def weather_hourly(teamID,gametime): #gametime is formatted as 7:30 AM or 7:00 pm,
team_map=Ugen.mlb_map(4,8)
batter_map=Ugen.mlb_map(4,9)
batter_dir=batter_map[teamID]
zipcode=str(team_map[teamID])
try:
gametime_hour=convert_time_24(int(gametime.split()[0].split(":")[0]),int(gametime.split()[0].split(":")[1]),gametime.split()[1])
except:
print 'gametime %s is not in proper format' % gametime
return None
data=weather_response('hourly',zipcode)
weather_dict={}
tzone=data['hourly_forecast'][1]['FCTTIME']['pretty'].split()[2]
gametime_hour=convert_timezones(gametime_hour,'ET',tzone)
temp_list,hum_list,wspd_list,wdir_list,pop_list=([] for i in range(5))
i=1 #counter to make sure you don't take forecast from the next day
for e in data['hourly_forecast']: #each e is one hours time
forecast_hour=e['FCTTIME']['hour_padded']
if forecast_hour[0]=='0' and forecast_hour[1]=='0':
forecast_hour=forecast_hour.replace("0","",1)
elif forecast_hour[0]=='0':
forecast_hour=forecast_hour.replace("0","")
if i<4 and (forecast_hour==str(gametime_hour) or forecast_hour==str(int(gametime_hour)+1) or forecast_hour==str(int(gametime_hour)+2)):
try:
temp_list.append(float(e['temp']['english'])) #temp in degF
except:
pass
try:
hum_list.append(float(e['humidity']))
except:
pass
try:
pop_list.append(float(e['pop']))
except:
pass
wdir=e['wdir']['degrees']
try:
if int(wdir)<180: #wdir is the direction the wind is blowing from
wdir_list.append(abs(float((int(wdir)+180)-batter_dir))) #right now do abs value
else:
wdir_list.append(abs(float((int(wdir)-180)-batter_dir)))
except:
pass
try:
wspd_list.append(float(e['wspd']['english'])) #mph or wspdm for metric
except:
pass
i=i+1
weather_dict['temp']=round(np.mean(temp_list),2)
weather_dict['humidity']=round(np.mean(hum_list),2)
weather_dict['wind']={}
weather_dict['wind']['wind_dir']=round(np.mean(wdir_list),2)
weather_dict['wind']['wind_speed']=round(np.mean(wspd_list),2)
weather_dict['pop']=round(np.mean(pop_list),2)
return weather_dict
def weather_hist(teamID, weather_date, gametime):
weather_dict = {}
team_map = Ugen.mlb_map(4, 8)
batter_map = Ugen.mlb_map(4, 9)
zipcode = str(team_map[teamID])
batter_dir = float(batter_map[teamID])
try:
gametime_hour = convert_time_24(
int(gametime.split()[0].split(":")[0]), int(gametime.split()[0].split(":")[1]), gametime.split()[1]
)
except:
print "gametime %s is not in proper format" % gametime
return None
data = weather_response("history" + "_" + weather_date.replace("-", ""), zipcode)
tzone = data["history"]["observations"][1]["date"]["pretty"].split()[2]
gametime_hour = convert_timezones(gametime_hour, "ET", tzone)
temp_list, hum_list, wspd_list, wdir_list = ([] for i in range(4))
for e in data["history"]["observations"]:
forecast_hour = e["date"]["hour"]
if forecast_hour[0] == "0" and forecast_hour[1] == "0":
forecast_hour = forecast_hour.replace("0", "", 1)
elif forecast_hour[0] == "0":
forecast_hour = forecast_hour.replace("0", "")
if (
forecast_hour == str(gametime_hour)
or forecast_hour == str(int(gametime_hour) + 1)
or forecast_hour == str(int(gametime_hour) + 2)
):
try:
temp_list.append(float(e["tempi"])) # temp in degF
except:
pass
try:
hum_list.append(float(e["hum"]))
except:
pass
wdir = e["wdird"]
try:
if int(wdir) < 180:
wdir_list.append(abs(float((int(wdir) + 180) - batter_dir))) # right now do abs value
else:
wdir_list.append(abs(float((int(wdir) - 180) - batter_dir)))
except:
pass
try:
wspd_list.append(float(e["wspdi"])) # mph or wspdm for metric
except:
pass
weather_dict["temp"] = round(np.mean(temp_list), 2)
weather_dict["humidity"] = round(np.mean(hum_list), 2)
weather_dict["wind"] = {}
weather_dict["wind"]["wind_dir"] = round(np.mean(wdir_list), 2)
weather_dict["wind"]["wind_speed"] = round(np.mean(wspd_list), 2)
return weather_dict
def weather_hist(teamID,weather_date,gametime):
weather_dict={}
team_map=Ugen.mlb_map(4,8)
batter_map=Ugen.mlb_map(4,9)
zipcode=str(team_map[teamID])
batter_dir=float(batter_map[teamID])
try:
gametime_hour=convert_time_24(int(gametime.split()[0].split(":")[0]),int(gametime.split()[0].split(":")[1]),gametime.split()[1])
except:
print 'gametime %s is not in proper format' % gametime
return None
data=weather_response('history'+'_'+weather_date.replace('-',''),zipcode)
tzone=data['history']['observations'][1]['date']['pretty'].split()[2]
gametime_hour=convert_timezones(gametime_hour,"ET",tzone)
temp_list,hum_list,wspd_list,wdir_list=([] for i in range(4))
for e in data['history']['observations']:
