Serega Ruby Serializer

The Serega Ruby Serializer provides easy and powerful DSL to describe your objects and to serialize them to Hash or JSON.

---

📌 Serega does not depend on any gem and works with any framework

---

It has some great features:

• Manually select serialized fields
• Secure from malicious queries with depth_limit plugin
• Solutions for N+1 problem (via batch, preloads or activerecord_preloads plugins)
• Built-in object presenter (presenter plugin)
• Adding custom metadata (via metadata or context_metadata plugins)
• Value formatters (formatters plugin) helps to transform time, date, money, percentage and any other values same way keeping code dry
• Conditional attributes - (if plugin)
• Auto camelCase keys - camel_case plugin

Installation

bundle add serega

Define serializers

Most apps should define base serializer with common plugins and settings to not repeat them in each serializer.

Serializers will inherit everything (plugins, config, attributes) from their superclasses.

class AppSerializer < Serega
  # plugin :one
  # plugin :two

  # config.one = :one
  # config.two = :two
end

class UserSerializer < AppSerializer
  # attribute :one
  # attribute :two
end

class CommentSerializer < AppSerializer
  # attribute :one
  # attribute :two
end

Adding attributes

class UserSerializer < Serega
  # Regular attribute
  attribute :first_name

  # Option :method specifies method that must be called on serialized object
  attribute :first_name, method: :old_first_name

  # Block is used to define attribute value
  attribute(:first_name) { |user| user.profile&.first_name }

  # Option :value can be used with proc or callable object to define attribute value
  attribute :first_name, value: UserProfile.new # must have #call method
  attribute :first_name, value: proc { |user| user.profile&.first_name }

  # Option :delegate can be used to define attribute value.
  # Sub-option :allow_nil by default is false
  attribute :first_name, delegate: { to: :profile, allow_nil: true }

  # Option :delegate can be used with :method sub-option, so method chain here
  # is user.profile.fname
  attribute :first_name, delegate: { to: :profile, method: :fname }

  # Option :const specifies attribute with specific constant value
  attribute(:type, const: 'user')

  # Option :hide specifies attributes that should not be serialized by default
  attribute :tags, hide: true

  # Option :serializer specifies nested serializer for attribute
  # We can specify serializer as Class, String or Proc.
  # Use String or Proc if you have cross references in serializers.
  attribute :posts, serializer: PostSerializer
  attribute :posts, serializer: "PostSerializer"
  attribute :posts, serializer: -> { PostSerializer }

  # Option `:many` specifies a has_many relationship
  # Usually it is defined automatically by checking `is_a?(Enumerable)`
  attribute :posts, serializer: PostSerializer, many: true

  # Option `:preload` can be specified when enabled `:preloads` plugin
  # It allows to specify associations to preload to attribute value
  attribute(:email, preload: :emails) { |user| user.emails.find(&:verified?) }

  # Options `:if`, `:unless`, `:if_value`, `:unless_value` can be specified
  # when enabled `:if` plugin. They hide attribute key and value from response.
  # See more usage examples in :if plugin section.
  attribute :email, if: proc { |user, ctx| user == ctx[:current_user] }
  attribute :email, if_value: :present?

  # Option `:format` can be specified when enabled `:formatters` plugin
  # It changes attribute value
  attribute :created_at, format: :iso_time
  attribute :updated_at, format: :iso_time

  # Option `:format` also can be used as Proc
  attribute :created_at, format: proc { |time| time.strftime("%Y-%m-%d")}
end

---

⚠️ Attribute names are checked to include only "a-z", "A-Z", "0-9", "_", "-", "~" characters.

We allow ONLY this characters as we want to be able to use attributes names in URLs without escaping.

This check can be disabled this way:

# Disable globally
Serega.config.check_attribute_name = false

# Disable for specific serializer
class SomeSerializer < Serega
  config.check_attribute_name = false
end

Serializing

We can serialize objects using class methods .to_h, .to_json, .as_json and same instance methods #to_h, #to_json, #as_json. to_h method is also aliased as call.

user = OpenStruct.new(username: 'serega')

class UserSerializer < Serega
  attribute :username
end

UserSerializer.to_h(user) # => {username: "serega"}
UserSerializer.to_h([user]) # => [{username: "serega"}]

UserSerializer.to_json(user) # => '{"username":"serega"}'
UserSerializer.to_json([user]) # => '[{"username":"serega"}]'

UserSerializer.as_json(user) # => {"username":"serega"}
UserSerializer.as_json([user]) # => [{"username":"serega"}]

If you always serialize same attributes it will make sense to save instance of serializer and reuse this instance, it will be a bit faster (fields will be prepared only once).

