Python实现实时数据采集新型冠状病毒数据实例


Posted in Python onFebruary 04, 2020

Python实时数据采集-新型冠状病毒

源代码 来源:https://github.com/Programming-With-Love/2019-nCoV

疫情数据时间为:2020.2.1

Python实现实时数据采集新型冠状病毒数据实例

项目相关截图:

全国数据展示

Python实现实时数据采集新型冠状病毒数据实例

国内数据展示

Python实现实时数据采集新型冠状病毒数据实例

国外数据展示

Python实现实时数据采集新型冠状病毒数据实例

查看指定区域详细数据

Python实现实时数据采集新型冠状病毒数据实例

源代码,注意安装所需模块(例如 pip install 模块名)

import requests

import re

from bs4 import BeautifulSoup

from time import sleep

import json

from prettytable import ALL

from prettytable import PrettyTable

 

hubei = {}

guangdong = {}

zhejiang = {}

beijing = {}

shanghai = {}

hunan = {}

anhui = {}

chongqing = {}

sichuan = {}

shandong = {}

guangxi = {}

fujian = {}

jiangsu = {}

henan = {}

hainan = {}

tianjin = {}

jiangxi = {}

shanxi1 = {} # 陕西

guizhou = {}

liaoning = {}

xianggang = {}

heilongjiang = {}

aomen = {}

xinjiang = {}

gansu = {}

yunnan = {}

taiwan = {}

shanxi2 = {} # 山西

jilin = {}

hebei = {}

ningxia = {}

neimenggu = {}

qinghai = {} # none

xizang = {} # none

provinces_idx = [hubei, guangdong, zhejiang, chongqing, hunan, anhui, beijing,

         shanghai, henan, guangxi, shandong, jiangxi, jiangsu, sichuan,

         liaoning, fujian, heilongjiang, hainan, tianjin, hebei, shanxi2,

         yunnan, xianggang, shanxi1, guizhou, jilin, gansu, taiwan,

         xinjiang, ningxia, aomen, neimenggu, qinghai, xizang]

map = {

  '湖北':0, '广东':1, '浙江':2, '北京':3, '上海':4, '湖南':5, '安徽':6, '重庆':7,

  '四川':8, '山东':9, '广西':10, '福建':11, '江苏':12, '河南':13, '海南':14,

  '天津':15, '江西':16, '陕西':17, '贵州':18, '辽宁':19, '香港':20, '黑龙江':21,

  '澳门':22, '新疆':23, '甘肃':24, '云南':25, '台湾':26, '山西':27, '吉林':28,

  '河北':29, '宁夏':30, '内蒙古':31, '青海':32, '西藏':33

}

 

 

def getTime(text):

  TitleTime = str(text)

  TitleTime = re.findall('<span>(.*?)</span>', TitleTime)

  return TitleTime[0]

 

def getAllCountry(text):

  AllCountry = str(text)

  AllCountry = AllCountry.replace("[<p class=\"confirmedNumber___3WrF5\"><span class=\"content___2hIPS\">", "")

  AllCountry = AllCountry.replace("<span style=\"color: #4169e2\">", "")

  AllCountry = re.sub("</span>", "", AllCountry)

  AllCountry = AllCountry.replace("</p>]", "")

   

  AllCountry = AllCountry.replace("<span style=\"color: rgb(65, 105, 226);\">", "")

  AllCountry = re.sub("<span>", "", AllCountry)

  AllCountry = re.sub("<p>", "", AllCountry)

  AllCountry = re.sub("</p>", "", AllCountry)

  return AllCountry 

 

def query(province):

  table = PrettyTable(['地区', '确诊', '死亡', '治愈'])

 

  for (k, v) in province.items():

    name = k

    table.add_row([name, v[0] if v[0] != 0 else '-', v[1] if v[1] != 0 else '-', v[2] if v[2] != 0 else '-'])

  if len(province.keys()) != 0:

    print(table)

  else:

    print("暂无")

 

def getInfo(text):

  text = str(text)

  text = re.sub("<p class=\"descText___Ui3tV\">", "", text)

  text = re.sub("</p>", "", text)

  return text

 

def is_json(json_str):

  try:

    json.loads(json_str)

  except ValueError:

    return False

  return True

 

def ff(str, num):

  return str[:num] + str[num+1:]

