Notice
Inherited Resources is no longer actively maintained by the original author and has been transferred to the ActiveAdmin organization for maintenance. New feature requests are not encouraged.
If you are not already using Inherited Resources we suggest instead using Rails'
respond_with
feature alongside the responders gem.
Inherited Resources
Inherited Resources speeds up development by making your controllers inherit all restful actions so you just have to focus on what is important. It makes your controllers more powerful and cleaner at the same time.
In addition to making your controllers follow a pattern, it helps you to write better code by following fat models and skinny controllers convention. There are two screencasts available besides this README:
- http://railscasts.com/episodes/230-inherited-resources
- https://www.akitaonrails.com/2009/09/01/screencast-real-thin-restful-controllers-with-inherited-resources
Installation
You can let bundler install Inherited Resources by adding this line to your application's Gemfile:
gem 'inherited_resources'
And then execute:
$ bundle install
Or install it yourself with:
$ gem install inherited_resources
HasScope
Since Inherited Resources 1.0, has_scope is not part of its core anymore but a gem dependency. Be sure to check the documentation to see how you can use it:
And it can be installed as:
$ gem install has_scope
Responders
Since Inherited Resources 1.0, responders are not part of its core anymore, but is set as Inherited Resources dependency and it's used by default by InheritedResources controllers. Be sure to check the documentation to see how it will change your application:
And it can be installed with:
$ gem install responders
Using responders will set the flash message to :notice and :alert. You can change that through the following configuration value:
InheritedResources.flash_keys = [ :success, :failure ]
Notice the CollectionResponder won't work with InheritedResources, as InheritedResources hardcodes the redirect path based on the current scope (like belongs to, polymorphic associations, etc).
Basic Usage
To use Inherited Resources you just have to inherit (duh) it:
class ProjectsController < InheritedResources::Base
end
And all actions are defined and working, check it! Your projects collection
(in the index action) is still available in the instance variable @projects
and your project resource (all other actions) is available as @project
.
The next step is to define which mime types this controller provides:
class ProjectsController < InheritedResources::Base
respond_to :html, :xml, :json
end
You can also specify them per action:
class ProjectsController < InheritedResources::Base
respond_to :html, :xml, :json
respond_to :js, :only => :create
respond_to :iphone, :except => [ :edit, :update ]
end
For each request, it first checks if the "controller/action.format" file is
available (for example "projects/create.xml") and if it's not, it checks if
the resource respond to :to_format (in this case, :to_xml
). Otherwise returns 404.
Another option is to specify which actions the controller will inherit from
the InheritedResources::Base
:
class ProjectsController < InheritedResources::Base
actions :index, :show, :new, :create
end
Or:
class ProjectsController < InheritedResources::Base
actions :all, :except => [ :edit, :update, :destroy ]
end
In your views, you will get the following helpers:
resource #=> @project
collection #=> @projects
resource_class #=> Project
As you might expect, collection (@projects
instance variable) is only available
on index actions.
If for some reason you cannot inherit from InheritedResources::Base
, you can
call inherit_resources in your controller class scope:
class AccountsController < ApplicationController
inherit_resources
end
One reason to use the inherit_resources
macro would be to ensure that your controller
never responds with the html mime-type. InheritedResources::Base
already
responds to :html
, and the respond_to
macro is strictly additive.
Therefore, if you want to create a controller that, for example, responds ONLY via :js
,
you will have to write it this way:
class AccountsController < ApplicationController
respond_to :js
inherit_resources
end
By default, InheritedResources::Base
will inherit from ::ApplicationController
. You can change this in a rails initializer:
# config/initializers/inherited_resources.rb
InheritedResources.parent_controller = 'MyController'
Overwriting defaults
Whenever you inherit from InheritedResources, several defaults are assumed.
For example you can have an AccountsController
for account management while the
resource is a User
:
class AccountsController < InheritedResources::Base
defaults :resource_class => User, :collection_name => 'users', :instance_name => 'user'
end
In the case above, in your views you will have @users
and @user
variables, but
the routes used will still be accounts_url
and account_url
. If you plan also to
change the routes, you can use :route_collection_name
and :route_instance_name
.
