def parse(self, response):
# Get this game code from file
with open(os.getcwd() + "/tmpfiles/" + ''.join(e for e in response.url if e.isalnum()) + ".txt") as f:
data = f.read()
m = re.search(r"Code: (?P<code>\d+)", data)
code = str(m.group('code')).zfill(16)
# Scrape box score and save raw file
table = response.xpath('//table[contains(@class, "mod-data")]')
rows = []
visitor = int(code) / 1000000000000
home = (int(code) / 100000000) % 1000
date = int(code) % 100000000
for row in table.xpath('.//tr'):
new_rows1 = [x.xpath('.//text()').extract() for x in row.xpath('.//td')]
if len(new_rows1) > 0:
rows.append(new_rows1)
new_rows2 = [x.xpath('.//text()').extract() for x in row.xpath('.//th')]
if len(new_rows2) > 0:
if len(new_rows2) == 3:
new_rows2 = [new_rows2[0], "", new_rows2[1], new_rows2[2]]
rows.append(new_rows2)
for i in range(0, len(rows[len(rows)-1])):
rows[len(rows)-1][i] = ''.join([re.sub(r"\[u'\\xa0'\]|', |\[u'|u'|'\]|\[|\]", '', str(rows[len(rows)-1][i]))])
Write_CSV(rows, "box/" + str(visitor).zfill(4) + str(home).zfill(4) + str(date) + ".csv")
# Convert to team-game-statistics format
visitor_TGS = Team_Game_Statistics(code, visitor)
home_TGS = Team_Game_Statistics(code, home)
team_names = Read_CSV("2014 Stats/team.csv")
team_names = team_names[1:]
team_abbvs = Read_CSV("2014 Stats/abbrevations.csv")
# Get score
for i in range(0, len(rows)):
first_qtr = re.search(r"FIRST QUARTER", rows[i][0])
if first_qtr:
while len(rows[i+1]) >= 5:
i += 1
visitor_TGS.Points = rows[i][4]
home_TGS.Points = rows[i][5]
second_qtr = re.search(r"SECOND QUARTER", rows[i][0])
if second_qtr:
while len(rows[i+1]) >= 5:
i += 1
visitor_TGS.Points = rows[i][4]
home_TGS.Points = rows[i][5]
third_qtr = re.search(r"THIRD QUARTER", rows[i][0])
if third_qtr:
while len(rows[i+1]) >= 5:
i += 1
visitor_TGS.Points = rows[i][4]
home_TGS.Points = rows[i][5]
fourth_qtr = re.search(r"FOURTH QUARTER", rows[i][0])
if fourth_qtr:
while len(rows[i+1]) >= 5:
i += 1
visitor_TGS.Points = rows[i][4]
home_TGS.Points = rows[i][5]
# Box score stats
for i in range(0, len(rows)):
# Total 1st downs
first_downs = re.search(r"1st Downs", rows[i][0])
if first_downs:
visitor_TGS.First_Down_Total = rows[i][1]
home_TGS.First_Down_Total = rows[i][2]
# 3rd down conversions
third_downs = re.search(r"3rd down efficiency", rows[i][0])
if third_downs:
eff = re.match(r"(?P<conv>\d+)\-(?P<att>\d+)", rows[i][1])
visitor_TGS.Third_Down_Att = eff.group("att")
visitor_TGS.Third_Down_Conv = eff.group("conv")
eff = re.match(r"(?P<conv>\d+)\-(?P<att>\d+)", rows[i][2])
home_TGS.Third_Down_Att = eff.group("att")
home_TGS.Third_Down_Conv = eff.group("conv")
# 4th down conversions
fourth_downs = re.search(r"4th down efficiency", rows[i][0])
if fourth_downs:
eff = re.match(r"(?P<conv>\d+)\-(?P<att>\d+)", rows[i][1])
visitor_TGS.Fourth_Down_Att = eff.group("att")
visitor_TGS.Fourth_Down_Conv = eff.group("conv")
eff = re.match(r"(?P<conv>\d+)\-(?P<att>\d+)", rows[i][2])
home_TGS.Fourth_Down_Att = eff.group("att")
home_TGS.Fourth_Down_Conv = eff.group("conv")
# Penalties
penalties = re.search(r"Penalties", rows[i][0])
if penalties:
num_yrds = re.search(r"(?P<num>\d+)\-(?P<yrds>\d+)", rows[i][1])
visitor_TGS.Penalty = num_yrds.group("num")
visitor_TGS.Penalty_Yard = num_yrds.group("yrds")
num_yrds = re.search(r"(?P<num>\d+)\-(?P<yrds>\d+)", rows[i][2])
home_TGS.Penalty = num_yrds.group("num")
home_TGS.Penalty_Yard = num_yrds.group("yrds")
# Possession
possession = re.search(r"Possession", rows[i][0])
if possession:
visitor_TGS.Time_Of_Possession = rows[i][1]
home_TGS.Time_Of_Possession = rows[i][2]
