[No.003-6]爬蟲網(wǎng)易賠率數(shù)據(jù)并導(dǎo)入到mysql數(shù)據(jù)庫(kù)

lbool 發(fā)布于2019-07-24 17:38 / 808人閱讀

摘要：文件的全路徑返回對(duì)象獲取所有即將讀取的的全路徑獲取場(chǎng)次獲取賽事類型獲取比賽日期獲取主隊(duì)客隊(duì)名稱獲取勝平負(fù)賠率刪除首尾兩個(gè)無效數(shù)據(jù)比分結(jié)果以及比分結(jié)果賠率更新勝平負(fù)勝其他平其他負(fù)其他更新總進(jìn)球勝其他負(fù)其他場(chǎng)次賽事類型比賽日期比

#encoding:utf-8
#!/usr/local/bin/python2.7
import urllib2
from bs4 import BeautifulSoup
import MySQLdb as mdb
import sys
import re
import os

#URL:HTML文件的全路徑
#返回：BeautifulSoup對(duì)象
def getSoup(url):
    return BeautifulSoup(open(url))

#獲取所有即將讀取的HTML的全路徑
def fileNames():
    temp_f = open("/root/bet/names.txt")
    temp = []
    for line in temp_f:
        temp.append("/root/bet/urls/"+line.strip())
    temp_f.close()
    return temp

#獲取場(chǎng)次
def getScr(soup):
    scr = []
    temp =[]
    for trs in soup.findAll("tr"):
        for tds in trs.findAll("td",{"width":"50"}):
            temp.append(tds.string)
    lt = len(temp)/4
    for i in range(lt):
        scr.append(temp[4*i+3])
    return scr

#獲取賽事類型
def getLea(soup):
    league = []
    for item in soup.findAll("tr"):
        for item1 in item.findAll("td",{"width":"70"}):
            league.append(item1.string.encode("utf-8"))
    return league

#獲取比賽日期
def getGmdate(soup):
    temp = []
    gmdate = []
    for item in soup.findAll("tr"):
        for item1 in item.findAll("td",{"width":"61"}):
            temp.append(item1)
    temp =temp[3:]
    for item in temp:
        gmdate.append(re.search("w{4}-w{2}-w{2}",str(item)).group())
    return gmdate

#獲取主隊(duì)客隊(duì)名稱
def getTeam(soup):
    team=[]
    for item in soup.findAll("tr"):
        for teams in item.findAll("a",{"class":"dui"}):
            team.append(teams.string.strip().encode("utf-8"))
    return team

#獲取勝平負(fù)賠率
def getSpfpl(soup):
    spfpl =[]
    temp = []
    for trs in soup.findAll("tr"):
        for tds in trs.findAll("span"):
            temp.append(tds.string)
#刪除首尾兩個(gè)無效數(shù)據(jù)
    temp =temp[1:-1]
    for i in range(len(temp)):
        for item in temp[7*i+4:7*i+7]:
            spfpl.append(item)
    return spfpl

#6.比分結(jié)果以及比分結(jié)果賠率
def getResult(soup):
    bfjg = []
    temp = []
    for trs in soup.findAll("tr"):
        for item in trs.findAll("div",{"align":"center"}):
            for item1 in item.findAll("strong"):
                bfjg.append(item1.string.encode("utf-8"))
    return bfjg