forecast_hour=e['date']['hour']
if forecast_hour[0]=='0' and forecast_hour[1]=='0':
forecast_hour=forecast_hour.replace("0","",1)
elif forecast_hour[0]=='0':
forecast_hour=forecast_hour.replace("0","")
if forecast_hour==str(gametime_hour) or forecast_hour==str(int(gametime_hour)+1) or forecast_hour==str(int(gametime_hour)+2):
try:
temp_list.append(float(e['tempi'])) #temp in degF
except:
pass
try:
hum_list.append(float(e['hum']))
except:
pass
wdir=e['wdird']
try:
if int(wdir)<180:
wdir_list.append(abs(float((int(wdir)+180)-batter_dir))) #right now do abs value
else:
wdir_list.append(abs(float((int(wdir)-180)-batter_dir)))
except:
pass
try:
wspd_list.append(float(e['wspdi'])) #mph or wspdm for metric
except:
pass
weather_dict['temp']=round(np.mean(temp_list),2)
weather_dict['humidity']=round(np.mean(hum_list),2)
weather_dict['wind']={}
weather_dict['wind']['wind_dir']=round(np.mean(wdir_list),2)
weather_dict['wind']['wind_speed']=round(np.mean(wspd_list),2)
return weather_dict
def vegas_odds_sportsbook(date): #takes in date YYYY-MM-DD or YYYYMMDD
print 'getting odds for %s' % date
url='http://www.sportsbookreview.com/betting-odds/mlb-baseball/merged/?date='+date.replace("-","")
content=urllib2.urlopen(url).read()
soup=BeautifulSoup(content)
team_map=Ugen.mlb_map(10,4)
table=soup.find("div",{"class":"eventGroup class-mlb-baseball"})
odds_dict={}
for row in soup.findAll("div",{"class":"event-holder holder-complete"}):
row_data=row.find("div",{"class":"eventLine odd status-complete "})
if not row_data:
row_data=row.find("div",{"class":"eventLine status-complete "})
hyperlink=row_data.find("meta",{"itemprop":"url"})
matchup_dict=get_gameday_odds(hyperlink.get('content'),team_map)
for team,odds in matchup_dict.iteritems():
if team not in odds_dict.keys(): #Ian: this is here so we don't take the second game of a double header (FD usually does first)
odds_dict[team]=odds
return odds_dict
def pitcher_season_averages(self,hist_data,home_away,year):
team_map=Ugen.mlb_map(11,4)
season_avg={}
if not home_away and stat=='innings_pitched':
try:
season_avg['K9']=numpy.mean([float(SO/IP*9) for SO,date,IP in zip(hist_data['strike_outs'],hist_data['Date'],hist_data['innings_pitched']) if str(date).split("-")[0]==year and IP>3])
season_avg['IP']=numpy.mean([IP for AT,HT,IP,date in zip(hist_data['away_starting_lineup'],hist_data['home_starting_lineup'], \
hist_data['innings_pitched'],hist_data['Date']) if str(date).split("-")[0]==year and (player in AT or player in HT)])
except:
season_avg['IP']=numpy.mean([IP for IP,date in zip(hist_data['innings_pitched'],hist_data['Date']) if str(date).split("-")[0]=='2015'and IP>3])
elif home_away=='home':
try:
season_avg['K9']=numpy.mean([float(SO/IP*9) for SO,date,IP,team,home_team,away_team \
in zip(hist_data['strike_outs'],hist_data['Date'],hist_data['innings_pitched'],hist_data['Team'],hist_data['home_team'],hist_data['away_team']) \
if str(date).split("-")[0]==year and IP>3 and team==team_map[home_team]])
season_avg['IP']=numpy.mean([IP for AT,HT,IP,date,team,home_team,away_team in zip(hist_data['away_starting_lineup'],hist_data['home_starting_lineup'], \
hist_data['innings_pitched'],hist_data['Date'],hist_data['Team'],hist_data['home_team'],hist_data['away_team']) \
if str(date).split("-")[0]==year and (player in AT or player in HT) and team==team_map[home_team]])
season_avg['ER']=numpy.mean([ER for AT,HT,ER,date,team,home_team,away_team in zip(hist_data['away_starting_lineup'],hist_data['home_starting_lineup'], \
hist_data['earned_runs'],hist_data['Date'],hist_data['Team'],hist_data['home_team'],hist_data['away_team']) \
if str(date).split("-")[0]==year and (player in AT or player in HT) and team==team_map[home_team]])
except:
season_avg['IP']=numpy.mean([IP for team,home_team,away_team,IP,date in zip(hist_data['Team'],hist_data['home_team'],hist_data['away_team'],\
hist_data['innings_pitched'],hist_data['Date']) if str(date).split("-")[0]==year and IP>3 and team==team_map[home_team]])
season_avg['ER']=numpy.mean([ER for team,home_team,away_team,IP,date,ER in zip(hist_data['Team'],hist_data['home_team'],hist_data['away_team'],\
hist_data['innings_pitched'],hist_data['Date'],hist_data['earned_runs']) if str(date).split("-")[0]==year and IP>3 and team==team_map[home_team]])
elif home_away=='away':
try:
season_avg['K9']=numpy.mean([float(SO/IP*9) for SO,date,IP,team,home_team,away_team \
in zip(hist_data['strike_outs'],hist_data['Date'],hist_data['innings_pitched'],hist_data['Team'],hist_data['home_team'],hist_data['away_team']) \
