Skip to content

Latest commit

 

History

History
647 lines (437 loc) · 22.4 KB

Guide.md

File metadata and controls

647 lines (437 loc) · 22.4 KB

FriendlyId Guide

  • Table of Contents {:toc}

Overview

FriendlyId is an ORM-centric Ruby library that lets you work with human-friendly strings as if they were numeric ids. Among other things, this facilitates replacing "unfriendly" URL's like:

http://example.com/states/4323454

with "friendly" ones such as:

http://example.com/states/washington

FriendlyId is typically used with Rails and Active Record, but can also be used in non-Rails applications, and with Sequel and DataMapper.

Simple Models

The simplest way to use FriendlyId is with a model that has a uniquely indexed column with no spaces or special characters, and that is seldom or never updated. The most common example of this is a user name or login column:

class User < ActiveRecord::Base
  validates_format_of :login, :with => /\A[a-z0-9]+\z/i
  has_friendly_id :login
end

@user = User.find "joe"   # the old User.find(1) still works, too
@user.to_param            # returns "joe"
redirect_to @user         # the URL will be /users/joe

In this case, FriendlyId assumes you want to use the column as-is; it will never modify the value of the column, and your application should ensure that the value is admissible in a URL:

class City < ActiveRecord::Base
  has_friendly_id :name
end

@city.find "Viña del Mar"
redirect_to @city # the URL will be /cities/Viña%20del%20Mar

For this reason, it is often more convenient to use Slugs rather than a single column.

Slugged Models

FriendlyId uses a separate table to store slugs for models which require some processing of the friendly_id text. The most common example is a blog post's title, which may have spaces, uppercase characters, or other attributes you wish to modify to make them more suitable for use in URL's.

class Post < ActiveRecord::Base
  has_friendly_id :title, :use_slug => true
end

@post = Post.create(:title => "This is the first post!")
@post.friendly_id   # returns "this-is-the-first-post"
redirect_to @post   # the URL will be /posts/this-is-the-first-post

If you are unsure whether to use slugs, then your best bet is to use them, because FriendlyId provides many useful features that only work with this feature. These features are explained in detail {file:Guide.md#features below}.

Installation

FriendlyId can be installed as a gem, or as a Rails plugin. It is compatible with Rails 2.3.x. and 3.0.

As a Gem

gem install friendly_id

Rails 2.3.x

After installing the gem, add an entry in environment.rb:

config.gem "friendly_id", :version => "~> 3.2.1"

Rails 3.0

After installing the gem, add an entry in the Gemfile:

gem "friendly_id", "~> 3.2.1"

As a Plugin

Plugin installation is simple for all supported versions of Rails:

./script/plugin install git://github.com/norman/friendly_id.git

However, installing as a gem offers simpler version control than plugin installation. Whenever possible, install as a gem instead. Plugin support may eventually be removed in a future version.

Future Compatibility

FriendlyId will always remain compatible with the current release of Rails, and at least one stable release behind. That means that support for 2.3.x will not be dropped until a stable release of 3.1 is out, or possibly longer.

Setup

After installing either as a gem or plugin, run:

rails generate friendly_id
# or "./script generate friendly_id" on Rails 2.3
rake db:migrate

This will install the Rake tasks and slug migration for FriendlyId. If you are not going to use slugs, you can use the skip-migration option:

rails generate friendly_id --skip-migration

FriendlyId is now set up and ready for you to use.

Configuration

FriendlyId is configured in your model using the has_friendly_id method:

has_friendly_id :a_column_or_method options_hash

class Post < ActiveRecord::Base
  # use the "title" column as the basis of the friendly_id, and use slugs
  has_friendly_id :title, :use_slug => true,
    # remove accents and other diacritics from Latin characters
    :approximate_ascii => true,
    # don't use slugs larger than 50 bytes
    :max_length => 50
end

Read on to learn about the various features that can be configured. For the full list of valid configuration options, see the instance attribute summary for {FriendlyId::Configuration}.

Features

FriendlyId Strings

FriendlyId uses the Babosa library for generating slug strings. When using slugs, FriendlyId/Babosa will automatically modify the slug text to make it more suitable for use in a URL:

class City < ActiveRecord::Base
  has_friendly_id :name, :use_slug => true
end

@city.create :name => "Viña del Mar"
@city.friendly_id  # will be "viña-del-mar"

By default, the string is downcased and stripped, spaces are replaced with dashes, and non-word characters other than "-" are removed.