if __name__=="__main__":
    reload(sys)
    sys.setdefaultencoding("utf-8")
    names = fileNames()
    conn=mdb.connect(host="localhost",user="root",passwd="oracle",db="betdb",port=3306)
    cur = conn.cursor()
    SQL="insert into results(id,lea,gmd,hos,gue,win,dog,los,res,odd) values(%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)"
    #更新勝平負(fù)
    sql_update3="update results set spf=3 where res="1:0" or res="2:0" or res="2:1" or res="3:0" or res="3:1" or res="3:2" or res="4:0" or res="4:1" or res="4:2" or res="5:0" or res="5:1" or res="5:2" or res="勝其他";"
    sql_update1="update results set spf=1 where res="0:0" or res="1:1" or res="2:2" or res="3:3" or res="平其他";"
    sql_update0="update results set spf=0 where res="0:1" or res="0:2" or res="1:2" or res="0:3" or res="1:3" or res="2:3" or res="0:4" or res="1:4" or res="2:4" or res="0:5" or res="1:5" or res="2:5" or res="負(fù)其他";"
    #更新總進(jìn)球
    sql_updatezjq0="update results set zjq=0 where res="0:0""
    sql_updatezjq1="update results set zjq=1 where res="1:0" or res="0:1""
    sql_updatezjq2="update results set zjq=2 where res="2:0" or res="1:1" or res="0:2""
    sql_updatezjq3="update results set zjq=3 where res="3:0" or res="2:1" or res="1:2" or res="0:3""
    sql_updatezjq4="update results set zjq=4 where res="4:0" or res="3:1" or res="2:2" or res="1:3" or res="0:4""
    sql_updatezjq5="update results set zjq=5 where res="5:0" or res="4:1" or res="3:2" or res="2:3" or res="0:5" or res="1:4""
    sql_updatezjq6="update results set zjq=6 where res="5:1" or res="3:3" or res="4:2" or res="2:4" or res="1:5""
    sql_updatezjq7="update results set zjq=7 where res="勝其他" or res="負(fù)其他""
    for htmlfilename in names:
        soup = getSoup(htmlfilename)
                print "Reading %s now ..." % htmlfilename
        #1.場(chǎng)次
        scr = getScr(soup)
        #2.賽事類型
        lea = getLea(soup)
        #3.比賽日期
        gmd = getGmdate(soup)
        #4.比賽隊(duì)伍
        tea = getTeam(soup)
        #5.勝平負(fù)賠率
        spf = getSpfpl(soup)
        #6.比賽結(jié)果
        bfj = getResult(soup)
        #7.形成場(chǎng)次日期唯一id
        ids=[]
        for i in range(len(scr)):
            ids.append(gmd[i]+"-"+scr[i])
        #裝配結(jié)果集
        res =[]
        for i in range(len(scr)):
            res.append(ids[i])
            res.append(lea[i])
            res.append(gmd[i])
            res.append(tea[2*i])
            res.append(tea[2*i+1])
            res.append(spf[3*i])
            res.append(spf[3*i+1])
            res.append(spf[3*i+2])
            res.append(bfj[2*i])
            res.append(bfj[2*i+1])
        l =len(res)/10
        for i in range(l):
            for item in res[10*i:10*i+1]:
                cur.execute(SQL,res[10*i:10*i+10])
                i+=1
        cur.execute(sql_update3)
        cur.execute(sql_update1)
        cur.execute(sql_update0)    
        cur.execute(sql_updatezjq0)
        cur.execute(sql_updatezjq1)
        cur.execute(sql_updatezjq2)
        cur.execute(sql_updatezjq3)
        cur.execute(sql_updatezjq4)
        cur.execute(sql_updatezjq5)
        cur.execute(sql_updatezjq6)
        cur.execute(sql_updatezjq7)

#SQL創(chuàng)建語句
CREATE TABLE `results` (
  `id` char(20) NOT NULL,
  `lea` char(100) DEFAULT NULL,
  `gmd` date DEFAULT NULL,
  `hos` char(100) DEFAULT NULL,
  `gue` char(100) DEFAULT NULL,
  `win` float(5,2) DEFAULT NULL,
  `dog` float(5,2) DEFAULT NULL,
  `los` float(5,2) DEFAULT NULL,
  `res` char(10) DEFAULT NULL,
  `odd` float(5,2) DEFAULT NULL,
  `zjq` tinyint(4) DEFAULT NULL,
  `spf` tinyint(4) DEFAULT NULL
) ENGINE=MyISAM DEFAULT CHARSET=utf8;