处理爬虫模型

process_data.py

@@ -74,7 +74,10 @@ class DataProcessor(object):
             case "useful":
                 item_features = [item[i] for i in useful_features]
             case "user":
-                item_features = [item[i] for i in user_features]
+                if item['type'] == "userupload":
+                    item_features = [item[i] for i in user_features]
+                else:
+                    return None, None
             case "spider":
                 if item['type'] == "spider":
                     item_features = [item[i] for i in spider_features]
@@ -151,10 +154,10 @@ if __name__ == "__main__":
     flag = int(input("please input method train or predict:\n "))
     if flag == 1:
         t = "train"
-        D = DataProcessor(flag=t, c="spider")
+        D = DataProcessor(flag=t, c="user")
         D.producer(dt="whole")
     else:
         t = "predict"
-        D = DataProcessor(flag=t, c="spider")
+        D = DataProcessor(flag=t, c="user")
         for d in range(16, 22):
             D.producer(d)