def GetSeason(self):
soup = Config.GetHTMLSoup(Config.URL_roster)
self.SeasonNumber = -1
for x in soup.find_all('option'):
val = int(x.get('value'))
sn = max(val, self.SeasonNumber)
return sn
def GetActivePlayers(self, season):
# If the player list is empty, go retrieve the lastest list
if (self.LastUpdateDateTime == time.gmtime(0)):
log.info('Reloading Players')
self.GetAllPlayers()
# Get the latest list of eligible players
url = COnfig.URL_roster
soup = Config.GetHTMLSoup(url)
x = soup.find_all(class_="table_1a")
def GetAllPlayers(self):
self.Players = []
url = Config.URL_player
soup = Config.GetHTMLSoup(url)
playerHTML = soup.find_all('option')
for index, x in enumerate(playerHTML):
if (index < 3):
continue
id_str = x.get('value')
player_name = x.get_text()
try:
player_id = int(id_str)
if (player_name != ''):
newPlayer = Player(player_id, player_name)
self.Players.append(newPlayer)
except:
log.warning('Failed to convert')
continue
self.LastUpdateDateTime = datetime.datetime.now()
self.Size = len(self.Players)
log.info('Found %d players' % (self.Size))
def GetStatsByPosition(self, position):
if (position == Positions.Center):
url = self.URL + '&filter_pos=center'
elif (position == Positions.Wing):
url = self.URL + '&filter_pos=wing'
elif (position == Positions.Defense):
url = self.URL + '&filter_pos=defense'
elif (position == Positions.Goalie):
url = self.URL + '&filter_pos=goalie'
else:
log.warning(
'Position entered that is not in the Config.Positions ENUM')
soup = Config.GetHTMLSoup(url)
tags = soup.find_all(class_="table_1")[0]
#loop through all tags
for index, tag in enumerate(tags):
#clean all variables
eligible = False
#skip all empty tags
if (tag != '\n'):
log.info('Tag #%s - Length = %s' % (index, len(tag)))
#only grab the first detailed contents and size
row = tag.contents[1].get_text().splitlines()
rowlen = len(row)
log.info('Row Length = %s' % (len(row)))
log.info(row)
#find their position
if (rowlen == 1 and 'Statistics' in row[0]):
#Get their position
pos = row[0].replace(' Statistics', '')
elif (rowlen >= 20 and 'Season' in row[2]):
log.info('Season Eligible')
eligible = True
season = row[2][7:]
SeasonType = 'Regular'
gp = row[5]
g = row[6]
a = row[7]
pim = row[9]
hits = row[10]
shots = row[11]
gwg = row[12]
fop = row[13]
pm = row[14]
corsi = row[15]
w = row[16]
l = row[17]
otl = row[18]
sal = row[20].replace('$', '').replace(',', '')
elif (rowlen >= 17 and 'Playoffs' in row[2]):
log.info('Playoff Eligible')
eligible = True
SeasonType = 'Playoff'
gp = row[4]
g = row[5]
a = row[6]
pim = row[7]
hits = row[8]
shots = row[9]
gwg = row[10]
fop = row[11]
pm = row[12]
corsi = row[13]
w = row[14]
l = row[15]
otl = row[16]
sal = sal
elif (rowlen >= 19 and 'International' in row[2]):
log.info('International Eligible')
eligible = True
SeasonType = 'International'
gp = row[4]
g = row[5]
a = row[6]
pim = row[7]
hits = row[8]
shots = row[9]
gwg = row[10]
fop = row[11]
pm = row[12]
corsi = row[13]
w = row[14]
l = row[15]
otl = row[16]
sal = sal
else:
log.info('Not Eligible')
if (eligible):
newSeasonStat = PlayerSeasonStat(self.PID, self.Name, pos,
season, SeasonType, gp, g,
a, pim, hits, shots, gwg,
fop, pm, corsi, w, l, otl,
sal)
row = pd.Series(newSeasonStat.ToList(), self.Columns)
self.Stats.loc[len(self.Stats)] = row
log.info(self.Stats)
def GetPlayerDetails(self):
log.info("Extracting data for: %s (%s)" % (self.Name, self.PID))
url_player = Config.GetPlayerURL(self.PID)
soup = Config.GetHTMLSoup(url_player)
return soup
