Guides.rubyonrails.org

Chapters

1. How to Load Core Extensions
   • Stand-Alone Active Support
   • Active Support Within a Ruby on Rails Application
2. Extensions to All Objects
   • blank? and present?
   • presence
   • duplicable?
   • try
   • singleton_class
   • class_eval(*args, &block)
   • acts_like?(duck)
   • to_param
   • to_query
   • with_options
   • Instance Variables
   • Silencing Warnings, Streams, and Exceptions
   • require_library_or_gem
3. Extensions to Module
   • alias_method_chain
   • Attributes
   • Parents
   • Constants
   • Synchronization
   • Reachable
   • Anonymous
   • Method Delegation
   • Method Names
   • Redefining Methods
4. Extensions to Class
   • Class Attributes
   • Class Inheritable Attributes
   • Subclasses & Descendants
5. Extensions to String
   • Output Safety
   • squish
   • truncate
   • Key-based Interpolation
   • starts_with? and ends_width?
   • strip_heredoc
   • Access
   • Inflections
   • Conversions
6. Extensions to Numeric
   • Bytes
7. Extensions to Integer
   • multiple_of?
   • ordinalize
8. Extensions to Float
   • round
9. Extensions to BigDecimal
10. Extensions to Enumerable
    • group_by
    • sum
    • each_with_object
    • index_by
    • many?
    • exclude?
11. Extensions to Array
    • Accessing
    • Random Access
    • Options Extraction
    • Conversions
    • Wrapping
    • Grouping
12. Extensions to Hash
    • Conversions
    • Merging
    • Diffing
    • Working with Keys
    • Slicing
    • Indifferent Access
13. Extensions to Regexp
    • multiline?
14. Extensions to Range
    • to_s
    • step
    • include?
    • overlaps?
15. Extensions to Proc
    • bind
16. Extensions to Date
    • Calculations
    • Conversions
17. Extensions to DateTime
    • Calculations
18. Extensions to Time
    • Calculations
    • Time Constructors
19. Extensions to Process
    • daemon
20. Extensions to File
    • atomic_write
21. Extensions to NameError
22. Extensions to LoadError
23. Changelog

1 How to Load Core Extensions

1.1 Stand-Alone Active Support

In order to have a near zero default footprint, Active Support does not load anything by default. It is broken in small pieces so that you may load just what you need, and also has some convenience entry points to load related extensions in one shot, even everything.

Thus, after a simple require like:

require 'active_support'

objects do not even respond to blank?, let’s see how to load its definition.

1.1.1 Cherry-picking a Definition

The most lightweight way to get blank? is to cherry-pick the file that defines it.

For every single method defined as a core extension this guide has a note that says where such a method is defined. In the case of blank? the note reads:

That means that this single call is enough:

require 'active_support/core_ext/object/blank'

Active Support has been carefully revised so that cherry-picking a file loads only strictly needed dependencies, if any.

1.1.2 Loading Grouped Core Extensions

The next level is to simply load all extensions to Object. As a rule of thumb, extensions to SomeClass are available in one shot by loading active_support/core_ext/some_class.

Thus, if that would do, to have blank? available we could just load all extensions to Object:

require 'active_support/core_ext/object'

1.1.3 Loading All Core Extensions

You may prefer just to load all core extensions, there is a file for that:

require 'active_support/core_ext'

1.1.4 Loading All Active Support

And finally, if you want to have all Active Support available just issue:

require 'active_support/all'

That does not even put the entire Active Support in memory upfront indeed, some stuff is configured via autoload, so it is only loaded if used.

1.2 Active Support Within a Ruby on Rails Application

A Ruby on Rails application loads all Active Support unless config.active_support.bare is true. In that case, the application will only load what the framework itself cherry-picks for its own needs, and can still cherry-pick itself at any granularity level, as explained in the previous section.

2 Extensions to All Objects

2.1 blank? and present?

The following values are considered to be blank in a Rails application:

• nil and false,

• strings composed only of whitespace, i.e. matching /\A\s*\z/,

• empty arrays and hashes, and

• any other object that responds to empty? and it is empty.

For example, this method from ActionDispatch::Session::AbstractStore uses blank? for checking whether a session key is present:

def ensure_session_key!
  if @key.blank?
    raise ArgumentError, 'A key is required...'
  end
end

The method present? is equivalent to !blank?. This example is taken from ActionDispatch::Http::Cache::Response:

def set_conditional_cache_control!
  return if self["Cache-Control"].present?
  ...
end

2.2 presence

The presence method returns its receiver if present?, and nil otherwise. It is useful for idioms like this:

host = config[:host].presence || 'localhost'

2.3 duplicable?

A few fundamental objects in Ruby are singletons. For example, in the whole life of a program the integer 1 refers always to the same instance:

1.object_id                 # => 3
Math.cos(0).to_i.object_id  # => 3

Hence, there’s no way these objects can be duplicated through dup or clone:

true.dup  # => TypeError: can't dup TrueClass

Some numbers which are not singletons are not duplicable either:

0.0.clone        # => allocator undefined for Float
(2**1024).clone  # => allocator undefined for Bignum

Active Support provides duplicable? to programmatically query an object about this property:

"".duplicable?     # => true
false.duplicable?  # => false

By definition all objects are duplicable? except nil, false, true, symbols, numbers, and class and module objects.

2.4 try

Sometimes you want to call a method provided the receiver object is not nil, which is something you usually check first. try is like Object#send except that it returns nil if sent to nil.

For instance, in this code from ActiveRecord::ConnectionAdapters::AbstractAdapter @logger could be nil, but you save the check and write in an optimistic style:

def log_info(sql, name, ms)
  if @logger.try(:debug?)
    name = '%s (%.1fms)' % [name || 'SQL', ms]
    @logger.debug(format_log_entry(name, sql.squeeze(' ')))
  end
end

2.5 singleton_class

The method singleton_class returns the singleton class of the receiver:

String.singleton_class     # => #<Class:String>
String.new.singleton_class # => #<Class:#<String:0x17a1d1c>>

2.6 class_eval(*args, &block)

You can evaluate code in the context of any object’s singleton class using class_eval:

class Proc
  def bind(object)
    block, time = self, Time.now
    object.class_eval do
      method_name = "__bind_#{time.to_i}_#{time.usec}"
      define_method(method_name, &block)
      method = instance_method(method_name)
      remove_method(method_name)
      method
    end.bind(object)
  end
end

2.7 acts_like?(duck)

The method acts_like provides a way to check whether some class acts like some other class based on a simple convention: a class that provides the same interface as String defines

def acts_like_string?
end

which is only a marker, its body or return value are irrelevant. Then, client code can query for duck-type-safeness this way:

some_klass.acts_like?(:string)

Rails has classes that act like Date or Time and follow this contract.

2.8 to_param

All objects in Rails respond to the method to_param, which is meant to return something that represents them as values in a query string, or as URL fragments.

By default to_param just calls to_s:

7.to_param # => "7"

The return value of to_param should not be escaped:

"Tom & Jerry".to_param # => "Tom & Jerry"

Several classes in Rails overwrite this method.

For example nil, true, and false return themselves. Array#to_param calls to_param on the elements and joins the result with “/”:

[0, true, String].to_param # => "0/true/String"

Notably, the Rails routing system calls to_param on models to get a value for the :id placeholder. ActiveRecord::Base#to_param returns the id of a model, but you can redefine that method in your models. For example, given

class User
  def to_param
    "#{id}-#{name.parameterize}"
  end
end

we get:

user_path(@user) # => "/users/357-john-smith"

2.9 to_query

Except for hashes, given an unescaped key this method constructs the part of a query string that would map such key to what to_param returns. For example, given

class User
  def to_param
    "#{id}-#{name.parameterize}"
  end
end

we get:

current_user.to_query('user') # => user=357-john-smith

This method escapes whatever is needed, both for the key and the value:

account.to_query('company[name]')
# => "company%5Bname%5D=Johnson+%26+Johnson"

so its output is ready to be used in a query string.

Arrays return the result of applying to_query to each element with key[] as key, and join the result with “&”:

[3.4, -45.6].to_query('sample')
# => "sample%5B%5D=3.4&sample%5B%5D=-45.6"

Hashes also respond to to_query but with a different signature. If no argument is passed a call generates a sorted series of key/value assignments calling to_query(key) on its values. Then it joins the result with “&”:

{:c => 3, :b => 2, :a => 1}.to_query # => "a=1&b=2&c=3"

The method Hash#to_query accepts an optional namespace for the keys:

{:id => 89, :name => "John Smith"}.to_query('user')
# => "user%5Bid%5D=89&user%5Bname%5D=John+Smith"

2.10 with_options

The method with_options provides a way to factor out common options in a series of method calls.

Given a default options hash, with_options yields a proxy object to a block. Within the block, methods called on the proxy are forwarded to the receiver with their options merged. For example, you get rid of the duplication in:

class Account < ActiveRecord::Base
  has_many :customers, :dependent => :destroy
  has_many :products,  :dependent => :destroy
  has_many :invoices,  :dependent => :destroy
  has_many :expenses,  :dependent => :destroy
end

this way:

class Account < ActiveRecord::Base
  with_options :dependent => :destroy do |assoc|
    assoc.has_many :customers
    assoc.has_many :products
    assoc.has_many :invoices
    assoc.has_many :expenses
  end
end

That idiom may convey grouping to the reader as well. For example, say you want to send a newsletter whose language depends on the user. Somewhere in the mailer you could group locale-dependent bits like this:

I18n.with_options :locale => user.locale, :scope => "newsletter" do |i18n|
  subject i18n.t :subject
  body    i18n.t :body, :user_name => user.name
end

2.11 Instance Variables

Active Support provides several methods to ease access to instance variables.

2.11.1 instance_variable_names

Ruby 1.8 and 1.9 have a method called instance_variables that returns the names of the defined instance variables. But they behave differently, in 1.8 it returns strings whereas in 1.9 it returns symbols. Active Support defines instance_variable_names as a portable way to obtain them as strings:

class C
  def initialize(x, y)
    @x, @y = x, y
  end
end

C.new(0, 1).instance_variable_names # => ["@y", "@x"]

2.11.2 instance_values

The method instance_values returns a hash that maps instance variable names without “@” to their corresponding values. Keys are strings both in Ruby 1.8 and 1.9:

class C
  def initialize(x, y)
    @x, @y = x, y
  end
end

C.new(0, 1).instance_values # => {"x" => 0, "y" => 1}

2.11.3 copy_instance_variables_from(object, exclude = [])

Copies the instance variables of object into self.

Instance variable names in the exclude array are ignored. If object responds to protected_instance_variables the ones returned are also ignored. For example, Rails controllers implement that method.

