View Source Dataloader

Dataloader provides an easy way efficiently load data in batches. It's inspired by


def deps do
    {:dataloader, "~> 1.0.0"}


The core concept of dataloader is a data source which is just a struct that encodes a way of retrieving data. More info in the Sources section.


Absinthe provides some dataloader helpers out of the box that you can import into your schema

import Absinthe.Resolution.Helpers, only: [dataloader: 1]

This is needed to use the various dataloader helpers to resolve a field:

field(:posts, list_of(:post), resolve: dataloader(Blog))

Let's start with a data source. Dataloader data sources are just structs that encode a way of retrieving data in batches. In a Phoenix application you'll generally have one source per context, so that each context can control how its data is loaded.

Here is a hypothetical Blog context and a dataloader ecto source:

defmodule MyApp.Blog do
  def data() do, query: &query/2)

  def query(queryable, _params) do

In this example, the query returned by query/2 is used as a starting point by Dataloader to build the final query, which it does by traversing schema associations. In other words, Dataloader can determine that an author has many posts, and that to retrieve posts it needs to get those with the relevant author_id. If that's sufficient for your needs, query/2 need not modify the query it's given. But if you only want to load published posts, query/2 can narrow the query accordingly.

When integrating Dataloader with GraphQL, we want to place it in our context so that we can access it in our resolvers. In your schema module add:

alias MyApp.{Blog, Foo}

def context(ctx) do
  loader =
    |> Dataloader.add_source(Blog,
    |> Dataloader.add_source(Foo, # Foo source could be a Redis source

  Map.put(ctx, :loader, loader)

def plugins do
  [Absinthe.Middleware.Dataloader] ++ Absinthe.Plugin.defaults()

The context/1 function is a callback specified by the Absinthe.Schema behaviour that gives the schema itself an opportunity to set some values in the context that it may need in order to run.

The plugins/0 function has been around for a while, and specifies what plugins the schema needs to resolve. See the documentation for more.

Unpacking Dataloader

The data/0 function creates an Ecto data source, to which you pass your repo and a query function. This query function is called every time you want to load something, and provides an opportunity to apply arguments or set defaults. So for example if you always want to only load non-deleted posts you can do:

def query(Post, _), do: from p in Post, where: is_nil(p.deleted_at)

def query(queryable, _), do: queryable

Now any time you're loading posts, you'll just get posts that haven't been deleted.

We can also use the context to ensure access conditions, so we can only show deleted posts for admins:

def query(Post, %{has_admin_rights: true}), do: Post

def query(Post, _), do: from p in Post, where: is_nil(p.deleted_at)

def query(queryable, _), do: queryable

Helpfully, those rules are defined within your context, helping ensure that it has the final say about data access.


Dataloader ships with two different built in sources:

  • Dataloader.Ecto - for easily pulling out data with ecto
  • Dataloader.KV - a simple KV key value source.


Here is a simple example of a loader using the KV source in combination with absinthe:

defmodule MyProject.Loaders.Nhl do
  @teams [%{
    id: 1,
    name: "New Jersey Devils",
    abbreviation: "NJD"
    id: 2,
    name: "New York Islanders",
    abbreviation: "NYI"
  # etc.

  def data() do

  def fetch(:teams, [%{}]) do
      %{} => @teams

  def fetch(:team, args) do
   # must return a map keyed by the args
   # args is a list of the args used to resolve your field
   # for example, if you have arg(:foo, non_null(:string))
   # args will look like: [%{foo: "value of foo here")}]

    |> Enum.reduce(%{}, fn(%{id: id} = arg, result) ->
      Map.put(result, arg, find_team(id))

  def fetch(_batch, args) do
    args |> Enum.reduce(%{}, fn(arg, accum) -> Map.put(accum, arg, nil) end)

  defp find_team(id) do
    @teams |> Enum.find(fn(t) -> t |> Map.get(:id) == id end)

Dataloader.KV requires a load function that accepts a batch and args. It must return a map of values keyed by the args. This is the purpose of the fetch/2 function. The dataloader/1 helper we imported above uses the field name as the batch, and a map where the argument name is the key. For example: fetch(:team, [%{ id: 1 }])

Pattern matching can be used to fetch differently depending on the batch. For example, when the :teams batch is requested, the args will actually be an empty map (i.e. %{}).

If you’re interested in more generic use of Dataloader, see the dataloader project source.