def ProcessBRPage(filename, con):
dateStr = filename.split('_')[-1]
gameDate = date(int(dateStr[:4]),int(dateStr[4:6]),int(dateStr[6:8]))
html = open(filename).read().decode('utf-8').replace('·','*')
pbp = PlayByPlay()
pbp.inning = ''
playNum = 1
playInd = 0
#Game Stuff
b = GameTeamsParser()
b.feed(html)
pbp.hAbb = ConvertTeamAbb('BR',b.homeTeamAbb)
pbp.aAbb = ConvertTeamAbb('BR',b.awayTeamAbb)
pbp.hTeam = GetTeamfromAbb(pbp.hAbb, con)
pbp.aTeam = GetTeamfromAbb(pbp.aAbb, con)
currentGame = Game(pbp.hTeam, pbp.aTeam, gameDate, con)
b = GameTimeParser()
b.feed(html)
currentGame.time = b.time
lengthStr = b.gamelen.split(':')
currentGame.gameLength += int(lengthStr[0]) * 60 + int(lengthStr[1])
b = GameWeatherParser()
b.feed(html)
currentGame.temp = int(re.search(': -?([0-9]{2}) F', b.weather).group(1))
currentGame.windSpeed = int(re.search('Wind ([0-9]{1,2})mph', b.weather, flags=re.IGNORECASE).group(1))
windDir = re.search('mph ([^,]*),', b.weather, flags=re.IGNORECASE).group(1)
if windDir == 'out to Centerfield':
windDir = 'tocf'
elif windDir == 'out to Rightfield':
windDir = 'torf'
elif windDir == 'out to Leftfield':
windDir = 'tolf'
elif windDir == 'from Left to Right':
windDir = 'ltor'
elif windDir == 'from Right to Left':
windDir = 'rtol'
elif windDir == 'in from Centerfield':
windDir = 'fromcf'
elif windDir == 'in from Rightfield':
windDir = 'fromrf'
elif windDir == 'in from Leftfield':
windDir = 'fromlf'
weatherStr = b.weather.split(',')
if len(weatherStr) < 4:
precip = 'unknown'
sky = weatherStr[-1].replace('.','').lower()
else:
precip = weatherStr[-1].replace('.','').lower()
sky = weatherStr[-2].lower()
field = b.field.lower() if (b.field != '') else 'unknown'
currentGame.weather = 'Field = '+field+', Prec = '+precip+', Sky = '+sky
b = GameUmpParser()
b.feed(html)
currentGame.homeUmp = b.homeump
#Lineups
b = GameWinLossSaveParser()
b.feed(html)
winPitch = abb(b.winPitch)
lossPitch = abb(b.lossPitch)
savePitch = '' if len(b.savePitch) == 0 else abb(b.savePitch)
b = BRLineupParser()
b.feed(html)
for line in b.lineup:
if len(line) == 8 and line[0].isdigit():
aBatNum = int(line[0])
aUID = str(line[1])
aName = str(line[2])
aPos = str(line[3]).strip()
hBatNum = int(line[4])
hUID = str(line[5])
hName = str(line[6])
hPos = str(line[7]).strip()
if aPos == 'P':
pbp.aPitcher = abb(aName)
if hPos == 'P':
pbp.hPitcher = abb(hName)
pbp.lineup[abb(aName)] = Lineup(123, pbp.aTeam, aName, aBatNum, aPos, aUID, 'BR', con)
pbp.lineup[abb(hName)] = Lineup(123, pbp.hTeam, hName, hBatNum, hPos, hUID, 'BR', con)
elif len(line) == 6:
pbp.aPitcher = abb(str(line[1]))
pbp.hPitcher = abb(str(line[4]))
#Pitch Rosters
bp = BRPitcherParser()
bp.feed(html)
pitchIDLookup = {}
erLookup = {}
for pitcher in bp.roster:
uid = str(pitcher[0])
name = str(pitcher[1])
pitchIDLookup[abb(name)] = uid
er = int(pitcher[-2])
erLookup[abb(name)] = er
team = pbp.aTeam if pitcher[-1] == 'A' else pbp.hTeam