In both arrays strings and symbols are understood, and they have to include the at sign.

class C
  def initialize(x, y, z)
    @x, @y, @z = x, y, z
  end

  def protected_instance_variables
    %w(@z)
  end
end

a = C.new(0, 1, 2)
b = C.new(3, 4, 5)

a.copy_instance_variables_from(b, [:@y])
# a is now: @x = 3, @y = 1, @z = 2

In the example object and self are of the same type, but they don’t need to.

2.12 Silencing Warnings, Streams, and Exceptions

The methods silence_warnings and enable_warnings change the value of $VERBOSE accordingly for the duration of their block, and reset it afterwards:

silence_warnings { Object.const_set "RAILS_DEFAULT_LOGGER", logger }

You can silence any stream while a block runs with silence_stream:

silence_stream(STDOUT) do
  # STDOUT is silent here
end

Silencing exceptions is also possible with suppress. This method receives an arbitrary number of exception classes. If an exception is raised during the execution of the block and is kind_of? any of the arguments, suppress captures it and returns silently. Otherwise the exception is reraised:

# If the user is locked the increment is lost, no big deal.
suppress(ActiveRecord::StaleObjectError) do
  current_user.increment! :visits
end

2.13 require_library_or_gem

The convenience method require_library_or_gem tries to load its argument with a regular require first. If it fails loads rubygems and tries again.

If the first attempt is a failure and rubygems can’t be loaded the method raises LoadError. On the other hand, if rubygems is available but the argument is not loadable as a gem, the method gives up and LoadError is also raised.

For example, that’s the way the MySQL adapter loads the MySQL library:

require_library_or_gem('mysql')

3 Extensions to Module

3.1 alias_method_chain

Using plain Ruby you can wrap methods with other methods, that’s called alias chaining.

For example, let’s say you’d like params to be strings in functional tests, as they are in real requests, but still want the convenience of assigning integers and other kind of values. To accomplish that you could wrap ActionController::TestCase#process this way in test/test_helper.rb:

ActionController::TestCase.class_eval do
  # save a reference to the original process method
  alias_method :original_process, :process

  # now redefine process and delegate to original_process
  def process(action, params=nil, session=nil, flash=nil, http_method='GET')
    params = Hash[*params.map {|k, v| [k, v.to_s]}.flatten]
    original_process(action, params, session, flash, http_method)
  end
end

That’s the method get, post, etc., delegate the work to.

That technique has a risk, it could be the case that :original_process was taken. To try to avoid collisions people choose some label that characterizes what the chaining is about:

ActionController::TestCase.class_eval do
  def process_with_stringified_params(...)
    params = Hash[*params.map {|k, v| [k, v.to_s]}.flatten]
    process_without_stringified_params(action, params, session, flash, http_method)
  end
  alias_method :process_without_stringified_params, :process
  alias_method :process, :process_with_stringified_params
end

The method alias_method_chain provides a shortcut for that pattern:

ActionController::TestCase.class_eval do
  def process_with_stringified_params(...)
    params = Hash[*params.map {|k, v| [k, v.to_s]}.flatten]
    process_without_stringified_params(action, params, session, flash, http_method)
  end
  alias_method_chain :process, :stringified_params
end

Rails uses alias_method_chain all over the code base. For example validations are added to ActiveRecord::Base#save by wrapping the method that way in a separate module specialised in validations.

3.2 Attributes

3.2.1 alias_attribute

Model attributes have a reader, a writer, and a predicate. You can aliase a model attribute having the corresponding three methods defined for you in one shot. As in other aliasing methods, the new name is the first argument, and the old name is the second (my mnemonic is they go in the same order as if you did an assignment):

class User < ActiveRecord::Base
  # let me refer to the email column as "login",
  # much meaningful for authentication code
  alias_attribute :login, :email
end

3.2.2 attr_accessor_with_default

The method attr_accessor_with_default serves the same purpose as the Ruby macro attr_accessor but allows you to set a default value for the attribute:

class Url
  attr_accessor_with_default :port, 80
end

Url.new.port # => 80

The default value can be also specified with a block, which is called in the context of the corresponding object:

class User
  attr_accessor :name, :surname
  attr_accessor_with_default(:full_name) {
	[name, surname].compact.join(" ")
  }
end

u = User.new
u.name = 'Xavier'
u.surname = 'Noria'
u.full_name # => "Xavier Noria"

The result is not cached, the block is invoked in each call to the reader.

You can overwrite the default with the writer:

url = Url.new
url.host # => 80
url.host = 8080
url.host # => 8080

The default value is returned as long as the attribute is unset. The reader does not rely on the value of the attribute to know whether it has to return the default. It rather monitors the writer: if there’s any assignment the value is no longer considered to be unset.

Active Resource uses this macro to set a default value for the :primary_key attribute:

attr_accessor_with_default :primary_key, 'id'

3.2.3 Internal Attributes

When you are defining an attribute in a class that is meant to be subclassed name collisions are a risk. That’s remarkably important for libraries.

Active Support defines the macros attr_internal_reader, attr_internal_writer, and attr_internal_accessor. They behave like their Ruby builtin attr_* counterparts, except they name the underlying instance variable in a way that makes collisions less likely.

The macro attr_internal is a synonym for attr_internal_accessor:

# library
class ThirdPartyLibrary::Crawler
  attr_internal :log_level
end

# client code
class MyCrawler < ThirdPartyLibrary::Crawler
  attr_accessor :log_level
end

In the previous example it could be the case that :log_level does not belong to the public interface of the library and it is only used for development. The client code, unaware of the potential conflict, subclasses and defines its own :log_level. Thanks to attr_internal there’s no collision.

By default the internal instance variable is named with a leading underscore, @_log_level in the example above. That’s configurable via Module.attr_internal_naming_format though, you can pass any sprintf-like format string with a leading @ and a %s somewhere, which is where the name will be placed. The default is "@_%s".

Rails uses internal attributes in a few spots, for examples for views:

module ActionView
  class Base
    attr_internal :captures
    attr_internal :request, :layout
    attr_internal :controller, :template
  end
end

3.2.4 Module Attributes

The macros mattr_reader, mattr_writer, and mattr_accessor are analogous to the cattr_* macros defined for class. Check Class Attributes.

For example, the dependencies mechanism uses them:

module ActiveSupport
  module Dependencies
    mattr_accessor :warnings_on_first_load
    mattr_accessor :history
    mattr_accessor :loaded
    mattr_accessor :mechanism
    mattr_accessor :load_paths
    mattr_accessor :load_once_paths
    mattr_accessor :autoloaded_constants
    mattr_accessor :explicitly_unloadable_constants
    mattr_accessor :logger
    mattr_accessor :log_activity
    mattr_accessor :constant_watch_stack
    mattr_accessor :constant_watch_stack_mutex
  end
end

3.3 Parents

3.3.1 parent

The parent method on a nested named module returns the module that contains its corresponding constant:

module X
  module Y
    module Z
    end
  end
end
M = X::Y::Z

X::Y::Z.parent # => X::Y
M.parent       # => X::Y

If the module is anonymous or belongs to the top-level, parent returns Object.

3.3.2 parent_name

The parent_name method on a nested named module returns the fully-qualified name of the module that contains its corresponding constant:

module X
  module Y
    module Z
    end
  end
end
M = X::Y::Z

X::Y::Z.parent_name # => "X::Y"
M.parent_name       # => "X::Y"

For top-level or anonymous modules parent_name returns nil.

3.3.3 parents

The method parents calls parent on the receiver and upwards until Object is reached. The chain is returned in an array, from bottom to top:

module X
  module Y
    module Z
    end
  end
end
M = X::Y::Z

X::Y::Z.parents # => [X::Y, X, Object]
M.parents       # => [X::Y, X, Object]

3.4 Constants

The method local_constants returns the names of the constants that have been defined in the receiver module:

module X
  X1 = 1
  X2 = 2
  module Y
    Y1 = :y1
    X1 = :overrides_X1_above
  end
end

X.local_constants    # => ["X2", "X1", "Y"], assumes Ruby 1.8
X::Y.local_constants # => ["X1", "Y1"], assumes Ruby 1.8

The names are returned as strings in Ruby 1.8, and as symbols in Ruby 1.9. The method local_constant_names returns always strings.

3.5 Synchronization

The synchronize macro declares a method to be synchronized:

class Counter
  @@mutex = Mutex.new
  attr_reader :value

  def initialize
    @value = 0
  end

  def incr
    @value += 1 # non-atomic
  end
  synchronize :incr, :with => '@@mutex'
end

The method receives the name of an action, and a :with option with code. The code is evaluated in the context of the receiver each time the method is invoked, and it should evaluate to a Mutex instance or any other object that responds to synchronize and accepts a block.

3.6 Reachable

A named module is reachable if it is stored in its corresponding constant. It means you can reach the module object via the constant.

That is what ordinarily happens, if a module is called “M”, the M constant exists and holds it:

module M
end

M.reachable? # => true

But since constants and modules are indeed kind of decoupled, module objects can become unreachable:

module M
end

orphan = Object.send(:remove_const, :M)

# The module object is orphan now but it still has a name.
orphan.name # => "M"

# You cannot reach it via the constant M because it does not even exist.
orphan.reachable? # => false

# Let's define a module called "M" again.
module M
end

# The constant M exists now again, and it stores a module
# object called "M", but it is a new instance.
orphan.reachable? # => false

3.7 Anonymous

A module may or may not have a name:

module M
end
M.name # => "M"

N = Module.new
N.name # => "N"

Module.new.name # => "" in 1.8, nil in 1.9

You can check whether a module has a name with the predicate anonymous?:

module M
end
M.anonymous? # => false

Module.new.anonymous? # => true

Note that being unreachable does not imply being anonymous:

module M
end

m = Object.send(:remove_const, :M)

m.reachable? # => false
m.anonymous? # => false

though an anonymous module is unreachable by definition.

3.8 Method Delegation

The macro delegate offers an easy way to forward methods.

Let’s imagine that users in some application have login information in the User model but name and other data in a separate Profile model:

class User < ActiveRecord::Base
  has_one :profile
end

With that configuration you get a user’s name via his profile, user.profile.name, but it could be handy to still be able to access such attribute directly:

class User < ActiveRecord::Base
  has_one :profile

  def name
    profile.name
  end
end

That is what delegate does for you:

class User < ActiveRecord::Base
  has_one :profile

  delegate :name, :to => :profile
end

It is shorter, and the intention more obvious.