if str(date).split("-")[0]==year and IP>3 and team==team_map[home_team]])
season_avg['IP']=numpy.mean([IP for AT,HT,IP,date,team,home_team,away_team in zip(hist_data['away_starting_lineup'],hist_data['home_starting_lineup'], \
hist_data['innings_pitched'],hist_data['Date'],hist_data['Team'],hist_data['home_team'],hist_data['away_team']) \
if str(date).split("-")[0]==year and (player in AT or player in HT) and team==team_map[away_team]])
season_avg['ER']=numpy.mean([ER for AT,HT,ER,date,team,home_team,away_team in zip(hist_data['away_starting_lineup'],hist_data['home_starting_lineup'], \
hist_data['earned_runs'],hist_data['Date'],hist_data['Team'],hist_data['home_team'],hist_data['away_team']) \
if str(date).split("-")[0]==year and (player in AT or player in HT) and team==team_map[away_team]])
except:
season_avg['IP']=numpy.mean([IP for team,home_team,away_team,IP,date in zip(hist_data['Team'],hist_data['home_team'],hist_data['away_team'],\
hist_data['innings_pitched'],hist_data['Date']) if str(date).split("-")[0]==year and IP>3 and team==team_map[away_team]])
season_avg['ER']=numpy.mean([ER for team,home_team,away_team,IP,date,ER in zip(hist_data['Team'],hist_data['home_team'],hist_data['away_team'],\
hist_data['innings_pitched'],hist_data['Date'],hist_data['earned_runs']) if str(date).split("-")[0]==year and IP>3 and team==team_map[away_team]])
return season_avg
def vegas_odds_sportsbook(date): #takes in date YYYY-MM-DD or YYYYMMDD
print 'getting odds for %s' % date
url = 'http://www.sportsbookreview.com/betting-odds/mlb-baseball/merged/?date=' + date.replace(
"-", "")
content = urllib2.urlopen(url).read()
soup = BeautifulSoup(content)
team_map = Ugen.mlb_map(10, 4)
table = soup.find("div", {"class": "eventGroup class-mlb-baseball"})
odds_dict = {}
for row in soup.findAll("div", {"class": "event-holder holder-complete"}):
row_data = row.find("div", {"class": "eventLine odd status-complete "})
if not row_data:
row_data = row.find("div",
{"class": "eventLine status-complete "})
hyperlink = row_data.find("meta", {"itemprop": "url"})
matchup_dict = get_gameday_odds(hyperlink.get('content'), team_map)
for team, odds in matchup_dict.iteritems():
if team not in odds_dict.keys(
): #Ian: this is here so we don't take the second game of a double header (FD usually does first)
odds_dict[team] = odds
return odds_dict
def mlb_starting_lineups(date=time.strftime("%Y-%m-%d")): #take date as string 'YYYY-MM-DD'.
print date
url='http://www.baseballpress.com/lineups/'+date
content=urllib2.urlopen(url).read()
soup=BeautifulSoup(content,"html.parser") #Ian: added this html.parser option based on suggestion from OSX terminal...may not be necessary on windows??
team_map=Ugen.mlb_map(6,4)
player_map=Ugen.mlb_map(2,0)
team_list,pitcher_list,lineups_list,gametime_list,weather_list,pitcher_arm_list,player_arm_list=([] for i in range(7))
teamid_dict={}
playerid_dict={}
gametime_list=[event_date.text for event_date in soup.findAll("div",{"class":"game-time"})]
gametime_list=[x for pair in zip(gametime_list,gametime_list) for x in pair] #duplicate elements in list
for forecast in soup.findAll("a",{"target":"forecast"}):
forecast_string=filter(lambda x: x in string.printable, forecast.text).split('Forecast: ')[1].split(' PoP')[0].replace(" ","-").replace("--","-").replace('F','degF')
if len(forecast_string.split('-'))==4:
forecast_string=forecast_string.split('-')[0]+'-'+forecast_string.split('-')[1]+' '+forecast_string.split('-')[2]+'-'+forecast_string.split('-')[3]
weather_list.append(forecast_string)
weather_list.append(forecast_string)
for team in soup.findAll("div",{"class":"team-data"}):
team_name=team.find("div",{"class":"team-name"}).get_text()
pitcher_name=team.find("a",{"class":"player-link"}).get_text()
pitcher_arm=team.find('div',{"class":"text"}).get_text().split('(')[1].split(')')[0]
if team_name in team_map: #Ian: Check if team name has been listed
team_name=team_map[team_name]
if pitcher_name in player_map:
pitcher_name=player_map[pitcher_name]
team_list.append(team_name)
pitcher_list.append(pitcher_name.replace("'","")+'_'+'pitcher')
pitcher_arm_list.append(pitcher_arm)
for table in soup.findAll("div",{"class":"cssDialog clearfix"}):
table_string=table.get_text()
home_lineup,away_lineup,home_lineup_arms,away_lineup_arms=([] for i in range(4))
if table_string.count("9. ")==2: #Ian: rethink; checks that both teams lineups have been listed. What if only one has been..
