[-1:][0]
# ev['venueLink'] = event.xpath('.//*/ul/li/a/@href')[1]
label= event.xpath('.//*/li[@class="elementor-repeater-item-46edd7d flexify ts-action"]/div/text()')
ev['label'] = ''.join([x.replace('\t', '').replace('\n', '') for x in label]).strip()
newLinks.append(ev)
# ppr(ev)
except Exception as e:
print("Error: ", ev, e)
links = links + newLinks
return links
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
@@ -89,11 +91,15 @@ for url in urls:
allLinks = getLinks(br, url, allLinks)
totalLinks = list({v['title']:v for v in allLinks}.values())
ppr(totalLinks)
ppr(len(totalLinks))
# sortedlinks = allLinks.sort()
# ppr(sortedlinks)
for event in totalLinks:
# quit()
for event in totalLinks[175:250]:
br.get(event['link'])
sleep(1)
ps = html.fromstring(br.page_source)
@@ -156,6 +162,10 @@ for d in new_data:
cal = Calendar.objects.get(shortcode='mde')
print("run paisa")
ppr(paisa)
for d in paisa:
try:
nvenue, created = Organization.objects.get_or_create(
@@ -170,17 +180,20 @@ for d in paisa:
nvenue.address_complete = d['address']
nvenue.save()
new_event, created = Event.objects.update_or_create(
event_type = d['category'],
show_title = d['title'],
show_link = d['link'],
show_date = d['dateStamp'],
show_day = d['dateStamp'],
scraper = scraper,
venue = nvenue
)
new_event.calendar.add(cal)
new_event.save()
# print("Success:", new_event)
try:
new_event, created = Event.objects.update_or_create(
event_type = d['category'],
show_title = d['title'],
show_link = d['link'],
show_date = d['dateStamp'],
scraper = scraper,
venue = nvenue
)
new_event.calendar.add(cal)
new_event.save()
print(new_event, created, new_event.scraper)
# print("Success:", new_event)
except Execption as e:
print(e)
digitools.updateScraper(scraper, item_count_start)

2
Working/venues/mn.mpls.HookNLadder.py
View File

@@ -81,11 +81,11 @@ for event in events:
show_title = event["show_title"],
show_link = event["link"],
show_date = event["date_time"],
show_day = event["date_time"],
guests = " ".join(event["subtitle"]),
venue = venue
)
digitools.add_calendar(new_event, 'msp')
ppr(new_event)
except Exception as e:
print("oops ", e, "\n\n", "Scraper:", scraper)

2
Working/venues/mn.mpls.ParkwayTheater.py
View File

@@ -55,7 +55,6 @@ def get_events(ps, event_type):
show_title = event['title'],
show_link = event['link'],
show_date = datetime.strptime(event['date'] + tz_str, DATETIME_FORMAT),
show_day = datetime.strptime(event['date'] + tz_str, DATETIME_FORMAT),
more_details = event["details"],
venue = venue
)
@@ -71,7 +70,6 @@ def get_events(ps, event_type):
show_title = event['title'],
show_link = event['link'],
show_date = datetime.strptime(event['date'] + tz_str, DATETIME_FORMAT),
show_day = datetime.strptime(event['date'] + tz_str, DATETIME_FORMAT),
more_details = event["details"],
venue = venue
)

13
Working/venues/mn.mpls.UptownVFW.py
View File

@@ -26,8 +26,8 @@ venue, created = Organization.objects.get_or_create(
scraper,item_count_start, virtcal = digitools.getScraper(venue, venue.website, 'msp')
tz_str = " -0600 UTC"
DATETIME_FORMAT = '%a %B %d @ %I:%M %p %Y %z %Z'
DATETIME_FORMAT_2 = '%a %B %d, %Y @ %I:%M %p %z %Z'
DATETIME_FORMAT = '%a %B %d @ %I:%M %p %Y %z %Z'
DATETIME_FORMAT_2 = '%a %B %d @ %I:%M %p %Y %z %Z'
calendar_url = 'https://noboolpresents.com/venues/uptown-vfw/'
current_year = str(datetime.now().year)
@@ -50,32 +50,33 @@ def getEvents(br):
title = event.xpath('.//*/h2[@class="alt-font"]/a/text()')[0].replace("\n", "").replace("\t", "")
link = event.xpath('.//*/h2[@class="alt-font"]/a/@href')[0]
try:
print("First Try", dateTime)
new_event, created = Event.objects.update_or_create(
scraper = scraper,
event_type = 'Mu',
show_title = title,
show_link = link,
show_date = datetime.strptime(dateTime + tz_str, DATETIME_FORMAT),
show_day = datetime.strptime(dateTime + tz_str, DATETIME_FORMAT),
venue = venue
)
digitools.add_calendar(new_event, 'msp')
scraper.items+=1
except Exception as e:
try:
print("New Try", dateTime)
new_event, created = Event.objects.update_or_create(
scraper = scraper,
event_type = 'Mu',
show_title = title,
show_link = link,
show_date = datetime.strptime(dateTime[:-4].strip() + tz_str, DATETIME_FORMAT_2),
show_day = datetime.strptime(dateTime[:-4].strip() + tz_str, DATETIME_FORMAT_2),
show_date = datetime.strptime(dateTime.strip() + tz_str, DATETIME_FORMAT_2),
venue = venue
)
digitools.add_calendar(new_event, 'msp')
scraper.items+=1
except Exception as e:
print("oops", e, dateTime[:-4])
print("oops", e)
ppr(event)
getEvents(br)
br.find_element(By.XPATH, './/*/li[@class="tribe-events-c-nav__list-item tribe-events-c-nav__list-item--next"]/a').click()