update

new_user_analysis/new_user_has_protratit_rate.py

@@ -240,3 +240,30 @@ WHERE spam_pv.device_id IS NULL
         print(res)
     # cursor.executemany()
     db.close()
+
+    res_dict = {}
+    maidian_sql = """select t1.device_id,t2.track as track from 
+        (select device_id from device_id_view where device_id in {device_id_tuple})t1 
+        left join 
+        (select cl_id, concat_ws(',', collect_list(action)) as track from
+    (select * from online.bl_hdfs_maidian_updates where partition_date = {partition_date} and cl_id is not null ) group by cl_id) t2 
+    on t1.device_id = t2.cl_id""".format(partition_date=today_str,device_id_tuple=str(tuple(no_portrait_device_id_list)))
+    print(maidian_sql)
+    track_df = spark.sql(maidian_sql)
+    # track_df.createOrReplaceTempView("device_id_view")
+    track_df.show(1)
+    sql_res = track_df.collect()
+
+    print("-------------------------------")
+    for count, res in enumerate(sql_res):
+        # print(count, res)
+        track = res.track
+        if not track:
+            continue
+        track_list = track.split(",")
+        for one_key_word in track_list:
+            if one_key_word in res_dict:
+                res_dict[one_key_word] += 1
+            else:
+                res_dict[one_key_word] = 1
+    print(res_dict)