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 requestsimport refrom bs4 import BeautifulSoupfrom time import sleepimport Jsonfrom prettytable import ALLfrom prettytable import prettytablehubei = {}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 = {} # nonexizang = {} # noneprovinces_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 textdef is_Json(Json_str):    try:        Json.loads(Json_str)    except ValueError:        return False    return Truedef 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)

最后，祝大家百毒不侵，中国加油！！一定能够度过难关！！ 总结

以上是内存溢出为你收集整理的Python实时数据采集-新型冠状病毒全部内容，希望文章能够帮你解决Python实时数据采集-新型冠状病毒所遇到的程序开发问题。

如果觉得内存溢出网站内容还不错，欢迎将内存溢出网站推荐给程序员好友。