add printer

word_vector/answer.py

@@ -35,7 +35,7 @@ def get_answer_click_data(spark, start, end):
           where action = 'page_view'
             AND partition_date BETWEEN '{}' AND '{}'
             AND page_name='answer_detail'
-            AND page_stay>=0.5
+            AND page_stay>=1
             AND cl_id is not null
             AND cl_id != ''
             AND business_id is not null
@@ -145,11 +145,13 @@ def get_online_answer_ids():
 
 def get_device_click_answer_ids_dict(click_df):
     online_ids = get_online_answer_ids()
+    print("online ids: {}".format(len(online_ids)))
     res = defaultdict(list)
     cols = click_df.orderBy("partition_date", ascending=False).collect()
     for i in cols:
         card_id = i["card_id"]
         session_id = i["app_session_id"]
+        print("card_id: {} {}".format(int(card_id), int(card_id) in online_ids)))
         if (card_id not in res[session_id]) and int(card_id) in online_ids:
             res[session_id].append(card_id)
     return res