for j in range(1,10):
name_list_raw=table_string[table_string.find(str(j)+". ")+3:].split(" (")
player=name_list_raw[0]
home_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player=player_map[player]
home_lineup.append(player.replace("'","")+'_'+'batter')
name_list_raw=table_string[table_string.find((str(j)+". "),table_string.find(str(j)+". ")+3)+3:].split(" (")
player=name_list_raw[0]
away_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player=player_map[player]
away_lineup.append(player.replace("'","")+'_'+'batter')
lineups_list.append(home_lineup)
lineups_list.append(away_lineup)
player_arm_list.append(home_lineup_arms)
player_arm_list.append(away_lineup_arms)
else:
lineups_list.append(['no home_lineup listed'])
lineups_list.append(['no away_lineup listed'])
player_arm_list.append(['no hitting style listed'])
player_arm_list.append(['no hitting style listed'])
opponent_dict={team_list[i]:{'home_team':team_list[i],'opponent':team_list[i-1]} for i in range(0,len(team_list)) if i%2 !=0}
opponent_dict.update({team_list[i]:{'home_team':team_list[i+1],'opponent':team_list[i+1]} for i in range(0,len(team_list)) if team_list[i] not in opponent_dict})
teamid_dict={}
for i in range(0,len(lineups_list)):
batting_order=range(1,len(lineups_list[i])+1)
lineups_list[i].append(pitcher_list[i])
player_arm_list[i].append(pitcher_arm_list[i])
player_arm_dict={player:{'arm':arm,'batting_order':order} for player,arm,order in zip(lineups_list[i],player_arm_list[i],batting_order)}
teamid_dict.update({team_list[i]:{'start_time':gametime_list[i],'date':date,'lineup':player_arm_dict,'home_teamid':opponent_dict[team_list[i]]['home_team'], \
'opponent':opponent_dict[team_list[i]]['opponent'],'weather_forecast':weather_list[i]}})
playerid_dict={player:{'start_time':teamid_dict[team_id]['start_time'],'weather_forecast':teamid_dict[team_id]['weather_forecast'], \
'teamid':team_id,'opposing_lineup': teamid_dict[teamid_dict[team_id]['opponent']]['lineup'],'arm':teamid_dict[team_id]['lineup'][player]['arm'], \
'batting_order':teamid_dict[team_id]['lineup'][player]['batting_order'],'home_teamid':teamid_dict[team_id]['home_teamid']} \
for team_id in teamid_dict for player in teamid_dict[team_id]['lineup']}
return teamid_dict,playerid_dict
def batter_lineup_stats(self,date,lineup_data,player_arm): #Ian: needs refactoring!!
lineup_stats_dict={}
hist_lineup_strikeout_rate,hist_lineup_ops,hist_lineup_slg,strikeout_PAs_list,ops_PAs_list,slg_PAs_list=([] for i in range(6))
team_map=Ugen.mlb_map(11,4)
for player in lineup_data[0]:
player_strikeout_rate_splits,player_ops_splits,player_slg_splits=([] for i in range(3))
if player.split("_")[1]!='pitcher':
player_data=self.get_db_gamedata(player.split("_")[0],"20130301",Ugen.previous_day(str(date)).replace("-","")) #may need to play with how much data you use to get batter's K avg
try:
player_data=player_data[player]
except KeyError:
print ('player %s not in db, needs new player map' % player)
rw=2
map_list=[]
while Cell("Output",rw,7).value:
map_list.append(Cell("Output",rw,7).value)
rw+=1
if player not in map_list:
Cell("Output",rw,7).value=player
Cell("Output",rw,8).value=Ugen.previous_day(str(date)).replace("-","")
continue
for indx,event in enumerate(player_data['event_id']):
try:
reverse_index = len(player_data['event_id'])-indx-1
team=player_data['Team'][reverse_index]
home_away=lineup_data[1] #This tells us whether the batter facing the current pitcher is at home or away. We want his K% splits for H/A
home_team=team_map[player_data['home_team'][reverse_index]]
try:
if team==home_team:
matchup='home'
opposing_lineup=ast.literal_eval(player_data['away_starting_lineup'][reverse_index])
else:
matchup='away'
opposing_lineup=ast.literal_eval(player_data['home_starting_lineup'][reverse_index])
opposing_pitcher_hand=[opposing_player_data['arm'] for opposing_player,opposing_player_data in opposing_lineup.iteritems() if 'pitcher' in opposing_player][0]
if matchup==home_away and (player_arm==opposing_pitcher_hand or player_arm=='S'): #
