爬虫网易财经上市公司财务数据

声明:代码仅作学习交流用途,代码分享者与创作者不承担任何由他人恶意运行而导致的责任,勿擅自修改限制频率的参数,勿恶意攻击网页,请学习浏览者遵守社会公德与法律秩序,爬虫导致的网页崩溃等损失由计算机操作者负全部责任,造成严重后果的需要承担刑事责任
爬虫代写:邮箱 [email protected]

import requests
from lxml import etree
from time import sleep
from fake_useragent import UserAgent
import pandas as pd
from random import randint

place_name = pd.read_csv(r'C:\Users\Admin\PycharmProjects\untitled\股票数据\上市公司代码.csv',encoding='utf-8')
length = len(place_name)
for i in range(length):
    print(i)
    UA = UserAgent().edge
    url_base='http://quotes.money.163.com/f10/zycwzb_{},year.html'
    name_code = place_name.iloc[i,0]
    url = url_base.format(name_code)
    headers={
    
    
        'User-Agent':UA
    }
    response = requests.get(url,headers=headers)
 #   print(response.status_code)
 #   print(response.encoding)
    sleep(randint(1,2))
  #  print(response.text)
    e = etree.HTML(response.text)
    v1 = e.xpath('''//div[@class='col_r']/table/tr[2]/td/text()''')
    v2 = e.xpath('''//div[@class='col_r']/table/tr[3]/td/text()''')
    v3 = e.xpath('''//div[@class='col_r']/table/tr[4]/td/text()''')
    v4 = e.xpath('''//div[@class='col_r']/table/tr[5]/td/text()''')
    v5 = e.xpath('''//div[@class='col_r']/table/tr[6]/td/text()''')
    v6 = e.xpath('''//div[@class='col_r']/table/tr[7]/td/text()''')
    v7 = e.xpath('''//div[@class='col_r']/table/tr[8]/td/text()''')
    v8= e.xpath('''//div[@class='col_r']/table/tr[9]/td/text()''')
    v9 = e.xpath('''//div[@class='col_r']/table/tr[10]/td/text()''')
    v10 = e.xpath('''//div[@class='col_r']/table/tr[11]/td/text()''')
    v11 = e.xpath('''//div[@class='col_r']/table/tr[12]/td/text()''')
    v12 = e.xpath('''//div[@class='col_r']/table/tr[13]/td/text()''')
    v13 = e.xpath('''//div[@class='col_r']/table/tr[14]/td/text()''')
    v14 = e.xpath('''//div[@class='col_r']/table/tr[15]/td/text()''')
    v15= e.xpath('''//div[@class='col_r']/table/tr[16]/td/text()''')
    v16 = e.xpath('''//div[@class='col_r']/table/tr[17]/td/text()''')
    v17 = e.xpath('''//div[@class='col_r']/table/tr[18]/td/text()''')
    v18 =e.xpath('''//div[@class='col_r']/table/tr[19]/td/text()''')
    v19 = e.xpath('''//div[@class='col_r']/table/tr[20]/td/text()''')
    all_info = {
    
    
        '基本每股收益(元)': v1,
        '每股净资产(元)': v2,
        '每股经营活动产生的现金流量净额(元)':v3,
        '主营业务收入(万元)': v4,
        '主营业务利润(万元)': v5,
        '营业利润(万元)': v6,
        '投资收益(万元)':v7,
        '营业外收支净额(万元)':v8,
        '利润总额(万元)':v9,
        '净利润(万元)':v10,
        '净利润(扣除非经常性损益后)(万元)':v11,
        '经营活动产生的现金流量净额(万元)':v12,
        '现金及现金等价物净增加额(万元)':v13,
        '总资产(万元)':v14,
        '流动资产(万元)':v15,
        '总负债(万元)':v16,
        '流动负债(万元)':v17,
        '股东权益不含少数股东权益(万元)':v18,
        '净资产收益率加权(%)':v19
    }
    outdata = pd.DataFrame(all_info)
    outdata.to_csv('C:\\Users\\Admin\\PycharmProjects\\untitled\\股票数据\\{}财务1.csv'.format(name_code), encoding='GBK')
    print('已保存111111111111111111111111111')


    v1 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[1]/td/text()''')
    v2 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[2]/td/text()''')
    v3 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[3]/td/text()''')
    v4 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[4]/td/text()''')
    v5 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[5]/td/text()''')
    v6 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[6]/td/text()''')
    v7 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[7]/td/text()''')
    v8 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[8]/td/text()''')
    v9 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[9]/td/text()''')
    v10 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[10]/td/text()''')
    v11 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[11]/td/text()''')
    v12 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[12]/td/text()''')
    v13 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[13]/td/text()''')
    v14 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[14]/td/text()''')
    v15 = e.xpath('''/html/body/div[2]/div[4]/div[3]/div[4]/table/tr[15]/td/text()''')

