more scrapers
This commit is contained in:
@@ -34,18 +34,17 @@ def get_info(pse):
|
||||
except Exception as e:
|
||||
print("details issue: ", e)
|
||||
try:
|
||||
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"], DATETIME_FORMAT)
|
||||
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"] + tz_str, 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)
|
||||
event["date_time"] = datetime.strptime(" ".join(event["date"]) + " " + event["details"]["Doors Open"] + tz_str, 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)
|
||||
event["date_time"] = datetime.strptime(" ".join(event["date"]) + tz_str, DATETIME_FORMAT_3)
|
||||
print("The Event:")
|
||||
ppr(event)
|
||||
return event
|
||||
|
||||
def get_date(pse, event):
|
||||
@@ -78,10 +77,10 @@ venue, created = Organization.objects.get_or_create(
|
||||
scraper,item_count_start,virtcal = digitools.getScraper(venue, venue.website, 'msp')
|
||||
|
||||
tz = pytz.timezone("US/Central")
|
||||
|
||||
DATETIME_FORMAT = '%b %d %Y %I%p'
|
||||
DATETIME_FORMAT_2 = '%b %d %Y %I:%M%p'
|
||||
DATETIME_FORMAT_3 = '%b %d %Y'
|
||||
tz_str = " -0600 UTC"
|
||||
DATETIME_FORMAT = '%b %d %Y %I%p %z %Z'
|
||||
DATETIME_FORMAT_2 = '%b %d %Y %I:%M%p %z %Z'
|
||||
DATETIME_FORMAT_3 = '%b %d %Y %z %Z'
|
||||
|
||||
# Set initial variables for City, etc
|
||||
month = int(datetime.now().month)
|
||||
@@ -112,8 +111,6 @@ 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")
|
||||
|
||||
if len(sys.argv) >= 2:
|
||||
arg1 = sys.argv[1]
|
||||
br = digitools.getBrowser(arg1)
|
||||
@@ -128,6 +125,8 @@ if datetime.now().day < 8:
|
||||
elif 7 < datetime.now().day < 15:
|
||||
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')[:15]
|
||||
elif 14 < datetime.now().day < 21:
|
||||
ps = digitools.getSource(br, calendar_url)
|
||||
shows = ps.xpath('.//*/div[@class="show_name content flex-fill"]/div/div/h4/a/@href')[:95]
|
||||
@@ -176,4 +175,3 @@ br.close()
|
||||
|
||||
digitools.updateScraper(scraper, item_count_start)
|
||||
|
||||
# br.find_element_by_class_name('fc-btn_allCalendars-button').click()
|
||||
|
||||
Reference in New Issue
Block a user