def getChampions(soup, season):
global allChampions
p = soup.select_one('#meta > div:nth-child(2) > p')
a = p.select_one('a')
name = a.get_text()
team_id = int(searchTeamByName(name)[0])
allChampions.append([team_id, season])
def helperTwo(row, id, season):
team_id = int(
searchTeamByName(
row.select_one('[data-stat="team_name"] > a').get_text())[0])
g = toIntOrFloat(row, '[data-stat="g"]')
mp = toIntOrFloat(row, '[data-stat="mp"]')
fg = toIntOrFloat(row, '[data-stat="opp_fg"]')
fga = toIntOrFloat(row, '[data-stat="opp_fga"]')
fg_pct = toIntOrFloat(row, '[data-stat="opp_fg_pct"]')
fg3 = toIntOrFloat(row, '[data-stat="opp_fg3"]')
fg3a = toIntOrFloat(row, '[data-stat="opp_fg3a"]')
fg3_pct = toIntOrFloat(row, '[data-stat="opp_fg3_pct"]')
fg2 = toIntOrFloat(row, '[data-stat="opp_fg2"]')
fg2a = toIntOrFloat(row, '[data-stat="opp_fg2a"]')
fg2_pct = toIntOrFloat(row, '[data-stat="opp_fg2_pct"]')
ft = toIntOrFloat(row, '[data-stat="opp_ft"]')
fta = toIntOrFloat(row, '[data-stat="opp_fta"]')
ft_pct = toIntOrFloat(row, '[data-stat="opp_ft_pct"]')
orb = toIntOrFloat(row, '[data-stat="opp_orb"]')
drb = toIntOrFloat(row, '[data-stat="opp_drb"]')
trb = toIntOrFloat(row, '[data-stat="opp_trb"]')
ast = toIntOrFloat(row, '[data-stat="opp_ast"]')
stl = toIntOrFloat(row, '[data-stat="opp_stl"]')
blk = toIntOrFloat(row, '[data-stat="opp_blk"]')
tov = toIntOrFloat(row, '[data-stat="opp_tov"]')
pf = toIntOrFloat(row, '[data-stat="opp_pf"]')
pts = toIntOrFloat(row, '[data-stat="opp_pts"]')
return [
id, team_id, season, g, mp, fg, fga, fg_pct, fg3, fg3a, fg3_pct, fg2,
fg2a, fg2_pct, ft, fta, ft_pct, orb, drb, trb, ast, stl, blk, tov, pf,
pts
]
def helperFive(row, id, season):
# 'id', 'team_id', 'season', 'g', 'mp' 'fg_pct', avg_dist, fg2a_ct_fga, pct_fga_00_03, pct_fga_03_10, pct_fga_10_16, pct_fga_16_xx, fg3a_pct_fga, fg2_pct, fg_pct_00_03, fg_pct_03_10, fg_pct_10_16, fg_pct_16_xx, fg3_pct, fg2_pct_ast, pct_fg2_dunk, fg2_dunk, pct_fg2_layup, fg2_layup, fg3_pct_ast, pct_fg3a_corner, fg3_pct_corner, fg3a_heave, fg3_heave
team_id = int(
searchTeamByName(
row.select_one('[data-stat="team_name"] > a').get_text())[0])
g = toIntOrFloat(row, '[data-stat="g"]')
mp = toIntOrFloat(row, '[data-stat="mp"]')
fg_pct = toIntOrFloat(row, '[data-stat="opp_fg_pct"]')
avg_dist = toIntOrFloat(row, '[data-stat="opp_avg_dist"]')
fg2a_ct_fga = toIntOrFloat(row, '[data-stat="opp_fg2a_ct_fga"]')
pct_fga_00_03 = toIntOrFloat(row, '[data-stat="opp_pct_fga_00_03"]')
pct_fga_03_10 = toIntOrFloat(row, '[data-stat="opp_pct_fga_03_10"]')
pct_fga_10_16 = toIntOrFloat(row, '[data-stat="opp_pct_fga_10_16"]')
pct_fga_16_xx = toIntOrFloat(row, '[data-stat="opp_pct_fga_16_xx"]')
fg3a_pct_fga = toIntOrFloat(row, '[data-stat="opp_fg3a_pct_fga"]')
fg2_pct = toIntOrFloat(row, '[data-stat="opp_fg2_pct"]')
fg_pct_00_03 = toIntOrFloat(row, '[data-stat="opp_fg_pct_00_03"]')
fg_pct_03_10 = toIntOrFloat(row, '[data-stat="opp_fg_pct_03_10"]')
fg_pct_10_16 = toIntOrFloat(row, '[data-stat="opp_fg_pct_10_16"]')
fg_pct_16_xx = toIntOrFloat(row, '[data-stat="opp_fg_pct_16_xx"]')
fg3_pct = toIntOrFloat(row, '[data-stat="opp_fg3_pct"]')
fg2_pct_ast = toIntOrFloat(row, '[data-stat="opp_fg2_pct_ast"]')
pct_fg2_dunk = toIntOrFloat(row, '[data-stat="opp_pct_fg2_dunk"]')
fg2_dunk = toIntOrFloat(row, '[data-stat="opp_fg2_dunk"]')
pct_fg2_layup = toIntOrFloat(row, '[data-stat="opp_pct_fg2_layup"]')