pbp.pitchers[abb(name)] = PitchRoster(123, team, name, uid, 'BR', con)
pbp.pitchers[abb(name)].IP = 0
pbp.pitchers[abb(name)].pitcherRole = 'Starter'
#Hitter ID Lookup
bb = BRBatterParser()
bb.feed(html)
batID = {}
for x in bb.batID.keys():
batID[abb(x)] = bb.batID[x]
for y in pbp.lineup.keys():
if abb(x) == y:
break
elif abb(x)[:11] == y:
pbp.lineup[abb(x)] = pbp.lineup[y]
del pbp.lineup[y]
#Plays
b = BRPlayParser()
b.feed(html)
for playNum in range(1, len(b.plays)+1):
pbp.plays[playInd] = Play()
pbp.plays[playInd].hitterID = b.plays[playNum][7]
pbp.plays[playInd].pitcherID = b.plays[playNum][8]
#Inning Stuff
innPre = 'Top' if b.plays[playNum][0][0] == 't' else 'Bot'
prevInn = pbp.inning
pbp.inning = innPre + ' ' + str(b.plays[playNum][0][1])
pbp.plays[playInd].inning = pbp.inning
if pbp.inning != prevInn:
pbp.outs = 0
pbp.firstBase = None
pbp.secondBase = None
pbp.thirdBase = None
pbp.plays[playInd].startSit = pbp.ReturnSit()
#Pitches
pitchStr = b.plays[playNum][4]
if len(pitchStr.split(' ')) > 1:
pbp.plays[playInd].pitchSeq = pitchStr.split(' ')[1]
pbp.plays[playInd].strikes = int(pitchStr.split('-')[1][0])
pbp.plays[playInd].balls = int(pitchStr.split('-')[0][-1])
pbp.plays[playInd].resultOuts = b.plays[playNum][5].count('O')
pbp.outs += b.plays[playNum][5].count('O')
pbp.plays[playInd].runsScored = b.plays[playNum][5].count('R')
if pbp.ProcessBRPlay(b.plays[playNum][11], playInd):
pbp.lineup[b.plays[playNum][7]].RBI += pbp.plays[playInd].runsScored
pbp.plays[playInd].endSit = pbp.ReturnSit()
pbp.plays[playInd].playNum = playInd + 1
#Subs
if playNum in b.subs.keys():
subs = b.subs[playNum].split(';')
for sub in subs:
#Pitching Change
if re.search('(.*) replaces (.*) pitching', sub) is not None:
newP = re.search('(.*) replaces (.*) pitching', sub).group(1)
p = pbp.aPitcher if b.plays[playNum+1][0][0] == 'b' else pbp.hPitcher
pbp.pitchers[p].IP += int(b.plays[playNum+1][0][-1])-1
for x in pbp.pitchers.keys():
if pbp.pitchers[x].team == pbp.pitchers[p].team and x != p:
pbp.pitchers[p].IP -= float(pbp.pitchers[x].IP)
if int(pbp.outs) in range(1,3):
pbp.pitchers[p].IP += int(pbp.outs) / 3.0
if b.plays[playNum+1][0][0] == 'b':
pbp.aPitcher = abb(newP)
team = pbp.aTeam
else:
pbp.hPitcher = abb(newP)
team = pbp.hTeam
pbp.pitchers[abb(newP)] = PitchRoster(123, team, newP, pitchIDLookup[abb(newP)], 'BR', con)
pbp.pitchers[abb(newP)].pitcherRole = 'Reliever'
batnum = int(re.search('batting ([0-9])',sub).group(1)) \
if re.search('batting ([0-9])', sub) is not None else 0
pbp.lineup[abb(newP)] = Lineup(123, team, newP, batnum, 'P', pitchIDLookup[abb(newP)], 'BR', con)
#Pinch Hitter
if re.search('(.*) pinch hits for (.*)', sub) is not None:
newP = re.search('(.*) pinch hits for (.*)', sub).group(1)
team = pbp.hTeam if b.plays[playNum+1][0][0] == 'b' else pbp.aTeam
batnum = int(re.search('batting ([0-9])', sub).group(1)) \
if re.search('batting ([0-9])', sub) is not None else 0
pbp.lineup[abb(newP)] = Lineup(123, team, newP, batnum, 'PH', bb.batID[newP], 'BR', con)