Namespaced controllers work out of the box, but if you need to specify a different route prefix you can do the following:
class Administrators::PeopleController < InheritedResources::Base
defaults :route_prefix => :admin
end
Then your named routes will be: admin_people_url
, admin_person_url
instead
of administrators_people_url
and administrators_person_url
.
If you want to customize how resources are retrieved you can overwrite collection and resource methods. The first is called on index action and the second on all other actions. Let's suppose you want to add pagination to your projects collection:
class ProjectsController < InheritedResources::Base
protected
def collection
get_collection_ivar || set_collection_ivar(end_of_association_chain.paginate(:page => params[:page]))
end
end
The end_of_association_chain
returns your resource after nesting all associations
and scopes (more about this below).
InheritedResources also introduces another method called begin_of_association_chain
.
It's mostly used when you want to create resources based on the @current_user
and
you have urls like "account/projects". In such cases you have to do
@current_user.projects.find
or @current_user.projects.build
in your actions.
You can deal with it just by doing:
class ProjectsController < InheritedResources::Base
protected
def begin_of_association_chain
@current_user
end
end
Overwriting actions
Let's suppose that after destroying a project you want to redirect to your root url instead of redirecting to projects url. You just have to do:
class ProjectsController < InheritedResources::Base
def destroy
super do |format|
format.html { redirect_to root_url }
end
end
end
You are opening your action and giving the parent action a new behavior. On the other hand, I have to agree that calling super is not very readable. That's why all methods have aliases. So this is equivalent:
class ProjectsController < InheritedResources::Base
def destroy
destroy! do |format|
format.html { redirect_to root_url }
end
end
end
Since most of the time when you change a create, update or destroy action you do so because you want to change its redirect url, a shortcut is provided. So you can do:
class ProjectsController < InheritedResources::Base
def destroy
destroy! { root_url }
end
end
If you simply want to change the flash message for a particular action, you can
pass the message to the parent action using the keys :notice
and :alert
(as you
would with flash):
class ProjectsController < InheritedResources::Base
def create
create!(:notice => "Dude! Nice job creating that project.")
end
end
You can still pass the block to change the redirect, as mentioned above:
class ProjectsController < InheritedResources::Base
def create
create!(:notice => "Dude! Nice job creating that project.") { root_url }
end
end
Now let's suppose that before create a project you have to do something special but you don't want to create a before filter for it:
class ProjectsController < InheritedResources::Base
def create
@project = Project.new(params[:project])
@project.something_special!
create!
end
end
Yes, it's that simple! The nice part is since you already set the instance variable
@project
, it will not build a project again.
Same goes for updating the project:
class ProjectsController < InheritedResources::Base
def update
@project = Project.find(params[:id])
@project.something_special!
update!
end
end
Before we finish this topic, we should talk about one more thing: "success/failure blocks". Let's suppose that when we update our project, in case of failure, we want to redirect to the project url instead of re-rendering the edit template.
Our first attempt to do this would be:
class ProjectsController < InheritedResources::Base
def update
update! do |format|
unless @project.errors.empty? # failure
format.html { redirect_to project_url(@project) }
end
end
end
end
Looks too verbose, right? We can actually do:
class ProjectsController < InheritedResources::Base
def update
update! do |success, failure|
failure.html { redirect_to project_url(@project) }
end
end
end
Much better! So explaining everything: when you give a block which expects one argument it will be executed in both scenarios: success and failure. But if you give a block that expects two arguments, the first will be executed only in success scenarios and the second in failure scenarios. You keep everything clean and organized inside the same action.
Smart redirects
Although the syntax above is a nice shortcut, you won't need to do it frequently because (since version 1.2) Inherited Resources has smart redirects. Redirects in actions calculates depending on the existent controller methods.
Redirects in create and update actions calculates in the following order: resource_url
,
collection_url
, parent_url
(which we are going to see later), and root_url
. Redirect
in destroy action calculate in following order collection_url
, parent_url
, root_url
.
Example:
class ButtonsController < InheritedResources::Base
belongs_to :window
actions :all, :except => [:show, :index]
end
This controller redirect to parent window after all CUD actions.
