Arel

This introduction is also helpful: Using Arel to Compose SQL Queries.

You can get the ast from ActiveRecord::Relation to debug SQL queries:

relation = User.all
relation.arel.ast

The arel README from commit rails/arel README.md @ 19c3c1c (2017 Dec. 6) for myself to lookup quickly. Note that Arel is not a public API, so there is no public documentation. Black magic!


Arel Really Exasperates Logicians

Arel is a SQL AST manager for Ruby. It

  1. simplifies the generation of complex SQL queries, and
  2. adapts to various RDBMSes.

It is intended to be a framework framework; that is, you can build your own ORM
with it, focusing on innovative object and collection modeling as opposed to
database compatibility and query generation.

For the moment, Arel uses Active Record's connection adapters to connect to the various engines and perform connection pooling, quoting, and type conversion.

Generating a query with Arel is simple. For example, in order to produce

SELECT * FROM users
;

you construct a table relation and convert it to SQL:

users = Arel::Table.new(:users)
query = users.project(Arel.sql('*'))
query.to_sql

Here is a whirlwind tour through the most common SQL operators. These will probably cover 80% of all interaction with the database.

First is the 'restriction' operator, where:

users.where(users[:name].eq('amy'))
# => SELECT * FROM users WHERE users.name = 'amy'

What would, in SQL, be part of the SELECT clause is called in Arel a projection:

users.project(users[:id])
# => SELECT users.id FROM users

Comparison operators =, !=, <, >, <=, >=, IN:

users.where(users[:age].eq(10)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" = 10

users.where(users[:age].not_eq(10)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" != 10

users.where(users[:age].lt(10)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" < 10

users.where(users[:age].gt(10)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" > 10

users.where(users[:age].lteq(10)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" <= 10

users.where(users[:age].gteq(10)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" >= 10

users.where(users[:age].in([20, 16, 17])).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE "users"."age" IN (20, 16, 17)

Bitwise operators &, |, ^, <<, >>:

users.where((users[:bitmap] & 16).gt(0)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE ("users"."bitmap" & 16) > 0

users.where((users[:bitmap] | 16).gt(0)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE ("users"."bitmap" | 16) > 0

users.where((users[:bitmap] ^ 16).gt(0)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE ("users"."bitmap" ^ 16) > 0

users.where((users[:bitmap] << 1).gt(0)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE ("users"."bitmap" << 1) > 0

users.where((users[:bitmap] >> 1).gt(0)).project(Arel.sql('*'))
# => SELECT * FROM "users"  WHERE ("users"."bitmap" >> 1) > 0

users.where((~ users[:bitmap]).gt(0)).project(Arel.sql('*'))
# => SELECT * FROM "users" WHERE  ~ "users"."bitmap" > 0

Joins resemble SQL strongly:

users.join(photos).on(users[:id].eq(photos[:user_id]))
# => SELECT * FROM users INNER JOIN photos ON users.id = photos.user_id

Left joins:

users.join(photos, Arel::Nodes::OuterJoin).on(users[:id].eq(photos[:user_id]))
# => SELECT FROM users LEFT OUTER JOIN photos ON users.id = photos.user_id

What are called LIMIT and OFFSET in SQL are called take and skip in Arel:

users.take(5) # => SELECT * FROM users LIMIT 5
users.skip(4) # => SELECT * FROM users OFFSET 4

GROUP BY is called group:

users.project(users[:name]).group(users[:name])
# => SELECT users.name FROM users GROUP BY users.name

The best property of Arel is its "composability," or closure under all operations. For example, to restrict AND project, just "chain" the method invocations:

users                                 \
  .where(users[:name].eq('amy'))      \
  .project(users[:id])                \
# => SELECT users.id FROM users WHERE users.name = 'amy'

All operators are chainable in this way, and they are chainable any number of times, in any order.

users.where(users[:name].eq('bob')).where(users[:age].lt(25))

The OR operator works like this:

users.where(users[:name].eq('bob').or(users[:age].lt(25)))

