Python实时数据采集-新型冠状病毒

angryTom
发布: 2020-02-03 17:51:00
转载
4014 人浏览过

Python实时数据采集-新型冠状病毒

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

项目相关截图:

全国数据展示

国内数据展示

国外数据展示

查看指定区域详细数据

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(&#39;<span>(.*?)</span>&#39;, 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([&#39;地区&#39;, &#39;确诊&#39;, &#39;死亡&#39;, &#39;治愈&#39;])

    for (k, v) in province.items():
        name = k
        table.add_row([name, v[0] if v[0] != 0 else &#39;-&#39;, v[1] if v[1] != 0 else &#39;-&#39;, v[2] if v[2] != 0 else &#39;-&#39;])
    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[&#39;user-agent&#39;] = &#39;Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.77 Safari/537.36&#39; #http头大小写不敏感
        headers[&#39;accept&#39;] = &#39;text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8&#39;
        headers[&#39;Connection&#39;] = &#39;keep-alive&#39;
        headers[&#39;Upgrade-Insecure-Requests&#39;] = &#39;1&#39;

        r = requests.get(url, headers=headers)
        r.raise_for_status()
        r.encoding = r.apparent_encoding
        soup = BeautifulSoup(r.text,&#39;lxml&#39;)
        table = PrettyTable([&#39;地区&#39;, &#39;确诊&#39;, &#39;死亡&#39;, &#39;治愈&#39;])
        table.hrules = ALL

        #### 截至时间
        # TitleTime = getTime(soup.select(&#39;.title___2d1_B&#39;))
        
        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[&#39;error&#39;] == 0:
                break

        print("==================================全国数据==================================")
        print()
        
        print("     确诊 " + str(json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;confirmedCount&#39;]) + " 例"
            + "       " + "疑似 " + str(json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;suspectedCount&#39;]) + " 例"
            + "       " + "死亡" + str(json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;deadCount&#39;]) + " 例"
            + "       " + "治愈" + str(json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;curedCount&#39;]) + " 例\n")

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

        print("传染源:" + json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;infectSource&#39;])
        print("病毒:" + json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;virus&#39;])
        print("传播途径:" + json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;passWay&#39;])
        print(json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;remark1&#39;])
        print(json_str[&#39;data&#39;][&#39;statistics&#39;][&#39;remark2&#39;] + "\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[&#39;provinceShortName&#39;] if province[&#39;provinceShortName&#39;] != 0 else &#39;-&#39;
            confirmed = province[&#39;confirmedCount&#39;] if province[&#39;confirmedCount&#39;] != 0 else &#39;-&#39;
            suspected = province[&#39;suspectedCount&#39;] if province[&#39;suspectedCount&#39;] != 0 else &#39;-&#39;
            cured = province[&#39;curedCount&#39;] if province[&#39;curedCount&#39;] != 0 else &#39;-&#39;
            dead = province[&#39;deadCount&#39;] if province[&#39;deadCount&#39;] != 0 else &#39;-&#39;
            table.add_row([province_name, confirmed, dead, cured])
            map[province_name] = idx
            idx = idx + 1
            for city in province[&#39;cities&#39;]:
                provinces_idx[map[province_name]][city[&#39;cityName&#39;]] = [city[&#39;confirmedCount&#39;], city[&#39;deadCount&#39;], city[&#39;curedCount&#39;]]

        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([&#39;地区&#39;, &#39;确诊&#39;, &#39;死亡&#39;, &#39;治愈&#39;])
        for province in provinces:
            confirmed = province[&#39;confirmedCount&#39;] if province[&#39;confirmedCount&#39;] != 0 else &#39;-&#39;
            dead = province[&#39;deadCount&#39;] if province[&#39;deadCount&#39;] != 0 else &#39;-&#39;
            cured = province[&#39;curedCount&#39;] if province[&#39;curedCount&#39;] != 0 else &#39;-&#39;
            table.add_row([province[&#39;provinceName&#39;], confirmed, dead, cured])
        
        print(table)
        print()
        
        print("==================================最新消息==================================")
        print()
        
            
        idx = 0
        for news in json_str[&#39;data&#39;][&#39;timeline&#39;]:
            if idx == 5:
                break
            print(news[&#39;pubDateStr&#39;] + "  " + news[&#39;title&#39;])
            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__ == &#39;__main__&#39;:
    main()
    sleep(30)
登录后复制

 最后,祝大家百毒不侵,中国加油!!一定能够度过难关!!

以上是Python实时数据采集-新型冠状病毒的详细内容。更多信息请关注PHP中文网其他相关文章!

相关标签:
来源:csdn.net
本站声明
本文内容由网友自发贡献,版权归原作者所有,本站不承担相应法律责任。如您发现有涉嫌抄袭侵权的内容,请联系admin@php.cn
作者最新文章
热门教程
更多>
最新下载
更多>
网站特效
网站源码
网站素材
前端模板
关于我们 免责声明 Sitemap
PHP中文网:公益在线PHP培训,帮助PHP学习者快速成长!