FriendlyId is the “Swiss Army bulldozer” of slugging and permalink plugins for Ruby on Rails. It allows you to create pretty URL's and work with human-friendly strings as if they were numeric ids for ActiveRecord models.
Using FriendlyId, it's easy to make your application use URL's like:
Text-based id's look better
They make URL's easier to remember.
They give no hint about the number of records in your database.
They are better for search engine optimization.
They can change, breaking your URL's and your SEO.
It can be tricky to ensure they're always unique.
They can become a pain to manage in large Rails applications.
They can conflict with your application's namespace.
FriendlyId tries to offer you the all the advantages, and avoid or soften the potential impact of the disadvantages.
User names (“non-slugged” models)
Usually users have unique user names stored in a column with a unique constraint or index. In this case, all you need to do is add this to your model:
and you can then write code like this:
@member = Member.find("joe") # the old Member.find(1) still works, too. @member.to_param # returns "joe" redirect_to @member # The URL would be /members/joe
Blog posts (“slugged” models)
Blog posts generally have titles which are distinctive but not necessarily unique. In this and similar cases, FriendlyId provides a Slug model separate from your Post model. The Slug model handles duplicate friendly_ids, as well as versioning.
Your model code would look something like this:
has_friendly_id :title, :use_slug => true
and you can then write code like this:
@post = Post.find("new-version-released") # Post.find(1) still works, too @post.to_param # returns "new-version-released" redirect_to @post # The URL would be /posts/new-version-released
Now in your controllers, if you want to prevent people from accessing your models by numeric id, you can detect whether they were found by the friendly_id:
@post = Post.find(params[:id]) raise "some error" if !@post.found_using_friendly_id?
or, you can 301 redirect if the model was found by the numeric id if you don't care about numeric access, but want the SEO value of the friendly_id:
@post = Post.find(params[:id]) redirect_to @post, :status => 301 if @post.has_better_id?
The “has_better_id?” method returns true if the model was found with the numeric id, or with an outdated slug.
FriendlyId will record changes to slugs so that you can tell when the model is found with an older slug, or by the numeric id. This can be useful if you want to do a 301 redirect to your updated URL.
class PostsController < ApplicationController before_filter ensure_current_post_url, :only => :show ... def ensure_current_post_url redirect_to @post, :status => :moved_permanently if @post.has_better_id? end end
This is particularly useful when implementing FrindlyId on an existing website that already has many URL's with the old numeric id listed on search engines. When the search engine spiders crawl your site, they will eventually pick up the new, more SEO-friendly URL's.
Non-unique Slug Names
FriendlyId will append a arbitrary number to the end of the id to keep it unique if necessary:
/posts/new-version-released /posts/new-version-released--2 /posts/new-version-released--3 ... etc.
Note that the number is preceeded by two dashes to distinguish it from the rest of the slug. This is important to enable having slugs like:
You can mark off some strings as reserved so that, for example, you don't end up with this problem:
/users/joe-schmoe # A user chose "joe schmoe" as his user name - no worries. /users/new # A user chose "new" as his user name, and now no one can sign up.
Here's how to do it:
class Restaurant < ActiveRecord::Base belongs_to :city has_friendly_id :name, :use_slug => true, :reserved => ["my", "values"] end
As of FriendlyId version 2.0.2, “new” and “index” are reseved by default. When you attempt to store a reserved value, FriendlyId raises a FriendlyId::SlugGenerationError.
Checking the slugs table all the time has an impact on performance, so as of 2.3.0, FriendlyId offers slug caching.
This feature can improve the performance of some views by about 25%, and reduce memory consumption by about 40% as compared to the same view without cached slugs. The biggest improvement will be for “index” type views with many links that depend on slugs to generate the URL.
To enable slug caching, simply add a column named “cached_slug” to your model. FriendlyId will automagically use this column if it detects it:
class AddCachedSlugToUsers < ActiveRecord::Migration def self.up add_column :users, :cached_slug, :string end def self.down remove_column :users, :cached_slug end end
Then, redo the slugs:
rake friendly_id:redo_slugs MODEL=User
This feature exists largely to improve the performance of URL generation, the part of Rails where FriendlyId has the biggest performance impact. FriendlyId never queries against this column, so it's not necessary to add an index on it unless your application does.
Two warnings when using this feature:
*DO NOT* forget to redo the slugs, or else this feature will not work!
Also, this feature uses attr_protected to protect the cached_slug column, unless you have already invoked attr_accessible. So if you wish to use attr_accessible, you must invoke it BEFORE you invoke has_friendly_id in your class.