Success and failure scenarios on destroy
The destroy action can also fail, this usually happens when you have a
before_destroy
callback in your model which returns false. However, in
order to tell InheritedResources that it really failed, you need to add
errors to your model. So your before_destroy
callback on the model should
be something like this:
def before_destroy
if cant_be_destroyed?
errors.add(:base, "not allowed")
false
end
end
Belongs to
Finally, our Projects are going to get some Tasks. Then you create a
TasksController
and do:
class TasksController < InheritedResources::Base
belongs_to :project
end
belongs_to
accepts several options to be able to configure the association.
For example, if you want urls like "/projects/:project_title/tasks", you can
customize how InheritedResources find your projects:
class TasksController < InheritedResources::Base
belongs_to :project, :finder => :find_by_title!, :param => :project_title
end
It also accepts :route_name
, :parent_class
and :instance_name
as options.
Check the lib/inherited_resources/class_methods.rb
for more.
Nested belongs to
Now, our Tasks get some Comments and you need to nest even deeper. Good practices says that you should never nest more than two resources, but sometimes you have to for security reasons. So this is an example of how you can do it:
class CommentsController < InheritedResources::Base
nested_belongs_to :project, :task
end
If you need to configure any of these belongs to, you can nest them using blocks:
class CommentsController < InheritedResources::Base
belongs_to :project, :finder => :find_by_title!, :param => :project_title do
belongs_to :task
end
end
Warning: calling several belongs_to
is the same as nesting them:
class CommentsController < InheritedResources::Base
belongs_to :project
belongs_to :task
end
In other words, the code above is the same as calling nested_belongs_to
.
Polymorphic belongs to
We can go even further. Let's suppose our Projects can now have Files, Messages and Tasks, and they are all commentable. In this case, the best solution is to use polymorphism:
class CommentsController < InheritedResources::Base
belongs_to :task, :file, :message, :polymorphic => true
# polymorphic_belongs_to :task, :file, :message
end
You can even use it with nested resources:
class CommentsController < InheritedResources::Base
belongs_to :project do
belongs_to :task, :file, :message, :polymorphic => true
end
end
The url in such cases can be:
/project/1/task/13/comments
/project/1/file/11/comments
/project/1/message/9/comments
When using polymorphic associations, you get some free helpers:
parent? #=> true
parent_type #=> :task
parent_class #=> Task
parent #=> @task
Right now, Inherited Resources is limited and does not allow you to have two polymorphic associations nested.
Optional belongs to
Later you decide to create a view to show all comments, independent if they belong to a task, file or message. You can reuse your polymorphic controller just doing:
class CommentsController < InheritedResources::Base
belongs_to :task, :file, :message, :optional => true
# optional_belongs_to :task, :file, :message
end
This will handle all those urls properly:
/comment/1
/tasks/2/comment/5
/files/10/comment/3
/messages/13/comment/11
This is treated as a special type of polymorphic associations, thus all helpers are available. As you expect, when no parent is found, the helpers return:
parent? #=> false
parent_type #=> nil
parent_class #=> nil
parent #=> nil
Singletons
Now we are going to add manager to projects. We say that Manager
is a singleton
resource because a Project
has just one manager. You should declare it as
has_one
(or resource) in your routes.
To declare an resource of current controller as singleton, you just have to give the
:singleton
option in defaults.
class ManagersController < InheritedResources::Base
defaults :singleton => true
belongs_to :project
# singleton_belongs_to :project
end
So now you can use urls like "/projects/1/manager".
In the case of nested resources (when some of the can be singletons) you can declare it separately
class WorkersController < InheritedResources::Base
#defaults :singleton => true #if you have only single worker
belongs_to :project
belongs_to :manager, :singleton => true
end
This is correspond urls like "/projects/1/manager/workers/1".
It will deal with everything again and hide the action :index from you.
