Module: ActiveRecord::QueryMethods

Adds an SQL comment to queries generated from this relation. For example:

User.annotate("selecting user names").select(:name)
# SELECT "users"."name" FROM "users" /* selecting user names */

User.annotate("selecting", "user", "names").select(:name)
# SELECT "users"."name" FROM "users" /* selecting */ /* user */ /* names */

The SQL block comment delimiters, “/*” and “*/”, will be added automatically.

Some escaping is performed, however untrusted user input should not be used.

Sets attributes to be used when creating new records from a relation object.

users = User.where(name: 'Oscar')
users.new.name # => 'Oscar'

users = users.create_with(name: 'DHH')
users.new.name # => 'DHH'

You can pass nil to #create_with to reset attributes:

users = users.create_with(nil)
users.new.name # => 'Oscar'

Specifies whether the records should be unique or not. For example:

User.select(:name)
# Might return two records with the same name

User.select(:name).distinct
# Returns 1 record per distinct name

User.select(:name).distinct.distinct(false)
# You can also remove the uniqueness

Specify associations args to be eager loaded using a LEFT OUTER JOIN. Performs a single query joining all specified associations. For example:

users = User.eager_load(:address).limit(5)
users.each do |user|
  user.address.city
end

# SELECT "users"."id" AS t0_r0, "users"."name" AS t0_r1, ... FROM "users"
#   LEFT OUTER JOIN "addresses" ON "addresses"."id" = "users"."address_id"
#   LIMIT 5

Instead of loading the 5 addresses with 5 separate queries, all addresses are loaded with a single joined query.

Loading multiple and nested associations is possible using Hashes and Arrays, similar to #includes:

User.eager_load(:address, friends: [:address, :followers])
# SELECT "users"."id" AS t0_r0, "users"."name" AS t0_r1, ... FROM "users"
#   LEFT OUTER JOIN "addresses" ON "addresses"."id" = "users"."address_id"
#   LEFT OUTER JOIN "friends" ON "friends"."user_id" = "users"."id"
#   ...

NOTE: Loading the associations in a join can result in many rows that contain redundant data and it performs poorly at scale.

Excludes the specified record (or collection of records) from the resulting relation. For example:

Post.excluding(post)
# SELECT "posts".* FROM "posts" WHERE "posts"."id" != 1

Post.excluding(post_one, post_two)
# SELECT "posts".* FROM "posts" WHERE "posts"."id" NOT IN (1, 2)

This can also be called on associations. As with the above example, either a single record of collection thereof may be specified:

post = Post.find(1)
comment = Comment.find(2)
post.comments.excluding(comment)
# SELECT "comments".* FROM "comments" WHERE "comments"."post_id" = 1 AND "comments"."id" != 2

This is short-hand for .where.not(id: post.id) and .where.not(id: [post_one.id, post_two.id]).

An ArgumentError will be raised if either no records are specified, or if any of the records in the collection (if a collection is passed in) are not instances of the same model that the relation is scoping.

Used to extend a scope with additional methods, either through a module or through a block provided.

The object returned is a relation, which can be further extended.

Using a Module

module Pagination
  def page(number)
    # pagination code goes here
  end
end

scope = Model.all.extending(Pagination)
scope.page(params[:page])

You can also pass a list of modules:

scope = Model.all.extending(Pagination, SomethingElse)

Using a Block

scope = Model.all.extending do
  def page(number)
    # pagination code goes here
  end
end
scope.page(params[:page])

You can also use a block and a module list:

scope = Model.all.extending(Pagination) do
  def per_page(number)
    # pagination code goes here
  end
end

Extracts a named association from the relation. The named association is first preloaded, then the individual association records are collected from the relation. Like so:

account.memberships.extract_associated(:user)
# => Returns collection of User records

This is short-hand for:

account.memberships.preload(:user).collect(&:user)

Specifies the table from which the records will be fetched. For example:

Topic.select('title').from('posts')
# SELECT title FROM posts

Can accept other relation objects. For example:

Topic.select('title').from(Topic.approved)
# SELECT title FROM (SELECT * FROM topics WHERE approved = 't') subquery

Passing a second argument (string or symbol), creates the alias for the SQL from clause. Otherwise the alias “subquery” is used:

Topic.select('a.title').from(Topic.approved, :a)
# SELECT a.title FROM (SELECT * FROM topics WHERE approved = 't') a

It does not add multiple arguments to the SQL from clause. The last from chained is the one used:

Topic.select('title').from(Topic.approved).from(Topic.inactive)
# SELECT title FROM (SELECT topics.* FROM topics WHERE topics.active = 'f') subquery

For multiple arguments for the SQL from clause, you can pass a string with the exact elements in the SQL from list:

color = "red"
Color
  .from("colors c, JSONB_ARRAY_ELEMENTS(colored_things) AS colorvalues(colorvalue)")
  .where("colorvalue->>'color' = ?", color)
  .select("c.*").to_a
# SELECT c.*
# FROM colors c, JSONB_ARRAY_ELEMENTS(colored_things) AS colorvalues(colorvalue)
# WHERE (colorvalue->>'color' = 'red')

Allows to specify a group attribute:

User.group(:name)
# SELECT "users".* FROM "users" GROUP BY name

Returns an array with distinct records based on the group attribute:

User.select([:id, :name])
# => [#<User id: 1, name: "Oscar">, #<User id: 2, name: "Oscar">, #<User id: 3, name: "Foo">]

User.group(:name)
# => [#<User id: 3, name: "Foo", ...>, #<User id: 2, name: "Oscar", ...>]

User.group('name AS grouped_name, age')
# => [#<User id: 3, name: "Foo", age: 21, ...>, #<User id: 2, name: "Oscar", age: 21, ...>, #<User id: 5, name: "Foo", age: 23, ...>]

Passing in an array of attributes to group by is also supported.

User.select([:id, :first_name]).group(:id, :first_name).first(3)
# => [#<User id: 1, first_name: "Bill">, #<User id: 2, first_name: "Earl">, #<User id: 3, first_name: "Beto">]

Allows to specify a HAVING clause. Note that you can’t use HAVING without also specifying a GROUP clause.

Order.having('SUM(price) > 30').group('user_id')

Allows to specify an order by a specific set of values.

User.in_order_of(:id, [1, 5, 3])
# SELECT "users".* FROM "users"
#   WHERE "users"."id" IN (1, 5, 3)
#   ORDER BY CASE
#     WHEN "users"."id" = 1 THEN 1
#     WHEN "users"."id" = 5 THEN 2
#     WHEN "users"."id" = 3 THEN 3
#   END ASC

Specify associations args to be eager loaded to prevent N + 1 queries. A separate query is performed for each association, unless a join is required by conditions.

For example:

users = User.includes(:address).limit(5)
users.each do |user|
  user.address.city
end

# SELECT "users".* FROM "users" LIMIT 5
# SELECT "addresses".* FROM "addresses" WHERE "addresses"."id" IN (1,2,3,4,5)

Instead of loading the 5 addresses with 5 separate queries, all addresses are loaded with a single query.

Loading the associations in a separate query will often result in a performance improvement over a simple join, as a join can result in many rows that contain redundant data and it performs poorly at scale.

You can also specify multiple associations. Each association will result in an additional query:

User.includes(:address, :friends).to_a
# SELECT "users".* FROM "users"
# SELECT "addresses".* FROM "addresses" WHERE "addresses"."id" IN (1,2,3,4,5)
# SELECT "friends".* FROM "friends" WHERE "friends"."user_id" IN (1,2,3,4,5)

Loading nested associations is possible using a ::Hash:

User.includes(:address, friends: [:address, :followers])

Conditions

If you want to add string conditions to your included models, you’ll have to explicitly reference them. For example:

User.includes(:posts).where('posts.name = ?', 'example').to_a

Will throw an error, but this will work:

User.includes(:posts).where('posts.name = ?', 'example').references(:posts).to_a
# SELECT "users"."id" AS t0_r0, ... FROM "users"
#   LEFT OUTER JOIN "posts" ON "posts"."user_id" = "users"."id"
#   WHERE "posts"."name" = ?  [["name", "example"]]

As the LEFT OUTER JOIN already contains the posts, the second query for the posts is no longer performed.

