My target SQL is the following, which is valid,
SELECT a.agreement_group_id,
(select id from agreement_t where agreement_group_id = a.agreement_group_id and
active = 'Y'),
...
FROM ets.agreement_t a
WHERE requester_uniqueidentifier = '0010079170'
GROUP BY a.agreement_group_id
ORDER BY a.agreement_group_id
But SqlAlchemy is producing the following -- and complaining that I don't have anon_1
in GROUP BY due to its placement of the sub-select in FROM
,
SELECT agreement_t_1.agreement_group_id AS agreement_t_1_agreement_group_id,
anon_1.id AS anon_1_id,
...
FROM ets.agreement_t AS agreement_t_1,
(SELECT ets.agreement_t.id AS id
FROM ets.agreement_t, ets.agreement_t AS agreement_t_1
WHERE ets.agreement_t.agreement_group_id = agreement_t_1.agreement_group_id AND
ets.agreement_t.active = 'Y') AS anon_1
WHERE agreement_t_1.requester_uniqueidentifier = '0010079170'
GROUP BY agreement_t_1.agreement_group_id, anon_1.id
ORDER BY agreement_t_1.agreement_group_id
Python SqlAlchemy code:
agreement = aliased(AgreementT)
subqueryActive = db_session.query(AgreementT.id).filter(
(AgreementT.agreement_group_id == agreement.agreement_group_id),
(AgreementT.active == 'Y')
).subquery()
result = (db_session.query(
agreement.agreement_group_id,
subqueryActive,
...
.filter(*filters)
.group_by(agreement.agreement_group_id)
.order_by(agreement.agreement_group_id)
.all())
I don't need any other Joins. As you can see, the subquery subqueryActive
already references the alias agreement
which is used in the main query. So why is the Sub-Select not placed properly in the SELECT
, but rather in the FROM
, with the following error?
psycopg2.errors.GroupingError: column "anon_1.id" must appear in the GROUP BY clause or be used in an aggregate function
LINE 1: ...ent_group_id AS agreement_t_1_agreement_group_id, anon_1.id ...
^