fg2_layup = toIntOrFloat(row, '[data-stat="opp_fg2_layup"]')
fg3_pct_ast = toIntOrFloat(row, '[data-stat="opp_fg3_pct_ast"]')
pct_fg3a_corner = toIntOrFloat(row, '[data-stat="opp_pct_fg3a_corner"]')
fg3_pct_corner = toIntOrFloat(row, '[data-stat="opp_fg3_pct_corner"]')
fg3a_heave = toIntOrFloat(row, '[data-stat="opp_fg3a_heave"]')
fg3_heave = toIntOrFloat(row, '[data-stat="opp_fg3_heave"]')
return [
id, team_id, season, g, mp, fg_pct, avg_dist, fg2a_ct_fga,
pct_fga_00_03, pct_fga_03_10, pct_fga_10_16, pct_fga_16_xx,
fg3a_pct_fga, fg2_pct, fg_pct_00_03, fg_pct_03_10, fg_pct_10_16,
fg_pct_16_xx, fg3_pct, fg2_pct_ast, pct_fg2_dunk, fg2_dunk,
pct_fg2_layup, fg2_layup, fg3_pct_ast, pct_fg3a_corner, fg3_pct_corner,
fg3a_heave, fg3_heave
]
def helperThree(row, id, season):
# id team_id season age wins losses wins_pyth losses_pyth mov sos srs off_rtg def_rtg net_rtg pace fta_per_fga_pct fg3a_per_fga_pct ts_pct efg_pct tov_pct orb_pct ft_rate opp_efg_pct opp_tov_pct drb_pct opp_ft_rate arena_name attendance attendance_per_g
team_id = int(
searchTeamByName(
row.select_one('[data-stat="team_name"] > a').get_text())[0])
age = toIntOrFloat(row, '[data-stat="age"]')
wins = toIntOrFloat(row, '[data-stat="wins"]')
losses = toIntOrFloat(row, '[data-stat="g"]')
wins_pyth = toIntOrFloat(row, '[data-stat="wins_pyth"]')
losses_pyth = toIntOrFloat(row, '[data-stat="losses_pyth"]')
mov = toIntOrFloat(row, '[data-stat="mov"]')
sos = toIntOrFloat(row, '[data-stat="sos"]')
srs = toIntOrFloat(row, '[data-stat="srs"]')
off_rtg = toIntOrFloat(row, '[data-stat="off_rtg"]')
def_rtg = toIntOrFloat(row, '[data-stat="def_rtg"]')
net_rtg = toIntOrFloat(row, '[data-stat="net_rtg"]')
pace = toIntOrFloat(row, '[data-stat="pace"]')
fta_per_fga_pct = toIntOrFloat(row, '[data-stat="fta_per_fga_pct"]')
fg3a_per_fga_pct = toIntOrFloat(row, '[data-stat="fg3a_per_fga_pct"]')
ts_pct = toIntOrFloat(row, '[data-stat="ts_pct"]')
efg_pct = toIntOrFloat(row, '[data-stat="efg_pct"]')
tov_pct = toIntOrFloat(row, '[data-stat="tov_pct"]')
orb_pct = toIntOrFloat(row, '[data-stat="orb_pct"]')
ft_rate = toIntOrFloat(row, '[data-stat="ft_rate"]')
opp_efg_pct = toIntOrFloat(row, '[data-stat="opp_efg_pct"]')
opp_tov_pct = toIntOrFloat(row, '[data-stat="opp_tov_pct"]')
drb_pct = toIntOrFloat(row, '[data-stat="drb_pct"]')
opp_ft_rate = toIntOrFloat(row, '[data-stat="opp_ft_rate"]')
arena_name = row.select_one('[data-stat="arena_name"]').get_text()
attendance = toIntOrFloat(row, '[data-stat="attendance"]')
attendance_per_g = toIntOrFloat(row, '[data-stat="attendance_per_g"]')
return [
id, team_id, season, age, wins, losses, wins_pyth, losses_pyth, mov,
sos, srs, off_rtg, def_rtg, net_rtg, pace, fta_per_fga_pct,
fg3a_per_fga_pct, ts_pct, efg_pct, tov_pct, orb_pct, ft_rate,
opp_efg_pct, opp_tov_pct, drb_pct, opp_ft_rate, arena_name, attendance,
attendance_per_g
]
def getGameData(url, year):
global playoff
global id
html_source = requests.get(url)
soup = BeautifulSoup(html_source.content, 'html.parser')
if soup.find('h1').get_text() == 'Page Not Found (404 error)':
return
season = soup.select_one(
'ul.hoversmooth > li.index > a > u').get_text()[:7]
table_rows = soup.select('#schedule > tbody > tr')
for tr in table_rows:
try:
if (
type(tr.contents[0]) is Tag
and tr.contents[0].get_text() == 'Playoffs'
) or url == 'https://www.basketball-reference.com/leagues/NBA_1980_games-april.html':
playoff = True