# Fumbles Lost
fum_lost = re.search(r"Fumbles lost", rows[i][0])
if fum_lost:
visitor_TGS.Fum_Lost = rows[i][1]
home_TGS.Fum_Lost = rows[i][2]
visitor_TGS.Fum_Ret = home_TGS.Fum_Lost
home_TGS.Fum_Ret = visitor_TGS.Fum_Lost
# Find stats
visitor_TGS = Parse_Box(rows, visitor_TGS, team_abbvs)
# START DEBUG --
#if int(visitor_TGS.Rush_Att) + int(visitor_TGS.Pass_Att) == 0:
#pdb.set_trace()
#visitor_TGS = Parse_Box(rows, visitor_TGS, team_abbvs)
# END DEBUG --
home_TGS = Parse_Box(rows, home_TGS, team_abbvs)
# START DEBUG --
#if int(home_TGS.Rush_Att) + int(home_TGS.Pass_Att) == 0:
#pdb.set_trace()
#home_TGS = Parse_Box(rows, visitor_TGS, team_abbvs)
# END DEBUG --
if os.path.isfile("2014 Stats/team-game-statistics.csv"):
f = open("2014 Stats/team-game-statistics.csv","a")
data_writer = csv.writer(f, lineterminator = '\n')
new_rows = []
new_rows.append(visitor_TGS.Compile())
new_rows.append(home_TGS.Compile())
data_writer.writerows(new_rows)
f.close()
else:
new_rows = []
new_rows.append(visitor_TGS.Header())
new_rows.append(visitor_TGS.Compile())
new_rows.append(home_TGS.Compile())
Write_CSV(new_rows, "2014 Stats/team-game-statistics.csv")
def parse(self, response):
# Get this game code from file
with open(os.getcwd() + "/tmpfiles/" + ''.join(e for e in response.url if e.isalnum()) + ".txt") as f:
data = f.read()
m = re.search(r"Code: (?P<code>\d+)", data)
code = str(m.group('code')).zfill(16)
# Scrape box score
away = int(long(code) / 1e12)
home = int((long(code) / 1e8) % 1e3)
date = int(long(code) % 1e8)
away_TGS = Team_Game_Statistics(code, away)
home_TGS = Team_Game_Statistics(code, home)
# MOVE SOME OF THESE TO MATCHUP SCRAPER
# Scrape passing
pass_div = response.xpath('//div[@id="gamepackage-passing"]')
# away
away_pass_div = pass_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_Passing(away_TGS, away_pass_div)
# home
home_pass_div = pass_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_Passing(home_TGS, home_pass_div)
# Scrape rushing
rush_div = response.xpath('//div[@id="gamepackage-rushing"]')
# away
away_rush_div = rush_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_Rushing(away_TGS, away_rush_div)
# home
home_rush_div = rush_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_Rushing(home_TGS, home_rush_div)
# Scrape kick returns
kr_div = response.xpath('//div[@id="gamepackage-kickReturns"]')
# away
away_kr_div = kr_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_KickReturns(away_TGS, away_kr_div)
# home
home_kr_div = kr_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_KickReturns(home_TGS, home_kr_div)
# Scrape punt returns
pr_div = response.xpath('//div[@id="gamepackage-puntReturns"]')
# away
away_pr_div = pr_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_PuntReturns(away_TGS, away_pr_div)
# home
home_pr_div = pr_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_PuntReturns(home_TGS, home_pr_div)
# Scrape interception returns
int_div = response.xpath('//div[@id="gamepackage-interceptions"]')
# away
away_int_div = int_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_Interceptions(away_TGS, away_int_div)
# home
home_int_div = int_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_Interceptions(home_TGS, home_int_div)
# Scrape kicking
kick_div = response.xpath('//div[@id="gamepackage-kicking"]')
# away
away_kick_div = kick_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_Kicking(away_TGS, away_kick_div)
# home
home_kick_div = kick_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_Kicking(home_TGS, home_kick_div)
