应用python创建IP代理池，并存入excel表格中

最新推荐文章于 2020-07-11 12:59:01 发布

qq_43279936

最新推荐文章于 2020-07-11 12:59:01 发布

阅读量225

点赞数

本文链接：https://blog.csdn.net/qq_43279936/article/details/99702376

版权

直接上代码：

import requests
from bs4 import BeautifulSoup

def get_content(url):
    user_agent='Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Firefox/45.0'
    response=requests.get(url,headers={'User-Agent':user_agent})
    response.raise_for_status()
    response.encoding=response.apparent_encoding
    return response.content

def parser_html(htmlcontent):
    soup=BeautifulSoup(htmlcontent,'html.parser')
    tableol=soup.find_all('table',{'id':'ip_list'})[0]
    trobjs=tableol.find_all('tr',{'class':'odd'})
    # print(trobjs)
    for trobj in trobjs:
        # print(trobj)
        ipname=trobj.find_all('td',limit=3)[1].get_text()

    ipport=trobj.find_all('td',limit=3)[2].get_text()
    # print(ipport)
    global xiciinfo
    xiciinfo.append((ipname,ipport))
return xiciinfo
import openpyxl

def create_to_excel(wbname,data,sheetname='Sheet1'):
    print('正在创建表格%s'%(wbname))
    wb=openpyxl.Workbook()
    sheet=wb.active
    sheet.title=sheetname
    print("正在写入数据...")
    for row,item in enumerate(data):
        # print(item)

    cell=sheet.cell(row=row+1,column=1,value=item)
    cell.value=item
wb.save(wbname)
print('保存工作薄%s成功...'%(wbname))






if __name__ == '__main__':
    xiciinfo=[]
    url='https://www.xicidaili.com/'
    htmlcontent=get_content(url)
    contents=parser_html(htmlcontent)
    # mulu=zip(contents)
    mulu=[]
    # print(mulu)
    for content in contents:
        dd=content[0]+':'+content[1]
        # print(dd)
        mulu.append(dd)
    # print(mulu)
    create_to_excel('doc/ipaddr.xlsx',mulu,sheetname='西刺IP池')