strike_outs=player_data['strike_outs'][reverse_index]
plate_appearances=player_data['plate_appearances'][reverse_index]
if plate_appearances>0:
strikeout_rate=float(strike_outs/plate_appearances)
if numpy.isnan(strikeout_rate):
print ("isnan error for calculated strikeout rate %s,%s" % (player,player_data['GameID'][reverse_index]))
player_strikeout_rate_splits.append(0.200)
else:
player_strikeout_rate_splits.append(strikeout_rate)
if player_data['ops'][reverse_index]>=0:
player_ops_splits.append(player_data['ops'][reverse_index])
else:
player_ops_splits.append(0)
if player_data['slg'][reverse_index]>=0:
player_slg_splits.append(player_data['slg'][reverse_index])
else:
player_slg_splits.append(0)
except ValueError: #This is when there is no starting lineups data (usually)
#print ('Value error %s %s' %(player_data["Date"][reverse_index],player))
pass
except IndexError:
print ('index error')
break
batting_order=lineup_data[0][player]['batting_order']
if len(player_strikeout_rate_splits)>2: #say we need 3 min values to incorporate the players strikeout rate into feature
hist_lineup_strikeout_rate.append(float(self.avg_plate_appearances[batting_order]*numpy.mean(player_strikeout_rate_splits)))
strikeout_PAs_list.append(self.avg_plate_appearances[batting_order])
if len(player_ops_splits)>2:
hist_lineup_ops.append(float(self.avg_plate_appearances[batting_order]*numpy.mean(player_ops_splits)))
ops_PAs_list.append(self.avg_plate_appearances[batting_order])
if len(player_slg_splits)>2:
hist_lineup_slg.append(float(self.avg_plate_appearances[batting_order]*numpy.mean(player_slg_splits)))
slg_PAs_list.append(self.avg_plate_appearances[batting_order])
lineup_stats_dict['strikeout_rate']=numpy.sum(hist_lineup_strikeout_rate)/numpy.sum(strikeout_PAs_list)
lineup_stats_dict['ops']=numpy.sum(hist_lineup_ops)/numpy.sum(ops_PAs_list)
lineup_stats_dict['slg']=numpy.sum(hist_lineup_slg)/numpy.sum(slg_PAs_list)
# print (lineup_stats_dict)
# os.system('pause')
return lineup_stats_dict
def mlb_starting_lineups(date=time.strftime("%Y-%m-%d")): #take date as string 'YYYY-MM-DD'. [desperately] Needs refactoring.
print date
url='http://www.baseballpress.com/lineups/'+date
content=urllib2.urlopen(url).read()
soup=BeautifulSoup(content)
team_map=Ugen.mlb_map(6,4)
player_map=Ugen.mlb_map(2,0)
# print 'player/team maps complete'
team_list,pitcher_list,lineups_list,gametime_list,weather_list,pitcher_arm_list,player_arm_list=([] for i in range(7))
teamid_dict={}
playerid_dict={}
for event_date in soup.findAll("div",{"class":"game-time"}):
gametime_list.append(event_date.text)
for forecast in soup.findAll("a",{"target":"forecast"}):
forecast_string=filter(lambda x: x in string.printable, forecast.text).split('Forecast: ')[1].split(' PoP')[0].replace(" ","-").replace("--","-").replace('F','degF')
if len(forecast_string.split('-'))==4:
forecast_string=forecast_string.split('-')[0]+'-'+forecast_string.split('-')[1]+' '+forecast_string.split('-')[2]+'-'+forecast_string.split('-')[3]
weather_list.append(forecast_string)
for team in soup.findAll("div",{"class":"team-data"}):
try:
team_name=team.find("div",{"class":"team-name"}).get_text()
except:
team_name=''
try:
pitcher_name=team.find("a",{"class":"player-link"}).get_text()
except:
pitcher_name=''
try:
pitcher_arm=team.find('div',{"class":"text"}).get_text().split('(')[1].split(')')[0]
except:
pitcher_arm=''
if team_name in team_map: #Ian: Check if team name has been listed
team_list.append(team_map[team_name])
else:
team_list.append(team_name)
if pitcher_name in player_map:
pitcher_name=player_map[pitcher_name]
pitcher_list.append(pitcher_name.replace("'","")+'_'+'pitcher')
pitcher_arm_list.append(pitcher_arm)
for table in soup.findAll("div",{"class":"cssDialog clearfix"}):
table_string=table.get_text()
home_lineup,away_lineup,home_lineup_arms,away_lineup_arms=([] for i in range(4))
if table_string.count("9. ")==2: #Ian: rethink-checks that both teams lineups have been listed. What if only one has been..