# Scrape punting
punt_div = response.xpath('//div[@id="gamepackage-punting"]')
# away
away_punt_div = punt_div.xpath('.//div[contains(@class,"gamepackage-away-wrap")]')
away_TGS = Extract_Punting(away_TGS, away_punt_div)
# home
home_punt_div = punt_div.xpath('.//div[contains(@class,"gamepackage-home-wrap")]')
home_TGS = Extract_Punting(home_TGS, home_punt_div)
# Get points
points_div = response.xpath('//div[@class="competitors"]')
away_points = points_div.xpath('.//div[contains(@class,"away")]')
away_TGS = Extract_Points(away_TGS, away_points)
home_points = points_div.xpath('.//div[contains(@class,"home")]')
home_TGS = Extract_Points(home_TGS, home_points)
# Write stats to file
if os.path.isfile(str(year) + " Stats/boxscore-stats.csv"):
f = open(str(year) + " Stats/boxscore-stats.csv","a")
data_writer = csv.writer(f, lineterminator = '\n')
new_rows = []
new_rows.append(away_TGS.Compile())
new_rows.append(home_TGS.Compile())
data_writer.writerows(new_rows)
f.close()
else:
new_rows = []
new_rows.append(away_TGS.Header())
new_rows.append(away_TGS.Compile())
new_rows.append(home_TGS.Compile())
Write_CSV(new_rows, str(year) + " Stats/boxscore-stats.csv")
play_data.append(play)
Write_CSV(play_data, str(year) + " Stats temp/unparsed_plays.csv")
# Build team-game-statistics
prev_game_code = 0
allTGS = []
for play in allPlays:
# found a new game
if float(play.Game_Code) != prev_game_code:
# save old data
if prev_game_code != 0:
allTGS.append(home_tgs)
allTGS.append(visitor_tgs)
visitor_code = int(math.floor(float(play.Game_Code) / 1e12))
home_code = int(math.floor(float(play.Game_Code) / 1e8)) % 1e4
home_tgs = Team_Game_Statistics(play.Game_Code, home_code)
visitor_tgs = Team_Game_Statistics(play.Game_Code, visitor_code)
prev_game_code = float(play.Game_Code)
# increment data
if play.Offense == home_tgs.Team_Code:
home_tgs.Extract_Play_Offense(play)
elif play.Offense == visitor_tgs.Team_Code:
visitor_tgs.Extract_Play_Offense(play)
# Write team-game-statistics to file
tgs_data = []
tgs_data.append(allTGS[0].Header())
for tgs in allTGS:
tgs_data.append(tgs.Compile_Stats())
Write_CSV(tgs_data, str(year) + " Stats temp/play_TGS.csv")
def parse(self, response):
# Get this game code from file
with open(os.getcwd() + "/tmpfiles/" +
''.join(e
for e in response.url if e.isalnum()) + ".txt") as f:
data = f.read()
m = re.search(r"Code: (?P<code>\d+)", data)
code = str(m.group('code')).zfill(16)
# Scrape box score
away = int(long(code) / 1e12)
home = int((long(code) / 1e8) % 1e3)
date = int(long(code) % 1e8)
away_TGS = Team_Game_Statistics(code, away)
home_TGS = Team_Game_Statistics(code, home)
# Scrape first downs
first_div = response.xpath('//tr[@data-stat-attr="firstDowns"]')
away_TGS.First_Down_Total = re.sub(
r'[\\\t|\\\n]', '',
first_div.xpath('.//td/text()').extract()[1])
home_TGS.First_Down_Total = re.sub(
r'[\\\t|\\\n]', '',
first_div.xpath('.//td/text()').extract()[2])
# Scrape turnovers
fumble_div = response.xpath('//tr[@data-stat-attr="fumblesLost"]')
away_TGS.Fum_Lost = re.sub(
r'[\\\t|\\\n]', '',
fumble_div.xpath('.//td/text()').extract()[1])
home_TGS.Fum_Lost = re.sub(
r'[\\\t|\\\n]', '',
fumble_div.xpath('.//td/text()').extract()[2])
away_TGS.Fum_Ret = home_TGS.Fum_Lost
home_TGS.Fum_Ret = away_TGS.Fum_Lost
# Scrape first down efficiency on 3rd down
eff_div = response.xpath('//tr[@data-stat-attr="thirdDownEff"]')
fatt_away = re.sub(r'[\\\t|\\\n]', '',
eff_div.xpath('.//td/text()').extract()[1])
fatt_home = re.sub(r'[\\\t|\\\n]', '',
eff_div.xpath('.//td/text()').extract()[2])
# away
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_away)
