def datestr_to_date(datestr):
"""
@param datestr [str]: "MON DD, H:MM PM EST"
(e.g. "NOV 29, 6:00 PM EST")
"MM/DD H:MM PM EST"
(e.g. "02/18 7:00 PM EST")
@return [datetime.date]
"""
if "," in datestr:
return get_date_yearless(datestr.split(",")[0])
else:
datenum = datestr.split(" ")[0]
month, day = [int(s) for s in datenum.split("/")]
monthstr = datetime.date(1900, month, 1).strftime("%b")
return get_date_yearless("%s %s" % (monthstr, day))
def run():
response = requests.get(URL)
soup = BeautifulSoup(response.text, "html5lib")
rows = []
player = date = status = comment = None
for tr in soup.find_all("tbody")[0].children:
row = [td for td in tr if hasattr(tr, "children")]
if player and date and status:
if len(row) == 1 and "Comment:" in row[0].contents[0].string:
Injury.objects.update_or_create(
player=player,
date=date,
comment=unicode(row[0].contents[-1]),
defaults={"status": status},
)
else:
Injury.objects.update_or_create(player=player,
date=date,
defaults={"status": status})
player = date = status = comment = None
if len(row) == 3:
name, status, datestr = [r.string for r in row]
if name != "NAME" and status != "STATUS" and datestr != "DATE":
date = get_date_yearless(datestr)
try:
player = Player.get_by_name(name)
except Player.DoesNotExist:
print(f"Couldn't find player {name}")
def datestr_to_date(datestr):
"""
@param datestr [str]: "MON DD, H:MM PM EST"
(e.g. "NOV 29, 6:00 PM EST")
@return [datetime.date]
"""
return get_date_yearless(datestr.split(',')[0])
def run():
response = requests.get(URL)
soup = BeautifulSoup(response.text, 'html5lib')
rows = []
player = date = status = comment = None
for tr in soup.find_all('tbody')[0].children:
row = [td for td in tr if hasattr(tr, 'children')]
if player and date and status:
if len(row) == 1 and 'Comment:' in row[0].contents[0].string:
Injury.objects.update_or_create(
player=player,
date=date,
comment=unicode(row[0].contents[-1]),
defaults={ 'status': status }
)
else:
Injury.objects.update_or_create(
player=player,
date=date,
defaults={ 'status': status }
)
player = date = status = comment = None
if len(row) == 3:
name, status, datestr = [r.string for r in row]
if name != 'NAME' and status != 'STATUS' and datestr != 'DATE':
date = get_date_yearless(datestr)
try:
player = Player.get_by_name(name)
except Player.DoesNotExist:
print 'Couldn\'t find player %s' % name
