updated naming conventions

This commit is contained in:
2026-01-09 22:49:19 -05:00
parent 88396f48ae
commit 7013d8327a
60 changed files with 999 additions and 2004 deletions

View File

@@ -1,73 +0,0 @@
import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
from time import sleep
from pprint import pprint as ppr
import pytz
from events.models import Organization, Scraper
import events.digitools as digitools
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%Y-%m-%d %I:%M %p'
venue, created = Organization.objects.get_or_create(
name="White Squirrel",
city="St. Paul",
website="https://whitesquirrelbar.com",
is_venue = True
)
scraper,item_count_start, virtcal = digitools.getScraper(venue, 'msp')
# Set initial variables for City, etc
calendar_url = [
'https://whitesquirrelbar.com/calendar/list/page/1/',
'https://whitesquirrelbar.com/calendar/list/page/2/',
'https://whitesquirrelbar.com/calendar/list/page/3/'
]
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
else:
print("No run_env")
quit()
def get_listings(pse, events):
listings = pse.xpath('.//*/li[@class="tribe-common-g-row tribe-events-calendar-list__event-row"]')
print("get listings", listings)
for l in listings:
event = {}
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
print("1: ", event)
try:
event["image"] = l.xpath('.//*/img/@src')[0]
except:
event["image"] = "none"
event["date"] = l.xpath('.//time/@datetime')[0]
event["time"] = l.xpath('.//*/span[@class="tribe-event-date-start"]/text()')[0].split("@")[1]
event["title"] = l.xpath('.//*/h3/a/text()')[0].replace("\t", "").replace("\n", "")
event["link"] = l.xpath('.//*/h3/a/@href')[0]
event['datetime'] = event['date'] + " " + event['time']
event["dateStamp"] = datetime.strptime(event['datetime'] , DATETIME_FORMAT)
events.append(event)
digitools.createBasicEvent(event, "Mu", venue)
scraper.items+=1
events = []
for cal in calendar_url:
ps = digitools.getSource(br, cal)
get_listings(ps, events)
br.close()
digitools.updateScraper(scraper, item_count_start)

View File

@@ -22,7 +22,7 @@ import events.digitools as digitools
from events.models import Organization, Scraper, Calendar, Event
venue, created = Organization.objects.get_or_create(
name="Acme Comedy Club",
name="Comuniful",
city="Minneapolis",
website="https://acmecomedycompany.com/the-club/calendar/",
is_venue = True

View File

@@ -6,10 +6,9 @@ import json
from selenium.webdriver.common.by import By
from lxml import html
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -25,7 +24,7 @@ venue, created = Organization.objects.get_or_create(
is_venue=True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'mde')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'mde')
DATETIME_FORMAT = '%B %d %Y %I:%M%p'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'

View File

@@ -0,0 +1,70 @@
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")
DATETIME_FORMAT = '%b %d %Y %I:%M %p'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'
venue, created = Organization.objects.get_or_create(
name="Acme Comedy Club",
city="Minneapolis",
website="https://acmecomedycompany.com/the-club/calendar/",
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
scraper.items = 0
scraper.save()
def get_events(ps, event_type):
contents = ps.xpath('.//*/li[@class="event"]')
for c in contents:
try:
event = {}
day = c.xpath('.//*/span[@class="day"]/text()')[0]
month = c.xpath('.//*/span[@class="mth"]/text()')[0]
year = datetime.now().year
if month == "Jan":
year = int(year) + 1
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event['title'] = c.xpath('.//*/span[@class="event_title"]/a/text()')[0]
event['date'] = [month, day, str(year), c.xpath('.//*/span[@class="event_time"]/text()')[0].strip()]
event['date'] = " ".join(event['date'])
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['link'] = c.xpath('.//*/span[@class="event_title"]/a/@href')[0]
digitools.createBasicEvent(event, "Co", 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")
br.close()
quit()
links = digitools.createURL("https://acmecomedycompany.com/the-club/calendar/")
for link in links:
ps = digitools.getSource(br, link)
get_events(ps, "Co")
digitools.updateScraper(scraper, item_count_start)
br.close()

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -22,7 +21,7 @@ venue, created = Organization.objects.get_or_create(
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
event_type = "Mu"

