def __init__(self, ids):
self.gameidlist = []
self.gameinseason = []
# self.teamname = input('What team? ')
# season = input('What season? ')
self.teamname = "Virginia Tech"
season = "2013"
url = (
"http://espn.go.com/college-football/team/schedule/_/id/"
+ ids[self.teamname][0]
+ "/year/"
+ str(season)
+ "/"
+ ids[self.teamname][1]
)
soup = muf.cleanhtml(url)
for link in soup.find_all("a"):
if link.get("href")[:14] == "/ncf/recap?id=":
# link should look like:
# http://espn.go.com/ncf/recap?id=292620259
self.gameidlist.append(link.get("href")[14:])
if link.get("href")[:17] == "/ncf/boxscore?id=":
# link should look like:
# http://espn.go.com/ncf/boxscore?id=262990259
self.gameidlist.append(link.get("href")[17:])
# open and parse for tuple of games, and return
for games in self.gameidlist:
self.gameinseason.append(gpbp.Game(games, self.teamname, season))
def __init__(self, gameid, teamname, season):
#http://espn.go.com/ncf/playbyplay?gameId=292620259&period=0
# url = 'http://espn.go.com/ncf/playbyplay?gameId=' + gameid + '&period=0'
#apparent update to format
#http://espn.go.com/college-football/playbyplay?gameId=242410259
url = 'http://espn.go.com/ncf/playbyplay?gameId=' + gameid
soup = muf.cleanhtml(url)
game_pbp = ""
self.teamname = ""
self.opponent = ""
self.season = season
#title from the ESPN page, which typically includes both team names in full, plus game date
# <title>Western Michigan Broncos vs. Virginia Tech Hokies - Play By Play - September 27, 2014 - ESPN</title>
self.game_title = ""
self.home = None
for title in soup.findAll("title"):
titletxt = title.get_text()
team1 = titletxt[:titletxt.find(' v')]
team2 = titletxt[titletxt.find('.')+2:titletxt.find(' -')]
#print(team1, " ", team2)
if all(word in team1 for word in teamname.split()):
self.teamname = team1
self.opponent = team2
self.home = False
elif all(word in team2 for word in teamname.split()):
self.teamname = team2
self.opponent = team1
self.home = True
self.game_result = ""
for game in soup.findAll("div", { "class" : "competitors" }):
for team in game.findAll("tr"):
for teams in team.findAll("td", { "class" : "team-name" }):
#print('teams: ', teams)
self.game_result += teams.get_text() + " "
for scores in team.find_all("td", { "class" : "final-score" }):
#print('scores: ', scores)
self.game_result += scores.get_text() + " "
#print('game_result: ', self.game_result)
#list of team and opponents names and abbreviation that are likely to be used
self.tmnmlist = [self.teamname]
self.oppnmlist = [self.opponent]
gameresultlist = self.game_result.split()
#print('gameresultlist: ', gameresultlist)
if len(gameresultlist)>4:
while len(gameresultlist)>4:
if not gameresultlist[0].isdigit() and not gameresultlist[1].isdigit():
gameresultlist[0] = gameresultlist[0] + " " + gameresultlist[1]
gameresultlist.remove(gameresultlist[1])
elif not gameresultlist[2].isdigit() and not gameresultlist[3].isdigit():
gameresultlist[2] = gameresultlist[2] + " " + gameresultlist[3]
gameresultlist.remove(gameresultlist[3])
if self.home:
if muf.is_abbrev(gameresultlist[2], self.teamname) and muf.is_abbrev(gameresultlist[0], self.opponent):
self.tmnmlist.append(gameresultlist[2])
self.oppnmlist.append(gameresultlist[0])
elif muf.is_abbrev(gameresultlist[2], self.opponent) and muf.is_abbrev(gameresultlist[0], self.teamname):
self.oppnmlist.append(gameresultlist[2])
self.tmnmlist.append(gameresultlist[0])
elif muf.is_abbrev(gameresultlist[2], self.teamname) and not muf.is_abbrev(gameresultlist[2], self.opponent):
self.tmnmlist.append(gameresultlist[2])
self.oppnmlist.append(gameresultlist[0])
elif muf.is_abbrev(gameresultlist[2], self.opponent) and not muf.is_abbrev(gameresultlist[2], self.teamname):
self.oppnmlist.append(gameresultlist[2])
self.tmnmlist.append(gameresultlist[0])
elif muf.is_abbrev(gameresultlist[0], self.teamname) and not muf.is_abbrev(gameresultlist[0], self.opponent):
self.tmnmlist.append(gameresultlist[0])
self.oppnmlist.append(gameresultlist[2])
elif muf.is_abbrev(gameresultlist[0], self.opponent) and not muf.is_abbrev(gameresultlist[0], self.teamname):
self.oppnmlist.append(gameresultlist[0])
self.tmnmlist.append(gameresultlist[2])
else:
print('could not make sense of abbrevs')
else:
if muf.is_abbrev(gameresultlist[2], self.opponent) and muf.is_abbrev(gameresultlist[0], self.teamname):