Note that #includes works with association names while #references needs the actual table name.

If you pass the conditions via a ::Hash, you don’t need to call #references explicitly, as #where references the tables for you. For example, this will work correctly:

User.includes(:posts).where(posts: { name: 'example' })

NOTE: Conditions affect both sides of an association. For example, the above code will return only users that have a post named “example”, and will only include posts named “example”, even when a matching user has other additional posts.

Allows you to invert an entire where clause instead of manually applying conditions.

class User
  scope :active, -> { where(accepted: true, locked: false) }
end

User.where(accepted: true)
# WHERE `accepted` = 1

User.where(accepted: true).invert_where
# WHERE `accepted` != 1

User.active
# WHERE `accepted` = 1 AND `locked` = 0

User.active.invert_where
# WHERE NOT (`accepted` = 1 AND `locked` = 0)

Be careful because this inverts all conditions before invert_where call.

class User
  scope :active, -> { where(accepted: true, locked: false) }
  scope :inactive, -> { active.invert_where } # Do not attempt it
end

# It also inverts `where(role: 'admin')` unexpectedly.
User.where(role: 'admin').inactive
# WHERE NOT (`role` = 'admin' AND `accepted` = 1 AND `locked` = 0)

Performs JOINs on args. The given symbol(s) should match the name of the association(s).

User.joins(:posts)
# SELECT "users".*
# FROM "users"
# INNER JOIN "posts" ON "posts"."user_id" = "users"."id"

Multiple joins:

User.joins(:posts, :account)
# SELECT "users".*
# FROM "users"
# INNER JOIN "posts" ON "posts"."user_id" = "users"."id"
# INNER JOIN "accounts" ON "accounts"."id" = "users"."account_id"

Nested joins:

User.joins(posts: [:comments])
# SELECT "users".*
# FROM "users"
# INNER JOIN "posts" ON "posts"."user_id" = "users"."id"
# INNER JOIN "comments" ON "comments"."post_id" = "posts"."id"

You can use strings in order to customize your joins:

User.joins("LEFT JOIN bookmarks ON bookmarks.bookmarkable_type = 'Post' AND bookmarks.user_id = users.id")
# SELECT "users".* FROM "users" LEFT JOIN bookmarks ON bookmarks.bookmarkable_type = 'Post' AND bookmarks.user_id = users.id

Alias for #left_outer_joins.

Performs LEFT OUTER JOINs on args:

User.left_outer_joins(:posts)
# SELECT "users".* FROM "users" LEFT OUTER JOIN "posts" ON "posts"."user_id" = "users"."id"

Specifies a limit for the number of records to retrieve.

User.limit(10) # generated SQL has 'LIMIT 10'

User.limit(10).limit(20) # generated SQL has 'LIMIT 20'

Specifies locking settings (default to true). For more information on locking, please see Locking.

Returns a chainable relation with zero records.

The returned relation implements the Null Object pattern. It is an object with defined null behavior and always returns an empty array of records without querying the database.

Any subsequent condition chained to the returned relation will continue generating an empty relation and will not fire any query to the database.

Used in cases where a method or scope could return zero records but the result needs to be chainable.

For example:

@posts = current_user.visible_posts.where(name: params[:name])
# the visible_posts method is expected to return a chainable Relation

def visible_posts
  case role
  when 'Country Manager'
    Post.where(country: country)
  when 'Reviewer'
    Post.published
  when 'Bad User'
    Post.none # It can't be chained if [] is returned.
  end
end

Specifies the number of rows to skip before returning rows.

User.offset(10) # generated SQL has "OFFSET 10"

Should be used with order.

User.offset(10).order("name ASC")

Specify optimizer hints to be used in the SELECT statement.

