京东衣服评论爬取

赵乾舟

1. import requests
   
2. import pandas as pd
   
3. import json
   
4. import time
   
5. id = input('请输入ID：')
   
6. url = f'https://club.jd.com/comment/productPageComments.action?callback=fetchJSON_comment98&productId={id}&score=0&sortType=5&page=0&pageSize=10&isShadowSku=0&fold=1'
   
7. UA = {'user-agent':'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36'}
   
8. respon = requests.get( url = url,headers=UA).text
   
9. respon = respon.replace('fetchJSON_comment98(','')
   
10. respon = respon.replace(');','')
    
11. zidian = json.loads(respon)
    
12. yeshu = zidian['maxPage']
    
13. print(yeshu)
    
14. for ye in range(1,yeshu+1):
    
15.     url = f'https://club.jd.com/comment/productPageComments.action?callback=fetchJSON_comment98&productId={id}&score=0&sortType=5&page={ye-1}&pageSize=10&isShadowSku=0&fold=1'
    
16.     respon = requests.get(url=url, headers=UA).text
    
17.     respon = respon.replace('fetchJSON_comment98(', '')
    
18.     respon = respon.replace(');', '')
    
19.     zidian = json.loads(respon)
    
20.     pinglun = zidian['comments']
    
21.     neirong = [pinglun['content'] for pinglun in pinglun]
    
22.     yanse = [pinglun['productColor'] for pinglun in pinglun]
    
23.     size = [pinglun['productSize'] for pinglun in pinglun]
    
24.     shuju = pd.DataFrame({'评价':neirong,'颜色':yanse,'大小':size})
    
25.     shuju.to_csv('d:/j3d.csv',mode='a',header=0,encoding='ANSI')
    
26.     time.sleep(3)

复制代码

赵乾舟

第14行，控制翻多少页