# Example with all fields
serializer = UserSerializer.new
serializer.to_h(user1)
serializer.to_h(user2)

# Example with custom fields
serializer = UserSerializer.new(only: [:username, :avatar])
serializer.to_h(user1)
serializer.to_h(user2)

---

⚠️ When you serialize Struct object, specify manually many: false. As Struct is Enumerable and we check object.is_a?(Enumerable) to detect if we should return array.

UserSerializer.to_h(user_struct, many: false)

Selecting Fields

By default all attributes are serialized (except marked as hide: true).

We can provide modifiers to select only needed attributes:

• only - lists attributes to serialize;
• except - lists attributes to not serialize;
• with - lists attributes to serialize additionally (By default all attributes are exposed and will be serialized, but some attributes can be hidden when they are defined with hide: true option, more on this below. with modifier can be used to expose such attributes).

Modifiers can be provided as Hash, Array, String, Symbol or their combinations.

With plugin string_modifiers we can provide modifiers as single String with attributes split by comma , and nested values inside brackets (), like: username,enemies(username,email). This can be very useful to accept list of fields in GET requests.

When provided non-existing attribute, Serega::AttributeNotExist error will be raised. This error can be muted with check_initiate_params: false parameter.

class UserSerializer < Serega
  plugin :string_modifiers # to send all modifiers in one string

  attribute :username
  attribute :first_name
  attribute :last_name
  attribute :email, hide: true
  attribute :enemies, serializer: UserSerializer, hide: true
end

joker = OpenStruct.new(
  username: 'The Joker',
  first_name: 'jack',
  last_name: 'Oswald White',
  email: 'joker@mail.com',
  enemies: []
)

bruce = OpenStruct.new(
  username: 'Batman',
  first_name: 'Bruce',
  last_name: 'Wayne',
  email: 'bruce@wayneenterprises.com',
  enemies: []
)

joker.enemies << bruce
bruce.enemies << joker

# Default
UserSerializer.to_h(bruce)
# => {:username=>"Batman", :first_name=>"Bruce", :last_name=>"Wayne"}

# With `:only` modifier
fields = [:username, { enemies: [:username, :email] }]
fields_as_string = 'username,enemies(username,email)'

UserSerializer.to_h(bruce, only: fields)
UserSerializer.new(only: fields).to_h(bruce)
UserSerializer.new(only: fields_as_string).to_h(bruce)
# =>
# {
#   :username=>"Batman",
#   :enemies=>[{:username=>"The Joker", :email=>"joker@mail.com"}]
# }

# With `:except` modifier
fields = %i[first_name last_name]
fields_as_string = 'first_name,last_name'
UserSerializer.new(except: fields).to_h(bruce)
UserSerializer.to_h(bruce, except: fields)
UserSerializer.to_h(bruce, except: fields_as_string)
# => {:username=>"Batman"}

# With `:with` modifier
fields = %i[email enemies]
fields_as_string = 'email,enemies'
UserSerializer.new(with: fields).to_h(bruce)
UserSerializer.to_h(bruce, with: fields)
UserSerializer.to_h(bruce, with: fields_as_string)
# =>
# {
#   :username=>"Batman",
#   :first_name=>"Bruce",
#   :last_name=>"Wayne",
#   :email=>"bruce@wayneenterprises.com",
#   :enemies=>[
#     {:username=>"The Joker", :first_name=>"jack", :last_name=>"Oswald White"}
#   ]
# }

# With not existing attribute
fields = %i[first_name enemy]
fields_as_string = 'first_name,enemy'
UserSerializer.new(only: fields).to_h(bruce)
UserSerializer.to_h(bruce, only: fields)
UserSerializer.to_h(bruce, only: fields_as_string)
# => raises Serega::AttributeNotExist, "Attribute 'enemy' not exists"