The macro accepts several methods:

delegate :name, :age, :address, :twitter, :to => :profile

When interpolated into a string, the :to option should become an expression that evaluates to the object the method is delegated to. Typically a string or symbol. Such a expression is evaluated in the context of the receiver:

# delegates to the Rails constant
delegate :logger, :to => :Rails

# delegates to the receiver's class
delegate :table_name, :to => 'self.class'

By default, if the delegation raises NoMethodError and the target is nil the exception is propagated. You can ask that nil is returned instead with the :allow_nil option:

delegate :name, :to => :profile, :allow_nil => true

With :allow_nil the call user.name returns nil if the user has no profile.

The option :prefix adds a prefix to the name of the generated method. This may be handy for example to get a better name:

delegate :street, :to => :address, :prefix => true

The previous example generates address_street rather than street.

A custom prefix may also be configured:

delegate :size, :to => :attachment, :prefix => :avatar

In the previous example the macro generates avatar_size rather than size.

3.9 Method Names

The builtin methods instance_methods and methods return method names as strings or symbols depending on the Ruby version. Active Support defines instance_method_names and method_names to be equivalent to them, respectively, but always getting strings back.

For example, ActionView::Helpers::FormBuilder knows this array difference is going to work no matter the Ruby version:

self.field_helpers = (FormHelper.instance_method_names - ['form_for'])

3.10 Redefining Methods

There are cases where you need to define a method with define_method, but don’t know whether a method with that name already exists. If it does, a warning is issued if they are enabled. No big deal, but not clean either.

The method redefine_method prevents such a potential warning, removing the existing method before if needed. Rails uses it in a few places, for instance when it generates an association’s API:

redefine_method("#{reflection.name}=") do |new_value|
  association = association_instance_get(reflection.name)

  if association.nil? || association.target != new_value
    association = association_proxy_class.new(self, reflection)
  end

  association.replace(new_value)
  association_instance_set(reflection.name, new_value.nil? ? nil : association)
end

4 Extensions to Class

4.1 Class Attributes

4.1.1 class_attribute

The method class_attribute declares one or more inheritable class attributes that can be overridden at any level down the hierarchy:

class A
  class_attribute :x
end

class B < A; end

class C < B; end

A.x = :a
B.x # => :a
C.x # => :a

B.x = :b
A.x # => :a
C.x # => :b

C.x = :c
A.x # => :a
B.x # => :b

For example ActionMailer::Base defines:

class_attribute :default_params
self.default_params = {
  :mime_version => "1.0",
  :charset      => "UTF-8",
  :content_type => "text/plain",
  :parts_order  => [ "text/plain", "text/enriched", "text/html" ]
}.freeze

They can be also accessed and overridden at the instance level:

A.x = 1

a1 = A.new
a2 = A.new
a2.x = 2

a1.x # => 1, comes from A
a2.x # => 2, overridden in a2

The generation of the writer instance method can be prevented by setting the option :instance_writer to false, as in

module AcitveRecord
  class Base
    class_attribute :table_name_prefix, :instance_writer => false
    self.table_name_prefix = ""
  end
end

A model may find that option useful as a way to prevent mass-assignment from setting the attribute.

For convenience class_attribute defines also an instance predicate which is the double negation of what the instance reader returns. In the examples above it would be called x?.

4.1.2 cattr_reader, cattr_writer, and cattr_accessor

The macros cattr_reader, cattr_writer, and cattr_accessor are analogous to their attr_* counterparts but for classes. They initialize a class variable to nil unless it already exists, and generate the corresponding class methods to access it:

class MysqlAdapter < AbstractAdapter
  # Generates class methods to access @@emulate_booleans.
  cattr_accessor :emulate_booleans
  self.emulate_booleans = true
end

Instance methods are created as well for convenience, they are just proxies to the class attribute. So, instances can change the class attribute, but cannot override it as it happens with class_attribute (see above). For example given

module ActionView
  class Base
    cattr_accessor :field_error_proc
    @@field_error_proc = Proc.new{ ... }
  end
end

we can access field_error_proc in views. The generation of the writer instance method can be prevented by setting :instance_writer to false (not any false value, but exactly false):

module ActiveRecord
  class Base
    # No pluralize_table_names= instance writer is generated.
    cattr_accessor :pluralize_table_names, :instance_writer => false
  end
end

A model may find that option useful as a way to prevent mass-assignment from setting the attribute.

4.2 Class Inheritable Attributes

Class variables are shared down the inheritance tree. Class instance variables are not shared, but they are not inherited either. The macros class_inheritable_reader, class_inheritable_writer, and class_inheritable_accessor provide accessors for class-level data which is inherited but not shared with children:

module ActionController
  class Base
    # FIXME: REVISE/SIMPLIFY THIS COMMENT.
    # The value of allow_forgery_protection is inherited,
    # but its value in a particular class does not affect
    # the value in the rest of the controllers hierarchy.
    class_inheritable_accessor :allow_forgery_protection
  end
end

They accomplish this with class instance variables and cloning on subclassing, there are no class variables involved. Cloning is performed with dup as long as the value is duplicable.

There are some variants specialised in arrays and hashes:

class_inheritable_array
class_inheritable_hash

Those writers take any inherited array or hash into account and extend them rather than overwrite them.

As with vanilla class attribute accessors these macros create convenience instance methods for reading and writing. The generation of the writer instance method can be prevented setting :instance_writer to false (not any false value, but exactly false):

module ActiveRecord
  class Base
    class_inheritable_accessor :default_scoping, :instance_writer => false
  end
end

Since values are copied when a subclass is defined, if the base class changes the attribute after that, the subclass does not see the new value. That’s the point.

4.3 Subclasses & Descendants

4.3.1 subclasses

The subclasses method returns the subclasses of the receiver:

class C; end
C.subclasses # => []

class B < C; end
C.subclasses # => [B]

class A < B; end
C.subclasses # => [B]

class D < C; end
C.subclasses # => [B, D]

The order in which these classes are returned is unspecified.

4.3.2 descendants

The descendants method returns all classes that are < than its receiver:

class C; end
C.descendants # => []

class B < C; end
C.descendants # => [B]

class A < B; end
C.descendants # => [B, A]

class D < C; end
C.descendants # => [B, A, D]

The order in which these classes are returned is unspecified.

5 Extensions to String

5.1 Output Safety

Inserting data into HTML templates needs extra care. For example you can’t just interpolate @review.title verbatim into an HTML page. On one hand if the review title is “Flanagan & Matz rules!” the output won’t be well-formed because an ampersand has to be escaped as “&amp;”. On the other hand, depending on the application that may be a big security hole because users can inject malicious HTML setting a hand-crafted review title. Check out the section about cross-site scripting in the Security guide for further information about the risks.

Active Support has the concept of (html) safe strings since Rails 3. A safe string is one that is marked as being insertable into HTML as is. It is trusted, no matter whether it has been escaped or not.

Strings are considered to be unsafe by default:

"".html_safe? # => false

You can obtain a safe string from a given one with the html_safe method:

s = "".html_safe
s.html_safe? # => true

It is important to understand that html_safe performs no escaping whatsoever, it is just an assertion:

s = "<script>...</script>".html_safe
s.html_safe? # => true
s            # => "<script>...</script>"

It is your responsibility to ensure calling html_safe on a particular string is fine.

If you append onto a safe string, either in-place with concat/<<, or with +, the result is a safe string. Unsafe arguments are escaped:

"".html_safe + "<" # => "&lt;"

Safe arguments are directly appended:

"".html_safe + "<".html_safe # => "<"

These methods should not be used in ordinary views. In Rails 3 unsafe values are automatically escaped:

<%= @review.title %> <%# fine in Rails 3, escaped if needed %>

To insert something verbatim use the raw helper rather than calling html_safe:

<%= raw @cms.current_template %> <%# inserts @cms.current_template as is %>

The raw helper calls html_safe for you:

def raw(stringish)
  stringish.to_s.html_safe
end

5.2 squish

The method String#squish strips leading and trailing whitespace, and substitutes runs of whitespace with a single space each:

" \n  foo\n\r \t bar \n".squish # => "foo bar"

There’s also the destructive version String#squish!.

5.3 truncate

The method truncate returns a copy of its receiver truncated after a given length:

"Oh dear! Oh dear! I shall be late!".truncate(20)
# => "Oh dear! Oh dear!..."

Ellipsis can be customized with the :omission option:

"Oh dear! Oh dear! I shall be late!".truncate(20, :omission => '…')
# => "Oh dear! Oh …"

Note in particular that truncation takes into account the length of the omission string.

Pass a :separator to truncate the string at a natural break:

"Oh dear! Oh dear! I shall be late!".truncate(18)
# => "Oh dear! Oh dea..."
"Oh dear! Oh dear! I shall be late!".truncate(18, :separator => ' ')
# => "Oh dear! Oh..."

In the above example “dear” gets cut first, but then :separator prevents it.

5.4 Key-based Interpolation

In Ruby 1.9 the % string operator supports key-based interpolation, both formatted and unformatted:

"Total is %<total>.02f" % {:total => 43.1}  # => Total is 43.10
"I say %{foo}" % {:foo => "wadus"}          # => "I say wadus"
"I say %{woo}" % {:foo => "wadus"}          # => KeyError

Active Support adds that functionality to % in previous versions of Ruby.

5.5 starts_with? and ends_width?

Active Support defines 3rd person aliases of String#start_with? and String#end_with?:

"foo".starts_with?("f") # => true
"foo".ends_with?("o")   # => true

5.6 strip_heredoc

The method strip_heredoc strips indentation in heredocs.

For example in

if options[:usage]
  puts <<-USAGE.strip_heredoc
    This command does such and such.

    Supported options are:
      -h         This message
      ...
  USAGE
end

the user would see the usage message aligned against the left margin.

Technically, it looks for the least indented line in the whole string, and removes that amount of leading whitespace.

5.7 Access

5.7.1 at(position)

Returns the character of the string at position position:

"hello".at(0)  # => "h"
"hello".at(4)  # => "o"
"hello".at(-1) # => "o"
"hello".at(10) # => ERROR if < 1.9, nil in 1.9

5.7.2 from(position)

Returns the substring of the string starting at position position:

"hello".from(0)  # => "hello"
"hello".from(2)  # => "llo"
"hello".from(-2) # => "lo"
"hello".from(10) # => "" if < 1.9, nil in 1.9

5.7.3 to(position)

Returns the substring of the string up to position position:

"hello".to(0)  # => "h"
"hello".to(2)  # => "hel"
"hello".to(-2) # => "hell"
"hello".to(10) # => "hello"

5.7.4 first(limit = 1)

The call str.first(n) is equivalent to str.to(n-1) if n > 0, and returns an empty string for n == 0.

5.7.5 last(limit = 1)

The call str.last(n) is equivalent to str.from(-n) if n > 0, and returns an empty string for n == 0.