except:
pass
# try:
if not tr.get('class'):
date = tr.select_one('[data-stat="date_game"]').get_text()
awayTeamName = tr.select_one(
'[data-stat="visitor_team_name"]').get_text()
awayTeamId = int(searchTeamByName(awayTeamName)[0])
awayTeamScore = tr.select_one(
'[data-stat="visitor_pts"]').get_text()
homeTeamName = tr.select_one(
'[data-stat="home_team_name"]').get_text()
homeTeamId = int(searchTeamByName(homeTeamName)[0])
homeTeamScore = tr.select_one('[data-stat="home_pts"]').get_text()
overtime = tr.select_one('[data-stat="overtimes"]').get_text()
attendance = tr.select_one('[data-stat="attendance"]').get_text()
if attendance != '':
attendance = int(attendance.replace(',', ''))
notes = tr.select_one('[data-stat="game_remarks"]').get_text()
boxScoreLink = ''
try:
boxScoreLink = 'https://www.basketball-reference.com' + tr.select_one(
'[data-stat="box_score_text"] > a').get("href")
except:
date2 = datetime.strptime(date, '%a, %b %d, %Y')
month = date2.month
day = date2.day
if month < 10:
month = '0' + str(month)
else:
month = str(month)
if day < 10:
day = '0' + str(day)
else:
day = str(day)
homeShortName = searchTeamByName(homeTeamName)[2]
boxScoreLink = 'https://www.basketball-reference.com/boxscores/' + \
str(date2.year) + month + day + \
'0' + homeShortName + '.html'
storeValue = [
id, homeTeamId, homeTeamScore, awayTeamId, awayTeamScore, date,
overtime, attendance, playoff, season, notes
]
boxScoreLinks.append([
id, boxScoreLink, awayTeamId, homeTeamId, season, playoff, date
])
id += 1
# if playoff:
# playoffGames.append(storeValue)
# else:
games.append(storeValue)
def getPlayerData(url):
html = requests.get(url[1])
soup = BeautifulSoup(html.content, 'html.parser')
team_ids = soup.select('#per_game > tfoot > tr [data-stat=team_id]')
div = soup.select_one('#meta > div[itemtype="https://schema.org/Person"]')
teams = []
fullName = ''
name = ''
current_team_id = ''
birth_place = ''
birth_date = ''
height = ''
weight = ''
playerUrl = '/'.join(url[1].split('/')[-3:])
retired = url[3]
name = div.select_one('h1').get_text().strip()
injury = False
for team_id in team_ids:
shortName = team_id.get_text()
if searchTeamByShortName(shortName) != None:
teams.append(int(searchTeamByShortName(shortName)))
if div.select_one(
'p:nth-child(2) > strong').get_text().strip() == 'Pronunciation':
fullName = div.select_one('p:nth-child(3) > strong').get_text().strip()
else:
fullName = div.select_one('p:nth-child(2) > strong').get_text().strip()
try:
height = div.select_one('span[itemprop="height"]').get_text()
except:
pass
try:
weight = div.select_one('span[itemprop="weight"]').get_text()[:-2]
except:
pass
try:
birth_date = div.select_one('#necro-birth > a:first-child').get_text(
).strip() + " " + div.select_one(
'#necro-birth > a:nth-child(2)').get_text().strip()
except:
pass
try:
find_current_team = div.select_one('[href*="/2020.html"]').get_text()
current_team_id = searchTeamByName(find_current_team)[0]
if len(teams) == 0 and current_team_id:
teams.append(current_team_id)
except:
pass
try:
birth_place = div.select_one(
'span[itemprop="birthPlace"]').get_text().strip()[3:]
except:
pass
if soup.select_one('#injury'):
injury = True
with open('players.csv', 'a', newline='') as file:
writer = csv.writer(file)
writer.writerow([
url[0], name, fullName, url[2], height, weight, birth_date,
birth_place, retired, playerUrl, current_team_id, teams, injury
])