Using a custom column name:
You can also use a different name for the column if you choose, via the :cache_column config option:
class User < ActiveRecord::Base has_friendly_id :name, :use_slug => true, :cache_column => 'my_cached_slug' end
FriendlyId can generate unique slugs within a given scope. For example:
class Restaurant < ActiveRecord::Base belongs_to :city has_friendly_id :name, :use_slug => true, :scope => :city end class City < ActiveRecord::Base has_many :restaurants has_friendly_id :name, :use_slug => true end http://example.org/cities/seattle/restaurants/joes-diner http://example.org/cities/chicago/restaurants/joes-diner Restaurant.find("joes-diner", :scope => "seattle") # returns 1 record Restaurant.find("joes-diner", :scope => "chicago") # returns 1 record Restaurant.find("joes-diner") # returns both records
The value for the :scope key in your model can be a custom method you define, or the name of a relation. If it's the name of a relation, then the scope's text value will be the result of calling to_param on the related model record. In the example above, the city model also uses FriendlyId and so its to_param method returns its friendly_id: chicago or seattle.
This feature is new in FriendlyId 2 and should be considered of experimental quality. Please don't use this for code that needs to run on the Space Shuttle.
FriendlyId's slugging can strip diacritics from Western European characters, so that you can have ASCII-only URL's; for example, conveting “ñøîéçü” to “noiecu.”
has_friendly_id :title, :use_slug => true, :strip_diacritics => true
If you are not using slugs, you'll have to do this manually for whatever value you're using as the friendly_id.
FriendlyId can also normalize slug text while preserving accented characters, if you prefer to leave them in your URL's:
has_friendly_id :title, :use_slug => true ... @post = Post.create(:title => "¡Feliz Año!") @post.friendly_id # "feliz-año"
FriendlyId can generate slugs in any language that can be written with Unicode. It does its best to strip away punctuation regardless of the language being used. Since the authors only speak English, Spanish, Portuguese and German, this has not been extensively tested with anything like Chinese, Russian, Greek, etc, but it “should work.” If you're a speaker of a language that uses a non-Roman writing system, your feedback would be most welcome.
has_friendly_id :title, :use_slug => true ... @post = Post.create(:title => "友好编号在中国") @post.friendly_id # "友好编号在中国" @post2 = Post.create(:title => "友好编号在中国") @post2.friendly_id # "友好编号在中国--2"
Custom Slug Generation
While FriendlyId's slug generation options work for most people, you may need something else. As of version 2.0.4 you can pass in your own custom slug generation block:
class Post < ActiveRecord::Base has_friendly_id :title, :use_slug => true do |text| MySlugGeneratorClass::my_slug_method(text) end end
FriendlyId will still respect your settings for max length and reserved words, but will use your block rather than the baked-in methods to normalize the friendly_id text.
FriendlyId is best installed as a Ruby Gem:
gem install friendly_id
Alternatively, you can install it as a Rails plugin, though this is discouraged:
./script/plugin install git://github.com/norman/friendly_id.git
Setting it up
The current release works with Rails 2.2 and above, and is compatible with Ruby 1.8 and 1.9. If you need support for Rails 2.0 - 2.1, you need to install an older version of FriendlyId. Here's how to set it up.
1) Install the Gem:
sudo gem install friendly_id cd my_app script/generate friendly_id rake db:migrate
2) Add the gem to config/environment.rb:
3) Add some code to your models:
class Post < ActiveRecord::Base has_friendly_id :title, :use_slug => true end
4) If you are using slugs, you can use a Rake task to generate slugs for your existing records:
rake friendly_id:make_slugs MODEL=MyModelName
If you eventually want to expire old slugs every so often, or perhaps every day via cron, you can do:
The default is to remove dead slugs older than 45 days, but is configurable:
rake friendly_id:remove_old_slugs MODEL=MyModelName DAYS=60
Installing an older version
You can download older versions of FriendlyId from Github that work with Rails < 2.2. These versions are, however, no longer maintained or supported.
FriendlyId is hosted on Github, and we love pull requests. :-)
Please report them on the Github issue tracker for this project.
If you have a bug to report, please include the following information:
Stack trace and error message.
Version information for FriendlyId, Rails and Ruby.
Any snippets of relevant model, view or controller code that shows how your are using FriendlyId.
If you are able to, it helps even more if you can fork FriendlyId on Github, and add a test that reproduces the error you are experiencing.
We are grateful for many contributions from the Ruby and Rails community, in particular from the following people:
Andrew Loe III
Joe Van Dyk
Copyright © 2008 Norman Clarke, Adrian Mugnolo and Emilio Tagua, released under the MIT license.