def get_game_info(url_path, match_date):
url_path = 'http://www.nba.com' + url_path
print url_path
html = get_response(url_path)
get_box_score(html, url_path, match_date)
pass
# get_game_info('http://www.nba.com/games/20111220/DETCLE/gameinfo.html', '2011-12-30')
def print_url(match_date, format_match_date):
url_path = 'http://www.nba.com/gameline/' + match_date + '/'
print url_path
html = get_response(url_path)
if html:
soup = BeautifulSoup(html)
game_set = set()
for tag in soup.find_all(href=re.compile("^/games/" + match_date + "/")):
if tag.string == 'recap':
game_set.add(tag['href'])
for tag in game_set:
get_game_info(tag, format_match_date)
pass
__author__ = 'chenghao'
from my_urllib import get_response
import json
import time
import datetime
from mysql_connector import get_connector
import mysql.connector
conn = get_connector()
while True:
url_path = 'http://china.nba.com/wap/static/data/season/schedule_7.json'
html = get_response(url_path)
if html:
dates = json.loads(html)['payload']['dates']
for date in dates:
utcMillis = date['utcMillis']
utcMillis = utcMillis.strip()
utcMillis = int(utcMillis)
game_date = datetime.date.fromtimestamp(utcMillis / 1000).isoformat()
games = date['games']
for game in games:
status = game['boxscore']['status']
status = status.strip()
status = int(status)
if status != 1:
sql = 'update next_match set status = %s'
cursor = conn.cursor()
cursor.execute(sql, (status,))
cursor.close()
conn.commit()
def game_live(conn,gameid):
request_time = 0
clean_flag = True
init = False
sql = 'update next_match set status = 2 where gameid = %s'
cursor = conn.cursor()
cursor.execute(sql,(gameid,))
cursor.close()
conn.commit()
conn.close()
while True:
url_path = 'http://china.nba.com/wap/static/data/game/snapshot_'+str(gameid)+'.json'
html = get_response(url_path)
if not html :
request_time += 1
clean_flag = True
if request_time < 30:
continue
else:
break
html = html.decode("utf-8-sig")
payload = json.loads(html)
payload = payload['payload']
status = payload['boxscore']['status']
status = int(status)
if not init:
if clean_flag:
sql = 'delete from live where gameid = %s'
cursor = conn.cursor()
cursor.execute(sql,(gameid,))
cursor.close()
conn.commit()
clean_flag = False
period = payload['boxscore']['period']
period = int(period)
for i in range(1,period,1):
data = []
url_path = 'http://china.nba.com/wap/static/data/game/playbyplay_'+gameid+'_'+unicode(i)+'.json'
html = get_response(url_path)
if not html:
request_time += 1
clean_flag = True
if request_time < 10:
continue
else:
break
html = html.decode("utf-8-sig")
playbyplay = json.loads(html)
payload = playbyplay['payload']
events = payload['playByPlays'][0]['events']
for j in range(len(events) - 1,-1,-1):
event = events[j]
awayScore = unicode(event['awayScore'])
homeScore = unicode(event['homeScore'])
gameClock = unicode(event['gameClock'])
description = unicode(event['description'])
playerId = unicode(event['playerId'])
teamId = unicode(event['teamId'])
m = (gameid,i,gameClock,awayScore,homeScore,description,teamId,playerId,len(events) - j)
print m
data.append(m)
cursor = conn.cursor()
sql = 'insert into live VALUES (%s,%s,%s,%s,%s,%s,%s,%s,%s)'
cursor.executemany(sql,data)
cursor.close()
conn.commit()
if clean_flag:
continue
init = True
loop_time404 = 0
while True:
data = []
url_path = 'http://china.nba.com/wap/static/data/game/playbyplay_'+gameid+'_'+unicode(period)+'.json'
html = get_response(url_path)
if not html:
loop_time404 += 1
if loop_time404 < 1000:
continue
else:
break
html = html.decode("utf-8-sig")
playbyplay = json.loads(html)
payload = playbyplay['payload']
events = payload['playByPlays'][0]['events']
sql = 'select * from live where gameid = %s and period = %s'
cursor = conn.cursor()
cursor.execute(sql,(gameid,period,))
values = cursor.fetchall()
for i in range(len(events) - len(values) -1,-1,-1):
event = events[i]
awayScore = unicode(event['awayScore'])
homeScore = unicode(event['homeScore'])
gameClock = unicode(event['gameClock'])
description = unicode(event['description'])
playerId = unicode(event['playerId'])
teamId = unicode(event['teamId'])
m = (gameid,period,gameClock,awayScore,homeScore,description,teamId,playerId,len(events)-i)
print m
data.append(m)
cursor = conn.cursor()
sql = 'insert into live VALUES (%s,%s,%s,%s,%s,%s,%s,%s,%s)'
cursor.executemany(sql,data)
cursor.close()
conn.commit()
now_period = payload['boxscore']['period']
now_period = int(now_period)
if now_period > period:
period = now_period
continue
if status == 3:
sql = 'update next_match set status = 3 where gameid = %s'
cursor = conn.cursor()
cursor.execute(sql,(gameid,))
cursor.close()
conn.commit()
conn.close()
return
conn.close()