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -13,8 +12,6 @@ import pytz
from events.models import Organization, Scraper, Event
import events.digitools as digitools
from lxml import html
count = 0
@@ -26,7 +23,7 @@ venue, created = Organization.objects.get_or_create(
is_venue=True,
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
ppr(scraper)
tz = pytz.timezone("US/Central")

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -21,7 +20,7 @@ venue, created = Organization.objects.get_or_create(
is_venue=False
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
DATETIME_FORMAT = '%A, %B %d , %Y %I:%M %p'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -14,7 +13,6 @@ import pytz
from events.models import Organization, Scraper, Event as DSEvent
import events.digitools as digitools
try:
venue, created = Organization.objects.get_or_create(
name="Parkway Theater",
@@ -25,7 +23,7 @@ try:
except Exception as e:
venue = Organization.objects.get(name="Parkway Theater")
scraper,item_count_start, virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start, virtcal = digitools.getScraper(venue, venue.website, 'msp')
tz = pytz.timezone("US/Central")

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -25,13 +24,14 @@ venue, created = Organization.objects.get_or_create(
is_venue = True
)
scraper,item_count_start, virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start, virtcal = digitools.getScraper(venue, venue.website, 'msp')
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%a %B %d @ %I:%M %p %Y'
DATETIME_FORMAT_2 = '%b %d %I:%M%p %Y'
DATETIME_FORMAT_3 = '%b %d %Y'
DATETIME_FORMAT_2 = '%a %B %d, %Y @ %I:%M %p'
# DATETIME_FORMAT_2 = '%b %d %I:%M%p %Y'
# DATETIME_FORMAT_3 = '%b %d %Y'
# Set initial variables for City, etc
calendar_url = 'https://noboolpresents.com/venues/uptown-vfw/'
current_year = str(datetime.now().year)
@@ -69,7 +69,21 @@ def getEvents(br):
digitools.add_calendar(new_event, 'msp')
scraper.items+=1
except Exception as e:
print("oops", e)
try:
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(), DATETIME_FORMAT_2),
show_day = datetime.strptime(dateTime[:-4].strip(), DATETIME_FORMAT_2),
# more_details = deets["tickets"],
venue = venue
)
digitools.add_calendar(new_event, 'msp')
scraper.items+=1
except Exception as e:
print("oops", e, dateTime[:-4])
getEvents(br)
br.find_element(By.XPATH, './/*/li[@class="tribe-events-c-nav__list-item tribe-events-c-nav__list-item--next"]/a').click()

View File

@@ -0,0 +1,71 @@
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
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%b %d %I:%M %p %Y'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'
venue, created = Organization.objects.get_or_create(
name="Cabooze",
city="Minneapolis",
website="https://www.cabooze.com/#/events",
is_venue=True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
def get_events(ps, event_type):
print("Getting events ...")
contents = ps.xpath('.//*/div[@class="vp-event-row vp-widget-reset vp-venue-thecabooze"]')
ppr(contents)
for c in contents:
try:
event = {}
date = c.xpath('.//*/span[@class="vp-month-n-day"]/text()')[0]
print(date)
month = date.split(" ")[0]
time = c.xpath('.//*/span[@class="vp-time"]/text()')[0]
year = datetime.now().year
if month == "Jan":
year = int(year) + 1
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event['title'] = c.xpath('.//*/div[@class="vp-event-name"]/text()')[0]
event['date'] = [date, time, str(year)]
event['date'] = " ".join(event['date'])
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['link'] = "https://www.cabooze.com/" + c.xpath('.//a[@class="vp-event-link"]/@href')[0]
# print("Event Dict Created")
# 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()
ps = digitools.getSource(br, venue.website)
get_events(ps, "Mu")
sleep(3)
digitools.updateScraper(scraper, item_count_start)
br.close()

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -21,9 +20,7 @@ venue, created = Organization.objects.get_or_create(
is_venue=True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
tz = pytz.timezone("US/Central")
scraper,item_count_start,virtcal = digitools.getScraper(venue,venue.website, 'msp')
DATETIME_FORMAT = '%A, %B %d, %Y %I:%M %p'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'
@@ -33,6 +30,9 @@ DATETIME_FORMAT_5 = '%A, %B %d @%I%p %Y'
def get_events(ps):
links = ps.xpath('.//*/div[@class="summary-title"]/a/@href')
print("Length of Links: ", len(links))
links = list(set(links))
print("New Length of Links: ", len(links))
for l in links:
if "cedar-news-blog" in l:
continue