5.8 Inflections

5.8.1 pluralize

The method pluralize returns the plural of its receiver:

"table".pluralize     # => "tables"
"ruby".pluralize      # => "rubies"
"equipment".pluralize # => "equipment"

As the previous example shows, Active Support knows some irregular plurals and uncountable nouns. Built-in rules can be extended in config/initializers/inflections.rb. That file is generated by the rails command and has instructions in comments.

Active Record uses this method to compute the default table name that corresponds to a model:

# active_record/base.rb
def undecorated_table_name(class_name = base_class.name)
  table_name = class_name.to_s.demodulize.underscore
  table_name = table_name.pluralize if pluralize_table_names
  table_name
end

5.8.2 singularize

The inverse of pluralize:

"tables".singularize    # => "table"
"rubies".singularize    # => "ruby"
"equipment".singularize # => "equipment"

Associations compute the name of the corresponding default associated class using this method:

# active_record/reflection.rb
def derive_class_name
  class_name = name.to_s.camelize
  class_name = class_name.singularize if collection?
  class_name
end

5.8.3 camelize

The method camelize returns its receiver in camel case:

"product".camelize    # => "Product"
"admin_user".camelize # => "AdminUser"

As a rule of thumb you can think of this method as the one that transforms paths into Ruby class or module names, where slashes separate namespaces:

"backoffice/session".camelize # => "Backoffice::Session"

For example, Action Pack uses this method to load the class that provides a certain session store:

# action_controller/metal/session_management.rb
def session_store=(store)
  if store == :active_record_store
    self.session_store = ActiveRecord::SessionStore
  else
    @@session_store = store.is_a?(Symbol) ?
      ActionDispatch::Session.const_get(store.to_s.camelize) :
      store
  end
end

camelize accepts an optional argument, it can be :upper (default), or :lower. With the latter the first letter becomes lowercase:

"visual_effect".camelize(:lower) # => "visualEffect"

That may be handy to compute method names in a language that follows that convention, for example JavaScript.

camelize is aliased to camelcase.

5.8.4 underscore

The method underscore goes the other way around, from camel case to paths:

"Product".underscore   # => "product"
"AdminUser".underscore # => "admin_user"

Also converts “::” back to “/”:

"Backoffice::Session".underscore # => "backoffice/session"

and understands strings that start with lowercase:

"visualEffect".underscore # => "visual_effect"

underscore accepts no argument though.

Rails class and module autoloading uses underscore to infer the relative path without extension of a file that would define a given missing constant:

# active_support/dependencies.rb
def load_missing_constant(from_mod, const_name)
  ...
  qualified_name = qualified_name_for from_mod, const_name
  path_suffix = qualified_name.underscore
  ...
end

5.8.5 titleize

The method titleize capitalizes the words in the receiver:

"alice in wonderland".titleize # => "Alice In Wonderland"
"fermat's enigma".titleize     # => "Fermat's Enigma"

titleize is aliased to titlecase.

5.8.6 dasherize

The method dasherize replaces the underscores in the receiver with dashes:

"name".dasherize         # => "name"
"contact_data".dasherize # => "contact-data"

The XML serializer of models uses this method to dasherize node names:

# active_model/serializers/xml.rb
def reformat_name(name)
  name = name.camelize if camelize?
  dasherize? ? name.dasherize : name
end

5.8.7 demodulize

Given a string with a qualified constant reference expression, demodulize returns the very constant name, that is, the rightmost part of it:

"Product".demodulize                        # => "Product"
"Backoffice::UsersController".demodulize    # => "UsersController"
"Admin::Hotel::ReservationUtils".demodulize # => "ReservationUtils"

Active Record for example uses this method to compute the name of a counter cache column:

# active_record/reflection.rb
def counter_cache_column
  if options[:counter_cache] == true
    "#{active_record.name.demodulize.underscore.pluralize}_count"
  elsif options[:counter_cache]
    options[:counter_cache]
  end
end

5.8.8 parameterize

The method parameterize normalizes its receiver in a way that can be used in pretty URLs.

"John Smith".parameterize # => "john-smith"
"Kurt Gödel".parameterize # => "kurt-godel"

In fact, the result string is wrapped in an instance of ActiveSupport::Multibyte::Chars.

5.8.9 tableize

The method tableize is underscore followed by pluralize.

"Person".tableize      # => "people"
"Invoice".tableize     # => "invoices"
"InvoiceLine".tableize # => "invoice_lines"

As a rule of thumb, tableize returns the table name that corresponds to a given model for simple cases. The actual implementation in Active Record is not straight tableize indeed, because it also demodulizes de class name and checks a few options that may affect the returned string.

5.8.10 classify

The method classify is the inverse of tableize. It gives you the class name corresponding to a table name:

"people".classify        # => "Person"
"invoices".classify      # => "Invoice"
"invoice_lines".classify # => "InvoiceLine"

The method understands qualified table names:

"highrise_production.companies".classify # => "Company"

Note that classify returns a class name as a string. You can get the actual class object invoking constantize on it, explained next.

5.8.11 constantize

The method constantize resolves the constant reference expression in its receiver:

"Fixnum".constantize # => Fixnum

module M
  X = 1
end
"M::X".constantize # => 1

If the string evaluates to no known constant, or its content is not even a valid constant name, constantize raises NameError.

Constant name resolution by constantize starts always at the top-level Object even if there is no leading “::”.

X = :in_Object
module M
  X = :in_M

  X                 # => :in_M
  "::X".constantize # => :in_Object
  "X".constantize   # => :in_Object (!)
end

So, it is in general not equivalent to what Ruby would do in the same spot, had a real constant be evaluated.

Mailer test cases obtain the mailer being tested from the name of the test class using constantize:

# action_mailer/test_case.rb
def determine_default_mailer(name)
  name.sub(/Test$/, '').constantize
rescue NameError => e
  raise NonInferrableMailerError.new(name)
end

5.8.12 humanize

The method humanize gives you a sensible name for display out of an attribute name. To do so it replaces underscores with spaces, removes any “_id” suffix, and capitalizes the first word:

"name".humanize           # => "Name"
"author_id".humanize      # => "Author"
"comments_count".humanize # => "Comments count"

The helper method full_messages uses humanize as a fallback to include attribute names:

def full_messages
  full_messages = []

  each do |attribute, messages|
    ...
    attr_name = attribute.to_s.gsub('.', '_').humanize
    attr_name = @base.class.human_attribute_name(attribute, :default => attr_name)
    ...
  end

  full_messages
end

5.8.13 foreign_key

The method foreign_key gives a foreign key column name from a class name. To do so it demodulizes, underscores, and adds “_id”:

"User".foreign_key           # => "user_id"
"InvoiceLine".foreign_key    # => "invoice_line_id"
"Admin::Session".foreign_key # => "session_id"

Pass a false argument if you do not want the underscore in “_id”:

"User".foreign_key(false) # => "userid"

Associations use this method to infer foreign keys, for example has_one and has_many do this:

# active_record/associations.rb
foreign_key = options[:foreign_key] || reflection.active_record.name.foreign_key

5.9 Conversions

5.9.1 ord

Ruby 1.9 defines ord to be the codepoint of the first character of the receiver. Active Support backports ord for single-byte encondings like ASCII or ISO-8859-1 in Ruby 1.8:

"a".ord # => 97
"à".ord # => 224, in ISO-8859-1

In Ruby 1.8 ord doesn’t work in general in UTF8 strings, use the multibyte support in Active Support for that:

"a".mb_chars.ord # => 97
"à".mb_chars.ord # => 224, in UTF8

Note that the 224 is different in both examples. In ISO-8859-1 “à” is represented as a single byte, 224. Its single-character representattion in UTF8 has two bytes, namely 195 and 160, but its Unicode codepoint is 224. If we call ord on the UTF8 string “à” the return value will be 195 in Ruby 1.8. That is not an error, because UTF8 is unsupported, the call itself would be bogus.

5.9.2 getbyte

Active Support backports getbyte from Ruby 1.9:

"foo".getbyte(0)  # => 102, same as "foo".ord
"foo".getbyte(1)  # => 111
"foo".getbyte(9)  # => nil
"foo".getbyte(-1) # => 111

5.9.3 to_date, to_time, to_datetime

The methods to_date, to_time, and to_datetime are basically convenience wrappers around Date._parse:

"2010-07-27".to_date              # => Tue, 27 Jul 2010
"2010-07-27 23:37:00".to_time     # => Tue Jul 27 23:37:00 UTC 2010
"2010-07-27 23:37:00".to_datetime # => Tue, 27 Jul 2010 23:37:00 +0000

to_time receives an optional argument :utc or :local, to indicate which time zone you want the time in:

"2010-07-27 23:42:00".to_time(:utc)   # => Tue Jul 27 23:42:00 UTC 2010
"2010-07-27 23:42:00".to_time(:local) # => Tue Jul 27 23:42:00 +0200 2010

Default is :utc.

Please refer to the documentation of Date._parse for further details.

6 Extensions to Numeric

6.1 Bytes

All numbers respond to these methods:

bytes
kilobytes
megabytes
gigabytes
terabytes
petabytes
exabytes

They return the corresponding amount of bytes, using a conversion factor of 1024:

2.kilobytes   # => 2048
3.megabytes   # => 3145728
3.5.gigabytes # => 3758096384
-4.exabytes   # => -4611686018427387904

Singular forms are aliased so you are able to say:

1.megabyte # => 1048576

7 Extensions to Integer

7.1 multiple_of?

The method multiple_of? tests whether an integer is multiple of the argument:

2.multiple_of?(1) # => true
1.multiple_of?(2) # => false

7.2 ordinalize

The method ordinalize returns the ordinal string corresponding to the receiver integer:

1.ordinalize    # => "1st"
2.ordinalize    # => "2nd"
53.ordinalize   # => "53rd"
2009.ordinalize # => "2009th"

8 Extensions to Float

8.1 round

The built-in method Float#round rounds a float to the nearest integer. Active Support adds an optional parameter to let you specify a precision:

Math::E.round(4) # => 2.7183

9 Extensions to BigDecimal

…

10 Extensions to Enumerable

10.1 group_by

Active Support redefines group_by in Ruby 1.8.7 so that it returns an ordered hash as in 1.9:

entries_by_surname_initial = address_book.group_by do |entry|
  entry.surname.at(0).upcase
end

Distinct block return values are added to the hash as they come, so that’s the resulting order.