Replacing Accented Characters

If your strings use Latin characters, you can use the :approximate_ascii option to remove accents and other diacritics:

class City < ActiveRecord::Base
  has_friendly_id :name, :use_slug => true, :approximate_ascii => true
end

@city.create :name => "Łódź, Poland"
@city.friendly_id  # will be "lodz-poland"

There are special options for some languages:

class Person < ActiveRecord::Base
  has_friendly_id :name, :use_slug => true, :approximate_ascii => true,
    :ascii_approximation_options => :german
end

@person.create :name => "Jürgen Müller"
@person.friendly_id  # will be "juergen-mueller"

FriendlyId supports whatever languages are supported by Babosa; at the time of writing, this includes Danish, German, Serbian and Spanish.

Unicode Slugs

By default, any character outside the Unicode Latin character range will be passed through untouched, allowing you to have slugs in Arabic, Japanese, Greek, etc:

@post.create :title => "katakana: ゲコゴサザシジ!"
@post.friendly_id # will be: "katakana-ゲコゴサザシジ"

ASCII Slugs

You can also configure FriendlyId using :strip_non_ascii to simply delete any non-ascii characters:

class Post < ActiveRecord::Base
  has_friendly_id :title, :use_slug => true, :strip_non_ascii => true
end

@post.create :title => "katakana: ゲコゴサザシジ!"
@post.friendly_id # will be: "katakana"

Using a Custom Method to Generate the Slug Text

FriendlyId can use either a column or a method to generate the slug text for your model:

class City < ActiveRecord::Base

  belongs_to :country
  has_friendly_id :name_and_country, :use_slug => true

  def name_and_country
    #{name} #{country.name}
  end

end

@country = Country.create(:name => "Argentina")
@city = City.create(:name => "Buenos Aires", :country => @country)
@city.friendly_id # will be "buenos-aires-argentina"

One word of caution: in the example above, if the country's name were updated, say, to "Argentine Republic", the city's friendly_id would not be automatically updated. For this reason, it's a good idea to avoid using frequently-updated relations as a part of the friendly_id.

Using a Custom Method to Process the Slug Text

If the built-in slug text handling options don't work for your application, you can override the normalize_friendly_id method in your model class in order to fine-tune the output:

class City < ActiveRecord::Base

  def normalize_friendly_id(text)
    my_text_modifier_method(text)
  end

end

The normalize_friendly_id method takes a single argument and receives an instance of {FriendlyId::SlugString}, a class which wraps a regular Ruby string with additional formatting options.

Converting non-Latin characters to ASCII with Stringex

Stringex is a library which provides some interesting options for transliterating non-Latin strings to ASCII:

"你好".to_url => "ni-hao"

Using Stringex with FriendlyId is a simple matter of installing and requiring the stringex gem, and overriding the normalize_friendly_id method in your model:

class City < ActiveRecord::Base
  def normalize_friendly_id(text)
    text.to_url
  end
end

However, be aware of some limitations of Stringex - it just does a context-free character-by-character approximation for Unicode strings without sensitivity to the string's language. This means, for example, that the Han characters used by Japanese, Mandarin, Cantonese, and other languages are all replaced with the same ASCII text. For Han characters, Stringex uses Mandarin, which makes its output on Japanese text useless. You can read more about the limitations of Stringex in the documentation for Unidecoder, the Perl library upon which Stringex is based.

Redirecting to the Current Friendly URL

FriendlyId maintains a history of your record's older slugs, so if your record's friendly_id changes, your URL's won't break. It offers several methods to determine whether the model instance was found using the most recent friendly_id. This helps you redirect to your "unfriendly" URL's to your new "friendly" ones when adding FriendlyId to an existing application:

class PostsController < ApplicationController

  before_filter ensure_current_post_url, :only => :show

  ...

  def ensure_current_post_url
    redirect_to @post, :status => :moved_permanently unless @post.friendly_id_status.best?
  end

end

For more information, take a look at the documentation for {FriendlyId::Status}.

Non-unique Slugs

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 preceded by "--" rather than "-" to distinguish it from the rest of the slug. This is important to enable having slugs like:

/cars/peugeot-206
/cars/peugeot-206--2

You can configure the separator string used by your model by setting the :sequence_separator option in has_friendly_id:

has_friendly_id :title, :use_slug => true, :sequence_separator => ":"

You can also override the default used in {FriendlyId::Configuration::DEFAULTS} to set the value for any model using FriendlyId. If you change this value in an existing application, be sure to {file:Guide.md#regenerating_slugs regenerate the slugs} afterwards.

For reasons I hope are obvious, you can't change this value to "-". If you try, FriendlyId will raise an error.

