update

new_user_analysis/new_user_has_protratit_rate.py

@@ -178,30 +178,41 @@ WHERE spam_pv.device_id IS NULL
     new_urser_device_id_df.createOrReplaceTempView("device_id_view")
     new_urser_device_id_df.show(1)
     sql_res = new_urser_device_id_df.collect()
+    res_dict = {}
+    portrait_dict = {
+        "first_demands": {},
+        "second_demands": {},
+        "first_solutions": {},
+        "second_solutions": {},
+        "first_positions": {},
+        "second_positions": {},
+        "projects": {},
+    }
 
     print("-------------------------------")
-    for count, res in enumerate(sql_res):
+    count_not_has_portratit = 0
+    for count_user_count, res in enumerate(sql_res):
         # print(count, res)
         portratit_res = get_user_portrait_tag3_from_redis(res.device_id)
-        print(count, res, portratit_res)
-
-    # res_dict = {}
-    #
+        # print(count_user_count, res, portratit_res)
+        temp_count = 0
+        for demand in portratit_res:
+            if portratit_res[demand]:
+                for tag in portratit_res[demand][0:3]:
+                   if tag in portrait_dict[demand]:
+                       portrait_dict[demand][tag] += 1
+                   else:
+                       portrait_dict[demand][tag] = 1
+                temp_count += 1
+        if not temp_count:
+            count_not_has_portratit += 1
+
+
+    print(portrait_dict)
+    print(count_user_count+1,count_not_has_portratit)
     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)
-    #
-    # exception_list = ("report_status","is_open_push")
+
+
     # for action in res_dict:
     #     partition_date = yesterday_str
     #     pid = hashlib.md5((partition_date + action).encode("utf8")).hexdigest()