Example (for MySQL):

Topic.optimizer_hints("MAX_EXECUTION_TIME(50000)", "NO_INDEX_MERGE(topics)")
# SELECT /*+ MAX_EXECUTION_TIME(50000) NO_INDEX_MERGE(topics) */ `topics`.* FROM `topics`

Example (for PostgreSQL with pg_hint_plan):

Topic.optimizer_hints("SeqScan(topics)", "Parallel(topics 8)")
# SELECT /*+ SeqScan(topics) Parallel(topics 8) */ "topics".* FROM "topics"

Returns a new relation, which is the logical union of this relation and the one passed as an argument.

The two relations must be structurally compatible: they must be scoping the same model, and they must differ only by #where (if no #group has been defined) or #having (if a #group is present).

Post.where("id = 1").or(Post.where("author_id = 3"))
# SELECT `posts`.* FROM `posts` WHERE ((id = 1) OR (author_id = 3))

Applies an ORDER BY clause to a query.

#order accepts arguments in one of several formats.

symbols

The symbol represents the name of the column you want to order the results by.

User.order(:name)
# SELECT "users".* FROM "users" ORDER BY "users"."name" ASC

By default, the order is ascending. If you want descending order, you can map the column name symbol to :desc.

User.order(email: :desc)
# SELECT "users".* FROM "users" ORDER BY "users"."email" DESC

Multiple columns can be passed this way, and they will be applied in the order specified.

User.order(:name, email: :desc)
# SELECT "users".* FROM "users" ORDER BY "users"."name" ASC, "users"."email" DESC

strings

Strings are passed directly to the database, allowing you to specify simple SQL expressions.

This could be a source of SQL injection, so only strings composed of plain column names and simple function(column_name) expressions with optional ASC/DESC modifiers are allowed.

User.order('name')
# SELECT "users".* FROM "users" ORDER BY name

User.order('name DESC')
# SELECT "users".* FROM "users" ORDER BY name DESC

User.order('name DESC, email')
# SELECT "users".* FROM "users" ORDER BY name DESC, email

Arel

If you need to pass in complicated expressions that you have verified are safe for the database, you can use ::Arel.

User.order(Arel.sql('end_date - start_date'))
# SELECT "users".* FROM "users" ORDER BY end_date - start_date

Custom query syntax, like JSON columns for PostgreSQL, is supported in this way.

User.order(Arel.sql("payload->>'kind'"))
# SELECT "users".* FROM "users" ORDER BY payload->>'kind'

Specify associations args to be eager loaded using separate queries. A separate query is performed for each association.

users = User.preload(:address).limit(5)
users.each do |user|
  user.address.city
end

# SELECT "users".* FROM "users" LIMIT 5
# SELECT "addresses".* FROM "addresses" WHERE "addresses"."id" IN (1,2,3,4,5)

Instead of loading the 5 addresses with 5 separate queries, all addresses are loaded with a separate query.

Loading multiple and nested associations is possible using Hashes and Arrays, similar to #includes:

User.preload(:address, friends: [:address, :followers])
# SELECT "users".* FROM "users"
# SELECT "addresses".* FROM "addresses" WHERE "addresses"."id" IN (1,2,3,4,5)
# SELECT "friends".* FROM "friends" WHERE "friends"."user_id" IN (1,2,3,4,5)
# SELECT ...

Mark a relation as readonly. Attempting to update a record will result in an error.

users = User.readonly
users.first.save
#=> ActiveRecord::ReadOnlyRecord: User is marked as readonly

To make a readonly relation writable, pass false.

users.readonly(false)
users.first.save
#=> true

Use to indicate that the given table_names are referenced by an SQL string, and should therefore be JOINed in any query rather than loaded separately. This method only works in conjunction with #includes. See #includes for more details.

User.includes(:posts).where("posts.name = 'foo'")
# Doesn't JOIN the posts table, resulting in an error.

User.includes(:posts).where("posts.name = 'foo'").references(:posts)
# Query now knows the string references posts, so adds a JOIN

Allows you to change a previously set group statement.