Namespaced Controllers
Namespaced controllers works out the box.
class Forum::PostsController < InheritedResources::Base
end
Inherited Resources prioritizes the default resource class for the namespaced controller in this order:
Forum::Post
ForumPost
Post
URL Helpers
When you use InheritedResources it creates some URL helpers. And they handle everything for you. :)
# /posts/1/comments
resource_url # => /posts/1/comments/#{@comment.to_param}
resource_url(comment) # => /posts/1/comments/#{comment.to_param}
new_resource_url # => /posts/1/comments/new
edit_resource_url # => /posts/1/comments/#{@comment.to_param}/edit
edit_resource_url(comment) # => /posts/1/comments/#{comment.to_param}/edit
collection_url # => /posts/1/comments
parent_url # => /posts/1
# /projects/1/tasks
resource_url # => /projects/1/tasks/#{@task.to_param}
resource_url(task) # => /projects/1/tasks/#{task.to_param}
new_resource_url # => /projects/1/tasks/new
edit_resource_url # => /projects/1/tasks/#{@task.to_param}/edit
edit_resource_url(task) # => /projects/1/tasks/#{task.to_param}/edit
collection_url # => /projects/1/tasks
parent_url # => /projects/1
# /users
resource_url # => /users/#{@user.to_param}
resource_url(user) # => /users/#{user.to_param}
new_resource_url # => /users/new
edit_resource_url # => /users/#{@user.to_param}/edit
edit_resource_url(user) # => /users/#{user.to_param}/edit
collection_url # => /users
parent_url # => /
Those urls helpers also accepts a hash as options, just as in named routes.
# /projects/1/tasks
collection_url(:page => 1, :limit => 10) #=> /projects/1/tasks?page=1&limit=10
In polymorphic cases, you can also give the parent as parameter to collection_url
.
Another nice thing is that those urls are not guessed during runtime. They are
all created when your application is loaded (except for polymorphic
associations, that relies on Rails' polymorphic_url
).
Custom actions
Since version 1.2, Inherited Resources allows you to define custom actions in controller:
class ButtonsController < InheritedResources::Base
custom_actions :resource => :delete, :collection => :search
end
This code creates delete and search actions in controller (they behaves like show and
index actions accordingly). Also, it will produce delete_resource_{path,url}
and
search_resources_{path,url}
url helpers.
What about views?
Sometimes just DRYing up the controllers is not enough. If you need to DRY up your views, check this Wiki page:
https://github.com/activeadmin/inherited_resources/wiki/Views-Inheritance
Notice that Rails 3.1 ships with view inheritance built-in.
Some DSL
For those DSL lovers, InheritedResources won't leave you alone. You can overwrite your success/failure blocks straight from your class binding. For it, you just need to add a DSL module to your application controller:
class ApplicationController < ActionController::Base
include InheritedResources::DSL
end
And then you can rewrite the last example as:
class ProjectsController < InheritedResources::Base
update! do |success, failure|
failure.html { redirect_to project_url(@project) }
end
end
Strong Parameters
If your controller defines a method named permitted_params
, InheritedResources
will call it where it would normally call params. This allows for easy
integration with the strong_parameters gem:
def permitted_params
params.permit(:widget => [:permitted_field, :other_permitted_field])
end
Remember that if your field is sent by client to server as an array, you have to write :permitted_field => []
, not just :permitted_field
.
Note that this doesn't work if you use strong_parameters' require method instead of permit, because whereas permit returns the entire sanitized parameter hash, require returns only the sanitized params below the parameter you required.
If you need params.require
you can do it like this:
def permitted_params
{:widget => params.fetch(:widget, {}).permit(:permitted_field, :other_permitted_field)}
end
Or better yet just override #build_resource_params
directly:
def build_resource_params
[params.fetch(:widget, {}).permit(:permitted_field, :other_permitted_field)]
end
Instead you can stick to a standard Rails 4 notation (as rails scaffold generates) and write:
def
params.require(:widget).permit(:permitted_field, :other_permitted_field)
end
In such case you should remove #permitted_params method because it has greater priority.
Funding
If you want to support us financially, you can help fund the project through a Tidelift subscription. By buying a Tidelift subscription you make sure your whole dependency stack is properly maintained, while also getting a comprehensive view of outdated dependencies, new releases, security alerts, and licensing compatibility issues.
Bugs and Feedback
If you discover any bugs, please describe it in the issues tracker, including Rails and InheritedResources versions.
Questions are better handled on StackOverflow.
MIT License. Copyright (c) 2009-2017 José Valim.
Security contact information
Please use the Tidelift security contact to report a security vulnerability. Tidelift will coordinate the fix and disclosure.