away_TGS.Third_Down_Att = m.group('a')
away_TGS.Third_Down_Conv = m.group('f')
# home
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_home)
home_TGS.Third_Down_Att = m.group('a')
home_TGS.Third_Down_Conv = m.group('f')
# Scrape first down efficiency on 4th down
eff_div = response.xpath('//tr[@data-stat-attr="fourthDownEff"]')
fatt_away = re.sub(r'[\\\t|\\\n]', '',
eff_div.xpath('.//td/text()').extract()[1])
fatt_home = re.sub(r'[\\\t|\\\n]', '',
eff_div.xpath('.//td/text()').extract()[2])
# away
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_away)
away_TGS.Fourth_Down_Att = m.group('a')
away_TGS.Fourth_Down_Conv = m.group('f')
# home
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_home)
home_TGS.Fourth_Down_Att = m.group('a')
home_TGS.Fourth_Down_Conv = m.group('f')
# Scrape time of possession
top_div = response.xpath('//tr[@data-stat-attr="possessionTime"]')
try:
top_away = re.sub(r'[\\\t|\\\n]', '',
top_div.xpath('.//td/text()').extract()[1])
except:
top_away = "30:00"
try:
top_home = re.sub(r'[\\\t|\\\n]', '',
top_div.xpath('.//td/text()').extract()[2])
except:
top_home = "30:00"
# away
m_away = re.search(r'(?P<h>\d+)\:(?P<m>\d+)', top_away)
# home
m_home = re.search(r'(?P<h>\d+)\:(?P<m>\d+)', top_home)
try:
away_TGS.Time_Of_Possession = str(60 * int(m_away.group('h')) +
int(m_away.group('m')))
home_TGS.Time_Of_Possession = str(60 * int(m_home.group('h')) +
int(m_home.group('m')))
except:
away_TGS.Time_Of_Possession = 1800
home_TGS.Time_Of_Possession = 1800
if int(away_TGS.Time_Of_Possession) == 1800 and int(
home_TGS.Time_Of_Possession) != 1800:
away_TGS.Time_Of_Possession = str(3600 -
int(home_TGS.Time_Of_Possession))
elif int(home_TGS.Time_Of_Possession) == 1800 and int(
away_TGS.Time_Of_Possession) != 1800:
home_TGS.Time_Of_Possession = str(3600 -
int(away_TGS.Time_Of_Possession))
# Scrape penalties
pen_div = response.xpath('//tr[@data-stat-attr="totalPenaltiesYards"]')
pen_away = re.sub(r'[\\\t|\\\n]', '',
pen_div.xpath('.//td/text()').extract()[1])
pen_home = re.sub(r'[\\\t|\\\n]', '',
pen_div.xpath('.//td/text()').extract()[2])
# away
m = re.search(r'(?P<tot>\d+)\-(?P<yds>\d+)', pen_away)
away_TGS.Penalty = m.group('tot')
away_TGS.Penalty_Yard = m.group('yds')
# home
m = re.search(r'(?P<tot>\d+)\-(?P<yds>\d+)', pen_home)
home_TGS.Penalty = m.group('tot')
home_TGS.Penalty_Yard = m.group('yds')
# Write stats to file
if os.path.isfile(str(year) + " Stats/matchup-stats.csv"):
f = open(str(year) + " Stats/matchup-stats.csv", "a")
data_writer = csv.writer(f, lineterminator='\n')
new_rows = []
new_rows.append(away_TGS.Compile())
new_rows.append(home_TGS.Compile())
data_writer.writerows(new_rows)
f.close()
else:
new_rows = []
new_rows.append(away_TGS.Header())
new_rows.append(away_TGS.Compile())
new_rows.append(home_TGS.Compile())
Write_CSV(new_rows, str(year) + " Stats/matchup-stats.csv")
def parse(self, response):
# Get this game code from file
with open(os.getcwd() + "/tmpfiles/" + ''.join(e for e in response.url if e.isalnum()) + ".txt") as f:
data = f.read()
m = re.search(r"Code: (?P<code>\d+)", data)
code = str(m.group('code')).zfill(16)
# Scrape box score
away = int(long(code) / 1e12)
home = int((long(code) / 1e8) % 1e3)
date = int(long(code) % 1e8)
away_TGS = Team_Game_Statistics(code, away)
home_TGS = Team_Game_Statistics(code, home)
# Scrape first downs
first_div = response.xpath('//tr[@data-stat-attr="firstDowns"]')
away_TGS.First_Down_Total = re.sub(r'[\\\t|\\\n]','',first_div.xpath('.//td/text()').extract()[1])
