Opened 2 years ago

Last modified 21 months ago

#25534 new New feature

Allow using datetime lookups in QuerySets aggregate calls

Reported by: Raúl Pedro Santos Owned by: nobody
Component: Database layer (models, ORM) Version: master
Severity: Normal Keywords:
Cc: josh.smeaton@…, info+coding@… Triage Stage: Accepted
Has patch: no Needs documentation: no
Needs tests: no Patch needs improvement: no
Easy pickings: no UI/UX: no

Description (last modified by Raúl Pedro Santos)

I've been scouring the web for an answer to this but the closest I can find is #25339, which is almost what I need but not quite, so I think I can safely conclude that it is just not possible with the currently available functionality and thus I'm opening this ticket to suggest adding it.

I would like to be able to use datetime lookups in an aggregate() call on a QuerySet.

My specific use case is this: I have a set of electricity consumption readings, each with a datetime field (and a few others). I need to sum the consumption and cost values grouped by month, day, year, week, etc. In other words, I need to be able to get the total energy consumption value and corresponding cost for each month, day, year, week, etc.

This is my ElectricityReading model and its parent Reading model (separated because we also have consumption readings for water and gas, which also derive from Reading):

from model_utils.models import TimeStampedModel
# Other imports here...

class Reading(TimeStampedModel):
    device = models.ForeignKey(Device)
    datetime = models.DateTimeField() # Terrible property name, I know :)
    manual = models.BooleanField(default=False)
    inserted_by = models.ForeignKey(User)

    class Meta:
        abstract = True

class ElectricityReading(Reading):
    vph1 = models.DecimalField(max_digits=18, decimal_places=3, null=True)
    vph2 = models.DecimalField(max_digits=18, decimal_places=3, null=True)
    vph3 = models.DecimalField(max_digits=18, decimal_places=3, null=True)
    wh_imp = models.DecimalField(max_digits=18, decimal_places=3)
    varh = models.DecimalField(max_digits=18, decimal_places=3, null=True)
    pf = models.DecimalField(max_digits=18, decimal_places=3, null=True)

    price = models.ForeignKey(ElectricityPrice)
    consumption = models.DecimalField(max_digits=18, decimal_places=3,
                                      null=True, blank=True, default=None)
    cost = models.DecimalField(max_digits=18, decimal_places=3, null=True,
                               blank=True, default=None)

I think the code I need is something along the lines of the following:

result = ElectricityReading.objects\
        .filter(device__grid__building_id=1) \
        .annotate(num_readings=Count('id'))\
        .annotate(total_consumption=Sum('consumption'))\
        .annotate(total_cost=Sum('cost'))\
        .aggregate(total=Count('datetime__month'))

Right now I'm doing this with this raw SQL:

SELECT
        (EXTRACT(YEAR FROM datetime)) AS reading_date_year,
        (EXTRACT(MONTH FROM datetime)) AS reading_date_month,
        (EXTRACT(DAY FROM datetime)) AS reading_date_day,
        (EXTRACT(HOUR FROM datetime)) AS reading_date_hour,
        SUM(consumption) as total,
        COUNT(id) as num_readings,
        SUM(cost) as total_cost,
        price_id
FROM electricity_reading
WHERE device_id IN (1, 2, 3)
        AND datetime >= '2015-10-01'
        AND datetime <= '2015-10-10'
GROUP BY reading_date_year, reading_date_month, reading_date_day,reading_date_hour, price_id

The part I can't seem to replicate with Django's ORM is the GROUP BY clause at the end, which is what I was expecting to be able to achieve using the aggregate(total=Count('datetime__month')) but instead I get the following error:

FieldError: Cannot resolve keyword 'datetime' into field. Choices are: consumption, cost, created, datetime, device, device_id, id, inserted_by, inserted_by_id, manual, modified, pf, price, price_id, varh, vph1, vph2, vph3, wh_imp, num_readings, total_consumption, total_cost

I would love that someone would tell me I am missing something, and if that's the case, please do! :)

Otherwise, I believe it would be beneficial to add this.

Change History (8)

comment:1 Changed 2 years ago by Raúl Pedro Santos

Description: modified (diff)

comment:2 Changed 2 years ago by Tim Graham

It would be helpful if you could include the simplest set of models so we can try the query.

comment:3 Changed 2 years ago by Tim Graham

Is this a duplicate of #10302?

comment:4 Changed 2 years ago by Raúl Pedro Santos

Sorry, I should have included the model. I'll edit the original post and add it.

As for the possible duplicate, I did see that post but it didn't look like the same thing (correct me if I'm wrong).

comment:5 Changed 2 years ago by Raúl Pedro Santos

Description: modified (diff)

comment:6 Changed 2 years ago by Josh Smeaton

Keywords: QuerySet.extra removed
Triage Stage: UnreviewedAccepted
Version: 1.7master

It's not a duplicate, no, but they are related. #10302 wants transform/lookup support in values() whereas this ticket is asking for support in aggregates/expressions.

1.9 converts transforms into func expressions, so we'll be able to do something like (simplifying the model here..):

from django.db.models.lookups import MonthTransform as Month
result = ElectricityReading.objects.aggregate(total=Count(Month('datetime')))

Which isn't quite as nice as Count('datetime__month'). It should be possible to convert the latter into the former internally though. I would imagine this would be handled internally within F(). Detect if we're trying to access a transform, extract the transform, wrap the original field, and continue as normal. This example (datetime part extraction) is probably the canonical usecase for transform support in aggregates.

If transforms can be supported with underscore syntax within F() objects, then that should solve #10302 as well. There are probably a few more tickets that could be closed with this implementation.

comment:7 Changed 2 years ago by Josh Smeaton

Cc: josh.smeaton@… added

comment:8 Changed 21 months ago by Markus Holtermann

Cc: info+coding@… added
Note: See TracTickets for help on using tickets.
Back to Top