In my app I've got various resources that I want to access via multiple URLs. For example, an invoice can be accessed via:
/invoices/:id
By the issuer of the invoice, and also via:
/pay/:payment_key
By the payer.
The latter URL does not require authentication (hence the secrefied payment_key).
The issue is how to get automatic URL helper methods for the custom URL. Usually you could use to_param to customize the resource URL (as described here), but that's not adequate in this case because I still want to retain the default resource URL.
If I create the helper methods by declaring a named route:
/pay/:payment_key, :as => :invoice_payment
Then I would expect invoice_payment_url(invoice) to include invoice.payment_key but it doesn't. Rails uses the invoice ID instead (similar to the behaviour reported here)
This seems like it's broken.
So I've been defining the url helpers for this resource manually.
def invoice_payment_path (invoice)
url_for :controller => "invoices",
:only_path => true,
:action => "pay",
:payment_key => invoice.payment_key
end
def invoice_payment_url (invoice)
url_for :controller => "invoices",
:only_path => false,
:action => "pay",
:payment_key => invoice.payment_key
end
Wondering if there is a DRYer way to do this?
RESTful is about resources. So are you sure the "payment" and "invoice" resources the same thing in your system? To me, it is more like a system design issue than a routing issue.
Another thought is using nested resources. You can view either "/payments/:id/invoices" or "/invoices/:id/payments", both make sense to me.
Yan
Related
Still pretty new to Rails, so if I'm taking the completely wrong approach, please feel free to straiten me out.
How do I make routes.rb aware that there's a root controller full of actions that don't manipulate models, while preserving the route helper methods? I'd like it to respond to requests like these:
http://localhost/download
http://localhost/share
With route helpers like
download_app_path
share_path
but without static named routes like these:
match '/download' => 'site#download', :as => :download_app
match '/share' => 'site#share', :as => :share
from a SiteController that doesn't create, show, or otherwise manipulates models from my app.
I've tried using an approach like this, but it works without generating the route helpers ( naturally )
match '/:action', :controller => 'site'
I could theoretically do without the route helpers, but I think they're a bit easier to read than passing hashes of url options to link_to or form methods.
Is there a way to accomplish a more resourceful root controller, or is what I'm trying to do unconventional for Rails?
Edit
For clarity, here's what this SiteController class looks like:
class SiteController < ApplicationController
def download
#custom_options = { .. }
end
def share
#custom_options = { .. }
end
def about
end
end
Its purpose is to allow me to collect pages that don't interact with resources ( such as Users or Friendships ) into a single controller and maintain them all in one place. I'm trying to set this controller up as the application root controller - so all paths from this controller will be directly off the app host ( myapp.com/download )
Thanks in advance!
routes and resources are not tied to models. it's just a RESTful convention. if you just want to use the index actions, in your example download and share could be done like
resouce :download, only: [:index]
resouce :share, only: [:index]
see all the examples in the guides http://guides.rubyonrails.org/routing.html
if you want to add the download and share functionality to some "resource" like, say a picture, then you would do something like:
resources :pictures do
get 'download', :on => :member
get 'share', :on => :member
end
a resource always has and endpoint /pictures for example. so if you want to have paths directly to your host, then you need to provide custom matchers in your routes like you did in your examples.
I have two resource controllers where I am using a slug to represent the ID. (friendly_id gem).
I am able to have the show path for one resource on the route but not for two at the same time. ie.
root :to => 'home#index'
match '/:id' => "properties#show"
match '/:id' => "contents#show"
Basically I want urls like,
# Content
domain.com/about-us
domain.com/terms
# Property
domain.com/unique-property-name
domain.com/another-unique-property-name
Whatever resource I put on top works. Is there a way to do this?
Thanks in advace if you can help.
