Django Group par requête?

Question

J'utilise django ORM pour extraire des données en utilisant le groupe par requête. La requête postgresql brute est-

select date_trunc('day', time) as Time ,count(status) from table where group_id='2177' and status=1 group by date_trunc('day', time) order by time desc limit 7 offset 0; 

et retourne la sortie correcte ..

 time    | count 
---------------------------+------- 
2017-05-04 00:00:00+05:30 | 12 
2017-05-03 00:00:00+05:30 | 26 
2017-05-02 00:00:00+05:30 | 25 
2017-05-01 00:00:00+05:30 | 26 
2017-04-30 00:00:00+05:30 | 26 
2017-04-29 00:00:00+05:30 | 26 
2017-04-28 00:00:00+05:30 | 26 

(7 lignes)

J'utilise la fonction django annoter à acheive cela - voici la requête django -

records = TableModel.objects.filter(     
         group_id=group_id,           
         status=1,                  
         time__range = get_time_range(range)                  
         ).annotate(
            period=DateTrunc('day', 'time') , 
            count=Count('status') 
           ).annotate(
             period = DateTrunc('day', 'time') 
           ).order_by('-time') 

Sur Debug, j'ai découvert que django est i nternally convertir en -

SELECT "table"."id", "table"."time", "table"."status", "table"."group_id", DATE_TRUNC('day', "table"."time") AS "period", COUNT("table"."status") AS "count" FROM "table" WHERE ("table"."group_id" = '2177' AND "table"."status" = 1 AND "table"."time" BETWEEN '2017-04-28 11:47:21.421755+00:00' AND '2017-05-05 11:47:21.421755+00:00') GROUP BY "table"."id", DATE_TRUNC('day', "table"."time") ORDER BY "table"."time" DESC; 

ici est l'o/p (il contient plus de nombre de lignes) -

 id |   time   | status    | group_id  |   period   | count 
    --------+---------------------------+----------------------+------------+---------------------------+------- 
    267821 | 2017-05-04 10:36:13+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 
    267790 | 2017-05-04 09:36:35+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 
    267786 | 2017-05-04 09:30:44+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 
    267735 | 2017-05-04 08:36:09+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 
    267696 | 2017-05-04 07:36:32+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 
    267650 | 2017-05-04 06:36:14+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 
    267603 | 2017-05-04 05:36:14+05:30 |     1 | 2177  | 2017-05-04 00:00:00+05:30 |  1 

    .... 
    .... 
    (149 rows) 

-à-dire par défaut, il commence son groupe en utilisant (id, date_trunc ('day', "table". "time")) bien que je ne l'ai nulle part mentionné dans y Quark Django Query. Par conséquent, la sortie sera modifiée.

Est-ce que Django ORM prend par défaut le champ id? Y a-t-il un moyen de le résoudre?

Answer 1

Il peut être fixé par -

SELECT 1 as id ,DATE_TRUNC('day', "table"."time") AS "period", COUNT("table"."status") AS "count" FROM "table" WHERE "table"."group_id" = '2177' and "table"."status"=1 and time >= '2017-04-28 12:48:33.348682+00:00' and time <= '2017-05-05 12:48:33.348682+00:00' GROUP BY DATE_TRUNC('day', "table"."time"),period ORDER BY period DESC limit {} offset {}; 

par défaut, django doit revenir champ ID (primary_key).

Answer 2

Essayez cette

TableModel.objects.filter(     
     group_id=group_id,           
     status=1,                  
     time__range = get_time_range(range)                  
    ).annotate(
      period=DateTrunc('day', 'time') # Truncate to day and add to select query 
    ).values('period') # Group By period 
    .annotate(count=Count('status')) # add count of the grouping to select 
    .values('period', 'count') 
    .order_by('-time')