10.2 sum

The method sum adds the elements of an enumerable:

[1, 2, 3].sum # => 6
(1..100).sum  # => 5050

Addition only assumes the elements respond to +:

[[1, 2], [2, 3], [3, 4]].sum    # => [1, 2, 2, 3, 3, 4]
%w(foo bar baz).sum             # => "foobarbaz"
{:a => 1, :b => 2, :c => 3}.sum # => [:b, 2, :c, 3, :a, 1]

The sum of an empty collection is zero by default, but this is customizable:

[].sum    # => 0
[].sum(1) # => 1

If a block is given sum becomes an iterator that yields the elements of the collection and sums the returned values:

(1..5).sum {|n| n * 2 } # => 30
[2, 4, 6, 8, 10].sum    # => 30

The sum of an empty receiver can be customized in this form as well:

[].sum(1) {|n| n**3} # => 1

The method ActiveRecord::Observer#observed_subclasses for example is implemented this way:

def observed_subclasses
  observed_classes.sum([]) { |klass| klass.send(:subclasses) }
end

10.3 each_with_object

The inject method offers iteration with an accumulator:

[2, 3, 4].inject(1) {|acc, i| product*i } # => 24

The block is expected to return the value for the accumulator in the next iteration, and this makes building mutable objects a bit cumbersome:

[1, 2].inject({}) {|h, i| h[i] = i**2; h} # => {1 => 1, 2 => 4}

See that spurious “; h”?

Active Support backports each_with_object from Ruby 1.9, which addresses that use case. It iterates over the collection, passes the accumulator, and returns the accumulator when done. You normally modify the accumulator in place. The example above would be written this way:

[1, 2].each_with_object({}) {|i, h| h[i] = i**2} # => {1 => 1, 2 => 4}

10.4 index_by

The method index_by generates a hash with the elements of an enumerable indexed by some key.

It iterates through the collection and passes each element to a block. The element will be keyed by the value returned by the block:

invoices.index_by(&:number)
# => {'2009-032' => <Invoice ...>, '2009-008' => <Invoice ...>, ...}

10.5 many?

The method many? is shorthand for collection.size > 1:

<% if pages.many? %>
  <%= pagination_links %>
<% end %>

If an optional block is given many? only takes into account those elements that return true:

@see_more = videos.many? {|video| video.category == params[:category]}

10.6 exclude?

The predicate exclude? tests whether a given object does not belong to the collection. It is the negation of the builtin include?:

to_visit << node if visited.exclude?(node)

11 Extensions to Array

11.1 Accessing

Active Support augments the API of arrays to ease certain ways of accessing them. For example, to returns the subarray of elements up to the one at the passed index:

%w(a b c d).to(2) # => %w(a b c)
[].to(7)          # => []

Similarly, from returns the tail from the element at the passed index on:

%w(a b c d).from(2)  # => %w(c d)
%w(a b c d).from(10) # => nil
[].from(0)           # => []

The methods second, third, fourth, and fifth return the corresponding element (first is built-in). Thanks to social wisdom and positive constructiveness all around, forty_two is also available.

11.2 Random Access

Active Support backports sample from Ruby 1.9:

shape_type = [Circle, Square, Triangle].sample
# => Square, for example

shape_types = [Circle, Square, Triangle].sample(2)
# => [Triangle, Circle], for example

11.3 Options Extraction

When the last argument in a method call is a hash, except perhaps for a &block argument, Ruby allows you to omit the brackets:

User.exists?(:email => params[:email])

That syntactic sugar is used a lot in Rails to avoid positional arguments where there would be too many, offering instead interfaces that emulate named parameters. In particular it is very idiomatic to use a trailing hash for options.

If a method expects a variable number of arguments and uses * in its declaration, however, such an options hash ends up being an item of the array of arguments, where kind of loses its role.

In those cases, you may give an options hash a distinguished treatment with extract_options!. That method checks the type of the last item of an array. If it is a hash it pops it and returns it, otherwise returns an empty hash.

Let’s see for example the definition of the caches_action controller macro:

def caches_action(*actions)
  return unless cache_configured?
  options = actions.extract_options!
  ...
end

This method receives an arbitrary number of action names, and an optional hash of options as last argument. With the call to extract_options! you obtain the options hash and remove it from actions in a simple and explicit way.

11.4 Conversions

11.4.1 to_sentence

The method to_sentence turns an array into a string containing a sentence that enumerates its items:

%w().to_sentence                # => ""
%w(Earth).to_sentence           # => "Earth"
%w(Earth Wind).to_sentence      # => "Earth and Wind"
%w(Earth Wind Fire).to_sentence # => "Earth, Wind, and Fire"

This method accepts three options:

• :two_words_connector: What is used for arrays of length 2. Default is " and ".
• :words_connector: What is used to join the elements of arrays with 3 or more elements, except for the last two. Default is ", ".
• :last_word_connector: What is used to join the last items of an array with 3 or more elements. Default is ", and ".

The defaults for these options can be localised, their keys are:

Options :connector and :skip_last_comma are deprecated.

11.4.2 to_formatted_s

The method to_formatted_s acts like to_s by default.

If the array contains items that respond to id, however, it may be passed the symbol :db as argument. That’s typically used with collections of ARs, though technically any object in Ruby 1.8 responds to id indeed. Returned strings are:

[].to_formatted_s(:db)            # => "null"
[user].to_formatted_s(:db)        # => "8456"
invoice.lines.to_formatted_s(:db) # => "23,567,556,12"

Integers in the example above are supposed to come from the respective calls to id.

11.4.3 to_xml

The method to_xml returns a string containing an XML representation of its receiver:

Contributor.all(:limit => 2, :order => 'rank ASC').to_xml
# =>
# <?xml version="1.0" encoding="UTF-8"?>
# <contributors type="array">
#   <contributor>
#     <id type="integer">4356</id>
#     <name>Jeremy Kemper</name>
#     <rank type="integer">1</rank>
#     <url-id>jeremy-kemper</url-id>
#   </contributor>
#   <contributor>
#     <id type="integer">4404</id>
#     <name>David Heinemeier Hansson</name>
#     <rank type="integer">2</rank>
#     <url-id>david-heinemeier-hansson</url-id>
#   </contributor>
# </contributors>

To do so it sends to_xml to every item in turn, and collects the results under a root node. All items must respond to to_xml, an exception is raised otherwise.

By default, the name of the root element is the underscorized and dasherized plural of the name of the class of the first item, provided the rest of elements belong to that type (checked with is_a?) and they are not hashes. In the example above that’s “contributors”.

If there’s any element that does not belong to the type of the first one the root node becomes “records”:

[Contributor.first, Commit.first].to_xml
# =>
# <?xml version="1.0" encoding="UTF-8"?>
# <records type="array">
#   <record>
#     <id type="integer">4583</id>
#     <name>Aaron Batalion</name>
#     <rank type="integer">53</rank>
#     <url-id>aaron-batalion</url-id>
#   </record>
#   <record>
#     <author>Joshua Peek</author>
#     <authored-timestamp type="datetime">2009-09-02T16:44:36Z</authored-timestamp>
#     <branch>origin/master</branch>
#     <committed-timestamp type="datetime">2009-09-02T16:44:36Z</committed-timestamp>
#     <committer>Joshua Peek</committer>
#     <git-show nil="true"></git-show>
#     <id type="integer">190316</id>
#     <imported-from-svn type="boolean">false</imported-from-svn>
#     <message>Kill AMo observing wrap_with_notifications since ARes was only using it</message>
#     <sha1>723a47bfb3708f968821bc969a9a3fc873a3ed58</sha1>
#   </record>
# </records>

If the receiver is an array of hashes the root element is by default also “records”:

[{:a => 1, :b => 2}, {:c => 3}].to_xml
# =>
# <?xml version="1.0" encoding="UTF-8"?>
# <records type="array">
#   <record>
#     <b type="integer">2</b>
#     <a type="integer">1</a>
#   </record>
#   <record>
#     <c type="integer">3</c>
#   </record>
# </records>

The name of children nodes is by default the name of the root node singularized. In the examples above we’ve seen “contributor” and “record”. The option :children allows you to set these node names.

The default XML builder is a fresh instance of Builder::XmlMarkup. You can configure your own builder via the :builder option. The method also accepts options like :dasherize and friends, they are forwarded to the builder:

Contributor.all(:limit => 2, :order => 'rank ASC').to_xml(:skip_types => true)
# =>
# <?xml version="1.0" encoding="UTF-8"?>
# <contributors>
#   <contributor>
#     <id>4356</id>
#     <name>Jeremy Kemper</name>
#     <rank>1</rank>
#     <url-id>jeremy-kemper</url-id>
#   </contributor>
#   <contributor>
#     <id>4404</id>
#     <name>David Heinemeier Hansson</name>
#     <rank>2</rank>
#     <url-id>david-heinemeier-hansson</url-id>
#   </contributor>
# </contributors>

11.5 Wrapping

The method Array.wrap wraps its argument in an array unless it is already an array (or array-like).

Specifically:

• If the argument is nil an empty list is returned.
• Otherwise, if the argument responds to to_ary it is invoked, and its result returned.
• Otherwise, returns an array with the argument as its single element.

Array.wrap(nil)       # => []
Array.wrap([1, 2, 3]) # => [1, 2, 3]
Array.wrap(0)         # => [0]

This method is similar in purpose to Kernel#Array, but there are some differences:

• If the argument responds to to_ary the method is invoked. Kernel#Array moves on to try to_a if the returned value is nil, but Arraw.wrap returns such a nil right away.
• If the returned value from to_ary is neither nil nor an Array object, Kernel#Array raises an exception, while Array.wrap does not, it just returns the value.
• It does not call to_a on the argument, though special-cases nil to return an empty array.

The last point is particularly worth comparing for some enumerables:

Array.wrap(:foo => :bar) # => [{:foo => :bar}]
Array(:foo => :bar)      # => [[:foo, :bar]]

Array.wrap("foo\nbar")   # => ["foo\nbar"]
Array("foo\nbar")        # => ["foo\n", "bar"], in Ruby 1.8

There’s also a related idiom that uses the splat operator:

[*object]

which in Ruby 1.8 returns [nil] for nil, and calls to Array(object) otherwise. (Please if you know the exact behavior in 1.9 contact fxn.)

Thus, in this case the behavior is different for nil, and the differences with Kernel#Array explained above apply to the rest of objects.

11.6 Grouping

11.6.1 in_groups_of(number, fill_with = nil)

The method in_groups_of splits an array into consecutive groups of a certain size. It returns an array with the groups:

[1, 2, 3].in_groups_of(2) # => [[1, 2], [3, nil]]

or yields them in turn if a block is passed:

<% sample.in_groups_of(3) do |a, b, c| %>
  <tr>
    <td><%=h a %></td>
    <td><%=h b %></td>
    <td><%=h c %></td>
  </tr>
<% end %>

The first example shows in_groups_of fills the last group with as many nil elements as needed to have the requested size. You can change this padding value using the second optional argument:

[1, 2, 3].in_groups_of(2, 0) # => [[1, 2], [3, 0]]

And you can tell the method not to fill the last group passing false:

[1, 2, 3].in_groups_of(2, false) # => [[1, 2], [3]]

As a consequence false can’t be a used as a padding value.