This is untested, but try utilizing a constraint on your route.
root :to => 'home#index'
match '/:id', :to => "properties#show",
:constraints => lambda { |r| Property.find_by_id(r.params[:id]).present? }
match '/:id', :to => "contests#show",
:constraints => lambda { |r| Contest.find_by_id(r.params[:id]).present? }
Alternatively, you can create a separate class that responds to matches? instead of defining a lambda proc. (I recommend placing these classes into separate files that will autoload within your Rails app.)
# app/constraints/property_constraint.rb
class PropertyConstraint
def self.matches?(request)
property = Property.find_by_id(request.params[:id])
property.present?
end
end
# app/constraints/contest_constraint.rb
class ContestConstraint
def self.matches?(request)
contest = Contest.find_by_id(request.params[:id])
contest.present?
end
end
# config/routes.rb
root :to => 'home#index'
match '/:id', :to => "properties#show", :constraints => PropertyConstraint
match '/:id', :to => "contests#show", :constraints => ContestConstraint
Unfortunately this results in an extra DB query (once in the routes, and once more in your controller). If anyone has a suggestion on minimizing this, please share. :)
This Rails Engine does what you want:
Slug Engine at Github
Basically, the author's approach was to mount a Rails Engine inside his main app. This Engine incorporates both a controller for handling slugs that exist and a piece of middleware for filtering out and abstaining on slugs that don't exist.
He explains why he took this approach and other aborted solutions in a rather detailed and interesting blog post. This blog post and the slug engine source code should be enough detail for you to get your own code up and running, but that open-source engine seems to be exactly what you're looking for if you want a drop-in solution.
You can do it in a middleware
Detect slug in path
If Content with this slug exists - change request path to "contents/:id"
If Property with this slug exists - change request path to "properties/:id"
in your routing set:
match 'contents/:id' => "properties#show"
match 'properties/:id' => "contents#show"
You could write another controller which takes the id from the router and checks if the id belongs to properties or content and renders the appropriate view.
match '/:id' => "router#show"
The controller would do something like this:
def show
#property = Property.find(params[:id])
if #property then
render 'property/show'
else
#content = Content.find(params[:id])
render 'content/show
end
end
Havn't tested this code, but this idea should work.
I would suggest that you do this in a more RESTful way, if possible. Basically, you have two different resources and you should separate them:
match 'properties/:id' => "properties#show"
match 'contents/:id' => "contents#show"
This will give you many advantages down the road. One immediate benefit is that you can avoid clashes between ids for properties and content. (Note that friendly_id will not help you with inter-model slug clashes in your original scheme.)
I'm upgrading my app to rails 3, and I am a bit confused about some of the routes. The resourceful ones are easy enough, but how can I set a generic rule for all actions in a specific controller. I tried something like this:
get 'custom/:action/' => {:controller => :custom}
But that didn't work. It seems the new format is "controller#action", but how can I specify the action to be variable?
Also, other than using named routes or resources, is it possible to do shorthand notation to name routes in a specific controller?
i.e. rather than:
get '/tasks', :controller => :home, :action => :tasks, :as => 'tasks_home'
get '/accounts', :controller => :home, :action => :accounts, :as => 'accounts_home'
is it possible to do something a little cleaner, like:
controller => :home do
get :tasks
get :accounts
end
And that would automatically created the named routes?
You can use action as a variable like this:
resource :custom do
match ':action'
end
This will generate
/custom/:action(.:format) customs#:action
custom POST /custom(.:format) customs#create
new_custom GET /custom/new(.:format) customs#new
edit_custom GET /custom/edit(.:format) customs#edit
GET /custom(.:format) customs#show
PUT /custom(.:format) customs#update
DELETE /custom(.:format) customs#destroy
So it will handle your action as a variable URL-s and will add some default CRUD actions as well.
Note that the controller name here is in plural. If you would like to use a route for a controller which name is in singular, use resources instead of resource.