# With not existing attribute and disabled validation
fields = %i[first_name enemy]
fields_as_string = 'first_name,enemy'
UserSerializer.new(only: fields, check_initiate_params: false).to_h(bruce)
UserSerializer.to_h(bruce, only: fields, check_initiate_params: false)
UserSerializer.to_h(bruce, only: fields_as_string, check_initiate_params: false)
# => {:first_name=>"Bruce"}

Using Context

Sometimes you can decide to use some context during serialization, like current_user or any.

class UserSerializer < Serega
  attribute(:email) do |user, ctx|
    user.email if ctx[:current_user] == user
  end
end

user = OpenStruct.new(email: 'email@example.com')
UserSerializer.(user, context: {current_user: user})
# => {:email=>"email@example.com"}

UserSerializer.new.to_h(user, context: {current_user: user}) # same
# => {:email=>"email@example.com"}

Configuration

This is initial config options, other config options can be added by plugins

class AppSerializer < Serega
  # Configure adapter to serialize to JSON.
  # It is `JSON.dump` by default. When Oj gem is loaded then default is
  # `Oj.dump(data, mode: :compat)`
  config.to_json = ->(data) { Oj.dump(data, mode: :compat) }

  # Configure adapter to de-serialize JSON.
  # De-serialization is used only for `#as_json` method.
  # It is `JSON.parse` by default.
  # When Oj gem is loaded then default is `Oj.load(data)`
  config.from_json = ->(data) { Oj.load(data) }

  # Disable/enable validation of modifiers params `:with`, `:except`, `:only`
  # By default it is enabled. After disabling,
  # when provided not existed attribute it will be just skipped.
  config.check_initiate_params = false # default is true, enabled

  # Stores in memory prepared `plans` - list of serialized attributes.
  # Next time serialization happens with same modifiers (`only, except, with`),
  # we will reuse already prepared `plans`.
  # This defines storage size (count of stored `plans` with different modifiers).
  config.max_cached_plans_per_serializer_count = 50 # default is 0, disabled
end

Plugins

Plugin :preloads

Allows to define :preloads to attributes and then allows to merge preloads from serialized attributes and return single associations hash.

Plugin accepts options:

• auto_preload_attributes_with_delegate - default false
• auto_preload_attributes_with_serializer - default false
• auto_hide_attributes_with_preload - default false

This options are very handy if you want to forget about finding preloads manually.

Preloads can be disabled with preload: false attribute option option. Also automatically added preloads can be overwritten with manually specified preload: :another_value.

Some examples, please read comments in the code below

class AppSerializer < Serega
  plugin :preloads,
    auto_preload_attributes_with_delegate: true,
    auto_preload_attributes_with_serializer: true,
    auto_hide_attributes_with_preload: true
end

class UserSerializer < AppSerializer
  # No preloads
  attribute :username

  # `preload: :user_stats` added manually
  attribute :followers_count, preload: :user_stats,
    value: proc { |user| user.user_stats.followers_count }

  # `preload: :user_stats` added automatically, as
  # `auto_preload_attributes_with_delegate` option is true
  attribute :comments_count, delegate: { to: :user_stats }

  # `preload: :albums` added automatically as
  # `auto_preload_attributes_with_serializer` option is true
  attribute :albums, serializer: 'AlbumSerializer'
end

class AlbumSerializer < AppSerializer
  attribute :images_count, delegate: { to: :album_stats }
end

# By default preloads are empty, as we specify `auto_hide_attributes_with_preload`
# so attributes with preloads will be skipped so nothing should be preloaded
UserSerializer.new.preloads
# => {}

UserSerializer.new(with: :followers_count).preloads
# => {:user_stats=>{}}

UserSerializer.new(with: %i[followers_count comments_count]).preloads
# => {:user_stats=>{}}

UserSerializer.new(
  with: [:followers_count, :comments_count, { albums: :images_count }]
).preloads
# => {:user_stats=>{}, :albums=>{:album_stats=>{}}}

---

SPECIFIC CASE #1: Serializing same object as association

For example you decided to show your current user as "user" and "user_stats". Where stats rely on user fields and some other associations. You should specify preload: nil to preload nested associations, if any, to "user".

