I'm doing a HTTP PUT call to update the data of an object with a nested relationship, and I'm met by the following error:
HTTP 400 Bad Request
"AttributeChoice with this slug already exists."
The reason why this is confusing is because I'm doing a HTTP PUT
call and I expect it to treat it as an UPDATE
and not a CREATE
.
My Models look like this:
class Attribute(models.Model):
name = models.CharField(max_length=100)
text_input = models.BooleanField(default=False)
slug = models.SlugField(unique=True)
class AttributeChoice(models.Model):
attribute = models.ForeignKey(Attribute)
value = models.CharField(max_length=100)
slug = models.SlugField(unique=True)
My Serializers look like this:
class AttributeChoiceSerializer(serializers.ModelSerializer):
class Meta:
model = AttributeChoice
fields = '__all__'
extra_kwargs = {'id': {'read_only': False}}
class AttributeSerializer(serializers.ModelSerializer):
attributechoice_set = AttributeChoiceSerializer(many=True)
class Meta:
model = Attribute
fields = ('id', 'name', 'text_input', 'slug', 'attributechoice_set')
def update(self, instance, validated_data):
choice_data = validated_data.pop('attributechoice_set')
for choice in choice_data:
# If id is within the call, then update the object with matching id
if 'id' in choice:
try:
choice_obj = AttributeChoice.objects.get(pk=choice['id'])
choice_obj.value = choice['value']
choice_obj.slug = choice['slug']
choice_obj.attribute = instance
# If ID is not found, then create a new object
except AttributeChoice.DoesNotExist:
choice_obj = AttributeChoice(**choice)
# If no ID within the call, create a new object.
else:
choice_obj = AttributeChoice(**choice)
choice_obj.save()
return instance
Debug:
Even if I remove the update()
function, I still get the same error. I believe the error is reported from when .is_valid()
is called in the ViewSet. So it's not the update()
that causes it.
Also, if I remove attributechoice_set = AttributeChoiceSerializer(many=True)
and just include the attributechoice_set
in the fields = ()
, the error disappears, but I need that line for the rest of the code to work.
Even through you're doing an update, it doesn't mean the nested data will just be updated.
You're simply saying that you want to update the top most object.
In some cases, you'll be removing or creating new nested objects while updating the top most one.
Therefore DRF considers by default that nested objects are for creation. You can work around this by explicitly removing the unique constraint on the nested serializer:
I think it is because of the validators.
Like:
Django rest serializer Breaks when data exists
As my solution, I mark this nested field to
read_only=True
, And do my own update, create function to accessself.initial_data
to handle myself.