Reserved Words

You can configure a list of 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.

Reserved words are configured using the :reserved_words option:

class Restaurant < ActiveRecord::Base
  belongs_to :city
  has_friendly_id :name, :use_slug => true, :reserved_words => ["my", "values"]
end

The reserved words can be specified as an array or (since 3.1.7) as a regular expression.

The strings "new" and "index" are reserved by default. When you attempt to store a reserved value, FriendlyId raises a {FriendlyId::ReservedError}. You can also override the default reserved words in {FriendlyId::Configuration::DEFAULTS} to set the value for any model using FriendlyId.

Caching the FriendlyId Slug for Better Performance

Checking the slugs table all the time has an impact on performance, so as of 2.2.0, FriendlyId offers slug caching.

Automatic setup

To enable slug caching, simply add a column named "cached_slug" to your model. FriendlyId will automatically use this column if it detects it:

class AddCachedSlugToUsers < ActiveRecord::Migration
  def self.up
    add_column :users, :cached_slug, :string
    add_index  :users, :cached_slug, :unique => true
  end

  def self.down
    remove_column :users, :cached_slug
  end
end

Then, redo the slugs:

rake friendly_id:redo_slugs MODEL=User

FriendlyId will automatically query against the cache column if it's available, which will improve the performance of many queries.

A few warnings when using this feature:

  • DO NOT forget to redo the slugs, or else this feature will not work!
  • This feature uses attr_protected to protect the cached_slug column, unless you have already invoked attr_accessible. If you wish to use attr_accessible, you must invoke it BEFORE you invoke has_friendly_id in your class.
  • Cached slugs are incompatible with scopes and are ignored if your model uses the :scope option.

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

Don't use "slug" or "slugs" because FriendlyId needs those names for its own purposes.

Nil slugs and skipping validations

You can choose to allow nil friendly_ids via the :allow_nil config option:

class User < ActiveRecord::Base
  has_friendly_id :name, :allow_nil => true
end

This works whether the model uses slugs or not.

For slugged models, if the friendly_id text is nil, no slug will be created. This can be useful, for example, to only create slugs for published articles and avoid creating many slugs with sequences.

For models that don't use slugs, this will make FriendlyId skip all its validations when the friendly_id text is nil. This can be useful, for example, if you wish to add the friendly_id value in an :after_save callback.

For non-slugged models, if you simply wish to skip friendly_ids's validations for some reason, you can override the skip_friendly_id_validations method. Note that this method is not used by slugged models.

Scoped Slugs

Note that in FriendlyId prior to 3.2.0, you could specify a non-standard :scope argument on finds. This feature has been removed in 3.2.0 in favor of the query stategies described below.

FriendlyId can generate unique slugs within a given scope. For example, assume you have an application that displays restaurants. Without scoped slugs, if two restaurants are named "Joe's Diner," the second one will end up with "joes-diner--2" as its friendly_id. Using scoped allows you to keep the slug names unique for each city, so that the second "Joe's Diner" can also have the slug "joes-diner", as long as it's located in a different city:

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

City.find("seattle").restaurants.find("joes-diner")
City.find("chicago").restaurants.find("joes-diner")

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".

Complications with Scoped Slugs

Scoped Models and Cached Slugs

If you want to use cached slugs with scoped models, be sure not to create a unique index on the cached_slug column.

Finding Records by friendly_id

If you are using scopes your friendly ids may not be unique, so a simple find like

Restaurant.find("joes-diner")

may return the wrong record. In these cases when you want to use the friendly_id for queries, either query as a relation, or specify the scope in your query conditions:

# will only return restaurants named "Joe's Diner" in the given city
@city.restaurants.find("joes-diner")

# or

Restaurants.find("joes-diner", :include => :slugs, :conditions => {:slugs => {:scope => @city.to_param}})

Finding All Records That Match a Scoped ID

If you want to find all records with a particular friendly_id regardless of scope, the easiest way is to use cached slugs and query this column directly:

Restaurant.find_all_by_cached_slug("joes-diner")

If you're not using cached slugs, then this is slightly more complicated, but still doable:

name, sequence = params[:id].parse_friendly_id
Restaurant.all(:include => :slugs, :conditions => {
  :slugs => {:name => name, :sequence => sequence}
})

Updating a Relation's Scoped Slugs

When using a relation as the scope, updating the relation will update the slugs, but only if both models have specified the relationship. In the above example, updates to City will update the slugs for Restaurant because City specifies that it has_many :restaurants.