class AppSerializer < Serega
  plugin :preloads,
    auto_preload_attributes_with_delegate: true,
    auto_preload_attributes_with_serializer: true,
    auto_hide_attributes_with_preload: true
end

class UserSerializer < AppSerializer
  attribute :username
  attribute :user_stats,
    serializer: 'UserStatSerializer',
    value: proc { |user| user },
    preload: nil
end

SPECIFIC CASE #2: Serializing multiple associations as single relation

For example "user" has two relations - "new_profile", "old_profile", and also profiles have "avatar" association. And you decided to serialize profiles in one array. You can specify preload_path: [[:new_profile], [:old_profile]] to achieve this:

class AppSerializer < Serega
  plugin :preloads,
    auto_preload_attributes_with_delegate: true,
    auto_preload_attributes_with_serializer: true
end

class UserSerializer < AppSerializer
  attribute :username
  attribute :profiles,
    serializer: 'ProfileSerializer',
    value: proc { |user| [user.new_profile, user.old_profile] },
    preload: [:new_profile, :old_profile],
    preload_path: [[:new_profile], [:old_profile]] # <--- like here
end

class ProfileSerializer < AppSerializer
  attribute :avatar, serializer: 'AvatarSerializer'
end

class AvatarSerializer < AppSerializer
end

UserSerializer.new.preloads
# => {:new_profile=>{:avatar=>{}}, :old_profile=>{:avatar=>{}}}

SPECIFIC CASE #3: Preload association through another association

attribute :image,
  preload: { attachment: :blob }, # <--------- like this one
  value: proc { |record| record.attachment },
  serializer: ImageSerializer,
  preload_path: [:attachment] # or preload_path: [:attachment, :blob]

In this case we don't know if preloads defined in ImageSerializer, should be preloaded to attachment or blob, so please specify preload_path manually. You can specify preload_path: nil if you are sure that there are no preloads inside ImageSerializer.

---

📌 Plugin :preloads only allows to group preloads together in single Hash, but they should be preloaded manually.

There are only activerecord_preloads plugin that can be used to preload this associations automatically.

Plugin :activerecord_preloads

(depends on preloads plugin, that must be loaded first)

Automatically preloads associations to serialized objects.

It takes all defined preloads from serialized attributes (including attributes from serialized relations), merges them into single associations hash and then uses ActiveRecord::Associations::Preloader to preload associations to objects.

class AppSerializer < Serega
  plugin :preloads,
    auto_preload_attributes_with_delegate: true,
    auto_preload_attributes_with_serializer: true,
    auto_hide_attributes_with_preload: false

  plugin :activerecord_preloads
end

class UserSerializer < AppSerializer
  attribute :username
  attribute :comments_count, delegate: { to: :user_stats }
  attribute :albums, serializer: AlbumSerializer
end

class AlbumSerializer < AppSerializer
  attribute :title
  attribute :downloads_count, preload: :downloads,
    value: proc { |album| album.downloads.count }
end

UserSerializer.to_h(user)
# => preloads {users_stats: {}, albums: { downloads: {} }}

Plugin :batch

Must be used to omit N+1 when loading attributes values.

User must provide batch loader object to attribute - attribute :foo, batch: {loader: SomeLoader, id_method: :id}.

Result must be returned as Hash, where each key is one of provided ids.

class AppSerializer
  plugin :batch
end

class UserSerializer < AppSerializer
  attribute :comments_count,
    batch: { loader: SomeLoader, id_method: :id }

  attribute :company,
    batch: { loader: SomeLoader, id_method: :id },
    serializer: CompanySerializer
end

Option :loader

Loaders can be defined as a Proc, a callable value or a named Symbol Named loaders should be predefined with config.batch.define(:loader_name) { |ids| ... })

Loader can accept 1 to 3 arguments:

1. List of ids (each id will be found by using :id_method option)
2. Context
3. PlanPoint - a special object containing information about current attribute and all children and parent attributes. It can be used to preload required associations to batch values. See example how to find required preloads when using with :preloads plugin.

class AppSerializer < Serega
  plugin :batch, id_method: :id
end

class UserSerializer < Serega
  # Define loader as callable object
  attribute :comments_count,
    batch: { loader: CountLoader }

