update

checkData.py

@@ -61,6 +61,8 @@ def checkData(check_data):
         main_url = df['url'].strip()
         source = df['source']
         price = df['price']
+        brand = df['brand']
+        subcategory = df['subcategory']
         print(main_url) 
         if "jd" in str(main_url): 
             try:
@@ -92,9 +94,9 @@ def checkData(check_data):
                     if "自营" in str(ziying):
                             name = html.xpath(
                                     "//div[@class='sku-name']/text()")
-                            if ("定制"in str(name)) or ("防弹"in str(name)) or ("射击"in str(name)) or ("订制"in str(name)) or ("卫星"in str(name)) or ("靶"in str(name)) or ("企业定制"in str(name)) or ("军迷"in str(name)) or ("携行具"in str(name)) or ("出版社"in str(name)) or ("书籍"in str(name)):
-                                print("定制/专用类产品暂不通过")
-                                result.append("定制/专用类产品暂不通过")
+                            if ("定制"in str(name)) or ("防弹"in str(name)) or ("射击"in str(name)) or ("订制"in str(name)) or ("卫星"in str(name)) or ("靶"in str(name)) or ("企业定制"in str(name)) or ("军迷"in str(name)) or ("携行具"in str(name)) or ("出版社"in str(name)) or ("书籍"in str(name)) or ("出版社" in str(brand)) or ("书籍" in str(subcategory)):
+                                print("定制/专用/书籍类产品暂不通过")
+                                result.append("定制/专用/书籍类产品暂不通过")
                                 id_all.append(date_id)
                                 sku_list.append(sku)
                                 name_list.append(name)
@@ -182,9 +184,9 @@ def checkData(check_data):
                 if len(ziying) == 1:
                         name = html.xpath(
                                 "//*[@id='gm-prd-main']/div[1]/h1/text()")
-                        if ("定制"in str(name)) or ("防弹"in str(name)) or ("射击"in str(name)) or ("订制"in str(name)) or ("卫星"in str(name)) or ("靶"in str(name)) or ("企业定制"in str(name)) or ("出版社"in str(name)) or ("书籍"in str(name)):
-                            print("定制/专用类产品暂不通过")
-                            result.append("定制/专用类产品暂不通过")
+                        if ("定制"in str(name)) or ("防弹"in str(name)) or ("射击"in str(name)) or ("订制"in str(name)) or ("卫星"in str(name)) or ("靶"in str(name)) or ("企业定制"in str(name)) or ("出版社"in str(name)) or ("书籍"in str(name)) or ("出版社" in str(brand)) or ("书籍" in str(subcategory)):
+                            print("定制/专用/书籍类产品暂不通过")
+                            result.append("定制/专用/书籍类产品暂不通过")
                             id_all.append(date_id)
                             sku_list.append(sku)
                             name_list.append(name)
@@ -263,9 +265,10 @@ def checkData(check_data):
                                 or ("订制"in str(product_name)) or ("卫星"in str(product_name)) \
                                 or ("靶"in str(product_name)) or ("企业定制"in str(product_name)) \
                                 or ("军迷"in str(product_name)) or ("携行具"in str(product_name)) \
-                                or ("出版社"in str(name)) or ("书籍"in str(name)):
-                                print("定制/专用类产品暂不通过")
-                                result.append("定制/专用类产品暂不通过")
+                                or ("出版社"in str(name)) or ("书籍"in str(name)) \
+                                or ("出版社" in str(brand)) or ("书籍" in str(subcategory)):
+                                print("定制/专用/书籍类产品暂不通过")
+                                result.append("定制/专用/书籍类产品暂不通过")
                                 price_list.append(sn_price)
                                 id_all.append(date_id)
                                 sku_list.append(sku)

matchSKU.py

@@ -19,15 +19,42 @@ def matchSKU(sku_list,frm):
     cursor = conn.cursor()
     print('正在删除库中产品状态为6的sku，稍等。。。')#首先将库中错误状态产品的JD/SN/GM的SKU删除
     cursor.execute("DELETE from productcode_sku where productcode in (select productcode from info_product where state='6')")
+    print('正在删除库中重复的SKU，稍等。。。')#需先删除库中重复的SKU，避免匹配相同的编码
+    cursor.execute("declare OperCursor Cursor for\
+                      SELECT productcode,sku,frm,count(*) ca  FROM productcode_sku\
+                    GROUP BY productcode,sku,frm\
+                    HAVING COUNT(*)>1\
+                      open OperCursor\
+                      declare @PRODUCTCODE as nvarchar(20)\
+                      declare @sku as nvarchar(50)\
+                      declare @frm as nvarchar(20)\
+                      declare @ca	  as 	int\
+                      declare @return_value	 as int\
+                    fetch next from OperCursor into @PRODUCTCODE,@sku,@frm,@ca\
+                    IF @@fetch_status=0\
+                    delete from productcode_sku\
+                            where id  in (select top (@ca-1) id from productcode_sku\
+                    where  productcode=@PRODUCTCODE and sku=@sku and frm=@frm)\
+                    \
+                    while @@fetch_status=0\
+                    begin\
+                    fetch next from OperCursor into @PRODUCTCODE,@sku,@frm,@ca\
+                    IF @@fetch_status=0\
+                    delete from productcode_sku\
+                    where id  in (select top (@ca-1) id from productcode_sku\
+                    where  productcode=@PRODUCTCODE and sku=@sku and frm=@frm)\
+                    end\
+                    close OperCursor\
+                    deallocate OperCursor")
     conn.commit()
     print('正在获取当前库中所有sku,请稍等.....')#获取库中所有的SKU,以供检查
     try:
-        get_all_sku = f"select  * from productcode_sku where frm in {frm}"
+        get_all_sku = f"select  * from productcode_sku where frm in {frm} and productcode not like '78%' "
         cursor.execute(get_all_sku)
         data_sku = (cursor.fetchall())
         data_sku_file = pd.DataFrame(data_sku,columns=['id','productcode','sku','skuname','createddate','frm'])
     except:
-        get_all_sku = f"select  * from productcode_sku where frm='{frm}'"
+        get_all_sku = f"select  * from productcode_sku where frm='{frm}' and productcode not like '78%' "
         cursor.execute(get_all_sku)
         data_sku = (cursor.fetchall())
         data_sku_file = pd.DataFrame(data_sku,columns=['id','productcode','sku','skuname','createddate','frm'])