提交爬取百度国内省份以及地级市疫情数据python模块

CaseInland.py

@@ -0,0 +1,120 @@
+import requests
+import re
+from bs4 import BeautifulSoup
+import json
+import time
+
+class City:
+    def __init__(self,tempjs={},city='无数据',confiemed='无数据',died='无数据',crued='无数据',confirmedRelative='无数据',cityCode='无数据'):
+        self.city=city
+        self.confirmed=confiemed
+        self.died=died
+        self.crued=crued
+        self.confirmedRelative=confirmedRelative
+        self.cityCode=cityCode
+        js={'city': '无数据', 'confirmed': '无数据', 'died': '无数据', 'crued': '无数据', 'confirmedRelative': '无数据', 'curConfirm': '无数据', 'cityCode': '无数据'}
+        for k,v in tempjs.items():
+            if tempjs[k]!='':
+                js[k]=v
+        self.city=js["city"]
+        self.confirmed=js["confirmed"]
+        self.died=js["died"]
+        self.crued=js["crued"]
+        self.confirmedRelative=js["confirmedRelative"]
+        self.cityCode=js["cityCode"]
+    def __str__(self):
+        if self.city=='境外输入':
+            return "其它情况:{:<10}城市代码:{:<4}新增患者:{:<4}累积确诊:{:<4}累积死亡:{:<4}累积治愈:{:<4}".format(self.city,self.cityCode,self.confirmedRelative,
+                                                self.confirmed,self.died,self.crued)
+        return "城市名:{:<10}城市代码:{:<4}新增患者:{:<4}累积确诊:{:<4}累积死亡:{:<4}累积治愈:{:<4}".format(self.city,self.cityCode,self.confirmedRelative,
+                                                self.confirmed,self.died,self.crued)
+        
+class Provinces:
+    def __init__(self):
+        self.number=0
+        self.province=[]
+    def addProvince(self,province):
+        self.province.append(province)
+    def printEveryProvince(self):
+        for p in self.province:
+            print(p)
+    def __str__(self):
+        return str(self.province)
+    
+    
+class Province:
+    def __init__(self,tempjs={},confirmed='无数据',died='无数据',crued='无数据',relativeTime='无数据',confirmedRelative='无数据',
+                 diedRelative='无数据',curedRelative='无数据',asymptomaticRelative='无数据',asymptomatic='无数据',
+                 curConfirm='无数据',curConfirmRelative='无数据',icuDisable='无数据',area='无数据',subList=[]):
+        self.died=died
+        self.confirmed=confirmed
+        self.crued=crued
+        self.relativeTime=relativeTime
+        self.confirmedRelative=confirmedRelative
+        self.diedRelative=diedRelative
+        self.curedRelative=curedRelative
+        self.asymptomaticRelative=asymptomaticRelative
+        self.asymptomatic=asymptomatic
+        self.curConfirm=curConfirm
+        self.curConfirmRelative=curConfirmRelative
+        self.icuDisable=icuDisable
+        self.area=area
+        self.subList=subList
+        js={'confirmed': '无数据', 'died': '无数据', 'crued': '无数据', 'relativeTime': '无数据', 'confirmedRelative': '无数据', 
+            'diedRelative': '无数据', 'curedRelative': '无数据', 'asymptomaticRelative': '无数据', 'asymptomatic': '无数据', 'curConfirm': '无数据', 
+            'curConfirmRelative': '无数据', 'icuDisable': '无数据', 'area': '无数据', 'subList':[]}
+        for k,v in tempjs.items():
+            if tempjs[k]!='':
+                js[k]=v
+        #print(js)
+        self.died=js['died']
+        self.confirmed=js['confirmed']
+        self.crued=js['crued']
+        self.relativeTime=js['relativeTime']
+        self.confirmedRelative=js['confirmedRelative']
+        self.diedRelative=js['diedRelative']
+        self.curedRelative=js['curedRelative']
+        self.asymptomaticRelative=js['asymptomaticRelative']
+        self.asymptomatic=js['asymptomatic']
+        self.curConfirm=js['curConfirm']
+        self.curConfirmRelative=js['curConfirmRelative']
+        self.icuDisable=js['icuDisable']
+        self.area=js['area']
+        self.subList=js['subList']
+        cities=js["subList"]
+        for c in cities:
+            city=City(c)
+            subList.append(city)
+    def printEveryCity(self):
+        for c in self.subList:
+            city=City(c)
+            print(city)
+    def __str__(self):
+        localobj=time.localtime(int(self.relativeTime))
+        newtime=time.strftime("%Y-%m-%d %H:%M:%S",localobj)
+        return "省份：{:<4}现有患者：{:<4}现有新增患者：{:<4}累积无症状感染者：{:<4}新增无症状感染者：{:<4}累积确诊：{:<6}新增确诊：{:<4}累积治愈：{:<6}新增治愈：{:<4}累积死亡：{:<6}新增死亡：{:<4}累积重症：{:<4}更新时间：{:<10}".format(self.area,self.curConfirm,self.curConfirmRelative,
+    self.asymptomatic,self.asymptomaticRelative,self.confirmed,self.confirmedRelative,self.crued,self.curedRelative,self.died,self.diedRelative,
+    self.icuDisable,newtime)
+    
+    
+def start():
+    res=requests.get('https://voice.baidu.com/act/newpneumonia/newpneumonia/')
+    res=res.text
+    #script type="application/json"
+    soup=BeautifulSoup(res,'html.parser')
+    tag=soup.find('script',attrs={'id':'captain-config'})
+    tagstr=str(tag)
+    findinland=re.findall('caseList".*"caseOutsideList',tagstr)[0]
+    province=re.findall('\{"confirmed.*?subList":\[.*?\]\}',findinland)
+    provinces=Provinces()
+    for item in province:
+        js=json.loads(item)
+        p=Province(js)
+        provinces.addProvince(p)
+        print(p)
+        p.printEveryCity()
+        print("++++++++++++++++++++++++++++++"*4)
+
+    #provinces.printEveryProvince()
+    
+