    all_info = {
    
    
        '基本每股收益(元)': v1,
        '每股净资产(元)': v2,
        '每股经营活动产生的现金流量净额(元)':v3,
        '主营业务收入(万元)': v4,
        '主营业务利润(万元)': v5,
        '营业利润(万元)': v6,
        '投资收益(万元)':v7,
        '营业外收支净额(万元)':v8,
        '利润总额(万元)':v9,
        '净利润(万元)':v10,
        '净利润(扣除非经常性损益后)(万元)':v11,
        '经营活动产生的现金流量净额(万元)':v12,
        '现金及现金等价物净增加额(万元)':v13,
        '总资产(万元)':v14,
        '流动资产(万元)':v15
    }
    outdata = pd.DataFrame(all_info)
    outdata.to_csv('C:\\Users\\Admin\\PycharmProjects\\untitled\\股票数据\\{}财务2.csv'.format(name_code), encoding='GBK')
    print('已保存222222222222222222222222222222222222222222')



    v1 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[1]/td/text()''')
    v2 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[2]/td/text()''')
    v3 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[3]/td/text()''')
    v4 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[4]/td/text()''')
    v5 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[5]/td/text()''')
    v6 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[6]/td/text()''')
    v7 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[7]/td/text()''')
    v8 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[8]/td/text()''')
    v9 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[9]/td/text()''')
    v10 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[10]/td/text()''')
    v11 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[11]/td/text()''')
    v12 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[12]/td/text()''')
    v13 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[13]/td/text()''')
    v14 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[14]/td/text()''')
    v15 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[15]/td/text()''')
    v16 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[16]/td/text()''')
    v17 = e.xpath('''/html/body/div[2]/div[4]/div[5]/div[4]/table/tr[17]/td/text()''')

    all_info = {
    
    
        '基本每股收益(元)': v1,
        '每股净资产(元)': v2,
        '每股经营活动产生的现金流量净额(元)':v3,
        '主营业务收入(万元)': v4,
        '主营业务利润(万元)': v5,
        '营业利润(万元)': v6,
        '投资收益(万元)':v7,
        '营业外收支净额(万元)':v8,
        '利润总额(万元)':v9,
        '净利润(万元)':v10,
        '净利润(扣除非经常性损益后)(万元)':v11,
        '经营活动产生的现金流量净额(万元)':v12,
        '现金及现金等价物净增加额(万元)':v13,
        '总资产(万元)':v14,
        '流动资产(万元)':v15,
        '总负债(万元)':v16,
        '流动负债(万元)':v17
    }
    outdata = pd.DataFrame(all_info)
    outdata.to_csv('C:\\Users\\Admin\\PycharmProjects\\untitled\\股票数据\\{}财务3.csv'.format(name_code), encoding='GBK')
    print('已保存3333333333333333333333333333333333333333333333')



    v1 = e.xpath('''/html/body/div[2]/div[4]/div[7]/div[4]/table/tr[1]/td/text()''')
    v2 = e.xpath('''/html/body/div[2]/div[4]/div[7]/div[4]/table/tr[2]/td/text()''')
    v3 = e.xpath('''/html/body/div[2]/div[4]/div[7]/div[4]/table/tr[3]/td/text()''')
    v4 = e.xpath('''/html/body/div[2]/div[4]/div[7]/div[4]/table/tr[4]/td/text()''')
    all_info = {
    
    
        '基本每股收益(元)': v1,
        '每股净资产(元)': v2,
        '每股经营活动产生的现金流量净额(元)':v3,
        '主营业务收入(万元)': v4
    }
    outdata = pd.DataFrame(all_info)
    outdata.to_csv('C:\\Users\\Admin\\PycharmProjects\\untitled\\股票数据\\{}财务4.csv'.format(name_code), encoding='GBK')
    print('已保存44444444444444444444444444444444444444444444444444444')




    v1 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[1]/td/text()''')
    v2 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[2]/td/text()''')
    v3 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[3]/td/text()''')
    v4 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[4]/td/text()''')
    v5 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[5]/td/text()''')
    v6 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[6]/td/text()''')
    v7 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[7]/td/text()''')
    v8 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[8]/td/text()''')
    v9 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[9]/td/text()''')
    v10 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[10]/td/text()''')
    v11 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[11]/td/text()''')
    v12 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[12]/td/text()''')
    v13 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[13]/td/text()''')
    v14 = e.xpath('''/html/body/div[2]/div[4]/div[9]/div[4]/table/tr[14]/td/text()''')
    all_info = {
    
    
        '基本每股收益(元)': v1,
        '每股净资产(元)': v2,
        '每股经营活动产生的现金流量净额(元)':v3,
        '主营业务收入(万元)': v4,
        '主营业务利润(万元)': v5,
        '营业利润(万元)': v6,
        '投资收益(万元)':v7,
        '营业外收支净额(万元)':v8,
        '利润总额(万元)':v9,
        '净利润(万元)':v10,
        '净利润(扣除非经常性损益后)(万元)':v11,
        '经营活动产生的现金流量净额(万元)':v12,
        '现金及现金等价物净增加额(万元)':v13,
        '总资产(万元)':v14
    }
    outdata = pd.DataFrame(all_info)
    outdata.to_csv('C:\\Users\\Admin\\PycharmProjects\\untitled\\股票数据\\{}财务5.csv'.format(name_code), encoding='GBK')
    print('已保存55555555555')

猜你喜欢

转载自blog.csdn.net/qq_42830971/article/details/112378070