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -17,6 +16,16 @@ from lxml import html
from events.models import Organization, Scraper, Event
import events.digitools as digitools
def process_times(times):
time = []
for t in times:
t = t.replace("\n", "").replace("TBA", "")
if len(t) > 0 and t.endswith("pm"):
if "-" in t:
t = t.split("-")[0] + "pm"
time.append(t)
return time
venue, created = Organization.objects.get_or_create(
name="Club 331",
city="Minneapolis",
@@ -24,13 +33,12 @@ venue, created = Organization.objects.get_or_create(
is_venue=True,
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
tz = pytz.timezone("US/Central")
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
DATETIME_FORMAT = '%b %d %I%p %Y'
DATETIME_FORMAT_2 = '%b %d %I:%M%p %Y'
DATETIME_FORMAT_3 = '%b %d %Y'
# Set initial variables for City, etc
calendar_url = 'https://331club.com/#calendar'
current_year = str(datetime.now().year)
@@ -53,19 +61,7 @@ sleep(3)
dates = ps.xpath('.//*/div[@class="event"]')
dates = dates + ps.xpath('.//*/div[@class="event hidden"]')
def process_times(times):
# print("Times: ", times)
time = []
for t in times:
t = t.replace("\n", "").replace("TBA", "")
if len(t) > 0 and t.endswith("pm"):
if "-" in t:
t = t.split("-")[0] + "pm"
time.append(t)
return time
events = []
for d in dates:
event_date = d.xpath('.//div[@class="event-date"]/span/text()')[:2]
cols = d.xpath('.//div[@class="column"]')
@@ -101,5 +97,4 @@ for d in dates:
events.append(event)
br.close()
digitools.updateScraper(scraper, item_count_start)