The answer to the second question is almost identical to the first one, use resource:
resource :home do
get :tasks
get :accounts
end
generates:
tasks_home GET /home/tasks(.:format) homes#tasks
accounts_home GET /home/accounts(.:format) homes#accounts
home POST /home(.:format) homes#create
new_home GET /home/new(.:format) homes#new
edit_home GET /home/edit(.:format) homes#edit
GET /home(.:format) homes#show
PUT /home(.:format) homes#update
DELETE /home(.:format) homes#destroy
Note that the matched controller names are in plural again, because of the convention.
Looks like this is related to the persisted field being set to false on nested ActiveResource objects: https://github.com/rails/rails/pull/3107
I'm using a Rails 3 App with Kaminari for Pagination. Because of my language, SEO and friendly URLs I want to change my URLs looks like. But it seems I have to decide between friendly pagination links and or displaying the correct method names. But first things first:
I have a model, which is called "pages" and contains some pages which can't be applied to only one model or don't contain any model. The following is a snippet of my "routes.rb":
match'/neugikeiten', :to => 'pages#neuigkeiten'
scope(:path_names => { :new => "neu", :edit => "bearbeiten", :delete => "loeschen", :index => "index", :page => "seite" }) do
resources :news, :path => "neuigkeiten"
end
As you can see, I match every page from pages to a single name and match all my other models and methods (I only used one as example) with scope. All of this works just fine. But now I have to add the Kaminare routes, for which the route looks like this:
resources :pages do
get 'seite/:page', :action => :neuigkeiten, :on => :collection
end
So, basically what I want to get is an URL like this:
/neuigkeiten/seite/2
The 2 is just an example and seite is the german word for page. But what i do get, is this:
/pages/seite/2
So, I get the model name, instead of the name I defined for this single page. I already tried switching the different codes but either I just have the original model name in the URL or I don't get nice URLs for pagination, which look like this:
/neuigkeiten?page=2
And if I try to match the pages without the resource like this:
match'/tagebuch/seite/:page', :to => 'pages#tagebuch', :on => :collection
I get the following error.
can't use collection outside resources scope
It seems to me that there should be another method to do this, because I can't be the only one having this kind of problem.
I'm glad for any help!
If anybody is interested in an solution, I found one myself. It's not perfect, but it works.
I had to install an extra gem for the localisation called "i18n_routing". Then in my routes.rb I did:
localized do
resources :news, :path => "neuigkeiten" do
get 'seite/:page', :action => :index, :on => :collection
end
end
The rest is done in the localized language file (en.yml).
I currently have the following routes set up for photos:
resources :photos
match 'photos/:user' => 'photos#user', :as => :user_photo
match 'photos/:user/:key' => 'photos#show', :as => :show_photo
Those two match routes give me URLs like:
http://example.com/photos/joe_schmoe
http://example.com/photos/joe_schmoe/123xyz
...similar to the way Flickr formats its URLs.
The problem I'm having, though, is that the photos/:user route is either interpreted as the show method, or if I put those custom routes before the resources then routes like /new get interpreted as the user method.
How can I get around that without having to do a ton of custom routes?
You'll want to put this custom route above the resources :users, so that it is matched first in the routes.
match 'photos/:user' => 'photos#user', :as => :user_photo
resources :photos
Additionally you can disable the show action altogether:
resources :photos, :except => :show
It's not a good idea to mix restful routes with custom match routes on the same resource. As you observed these two routes will intercept each others actions. Your best choice is to pick only one routing system for a resource and stick with it. If you want flickr style routes you should remove the restful route and add the other necessary match routes for new/create/etc you might need. If you desperately want to keep both of these routes You either need to disable show from the rest route with the rest route being on top, or you disable new from the match route while match being on top. You can use regexp to filter out some requests from match like this:
match 'photos/:user' => 'photos#user', :as => :user_photo, :constraints => { :user => /.*[^n][^e][^w].*/ }
This gets ugly really fast tho and I suggest just not using the rest route at all.