rewrite session data

word_vector/tractate.py

@@ -45,9 +45,9 @@ def get_spark(app_name=""):
 def get_tracate_click_data(spark, start, end):
     reg = r"""^\\d+$"""
     sql = """
-        SELECT DISTINCT t1.partition_date, t1.cl_id, cast(t1.business_id as int) card_id
+        SELECT DISTINCT t1.partition_date, t1.cl_id, cast(t1.business_id as int) card_id, t1.app_session_id
           FROM
-          (select partition_date,cl_id,business_id,action,page_name,page_stay
+          (select partition_date,cl_id,business_id,action,page_name,page_stay, app_session_id
           from online.bl_hdfs_maidian_updates
           where action = 'page_view'
             AND partition_date BETWEEN '{}' AND '{}'
@@ -149,7 +149,10 @@ def get_device_click_tractate_ids_dict(click_df):
     res = defaultdict(list)
     cols = click_df.orderBy("partition_date", ascending=False).collect()
     for i in cols:
-        res[i["cl_id"]].append(i["card_id"])
+        card_id = i["card_id"]
+        session_id = i["app_session_id"]
+        if card_id not in res[session_id]:
+            res[session_id].append(card_id)
     return res