微博疫情数据爬取

冷回清风暖

1. import requests
   
2. import xlwt
   
3. 
   
4. def get_json():
   
5.     url = "https://interface.sina.cn/news/wap/fymap2020_data.d.json"
   
6.     headers = {
   
7.         'cookie': 'Apache=2885827998838.657.1585714305110; SINAGLOBAL=2885827998838.657.1585714305110; ULV=1585714305112:1:1:1:2885827998838.657.1585714305110:; NCP-SINA-CN=; genTime=1585715444; statuid=__10.41.1.25_1585715444_0.40395200; statuidsrc=Mozilla%2F5.0+%28Linux%3B+Android+10%3B+VOG-AL00+Build%2FHUAWEIVOG-AL00%3B+wv%29+AppleWebKit%2F537.36+%28KHTML%2C+like+Gecko%29+Version%2F4.0+Chrome%2F74.0.3729.186+Mobile+Safari%2F537.36%3B+HUAWEI-VOG-AL00__sinanews__7.33.5__android__10__525%3B+hybrid__0.8__online%6010.41.1.25%60http%3A%2F%2Finterface.sina.cn%2Fnews%2Fwap%2Ffyzt_newsv2.d.json%3Fcallback%3D_aLocalNewsFunction%26_%3D1585715443935%26type%3Dprovince%26code%3Dhenan%26page%3D1%60%60__10.41.1.25_1585715444_0.40395200; ustat=__10.41.1.25_1585715444_0.40395200; historyRecord={"href":"https://news.sina.cn/zt_d/yiqing0121","refer":"https://cn.bing.com/"}; vt=4',
   
8.         'referer': 'https://news.sina.cn/zt_d/yiqing0121',
   
9.         'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.149 Safari/537.36 Edg/80.0.361.69',
   
10.     }
    
11.     try:
    
12.         response = requests.get(url, headers=headers)
    
13.         if response.content:
    
14.             return response.json()
    
15.         else:
    
16.             return None
    
17.     except requests.ConnectionError:
    
18.         return "连接网页失败"
    
19. 
    
20. def get_data(json):
    
21.     data= json['data']['list']
    
22.     #['data']['list'][0]['city']
    
23.     for item in data:
    
24.         for city in item['city']:
    
25.             cit = city['name']
    
26.             conNum = city['conNum']#累计确诊病例
    
27.             cureNum = city['cureNum']#治愈
    
28.             deathNum = city['deathNum']#死亡
    
29.             if cit != '境外输入人员':
    
30.                 yield {
    
31.                     'city':cit,
    
32.                     'conNum':conNum,
    
33.                     'cureNum':cureNum,
    
34.                     'deathNum':deathNum,
    
35.                 }
    
36. 
    
37.     return data
    
38. 
    
39. def save_data(data):
    
40.     myxls = xlwt.Workbook()
    
41.     sheet1 = myxls.add_sheet(u'20200401', cell_overwrite_ok=True)
    
42.     sheet1.write(0, 0, "FID")
    
43.     sheet1.write(0, 1, "city")
    
44.     sheet1.write(0, 2, "conNum")
    
45.     sheet1.write(0, 3, "cureNum")
    
46.     sheet1.write(0, 4, "deathNum")
    
47.     i = 1
    
48.     for item in data:
    
49.         sheet1.write(i, 0, i)
    
50.         sheet1.write(i, 1, item['city'])
    
51.         sheet1.write(i, 2, item['conNum'])
    
52.         sheet1.write(i, 3, item['cureNum'])
    
53.         sheet1.write(i, 4, item['deathNum'])
    
54.         i = i + 1
    
55.     myxls.save('疫情0401.xls')
    
56. 
    
57. 
    
58. if __name__ == '__main__':
    
59.     json = get_json()
    
60.     data = get_data(json)
    
61.     save_data(data)
    
62.     for item in data:
    
63.         print(item)
    

复制代码

爬虫小白 由于作业要用到这些数据 所以做了下
然后给大家分享一下

风陵渡口杨过

厉害

妈妈l我错乐

牛啊