完成 data_cleansing 模块的集成

src/backend/JDSpiders/BoardUSuit/BoardUSuit.py

@@ -6,9 +6,8 @@ Description:
   shop_name ：店铺名
   评论数和好评率不爬，因为都是0
 Author: Fishermanykx
-Date: 2020-12-11 14:25:04 
 LastEditors: Fishermanykx
-LastEditTime: 2021-05-28 11:12:07
+LastEditTime: 2021-07-12 16:26:41
 '''
 import re
 import json

src/backend/JDSpiders/CaseFan/CaseFan.py

@@ -3,9 +3,8 @@ Description:
   introduction ：商品介绍的 .json
   Ptable_params ：规格与包装的 .json
 Author: Fishermanykx
-Date: 2020-12-08 20:45:47
 LastEditors: Fishermanykx
-LastEditTime: 2021-05-28 11:12:00
+LastEditTime: 2021-07-12 16:14:17
 '''
 
 import json

src/backend/JDSpiders/Spider/DBConfig.py

@@ -1,7 +1,6 @@
 '''
 Description: Data Base Configuration
 Author: Fishermanykx
-Date: 2020-12-30 15:33:37
 LastEditors: Fishermanykx
 LastEditTime: 2021-05-28 11:11:42
 '''

src/backend/JDSpiders/Spider/JDSpider.py

@@ -1,16 +1,15 @@
 '''
 Description:
 Author: Fishermanykx
-Date: 2020-12-29 08:21:41
 LastEditors: Fishermanykx
-LastEditTime: 2021-05-28 11:11:28
+LastEditTime: 2021-07-13 09:17:32
 '''
 
 import re
 import json
 import time
 import pymysql
-from pymysql.converters import escape_string
+# from pymysql.converters import escape_string
 
 from selenium.common.exceptions import ElementNotInteractableException
 from selenium.common.exceptions import ElementClickInterceptedException
@@ -659,7 +658,7 @@ class CPUSpider(JDSpider):
 
     page_num = 1
     start_page = 1
-    # self.productSpider(cpu_link, page_num, start_page)
+    self.productSpider(cpu_link, page_num, start_page)
 
     self.cleanCPU()
     print("Successfully get CPU data!")
@@ -824,7 +823,7 @@ class MotherboardSpider(JDSpider):
       # 判定是否为板-U套装
       record = data[i]
       if record['comment_num'] == 100:  # 抓到板-U套装了
-        continue      
+        continue
       try:
         if '套装' in record['title_name']:
           continue
@@ -959,7 +958,7 @@ class GraphicsCardSpider(JDSpider):
       else:
         generation = 1
     else:
-      if ('6900' in name) or ('6800' in name):
+      if ('6900' in name) or ('6800' in name) or ('6700' in name):
         generation = 3
       elif ('5700' in name) or ('5600' in name) or ('5500' in name):
         generation = 2
@@ -1187,7 +1186,7 @@ class MemorySpider(JDSpider):
         "Ptable_params, title_name) VALUES (%(id)s, %(name)s, %(comment_num)s, %(praise_rate)s, %(shop_name)s, %(price)s"\
         ", %(link)s, %(brand)s, %(frequency)s, %(total_capacity)s, %(memory_num)s, %(appearance)s, "\
         "%(ddr_gen)s, %(introduction)s, %(Ptable_params)s, %(title_name)s)"
-    # sql_insert = escape_string(sql_insert)
+    # sql_insert = pymysql.escape_string(sql_insert)
     # cursor.executemany(sql_insert, new_data)
     for i in range(len(new_data)):
       cursor.execute(sql_insert, new_data[i])
@@ -1894,7 +1893,7 @@ class PowerSupplySpider(JDSpider):
   def main(self):
     power_supply_link = "https://list.jd.com/list.html?cat=670%2C677%2C691&psort=3&psort=3&page="
     # page_num = 25  # 抓25页
-    page_num = 1  
+    page_num = 1
     start_page = 1
     # self.productSpider(power_supply_link, page_num, start_page)
 
@@ -2090,7 +2089,7 @@ class CaseSpider(JDSpider):
 
 if __name__ == "__main__":
   accessory_type = 'all'
-  accessory_type = 'motherboard'
+  accessory_type = 'graphics_card'
   if accessory_type == 'cpu':
     cpu_spider = CPUSpider('cpu')
     cpu_spider.main()