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -16,6 +15,59 @@ import pytz
from events.models import Organization, Scraper, Event
import events.digitools as digitools
def get_info(pse):
event = {}
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event["venue"] = pse.xpath('.//*/div[@class="content"]/div/div[@class="venue_name"]/text()')[0].replace('\t', '').replace('\n', '').strip()
event["show_title"] = pse.xpath('.//*/span[@class="show_title"]/text()')[0].replace('\t', '').replace('\n', '')
if event["show_title"] == "":
event["show_title"] = pse.xpath('.//*/span[@class="show_title"]/text()')[2].replace('\t', '').replace('\n', '')
event["guests"] = pse.xpath('.//*/div[@class="feature_details_main d-flex align-items-center"]/div/h4/text()')
event["flyer"] = pse.xpath('.//*/img[@class="gig_poster lazy loaded"]/@src')
try:
event = get_date(pse, event)
except Exception as e:
print("date issue: ", e)
try:
event = get_details(pse, event)
except Exception as e:
print("details issue: ", e)
try:
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"], DATETIME_FORMAT)
except Exception as e:
print("Using alt date format 2: ", e)
try:
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"], DATETIME_FORMAT_2)
ppr(event)
except Exception as e:
print("Using alt date format 3: ", e)
print(event['date'])
event["date_time"] = datetime.strptime(" ".join(event["date"]), DATETIME_FORMAT_3)
print("The Event:")
ppr(event)
return event
def get_date(pse, event):
month = pse.xpath('.//*/div[@class="date_container"]/div/div[@class="month"]/text()')[0].replace('\t', '').replace('\n', '')
day = pse.xpath('.//*/div[@class="date_container"]/div/div[@class="day"]/text()')[0].replace('\t', '').replace('\n', '')
year = pse.xpath('.//*/div[@class="date_container"]/div/div[@class="year"]/text()')[0].replace('\t', '').replace('\n', '')
event["date"] = [month, day, year]
return event
def get_details(pse, event):
try:
details = pse.xpath('.//*/div[@class="show_details text-center"]/div/div/h6/text()')
info = pse.xpath('.//*/div[@class="show_details text-center"]/div/div/h2/text()')
di = zip(details, info)
details = {}
for d,i in di:
details[d] = i
event["details"] = details
return event
except Exception as e:
print("details issue: ", e)
venue, created = Organization.objects.get_or_create(
name="First Avenue",
city="Minneapolis",
@@ -23,7 +75,7 @@ venue, created = Organization.objects.get_or_create(
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
tz = pytz.timezone("US/Central")
@@ -60,7 +112,7 @@ else:
calendar_url_2 = 'https://first-avenue.com/shows/?start_date=' + str(year) + next_month_string
print("\n\n", calendar_url, calendar_url_2, "\n\n")
# print("\n\n", calendar_url, calendar_url_2, "\n\n")
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
@@ -69,7 +121,7 @@ else:
print("No run_env")
quit()
# Get Events based on date of month
if datetime.now().day < 8:
ps = digitools.getSource(br, calendar_url)
shows = ps.xpath('.//*/div[@class="show_name content flex-fill"]/div/div/h4/a/@href')[:63]
@@ -84,63 +136,11 @@ elif 14 < datetime.now().day < 21:
else:
ps = digitools.getSource(br, calendar_url)
shows = ps.xpath('.//*/div[@class="show_name content flex-fill"]/div/div/h4/a/@href')
ps = digitools.getSource(br, calendar_url_2)
shows = shows + ps.xpath('.//*/div[@class="show_name content flex-fill"]/div/div/h4/a/@href')[:63]
events = []
def get_info(pse):
event = {}
event['scraper'] = scraper
event['calendars'] = [scraper.calendar]
event["venue"] = pse.xpath('.//*/div[@class="content"]/div/div[@class="venue_name"]/text()')[0].replace('\t', '').replace('\n', '').strip()
event["show_title"] = pse.xpath('.//*/span[@class="show_title"]/text()')[0].replace('\t', '').replace('\n', '')
if event["show_title"] == "":
event["show_title"] = pse.xpath('.//*/span[@class="show_title"]/text()')[2].replace('\t', '').replace('\n', '')
event["guests"] = pse.xpath('.//*/div[@class="feature_details_main d-flex align-items-center"]/div/h4/text()')
event["flyer"] = pse.xpath('.//*/img[@class="gig_poster lazy loaded"]/@src')
try:
event = get_date(pse, event)
except Exception as e:
print("date issue: ", e)
try:
event = get_details(pse, event)
except Exception as e:
print("details issue: ", e)
try:
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"], DATETIME_FORMAT)
except Exception as e:
print("Using alt date format 2: ", e)
try:
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"], DATETIME_FORMAT_2)
ppr(event)
except Exception as e:
print("Using alt date format 3: ", e)
print(event['date'])
event["date_time"] = datetime.strptime(" ".join(event["date"]), DATETIME_FORMAT_3)
return event
def get_date(pse, event):
month = pse.xpath('.//*/div[@class="date_container"]/div/div[@class="month"]/text()')[0].replace('\t', '').replace('\n', '')
day = pse.xpath('.//*/div[@class="date_container"]/div/div[@class="day"]/text()')[0].replace('\t', '').replace('\n', '')
year = pse.xpath('.//*/div[@class="date_container"]/div/div[@class="year"]/text()')[0].replace('\t', '').replace('\n', '')
event["date"] = [month, day, year]
return event
def get_details(pse, event):
try:
details = pse.xpath('.//*/div[@class="show_details text-center"]/div/div/h6/text()')
info = pse.xpath('.//*/div[@class="show_details text-center"]/div/div/h2/text()')
di = zip(details, info)
details = {}
for d,i in di:
details[d] = i
event["details"] = details
return event
except Exception as e:
print("details issue: ", e)
for show in shows:
br.get(show)
sleep(2)
@@ -153,6 +153,7 @@ for show in shows:
event = get_info(pse)
except Exception as e:
print("get_info error: ", e)
pass
try:
event["link"] = show
if event["venue"] in ["Palace Theater", "Turf Club", "The Fitzgerald Theater", "Amsterdam Bar & Hall"]:
@@ -168,7 +169,7 @@ for show in shows:
scraper.items+=1
except Exception as e:
print("event creation error: ", e, "\n\n", event, "\n\n", created)
quit()
# quit()
ppr(events)
br.close()