self.tmnmlist.append(gameresultlist[0])
self.oppnmlist.append(gameresultlist[2])
elif muf.is_abbrev(gameresultlist[2], self.teamname) and muf.is_abbrev(gameresultlist[0], self.opponent):
self.oppnmlist.append(gameresultlist[0])
self.tmnmlist.append(gameresultlist[2])
elif muf.is_abbrev(gameresultlist[2], self.opponent) and not muf.is_abbrev(gameresultlist[2], self.teamname):
self.tmnmlist.append(gameresultlist[0])
self.oppnmlist.append(gameresultlist[2])
elif muf.is_abbrev(gameresultlist[2], self.teamname) and not muf.is_abbrev(gameresultlist[2], self.opponent):
self.oppnmlist.append(gameresultlist[0])
self.tmnmlist.append(gameresultlist[2])
elif muf.is_abbrev(gameresultlist[0], self.opponent) and not muf.is_abbrev(gameresultlist[0], self.teamname):
self.tmnmlist.append(gameresultlist[2])
self.oppnmlist.append(gameresultlist[0])
elif muf.is_abbrev(gameresultlist[0], self.teamname) and not muf.is_abbrev(gameresultlist[0], self.opponent):
self.oppnmlist.append(gameresultlist[2])
self.tmnmlist.append(gameresultlist[0])
else:
print('could not make sense of abbrevs')
#pbps is a data container for each string from ESPNs playbyplay file
#which should include information about the quarter, drive, etc, in addition
#to the actual play data
self.pbps = []
#plays is a structure to store the actual play data, includeing all relevent
#data about the current state of the game
self.plays = []
self.ply = []
#here are my local variables for keeping track of game state information (ie quarter, drive, etc)
gameperiod = 1 #1,2,3,4,5,6,7,8... etc, with 5 representing a first OT period, and so on
drive = 1
possesion = self.teamname
gamestate = {'quarter': gameperiod, 'driveno': drive, 'teamname': self.tmnmlist, 'opponent': self.oppnmlist, 'season': self.season, 'home':self.home, 'offense': possesion,}
for gamestring in soup.findAll("div", { "id" : "gamepackage-play-by-play" }):
game_pbp = gamestring
for play in game_pbp.findAll("li"):
print("play: ", play)
self.pbps.append(play.get_text(separator=u'<>'))
if 'Quarter' in self.pbps[-1]:
if str(gameperiod) not in self.pbps[-1]:
for char in self.pbps[-1]:
if char.isdigit():
gameperiod = int(char)
print(self.pbps[-1], " ", gameperiod)
elif 'OT ' in self.pbps[-1]:
gameperiod = self.isDigitOT(self.pbps[-1])
print(self.pbps[-1], " ", gameperiod)
#if quarter or drive info, store to local variable, if play info, store info to plays, including quarter and drive info
#play by play entries that still need to be dealt with
elif any(word in self.pbps[-1] for word in ['coin toss', 'toss', 'kick', 'attempt failed', 'extra point', 'Extra Point', 'Extra point', 'Conversion', 'conversion', 'punt', 'FG', 'field goal', 'Field Goal', 'Penalty', 'PENALTY', 'penalty', 'Fumble']):
pass
#play by play entries I don't yet have plans for
elif any(word in self.pbps[-1] for word in ['DRIVE', 'ball on', 'End']):
pass
#timeouts contain game clock info
elif any(word in self.pbps[-1] for word in ['Timeout', 'timeout']):
pass
elif any(word in self.pbps[-1] for word in ['rush ', 'run ', 'Run ', 'sacked ', 'scramble ']):
currentplay = plys.Rushing(self.pbps[-1], gamestate)
self.plays.append(currentplay.getplayinfo())
self.ply.append(currentplay)
elif any(word in self.pbps[-1] for word in ['pass ', 'incompletion ', 'interception ', 'Interception ']):
currentplay = plys.Passing(self.pbps[-1], gamestate)
self.plays.append(currentplay.getplayinfo())
elif any(word in self.pbps[-1] for word in ['SAFETY', 'Safety', 'safety']):
pass
elif any(word in self.pbps[-1] for word in [':']):
#contains the team currently with the ball
possesion = self.pbps[-1].split()
while len(possesion)>1:
if not possesion[0][0].isdigit() and not possesion[1][0].isdigit():
if possesion[1] == 'at':
for n in range(len(possesion)-1):
possesion.remove(possesion[1])
else:
possesion[0] = possesion[0] + " " + possesion[1]
possesion.remove(possesion[1])
else:
possesion.remove(possesion[1])
#likely also containing the best game clock info for the game
pass
elif any(word in self.pbps[-1] for word in ['at','and']):
#these might be empty plays due to inconsistencies and errors in espns play by play
#may need to add some checks for this?
pass
else:
#pass
print("these are unhandled plays: ", self.pbps[-1])
gamestate['quarter'] = gameperiod
gamestate['driveno'] = drive
gamestate['offense'] = possesion