#Defensive Sub
if re.search('(.*) replaces (.*) playing (.*) batting', sub) is not None:
newP = re.search('(.*) replaces (.*) playing (.*) batting', b.subs[playNum]).group(1)
team = pbp.aTeam if b.plays[playNum+1][0][0] == 'b' else pbp.hTeam
batnum = int(re.search('batting ([0-9])', b.subs[playNum]).group(1)) \
if re.search('batting ([0-9])', b.subs[playNum]) is not None else 0
pos = re.search('(.*) replaces (.*) playing (.*) batting', b.subs[playNum]).group(3)
pbp.lineup[abb(newP)] = Lineup(123, team, newP, batnum, pos, bb.batID[newP], 'BR', con)
#Pinch Runner
if re.search('(.*) pinch runs for', b.subs[playNum]) is not None:
newP = re.search('(.*) pinch runs for', b.subs[playNum]).group(1)
team = pbp.hTeam if b.plays[playNum+1][0][0] == 'b' else pbp.aTeam
batnum = int(re.search('batting ([0-9])', b.subs[playNum]).group(1)) \
if re.search('batting ([0-9])', b.subs[playNum]) is not None else 0
pbp.lineup[abb(newP)] = Lineup(123, team, newP, batnum, 'PR', bb.batID[newP], 'BR', con)
playInd += 1
pbp.plays[playInd] = Play()
pbp.plays[playInd].hitterID = abb(re.search('(.*) pinch runs for (.*) \(', b.subs[playNum]).group(1))
pbp.plays[playInd].pitcherID = abb(pbp.aPitcher) if pbp.inning[0] == 'B' else abb(pbp.hPitcher)
pbp.plays[playInd].inning = pbp.inning
pbp.plays[playInd].startSit = pbp.plays[playInd].endSit = pbp.ReturnSit()
pbp.plays[playInd].playType = 'Pinch Runner'
replacee = re.search('(.*) pinch runs for (.*) \(', b.subs[playNum]).group(2)
if pbp.firstBase != None and pbp.firstBase[1] == abb(replacee):
pbp.firstBase = [playInd, abb(newP)]
elif pbp.secondBase != None and pbp.secondBase[1] == abb(replacee):
pbp.secondBase = [playInd, abb(newP)]
elif pbp.thirdBase != None and pbp.thirdBase[1] == abb(replacee):
pbp.thirdBase = [playInd, abb(newP)]
playInd += 1
#Calculate IP for Last Pitchers
currentGame.InsertBlankGame(con)
currentGame.totalInnings = int(pbp.inning.split(' ')[-1])
pbp.pitchers[pbp.hPitcher].IP += currentGame.totalInnings
pbp.pitchers[pbp.aPitcher].IP += currentGame.totalInnings
if pbp.inning[:3] == 'Top':
pbp.pitchers[pbp.aPitcher].IP -= 1
elif pbp.inning[:3] == 'Bot':
pbp.pitchers[pbp.aPitcher].IP -= float(3 - int(pbp.outs)) / 3.0
for x in pbp.pitchers.keys():
if pbp.pitchers[x].team == pbp.pitchers[pbp.hPitcher].team and x != pbp.hPitcher:
pbp.pitchers[pbp.hPitcher].IP -= float(pbp.pitchers[x].IP)
if pbp.pitchers[x].team == pbp.pitchers[pbp.aPitcher].team and x != pbp.aPitcher:
pbp.pitchers[pbp.aPitcher].IP -= float(pbp.pitchers[x].IP)
#Insert Game, Plays, Pitchers, and Lineups into DB
for x in pbp.plays.values():
if x.playType not in (
'No Play', 'Stolen Base', 'Caught Stealing', 'Pick Off', 'Balk', 'Passed Ball', 'Wild Pitch',
'Defensive Indifference', 'Error on Foul', 'Unknown Runner Activity'):
x.CalcPitches()
pbp.lineup[x.hitterID].PA += 1
pbp.pitchers[x.pitcherID].ContactStrikes += x.contactX
pbp.pitchers[x.pitcherID].SwingStrikes += x.swingX