11.6.2 in_groups(number, fill_with = nil)

The method in_groups splits an array into a certain number of groups. The method returns and array with the groups:

%w(1 2 3 4 5 6 7).in_groups(3)
# => [["1", "2", "3"], ["4", "5", nil], ["6", "7", nil]]

or yields them in turn if a block is passed:

%w(1 2 3 4 5 6 7).in_groups(3) {|group| p group}
["1", "2", "3"]
["4", "5", nil]
["6", "7", nil]

The examples above show that in_groups fills some groups with a trailing nil element as needed. A group can get at most one of these extra elements, the rightmost one if any. And the groups that have them are always the last ones.

You can change this padding value using the second optional argument:

%w(1 2 3 4 5 6 7).in_groups(3, "0")
# => [["1", "2", "3"], ["4", "5", "0"], ["6", "7", "0"]]

And you can tell the method not to fill the smaller groups passing false:

%w(1 2 3 4 5 6 7).in_groups(3, false)
# => [["1", "2", "3"], ["4", "5"], ["6", "7"]]

As a consequence false can’t be a used as a padding value.

11.6.3 split(value = nil)

The method split divides an array by a separator and returns the resulting chunks.

If a block is passed the separators are those elements of the array for which the block returns true:

(-5..5).to_a.split { |i| i.multiple_of?(4) }
# => [[-5], [-3, -2, -1], [1, 2, 3], [5]]

Otherwise, the value received as argument, which defaults to nil, is the separator:

[0, 1, -5, 1, 1, "foo", "bar"].split(1)
# => [[0], [-5], [], ["foo", "bar"]]

12 Extensions to Hash

12.1 Conversions

12.1.1 to_xml

The method to_xml returns a string containing an XML representation of its receiver:

{"foo" => 1, "bar" => 2}.to_xml
# =>
# <?xml version="1.0" encoding="UTF-8"?>
# <hash>
#   <foo type="integer">1</foo>
#   <bar type="integer">2</bar>
# </hash>

To do so, the method loops over the pairs and builds nodes that depend on the values. Given a pair key, value:

• If value is a hash there’s a recursive call with key as :root.

• If value is an array there’s a recursive call with key as :root, and key singularized as :children.

• If value is a callable object it must expect one or two arguments. Depending on the arity, the callable is invoked with the options hash as first argument with key as :root, and key singularized as second argument. Its return value becomes a new node.

• If value responds to to_xml the method is invoked with key as :root.

• Otherwise, a node with key as tag is created with a string representation of value as text node. If value is nil an attribute “nil” set to “true” is added. Unless the option :skip_types exists and is true, an attribute “type” is added as well according to the following mapping:

XML_TYPE_NAMES = {
  "Symbol"     => "symbol",
  "Fixnum"     => "integer",
  "Bignum"     => "integer",
  "BigDecimal" => "decimal",
  "Float"      => "float",
  "TrueClass"  => "boolean",
  "FalseClass" => "boolean",
  "Date"       => "date",
  "DateTime"   => "datetime",
  "Time"       => "datetime"
}

By default the root node is “hash”, but that’s configurable via the :root option.

The default XML builder is a fresh instance of Builder::XmlMarkup. You can configure your own builder with the :builder option. The method also accepts options like :dasherize and friends, they are forwarded to the builder.

12.2 Merging

Ruby has a built-in method Hash#merge that merges two hashes:

{:a => 1, :b => 1}.merge(:a => 0, :c => 2)
# => {:a => 0, :b => 1, :c => 2}

Active Support defines a few more ways of merging hashes that may be convenient.

12.2.1 reverse_merge and reverse_merge!

In case of collision the key in the hash of the argument wins in merge. You can support option hashes with default values in a compact way with this idiom:

options = {:length => 30, :omission => "..."}.merge(options)

Active Support defines reverse_merge in case you prefer this alternative notation:

options = options.reverse_merge(:length => 30, :omission => "...")

And a bang version reverse_merge! that performs the merge in place:

options.reverse_merge!(:length => 30, :omission => "...")

12.2.2 reverse_update

The method reverse_update is an alias for reverse_merge!, explained above.

12.2.3 deep_merge and deep_merge!

As you can see in the previous example if a key is found in both hashes the value in the one in the argument wins.

Active Support defines Hash#deep_merge. In a deep merge, if a key is found in both hashes and their values are hashes in turn, then their merge becomes the value in the resulting hash:

{:a => {:b => 1}}.deep_merge(:a => {:c => 2})
# => {:a => {:b => 1, :c => 2}}

The method deep_merge! performs a deep merge in place.

12.3 Diffing

The method diff returns a hash that represents a diff of the receiver and the argument with the following logic:

• Pairs key, value that exist in both hashes do not belong to the diff hash.

• If both hashes have key, but with different values, the pair in the receiver wins.

• The rest is just merged.

{:a => 1}.diff(:a => 1)
# => {}, first rule

{:a => 1}.diff(:a => 2)
# => {:a => 1}, second rule

{:a => 1}.diff(:b => 2)
# => {:a => 1, :b => 2}, third rule

{:a => 1, :b => 2, :c => 3}.diff(:b => 1, :c => 3, :d => 4)
# => {:a => 1, :b => 2, :d => 4}, all rules

{}.diff({})        # => {}
{:a => 1}.diff({}) # => {:a => 1}
{}.diff(:a => 1)   # => {:a => 1}

An important property of this diff hash is that you can retrieve the original hash by applying diff twice:

hash.diff(hash2).diff(hash2) == hash

Diffing hashes may be useful for error messages related to expected option hashes for example.

12.4 Working with Keys

12.4.1 except and except!

The method except returns a hash with the keys in the argument list removed, if present:

{:a => 1, :b => 2}.except(:a) # => {:b => 2}

If the receiver responds to convert_key, the method is called on each of the arguments. This allows except to play nice with hashes with indifferent access for instance:

{:a => 1}.with_indifferent_access.except(:a)  # => {}
{:a => 1}.with_indifferent_access.except("a") # => {}

The method except may come in handy for example when you want to protect some parameter that can’t be globally protected with attr_protected:

params[:account] = params[:account].except(:plan_id) unless admin?
@account.update_attributes(params[:account])

There’s also the bang variant except! that removes keys in the very receiver.

12.4.2 stringify_keys and stringify_keys!

The method stringify_keys returns a hash that has a stringified version of the keys in the receiver. It does so by sending to_s to them:

{nil => nil, 1 => 1, :a => :a}.stringify_keys
# => {"" => nil, "a" => :a, "1" => 1}

The result in case of collision is undefined:

{"a" => 1, :a => 2}.stringify_keys
# => {"a" => 2}, in my test, can't rely on this result though

This method may be useful for example to easily accept both symbols and strings as options. For instance ActionView::Helpers::FormHelper defines:

def to_check_box_tag(options = {}, checked_value = "1", unchecked_value = "0")
  options = options.stringify_keys
  options["type"] = "checkbox"
  ...
end

The second line can safely access the “type” key, and let the user to pass either :type or “type”.

There’s also the bang variant stringify_keys! that stringifies keys in the very receiver.

12.4.3 symbolize_keys and symbolize_keys!

The method symbolize_keys returns a hash that has a symbolized version of the keys in the receiver, where possible. It does so by sending to_sym to them:

{nil => nil, 1 => 1, "a" => "a"}.symbolize_keys
# => {1 => 1, nil => nil, :a => "a"}

The result in case of collision is undefined:

{"a" => 1, :a => 2}.symbolize_keys
# => {:a => 2}, in my test, can't rely on this result though

This method may be useful for example to easily accept both symbols and strings as options. For instance ActionController::UrlRewriter defines

def rewrite_path(options)
  options = options.symbolize_keys
  options.update(options[:params].symbolize_keys) if options[:params]
  ...
end

The second line can safely access the :params key, and let the user to pass either :params or “params”.

There’s also the bang variant symbolize_keys! that symbolizes keys in the very receiver.

12.4.4 to_options and to_options!

The methods to_options and to_options! are respectively aliases of symbolize_keys and symbolize_keys!.

12.4.5 assert_valid_keys

The method assert_valid_keys receives an arbitrary number of arguments, and checks whether the receiver has any key outside that white list. If it does ArgumentError is raised.

{:a => 1}.assert_valid_keys(:a)  # passes
{:a => 1}.assert_valid_keys("a") # ArgumentError

Active Record does not accept unknown options when building associations for example. It implements that control via assert_valid_keys:

mattr_accessor :valid_keys_for_has_many_association
@@valid_keys_for_has_many_association = [
  :class_name, :table_name, :foreign_key, :primary_key,
  :dependent,
  :select, :conditions, :include, :order, :group, :having, :limit, :offset,
  :as, :through, :source, :source_type,
  :uniq,
  :finder_sql, :counter_sql,
  :before_add, :after_add, :before_remove, :after_remove,
  :extend, :readonly,
  :validate, :inverse_of
]

def create_has_many_reflection(association_id, options, &extension)
  options.assert_valid_keys(valid_keys_for_has_many_association)
  ...
end

12.5 Slicing

Ruby has built-in support for taking slices out of strings and arrays. Active Support extends slicing to hashes:

{:a => 1, :b => 2, :c => 3}.slice(:a, :c)
# => {:c => 3, :a => 1}

{:a => 1, :b => 2, :c => 3}.slice(:b, :X)
# => {:b => 2} # non-existing keys are ignored

If the receiver responds to convert_key keys are normalized:

{:a => 1, :b => 2}.with_indifferent_access.slice("a")
# => {:a => 1}

There’s also slice! which in addition to perform a slice in place returns what’s removed:

hash = {:a => 1, :b => 2}
rest = hash.slice!(:a) # => {:b => 2}
hash                   # => {:a => 1}

12.6 Indifferent Access

The method with_indifferent_access returns an ActiveSupport::HashWithIndifferentAccess out of its receiver:

{:a => 1}.with_indifferent_access["a"] # => 1

13 Extensions to Regexp

13.1 multiline?

The method multiline? says whether a regexp has the /m flag set, that is, whether the dot matches newlines.

