How to group by two columns on queryset in Django2? - django

Question

I am getting little confused about how to use .annotate on quesryset.
To be quick: I have a model:
class Row(models.Model):
order = models.ForeignKey('order.Header', blank=True, null=True)
qty = models.IntegerField(blank=True, null=True, default=0)
name = models.CharField(default='', blank=True, null=True)
total = models.DecimalField(max_digits=10, decimal_places=2,default=0, blank=True, null=True)
profit = models.DecimalField(max_digits=10,decimal_places=2,default=0, blank=True, null=True)
profit_percent = models.DecimalField(max_digits=6,decimal_places=2,default=0, blank=True, null=True)
month_sold = models.IntegerField(default=0)
month_painted = models.IntegerField(default=0)
area_painted_1 = models.DecimalField(max_digits=5,decimal_places=2,default=0, blank=True, null=True)
area_painted_2 = models.DecimalField(max_digits=5,decimal_places=2,default=0, blank=True, null=True)
What I need to do is to create a kind of a summary, that will tell me month by month, a sum of Total, Profit Avg of profit, and also a sum of the painted area.
Something like that:
+-------+-------+--------+----------+--------+--------+
| month | Total | Profit | Profit % | area_1 | area_2 |
+-------+-------+--------+----------+--------+--------+
| 0 | 23000 | 3000 | 13% | 55 | 12 |
| Jan | 10000 | 1000 | 10% | 43 | 44 |
| April | 20000 | 1000 | 5% | 99 | 134 |
+-------+-------+--------+----------+--------+--------+
I tried to achieve that with .annotate:
result = Row.objects.values('month_sold') \
.annotate(total=Sum('total')+1) \
.annotate(profit=Sum('profit'))
.annotate(profit_percent=Round(F('profit')/F('total')*100, 2))
.annotate(area_2=Sum('area_painted_2'))
.annotate(area_1=Sum('area_painted_1'))
.values('month_sold', 'total', 'profit', 'profit_percent',
'area_1', 'area_2')
.order_by('moth_sold')
But obviously, it groups by month_sold. So total, profit values are good, but I don't know how to get area_1 and _2 by month_painted.
Any indications or ideas how can I solve it?

Answer 1

I'm not sure I've got you right. In your table "Something like that", do you want month to refer to different fields in your model (either month_sold or month_painted) depending on what aggregate you're looking at? So for Total and Profit, it's month_sold, and for area_1 and area_2 it's month_painted?
If that's the case, you're not going to achieve it with one single query. In raw SQL, you could join the table with itself on month_sold = month_painted; in Djano's ORM, I believe you'd need subqueries for each aggregate that is not grouped on the month type of the main query. For instance:
sq1 = (
Row.objects
.filter(month_painted=OuterRef('month_sold'))
.values('month_painted')
.annotate(area_1=Sum('area_painted_1'))
.values('area_1')
)
sq2 = (
Row.objects
.filter(month_painted=OuterRef('month_sold'))
.values('month_painted')
.annotate(area_2=Sum('area_painted_2'))
.values('area_2')
)
result = (
Row.objects
.values('month_sold')
.annotate(total=Sum('total')+1)
.annotate(profit=Sum('profit'))
.annotate(profit_percent=Round(F('profit')/F('total')*100, 2))
.annotate(area_1=Subquery(sq1, output_field=models.IntegerField()))
.annotate(area_2=Subquery(sq2, output_field=models.IntegerField()))
.values('month_sold', 'total', 'profit', 'profit_percent',
'area_1', 'area_2')
.order_by('month_sold')
)
Which month fields (month_sold or month_painted) the main query and the subqueries are base based on depends on which month type you want to be the outer part of the outer join, ie. which month type you want to include even if there are no corresponding values for the other month type. To include both (= FULL OUTER JOIN) using the ORM, you'd first have to get a list of all months (whether painted or sold), and then pull in the other columns as individual subqueries.