Skip to content

dataloaders.mdx

Latest commit

 

History

History
129 lines (92 loc) · 3.69 KB

dataloaders.mdx

File metadata and controls

129 lines (92 loc) · 3.69 KB
Raw
title DataLoaders

DataLoaders

DataLoaders help solve the N+1 query problem by batching and caching database queries. For basic information about DataLoaders, see the Strawberry DataLoaders documentation.

Tip

Strawberry Django's Query Optimizer handles most optimization scenarios automatically. Use DataLoaders when you need custom batching logic or are working with external data sources.

Using DataLoaders with Django

Basic Example

Here's a basic DataLoader for fetching Django models:

from strawberry.dataloader import DataLoader
from asgiref.sync import sync_to_async

from . import models


async def load_authors(keys: list[int]) -> list[models.Author | None]:
    """Batch load authors by their IDs."""
    # Build map from async queryset iteration
    author_map = {
        author.id: author async for author in models.Author.objects.filter(id__in=keys)
    }
    return [author_map.get(key) for key in keys]

Sharing DataLoaders Across a Request

Important: DataLoaders must be instantiated once per request and shared across all resolvers to enable batching. Store them in the GraphQL context:

from strawberry.dataloader import DataLoader
from .dataloaders import load_authors
from . import models


class GraphQLContext:
    def __init__(self, request):
        self.request = request
        self._author_loader: DataLoader[int, models.Author | None] | None = None

    @property
    def author_loader(self) -> DataLoader[int, models.Author | None]:
        if self._author_loader is None:
            self._author_loader = DataLoader(load_fn=load_authors)
        return self._author_loader


def get_context(request):
    return GraphQLContext(request)
from django.urls import path
from strawberry.django.views import AsyncGraphQLView
from .schema import schema
from .context import get_context

urlpatterns = [
    path(
        "graphql/", AsyncGraphQLView.as_view(schema=schema, context_getter=get_context)
    ),
]

Using the DataLoader

import strawberry_django
from strawberry import field
from strawberry.types import Info


@strawberry_django.type(models.Book)
class Book:
    id: strawberry.auto
    title: strawberry.auto
    author_id: strawberry.Private[int]

    @field
    async def author(self, info: Info) -> "Author":
        return await info.context.author_loader.load(self.author_id)

Common Patterns

One-to-Many Relationships

from collections import defaultdict


async def load_books_by_author(author_ids: list[int]) -> list[list[models.Book]]:
    """Load all books for multiple authors."""
    books_by_author = defaultdict(list)
    async for book in models.Book.objects.filter(author_id__in=author_ids):
        books_by_author[book.author_id].append(book)

    return [books_by_author[author_id] for author_id in author_ids]

Many-to-Many Relationships

async def load_tags(article_ids: list[int]) -> list[list[models.Tag]]:
    """Load tags for multiple articles."""
    tags_by_article = {
        article.id: await sync_to_async(list)(article.tags.all())
        async for article in models.Article.objects.filter(
            id__in=article_ids
        ).prefetch_related("tags")
    }

    return [tags_by_article.get(article_id, []) for article_id in article_ids]

See Also