Routes for Scoped Models

Note that FriendlyId does not set up any routes for scoped models; you must do this yourself in your application. Here's an example of one way to set this up:

# in routes.rb
resources :cities do
  resources :restaurants
end

# in views
<%= link_to 'Show', [@city, @restaurant] %>

# in controllers
@city = City.find(params[:city_id])
@restaurant = @city.restaurants.find(params[:id])

# URL's:
http://example.org/cities/seattle/restaurants/joes-diner
http://example.org/cities/chicago/restaurants/joes-diner

FriendlyId Rake Tasks

FriendlyId provides several tasks to help maintain your application.

Generating New Slugs For the First Time

friendly_id:make_slugs MODEL=<model name>

Use this task to generate slugs after installing FriendlyId in a new application.

Regenerating Slugs

friendly_id:redo_slugs MODEL=<model name>

Use this task to regenerate slugs after making any changes to your model's FriendlyId configuration options that affect slug generation. For example, if you introduce a cached_slug column or change the :seqence_separator.

Deleting Old Slugs

rake friendly_id:remove_old_slugs MODEL=<model name> DAYS=<days>

Use this task if you wish to delete expired slugs; manually or perhaps via cron. If you don't specify the days option, the default is to remove unused slugs older than 45 days.

Misc tips

Allowing Users to Override/Control Slugs

Would you like to mostly use default slugs, but allow the option of a custom user-chosen slug in your application? If so, then you're not the first to want this. Here's a demo application showing how it can be done.

Default Scopes

Whether you're using FriendlyId or not, a good rule of thumb for default scopes is to always use your model's table name. Otherwise any time you do a join, you risk having queries fail because of duplicate column names - particularly for a default scope like this one:

default_scope :order => "created_at DESC"

Instead, do this:

default_scope :order => = "#{quoted_table_name}.created_at DESC"

Or even better, unless you're using a custom primary key:

default_scope :order => = "#{quoted_table_name}.id DESC"

because sorting by a unique integer column is faster than sorting by a date column.

MySQL MyISAM tables

Currently, the default FriendlyId migration will not work with MyISAM tables because it creates an index that's too large. The easiest way to work around this is to change the generated migration to add limits on some column lengths. Please see this issue in the FriendlyId issue tracker for more information.

Hacking FriendlyId

A couple of notes for programmers intending to work on FriendlyId:

If you intend to send a pull request, in general it's best to make minor changes in the master branch, and major changes in the edge branch.

Before removing any public or protected methods, FriendlyId will deprecate them through one major release cycle. Private methods may, however, change at any time.

Some Benchmarks

These benchmarks can give you an idea of FriendlyId's impact on the performance of your application. Of course your results may vary.

Note that much of the performance difference can be attributed to finding an SQL record by a text column. Finding a single record by numeric primary key is always the fastest operation, and thus the best choice when possible. If you decide not to use FriendlyId for performance reasons, keep in mind that your own solution is unlikely to be any faster than FriendlyId with cached slugs enabled. But if it is, then your patches would be very welcome!

activerecord (2.3.8)
ruby 1.9.2p0 (2010-08-18 revision 29036) [x86_64-darwin10.4.0]
friendly_id (3.1.4)
sqlite3-ruby (1.3.1)
sqlite3 3.6.12 in-memory database

                                                   | DEFAULT | NO_SLUG |    SLUG | CACHED_SLUG |
------------------------------------------------------------------------------------------------
find model by id                             x1000 |   0.370 |   0.503 |   0.940 |       0.562 |
find model using array of ids                x1000 |   0.612 |   0.615 |   1.054 |       0.957 |
find model using id, then to_param           x1000 |   0.374 |   0.535 |   1.396 |       0.567 |
================================================================================================
Total                                              |   1.356 |   1.653 |   3.390 |       2.086 |


activerecord (3.0.0)
ruby 1.9.2p0 (2010-08-18 revision 29036) [x86_64-darwin10.4.0]
friendly_id (3.1.4)
sqlite3-ruby (1.3.1)
sqlite3 3.6.12 in-memory database

                                                   | DEFAULT | NO_SLUG |    SLUG | CACHED_SLUG |
------------------------------------------------------------------------------------------------
find model by id                             x1000 |   0.286 |   0.365 |   0.518 |       0.393 |
find model using array of ids                x1000 |   0.329 |   0.441 |   0.709 |       0.475 |
find model using id, then to_param           x1000 |   0.321 |   0.332 |   0.976 |       0.399 |
================================================================================================
Total                                              |   0.936 |   1.138 |   2.203 |       1.266 |