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

@@ -26,8 +26,8 @@ except Exception as e:
scraper,item_count_start, virtcal = digitools.getScraper(venue, venue.website, 'msp')
tz = pytz.timezone("US/Central")
DATETIME_FORMAT = '%b %d, %Y %I:%M %p'
tz_str = " -0600 UTC"
DATETIME_FORMAT = '%b %d, %Y %I:%M %p %z %Z'
def get_events(ps, event_type):
contents = ps.xpath('.//*/div[@class="summary-content sqs-gallery-meta-container"]')
@@ -40,7 +40,7 @@ def get_events(ps, event_type):
event['title'] = c.xpath('.//*/a[@class="summary-title-link"]/text()')[0]
event['link'] = "https://theparkwaytheater.com" + c.xpath('.//*/a[@class="summary-title-link"]/@href')[0]
event['date'] = c.xpath('.//div/div/time/text()')[0] + " " + c.xpath('.//*/span[@class="event-time-12hr"]/text()')[0].split("")[0].strip()
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['dateStamp'] = datetime.strptime(event['date'] + tz_str, DATETIME_FORMAT)
event['desc'] = c.xpath('.//*/p/text()')[0]
event['img_link'] = i
event['details'] = {
@@ -54,8 +54,8 @@ def get_events(ps, event_type):
event_type = event_type,
show_title = event['title'],
show_link = event['link'],
show_date = datetime.strptime(event['date'], DATETIME_FORMAT),
show_day = datetime.strptime(event['date'], DATETIME_FORMAT),
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
)
@@ -64,14 +64,14 @@ def get_events(ps, event_type):
except Exception as e:
try:
event['date'] = c.xpath('.//div/div/time/text()')[0].split("")[0] + " " + c.xpath('.//*/span[@class="event-time-12hr"]/text()')[0].split("")[0].strip()
event['dateStamp'] = datetime.strptime(event['date'], DATETIME_FORMAT)
event['dateStamp'] = datetime.strptime(event['date'] + tz_str, DATETIME_FORMAT)
new_event, created = DSEvent.objects.update_or_create(
scraper = scraper,
event_type = event_type,
show_title = event['title'],
show_link = event['link'],
show_date = datetime.strptime(event['date'], DATETIME_FORMAT),
show_day = datetime.strptime(event['date'], DATETIME_FORMAT),
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
)
@@ -97,8 +97,6 @@ get_events(ps, "Mu")
calendar_url = "https://theparkwaytheater.com/movies"
ps = digitools.getSource(br, calendar_url)
get_events(ps, "Th")
# ppr(events)
br.close()
digitools.updateScraper(scraper, item_count_start)
digitools.updateScraper(scraper, item_count_start)