home_TGS.First_Down_Total = re.sub(r'[\\\t|\\\n]','',first_div.xpath('.//td/text()').extract()[2])
# Scrape turnovers
fumble_div = response.xpath('//tr[@data-stat-attr="fumblesLost"]')
away_TGS.Fum_Lost = re.sub(r'[\\\t|\\\n]','',fumble_div.xpath('.//td/text()').extract()[1])
home_TGS.Fum_Lost = re.sub(r'[\\\t|\\\n]','',fumble_div.xpath('.//td/text()').extract()[2])
away_TGS.Fum_Ret = home_TGS.Fum_Lost
home_TGS.Fum_Ret = away_TGS.Fum_Lost
# Scrape first down efficiency on 3rd down
eff_div = response.xpath('//tr[@data-stat-attr="thirdDownEff"]')
fatt_away = re.sub(r'[\\\t|\\\n]','',eff_div.xpath('.//td/text()').extract()[1])
fatt_home = re.sub(r'[\\\t|\\\n]','',eff_div.xpath('.//td/text()').extract()[2])
# away
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_away)
away_TGS.Third_Down_Att = m.group('a')
away_TGS.Third_Down_Conv = m.group('f')
# home
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_home)
home_TGS.Third_Down_Att = m.group('a')
home_TGS.Third_Down_Conv = m.group('f')
# Scrape first down efficiency on 4th down
eff_div = response.xpath('//tr[@data-stat-attr="fourthDownEff"]')
fatt_away = re.sub(r'[\\\t|\\\n]','',eff_div.xpath('.//td/text()').extract()[1])
fatt_home = re.sub(r'[\\\t|\\\n]','',eff_div.xpath('.//td/text()').extract()[2])
# away
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_away)
away_TGS.Fourth_Down_Att = m.group('a')
away_TGS.Fourth_Down_Conv = m.group('f')
# home
m = re.search(r'(?P<f>\d+)\-(?P<a>\d+)', fatt_home)
home_TGS.Fourth_Down_Att = m.group('a')
home_TGS.Fourth_Down_Conv = m.group('f')
# Scrape time of possession
top_div = response.xpath('//tr[@data-stat-attr="possessionTime"]')
try:
top_away = re.sub(r'[\\\t|\\\n]','',top_div.xpath('.//td/text()').extract()[1])
except:
top_away = "30:00"
try:
top_home = re.sub(r'[\\\t|\\\n]','',top_div.xpath('.//td/text()').extract()[2])
except:
top_home = "30:00"
# away
m_away = re.search(r'(?P<h>\d+)\:(?P<m>\d+)', top_away)
# home
m_home = re.search(r'(?P<h>\d+)\:(?P<m>\d+)', top_home)
try:
away_TGS.Time_Of_Possession = str(60*int(m_away.group('h')) + int(m_away.group('m')))
home_TGS.Time_Of_Possession = str(60*int(m_home.group('h')) + int(m_home.group('m')))
except:
away_TGS.Time_Of_Possession = 1800
home_TGS.Time_Of_Possession = 1800
if int(away_TGS.Time_Of_Possession) == 1800 and int(home_TGS.Time_Of_Possession) != 1800:
away_TGS.Time_Of_Possession = str(3600 - int(home_TGS.Time_Of_Possession))
elif int(home_TGS.Time_Of_Possession) == 1800 and int(away_TGS.Time_Of_Possession) != 1800:
home_TGS.Time_Of_Possession = str(3600 - int(away_TGS.Time_Of_Possession))
# Scrape penalties
pen_div = response.xpath('//tr[@data-stat-attr="totalPenaltiesYards"]')
pen_away = re.sub(r'[\\\t|\\\n]','',pen_div.xpath('.//td/text()').extract()[1])
pen_home = re.sub(r'[\\\t|\\\n]','',pen_div.xpath('.//td/text()').extract()[2])
# away
m = re.search(r'(?P<tot>\d+)\-(?P<yds>\d+)', pen_away)
away_TGS.Penalty = m.group('tot')
away_TGS.Penalty_Yard = m.group('yds')
# home
m = re.search(r'(?P<tot>\d+)\-(?P<yds>\d+)', pen_home)
home_TGS.Penalty = m.group('tot')
home_TGS.Penalty_Yard = m.group('yds')
# Write stats to file
if os.path.isfile(str(year) + " Stats/matchup-stats.csv"):
f = open(str(year) + " Stats/matchup-stats.csv","a")
data_writer = csv.writer(f, lineterminator = '\n')
new_rows = []
new_rows.append(away_TGS.Compile())
new_rows.append(home_TGS.Compile())
data_writer.writerows(new_rows)
f.close()
else:
new_rows = []
new_rows.append(away_TGS.Header())
new_rows.append(away_TGS.Compile())
new_rows.append(home_TGS.Compile())
Write_CSV(new_rows, str(year) + " Stats/matchup-stats.csv")