for j in range(1,10):
name_list_raw=table_string[table_string.find(str(j)+". ")+3:].split(" (")
player=name_list_raw[0]
home_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player=player_map[player]
home_lineup.append(player.replace("'","")+'_'+'batter')
name_list_raw=table_string[table_string.find((str(j)+". "),table_string.find(str(j)+". ")+3)+3:].split(" (")
player=name_list_raw[0]
away_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player=player_map[player]
away_lineup.append(player.replace("'","")+'_'+'batter')
lineups_list.append(home_lineup)
lineups_list.append(away_lineup)
player_arm_list.append(home_lineup_arms)
player_arm_list.append(away_lineup_arms)
else:
lineups_list.append(['no home_lineup listed'])
lineups_list.append(['no away_lineup listed'])
player_arm_list.append(['no hitting style listed'])
player_arm_list.append(['no hitting style listed'])
i=j=0
while i<len(lineups_list):
#if pitcher_list[i] not in lineups_list[i]
lineups_list[i].append(pitcher_list[i])
player_arm_list[i].append(pitcher_arm_list[i])
teamid_dict[team_list[i]]={}
teamid_dict[team_list[i]]['start_time']=gametime_list[j]
teamid_dict[team_list[i]]['weather_forecast']=weather_list[j]
teamid_dict[team_list[i]]['date']=date
player_arm_dict={}
z=1
for player,arm in zip(lineups_list[i],player_arm_list[i]):
player_arm_dict[player]={}
player_arm_dict[player]['arm']=arm
player_arm_dict[player]['batting_order']=str(z)
z=z+1
teamid_dict[team_list[i]]['lineup']=player_arm_dict
if i%2 !=0:
j=j+1
teamid_dict[team_list[i]]['home_teamid']=team_list[i]
teamid_dict[team_list[i]]['opponent']=team_list[i-1]
else:
teamid_dict[team_list[i]]['home_teamid']=team_list[i+1]
teamid_dict[team_list[i]]['opponent']=team_list[i+1]
i=i+1
i=j=0
while i<len(lineups_list):
z=1
for player,arm in zip(lineups_list[i],player_arm_list[i]):
playerid_dict[player]={}
playerid_dict[player]['start_time']=gametime_list[j]
playerid_dict[player]['weather_forecast']=weather_list[j]
playerid_dict[player]['teamid']=team_list[i]
playerid_dict[player]['arm']=arm
playerid_dict[player]['home_teamid']=teamid_dict[playerid_dict[player]['teamid']]['home_teamid']
playerid_dict[player]['opposing_lineup']=teamid_dict[teamid_dict[playerid_dict[player]['teamid']]['opponent']]['lineup']
playerid_dict[player]['batting_order']=str(z)
z=z+1
if i%2 !=0:
j=j+1
i=i+1
return teamid_dict,playerid_dict
def weather_hourly(teamID, gametime): # gametime is formatted as 7:30 AM or 7:00 pm,
team_map = Ugen.mlb_map(4, 8)
batter_map = Ugen.mlb_map(4, 9)
batter_dir = batter_map[teamID]
zipcode = str(team_map[teamID])
try:
gametime_hour = convert_time_24(
int(gametime.split()[0].split(":")[0]), int(gametime.split()[0].split(":")[1]), gametime.split()[1]
)
except:
print "gametime %s is not in proper format" % gametime
return None
data = weather_response("hourly", zipcode)
weather_dict = {}
tzone = data["hourly_forecast"][1]["FCTTIME"]["pretty"].split()[2]
gametime_hour = convert_timezones(gametime_hour, "ET", tzone)
temp_list, hum_list, wspd_list, wdir_list, pop_list = ([] for i in range(5))
i = 1 # counter to make sure you don't take forecast from the next day
for e in data["hourly_forecast"]: # each e is one hours time
forecast_hour = e["FCTTIME"]["hour_padded"]
if forecast_hour[0] == "0" and forecast_hour[1] == "0":
forecast_hour = forecast_hour.replace("0", "", 1)
elif forecast_hour[0] == "0":
forecast_hour = forecast_hour.replace("0", "")
if i < 4 and (
forecast_hour == str(gametime_hour)
or forecast_hour == str(int(gametime_hour) + 1)
or forecast_hour == str(int(gametime_hour) + 2)
):
try:
temp_list.append(float(e["temp"]["english"])) # temp in degF
except:
pass
try:
hum_list.append(float(e["humidity"]))
except:
pass
try:
pop_list.append(float(e["pop"]))
except:
pass
wdir = e["wdir"]["degrees"]
try:
if int(wdir) < 180: # wdir is the direction the wind is blowing from
wdir_list.append(abs(float((int(wdir) + 180) - batter_dir))) # right now do abs value
else:
wdir_list.append(abs(float((int(wdir) - 180) - batter_dir)))
except:
pass
try:
wspd_list.append(float(e["wspd"]["english"])) # mph or wspdm for metric
except:
pass
i = i + 1
weather_dict["temp"] = round(np.mean(temp_list), 2)
weather_dict["humidity"] = round(np.mean(hum_list), 2)
weather_dict["wind"] = {}
weather_dict["wind"]["wind_dir"] = round(np.mean(wdir_list), 2)
weather_dict["wind"]["wind_speed"] = round(np.mean(wspd_list), 2)
weather_dict["pop"] = round(np.mean(pop_list), 2)
return weather_dict
def mlb_starting_lineups(date=time.strftime("%Y-%m-%d")): #take date as string 'YYYY-MM-DD'.