     

 

def main():

  url = "https://3g.dxy.cn/newh5/view/pneumonia"

 

  try:

    headers = {}

    headers['user-agent'] = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.77 Safari/537.36' #http头大小写不敏感

    headers['accept'] = 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8'

    headers['Connection'] = 'keep-alive'

    headers['Upgrade-Insecure-Requests'] = '1'

 

    r = requests.get(url, headers=headers)

    r.raise_for_status()

    r.encoding = r.apparent_encoding

    soup = BeautifulSoup(r.text,'lxml')

    table = PrettyTable(['地区', '确诊', '死亡', '治愈'])

    table.hrules = ALL

 

    #### 截至时间

    # TitleTime = getTime(soup.select('.title___2d1_B'))

     

    print()

    # print("       ",TitleTime + "\n")

 

    while True:

      r = requests.get("https://service-f9fjwngp-1252021671.bj.apigw.tencentcs.com/release/pneumonia")

      json_str = json.loads(r.text)

      if json_str['error'] == 0:

        break

 

    print("==================================全国数据==================================")

    print()

     

    print("   确诊 " + str(json_str['data']['statistics']['confirmedCount']) + " 例"

      + "    " + "疑似 " + str(json_str['data']['statistics']['suspectedCount']) + " 例"

      + "    " + "死亡" + str(json_str['data']['statistics']['deadCount']) + " 例"

      + "    " + "治愈" + str(json_str['data']['statistics']['curedCount']) + " 例\n")

 

    print("==================================相关情况==================================")

    print()

 

    print("传染源:" + json_str['data']['statistics']['infectSource'])

    print("病毒:" + json_str['data']['statistics']['virus'])

    print("传播途径:" + json_str['data']['statistics']['passWay'])

    print(json_str['data']['statistics']['remark1'])

    print(json_str['data']['statistics']['remark2'] + "\n")

       

    print("==================================国内情况==================================")

    print()

     

    json_provinces = re.findall("{\"provinceName\":(.*?)]}", str(soup))

 

    idx = 0

    for province in json_provinces:

      if is_json(province):

        pass

 

      else:

        province = "{\"provinceName\":" + province + "]}"

        province = json.loads(province)

         

      province_name = province['provinceShortName'] if province['provinceShortName'] != 0 else '-'

      confirmed = province['confirmedCount'] if province['confirmedCount'] != 0 else '-'

      suspected = province['suspectedCount'] if province['suspectedCount'] != 0 else '-'

      cured = province['curedCount'] if province['curedCount'] != 0 else '-'

      dead = province['deadCount'] if province['deadCount'] != 0 else '-'

      table.add_row([province_name, confirmed, dead, cured])

      map[province_name] = idx

      idx = idx + 1

      for city in province['cities']:

        provinces_idx[map[province_name]][city['cityName']] = [city['confirmedCount'], city['deadCount'], city['curedCount']]

 

    print(table)

     

     

    print()

    print("==================================国外情况==================================")

    print()

 

    json_provinces = str(re.findall("\"id\":949(.*?)]}", str(soup)))

    json_provinces = json_provinces[:1] + "{\"id\":949" + json_provinces[2:]

    json_provinces = json_provinces[:len(json_provinces) - 2] + json_provinces[len(json_provinces) - 1:]

    provinces = json.loads(json_provinces)

 

    table = PrettyTable(['地区', '确诊', '死亡', '治愈'])

    for province in provinces:

      confirmed = province['confirmedCount'] if province['confirmedCount'] != 0 else '-'

      dead = province['deadCount'] if province['deadCount'] != 0 else '-'

      cured = province['curedCount'] if province['curedCount'] != 0 else '-'

      table.add_row([province['provinceName'], confirmed, dead, cured])

     

    print(table)

    print()

     

    print("==================================最新消息==================================")

    print()

     

       

    idx = 0

    for news in json_str['data']['timeline']:

      if idx == 5:

        break

      print(news['pubDateStr'] + " " + news['title'])

      idx = idx + 1

     

 

    print()

    key = input("请输入您想查询详细信息的省份,例如 湖北\n")

    print()

    if key in map.keys():

      query(provinces_idx[map[key]])

    else:

      print("暂无相关信息")

