Primary key requirement in raw SQL complicates the query in Django

ε祈祈猫儿з 提交于 2020-01-01 09:39:18

问题


To get max value from a simple table of values, I can write the following query in Django:

MyTable.objects.aggregate(Max('value'))

The SQL generated is : 'SELECT MAX("mytable"."value") AS "value__max" FROM "mytable"'

Now if I write the same SQL using the raw query manager:

1. MyTable.objects.raw('SELECT max(value) FROM mytable')

Django throws an error InvalidQuery: Raw query must include the primary key. This is also mentioned in Django docs: "There is only one field that you can’t leave out - the primary key field". So after adding the id field, I need GROUP BY as well. The new query becomes:

2. MyTable.objects.raw('SELECT id, max(value) FROM mytable GROUP BY id')

This doesn't give me a single max value anymore because I'm forced to use GROUP BY id. Now I need to add an ORDER BY and LIMIT statement to get the expected answer for an otherwise simple SQL statement that work.

3. MyTable.objects.raw('SELECT id, max(value) AS mv FROM mytable GROUP BY id ORDER BY mv DESC LIMIT 1')

Is there a way simplify the above query i.e. not use ORDER/LIMIT/GROUP BY (FWIW, using PosgreSQL)?

Update:

Here's a hack that'll work. I alias the max value as id to make Django happy. Is there any issue here?

MyTable.objects.raw('SELECT max(value) AS id FROM mytable')

Update 2:

Here's the query plan for the simple SQL (1) vs the complicated final one (3):

"Aggregate  (cost=5.25..5.26 rows=1 width=2) (actual time=0.155..0.155 rows=1 loops=1)"
"  ->  Seq Scan on mytable  (cost=0.00..4.60 rows=260 width=2) (actual time=0.018..0.067 rows=260 loops=1)"
"Total runtime: 0.222 ms"


"Limit  (cost=9.80..9.80 rows=1 width=6) (actual time=0.548..0.548 rows=1 loops=1)"
"  ->  Sort  (cost=9.80..10.45 rows=260 width=6) (actual time=0.545..0.545 rows=1 loops=1)"
"        Sort Key: (max(value))"
"        Sort Method: top-N heapsort  Memory: 25kB"
"        ->  HashAggregate  (cost=5.90..8.50 rows=260 width=6) (actual time=0.328..0.432 rows=260 loops=1)"
"              ->  Seq Scan on mytable  (cost=0.00..4.60 rows=260 width=6) (actual time=0.018..0.069 rows=260 loops=1)"
"Total runtime: 0.638 ms"

P.S. The actual query is more complicated (somewhat related to this answer : https://dba.stackexchange.com/a/86404/52114)


回答1:


You should use custom SQL instead of Manager.raw() method:

from django.db import connection

cursor = connection.cursor()
cursor.execute('SELECT max(value) FROM mytable')
max_value = cursor.fetchone()[0]



回答2:


U can use

ModelName.objects.raw('SELECT 1 as id , max(value) FROM mytable')



回答3:


I would do something like:

select id, value from mytable order by value desc limit 1


来源:https://stackoverflow.com/questions/27560912/primary-key-requirement-in-raw-sql-complicates-the-query-in-django

易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!