  # Define loader as a Proc
  attribute :comments_count,
    batch: { loader: proc { |ids| CountLoader.call(ids) } }

  # Define loader as a Symbol
  config.batch.define(:comments_count_loader) { |ids| CountLoader.call(ids }
  attribute :comments_count, batch: { loader: :comments_count_loader }
end

class CountLoader
  def self.call(user_ids)
    Comment.where(user_id: user_ids).group(:user_id).count
  end
end

Option :id_method

Batch plugin can be added with global :id_method option. It can be a Symbol, Proc or any callable value, which can accept current object and current context.

class SomeSerializer
  plugin :batch, id_method: :id
end

class UserSerializer < AppSerializer
  attribute :comments_count,
    batch: { loader: CommentsCountBatchLoader } # no :id_method here anymore

  attribute :company,
    batch: { loader: UserCompanyBatchLoader }, # no :id_method here anymore
    serializer: CompanySerializer
end

However, global id_method option can be overwritten via config.batch.id_method= method or in specific attributes with id_method option.

class SomeSerializer
  plugin :batch, id_method: :id # global id_method is `:id`
end

class UserSerializer < AppSerializer
  # :user_id will be used as default `id_method` for all batch attributes
  config.batch.id_method = :user_id

  # id_method is :user_id
  attribute :comments_count,
    batch: { loader: CommentsCountBatchLoader }


  # id_method is :user_id
  attribute :company,
    batch: { loader: UserCompanyBatchLoader }, serializer: CompanySerializer

  # id_method is :uuid
  attribute :points_amount,
    batch: { loader: PointsBatchLoader, id_method: :uuid }
end

Option :default

The default value for attributes without found value can be specified via :default option. By default attributes without found value will be serialized as nil. Attribute marked as many: true will be serialized as empty array []

class UserSerializer < AppSerializer
  # Missing values become empty arrays, as `many: true` option specified
  attribute :companies,
    batch: {loader: proc {}},
    serializer: CompanySerializer,
    many: true

  # Missing values become `0` as specified directly
  attribute :points_amount,
    batch: { loader: proc {}, default: 0 }
end

Batch attributes can be marked as hidden by default if plugin specified with auto_hide option. Also auto_hide option can be changed with config.batch.auto_hide= method.

Look at select serialized fields for more information about hiding/showing attributes.

class AppSerializer
  plugin :batch, auto_hide: true
end

class UserSerializer < AppSerializer
  config.batch.auto_hide = false
end

---

⚠️ ATTENTION: Batch plugin must be added to all serializers that have :batch attributes inside nested serializers. For example when you serialize User -> Album -> Song and Song has batch attribute, then batch plugin must be added to the User serializer also.

Best way would be to create one parent AppSerializer < Serega serializer and add :batch plugin once to this parent serializer.

Plugin :root

Allows to add root key to your serialized data

Accepts options:

• :root - specifies root for all responses
• :root_one - specifies root for single object serialization only
• :root_many - specifies root for multiple objects serialization only

Adds additional config options:

• config.root.one
• config.root.many
• config.root.one=
• config.root_many=

Default root is :data.

Root also can be changed per serialization.

Also root can be removed for all responses by providing root: nil. In this case no root will be added to response, but you still can to add it per serialization

 #@example Define :root plugin with different options

 class UserSerializer < Serega
   plugin :root # default root is :data
 end

 class UserSerializer < Serega
   plugin :root, root: :users
 end

 class UserSerializer < Serega
   plugin :root, root_one: :user, root_many: :people
 end

 class UserSerializer < Serega
   plugin :root, root: nil # no root by default
 end

 # @example Change root per serialization:

 class UserSerializer < Serega
   plugin :root
 end

 UserSerializer.to_h(nil)              # => {:data=>nil}
 UserSerializer.to_h(nil, root: :user) # => {:user=>nil}
 UserSerializer.to_h(nil, root: nil)   # => nil

Plugin :metadata

Depends on: :root plugin, that must be loaded first

Adds ability to describe metadata and adds it to serialized response

Added class-level method :meta_attribute, to define metadata, it accepts:

• *path [Array of Symbols] - nested hash keys.

