统计hihoCoder挑战赛成绩的Python脚本

# -*- coding: utf-8 -*-
from bs4 import BeautifulSoup
import urllib
def updateScoreList(scoreList, url):
    """
    :param scoreList: 
    :param url: 
    :return:
    """
    page = urllib.urlopen(url)
    soup = BeautifulSoup(page)
    contestants = soup.find_all(class_="fn-ell")
    for person in contestants:
        userName = person.string
        row = person.parent.parent # table row for this contestant
        statics = row.find_all(‘td‘)
        scoreTag = statics[2]
        score = int(unicode(scoreTag.string))
        if userName in scoreList:
            scoreList[userName][‘score‘] += score
            scoreList[userName][‘numOfContests‘] += 1
        else:
            scoreList[userName] = {‘score‘:score,‘numOfContests‘:1}

def generateRankList(urlList, numOfWinners):
    """
    :param urlList: url of hihocoder challenge contest rank pages
    :param numOfWinners: number of winners
    :return: the first numOfWinners winners in these contests
    """
    scoreList = {}
    for url in urlList:
        updateScoreList(scoreList, url)
    rank = []
    for userName, person in scoreList.items():
        rank.append((userName, person[‘score‘], person[‘numOfContests‘]))
    rank.sort(cmp=lambda p1, p2:p2[1]-p1[1])
    return rank[:numOfWinners]

def main():
    urlList = []
    for num in range(8, 9):
        urlList.append("http://hihocoder.com/contest/challenge%d/rank" % num)
    rankList = generateRankList(urlList, 8)
    print u"用户名 总分 参赛次数"
    for person in rankList:
       print person

if __name__ == "__main__":
    main()