Datasource
- Automatically preload associations for your serializers
- Specify custom SQL snippets for virtual attributes (Query attributes)
- Write custom preloading logic in a reusable way
Install
Add to Gemfile
gem 'datasource'
bundle install
rails g datasource:install
Upgrade
rails g datasource:install
ORM support
- ActiveRecord
- Sequel
Serializer support
- active_model_serializers
Simple Mode
Datasource is configured to run in Simple mode by default, which makes it easier to start with, but disables some advanced optimizations. See Advanced mode for more information after you understand Simple mode.
Associations
The most noticable magic effect of using Datasource in Simple mode (Advanced mode has other benefits) is that associations will automatically be preloaded using a single query.
class PostSerializer < ActiveModel::Serializer
attributes :id, :title
end
class UserSerializer < ActiveModel::Serializer
attributes :id
has_many :posts
end
SELECT users.* FROM users
SELECT posts.* FROM posts WHERE id IN (?)
This means you do not need to call includes
yourself. It will be done
automatically by Datasource.
Show action
You will probably want to reuse the same preloading logic in your show action.
You will need to call for_serializer
on the scope before you call find
.
You can optionally give it the serializer class as an argument.
class PostsController < ApplicationController
def show
post = Post.for_serializer.find(params[:id])
# more explicit:
# post = Post.for_serializer(PostSerializer).find(params[:id])
render json: post
end
end
You can also use it on an existing record, but doing it this way may result in an additional SQL query (for example if you use query attributes).
class UsersController < ApplicationController
def show
user = current_user.for_serializer
render json: user
end
end
Query attribute
You can specify a SQL fragment for SELECT
and use that as an attribute on your
model. As a simple example you can concatenate 2 strings together in SQL:
class User < ActiveRecord::Base
datasource_module do
query :full_name do
"users.first_name || ' ' || users.last_name"
end
end
end
class UserSerializer < ActiveModel::Serializer
attributes :id, :full_name
end
SELECT users.*, (users.first_name || ' ' || users.last_name) AS full_name FROM users
Note: If you need data from another table, use a join in a loader (see below).
Loader
You might want to have some more complex preloading logic. In that case you can use a loader. A loader will receive ids of the records, and needs to return a hash. The key of the hash must be the id of the record for which the value is.
A loader will only be executed if a computed attribute depends on it. See
Advanced mode for
information about computed attributes (but this works the same way in Simple mode).
A more simple alternative to loader which doesn't require computed attributes is to use
Loaded.
If an attribute depends on multiple loaders, pass an array of loaders like
so computed :attr, loaders: [:loader1, :loader2]
.
Be careful that if your hash does not contain a value for the object ID, the loaded value
will be nil. However you can use the default
option for such cases (see below example).
class User < ActiveRecord::Base
datasource_module do
computed :post_count, loader: :post_counts
loader :post_counts, array_to_hash: true, default: 0 do |user_ids|
results = Post
.where(user_id: user_ids)
.group(:user_id)
.pluck("user_id, COUNT(id)")
end
end
end
class UserSerializer < ActiveModel::Serializer
attributes :id, :post_count
def post_count
# Will automatically give you the value for this user's ID
object.loaded_values[:post_counts]
end
end
SELECT users.* FROM users
SELECT user_id, COUNT(id) FROM posts WHERE user_id IN (?)
Datasource provides shortcuts to transform your data into a hash. Here are examples:
loader :stuff, array_to_hash: true do |ids|
[[1, "first"], [2, "second"]]
# will be transformed into
# { 1 => "first", 2 => "second" }
end
loader :stuff, group_by: :user_id do |ids|
Post.where(user_id: ids)
# will be transformed into
# { 1 => [#<Post>, #<Post>, ...], 2 => [ ... ], ... }
end
loader :stuff, group_by: :user_id, one: true do |ids|
Post.where(user_id: ids)
# will be transformed into
# { 1 => #<Post>, 2 => #<Post>, ... }
end
loader :stuff, group_by: "user_id", one: true do |ids|
# it works the same way on an array of hashes
# but be careful about Symbol/String difference
[{ "title" => "Something", "user_id" => 10 }]
# will be transformed into
# { 10 => { "title" => "Something", "user_id" => 10 } }
end
Loaded
Loaded is the same as loader, but it automatically creates a computed attribute and defines a method with the same name on your model.
Here is the previous example with loaded
instead of loader
:
class User < ActiveRecord::Base
datasource_module do
loaded :post_count, array_to_hash: true, default: 0 do |user_ids|
results = Post
.where(user_id: user_ids)
.group(:user_id)
.pluck("user_id, COUNT(id)")
end
end
end
class UserSerializer < ActiveModel::Serializer
attributes :id, :post_count
# Note that the User now has a generated post_count method
end
When using loaded
, if you already have the method with this name defined in your
model, datasource will automatically create a 'wrapper' method that will use the
loaded value if available (when you are using a serializer/datasource), otherwise
it will fallback to your original method. This way you can still use the same
method when you are not using a serializer/datasource. For example:
class User < ActiveRecord::Base
datasource_module do
loaded :post_count, array_to_hash: true, default: 0 do |user_ids|
results = Post
.where(user_id: user_ids)
.group(:user_id)
.pluck("user_id, COUNT(id)")
end
end
def post_count
posts.count
end
end
class UserSerializer < ActiveModel::Serializer
attributes :id, :post_count # <- post_count will be read from loaded_values
end
User.first.post_count # <- your method will be called
Getting Help
If you find a bug, please report an Issue.
If you have a question, you can also open an Issue.
Contributing
- Fork it ( https://github.com/mrbrdo/datasource/fork )
- Create your feature branch (
git checkout -b my-new-feature
) - Commit your changes (
git commit -am 'Add some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create a new Pull Request