Django DRF: how to groupby on a foreign fields? - django

Question

I have a model where users can upvote other users for specific topics. Something like:
#models.py
Class Topic(models.Model):
name = models.StringField()
def __str__(self):
return str(self.name)
Class UserUpvotes(models.Model):
"""Holds total upvotes by user and topic"""
user = models.ForeignKey(User)
topic= models.ForeignKey(Topic)
upvotes = models.PositiveIntegerField(default=0)
Using DRF, I have an API that returns the following: topic_id, topic_name, and upvotes, which is the total upvotes for a given topic.
One of the project requirements is for the API to use these field names specifically: topic_id, topic_name, and upvotes
#serializers.py
class TopicUpvotesSerializer(serializers.ModelSerializer):
topic_name = serializers.StringRelatedField(source="topic")
class Meta:
model = UserUpvotes
fields = ["topic_id", "topic_name", "upvotes"]
My trouble is aggregating these fields. I'm filtering the UserUpvotes by user or team and then aggregating by topic.
Desired output
This is the result I want to get. When I don't perform any aggregations (and there are views where this will be the case), it works.
[
{
"topic_id": 3,
"topic_name": "Korean Studies",
"upvotes": 14
},
{
"topic_id": 12,
"topic_name": "Inflation",
"upvotes": 3
},
]
At first, I tried creating a TopicSerializer, and then assigning it to the topic field in TopicUpvotesSerializer. But then, the resulting json would have a nested "topic" field and the aggragation would fail.
Attempt 1
#views.py
def get_queryset(self):
return (
UserUpvotes.objects.filter(user__team=team)
.values("topic")
.annotate(upvotes=models.Sum("upvotes"))
.order_by("-upvotes")
)
My problem is that the topic_id and topic_name fields are not showing. I get something like:
[
{
"topic_name": "3",
"upvotes": 14
},
{
"topic_name": "12",
"upvotes": 3
},
]
Attempt 2
Another queryset attempt:
# views.py
def get_queryset(self):
return (
UserUpvotes.objects.filter(user__team=team)
.values("topic__id", "topic__name")
.annotate(upvotes=models.Sum("upvotes"))
.order_by("-upvotes")
)
Which yields:
[
{
"upvotes": 14
},
{
"upvotes": 3
},
]
The aggregation worked on the queryset level, but the serializer failed to find the correct fields.
Attempt 3
This was the closest I got:
# views.py
def get_queryset(self):
return (
UserUpvotes.objects.filter(user__team=team)
.values("topic__id", "topic__name")
.annotate(upvotes=models.Sum("upvotes"))
.values("topic_id", "topic", "upvotes")
.order_by("-upvotes")[:n]
)
[
{
"topic_name": 3,
"topic_name": "3",
"upvotes": 14
},
{
"topic_name": 12,
"topic_name": "12",
"upvotes": 3
},
]
I have no idea why "topic_name" is simply transforming the "topic_id" into a string, instead of calling the string method.

Answer 1

Work with a serializer for the topic:
class TopicSerializer(serializers.ModelSerializer):
upvotes = serializers.IntegerField(read_only=True)
class Meta:
model = Topic
fields = ['id', 'name', 'upvotes']
then in the ModelViewSet, you annotate:
from django.db.models import Sum
from rest_framework.viewsets import ModelViewSet
class TopicViewSet(ModelViewSet):
serializer_class = TopicSerializer
queryset = Topic.objects.annotate(upvotes=Sum('userupvotes__upvotes'))

Answer 2

Desired output
This is the result I want to get. When I don't perform any aggregations (and there are views where this will be the case), it works.
[
{
"topic_name": 3,
"topic_name": "Korean Studies",
"upvotes": 14
},
{
"topic_name": 12,
"topic_name": "Inflation",
"upvotes": 3
},
]
The serialized FK will always give you the ID of the related model. I am not sure why you name it topic_name if that is equal to an ID. Now, if you really want to get the name field of the Topic model
in the topic_name = serializers.StringRelatedField(source="topic") you should give it a source="topic.name"
However, if you trying to get the ID of the relation you can still use ModelSerializer :
class TopicUpvotesSerializer(serializers.ModelSerializer):
class Meta:
model = UserUpvotes
fields = "__all__"

Answer 3

#willem-van-onsem's answer is the correct one for the problem as I had put it.
But... I had another use case (sorry! ◑﹏◐), for when the Users API used UserUpvotes serializer as a nested field. So I had to find another solution. This is was I eventually ended up with. I'm posting in case it helps anyone.
class UserUpvotesSerializer(serializers.ModelSerializer):
topic_name = serializers.SerializerMethodField()
def get_topic_name (self, obj):
try:
_topic_name = obj.topic.name
except TypeError:
_topic_name = obj.get("skill__name", None)
return _topic_name
class Meta:
model = UserUpvotes
fields = ["topic_id", "topic_name", "upvotes"]
I still have no idea why the SerializerMethodField works and the StringRelatedField field doesn't. It feels like a bug?
Anyways, the rub here is that, after the values().annotate() aggregation, obj is no longer a QuerySet, but a dict. So accessing namedirectly will give you a 'UserUpvotes' object is not subscriptable error.
I don’t know if there are any other edge cases I should be aware of (this is when I REALLY miss type hints in Django), but it works so far