• **options [Hash]

  • :const - describes metadata value (if it is constant)
  • :value - describes metadata value as any #callable instance
  • :hide_nil - does not show metadata key if value is nil, false by default
  • :hide_empty, does not show metadata key if value is nil or empty, false by default

• &block [Proc] - describes value for current meta attribute

class AppSerializer < Serega
  plugin :root
  plugin :metadata

  meta_attribute(:version, const: '1.2.3')
  meta_attribute(:ab_tests, :names, value: ABTests.new.method(:names))
  meta_attribute(:meta, :paging, hide_nil: true) do |records, ctx|
    next unless records.respond_to?(:total_count)

    {
      page: records.page,
      per_page: records.per_page,
      total_count: records.total_count
    }
  end
end

AppSerializer.to_h(nil)
# => {:data=>nil, :version=>"1.2.3", :ab_tests=>{:names=> ... }}

Plugin :context_metadata

Depends on: :root plugin, that must be loaded first

Allows to provide metadata and attach it to serialized response.

Accepts option :context_metadata_key with name of keyword that must be used to provide metadata. By default it is :meta

Key can be changed in children serializers using config config.context_metadata.key=(value)

class UserSerializer < Serega
  plugin :root, root: :data
  plugin :context_metadata, context_metadata_key: :meta

  # Same:
  # plugin :context_metadata
  # config.context_metadata.key = :meta
end

UserSerializer.to_h(nil, meta: { version: '1.0.1' })
# => {:data=>nil, :version=>"1.0.1"}

Plugin :formatters

Allows to define formatters and apply them on attribute values.

Config option config.formatters.add can be used to add formatters.

Attribute option :format can be used with name of formatter or with callable instance.

Formatters can accept up to 2 parameters (formatted object, context)

class AppSerializer < Serega
  plugin :formatters, formatters: {
    iso8601: ->(value) { time.iso8601.round(6) },
    on_off: ->(value) { value ? 'ON' : 'OFF' },
    money: ->(value, ctx) { value / 10**ctx[:digits) }
    date: DateTypeFormatter # callable
  }
end

class UserSerializer < Serega
  # Additionally we can add formatters via config in subclasses
  config.formatters.add(
    iso8601: ->(value) { time.iso8601.round(6) },
    on_off: ->(value) { value ? 'ON' : 'OFF' },
    money: ->(value) { value.round(2) }
  )

  # Using predefined formatter
  attribute :commission, format: :money
  attribute :is_logined, format: :on_off
  attribute :created_at, format: :iso8601
  attribute :updated_at, format: :iso8601

  # Using `callable` formatter
  attribute :score_percent, format: PercentFormmatter # callable class
  attribute :score_percent, format: proc { |percent| "#{percent.round(2)}%" }
end

Plugin :presenter

Helps to write clear code by adding attribute names as methods to Presenter

class UserSerializer < Serega
  plugin :presenter

  attribute :name
  attribute :address

  class Presenter
    def name
      [first_name, last_name].compact_blank.join(' ')
    end

    def address
      [country, city, address].join("\n")
    end
  end
end

Plugin :string_modifiers

Allows to specify modifiers as strings.

Serialized attributes must be split with , and nested attributes must be defined inside brackets (, ).

Modifiers can still be provided old way using nested hashes or arrays.

PostSerializer.plugin :string_modifiers
PostSerializer.new(only: "id,user(id,username)").to_h(post)
PostSerializer.new(except: "user(username,email)").to_h(post)
PostSerializer.new(with: "user(email)").to_h(post)

# Modifiers can still be provided old way using nested hashes or arrays.
PostSerializer.new(with: {user: %i[email, username]}).to_h(post)

Plugin :if

Plugin adds :if, :unless, :if_value, :unless_value options to attributes so we can remove attributes from response in various ways.

Use :if and :unless when you want to hide attributes before finding attribute value, and use :if_value and :unless_value to hide attributes after finding final value.

Options :if and :unless accept currently serialized object and context as parameters. Options :if_value and :unless_value accept already found serialized value and context as parameters.

Options :if_value and :unless_value cannot be used with :serializer option, as serialized objects have no "serialized value". Use :if and :unless in this case.

