问题
I am trying to run an aggregation query that is roughly equals to:
select
sum(impressions) as impressions,
sum(clicks) as clicks,
sum(clicks)/sum(impressions) as ctr
from stats
group by product
order by ctr;
The database used is PostgreSQL.
I made this query expression (Django 1.9):
Stats.objects.values('product').annotate(
impressions = models.Sum('impressions'),
clicks = models.Sum('clicks'),
ctr = models.ExpressionWrapper(
models.F('clicks')/models.F('impressions')), output_field = models.FloatField()
)
).order_by('ctr')
There are 2 problems with it:
ctr
is 0.0 because it divides integers at the database level- it throws
division by zero
if impressions are 0
What is the proper solution?
回答1:
Use conditional expressions:
from django.db.models import Case, F, Sum, When
Stats.objects.values('product').annotate(
tot_impressions=Sum('impressions'),
tot_clicks=Sum('clicks')
).annotate(
ctr=Case(When(tot_impressions=0, then=None), # or other value, e.g. then=0
# 1.0*... is to get float in SQL
default=1.0*F('tot_clicks')/F('tot_impressions'),
output_field=models.FloatField())
).order_by('ctr')
回答2:
The exact math on how you scale this is up to you, but adding in a python constant inside a nested ExpressionWrapper should suffice:
ctr = models.ExpressionWrapper(
models.F('clicks')/models.ExpressionWrapper(
models.F('impressions') + 1, output_field = models.FloatField()
),
output_field = models.FloatField()
)
回答3:
You can use the django-pg-utils
package for the divide
query expression that handles division by zero and returns float values
from pg_utils import divide
Stats.objects.annotate(ctr=divide(F('clicks'), F('impressions')))
来源:https://stackoverflow.com/questions/34733760/django-query-expression-for-calculated-fields-that-require-conditions-and-castin