%r{.}.multiline?  # => false
%r{.}m.multiline? # => true

Regexp.new('.').multiline?                    # => false
Regexp.new('.', Regexp::MULTILINE).multiline? # => true

Rails uses this method in a single place, also in the routing code. Multiline regexps are disallowed for route requirements and this flag eases enforcing that constraint.

def assign_route_options(segments, defaults, requirements)
  ...
  if requirement.multiline?
    raise ArgumentError, "Regexp multiline option not allowed in routing requirements: #{requirement.inspect}"
  end
  ...
end

14 Extensions to Range

14.1 to_s

Active Support extends the method Range#to_s so that it understands an optional format argument. As of this writing the only supported non-default format is :db:

(Date.today..Date.tomorrow).to_s
# => "2009-10-25..2009-10-26"

(Date.today..Date.tomorrow).to_s(:db)
# => "BETWEEN '2009-10-25' AND '2009-10-26'"

As the example depicts, the :db format generates a BETWEEN SQL clause. That is used by Active Record in its support for range values in conditions.

14.2 step

Active Support extends the method Range#step so that it can be invoked without a block:

(1..10).step(2) # => [1, 3, 5, 7, 9]

As the example shows, in that case the method returns and array with the corresponding elements.

14.3 include?

The method Range#include? says whether some value falls between the ends of a given instance:

(2..3).include?(Math::E) # => true

Active Support extends this method so that the argument may be another range in turn. In that case we test whether the ends of the argument range belong to the receiver themselves:

(1..10).include?(3..7)  # => true
(1..10).include?(0..7)  # => false
(1..10).include?(3..11) # => false
(1...9).include?(3..9)  # => false

14.4 overlaps?

The method Range#overlaps? says whether any two given ranges have non-void intersection:

(1..10).overlaps?(7..11)  # => true
(1..10).overlaps?(0..7)   # => true
(1..10).overlaps?(11..27) # => false

15 Extensions to Proc

15.1 bind

As you surely know Ruby has an UnboundMethod class whose instances are methods that belong to the limbo of methods without a self. The method Module#instance_method returns an unbound method for example:

Hash.instance_method(:delete) # => #<UnboundMethod: Hash#delete>

An unbound method is not callable as is, you need to bind it first to an object with bind:

clear = Hash.instance_method(:clear)
clear.bind({:a => 1}).call # => {}

Active Support defines Proc#bind with an analogous purpose:

Proc.new { size }.bind([]).call # => 0

As you see that’s callable and bound to the argument, the return value is indeed a Method.

Action Pack uses this trick in rescue_from for example, which accepts the name of a method and also a proc as callbacks for a given rescued exception. It has to call them in either case, so a bound method is returned by handler_for_rescue, thus simplifying the code in the caller:

def handler_for_rescue(exception)
  _, rescuer = Array(rescue_handlers).reverse.detect do |klass_name, handler|
    ...
  end

  case rescuer
  when Symbol
    method(rescuer)
  when Proc
    rescuer.bind(self)
  end
end

16 Extensions to Date

16.1 Calculations

16.1.1 Date.current

Active Support defines Date.current to be today in the current time zone. That’s like Date.today, except that it honors the user time zone, if defined. It also defines Date.yesterday and Date.tomorrow, and the instance predicates past?, today?, and future?, all of them relative to Date.current.

16.1.2 Named dates

16.1.2.1 prev_year, next_year

In Ruby 1.9 prev_year and next_year return a date with the same day/month in the last or next year:

d = Date.new(2010, 5, 8) # => Sat, 08 May 2010
d.prev_year              # => Fri, 08 May 2009
d.next_year              # => Sun, 08 May 2011

If date is the 29th of February of a leap year, you obtain the 28th:

d = Date.new(2000, 2, 29) # => Tue, 29 Feb 2000
d.prev_year               # => Sun, 28 Feb 1999
d.next_year               # => Wed, 28 Feb 2001

Active Support defines these methods as well for Ruby 1.8.

16.1.2.2 prev_month, next_month

In Ruby 1.9 prev_month and next_month return the date with the same day in the last or next month:

d = Date.new(2010, 5, 8) # => Sat, 08 May 2010
d.prev_month             # => Thu, 08 Apr 2010
d.next_month             # => Tue, 08 Jun 2010

If such a day does not exist, the last day of the corresponding month is returned:

Date.new(2000, 5, 31).prev_month # => Sun, 30 Apr 2000
Date.new(2000, 3, 31).prev_month # => Tue, 29 Feb 2000
Date.new(2000, 5, 31).next_month # => Fri, 30 Jun 2000
Date.new(2000, 1, 31).next_month # => Tue, 29 Feb 2000

Active Support defines these methods as well for Ruby 1.8.

16.1.2.3 beginning_of_week, end_of_week

The methods beginning_of_week and end_of_week return the dates for the beginning and end of week, assuming weeks start on Monday:

d = Date.new(2010, 5, 8) # => Sat, 08 May 2010
d.beginning_of_week      # => Mon, 03 May 2010
d.end_of_week            # => Sun, 09 May 2010

beginning_of_week is aliased to monday and at_beginning_of_week. end_of_week is aliased to sunday and at_end_of_week.

16.1.2.4 next_week

next_week receives a symbol with a day name in English (in lowercase, default is :monday) and it returns the date corresponding to that day in the next week:

d = Date.new(2010, 5, 9) # => Sun, 09 May 2010
d.next_week              # => Mon, 10 May 2010
d.next_week(:saturday)   # => Sat, 15 May 2010

16.1.2.5 beginning_of_month, end_of_month

The methods beginning_of_month and end_of_month return the dates for the beginning and end of the month:

d = Date.new(2010, 5, 9) # => Sun, 09 May 2010
d.beginning_of_month     # => Sat, 01 May 2010
d.end_of_month           # => Mon, 31 May 2010

beginning_of_month is aliased to at_beginning_of_month, and end_of_month is aliased to at_end_of_month.

16.1.2.6 beginning_of_quarter, end_of_quarter

The methods beginning_of_quarter and end_of_quarter return the dates for the beginning and end of the quarter of the receiver’s calendar year:

d = Date.new(2010, 5, 9) # => Sun, 09 May 2010
d.beginning_of_quarter   # => Thu, 01 Apr 2010
d.end_of_quarter         # => Wed, 30 Jun 2010

beginning_of_quarter is aliased to at_beginning_of_quarter, and end_of_quarter is aliased to at_end_of_quarter.

16.1.2.7 beginning_of_year, end_of_year

The methods beginning_of_year and end_of_year return the dates for the beginning and end of the year:

d = Date.new(2010, 5, 9) # => Sun, 09 May 2010
d.beginning_of_year      # => Fri, 01 Jan 2010
d.end_of_year            # => Fri, 31 Dec 2010

beginning_of_year is aliased to at_beginning_of_year, and end_of_year is aliased to at_end_of_year.

16.1.3 Other Date Computations

16.1.3.1 years_ago, years_since

The method years_ago receives a number of years and returns the same date those many years ago:

date = Date.new(2010, 6, 7)
date.years_ago(10) # => Wed, 07 Jun 2000

years_since moves forward in time:

date = Date.new(2010, 6, 7)
date.years_since(10) # => Sun, 07 Jun 2020

If such a day does not exist, the last day of the corresponding month is returned:

Date.new(2012, 2, 29).years_ago(3)     # => Sat, 28 Feb 2009
Date.new(2012, 2, 29).years_since(3)   # => Sat, 28 Feb 2015

16.1.3.2 months_ago, months_since

The methods months_ago and months_since work analogously for months:

Date.new(2010, 4, 30).months_ago(2)   # => Sun, 28 Feb 2010
Date.new(2010, 4, 30).months_since(2) # => Wed, 30 Jun 2010

If such a day does not exist, the last day of the corresponding month is returned:

Date.new(2010, 4, 30).months_ago(2)    # => Sun, 28 Feb 2010
Date.new(2009, 12, 31).months_since(2) # => Sun, 28 Feb 2010

16.1.3.3 advance

The most generic way to jump to other days is advance. This method receives a hash with keys :years, :months, :weeks, :days, and returns a date advanced as much as the present keys indicate:

date = Date.new(2010, 6, 6)
date.advance(:years => 1, :weeks => 2)  # => Mon, 20 Jun 2011
date.advance(:months => 2, :days => -2) # => Wed, 04 Aug 2010

Note in the previous example that increments may be negative.

To perform the computation the method first increments years, then months, then weeks, and finally days. This order is important towards the end of months. Say for example we are at the end of February of 2010, and we want to move one month and one day forward.

The method advance advances first one month, and then one day, the result is:

Date.new(2010, 2, 28).advance(:months => 1, :days => 1)
# => Sun, 29 Mar 2010

While if it did it the other way around the result would be different:

Date.new(2010, 2, 28).advance(:days => 1).advance(:months => 1)
# => Thu, 01 Apr 2010

16.1.4 Changing Components

The method change allows you to get a new date which is the same as the receiver except for the given year, month, or day:

Date.new(2010, 12, 23).change(:year => 2011, :month => 11)
# => Wed, 23 Nov 2011

This method is not tolerant to non-existing dates, if the change is invalid ArgumentError is raised:

Date.new(2010, 1, 31).change(:month => 2)
# => ArgumentError: invalid date

16.1.5 Durations

Durations can be added and substracted to dates:

d = Date.current
# => Mon, 09 Aug 2010
d + 1.year
# => Tue, 09 Aug 2011
d - 3.hours
# => Sun, 08 Aug 2010 21:00:00 UTC +00:00

They translate to calls to since or advance. For example here we get the correct jump in the calendar reform:

Date.new(1582, 10, 4) + 1.day
# => Fri, 15 Oct 1582

16.1.6 Timestamps

16.1.6.1 beginning_of_day, end_of_day

The method beginning_of_day returns a timestamp at the beginning of the day (00:00:00):

date = Date.new(2010, 6, 7)
date.beginning_of_day # => Sun Jun 07 00:00:00 +0200 2010

The method end_of_day returns a timestamp at the end of the day (23:59:59):

date = Date.new(2010, 6, 7)
date.end_of_day # => Sun Jun 06 23:59:59 +0200 2010

beginning_of_day is aliased to at_beginning_of_day, midnight, at_midnight.

16.1.6.2 ago, since

The method ago receives a number of seconds as argument and returns a timestamp those many seconds ago from midnight:

date = Date.current # => Fri, 11 Jun 2010
date.ago(1)         # => Thu, 10 Jun 2010 23:59:59 EDT -04:00

Similarly, since moves forward:

date = Date.current # => Fri, 11 Jun 2010
date.since(1)       # => Fri, 11 Jun 2010 00:00:01 EDT -04:00

16.1.7 Other Time Computations

16.2 Conversions

17 Extensions to DateTime

17.1 Calculations

The class DateTime is a subclass of Date so by loading active_support/core_ext/date/calculations.rb you inherit these methods and their aliases, except that they will always return datetimes:

yesterday
tomorrow
beginning_of_week (monday, at_beginning_of_week)
end_on_week (at_end_of_week)
next_week
months_ago
months_since
beginning_of_month (at_beginning_of_month)
end_of_month (at_end_of_month)
prev_month
next_month
beginning_of_quarter (at_beginning_of_quarter)
end_of_quarter (at_end_of_quarter)
beginning_of_year (at_beginning_of_year)
end_of_year (at_end_of_year)
years_ago
years_since
prev_year
next_year

The following methods are reimplemented so you do not need to load active_support/core_ext/date/calculations.rb for these ones:

beginning_of_day (midnight, at_midnight, at_beginning_of_day)
end_of_day
ago
since (in)

On the other hand, advance and change are also defined and support more options, they are documented below.