Post.group(:title, :body)
# SELECT `posts`.`*` FROM `posts` GROUP BY `posts`.`title`, `posts`.`body`

Post.group(:title, :body).regroup(:title)
# SELECT `posts`.`*` FROM `posts` GROUP BY `posts`.`title`

This is short-hand for unscope(:group).group(fields). Note that we’re unscoping the entire group statement.

Replaces any existing order defined on the relation with the specified order.

User.order('email DESC').reorder('id ASC') # generated SQL has 'ORDER BY id ASC'

Subsequent calls to order on the same relation will be appended. For example:

User.order('email DESC').reorder('id ASC').order('name ASC')

generates a query with ORDER BY id ASC, name ASC.

Allows you to change a previously set select statement.

Post.select(:title, :body)
# SELECT `posts`.`title`, `posts`.`body` FROM `posts`

Post.select(:title, :body).reselect(:created_at)
# SELECT `posts`.`created_at` FROM `posts`

This is short-hand for unscope(:select).select(fields). Note that we’re unscoping the entire select statement.

Reverse the existing order clause on the relation.

User.order('name ASC').reverse_order # generated SQL has 'ORDER BY name DESC'

Allows you to change a previously set where condition for a given attribute, instead of appending to that condition.

Post.where(trashed: true).where(trashed: false)
# WHERE `trashed` = 1 AND `trashed` = 0

Post.where(trashed: true).rewhere(trashed: false)
# WHERE `trashed` = 0

Post.where(active: true).where(trashed: true).rewhere(trashed: false)
# WHERE `active` = 1 AND `trashed` = 0

This is short-hand for unscope(where: conditions.keys).where(conditions). Note that unlike reorder, we’re only unscoping the named conditions – not the entire where statement.

Works in two unique ways.

First: takes a block so it can be used just like Array#select.

Model.all.select { |m| m.field == value }

This will build an array of objects from the database for the scope, converting them into an array and iterating through them using Array#select.

Second: Modifies the SELECT statement for the query so that only certain fields are retrieved:

Model.select(:field)
# => [#<Model id: nil, field: "value">]

Although in the above example it looks as though this method returns an array, it actually returns a relation object and can have other query methods appended to it, such as the other methods in QueryMethods.

The argument to the method can also be an array of fields.

Model.select(:field, :other_field, :and_one_more)
# => [#<Model id: nil, field: "value", other_field: "value", and_one_more: "value">]

The argument also can be a hash of fields and aliases.

Model.select(models: { field: :alias, other_field: :other_alias })
# => [#<Model id: nil, alias: "value", other_alias: "value">]

Model.select(models: [:field, :other_field])
# => [#<Model id: nil, field: "value", other_field: "value">]

You can also use one or more strings, which will be used unchanged as SELECT fields.

Model.select('field AS field_one', 'other_field AS field_two')
# => [#<Model id: nil, field_one: "value", field_two: "value">]

If an alias was specified, it will be accessible from the resulting objects:

Model.select('field AS field_one').first.field_one
# => "value"

Accessing attributes of an object that do not have fields retrieved by a select except id will throw ::ActiveModel::MissingAttributeError:

Model.select(:field).first.other_field
# => ActiveModel::MissingAttributeError: missing attribute 'other_field' for Model

Sets the returned relation to strict_loading mode. This will raise an error if the record tries to lazily load an association.

user = User.strict_loading.first
user.comments.to_a
#=> ActiveRecord::StrictLoadingViolationError

Checks whether the given relation is structurally compatible with this relation, to determine if it’s possible to use the #and and #or methods without raising an error. Structurally compatible is defined as: they must be scoping the same model, and they must differ only by #where (if no #group has been defined) or #having (if a #group is present).

Post.where("id = 1").structurally_compatible?(Post.where("author_id = 3"))
# => true

Post.joins(:comments).structurally_compatible?(Post.where("id = 1"))
# => false

Deduplicate multiple values.

Removes an unwanted relation that is already defined on a chain of relations. This is useful when passing around chains of relations and would like to modify the relations without reconstructing the entire chain.