The AND operator behaves similarly (same exact behaviour as chained calls to .where):

users.where(users[:name].eq('bob').and(users[:age].lt(25)))

Here is an example of the DISTINCT operator:

posts = Arel::Table.new(:posts)
posts.project(posts[:title])
posts.distinct
posts.to_sql # => 'SELECT DISTINCT "posts"."title" FROM "posts"'

Aggregate functions AVG, SUM, COUNT, MIN, MAX, HAVING:

photos.group(photos[:user_id]).having(photos[:id].count.gt(5))
# => SELECT FROM photos GROUP BY photos.user_id HAVING COUNT(photos.id) > 5

users.project(users[:age].sum)
# => SELECT SUM(users.age) FROM users

users.project(users[:age].average)
# => SELECT AVG(users.age) FROM users

users.project(users[:age].maximum)
# => SELECT MAX(users.age) FROM users

users.project(users[:age].minimum)
# => SELECT MIN(users.age) FROM users

users.project(users[:age].count)
# => SELECT COUNT(users.age) FROM users

Aliasing Aggregate Functions:

users.project(users[:age].average.as("mean_age"))
# => SELECT AVG(users.age) AS mean_age FROM users

The examples above are fairly simple and other libraries match or come close to matching the expressiveness of Arel (e.g. Sequel in Ruby).

Suppose we have a table products with prices in different currencies. And we have a table currency_rates, of constantly changing currency rates. In Arel:

products = Arel::Table.new(:products)
# Attributes: [:id, :name, :price, :currency_id]

currency_rates = Arel::Table.new(:currency_rates)
# Attributes: [:from_id, :to_id, :date, :rate]

Now, to order products by price in user preferred currency simply call:

products.
  join(:currency_rates).on(products[:currency_id].eq(currency_rates[:from_id])).
  where(currency_rates[:to_id].eq(user_preferred_currency), currency_rates[:date].eq(Date.today)).
  order(products[:price] * currency_rates[:rate])

Alias

Where Arel really shines is in its ability to handle complex joins and aggregations. As a first example, let's consider an "adjacency list", a tree represented in a table. Suppose we have a table comments, representing a threaded discussion:

comments = Arel::Table.new(:comments)

And this table has the following attributes:

# [:id, :body, :parent_id]

The parent_id column is a foreign key from the comments table to itself.
Joining a table to itself requires aliasing in SQL. This aliasing can be handled from Arel as below:

replies = comments.alias
comments_with_replies = \
  comments.join(replies).on(replies[:parent_id].eq(comments[:id])).where(comments[:id].eq(1))
# => SELECT * FROM comments INNER JOIN comments AS comments_2
#    WHERE comments_2.parent_id = comments.id AND comments.id = 1

This will return the reply for the first comment.

CTE

Common Table Expressions (CTE) support via:

Create a CTE

cte_table = Arel::Table.new(:cte_table)
composed_cte = Arel::Nodes::As.new(cte_table, photos.where(photos[:created_at].gt(Date.current)))

Use the created CTE:

users.
  join(cte_table).on(users[:id].eq(cte_table[:user_id])).
  project(users[:id], cte_table[:click].sum).
  with(composed_cte)

# => WITH cte_table AS (SELECT FROM photos  WHERE photos.created_at > '2014-05-02')
#    SELECT users.id, SUM(cte_table.click)
#    FROM users INNER JOIN cte_table ON users.id = cte_table.user_id

When your query is too complex for Arel, you can use Arel::SqlLiteral:

photo_clicks = Arel::Nodes::SqlLiteral.new(<<-SQL
    CASE WHEN condition1 THEN calculation1
    WHEN condition2 THEN calculation2
    WHEN condition3 THEN calculation3
    ELSE default_calculation END
SQL
)

photos.project(photo_clicks.as("photo_clicks"))
# => SELECT CASE WHEN condition1 THEN calculation1
#    WHEN condition2 THEN calculation2
#    WHEN condition3 THEN calculation3
#    ELSE default_calculation END
#    FROM "photos"