17.1.1 Named Datetimes

17.1.1.1 DateTime.current

Active Support defines DateTime.current to be like Time.now.to_datetime, except that it honors the user time zone, if defined. It also defines instance predicates past?, and future? relative to DateTime.current.

17.1.2 Other Extensions

17.1.2.1 seconds_since_midnight

The method seconds_since_midnight returns the number of seconds since midnight:

now = DateTime.current     # => Mon, 07 Jun 2010 20:26:36 +0000
now.seconds_since_midnight # => 73596

17.1.2.2 utc

The method utc gives you the same datetime in the receiver expressed in UTC.

now = DateTime.current # => Mon, 07 Jun 2010 19:27:52 -0400
now.utc                # => Mon, 07 Jun 2010 23:27:52 +0000

This method is also aliased as getutc.

17.1.2.3 utc?

The predicate utc? says whether the receiver has UTC as its time zone:

now = DateTime.now # => Mon, 07 Jun 2010 19:30:47 -0400
now.utc?           # => false
now.utc.utc?       # => true

17.1.2.4 advance

The most generic way to jump to another datetime is advance. This method receives a hash with keys :years, :months, :weeks, :days, :hours, :minutes, and :seconds, and returns a datetime advanced as much as the present keys indicate.

d = DateTime.current
# => Thu, 05 Aug 2010 11:33:31 +0000
d.advance(:years => 1, :months => 1, :days => 1, :hours => 1, :minutes => 1, :seconds => 1)
# => Tue, 06 Sep 2011 12:34:32 +0000

This method first computes the destination date passing :years, :months, :weeks, and :days to Date#advance documented above. After that, it adjusts the time calling since with the number of seconds to advance. This order is relevant, a different ordering would give different datetimes in some edge-cases. The example in Date#advance applies, and we can extend it to show order relevance related to the time bits.

If we first move the date bits (that have also a relative order of processing, as documented before), and then the time bits we get for example the following computation:

d = DateTime.new(2010, 2, 28, 23, 59, 59)
# => Sun, 28 Feb 2010 23:59:59 +0000
d.advance(:months => 1, :seconds => 1)
# => Mon, 29 Mar 2010 00:00:00 +0000

but if we computed them the other way around, the result would be different:

d.advance(:seconds => 1).advance(:months => 1)
# => Thu, 01 Apr 2010 00:00:00 +0000

17.1.3 Changing Components

The method change allows you to get a new datetime which is the same as the receiver except for the given options, which may include :year, :month, :day, :hour, :min, :sec, :offset, :start:

now = DateTime.current
# => Tue, 08 Jun 2010 01:56:22 +0000
now.change(:year => 2011, :offset => Rational(-6, 24))
# => Wed, 08 Jun 2011 01:56:22 -0600

If hours are zeroed, then minutes and seconds are too (unless they have given values):

now.change(:hour => 0)
# => Tue, 08 Jun 2010 00:00:00 +0000

Similarly, if minutes are zeroed, then seconds are too (unless it has given a value):

now.change(:min => 0)
# => Tue, 08 Jun 2010 01:00:00 +0000

This method is not tolerant to non-existing dates, if the change is invalid ArgumentError is raised:

DateTime.current.change(:month => 2, :day => 30)
# => ArgumentError: invalid date

17.1.4 Durations

Durations can be added and substracted to datetimes:

now = DateTime.current
# => Mon, 09 Aug 2010 23:15:17 +0000
now + 1.year
# => Tue, 09 Aug 2011 23:15:17 +0000
now - 1.week
# => Mon, 02 Aug 2010 23:15:17 +0000

They translate to calls to since or advance. For example here we get the correct jump in the calendar reform:

DateTime.new(1582, 10, 4, 23) + 1.hour
# => Fri, 15 Oct 1582 00:00:00 +0000

18 Extensions to Time

18.1 Calculations

Active Support adds to Time many of the methods available for DateTime:

past?
today?
future?
yesterday
tomorrow
seconds_since_midnight
change
advance
ago
since (in)
beginning_of_day (midnight, at_midnight, at_beginning_of_day)
end_of_day
beginning_of_week (monday, at_beginning_of_week)
end_on_week (at_end_of_week)
next_week
months_ago
months_since
beginning_of_month (at_beginning_of_month)
end_of_month (at_end_of_month)
prev_month
next_month
beginning_of_quarter (at_beginning_of_quarter)
end_of_quarter (at_end_of_quarter)
beginning_of_year (at_beginning_of_year)
end_of_year (at_end_of_year)
years_ago
years_since
prev_year
next_year

They are analogous. Please refer to their documentation above and take into account the following differences:

• change accepts an additional :usec option.
• Time understands DST, so you get correct DST calculations as in

Time.zone_default
# => #<ActiveSupport::TimeZone:0x7f73654d4f38 @utc_offset=nil, @name="Madrid", ...>

# In Barcelona, 2010/03/28 02:00 +0100 becomes 2010/03/28 03:00 +0200 due to DST.
t = Time.local_time(2010, 3, 28, 1, 59, 59)
# => Sun Mar 28 01:59:59 +0100 2010
t.advance(:seconds => 1)
# => Sun Mar 28 03:00:00 +0200 2010

• If since or ago jump to a time that can’t be expressed with Time a DateTime object is returned instead.

18.2 Time Constructors

Active Support defines Time.current to be Time.zone.now if there’s a user time zone defined, with fallback to Time.now:

Time.zone_default
# => #<ActiveSupport::TimeZone:0x7f73654d4f38 @utc_offset=nil, @name="Madrid", ...>
Time.current
# => Fri, 06 Aug 2010 17:11:58 CEST +02:00

Analogously to DateTime, the predicates past?, and future? are relative to Time.current.

Use the local_time class method to create time objects honoring the user time zone:

Time.zone_default
# => #<ActiveSupport::TimeZone:0x7f73654d4f38 @utc_offset=nil, @name="Madrid", ...>
Time.local_time(2010, 8, 15)
# => Sun Aug 15 00:00:00 +0200 2010

The utc_time class method returns a time in UTC:

Time.zone_default
# => #<ActiveSupport::TimeZone:0x7f73654d4f38 @utc_offset=nil, @name="Madrid", ...>
Time.utc_time(2010, 8, 15)
# => Sun Aug 15 00:00:00 UTC 2010

Both local_time and utc_time accept up to seven positional arguments: year, month, day, hour, min, sec, usec. Year is mandatory, month and day default to 1, and the rest default to 0.

If the time to be constructed lies beyond the range supported by Time in the runtime platform, usecs are discarded and a DateTime object is returned instead.

18.2.1 Durations

Durations can be added and substracted to time objects:

now = Time.current
# => Mon, 09 Aug 2010 23:20:05 UTC +00:00
now + 1.year
#  => Tue, 09 Aug 2011 23:21:11 UTC +00:00
now - 1.week
# => Mon, 02 Aug 2010 23:21:11 UTC +00:00

They translate to calls to since or advance. For example here we get the correct jump in the calendar reform:

Time.utc_time(1582, 10, 3) + 5.days
# => Mon Oct 18 00:00:00 UTC 1582

19 Extensions to Process

19.1 daemon

Ruby 1.9 provides Process.daemon, and Active Support defines it for previous versions. It accepts the same two arguments, whether it should chdir to the root directory (default, true), and whether it should inherit the standard file descriptors from the parent (default, false).

20 Extensions to File

20.1 atomic_write

With the class method File.atomic_write you can write to a file in a way that will prevent any reader from seeing half-written content.

The name of the file is passed as an argument, and the method yields a file handle opened for writing. Once the block is done atomic_write closes the file handle and completes its job.

For example, Action Pack uses this method to write asset cache files like all.css:

File.atomic_write(joined_asset_path) do |cache|
  cache.write(join_asset_file_contents(asset_paths))
end

To accomplish this atomic_write creates a temporary file. That’s the file the code in the block actually writes to. On completion, the temporary file is renamed, which is an atomic operation on POSIX systems. If the target file exists atomic_write overwrites it and keeps owners and permissions.

The auxiliary file is written in a standard directory for temporary files, but you can pass a directory of your choice as second argument.

21 Extensions to NameError

Active Support adds missing_name? to NameError, which tests whether the exception was raised because of the name passed as argument.

The name may be given as a symbol or string. A symbol is tested against the bare constant name, a string is against the fully-qualified constant name.

For example, when an action of PostsController is called Rails tries optimistically to use PostsHelper. It is OK that the helper module does not exist, so if an exception for that constant name is raised it should be silenced. But it could be the case that posts_helper.rb raises a NameError due to an actual unknown constant. That should be reraised. The method missing_name? provides a way to distinguish both cases:

def default_helper_module!
  module_name = name.sub(/Controller$/, '')
  module_path = module_name.underscore
  helper module_path
rescue MissingSourceFile => e
  raise e unless e.is_missing? "#{module_path}_helper"
rescue NameError => e
  raise e unless e.missing_name? "#{module_name}Helper"
end

22 Extensions to LoadError

Active Support adds is_missing? to LoadError, and also assigns that class to the constant MissingSourceFile for backwards compatibility.

Given a path name is_missing? tests whether the exception was raised due to that particular file (except perhaps for the “.rb” extension).

For example, when an action of PostsController is called Rails tries to load posts_helper.rb, but that file may not exist. That’s fine, the helper module is not mandatory so Rails silences a load error. But it could be the case that the helper module does exist and in turn requires another library that is missing. In that case Rails must reraise the exception. The method is_missing? provides a way to distinguish both cases:

def default_helper_module!
  module_name = name.sub(/Controller$/, '')
  module_path = module_name.underscore
  helper module_path
rescue MissingSourceFile => e
  raise e unless e.is_missing? "helpers/#{module_path}_helper"
rescue NameError => e
  raise e unless e.missing_name? "#{module_name}Helper"
end

23 Changelog

Lighthouse ticket

• August 10, 2010: Starts to take shape, added to the index.
• April 18, 2009: Initial version by Xavier Noria