       

    print("\n欢迎提出各种意见")

  except:

    print("连接失败")

 

if __name__ == '__main__':

  main()

  sleep(30)

以上就是Python实时数据采集-新型冠状病毒的详细内容,大家出门要做好安全措施,感谢对三水点靠木的支持。

Python 相关文章推荐
python冒泡排序简单实现方法
Jul 09 Python
使用Mixin设计模式进行Python编程的方法讲解
Jun 21 Python
通过5个知识点轻松搞定Python的作用域
Sep 09 Python
Python设计模式之模板方法模式实例详解
Jan 17 Python
python实现windows壁纸定期更换功能
Jan 21 Python
python+pyqt5实现24点小游戏
Jan 24 Python
树莓派实现移动拍照
Jun 22 Python
浅谈Pytorch中的torch.gather函数的含义
Aug 18 Python
OpenCV哈里斯(Harris)角点检测的实现
Jan 15 Python
pytorch对梯度进行可视化进行梯度检查教程
Feb 04 Python
Python 之 Json序列化嵌套类方式
Feb 27 Python
Python新手如何理解循环加载模块
May 29 Python
在tensorflow中实现屏蔽输出的log信息
Feb 04 #Python
Python变量作用域LEGB用法解析
Feb 04 #Python
如何在python开发工具PyCharm中搭建QtPy环境(教程详解)
Feb 04 #Python
TensorFlow基本的常量、变量和运算操作详解
Feb 03 #Python
Tensorflow轻松实现XOR运算的方式
Feb 03 #Python
Tensorflow不支持AVX2指令集的解决方法
Feb 03 #Python
基于Python3.6中的OpenCV实现图片色彩空间的转换
Feb 03 #Python
You might like
smarty的保留变量问题
2008/10/23 PHP
php在window iis的莫名问题的测试方法
2013/05/14 PHP
windows下配置php5.5开发环境及开发扩展
2014/12/25 PHP
php单文件版在线代码编辑器
2015/03/12 PHP
php判断当前操作系统类型
2015/10/28 PHP
Linux安装配置php环境的方法
2016/01/14 PHP
深入剖析浏览器退出之后php还会继续执行么
2016/05/17 PHP
在Z-Blog中运行代码[html][/html](纯JS版)
2007/03/25 Javascript
JavaScript Date对象 日期获取函数
2010/12/19 Javascript
js判断鼠标同时离开两个div的思路及代码
2013/05/31 Javascript
JS获取iframe中longdesc属性的方法
2015/04/01 Javascript
用JavaScript实现页面重定向功能的教程
2015/06/04 Javascript
Bootstrap每天必学之栅格系统(布局)
2015/11/25 Javascript
JavaScript学习小结(7)之JS RegExp
2015/11/29 Javascript
angularjs 实现带查找筛选功能的select下拉框实例
2017/01/11 Javascript
LayUi中接口传数据成功,表格不显示数据的解决方法
2018/08/19 Javascript
Vue中保存数据到磁盘文件的方法
2018/09/06 Javascript
js动态添加带圆圈序号列表的实例代码
2021/02/18 Javascript
Python标准库之多进程(multiprocessing包)介绍
2014/11/25 Python
python创建进程fork用法
2015/06/04 Python
Pandas 合并多个Dataframe(merge,concat)的方法
2018/06/08 Python
基于Python的ModbusTCP客户端实现详解
2019/07/13 Python
解决Python logging模块无法正常输出日志的问题
2020/02/21 Python
浅谈selenium如何应对网页内容需要鼠标滚动加载的问题
2020/03/14 Python
Nike加拿大官网:Nike.com (CA)
2019/04/09 全球购物
Booking.com缤客中国:全球酒店在线预订网站
2020/05/03 全球购物
工程现场管理求职自荐信
2013/10/02 职场文书
文明家庭先进事迹材
2014/01/27 职场文书
教师群众路线心得体会
2014/11/04 职场文书
工伤私了协议书范本
2014/11/24 职场文书
超强台风观后感
2015/06/09 职场文书
工资证明范本
2015/06/12 职场文书
优秀志愿者感言
2015/08/01 职场文书
我的中国梦主题班会
2015/08/14 职场文书
解析Redis Cluster原理
2021/06/21 Redis
Python+tkinter实现高清图片保存
2022/03/13 Python