Python实时数据采集-新型冠状病毒-Python教程-PHP中文网

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 = {
    &#39;湖北&#39;:0, &#39;广东&#39;:1, &#39;浙江&#39;:2, &#39;北京&#39;:3, &#39;上海&#39;:4, &#39;湖南&#39;:5, &#39;安徽&#39;:6, &#39;重庆&#39;:7,
    &#39;四川&#39;:8, &#39;山东&#39;:9, &#39;广西&#39;:10, &#39;福建&#39;:11, &#39;江苏&#39;:12, &#39;河南&#39;:13, &#39;海南&#39;:14,
    &#39;天津&#39;:15, &#39;江西&#39;:16, &#39;陕西&#39;:17, &#39;贵州&#39;:18, &#39;辽宁&#39;:19, &#39;香港&#39;:20, &#39;黑龙江&#39;:21,
    &#39;澳门&#39;:22, &#39;新疆&#39;:23, &#39;甘肃&#39;:24, &#39;云南&#39;:25, &#39;台湾&#39;:26, &#39;山西&#39;:27, &#39;吉林&#39;:28,
    &#39;河北&#39;:29, &#39;宁夏&#39;:30, &#39;内蒙古&#39;:31, &#39;青海&#39;:32, &#39;西藏&#39;: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！