# Django nested serializer prefetching before validation
**URL:** https://forum.djangoproject.com/t/django-nested-serializer-prefetching-before-validation/21774
**Category:** Using the ORM
**Created:** [June 22, 2023, 9:12am UTC](https://forum.djangoproject.com/t/django-nested-serializer-prefetching-before-validation/21774 "2023-06-22T09:12:22Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@arpika99](https://forum.djangoproject.com/u/arpika99)
#### Post date: [June 22, 2023, 9:12am UTC](https://forum.djangoproject.com/t/django-nested-serializer-prefetching-before-validation/21774/1 "2023-06-22T09:12:22Z")
Hello,
I am having some difficulty with prefetching nested model objects, so the number of repeated queries are causing big performance issues.
I already tried to learn from a couple of blog posts, but I did not succeed. Maybe I am just too simple minded.
example:
> **[Speed up Django Nested Foreign Key Serializers w/ prefetch\_related](https://medium.com/quant-five/speed-up-django-nested-foreign-key-serializers-w-prefetch-related-ae7981719d3f)**
>
> If you’re here just for the solution, JUMP HERE.
models.py
```auto
class Project(models.Model):
project_name = models.CharField(max_length=255)
status = models.CharField(max_length= 255)
owner = models.ForeignKey(User, null=True, on_delete=models.CASCADE)
user_id = models.PositiveIntegerField()
def __str__ (self) -> str:
return self.project_name
class Country(models.Model):
SUPPORTED = "supported"
NOT_SUPPORTED = "not_supported"
SUSPENDED = "suspended"
STATUS_CHOICES = [
(SUPPORTED, "Supported"),
(NOT_SUPPORTED, "Not supported"),
(SUSPENDED, "Suspended"),
]
name = models.CharField(max_length=255)
iso_code2 = models.CharField(max_length=2, primary_key=True)
iso_code3 = models.CharField(max_length=3, unique=True)
status = models.CharField(max_length=255, choices=STATUS_CHOICES, default=NOT_SUPPORTED)
class Meta:
verbose_name_plural = "Countries"
ordering = ["name"]
def __str__ (self) -> str:
return self.name
class Site(models.Model):
project = models.ForeignKey(Project, on_delete=models.CASCADE, null=True, related_name="sites")
site_id1 = models.CharField(max_length=255)
site_id2 = models.CharField(max_length=255, blank=True)
country_code = models.ForeignKey(Country, on_delete=models.CASCADE)
postal_code = models.CharField(max_length=255)
city = models.CharField(max_length=255)
address = models.CharField(max_length=512)
```
serializers.py
```auto
class SiteSerializer(serializers.ModelSerializer):
def validate(self, attrs):
error_details = {}
country = attrs.get("country_code")
if country.status != "supported":
error_details.setdefault("country", []).append(f"The given country_code: '{country.iso_code2}' ({country.name}) is currently not supported. Current status: {country.status}")
if error_details:
raise serializers.ValidationError(detail=error_details, code="config_error")
return super().validate(attrs)
class Meta:
model = Site
fields = [
"site_id1",
"site_id2",
"country_code",
"city",
"postal_code",
"address",
]
class FileProjectSerializer(serializers.ModelSerializer):
sites = SiteSerializer(many=True, min_length= 1, max_length= 3000, required=True)
class Meta:
model = Project
fields = [
"project_name",
"status",
"sites",
"owner",
"user_id",
]
read_only_fields = [
"owner",
"status",
]
def create(self, validated_data):
sites = validated_data.pop("sites",[])
project = Project.objects.create(**validated_data)
for site in sites:
Site.objects.create(project=project, **site)
return project
```
views.py
```auto
class FileUploadView(generics.CreateAPIView):
parser_classes = [FormParser, MultiPartParser, FileUploadParser]
permission_classes = [permissions.IsAuthenticated]
serializer_class = FileProjectSerializer
def perform_create(self, serializer):
return serializer.save()
def create(self, request, *args, **kwargs):
file_obj = request.FILES.get("file", None)
if file_obj is None:
raise ParseError(detail=f"No file found.")
#file converting I did not include these since it is only conversion to dict with pandas
sites = self.get_data_from_file_as_dict(file_obj)
sites = self.translate_file_keys_to_model(sites)
project_name = file_obj.name.replace(".xlsx", "").replace(".csv","")
project_name = project_name.encode("ascii", "replace").decode()
#preparing data for serializer (user_id will be sent via request, now its just static)
project = {
"project_name": project_name,
"sites": sites,
"user_id": 2345678543
}
#serializing
serializer = self.get_serializer(data=project)
serializer.is_valid(raise_exception=True)
instance = self.perform_create(serializer)
response_data = serializer.data
response = {
"project_name": response_data["project_name"],
"sites": response_data["sites"]
}
#deleting the instance because it is only a file upload which returns a validated json with additional default values
#maybe there is a better solution, but that might be another topic
instance.delete()
return Response(data= response, status=status.HTTP_200_OK)
```
Silky results
Any help would be appreciated.
---
### Author:  [@DanielGnzlzVll](https://forum.djangoproject.com/u/DanielGnzlzVll)
#### Post date: [June 22, 2023, 12:11pm UTC](https://forum.djangoproject.com/t/django-nested-serializer-prefetching-before-validation/21774/2 "2023-06-22T12:11:31Z")
Hi,
I couldn’t see your “get\_queryset” method, would you add it ?
one improvement i can see so far is using bulk create in your serializer, have you tried it ?
---
### Author:  [@arpika99](https://forum.djangoproject.com/u/arpika99)
#### Post date: [June 22, 2023, 1:18pm UTC](https://forum.djangoproject.com/t/django-nested-serializer-prefetching-before-validation/21774/3 "2023-06-22T13:18:23Z")
Hello Daniel,
I have not included it into my viewset, because after multiple tests I realised that it is not even called in case of post method.
I also looked after this on drf-docs.
According to drf-docs, it is used for list views and retrieve views.
[Django rest framework docs](https://www.django-rest-framework.org/api-guide/generic-views/#get_querysetself)
On stackoverflow I found the following:
[Stackoverflow: do we need query set in CreateAPIView?](https://stackoverflow.com/questions/70504043/do-we-need-query-set-in-createapiview)
If I remember correctly I had something as following, but scrapped it after realised it is not even called…
```auto
def get_queryset(self):
queryset = Project.objects.prefetch_related(
"sites",
"sites__country_code"
)
return queryset
```
“one improvement i can see so far is using bulk create in your serializer, have you tried it ?”
Regarding the bulk create method there is a slight change in my actual models. The nesting does not stop at Site model, but I did not include all the submodels, so it would be easier to understand.