See also a :hide option that is available without any plugins to hide attribute without conditions. Look at select serialized fields for :hide usage examples.

 class UserSerializer < Serega
   attribute :email, if: :active? # translates to `if user.active?`
   attribute :email, if: proc {|user| user.active?} # same
   attribute :email, if: proc {|user, ctx| user == ctx[:current_user]}
   attribute :email, if: CustomPolicy.method(:view_email?)

   attribute :email, unless: :hidden? # translates to `unless user.hidden?`
   attribute :email, unless: proc {|user| user.hidden?} # same
   attribute :email, unless: proc {|user, context| context[:show_emails]}
   attribute :email, unless: CustomPolicy.method(:hide_email?)

   attribute :email, if_value: :present? # if email.present?
   attribute :email, if_value: proc {|email| email.present?} # same
   attribute :email, if_value: proc {|email, ctx| ctx[:show_emails]}
   attribute :email, if_value: CustomPolicy.method(:view_email?)

   attribute :email, unless_value: :blank? # unless email.blank?
   attribute :email, unless_value: proc {|email| email.blank?} # same
   attribute :email, unless_value: proc {|email, context| context[:show_emails]}
   attribute :email, unless_value: CustomPolicy.method(:hide_email?)
 end

Plugin :camel_case

By default when we add attribute like attribute :first_name this means:

• adding a :first_name key to resulted hash
• adding a #first_name method call result as value

But its often desired to response with camelCased keys. By default this can be achieved by specifying attribute name and method directly for each attribute: attribute :firstName, method: first_name

This plugin transforms all attribute names automatically. We use simple regular expression to replace _x to X for the whole string. We make this transformation only once when attribute is defined.

You can provide your own callable transformation when defining plugin, for example plugin :camel_case, transform: ->(name) { name.camelize }

For any attribute camelCase-behavior can be skipped when camel_case: false attribute option provided.

This plugin transforms only attribute keys, without affecting root, metadata, context_metadata plugins keys.

If you wish to select serialized fields, you should provide them camelCased.

class AppSerializer < Serega
  plugin :camel_case
end

class UserSerializer < AppSerializer
  attribute :first_name
  attribute :last_name
  attribute :full_name, camel_case: false,
    value: proc { |user| [user.first_name, user.last_name].compact.join(" ") }
end

require "ostruct"
user = OpenStruct.new(first_name: "Bruce", last_name: "Wayne")
UserSerializer.to_h(user)
# => {firstName: "Bruce", lastName: "Wayne", full_name: "Bruce Wayne"}

UserSerializer.new(only: %i[firstName lastName]).to_h(user)
# => {firstName: "Bruce", lastName: "Wayne"}

Plugin :depth_limit

Helps to secure from malicious queries that require to serialize too much or from accidental serializing of objects with cyclic relations.

Depth limit is checked when constructing a serialization plan, that is when #new method is called, ex: SomeSerializer.new(with: params[:with]). It can be useful to instantiate serializer before any other business logic to get possible errors earlier.

Any class-level serialization methods also check depth limit as they also instantiate serializer.

When depth limit is exceeded Serega::DepthLimitError is raised. Depth limit error details can be found in additional Serega::DepthLimitError#details method

Limit can be checked or changed with next config options:

• config.depth_limit.limit
• config.depth_limit.limit=

There are no default limit, but it should be set when enabling plugin.

class AppSerializer < Serega
  plugin :depth_limit, limit: 10 # set limit for all child classes
end

class UserSerializer < AppSerializer
  config.depth_limit.limit = 5 # overrides limit for UserSerializer
end

Plugin :explicit_many_option

Plugin requires to add :many option when adding relationships (relationships are attributes with :serializer option specified)

Adding this plugin makes it clearer to find if relationship returns array or single object

  class BaseSerializer < Serega
    plugin :explicit_many_option
  end

  class UserSerializer < BaseSerializer
    attribute :name
  end

  class PostSerializer < BaseSerializer
    attribute :text
    attribute :user, serializer: UserSerializer, many: false
    attribute :comments, serializer: PostSerializer, many: true
  end

Errors

• Serega::SeregaError is a base error raised by this gem.
• Serega::AttributeNotExist error is raised when validating attributes in :only, :except, :with modifiers

Release

To release a new version, read RELEASE.md.

Development

• bundle install - install dependencies
• bin/console - open irb console with loaded gems
• bundle exec rspec - run tests
• bundle exec rubocop - check code standards
• yard stats --list-undoc --no-cache - view undocumented code
• yard server --reload - view code documentation

Contributing

Bug reports, pull requests and improvements ideas are very welcome!

License

The gem is available as open source under the terms of the MIT License.