User.order('email DESC').unscope(:order) == User.all

The method arguments are symbols which correspond to the names of the methods which should be unscoped. The valid arguments are given in VALID_UNSCOPING_VALUES. The method can also be called with multiple arguments. For example:

User.order('email DESC').select('id').where(name: "John")
    .unscope(:order, :select, :where) == User.all

One can additionally pass a hash as an argument to unscope specific :where values. This is done by passing a hash with a single key-value pair. The key should be :where and the value should be the where value to unscope. For example:

User.where(name: "John", active: true).unscope(where: :name)
    == User.where(active: true)

This method is similar to #except, but unlike #except, it persists across merges:

User.order('email').merge(User.except(:order))
    == User.order('email')

User.order('email').merge(User.unscope(:order))
    == User.all

This means it can be used in association definitions:

has_many :comments, -> { unscope(where: :trashed) }

Returns a new relation, which is the result of filtering the current relation according to the conditions in the arguments.

#where accepts conditions in one of several formats. In the examples below, the resulting SQL is given as an illustration; the actual query generated may be different depending on the database adapter.

String

A single string, without additional arguments, is passed to the query constructor as an SQL fragment, and used in the where clause of the query.

Client.where("orders_count = '2'")
# SELECT * from clients where orders_count = '2';

Note that building your own string from user input may expose your application to injection attacks if not done properly. As an alternative, it is recommended to use one of the following methods.

Array

If an array is passed, then the first element of the array is treated as a template, and the remaining elements are inserted into the template to generate the condition. Active Record takes care of building the query to avoid injection attacks, and will convert from the ruby type to the database type where needed. Elements are inserted into the string in the order in which they appear.

User.where(["name = ? and email = ?", "Joe", "joe@example.com"])
# SELECT * FROM users WHERE name = 'Joe' AND email = 'joe@example.com';

Alternatively, you can use named placeholders in the template, and pass a hash as the second element of the array. The names in the template are replaced with the corresponding values from the hash.

User.where(["name = :name and email = :email", { name: "Joe", email: "joe@example.com" }])
# SELECT * FROM users WHERE name = 'Joe' AND email = 'joe@example.com';

This can make for more readable code in complex queries.

Lastly, you can use sprintf-style % escapes in the template. This works slightly differently than the previous methods; you are responsible for ensuring that the values in the template are properly quoted. The values are passed to the connector for quoting, but the caller is responsible for ensuring they are enclosed in quotes in the resulting SQL. After quoting, the values are inserted using the same escapes as the Ruby core method Kernel.sprintf.

User.where(["name = '%s' and email = '%s'", "Joe", "joe@example.com"])
# SELECT * FROM users WHERE name = 'Joe' AND email = 'joe@example.com';

If #where is called with multiple arguments, these are treated as if they were passed as the elements of a single array.

User.where("name = :name and email = :email", { name: "Joe", email: "joe@example.com" })
# SELECT * FROM users WHERE name = 'Joe' AND email = 'joe@example.com';

When using strings to specify conditions, you can use any operator available from the database. While this provides the most flexibility, you can also unintentionally introduce dependencies on the underlying database. If your code is intended for general consumption, test with multiple database backends.

Hash

#where will also accept a hash condition, in which the keys are fields and the values are values to be searched for.

Fields can be symbols or strings. Values can be single values, arrays, or ranges.

User.where(name: "Joe", email: "joe@example.com")
# SELECT * FROM users WHERE name = 'Joe' AND email = 'joe@example.com'

User.where(name: ["Alice", "Bob"])
# SELECT * FROM users WHERE name IN ('Alice', 'Bob')

User.where(created_at: (Time.now.midnight - 1.day)..Time.now.midnight)
# SELECT * FROM users WHERE (created_at BETWEEN '2012-06-09 07:00:00.000000' AND '2012-06-10 07:00:00.000000')

In the case of a belongs_to relationship, an association key can be used to specify the model if an ::ActiveRecord object is used as the value.

author = Author.find(1)

# The following queries will be equivalent:
Post.where(author: author)
Post.where(author_id: author)

This also works with polymorphic belongs_to relationships:

treasure = Treasure.create(name: 'gold coins')
treasure.price_estimates << PriceEstimate.create(price: 125)

# The following queries will be equivalent:
PriceEstimate.where(estimate_of: treasure)
PriceEstimate.where(estimate_of_type: 'Treasure', estimate_of_id: treasure)

::Hash conditions may also be specified in a tuple-like syntax. ::Hash keys may be an array of columns with an array of tuples as values.

