Neomodel objects + REST framework

[ptchankue]

Hi there!
I would like to use neomodel to perform REST operations on a neo4j db.
From the documentation I can create a model and save nodes in the database.

I tried to get all the nodes by doing a normal django get all: persons = Person.objects.all(), but neomodel doesn't support objects. This makes it difficult to write a serializer that will feed a REST framework.

Do you have a solution for this issue?

Regards,
Patrick

[theY4Kman]

Each StructuredNode subclass has a nodes object, which serves the same purpose as objects from Django's ORM. Some examples can be found in the neomodel docs. Here's another example for your case:

from neomodel import StructuredNode, StringProperty

class Person(StructuredNode):
    name = StringProperty()

people = Person.nodes.all()

[bleib1dj]

@ptchankue expanding on @theY4Kman's answer I've used neomodel with the Django REST Framework utilizing something like Person.nodes.all() where you would normally use Person.objects.all() to access an object from the Django ORM. So you can quickly build a ViewSet by doing something along the lines of:

views.py

class PersonViewSet(viewsets.ModelViewSet):
    serializer_class = PersonSerializer
    lookup_field = "username"
    queryset = Person.nodes.all()

    def get_object(self):
        return Person.nodes.get(username=self.kwargs[self.lookup_field])

serializers.py

from rest_framework import serializers

class PlebSerializerNeo(serializers.Serializer):
    username = serializers.CharField()

If you are using an old version of DRF this may throw some errors due to it not using the core ORM but as of 3.1.2 which includes PR #2745 those issues should be resolved.

[auvipy]

just curious, is drf serializer capable of serializing neomodels datatype?

[robinedwards]

I have never tried so I guess not but patches would be most welcome

UPDATE: I had to make a custom serializer to get it to work.

I get stuck here in the serializer.

It looks like you would need to add something like:

class DjangoNode(StructuredNode):
        ...
        Meta = 
                concrete_model = ...

... in order to mimic what DRF is looking for from a regular model so that it can work with this line of code that the ModelSerializer is looking for. Otherwise this fails when running the serializer of your model.

# https://github.com/encode/django-rest-framework/blob/master/rest_framework/utils/model_meta.py
opts = model._meta.concrete_model._meta

So it looks like DRF wants data from the concrete_model meta of regular model metas, which includes the fields like pk, fields, and relationships.

def get_field_info(model):
    opts = model._meta.concrete_model._meta

    pk = _get_pk(opts)
    fields = _get_fields(opts)
    forward_relations = _get_forward_relationships(opts)
    reverse_relations = _get_reverse_relationships(opts)
    fields_and_pk = _merge_fields_and_pk(pk, fields)
    relationships = _merge_relationships(forward_relations, reverse_relations)

    return FieldInfo(pk, fields, forward_relations, reverse_relations,
                     fields_and_pk, relationships)

PK can just be the first field [0] in the model... you could start your relationship names in the model with rel_up* or rel_down and assume the rest are regular fields.

Bah, then would need to call those relationships to get the values... this is deep.

You know what, sounds like we would just write our own serializer for DRF.

[Ramel]

Is there a way to summarize the use of neomodel + DRF with a Django demo app code?

@Ramel yes, you could use Python to do this.

[Ramel]

I'm viewing here a lot of code pieces, and I cannot figure out the all picture to get it work. I assume you (@HashRocketSyntax) have write a custom serializer as you say in your update. But is there a way to document more your usage of DRF+Django? Is it different of this example: paradise-paper-tutorial-serialize?