update

match_program.py

@@ -20,7 +20,7 @@ def get_test_data():
     conn = pymssql.connect('123.56.115.207','zgcindex','jiayou2017+2018','ZI_DataBase')
     cursor = conn.cursor()
     #cursor.execute("select * from vw_electricity_source_price where ProductCode in "+ code_list +" and periods = (select top 1 periods from vw_entry_source_price order by periods desc)")
-    cursor.execute("select * from Product_Api_Data where batch = '20191224' and source = 'JD' and state = 4")
+    cursor.execute("select * from Product_Api_Data where batch = '2020-02-153' and state = 4")
     test_df = pd.DataFrame([v for v in cursor.fetchall()],columns=[tuple[0] for tuple in cursor.description])
         
     cursor.close()
@@ -70,7 +70,7 @@ def get_db_data(zgc_category_code):
         
         if db_file_name in file_list:
             print('获取本地已存数据库数据')
-            db_attr_data = pd.read_csv(file_path,converters={'productcode':str,'SubCategoryCode':str,'BrandCode':str})
+            db_attr_data = pd.read_csv(file_path,converters={'ProductCode':str,'SubCategoryCode':str,'BrandCode':str})
             db_attr_data = db_attr_data.rename(columns={"productcode": "ProductCode","productname":"ProductName","subtitle":"SubTitle","state":"State","value":"Value"})
             return db_attr_data
         
@@ -121,11 +121,14 @@ def match_with_db(params_df):
     match_res['productName'] = [list(params_df[params_df['productId'] == _id]['productName'])[0] for _id in list(match_res['productId'])]
     match_res['source'] = [list(params_df[params_df['productId'] == _id]['source'])[0] for _id in list(match_res['productId'])]
     match_res['batch'] = [list(params_df[params_df['productId'] == _id]['batch'])[0] for _id in list(match_res['productId'])]
+    match_res['subcategorycode'] = [list(params_df[params_df['productId'] == _id]['subcategorycode'])[0] for _id in list(match_res['productId'])]
+    match_res.sort_values("subcategorycode",inplace=True)
+    
     
     match_code_list = []
     match_name_list = []
     #classify data
-    category_list = params_df['subcategorycode'].unique().tolist()
+    category_list = match_res['subcategorycode'].unique().tolist()
     
     for category_code in category_list:
         print('开始匹配'+category_code+'数据')
@@ -136,7 +139,7 @@ def match_with_db(params_df):
         #get brand list
         brand_list = single_df['brandCode'].unique().tolist()
         #get productid
-        productid_list = single_df['productId'].unique().tolist()
+        productid_list = match_res[match_res['subcategorycode']==category_code]['productId'].unique().tolist()
         
         #get db data
         db_attr_data = get_db_data(category_code)
@@ -144,6 +147,7 @@ def match_with_db(params_df):
         
         #match
         for productid in productid_list:
+            
             #productid = '46262229631'
             matched_productcode_list = []
             matched_productname_list = []
@@ -173,6 +177,7 @@ def match_with_db(params_df):
             match_name_list.append(matched_productname_list)
     match_res['match_code'] = match_code_list
     match_res['match_name'] = match_name_list
+    del match_res['subcategorycode']
     
     return match_res
 
@@ -181,3 +186,4 @@ def match(df):
     params_df = get_params_df(df)
     res = match_with_db(params_df)
     return res
+

test_run.py

@@ -10,13 +10,14 @@ import match_program
 
 import pandas as pd
 import pymssql
+import pymysql
 
 '''
 test data load
 '''
 
 def get_test_data():
-    
+    '''
     conn = pymssql.connect('123.56.115.207','zgcindex','jiayou2017+2018','ZI_DataBase')
     cursor = conn.cursor()
     #cursor.execute("select * from vw_electricity_source_price where ProductCode in "+ code_list +" and periods = (select top 1 periods from vw_entry_source_price order by periods desc)")
@@ -25,6 +26,16 @@ def get_test_data():
         
     cursor.close()
     conn.close()
+    '''
+    conn = pymysql.connect('59.110.219.171','root','qwertyuiop1','product_chongqin')
+    cursor = conn.cursor()
+    #cursor.execute("select * from vw_electricity_source_price where ProductCode in "+ code_list +" and periods = (select top 1 periods from vw_entry_source_price order by periods desc)")
+    cursor.execute("select * from product")
+    test_df = pd.DataFrame([v for v in cursor.fetchall()],columns=[tuple[0] for tuple in cursor.description])
+        
+    cursor.close()
+    conn.close()
+    
     return test_df
 
 test_df = get_test_data()