环境：python3
目的：爬取15年的双色球开奖号以及期数和开奖日期：
上代码
#!/usr/bin/env python3
#-*-coding:utf-8-*-
# @Author : 杜文涛
# @Time : 2018/4/19 16:01
# @File : cpssq.py
#彩票双色球数据
import requests
import re
import xlwt
import time
def get_all_page():
global all_page
url = "http://kaijiang.zhcw.com/zhcw/html/ssq/list_1.html"
reponse = requests.get(url=url)
reponse.encoding='utf-8'
html = reponse.text
all_page = int(re.findall(r"class=\"pg\".*?<strong>(.*?)</strong>",html)[0])
return all_page
def get_num():
k = -1
f = xlwt.Workbook(encoding='utf-8')
sheet01 = f.add_sheet(u'sheel1', cell_overwrite_ok=True)
for page_num in range(1,all_page):
url = "http://kaijiang.zhcw.com/zhcw/html/ssq/list_"+str(page_num)+".html"
reponse = requests.get(url=url)
time.sleep(5)
reponse.encoding = 'utf-8'
html = reponse.text
rule = r"<tr>.*?<td align=\"center\">(.*?)</td>.*?<td align=\"center\">(.*?)</td>.*?<td align=\"center\" style=\"padding-left:10px;\">.*?<em class=\"rr\">(.*?)</em>.*?<em class=\"rr\">(.*?)</em>.*?<em class=\"rr\">(.*?)</em>.*?<em class=\"rr\">(.*?)</em>.*?<em class=\"rr\">(.*?)</em>.*?<em class=\"rr\">(.*?)</em>.*?<em>(.*?)</em></td>"
num = re.findall(rule, html, re.S | re.M)
# f = xlwt.Workbook(encoding='utf-8')
# sheet01 = f.add_sheet(u'sheel1', cell_overwrite_ok=True)
sheet01.write(0, 0, "日期")
sheet01.write(0, 1, "期数")
sheet01.write(0, 2, "第一个红球")
sheet01.write(0, 3, "第二个红球")
sheet01.write(0, 4, "第三个红球")
sheet01.write(0, 5, "第四个红球")
sheet01.write(0, 6, "第五个红球")
sheet01.write(0, 7, "第六个红球")
sheet01.write(0, 8, "蓝球")
print("正在写入第%s页" % (page_num))
for i in range(0,len(num)):
k += 1
sheet01.write(k + 1, 0, num[i][0])
sheet01.write(k + 1, 1, num[i][1])
sheet01.write(k + 1, 2, num[i][2])
sheet01.write(k + 1, 3, num[i][3])
sheet01.write(k + 1, 4, num[i][4])
sheet01.write(k + 1, 5, num[i][5])
sheet01.write(k + 1, 6, num[i][6])
sheet01.write(k + 1, 7, num[i][7])
sheet01.write(k + 1, 8, num[i][8])
f.save("双色球统计结果.xls")
if __name__ == '__main__':
get_all_page()
get_num()
运行后的结果：
说明：最后一页数据较少，就没要；如果想要修改成下面的代码：
for page_num in range(1,all_page+1):
关注微信公众号回复“彩票”获取源代码和数据
微信公众号：
原文地址：http://blog.51cto.com/tdcqvip/2105499
时间： 2024-07-29 16:20:12