pbp.pitchers[x.pitcherID].LookStrikes += x.lookX
pbp.pitchers[x.pitcherID].Strikes += x.lookX + x.swingX + x.contactX
pbp.pitchers[x.pitcherID].Balls += x.balls
pbp.pitchers[x.pitcherID].pitchCount += x.lookX + x.swingX + x.contactX + x.balls
if x.ballType in ('Ground Ball', 'Bunt Ground Ball'):
pbp.pitchers[x.pitcherID].GB += 1
elif x.ballType in ('Line Drive', 'Bunt Line Drive'):
pbp.pitchers[x.pitcherID].LD += 1
elif x.ballType in ('Fly Ball', 'Pop Up', 'Bunt Pop'):
pbp.pitchers[x.pitcherID].FB += 1
if x.playType in (
'Strikeout', 'Out', 'Double Play', 'Triple Play', "Fielders Choice", 'Reach On Error', 'Single',
'Double', 'Ground Rule Double', 'Triple', 'Home Run'):
pbp.lineup[x.hitterID].AB += 1
if x.playType in ('Single', 'Double', 'Ground Rule Double', 'Triple', 'Home Run'):
pbp.lineup[x.hitterID].Hits += 1
pbp.pitchers[x.pitcherID].Hits += 1
if x.playType == 'Single':
pbp.lineup[x.hitterID].Single += 1
elif x.playType in ('Double', 'Ground Rule Double'):
pbp.lineup[x.hitterID].Double += 1
elif x.playType == 'Triple':
pbp.lineup[x.hitterID].Triple += 1
elif x.playType == 'Home Run':
pbp.lineup[x.hitterID].HR += 1
elif x.playType == 'Strikeout':
pbp.pitchers[x.pitcherID].K += 1
else:
if x.playType in ('Walk', 'Intentional Walk'):
pbp.pitchers[x.pitcherID].BB += 1
pbp.lineup[x.hitterID].BB += 1
pbp.pitchers[x.pitcherID].pitchCount += 1
elif x.playType == 'Hit By Pitch':
pbp.pitchers[x.pitcherID].HBP += 1
pbp.lineup[x.hitterID].HBP += 1
if x.inning[:3] == 'Top':
if x.hit:
currentGame.awayHits += 1
if x.runScored:
pbp.lineup[x.hitterID].Runs += 1
currentGame.awayRuns += x.runsScored
else:
if x.hit:
currentGame.homeHits += 1
if x.runScored:
pbp.lineup[x.hitterID].Runs += 1
currentGame.homeRuns += x.runsScored
x.gameKey = currentGame.gameKey
x.hitterID = batID[x.hitterID]
x.pitcherID = pitchIDLookup[x.pitcherID]
x.InsertPlay('BR', con)
if currentGame.homeRuns > currentGame.awayRuns:
currentGame.homeTeamWin = True
elif currentGame.homeRuns == currentGame.awayRuns:
currentGame.tie = True
currentGame.UpdateStats(con)
# Go Through Pitchers
for x in pbp.pitchers.keys():
if x == winPitch:
pbp.pitchers[x].Win = True
elif x == lossPitch:
pbp.pitchers[x].Loss = True
elif x == savePitch:
pbp.pitchers[x].Save = True
if pbp.pitchers[x].IP == 9.0:
pbp.pitchers[x].CG = True
if pbp.pitchers[x].Runs == 0:
pbp.pitchers[x].SO = True
if pbp.pitchers[x].Hits == 0:
pbp.pitchers[x].NH = True
pbp.pitchers[x].gameKey = currentGame.gameKey
pbp.pitchers[x].earnedRuns = erLookup[x]
pbp.pitchers[x].InsertRosterRow(con)
# Go Through Hitters
for x in pbp.lineup.keys():
pbp.lineup[x].game = currentGame.gameKey
pbp.lineup[x].InsertLineupRow(con)
return currentGame, pbp
from BRParser import BRPitcherParser
import urllib2
b = BRPitcherParser()
url = "http://www.baseball-reference.com/boxes/DET/DET201504060.shtml"
html = urllib2.urlopen(url).read().decode('utf-8').replace('·','*')
b.feed(html)
for x in b.roster:
print x