Using Elixir/Absinthe/Ecto/Dataloader, how do you query/filter a source by multiple fields?
Let's say you would want to filter a schema (and dataloader source) named User
by two fields, one named is_admin
(only true
values) and one group_id
that can be any value in a list, e.g [1, 5, 9]
How would this look using Dataloader ?
This is from the schema definition:
alias App.Data
# ...
object :app do
field :users, list_of(:user) do
resolve fn app, _args, %{context: %{loader: loader}} ->
params = [is_admin: true, group_id: [1,2,3]]
loader
|> Data.load_many(:users, params)
|> on_load(fn loader ->
{:ok, Data.get_many(loader, :user, params)}
end)
end
end
end
and this is the data source module:
defmodule App.Data do
alias App.User
def data do
Dataloader.Ecto.new(Repo, run_batch: &run_batch/5)
end
def load_many(loader, :users, [is_admin: is_admin, group_id: ids]) do
Dataloader.load(loader, __MODULE__, {:many, {User, is_admin: is_admin}}, group_id: ids)
end
def get_many(loader, :users, [is_admin: is_admin, group_id: ids]) do
Dataloader.load(loader, __MODULE__, {:many, {User, is_admin: is_admin}}, group_id: ids)
end
def run_batch(queryable, query, col, inputs, repo_opts) do
Dataloader.Ecto.run_batch(Repo, queryable, query, col, inputs, repo_opts)
end
end
So to answer my own question, the solution that worked was to leverage the &query/2
of Dataloader so the resolver in GQL resolver would basically look like this:
loader
|> Data.load({:many, User, group_ids: ids}, is_admin: true)
|> on_load(fn loader ->
loader
|> Data.get({:many, User, group_ids: ids}, is_admin: true)
# ...
end)
... and the query/2
method in the App.Data
module would look like this:
defp query(User, %{group_ids: ids}) do
from u in User, where: u.group_id in ^ids
end
This is explained here:
https://hexdocs.pm/lazyloader/Dataloader.Ecto.html#module-filtering-ordering