View File

@@ -0,0 +1,70 @@
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
count = 0
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%a, %b %d %Y %I%p'
DATETIME_FORMAT_2 = '%a, %b %d %Y %I:%M %p SHOW'
venue, created = Organization.objects.get_or_create(
name="Icehouse",
city="Minneapolis",
website = "https://icehouse.turntabletickets.com",
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
def get_events(ps, event_type):
contents = ps.xpath('.//*/div[@class="performances whitespace-pre-line w-full md:w-3/4"]')
for c in contents:
try:
event = {}
event['scraper'] = scraper
event['calendars'] = scraper.calendar
event['title'] = c.xpath('.//*/h3[@class="text-3xl font-semibold font-heading mr-auto"]/text()')[0]
event['link'] = venue.website + c.xpath('.//*/a[@class="show-link"]/@href')[0]
event['date'] = c.xpath('.//*/h4[@class="day-of-week"]/text()')[0]
month = event['date'].split(' ')[1].strip()
print("MONTH: ", month)
year = int(datetime.today().year)
if month in ['Jan', 'Feb', 'Mar']:
year = year + 1
time = c.xpath('.//p/span/text()')[0][:4].strip()
if time[-1:] == 'm':
event['dateStamp'] = event['date'] + ' ' + str(year) + ' ' + time
event['dateStamp'] =datetime.strptime(event['dateStamp'], DATETIME_FORMAT)
digitools.createBasicEvent(event, event_type, venue)
else:
ppr(event)
print("MONTH: ", month)
except Exception as e:
print("What?",e,"\n\n+++")
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
else:
print("No run_env")
quit()
ps = digitools.getSource(br, venue.website)
br.execute_script("window.scrollTo(0, window.scrollY + 5000)")
get_events(ps, "Mu")
# ppr(events)
br.close()
digitools.updateScraper(scraper, item_count_start)

View File

@@ -0,0 +1,78 @@
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
current_year = str(datetime.now().year)
venue, created = Organization.objects.get_or_create(
name="Piller Forum",
city="Minneapolis",
website="https://www.pilllar.com/pages/events",
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
event_type = "Mu"
# Time Signatures
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%b. %d %Y %I:%M%p'
DATETIME_FORMAT_night = '%b. %d %Y %I:%M %p'
DATETIME_FORMAT_2 = '%b. %d %Y %I:%Mam'
def get_events(ps, event_type):
contents = ps.xpath('.//*/div[@class="sse-row sse-clearfix"]')
for c in contents:
year = datetime.today().year
try:
date = c.xpath('.//h1[@class="sse-size-64"]/text()')[0]
month = date.split(".")[0]
ppr(date)
if month in ['JAN', 'FEB', 'MAR']:
year = int(datetime.today().year) + 1
event = {}
event['scraper'] = scraper
event['calendars'] = scraper.calendar
event['link'] = venue.website
event['title'] = c.xpath('.//p/span/b/text()')[0]
event['deets'] = c.xpath('.//p/span/text()')[0]
event['title'] = event['title'] + ' ' + event['deets']
paras = c.xpath('.//p/text()')
times = paras[1].split(" ")[1]
event['datetime'] = "{0} {1} {2}".format(date, year, times)
event['dateStamp'] =datetime.strptime(event['datetime'], DATETIME_FORMAT)
ppr(event)
digitools.createBasicEvent(event, event_type, venue)
except Exception as e:
# print(e)
pass
if len(sys.argv) >= 2:
arg1 = sys.argv[1]
br = digitools.getBrowser(arg1)
else:
print("No run_env")
quit()
ps = digitools.getSource(br, venue.website)
get_events(ps, event_type)
sleep(3)
br.close()
digitools.updateScraper(scraper, item_count_start)

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -21,7 +20,7 @@ venue, created = Organization.objects.get_or_create(
is_venue=True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
DATETIME_FORMAT = '%B %d %Y %I:%M%p'
DATETIME_FORMAT_2 = '%A, %B %d @ %I%p %Y'

View File

@@ -2,10 +2,9 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
@@ -21,7 +20,7 @@ venue, created = Organization.objects.get_or_create(
is_venue = True
)
scraper,item_count_start,virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
event_type = ""

View File

@@ -2,10 +2,10 @@ import os, sys
from datetime import datetime
from dateutil import relativedelta
import django
sys.path.append('../../../../')
os.environ['DJANGO_SETTINGS_MODULE'] = 'config.django.local'
django.setup()
sys.path.append('/var/www/digisnaxx.ado/scrapers')
import dtss
dtss.getReady()
from time import sleep
from pprint import pprint as ppr
import pytz
@@ -21,7 +21,7 @@ venue, created = Organization.objects.get_or_create(
is_venue = False
)
scraper,item_count_start, virtcal = digitools.getScraper(venue, 'msp')
scraper,item_count_start, virtcal = digitools.getScraper(venue, venue.website, 'msp')
# Time Signatures
tz = pytz.timezone("US/Central")