def getTeamsInfo(): # base url url = "http://g.hupu.com/nba/players/" # teams teams = [ "Spurs", "Rockets", "Mavericks", "Grizzlies", "Pelicans", "Clippers", "Warriors", "Suns", "Lakers", "Kings", "Blazers", "Thunder", "Nuggets", "Timberwolves", "Jazz", "Raptors", "Nets", "Knicks", "Celtics", "76ers", "Heat", "Hawks", "Wizards", "Bobcats", "Magic", "Pacers", "Bulls", "Pistons", "Cavaliers", "Bucks", ] # team dictionary teamDic = {} # get all html info divide by teamName for team in teams: teamUrl = url + team teamPlayerInfos = CommonUtils.getSoupFromUrl(teamUrl) teamDic[team] = teamPlayerInfos return teamDic
def getTeamUrlContent(): teamUrl = "http://g.hupu.com/nba/teams/" #teams teams = ['Spurs','Rockets','Mavericks','Grizzlies','Pelicans','Clippers', 'Warriors','Suns','Lakers','Kings','Blazers','Thunder','Nuggets', 'Timberwolves','Jazz','Raptors','Nets','Knicks','Celtics','76ers', 'Heat','Hawks','Wizards','Bobcats','Magic','Pacers','Bulls', 'Pistons','Cavaliers','Bucks'] teamDic = {} for teamName in teams: url = teamUrl + teamName teamInfo = CommonUtils.getSoupFromUrl(url) teamDic[teamName] = teamInfo return teamDic
def getUrlContent(): return CommonUtils.getSoupFromUrl("http://g.hupu.com/nba/daily/boxscore_23325.html")
def init(): scheduleUrl = "http://nba.sports.sina.com.cn/match_result.php?day=all" return CommonUtils.getSoupFromUrl(scheduleUrl)
#coding=utf-8 #!/usr/bin/python import os, re import CommonUtils url = 'http://nba.sports.sina.com.cn/look_scores.php?id=2006110120' bs = CommonUtils.getSoupFromUrl(url) # client and home teams = bs.find_all("a", 'tlogo') # client client = teams[0].string.strip() # home home = teams[1].string.strip() # final points total_points = bs.find_all("td", "num") # sec points sec_points = bs.find_all("td", {"bgcolor": "#000000"}) # date date = url.split("=")[1][0:8] print client,home,total_points,sec_points