python爬取双色球每期中奖号码（原创）

#-*—coding:utf8-*-
import re
import requests
import mysql.connector
import sys
reload(sys)
sys.setdefaultencoding("utf-8")

url1= "http://kaijiang.500.com/shtml/ssq/"
url2=".shtml?0_ala_baidu"
print "數據庫開始連接"
conn=mysql.connector.connect(host='127.0.0.1',
                                   user='root',
                                   passwd='123456',
                                   db='test')
cursor = conn.cursor()
print "數據庫已經連接"
num=0
for j in range (3000,19000,1000):
    for i in range(j+1,j+1000,1):
        url=url1+bytes(i)+url2
        # print url
        html = requests.get(url)
        # print html.text
        title = re.findall('<li class="ball_red">(.*?)</li>', html.text, re.S)
        # print "第",i,"期"
        # print "红球是:"
        for each in title:
            print each,
        title1 = re.findall('<li class="ball_blue">(.*?)</li>', html.text, re.S)
        # print "篮球是:",
        # for each in title1:
        #     print each

        insert_ssq = ("INSERT INTO ssq(rq ,hq1,hq2,hq3,hq4,hq5,hq6,lq) VALUES(%s,%s,%s,%s,%s,%s,%s,%s)")
        data_ssq = None
        if len(title)<6:
            print   "第",i,"条数据开始插入失败"
            break
        data_ssq = [(i,title[0], title[1], title[2],title[3],title[4],title[5],title1[0])]
        # cursor.execute(insert_ssq, data_ssq)
        # num += 1
        # conn.commit()
        try:
            cursor.executemany(insert_ssq, data_ssq)
            num+=1
            conn.commit()
            print "成功插入第",i,"期"
        except:
            conn.rollback()
            print "第",i,"期,插入失败"
# 使用execute方法执行SQL语句
# cursor.execute('select * from student where 1=1')  # 表查询

# 使用 fetchone() 方法获取一条数据库。
# values = cursor.fetchall()

# print(values)

# 关闭数据库连接
print "关闭数据库连接"
print "插入了",num,"条数据"
cursor.close()

刚学习python，这个蛮有趣。

本人不研究彩票，只是别人所托。。。