Article.where([:author_id, :id] => [[15, 1], [15, 2]])
# SELECT * FROM articles WHERE author_id = 15 AND id = 1 OR author_id = 15 AND id = 2

Joins

If the relation is the result of a join, you may create a condition which uses any of the tables in the join. For string and array conditions, use the table name in the condition.

User.joins(:posts).where("posts.created_at < ?", Time.now)

For hash conditions, you can either use the table name in the key, or use a sub-hash.

User.joins(:posts).where("posts.published" => true)
User.joins(:posts).where(posts: { published: true })

No Argument

If no argument is passed, #where returns a new instance of QueryMethods::WhereChain, that can be chained with WhereChain#not, WhereChain#missing, or WhereChain#associated.

Chaining with WhereChain#not:

User.where.not(name: "Jon")
# SELECT * FROM users WHERE name != 'Jon'

Chaining with WhereChain#associated:

Post.where.associated(:author)
# SELECT "posts".* FROM "posts"
# INNER JOIN "authors" ON "authors"."id" = "posts"."author_id"
# WHERE "authors"."id" IS NOT NULL

Chaining with WhereChain#missing:

Post.where.missing(:author)
# SELECT "posts".* FROM "posts"
# LEFT OUTER JOIN "authors" ON "authors"."id" = "posts"."author_id"
# WHERE "authors"."id" IS NULL

Blank Condition

If the condition is any blank-ish object, then #where is a no-op and returns the current relation.

Add a Common Table Expression (CTE) that you can then reference within another SELECT statement.

Note: CTE’s are only supported in MySQL for versions 8.0 and above. You will not be able to use CTE’s with MySQL 5.7.

Post.with(posts_with_tags: Post.where("tags_count > ?", 0))
# => ActiveRecord::Relation
# WITH posts_with_tags AS (
#   SELECT * FROM posts WHERE (tags_count > 0)
# )
# SELECT * FROM posts

Once you define Common Table Expression you can use custom FROM value or JOIN to reference it.

Post.with(posts_with_tags: Post.where("tags_count > ?", 0)).from("posts_with_tags AS posts")
# => ActiveRecord::Relation
# WITH posts_with_tags AS (
#  SELECT * FROM posts WHERE (tags_count > 0)
# )
# SELECT * FROM posts_with_tags AS posts

Post.with(posts_with_tags: Post.where("tags_count > ?", 0)).joins("JOIN posts_with_tags ON posts_with_tags.id = posts.id")
# => ActiveRecord::Relation
# WITH posts_with_tags AS (
#   SELECT * FROM posts WHERE (tags_count > 0)
# )
# SELECT * FROM posts JOIN posts_with_tags ON posts_with_tags.id = posts.id

It is recommended to pass a query as Relation. If that is not possible and you have verified it is safe for the database, you can pass it as SQL literal using ::Arel.

Post.with(popular_posts: Arel.sql("... complex sql to calculate posts popularity ..."))

Great caution should be taken to avoid SQL injection vulnerabilities. This method should not be used with unsafe values that include unsanitized input.

To add multiple CTEs just pass multiple key-value pairs

Post.with(
  posts_with_comments: Post.where("comments_count > ?", 0),
  posts_with_tags: Post.where("tags_count > ?", 0)
)

or chain multiple #with calls

Post
  .with(posts_with_comments: Post.where("comments_count > ?", 0))
  .with(posts_with_tags: Post.where("tags_count > ?", 0))

Alias for #excluding.