SQLAlchemy: Select rows, grouped by a date column, whose datetime column is most recent for each date

Question

Given the following table in a PostgreSQL database.

value	created_at	date
2	2022-02-01 01:01:01	2022-03-15
10	2022-02-15 01:01:01	2022-03-15
5	2022-02-02 01:01:01	2022-03-16
6	2022-02-16 01:01:01	2022-03-16

How can you write sqlalchemy code which:

1. Groups by the date column
2. Get's the most recent created_at value for each date
3. Returns all rows with meet that criteria

The output of above query should be the following:

value	created_at	date
10	2022-02-15 01:01:01	2022-03-15
6	2022-02-16 01:01:01	2022-03-16

I have tried the following query:

query = (
            select(
                func.max(TABLE.created_at),
                TABLE.date,
                TABLE.value
            )
            .group_by(models.TABLE.date)
        )

But get the error "value" must appear in the GROUP BY clause or be used in an aggregate function

While, based on my understanding, using an aggregate function or including "value" in the GROUP BY clause will not provide the desired result.

What query can I use to get desired rows?

Answer 1

I was looking for the distinct clause

query = select(TABLE).distinct(
            TABLE.date
        ).order_by(
            TABLE.date,
            TABLE.created_at.desc(),
        )

Answer 2

I had the same problem at the project that I am doing, but after reading documentation about relationships and distinct i made this query. You can also try like this, hope it helps.

query = (
         select value, created_at, date, count(distinct(date)) from TABLE group by value, created_at, date
        )