print date
url='http://www.baseballpress.com/lineups/'+date
content=urllib2.urlopen(url).read()
soup=BeautifulSoup(content,"html.parser") #Ian: added this html.parser option based on suggestion from OSX terminal...may not be necessary on windows??
team_map=Ugen.mlb_map(6,4)
player_map=Ugen.mlb_map(2,0)
team_list,pitcher_list,lineups_list,gametime_list,weather_list,pitcher_arm_list,player_arm_list=([] for i in range(7))
teamid_dict={}
playerid_dict={}
gametime_list=[event_date.text for event_date in soup.findAll("div",{"class":"game-time"})]
gametime_list=[x for pair in zip(gametime_list,gametime_list) for x in pair] #duplicate elements in list
for forecast in soup.findAll("a",{"target":"forecast"}):
forecast_string=filter(lambda x: x in string.printable, forecast.text).split('Forecast: ')[1].split(' PoP')[0].replace(" ","-").replace("--","-").replace('F','degF')
if len(forecast_string.split('-'))==4:
forecast_string=forecast_string.split('-')[0]+'-'+forecast_string.split('-')[1]+' '+forecast_string.split('-')[2]+'-'+forecast_string.split('-')[3]
weather_list.append(forecast_string)
weather_list.append(forecast_string)
for team in soup.findAll("div",{"class":"team-data"}):
team_name=team.find("div",{"class":"team-name"}).get_text()
pitcher_name=team.find("a",{"class":"player-link"}).get_text()
pitcher_arm=team.find('div',{"class":"text"}).get_text().split('(')[1].split(')')[0]
if team_name in team_map: #Ian: Check if team name has been listed
team_name=team_map[team_name]
if pitcher_name in player_map:
pitcher_name=player_map[pitcher_name]
team_list.append(team_name)
pitcher_list.append(pitcher_name.replace("'","")+'_'+'pitcher')
pitcher_arm_list.append(pitcher_arm)
for table in soup.findAll("div",{"class":"cssDialog clearfix"}):
table_string=table.get_text()
home_lineup,away_lineup,home_lineup_arms,away_lineup_arms=([] for i in range(4))
if table_string.count("9. ")==2: #Ian: rethink; checks that both teams lineups have been listed. What if only one has been..
for j in range(1,10):
name_list_raw=table_string[table_string.find(str(j)+". ")+3:].split(" (")
player=name_list_raw[0]
home_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player=player_map[player]
home_lineup.append(player.replace("'","")+'_'+'batter')
name_list_raw=table_string[table_string.find((str(j)+". "),table_string.find(str(j)+". ")+3)+3:].split(" (")
player=name_list_raw[0]
away_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player=player_map[player]
away_lineup.append(player.replace("'","")+'_'+'batter')
lineups_list.append(home_lineup)
lineups_list.append(away_lineup)
player_arm_list.append(home_lineup_arms)
player_arm_list.append(away_lineup_arms)
else:
lineups_list.append(['no home_lineup listed'])
lineups_list.append(['no away_lineup listed'])
player_arm_list.append(['no hitting style listed'])
player_arm_list.append(['no hitting style listed'])
opponent_dict={team_list[i]:{'home_team':team_list[i],'opponent':team_list[i-1]} for i in range(0,len(team_list)) if i%2 !=0}
opponent_dict.update({team_list[i]:{'home_team':team_list[i+1],'opponent':team_list[i+1]} for i in range(0,len(team_list)) if team_list[i] not in opponent_dict})
teamid_dict={}
for i in range(0,len(lineups_list)):
batting_order=range(1,len(lineups_list[i])+1)
lineups_list[i].append(pitcher_list[i])
player_arm_list[i].append(pitcher_arm_list[i])
player_arm_dict={player:{'arm':arm,'batting_order':order} for player,arm,order in zip(lineups_list[i],player_arm_list[i],batting_order)}
teamid_dict.update({team_list[i]:{'start_time':gametime_list[i],'date':date,'lineup':player_arm_dict,'home_teamid':opponent_dict[team_list[i]]['home_team'], \
'opponent':opponent_dict[team_list[i]]['opponent'],'weather_forecast':weather_list[i]}})
playerid_dict={player:{'start_time':teamid_dict[team_id]['start_time'],'weather_forecast':teamid_dict[team_id]['weather_forecast'], \
'teamid':team_id,'opposing_lineup': teamid_dict[teamid_dict[team_id]['opponent']]['lineup'],'arm':teamid_dict[team_id]['lineup'][player]['arm'], \
'batting_order':teamid_dict[team_id]['lineup'][player]['batting_order'],'home_teamid':teamid_dict[team_id]['home_teamid']} \
for team_id in teamid_dict for player in teamid_dict[team_id]['lineup']}
return teamid_dict,playerid_dict
def mlb_starting_lineups(
date=time.strftime("%Y-%m-%d")
): #take date as string 'YYYY-MM-DD'. [desperately] Needs refactoring.
