graphene-django-optimizer

Optimize queries executed by graphene-django automatically, using select_related, prefetch_related and only methods of Django QuerySet.

Install

pip install graphene-django-optimizer

Usage

Having the following schema based on the tutorial of graphene-django (notice the use of gql_optimizer)

# cookbook/ingredients/schema.py
import graphene

from graphene_django.types import DjangoObjectType
import graphene_django_optimizer as gql_optimizer

from cookbook.ingredients.models import Category, Ingredient


class CategoryType(DjangoObjectType):
    class Meta:
        model = Category


class IngredientType(DjangoObjectType):
    class Meta:
        model = Ingredient


class Query(object):
    all_categories = graphene.List(CategoryType)
    all_ingredients = graphene.List(IngredientType)

    def resolve_all_categories(root, info):
        return gql_optimizer.query(Category.objects.all(), info)

    def resolve_all_ingredients(root, info):
        return gql_optimizer.query(Ingredient.objects.all(), info)

We will show some graphql queries and the queryset that will be executed.

Fetching all the ingredients with the related category:

{
  all_ingredients {
    id
    name
    category {
        id
        name
    }
  }
}

# optimized queryset:
ingredients = (
    Ingredient.objects
    .select_related('category')
    .only('id', 'name', 'category__id', 'category__name')
)

Fetching all the categories with the related ingredients:

{
  all_categories {
    id
    name
    ingredients {
        id
        name
    }
  }
}

# optimized queryset:
categories = (
    Category.objects
    .only('id', 'name')
    .prefetch_related(Prefetch(
        'ingredients',
        queryset=Ingredient.objects.only('id', 'name'),
    ))
)

Advanced usage

Sometimes we need to have a custom resolver function. In those cases, the field can't be auto optimized. So we need to use gql_optimizer.resolver_hints decorator to indicate the optimizations.

If the resolver returns a model field, we can use the model_field argument:

import graphene
from graphene_django.types import DjangoObjectType
import graphene_django_optimizer as gql_optimizer


class ItemType(DjangoObjectType):
    product = graphene.Field('ProductType')

    @gql_optimizer.resolver_hints(
        model_field='product',
    )
    def resolve_product(root, info):
        # check if user have permission for seeing the product
        if info.context.user.is_anonymous():
            return None
        return root.product

This will automatically optimize any subfield of product.

Now, if the resolver uses related fields, you can use the select_related argument:

import graphene
from graphene_django.types import DjangoObjectType
import graphene_django_optimizer as gql_optimizer


class ItemType(DjangoObjectType):
    name = graphene.String()

    @gql_optimizer.resolver_hints(
        select_related=('product', 'shipping'),
        only=('product__name', 'shipping__name'),
    )
    def resolve_name(root, info):
        return '{} {}'.format(root.product.name, root.shipping.name)

Finally, if your field has an argument for filtering results, you can use the prefetch_related argument with a function that returns a Prefetch instance as the value.

from django.db.models import Prefetch
import graphene
from graphene_django.types import DjangoObjectType
import graphene_django_optimizer as gql_optimizer


class CartType(DjangoObjectType):
    items = graphene.List(
        'ItemType',
        product_id=graphene.ID(),
    )

    @gql_optimizer.resolver_hints(
        prefetch_related=lambda info, product_id: Prefetch(
            'items',
            queryset=gql_optimizer.query(Item.objects.filter(product_id=product_id), info),
            to_attr='gql_product_id_' + product_id,
        ),
    )
    def resolve_items(root, info, product_id):
        return getattr(root, 'gql_product_id_' + product_id)

With these hints, any field can be optimized.

Optimize with non model fields

Sometimes we need to have a custom non model fields. In those cases, the optimizer would not optimize with the Django .only() method. So if we still want to optimize with the .only() method, we need to use disable_abort_only option:

class IngredientType(DjangoObjectType):
    calculated_calories = graphene.String()

    class Meta:
        model = Ingredient
    
    def resolve_calculated_calories(root, info):
        return get_calories_for_ingredient(root.id)


class Query(object):
    all_ingredients = graphene.List(IngredientType)

    def resolve_all_ingredients(root, info):
        return gql_optimizer.query(Ingredient.objects.all(), info, disable_abort_only=True)