print date
url = 'http://www.baseballpress.com/lineups/' + date
content = urllib2.urlopen(url).read()
soup = BeautifulSoup(content)
team_map = Ugen.mlb_map(6, 4)
player_map = Ugen.mlb_map(2, 0)
# print 'player/team maps complete'
team_list, pitcher_list, lineups_list, gametime_list, weather_list, pitcher_arm_list, player_arm_list = (
[] for i in range(7))
teamid_dict = {}
playerid_dict = {}
for event_date in soup.findAll("div", {"class": "game-time"}):
gametime_list.append(event_date.text)
for forecast in soup.findAll("a", {"target": "forecast"}):
forecast_string = filter(
lambda x: x in string.printable,
forecast.text).split('Forecast: ')[1].split(' PoP')[0].replace(
" ", "-").replace("--", "-").replace('F', 'degF')
if len(forecast_string.split('-')) == 4:
forecast_string = forecast_string.split(
'-')[0] + '-' + forecast_string.split(
'-')[1] + ' ' + forecast_string.split(
'-')[2] + '-' + forecast_string.split('-')[3]
weather_list.append(forecast_string)
for team in soup.findAll("div", {"class": "team-data"}):
try:
team_name = team.find("div", {"class": "team-name"}).get_text()
except:
team_name = ''
try:
pitcher_name = team.find("a", {"class": "player-link"}).get_text()
except:
pitcher_name = ''
try:
pitcher_arm = team.find('div', {
"class": "text"
}).get_text().split('(')[1].split(')')[0]
except:
pitcher_arm = ''
if team_name in team_map: #Ian: Check if team name has been listed
team_list.append(team_map[team_name])
else:
team_list.append(team_name)
if pitcher_name in player_map:
pitcher_name = player_map[pitcher_name]
pitcher_list.append(pitcher_name.replace("'", "") + '_' + 'pitcher')
pitcher_arm_list.append(pitcher_arm)
for table in soup.findAll("div", {"class": "cssDialog clearfix"}):
table_string = table.get_text()
home_lineup, away_lineup, home_lineup_arms, away_lineup_arms = (
[] for i in range(4))
if table_string.count(
"9. "
) == 2: #Ian: rethink-checks that both teams lineups have been listed. What if only one has been..
for j in range(1, 10):
name_list_raw = table_string[table_string.find(str(j) + ". ") +
3:].split(" (")
player = name_list_raw[0]
home_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player = player_map[player]
home_lineup.append(player.replace("'", "") + '_' + 'batter')
name_list_raw = table_string[table_string.find(
(str(j) + ". "),
table_string.find(str(j) + ". ") + 3) + 3:].split(" (")
player = name_list_raw[0]
away_lineup_arms.append(name_list_raw[1].split(')')[0])
if player in player_map:
player = player_map[player]
away_lineup.append(player.replace("'", "") + '_' + 'batter')
lineups_list.append(home_lineup)
lineups_list.append(away_lineup)
player_arm_list.append(home_lineup_arms)
player_arm_list.append(away_lineup_arms)
else:
lineups_list.append(['no home_lineup listed'])
lineups_list.append(['no away_lineup listed'])
player_arm_list.append(['no hitting style listed'])
player_arm_list.append(['no hitting style listed'])
i = j = 0
while i < len(lineups_list):
#if pitcher_list[i] not in lineups_list[i]
lineups_list[i].append(pitcher_list[i])
player_arm_list[i].append(pitcher_arm_list[i])
teamid_dict[team_list[i]] = {}
teamid_dict[team_list[i]]['start_time'] = gametime_list[j]
teamid_dict[team_list[i]]['weather_forecast'] = weather_list[j]
teamid_dict[team_list[i]]['date'] = date
player_arm_dict = {}
z = 1
for player, arm in zip(lineups_list[i], player_arm_list[i]):
player_arm_dict[player] = {}
player_arm_dict[player]['arm'] = arm
player_arm_dict[player]['batting_order'] = str(z)
z = z + 1
teamid_dict[team_list[i]]['lineup'] = player_arm_dict
if i % 2 != 0:
j = j + 1
teamid_dict[team_list[i]]['home_teamid'] = team_list[i]
teamid_dict[team_list[i]]['opponent'] = team_list[i - 1]
else:
teamid_dict[team_list[i]]['home_teamid'] = team_list[i + 1]
teamid_dict[team_list[i]]['opponent'] = team_list[i + 1]
i = i + 1
i = j = 0
while i < len(lineups_list):
z = 1
for player, arm in zip(lineups_list[i], player_arm_list[i]):
playerid_dict[player] = {}
playerid_dict[player]['start_time'] = gametime_list[j]
playerid_dict[player]['weather_forecast'] = weather_list[j]
playerid_dict[player]['teamid'] = team_list[i]
playerid_dict[player]['arm'] = arm
playerid_dict[player]['home_teamid'] = teamid_dict[
playerid_dict[player]['teamid']]['home_teamid']
playerid_dict[player]['opposing_lineup'] = teamid_dict[teamid_dict[
playerid_dict[player]['teamid']]['opponent']]['lineup']
playerid_dict[player]['batting_order'] = str(z)
z = z + 1
if i % 2 != 0:
j = j + 1
i = i + 1
return teamid_dict, playerid_dict
